How should you read a laboratory or imaging report? Understanding why the normal range is a property of the laboratory, not of your body

TL;DR That column on the right of your report headed "normal range" is a set of figures the laboratory issuing the report has set for itself, not a universal boundary of health. Ten chemical pathology laboratories in Hong Kong were measured once: on aspartate aminotransferase (AST), the upper limits of their reference intervals differed by as much as 47%, an absolute difference of 16 U/L; and even on the same analytical platform with the same reagents, of the three laboratories using the Abbott platform eleven of fourteen analytes had different reference intervals, of the four using Roche eleven likewise, and of the two using Beckman three (Hong Kong Med J 2019;25(4):295–304). The same paper also reports that for ten of those analytes the degree of measurement agreement between laboratories and between platforms was adequate. What is inconsistent is not the figure the machine produces but the ruler printed on the right of the report. So: read the range printed on your own report, and do not measure yourself against a range from anywhere else — including this article. "One item out of range" is the norm, not the exception. Australian Prescriber (1 April 2009) did the arithmetic: on a panel of 20 items, the chance that a healthy person has all twenty inside the ranges is only 36%, which is to say that 64% of healthy people will have at least one flagged as "abnormal". The other half matters more: a normal figure does not mean nothing is wrong. The World Health Organization (2011) states that a glycated haemoglobin below 6.5% does not exclude diabetes diagnosed by glucose testing; summary statement 4 of the American College of Gastroenterology (ACG, 2017) puts this whole article's argument in one line — a normal ALT does not necessarily exclude significant liver disease; the European Association for the Study of the Liver and two other societies (2024) state that people with MASLD and normal transaminases can still have significant steatohepatitis and progress to advanced fibrosis or cirrhosis; and the UK NICE clinical guideline CG95 states that acute coronary syndrome cannot be excluded on a normal resting twelve-lead electrocardiogram. Having sought the professional views of the Department of Health, the Hong Kong College of Pathologists and the Hong Kong Medical Association, the Consumer Council's own formulation in its report was: 「無論化驗結果正常與否亦應由醫生解讀」 — whether or not a laboratory result is normal, it should be interpreted by a doctor (14 July 2016). Want your own report back? Start with your phone. In 2025 the government added laboratory reports (9 January) and radiology reports (3 April) to the eHealth mobile app's "investigation records" function in two steps, and they can generally be seen in the app 14 days after the report is issued — but that route requires you to have registered for eHealth, to hold a valid eHealth account, and to be aged 16 or over. Microbiology and anatomical pathology records are not displayed at this stage, and radiology images cover only examinations from 1 January 2025 onwards. For a complete report, a paper copy or an image disc, the route is a data access request under the Personal Data (Privacy) Ordinance: the Hospital Authority's handling fee is 100 dollars per request and the statutory limit is 40 days, and that provision opens by stating that it is subject to sections 19(2), 20 and 28(5). This article will not, and cannot, tell you what your report means. It also deliberately publishes no set of "normal range" figures, for reasons given below. What it does is set out how reference ranges are arrived at, why every laboratory's differ, what each test actually measures, what can throw it off, what each imaging examination can and cannot see, the risk figures for contrast agents, and how someone in Hong Kong gets their own report back, at what cost and in what time.

---

#How is a "normal range" actually arrived at?

A reference range is a statistical line, not a physiological one. It is drawn from the results of a group of "reference individuals" by taking the middle 95% — so from the moment it is designed, healthy people are bound to fall outside it.

The local survey published by Chan Chun-hei and colleagues in the Hong Kong Medical Journal in 2019 defines it that way at the outset: the "normal range" is the middle 95% of a group treated as healthy. The remaining 5% are healthy too; they simply run high or low on that one test. [Note 1]

The same paper also sets out how a laboratory should establish those figures: it may group by sex and age and run a reference interval study with at least 120 reference individuals per group; but such a study takes enormous effort and resources, so a laboratory may instead adopt the reference interval of the reagent manufacturer or of the literature and verify it with the results of at least 20 reference individuals. [Note 1]

In other words, the range on your report is very often one the laboratory adopted from a manufacturer or the literature and verified to a limited extent, not one measured afresh in a Hong Kong population.

There is one further premise many take as inevitable and which is not: the 95% coverage is itself only a convention. Australian Prescriber (the Australian national prescribing journal, 1 April 2009) uses the words "by tradition" — a tradition, not a law. The same article gives a counter-example: the reference range commonly used for serum insulin may cover only one standard deviation either side of the mean, that is 68% of the reference population; on which 16% of normal people will be flagged as "high" and 16% as "low". [Note 2]

Who this section is most use to: anyone holding a report, seeing an asterisk or an H or L mark against a row, and wanting to know how much that mark weighs by itself.

---

#Why does every laboratory in Hong Kong print a different range for the same test?

Because Hong Kong has no common reference intervals at all. That is not conjecture but the result of a measurement Hong Kong made of itself, and the gaps are too large to be called slight.

Chan Chun-hei and colleagues' survey of 2019 covered ten chemical pathology laboratories in Hong Kong, fourteen general chemistry analytes and four analytical platforms (Abbott Architect, Beckman Coulter AU, Roche Cobas and Siemens Dimension EXL).

The Chinese abstract printed on the second page of that PDF gives the figures thus: between laboratories, the percentage difference in the upper limit of the reference interval reached 47% for AST (an absolute difference of 16 U/L), 29% for urea (1.8 mmol/L) and 18% for potassium (0.8 mmol/L); and in the lower limit, 24% for urea (0.6 mmol/L), 22% for phosphate (0.16 mmol/L) and 8% for total protein (5 g/L). [Note 3]

The English body then reports a still more awkward finding: it is not that the machines differ. Even on the same analytical platform, the same method and the same reagents, of the three laboratories using the Abbott platform eleven of the fourteen analytes had different reference intervals, of the four using Roche eleven likewise, and of the two using Beckman three. The paper's concrete example is potassium: laboratories 1 and 2 both use 3.6 to 5.2 mmol/L for men and women alike, while laboratory 3 uses 3.5 to 4.5 for men and 3.4 to 4.4 mmol/L for women. [Note 4]

The same paper opens the "results" of its English abstract with a finding running the other way, which this article must give as well or it would overstate the case: for ten of the analytes — albumin, alanine aminotransferase, aspartate aminotransferase, chloride, gamma-glutamyl transferase, phosphate, potassium, sodium, total protein and urea — the agreement between laboratories and between platforms was adequate; but the reference intervals adopted differed at every laboratory. [Note 5]

What is inconsistent is not the figure the machine produces but the range printed on the right of the report. That distinction has to be said plainly: this article is not saying that Hong Kong's laboratories measure wrongly. The figures they produce are consistent enough — and it is precisely because they are consistent enough that using different rulers to judge "normal" against "abnormal" is superfluous. The paper's own words are "redundant variation", not measurement error.

In one sentence: the same blood sample that is "normal" at laboratory 1 can become "high" at laboratory 3; the two use the same machine and the same reagents, and the figures they produce are much the same. The difference lies in two rulers.

The paper's own conclusion is a recommendation: that Hong Kong adopt common reference intervals for general chemistry analytes to reduce redundant variation between laboratories. [Note 6]

⚠️ That is a recommendation of 2019. Nothing among the sources cited here shows that Hong Kong has since implemented common reference intervals, so it cannot be said that Hong Kong has now standardised them.

The paper also states its own limitations, and this point must be given together with the finding: the quality assurance programme samples used for the retrospective methodological comparison were few in number and limited in selection, and the artificial materials of such samples produce greater variation through matrix effects. But immediately after stating its limitations comes its own "nevertheless" — this survey shows that methodological bias does not preclude the adoption of common reference intervals for seven general chemistry analytes. Quoting only the first half would leave a reader thinking the paper itself doubted its own finding. [Note 7]

Who this section is most use to: anyone who has changed doctor or laboratory, or is holding two reports from different institutions and trying to compare them. This section is the reason not to.

---

#On a panel of 20 items, what are the chances a healthy person has "at least one abnormal"?

36% all normal; 64% with at least one flagged abnormal. That is the real distribution of a body check report, not the "5%" of popular accounts.

Australian Prescriber (1 April 2009) treats this under a heading of its own: Why normal people often have abnormal results. Its arithmetic is that one machine running a multiple biochemical analysis can produce 20 results, and assuming those results are independent of one another and that the results in the reference population are normally distributed, only 36% of normal people will have all 20 inside the reference ranges, and 64% will have at least one "abnormal". [Note 8]

Following the source's own formula — each item independent, each with a 95% chance of falling inside the range, that is 0.95 to the power n:

• 2 items: 0.95 × 0.95 = 0.90 → 90% with both normal, and 10% with at least one abnormal (given by the source)
• 10 items: 0.95¹⁰ = 0.599 → about 60% with all ten normal, and about 40% with at least one abnormal (calculated by this article on the same formula)
• 20 items: 0.95²⁰ = 0.359 → about 36% with all twenty normal, and about 64% with at least one abnormal (given by the source)

⚠️ That formula is calculated on assumptions the source itself declares do not hold: the results are not independent of one another (the liver enzymes, for instance, are correlated), and the distribution in the reference population may not be normal. So these figures are an indication of order of magnitude, not a precise prediction. That is what the source itself writes.

But the same article goes straight on to say: the more extreme the deviation, and the more related tests are abnormal at once, the greater the chance that the abnormality has clinical meaning. So what this section explains is why there is an asterisk, not that the asterisk can be ignored.

The same article also explains why a repeat test a few months later usually comes back normal: a result at an extreme in the first place is the product of random variation running to an extreme in one direction; on a second measurement of the same person, variation of the same size in the same direction is unlikely to recur, so later measurements move towards the middle — regression to the mean. [Note 9]

As for when to repeat — nothing among the sources cited here sets a general interval for repeating an ordinary laboratory result. The only one that writes an interval down is the European Thyroid Association's, for the specific situation of subclinical hypothyroidism, of 「兩至三個月」 (see the thyroid section below). When to repeat is for a doctor to decide on the circumstances.

Who this section is most use to: someone who has had a catch-all body check package, has one or two rows flagged in red on the report, and has started searching for the words.

---

#Does "all the figures are normal" mean nothing is wrong?

It does not. This is the most important section in the article, because when "normal" is read as "safe", the error runs in the more dangerous direction.

Every sentence below is one a named, dated source says of itself.

Diabetes — the World Health Organization (2011). The boxed recommendation of the WHO's Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus (WHO/NMH/CHP/CPM/11.1): glycated haemoglobin can be used as a diagnostic test for diabetes provided that stringent quality assurance is in place, that assays are standardised to criteria aligned to the international reference values, and that there are no conditions present which preclude its accurate measurement; a cut-off point of 6.5% is recommended, and a value below 6.5% does not exclude diabetes diagnosed using glucose tests. The GRADE quality of evidence is moderate and the strength of the recommendation conditional. [Note 10]

⚠️ That 6.5% cut-off cannot be read alone as "6.6% means diabetes". Page 8 of the same report states that the diagnosis should be confirmed with a repeat glycated haemoglobin test. What this section is about is the other direction — below 6.5% does not exclude diabetes; and above 6.5% is not established on one measurement either.

The liver — the American College of Gastroenterology (2017) and EASL–EASD–EASO (2024). The ACG's Evaluation of Abnormal Liver Chemistries guideline of 2017 lists seven summary statements under "What are Truly Normal Liver Chemistry Tests?". Statement 4 is a single sentence: a normal ALT may not exclude significant liver disease. Statement 6 is about why the upper limit differs on every report: the range of the upper limit of normal for AST and ALT can differ between laboratories. [Note 11]

The specialist society's own words are "may not exclude" — which is to say that a report marked normal takes no responsibility for excluding anything for you.

And statement 2 in the same section runs the other way, reproduced here as well, because reading only statement 4 yields the opposite conclusion that being over the line does not matter: in a population without identifiable risk factors, an ALT or AST above the upper limit of normal is associated with increased liver-related mortality. [Note 11]

The EASL–EASD–EASO MASLD guideline of 2024, before giving its own upper limits, opens with the same thing: elevated liver enzymes — the transaminases — are associated with increased liver-related mortality. And having set the line, the next sentence turns to the other direction: people with MASLD and normal transaminases can still have significant steatohepatitis and progress to advanced fibrosis or cirrhosis; and as liver fibrosis worsens, the risk of liver-related outcomes — death, hospitalisation, hepatocellular carcinoma — rises as well. [Note 12]

Both directions have to be read together: normal does not exclude, and over the line is not meaningless either. The ending of the ACG's own statement 1 lands in the same place — levels above this should be assessed by a doctor. Both point towards seeing a doctor; neither points towards reassurance.

The kidney — KDIGO (2024). Page S137 of the KDIGO 2024 clinical practice guideline for chronic kidney disease (Kidney Int. 2024;105(4S):S117–S314) states: the definition of chronic kidney disease encompasses a variety of markers of kidney damage and is not limited to a reduced glomerular filtration rate and an albumin-to-creatinine ratio above 30 mg/g; but the staging system is based on only two dimensions, the glomerular filtration rate and the degree of albuminuria — and that subtle distinction is often overlooked by healthcare staff and students. [Note 13]

That is worth spelling out. The condition in the diagnostic criteria table of that KDIGO guideline (page S137) is: either of two pathways, persisting for at least three months. One is a glomerular filtration rate below 60 mL/min/1.73 m²; the other is a marker of kidney damage, including albuminuria, urine sediment abnormalities, electrolyte and other abnormalities due to tubular disorders, histological abnormalities, structural abnormalities on imaging, and a history of kidney transplantation. (This paragraph restates the content of that table in this article's own words rather than quoting it: the PDF renders the 「≥」 symbol as 「$」 on extraction, so it should not be transcribed verbatim.) In other words, a person with a normal eGFR but albumin in the urine is invisible if only the eGFR row is read.

The electrocardiogram — NICE CG95. Recommendation 1.2.2.5 of the UK National Institute for Health and Care Excellence clinical guideline CG95 (published 24 March 2010, last updated 30 November 2016) is a single sentence: do not exclude acute coronary syndrome on the basis of a normal resting twelve-lead electrocardiogram. Recommendation 1.2.2.4 supplies the other side: even in the absence of ST-segment change, other changes on a resting twelve-lead electrocardiogram — particularly Q wave and T wave changes — should raise suspicion. [Note 14]

Immediately below 1.2.2.5 comes 1.2.2.6, which answers a question a reader is bound to ask — does taking another electrocardiogram settle it? — and its own last sentence is the answer: note that the results may not be conclusive. [Note 14]

⚠️ Those words "may not be conclusive" are the guideline's own. So the idea that a few more electrocardiograms will tell you is something NICE itself does not say; and that provision is addressed to healthcare staff, who decide how often to repeat it, not to you.

And recommendation 1.2.1.8 states that a normal electrocardiogram can still trigger same-day emergency assessment: for suspected acute coronary syndrome with no reason for emergency referral, where there was chest pain within the past 12 hours but the pain has now gone and the resting twelve-lead electrocardiogram is normal, or where the last episode of pain was 12 to 72 hours before, refer for same-day emergency assessment. [Note 14]

But the words "no reason for emergency referral" in 1.2.1.8 point back at the previous recommendation, and it is that one that deals with emergency admission. Recommendation 1.2.1.7: for suspected acute coronary syndrome with current chest pain, or where the pain has now gone but there was chest pain within the past 12 hours and the resting twelve-lead electrocardiogram is abnormal or unavailable, refer as an emergency admission. And below 1.2.1.8 there is a third — 1.2.1.9 — dealing precisely with the situation where the pain has gone but it is not simply a matter of the pain having gone: where chest pain has resolved but there are signs of a complication (the example given there being pulmonary oedema), refer for hospital assessment, the urgency being a matter of clinical judgment. [Note 14]

All three recommendations have to be read together. Same-day emergency assessment is the arrangement for those whose pain has gone, whose electrocardiogram is normal, and who have no reason for emergency referral. If you have pain now, what NICE's recommendation says is emergency admission, not same-day assessment; and if the pain has stopped but something else is present, what NICE says is hospital assessment. Which is to say that "the pain has gone plus a normal electrocardiogram" does not automatically place you in the most relaxed cell. This section is not for matching yourself against: with chest pain, see a doctor or go to accident and emergency.

The chest X-ray — a systematic review. Bradley SH and colleagues (Br J Gen Pract 2019;69(689):e827-e835) systematically reviewed the sensitivity of chest X-ray for lung cancer in symptomatic people, and concluded: the highest quality studies show a sensitivity of only 77% to 80%, and general practitioners should consider whether a high-risk patient with a negative film needs further investigation. That review also states of itself that almost all the studies were of poor quality, and that the 77%–80% is for people with symptoms, not asymptomatic screening. [Note 15]

The Hong Kong government's own formulation. The blood tests page of the Cancer Online Resource Hub (cancer.gov.hk) states that even with advanced technology every blood test has its limitations, such as false positive or false negative results; and the eHealth frequently asked questions state that a patient should seek medical advice to interpret health information about themselves. [Note 16]

Who this section is most use to: someone holding an "all normal" report and planning on that basis to stop attending to some persistent symptom. This section's conclusion is one sentence: a normal report does not mean the problem does not exist; with symptoms, see a doctor, and do not use a normal report to cancel them out.

---

#Is there an authoritative Hong Kong document on how a body check report should be read? Yes, and it is from 2016

Yes. In 2016 the Consumer Council sought the professional views of the Department of Health, the Hong Kong College of Pathologists and the Hong Kong Medical Association, and the conclusion in its own report is the opposite of "the figures are normal, so we are done". (This is a Consumer Council document, not a government statement — the distinction is drawn below.)

The core sentence of the Consumer Council press release (the page self-dated 2016.07.14) is: 「故無論化驗結果正常與否亦應由醫生解讀，以免錯誤理解相關檢驗結果」. Which is to say that this is not "see a doctor if it is abnormal" but "see a doctor even if it is normal". [Note 17]

On the expert views about individual examinations, the opening sentence is the general proposition of the whole passage rather than a preamble: without clinical judgment and in the absence of any symptoms, many common body check items are of low effectiveness in detecting hidden disease. The two that follow are examples — a resting or exercise electrocardiogram used to detect coronary artery disease has a high false positive rate in people with a low incidence of disease and no symptoms whatever; and a chest X-ray used to screen for lung cancer has insufficient sensitivity and a higher probability of false negatives. [Note 17]

On blood tests, the page states: even where the results of a complete blood picture or complete blood count and the erythrocyte sedimentation rate are outside the normal range, if the person examined has no symptoms, that test alone is not enough to determine whether they have a disease. And the last sentence is the most useful in the whole statement, and speaks to the item most commonly done in a Hong Kong body check package: when assessing cardiovascular disease risk with a blood lipid test, other factors must be considered at the same time, including age, blood pressure level and whether the person smokes. One blood lipid figure alone shows nothing. [Note 17]

The same page has a further passage on tumour markers, aimed directly at what body check packages sell: tumour markers are particular substances in the blood, generally used by doctors to assist in diagnosis and in monitoring recovery from cancer, and are not an ideal tool for screening for cancer; and at present no tumour marker test has sufficient sensitivity and specificity to be used alone as a cancer screening test. [Note 17]

⚠️ Who said this document has to be made clear. The original of the press release is 「消委會就健康檢查徵詢衞生署、香港病理學專科學院及香港醫學會的專業意見」 (English version: 「The Council sought expert advice from the Department of Health (DH), the Hong Kong College of Pathologists (HKCPath), and the Hong Kong Medical Association (HKMA)」), after which the Council itself 「再將林林總總的檢查項目和專家的意見歸納成五大類別」 — grouped the many examination items and the experts' views into five categories. The verbs are "sought" and "grouped", not "co-signed". The author is the Consumer Council, and the three bodies are the side whose views were sought; this is not a co-signed document, not a joint statement and not a tripartite position paper, and neither the Chinese nor the English version contains 「聯署」, joint, jointly or co-signed. Note also that the Department of Health is a government department, not a professional body — the three cannot be lumped together as "three professional bodies".

That press release is dated 14 July 2016, which is ten years ago. It is the only Hong Kong document among the material cited here carrying the positions of the Department of Health, the Hong Kong College of Pathologists and the Hong Kong Medical Association on the limitations of body check items; and what it discusses is the properties of the tests themselves rather than a service that may since have ceased. But it must be cited with that year of 2016, and cannot be taken as current government advice.

Who this section is most use to: someone about to choose the items in a body check package, or assuming that more items is safer. The Consumer Council's conclusion is that you see a doctor first and let clinical judgment decide what to test, not the other way about.

---

#What does each test actually measure, and what can throw it off?

This section deliberately contains no "normal range" figures, for reasons given in the next section. What it discusses is the mechanism: what each test measures, what moves the figure, and what circumstances make that figure fail to reflect the state of your body.

#The complete blood count (CBC)

A complete blood count measures the numbers and characteristics of red cells, white cells and platelets. StatPearls' Normal and Abnormal Complete Blood Count With Differential (NBK604207, the source's own revision date 8 June 2024) notes that automated analysers produce "spurious" results, and that those spurious results look exactly like genuine abnormalities on the report: clumping of white cells can cause spurious leucopenia; the ethylenediaminetetraacetic acid (EDTA) in the collection tube can cause platelets to clump, producing spurious thrombocytopenia, for which it recommends repeating the test with a tube containing another anticoagulant such as citrate or oxalate. The chapter's concluding passage says why one figure is not enough: results must be interpreted in the clinical context and according to local laboratory guidance, because reference intervals belong to each laboratory; and a peripheral blood smear is essential in assessing any abnormal complete blood count. [Note 18]

An article in Mayo Clinic Proceedings in 2005 (Tefferi A et al., How to Interpret and Pursue an Abnormal Complete Blood Cell Count in Adults) takes the same reasoning to its furthest point: a person can vary considerably relative to their own baseline — their personal normal — without ever leaving the "normal" reference range. [Note 19]

The British Society of Gastroenterology's iron deficiency anaemia guideline of 2021 (Gut. 2021;70(11):2030–2051) simply declines to publish a universal haemoglobin line, recommending instead that anaemia be defined as a haemoglobin concentration below the lower limit of normal for the relevant population and for the laboratory performing the test. The same guideline also acknowledges a state in which the haemoglobin is within range while iron is already depleted, and names one phenomenon: serum ferritin is the single most useful marker of iron deficiency anaemia, but where a "falsely normal" ferritin is suspected, other blood tests — transferrin saturation, for instance — can help. [Note 20]

#Liver tests

The American College of Gastroenterology's guideline summary of 2017 begins with a question of nomenclature: the most commonly ordered liver tests are serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase and bilirubin, and these tests should be called liver chemistries or liver tests. [Note 21]

Which is to say that even the specialist society considers they should not be called "liver function" tests — because what they mainly measure is whether liver cells have been damaged, not the function of the liver.

The same summary also explains why they are to be read as a group rather than row by row: hepatocellular injury is defined as a disproportionate elevation of AST and ALT relative to alkaline phosphatase; and cholestatic injury as a disproportionate elevation of alkaline phosphatase relative to AST and ALT. [Note 21]

And the "upper limit of normal" for ALT is this article's most forceful demonstration — seven published sets of figures, with upper limits from 33 to 70 U/L in men and 23 to 45 U/L in women, and they are not even the same kind of figure.

The table is to be read like this:

1. Upper limits from 23 to 70 U/L are all in use. The lowest upper limit in the table is the Japanese common reference interval's 23 U/L for women and the highest the Nordic 70 U/L for men, more than threefold apart.
2. They are not measuring the same thing, and that is the point. The first two rows are specialist societies' clinical decision limits — "past this line it is worth following up" — each tied to one clinical context; the third is a value measured in a healthy population without liver disease risk factors; and the last four are reference intervals — "the middle 95% of a group of reference individuals falls here". Comparing three kinds of figure together is unfair, but the ordinary reader sees only one number on their report, with no marking of which kind it is.
3. Two rows in the table agree, and the reason they agree is itself a finding. The healthy adult upper limits in summary statement 1 of ACG 2017 are 29-33 IU/L in men and 19-25 IU/L in women; and the first sentence of the same passage of AASLD 2018 reports the same set. So the divergence is not a dispute between societies about "healthy normal" but lies between the healthy population research values and the reference intervals laboratories actually print. [Note 21]
4. The AASLD's 35 and 25 are not its account of "healthy normal", and its own sentence is not finished there. The opening half of the original is the scope it draws for itself: decisions in the management of chronic hepatitis B (CHB); and the closing half is its own qualification — variation between repeat measurements of the same sample has been described, and where a single elevated ALT is close to a treatment cut-off, that may prompt the clinician to repeat it. Treating it as "the normal ALT" and comparing it with the other rows mixes two figures of different purposes.
5. The Hong Kong figures in the last row are an academic proposal of 2019, not a range any Hong Kong laboratory currently uses, and not a government or Hospital Authority position. The column heading in that HKMJ table reads 「Hong Kong (proposed by this study)」.
6. The table above takes only the ALT row from HKMJ 2019 table 2; the original of that table has 18 rows of data covering 14 analytes, and lists the published common reference intervals of four jurisdictions (the United Kingdom, Australasia, Japan and the Nordic countries), plus the set that study proposed for Hong Kong. ALT is not the only row with divergence. (Of the fourteen analytes, ALT, AST, creatinine and gamma-glutamyl transferase each take two rows for men and women, so there are more rows than analytes.)
7. ⚠️ This table is about how inconsistent the upper limits are, not about it being all right to be over one. See the liver section above: an ALT or AST above the upper limit of normal is associated with increased liver-related mortality. The purpose of the table is to tell you not to measure your own report against a third party's figures, not to tell you to ignore the asterisk on it.

#Kidney tests

See the KDIGO section above: eGFR is not the only entrance to chronic kidney disease. Another common misconception is that a low creatinine means good kidneys. Page S148 of the KDIGO 2024 guideline states: a very low serum creatinine often reflects poor health, such as frailty or sarcopenia, limiting creatinine generation. Page S147 of the same guideline explains why estimating eGFR from creatinine has an inherent limitation: creatinine relates directly to muscle mass and can mislead at extremes of body size or in particular conditions — spinal cord injury, sarcopenia; while cystatin C is affected by another set of variables — steroid use, thyroid disease, cancer. [Note 22]

#Glycated haemoglobin (HbA1c)

Besides the WHO's sentence above that a value below 6.5% does not exclude diabetes diagnosed using glucose tests, there is something most reports never mention: HbA1c can be pushed up or pulled down by other conditions, and the figure itself will not show that it has been affected.

Annex one of the WHO's 2011 report (page 16) lists the factors in five classes — erythropoiesis, altered haemoglobin, glycation, erythrocyte destruction, and the assay itself. Iron deficiency and vitamin B12 deficiency push it up; erythropoietin, iron supplementation, B12 supplementation, reticulocytosis and chronic liver disease pull it down; haemoglobinopathies, fetal haemoglobin and methaemoglobin can move it either way; alcoholism and chronic renal failure push it up; aspirin and vitamins C and E pull it down; splenectomy pushes it up; splenomegaly, rheumatoid arthritis and certain drugs pull it down; hyperbilirubinaemia pushes it up; and hypertriglyceridaemia pulls it down. [Note 23]

⚠️ That table is not a complete list. The WHO's own heading is 「Some of the factors」, and its own asterisk notes that certain interferences are invisible on certain assays. The purpose of that table is not for you to match yourself against but to explain why an HbA1c figure has to be read by a doctor together with other information. That document is also from 2011.

The WHO also says one thing about the range below 6.5% that local material rarely quotes: the expert group considered that the current evidence is insufficient to make any formal recommendation on the interpretation of glycated haemoglobin levels below 6.5%. But that sentence cannot be taken out on its own, because the same report says two things before reaching that conclusion. [Note 24]

The first is about the 6.5% line itself: one test is not a diagnosis. Page 8 of that report restates the 2009 International Expert Committee report: a glycated haemoglobin of 6.5% or above is diagnostic, but the diagnosis should be confirmed with a repeat glycated haemoglobin test, unless the patient already has clinical symptoms and a plasma glucose above 11.1 mmol/l (200 mg/dl), in which case no repeat is needed. So a report reading 6.6% is, on that account, a result to be followed up and repeated, not a diagnosis already established. [Note 24]

The second is that the band below 6.5% does mean something. The same page goes straight on: levels just below 6.5% may indicate the presence of intermediate hyperglycaemia; the exact cut-off for that lower boundary remains to be defined, though the American Diabetes Association recommends 5.7–6.4% as the high-risk range, and the International Expert Committee suggests that people with a glycated haemoglobin between 6.0 and 6.5% are at particularly high risk and may be considered for diabetes prevention measures. [Note 24]

In one sentence: the WHO declines to make a formal recommendation on interpreting below 6.5%; it does not say the band is meaningless. Those two are very different — the first is an institution's restraint about pronouncing, and the second a conclusion telling people to relax. So a reader whose result is 6.1% should not read out of that WHO sentence that there is nothing to discuss.

Nothing in the material cited here is a Hong Kong Department of Health document defining that "prediabetes" band, so this article does not attribute the band to the Hong Kong Department of Health. This is a divergence between authorities, and this article reproduces both originals without choosing between them for a reader — but either way, that figure is one to take to a doctor.

#Thyroid function

StatPearls' Hypothyroidism (NBK519536, the source's own revision date 18 February 2024) states: central hypothyroidism can arise from the pituitary or the hypothalamus, and the thyroid-stimulating hormone produced in such cases can be biologically inactive, so the diagnosis should be based on free T4 rather than on thyroid-stimulating hormone. [Note 25]

In one sentence: thyroid-stimulating hormone (TSH) alone cannot settle the thyroid question — which is why doctors order free T4 with it. Central hypothyroidism is rare, and the point of the material above is not for a reader to diagnose themselves but to explain why a panel is read together.

The European Thyroid Association's 2013 guideline on subclinical hypothyroidism (Eur Thyroid J. 2013;2:215–228) defines it as a raised serum thyroid-stimulating hormone level with circulating thyroxine and triiodothyronine concentrations within the population reference range; and divides it by degree into two — mildly raised at 4.0–10.0 mU/l, and more severely raised at >10 mU/l. On a first finding of a raised thyroid-stimulating hormone with a free T4 within the reference range, the thyroid-stimulating hormone and free T4 should be measured again together with thyroid peroxidase antibodies, preferably 「兩至三個月」 apart. [Note 26]

⚠️ That 「兩至三個月」 is the European Thyroid Association's own recommendation in its 2013 guideline, for that specific situation, and is not a general repeat interval applying to every abnormal test.

Who this section is most use to: someone wanting to know what each row on a report actually measures, rather than whether their own figure "passes".

---

#Why does this article publish no set of "normal range" figures at all?

Because nothing in the material cited here can support a set of "normal ranges" applying to a Hong Kong reader; and the one Hong Kong paper that has measured the question concluded precisely that Hong Kong's laboratories are inconsistent about those figures.

That is not excessive caution but a necessary consequence of this article's argument:

• If a reference range is a property of the laboratory issuing the report (sections one and two above), then any set of figures published by a third party has no comparative force against your report.
• Between ten Hong Kong laboratories, the upper limit of the AST reference interval differed by as much as 47% (HKMJ 2019). Measuring your report against a set of figures from this article would be reading laboratory one's result by an eleventh laboratory's standard.
• ALT alone has seven published sets of figures, with upper limits from 33 to 70 U/L in men and 23 to 45 U/L in women, belonging to three different kinds of line (the table above). There is no "the" normal ALT — even the American College of Gastroenterology writes in terms that the range of the upper limit of normal for AST and ALT can differ between laboratories.
• Hong Kong government websites (*.gov.hk, 2 August 2026) publish no reference range table for the complete blood count or thyroid function. That is a bounded statement — the boundary is Hong Kong government websites, not a universal negative.

So this article's answer to "my report says X is 8.3, is that normal?" is only one thing: look at the column on the right of your report. If that column is not clear, or you want to know what that figure means for you, what is needed is a doctor, not another set of figures.

---

#What do all those English words on an electrocardiogram report mean?

This section explains terms, not your result. The machine-printed conclusion on an electrocardiogram report is provisional: a large study reviewing 2011 to 2023 showed that across the whole period 31.3% of automated electrocardiogram reports were amended after review by a doctor. That study reports a tested downward trend (trend test P < 0.001), from 42.2% in 2011–12 to 18.4% in 2023, but the period-by-period figures in its own table do not fall monotonically.

The terms first. The definitions of the following eight — sinus rhythm, sinus bradycardia, sinus tachycardia, atrial premature beats, ventricular premature beats, axis deviation, the ST segment and the T wave — are all from the corresponding StatPearls chapters, whose originals, chapter numbers and revision dates are at [Note 27]. Two points are worth making plain first: the normal resting heart rate in an adult is 60 to 100 beats per minute, and sinus bradycardia means a rhythm initiated by the sinoatrial node with normal myocardial depolarisation but fewer than 60 beats per minute; and "premature beats" on a report can mean either atrial or ventricular, which have different origins and cannot be treated as the same thing. This article gives only the definitions of these terms, and says nothing whatever about what any finding means, how often it occurs or whether it needs following up — those are for a doctor. [Note 27]

The line the machine prints is a first draft. Chiu IM and colleagues (Eur Heart J Digit Health, January 2026) reviewed 159 630 electrocardiograms, comparing the preliminary report produced by the GE Marquette™ 12SL automated analysis program word for word with the doctor's final report. The result: 31.3% of reports had been amended by a doctor in some form. On individual diagnostic terms, "prolonged QT interval" (added to 5.6% of the original reports) and "electronic ventricular pacemaker" (added to 3.6%) were frequently added by doctors; while of the automated reports carrying "inferior infarct" and "anterior infarct", 32.0% and 44.6% respectively had to be deleted. [Note 28]

⚠️ The 31.3% is the aggregate rate across the whole period 2011 to 2023, not a current rate. The body of that paper goes straight on: more recent electrocardiograms were processed by newer versions of the analysis program, and the physician amendment rate fell from 42.2% in 2011–12 to 25.6% in 2021–22 and 18.4% in 2023 (trend test P < 0.001). [Note 28]

And a downward trend is not a fall in every period. The amendment rates listed period by period in that paper's table 1 are: 42.23% in 2011/12, 30.86% in 2013/14, 47.57% in 2015/16, 39.38% in 2017/18, 21.72% in 2019/20, 25.59% in 2021/22 and 18.43% in 2023 — rising twice along the way (from 2013/14 to 2015/16, and from 2019/20 to 2021/22). What the paper itself reports is a decline established by a trend test, together with the three time points it selects in its abstract and its body, not a monotonically falling curve.

Which is to say: the machine's line has become much more accurate, but "more accurate" does not mean it need not be read by a person — in 2023 nearly a fifth of automated reports were still amended by a doctor.

This is a single-centre retrospective study (Cedars-Sinai Medical Center), one manufacturer's algorithm, over 2011 to 2023. The study also states of itself that the 159 630 analysed are not a census of all 769 088, and include a random sample of 10 000 from each of eight very high-volume readers, a step the authors themselves list as a limitation that may cause selection bias. And before inclusion the study excluded outlier physicians with extremely high amendment rates (its example being a physician who amended more than 95% of reports) — the direction of that step is to make the 31.3% lower: the excluded reports had a higher amendment rate than the rest, so the overall rate calculated is smaller than it would be with them included.

NICE CG95 has a recommendation for healthcare staff on the same point, 1.2.2.7: a resting twelve-lead electrocardiogram should, in addition to any automated interpretation, be reviewed by a healthcare professional qualified to interpret it. [Note 29]

In one sentence: the machine saying "Abnormal ECG" does not mean something is wrong with you, and its saying "Normal" does not mean nothing is. Both need a person.

What this article deliberately does not reproduce is stated here. Section 1.3.3 of NICE CG95 carries a chest pain scoring rule for clinicians — classifying chest pain as typical angina, atypical angina or non-anginal by the number of features present, with a table of features that make angina less likely. That rule is written for a doctor who has already taken a history, examined the patient and has an electrocardiogram in hand. This article will not reproduce it: a reader counting those features against themselves at two in the morning and then deciding not to go to hospital would be doing precisely what that rule does not permit. This problem cannot be solved at home — and that it cannot is itself the answer.

Who this section is most use to: someone whose body check report came with an electrocardiogram carrying a few English terms, and whose doctor spent thirty seconds saying it was nothing. If you have chest pain, breathlessness or any persistent symptom, do not use this section to judge your own case — per NICE CG95 recommendation 1.2.2.5 above, acute coronary syndrome cannot be excluded even on a completely normal electrocardiogram.

---

#What can an X-ray, an ultrasound, a CT, an MRI and a PET each see, and not see?

The difference is not which is more advanced but which physical property forms the image — and every physical property has something it inherently cannot see.

The principle and uses of each below are all from the Chinese originals of the corresponding pages of the Hong Kong government's Cancer Online Resource Hub (cancer.gov.hk), transcribed verbatim at [Note 30]. None of those pages carries a revision date, so all are given by retrieval date: 2 August 2026 for the pages below, and 3 August 2026 for the positron emission tomography page.

• X-ray imaging: an X-ray is a form of electromagnetic radiation invisible to the eye, which the patient does not feel as it passes through the body and which strikes a sensitised screen placed on the other side to produce an image; it is generally used to detect the outline, masses or lesions of the body part concerned. ⚠️ The English version of that same page is headed "X-ray Angiography", but the body text of both versions describes ordinary plain radiography, not angiography. The English heading does not match its own body; this article follows the Chinese version.
• Ultrasound: a high-frequency sound wave, at frequencies far above the range humans can hear; used normally, an ultrasound scan carries no radiation, is harmless and causes no pain. And its inherent blind spot is written on the same page: ultrasound is reflected when it meets air and bone, preventing it from penetrating tissues within the body, which is why a water-soluble gel is used as a medium during the examination. In one sentence — ultrasound cannot get past air or bone. The gel is not for slipperiness but to drive out the layer of air between the probe and the skin; and what is behind the lungs and the skull is hard for ultrasound to see.
• Computed tomography (CT): uses computer technology and a rotating X-ray gantry to render precise cross-sectional images for diagnosing disease within the body.
• Magnetic resonance imaging (MRI): uses a powerful magnetic field, radio waves and sophisticated computer technology to see through to the tissues inside the body, which that page describes as safe, accurate and without radiation or pain. But the risk passage on the same page also states: medical science does not yet know the long-term effects of a magnetic resonance examination on the human body or on a fetus.
• Positron emission tomography (PET): an advanced isotope imaging technology, used to detect whether a cancer has spread to other parts of the body. The rest of that page is one introductory sentence and three links (breast cancer, prostate cancer, oesophageal cancer).

⚠️ The Hong Kong government's own PET page publishes no radiation dose, preparation, risk or complication in either language. The computed tomography and magnetic resonance pages on the same site both have a set of sections on preparation, procedure, and risks or complications; the PET page has none. That is a bounded statement: it is about that page on cancer.gov.hk (retrieved 3 August 2026). Note also that the page is headed 「正電子電腦斷層掃描」 while the body text writes 「正電子電腦掃描」; and the introductory sentences of the Chinese and English versions do not correspond. This article transcribes the Chinese version.

What to volunteer before an MRI. The Cancer Online Resource Hub's magnetic resonance page lists eight things: having a cardiac pacemaker or an artificial heart valve or the like in the body; having any metal device fitted, implanted or worn on the body; working now or formerly in the metal trades; having been injured by a metal fragment or a bullet; having now or formerly a tattoo or tattooed eyebrows; having renal failure, kidney disease or previous kidney surgery; being or possibly being pregnant; and suffering from claustrophobia. [Note 31]

⚠️ That list is not a closed one. The first item ends with 「等」, which is open-ended in itself; and the same item in the English version lists four kinds of device beginning with "such as". Neither version claims the list is exhaustive. Not being on that list does not mean not mentioning it — any implant, surgical history or possibility of pregnancy should be raised before the examination.

Preparation before a CT. The computed tomography page on the same site lists seven things: tell the doctor beforehand of any history of asthma, urticaria, eczema, allergic rhinitis or allergic reaction to certain foods, drugs or X-ray contrast agents; if pregnancy is suspected or confirmed, be sure to tell the radiology staff before the examination; a diabetic patient should adjust their medication as directed by their doctor; do not eat or drink for four hours before the examination; most abdominal examinations require an oral contrast agent or water several hours before; an inpatient may take it orally on the ward beforehand; and where an intravenous contrast agent is required, the patient must sign a consent form. [Note 32]

The first item in the English version of that page ends with "etc.", explicitly marking it open; the Chinese version has no corresponding 「等」 on that item. Both versions should be treated as open lists throughout.

Who this section is most use to: someone who has been given a referral letter naming an examination they have never had, and wants to know what will happen when they get there.

---

#How risky are contrast agents? A government figure that exists in Chinese and not in English

Severe contrast reactions are rare, and the Hong Kong government has published specific probabilities for them — though one of those figures appears only in its Chinese version.

#The contrast agent used in computed tomography

The "risks or complications" section of the Cancer Online Resource Hub's computed tomography page has three tiers: a mild reaction is slight warmth, itching, nausea, vomiting, arm pain, sneezing, coughing and slight difficulty breathing, and some patients react later, usually feeling pain, itching, a rash or salivary gland swelling at the injection site within 24 hours, which is temporary; a moderate reaction is a rash, fever, chills, a rise or fall in blood pressure or palpitations, requiring special treatment and monitoring; and a severe reaction is breathlessness, an irregular heart rate, angina, renal failure, seizures and coma, requiring immediate treatment. Among the specific risks, extravasation of the contrast agent can occur during injection; and the chance of a patient dying from sensitivity to the contrast agent is about 「二十五萬分之一」 — about one in a quarter of a million. [Note 33]

⚠️ That list of reactions cannot be treated as closed. The page's own framing words are 「一般風險包括：」 and 「特別風險包括︰」 — "include" is open-ended, and neither the Chinese nor the English version claims the list is exhaustive. Not being on that list does not mean not mentioning it: if you feel unwell after a contrast injection, tell the radiology staff at once.

And that 「二十五萬分之一」 appears only in the Chinese version. At the corresponding place, the English version of the same page reads: in extremely rare cases, the contrast agent can cause death through a severe allergic reaction — with no figure at all. [Note 33]

This is a substantive difference between two languages of the same Hong Kong government web page about the same examination, and it is the Chinese version that carries the quantified risk. That is the opposite of the common assumption that the English version is usually the fuller one. This article follows the Chinese page and does not attribute that figure to the English page.

#Magnetic resonance: besides the contrast agent, the government lists two other risks of its own

Magnetic resonance involves no ionising radiation, but that does not mean it is without risk — and of the three items in the "general risks or complications" section of that government page, the first two have nothing to do with contrast agents at all: during the examination the machine makes a great deal of unavoidable noise, which can reach one hundred and thirty decibels or more (the English version of that page writes 「130dB or above」), and nerves may be stimulated, producing transient numbness, muscle twitching, warmth or pain; the examination may cause an anxiety reaction, excessive fear or a panic attack because of the enclosed space; and only the third concerns the contrast agent — about 1% of patients have a mild allergic reaction such as headache, dizziness, nausea or vomiting, and extravasation of the contrast agent can cause discomfort, swelling or pain at the injection site, symptoms that resolve after one to two days. [Note 34]

The "important risks or complications" section of the same page states: in rare cases a gadolinium-containing magnetic resonance contrast agent can cause a severe reaction such as seizures, severe anaphylactic shock and death, the risk of death being 「四十萬分之一」 — one in a little under half a million; incompatible metal devices or instruments in the body may move, heat up and cause burns, malfunction or adhere to the magnetic resonance machine; and a patient with renal failure receiving a gadolinium-containing contrast agent has a 1% to seven per cent chance of developing nephrogenic systemic fibrosis, in which the affected patient develops restricted movement, joint contractures and multiple organ failure, a reaction that rarely leads to death. [Note 34]

⚠️ Those two death risk figures cannot be used interchangeably. 「二十五萬分之一」 is for the iodinated contrast agent used in CT; 「四十萬分之一」 is for the gadolinium-containing agent used in MRI. They are two different drugs.

The same government website, two contrast agents, and two sets of figures it publishes for itself:

Who this section is most use to: someone booked for a CT or MRI "with contrast" who has to sign a consent form and wants to know what they are signing; someone with kidney disease who has been recommended a gadolinium-containing study — the last row of the table above is a point to raise with the doctor; and someone with claustrophobia or particular sensitivity to noise facing a magnetic resonance examination — the one hundred and thirty decibels is a general risk that government page lists itself, and is worth raising before the examination.

---

#Radiation dose: how much does one examination actually deliver?

You have to know what unit is being measured before there is anything to compare.

The United States Food and Drug Administration (FDA)'s What are the Radiation Risks from CT? (the page self-noting "Content current as of: 12/05/2017") states: the quantity most relevant for assessing the cancer risk of computed tomography is the "effective dose", in millisieverts (mSv), which allows the risk estimates of irradiating part of the body and the whole body to be compared with one another. The Department of Health Radiation Health Unit's Understanding Radiation states: the unit for measuring radiation dose is the sievert (Sv), and one millisievert is a thousandth of a sievert, one part in 1,000. [Note 35]

The FDA's own qualifications have to be read with that table, and there are two of them, not one — the second appears immediately beside Table 1 on that page, and the bounds it gives are much wider than the first's.

The first: the radiation dose of a computed tomography scan varies from person to person, depending on the size of the part examined, the type of procedure, and the model and operation of the machine; and the typical values quoted are to be regarded as estimates that cannot be applied precisely to any individual patient, examination or machine model, the actual dose being able to be two or three times larger or smaller than the estimate. The second: the estimated effective dose of a single diagnostic computed tomography examination can vary by a factor of ten or more depending on the type of scan, the patient's size, and the model and operation of the machine. [Note 36]

In one sentence: the table above is an order-of-magnitude reference and not the dose of your own examination — and the FDA gives two ranges of variation, which describe different things and must not be confused. Your actual dose can differ from the table's estimate by two or three times; but between different scan types, patient sizes, machine models and modes of operation, the estimated dose itself can vary by a factor of ten or more. The latter is wider than the former; it does not replace it.

The opening of the FDA's "risk estimates" section is not a reassuring comparison but an unreassuring one, reproduced here. The effective doses of diagnostic computed tomography are generally estimated at 1 to 10 millisieverts, a range not much lower than the lowest doses — 5 to 20 millisieverts — estimated to have been received by some Japanese atomic bomb survivors; and those survivors show a small but definite radiation-related excess relative risk of increased cancer mortality. [Note 37]

⚠️ That sentence does not say that one CT is the equivalent of an atomic bomb. What the FDA says is that the dose is of a comparable order of magnitude, and its reason for drawing that comparison is precisely that human population data exist at that order — the "one in two thousand" below is not extrapolated from nothing.

The FDA also translates that dose into a risk figure with a denominator and a comparator, and the first sentence of that passage is its own qualification: the risk of cancer from radiation exposure depends on the part of the body exposed and on age and sex at exposure. A computed tomography examination with an effective dose of 10 millisieverts may increase the possibility of a fatal cancer by about one in two thousand; and that increase can be compared with the natural incidence of fatal cancer in the United States population of about one in five, that is 400 in two thousand, making the combined risk possibly rise from 400 in two thousand to four hundred and one in two thousand. But the last sentence of the same passage is: nevertheless, this small increase in radiation-associated cancer risk for an individual can become a public health concern if large numbers of people undergo increased numbers of computed tomography screening procedures of uncertain benefit. [Note 38]

Working through the FDA's figures once: one in five = 400 in two thousand ✓; four hundred plus one = four hundred and one ✓. Which is to say that on the FDA's estimate, one 10 mSv computed tomography scan moves a person's lifetime chance of a fatal cancer from 20.0% to 20.05%.

⚠️ That calculation tells only half the story, and the other half is the FDA's own next sentence. "400 becomes 401" is an individual-level formulation; the FDA goes straight on to say that when large numbers of people have screening scans of uncertain benefit, that same small increase becomes a public health problem. That sentence points directly at the next section — self-financed "whole-body scans". And the first sentence also says the risk depends on the part of the body, on age and on sex; one in two thousand is not a figure that applies to everyone.

The FDA also states the uncertainty of such estimates: for the low levels of radiation exposure common to diagnostic radiological procedures, there is considerable uncertainty in the risk estimates; and some scientists take the view that low-dose radiation does not increase cancer risk at all, though that is a minority view. [Note 39]

The background radiation proportion in Hong Kong — two government bodies, two figures. The Department of Health Radiation Health Unit's Understanding Radiation says: natural radiation accounts for about 80% of the radiation dose we receive, and radiation levels vary between places. The proportion table on the Hong Kong Observatory's "Where does radiation come from?" page (retrieved 3 August 2026) says: natural 85.5% (radon 42.5%, gamma rays 18%, cosmic rays 14%, radioactive elements within the body 11%), and artificial 14.5% (medical 14%, radioactive fallout 0.2%, other 0.2%, nuclear industry discharges < 0.1%). Note that the "medical 14%" is a proportion of the total dose and not 14% of the artificial 14.5% — which is to say that almost all artificial radiation comes from medicine. [Note 40]

Which is to say: two Hong Kong government bodies have published two figures, 80% and 85.5%, for the same thing. This article reports both, attributing each to its source, and does not take an average.

⚠️ The material cited in the Department of Health's Understanding Radiation booklet includes the Department of Health Annual Report 1997/98, the Commissioner for Labour's Report 1997, ICRP Publication 60 (1990) and the NRPB's Living with Radiation (1998) — it is an old document, and this article takes only its qualitative statements, not treating its figures as current.

The same booklet also gives the thresholds for acute radiation injury, and the passage before those thresholds is itself a qualification, reproduced here as well: living things on earth are constantly exposed to a certain level of natural radiation, and although radiation may damage the body's cells and tissues, unless the dose received is extremely high the effect on health is negligible. Then come the thresholds: if the body is suddenly heavily irradiated (more than 1 000 millisieverts) acute radiation injury results, with short-term symptoms such as nausea, vomiting, extreme fatigue and hair loss; and if the whole-body dose reaches 10 000 millisieverts or more, life is very likely to be in danger even with proper treatment. [Note 41]

Those two figures are thresholds of acute injury, not "safe upper limits". The same page goes straight on: exposure to radiation increases the chance of cancer and of genetic defects in one's children. All three have to be read together — the Department of Health itself says the effect is negligible unless the dose is extremely high, says what 1 000 and 10 000 millisieverts do, and says that exposure to radiation increases the chance of cancer. The values of any diagnostic examination in the table above (0.02 to 16 millisieverts) are still a long way from 1 000 millisieverts, but "a long way" is not "equal to zero", and that is precisely what those three sentences of the Department of Health's mean.

NICE CG95 also has a recommendation bearing directly on a reader, 1.1.1.5: information on the risks of diagnostic investigations, including any radiation exposure, should be offered. [Note 42] Which is to say that asking about the radiation involved in your examination is something the guideline expects healthcare staff to volunteer.

Who this section is most use to: someone who has had several imaging examinations arranged in a short period and has begun to worry about having so many; and anyone concerned whether an MRI is necessarily safer than a CT — on radiation, magnetic resonance and ultrasound involve no ionising radiation, but each has other limitations and risks of its own, and which examination to choose is decided by a doctor on the clinical question.

---

#The clearer the picture, the more it picks up in passing — is that so?

It is. The clearer the picture, the more irrelevant things it picks up in passing — and those "incidental findings" have a probability, consequences and a cost.

In the section on the risks of computed tomography, what the United States Food and Drug Administration puts first is not radiation but incidental findings. It lists two main risks in all, the first being: results showing benign or incidental findings, leading to unnecessary follow-up examinations that may be invasive and carry risks of their own; and only the second being the increased possibility of inducing cancer from X-ray radiation exposure. [Note 43]

How common is it? The umbrella review published by O'Sullivan JW and colleagues in the BMJ (2018;361:k2387) covered 20 systematic reviews and 240 primary studies. Its definition: an incidental imaging finding is an imaging abnormality in a healthy asymptomatic patient, or an imaging abnormality in a symptomatic patient that is plainly unrelated to their symptoms. [Note 44]

The prevalence for each examination (table 3 of that review): the highest is chest computed tomography at 45% (95% confidence interval 36% to 55%), followed by computed tomography colonography at 38% (21% to 57%) and cardiac magnetic resonance at 34% (22% to 46%); the lowest are whole-body positron emission tomography or PET/CT at 2% (1% to 4%) and chest computed tomography for incidental pulmonary embolism at 2% (1% to 4%); with spinal magnetic resonance at 22% (19% to 26%) and brain magnetic resonance at 22% (14% to 31%) in the middle. [Note 44]

⚠️ That review's own heterogeneity qualification must go with those figures: 15 of the 20 meta-analyses had an I² above 50%, that is, high between-study heterogeneity.

Another bounded gap is one the review declares itself: it found no eligible systematic review or meta-analysis quantifying the prevalence of incidental findings on plain X-ray or ultrasound. Which is to say that the incidental finding rate for plain films and ultrasound is an unknown in that review, not a zero. [Note 44]

As for how many of those incidental findings are malignant — the review has figures organ by organ, but this article does not reproduce them one by one, because the table's own footnotes state that the analyses for the colon, the parotid and the prostate include patients known or suspected to have cancer, and that two versions of the value exist for the thyroid. A percentage detached from its footnote would misdescribe its population. Readers who need them should read table 4 of that paper and its footnotes directly.

The authors' own passage on the clinical implications: incidental findings can cause patient anxiety and lead to further investigation and treatment, some of which may do more harm than good; and incidental findings have financial consequences as well — beyond the cost of the subsequent management, they may change a patient's insurance status and premiums. [Note 44]

Who this section is most use to: someone considering paying for a "whole-body scan". This section and the Consumer Council's position of 2016 above are two sides of one question.

---

#How does someone in Hong Kong get their own laboratory or imaging report back? Open the app first, and then the statutory route

Two routes. The first needs an eHealth account but costs nothing: in 2025 the government added laboratory reports and radiology reports to the eHealth mobile app, and they can generally be seen in the app 14 days after being issued. The second is the statutory route: you have a right to a copy of your own medical record, within a limit of 40 days, and the Hospital Authority's published handling fee is 100 dollars per request and 300 dollars per image disc.

That 40-day line carries statutory conditions of its own: section 19(1) of the Personal Data (Privacy) Ordinance opens with 「在符合第(2)款及第20及28(5)條的規定下」, a qualification making it subject to a later subsection of its own and to two other sections of the ordinance — and where it cannot be met, the other side must notify you in writing before the limit expires with the reasons, while supplying whatever part it can.

#First stop: can you see your own laboratory and radiology reports inside the eHealth app?

Conditionally, yes. You must first register for eHealth, hold a valid eHealth account and be aged 16 or over; and once that is done, the laboratory reports and radiology reports the government added in two steps in 2025 can generally be seen for yourself in the app's "investigation records" function 14 days after the report is issued, without paying, without waiting 40 days and without paying 100 dollars. But two classes of laboratory record are not displayed at this stage, and radiology images have a date boundary.

The first thing: this route needs an eHealth account, and it has an age floor. The eHealth frequently asked questions page lists four conditions: having registered for eHealth and holding a valid eHealth account; being aged 16 or above; being capable of managing one's own affairs and matters relating to participation in eHealth; and providing a mobile telephone number or email address to receive eHealth notifications. [Note 45]

Which is to say that someone who has not registered for eHealth, or is under 16, cannot use the 14-day self-service access described below. The same page also states that for a child under 16, a person with parental responsibility for them may apply through the mobile app to become their "carer" and manage the account on their behalf.

As for money: this is eHealth's own statement and not this article's estimate — participation in and use of eHealth by healthcare providers and by patients alike is voluntary and free of charge.

The eHealth "Sharable Data" page lists the scope of sharable data as: personal particulars; allergies and adverse drug reactions; diagnosis, procedures and other medical procedures and medication; admission, encounter and appointment information; clinical summaries; birth and immunisation records; laboratory and radiology reports; other investigation reports; medical referral information; observations and lifestyle records; and medical certificates. [Note 46]

But "within the scope of sharing" and "visible to the patient in a mobile app" are two different things. The four examples in that passage of the eHealth frequently asked questions — medication, appointments, allergies, vaccination — do not mention laboratory or radiology; but the second sentence of the same passage is a signpost, directing you to app.ehealth.gov.hk, where the answer is. [Note 47]

The eHealth thematic website (app.ehealth.gov.hk, retrieved 3 August 2026), frequently asked questions 5 and 6, state: members of the public can view through the app the general routine investigation and diagnostic laboratory records deposited by private healthcare providers, the Hospital Authority and the Department of Health, while microbiology (culture and antibiotic sensitivity testing) and anatomical pathology records are not displayed in the app at this stage, and laboratory records generally appear 14 days after the result is issued; and on radiology, members of the public can view the radiology report PDF and the radiology images of examinations on or after 1 January 2025, generally available 14 days after the radiology report is issued, the radiology records available depending on what the healthcare provider has deposited into eHealth. [Note 48]

The government also issued a press release for each of those two extensions: the Health Bureau's "eHealth App Enhances Function for Viewing Healthcare Providers' Laboratory Reports" (9 January 2025) and "eHealth App Adds Radiology Report Viewing Function" (3 April 2025). The latter also updated the nine categories of electronic health record viewable by patients. [Note 49]

In one sentence: when asking how to get your laboratory report, open the app first rather than filing at once. But four things have to be kept distinct, all from the originals quoted above:

• Microbiology (culture and antibiotic sensitivity testing) and anatomical pathology records are not displayed in the app at this stage. Which is to say that culture reports and histology reports have to go by the data access request route below.
• On radiology, what the app shows is the radiology report PDF; the radiology images themselves cover only examinations on or after 1 January 2025.
• Which records are visible depends on the healthcare provider that deposited them. The original says healthcare provider, without distinguishing public from private. On the private side there is a further layer: whether you have given that provider a sharing consent.
• What the app displays is not your complete health record, and the first sentence states that premise. Question 3 of the "functions" section of the eHealth thematic website's frequently asked questions states: after logging in you can access part of your electronic health record, including medication records, appointment records, allergies, vaccination records, investigation records and dental condition; note that this is not your complete health record and cannot replace the records kept for you by healthcare providers. [Note 47]

So the app is not a substitute but a first stop. For a complete report, a paper copy, an image disc, or the two classes of laboratory record the app does not display, the statutory route is needed.

#What the law says

Section 18 of the Personal Data (Privacy) Ordinance (Chapter 486 of the Laws of Hong Kong) gives you two rights: to require a data user to inform you whether they hold your personal data; and, if they do, to require them to supply a copy. [Note 50]

Section 19(1) sets the time limit: subject to subsection (2) and to sections 20 and 28(5), a data user must comply within 40 days of receiving a data access request — where they hold the data, by informing you in writing and supplying a copy; and where they do not hold it, by informing you in writing that they do not. [Note 51]

⚠️ Note that opening phrase, 「在符合第(2)款及第20及28(5)條的規定下」: the ordinance itself states in its very first words that the 40 days have exceptions. Note also limb (b) — if the other side does not have your record at all, their statutory duty is to tell you in writing that they do not have it.

Subsection (3)(c) of the same section is about what the copy you receive has to be like. That subsection is rarely quoted, and every part of it carries a qualification of its own that cannot be cut short: the copy must so far as practicable be intelligible (except where it is a true copy of a document that is on its face not intelligible); must be comprehensible where any coding used by the data user has been sufficiently explained; and, except as provided in sub-subparagraph (B), must be in the language specified in the request. [Note 52]

Section 28(3) draws a line on fees: no fee imposed for complying with a data access request shall be excessive. [Note 53]

⚠️ The text of Chapter 486 cited here is the verified copy in the Department of Justice database as at 1 October 2022. If there have been amendments touching those provisions since that date, this article does not reflect them. Note also that section 20 sets out the circumstances in which a data user must refuse and may refuse to comply with a data access request; this article does not reproduce that section in full and does not present any part of it as a complete list — readers who need it should read section 20 of Chapter 486 in full.

#What to do in practice, and what it costs

The Hospital Authority's data access request page (the page footer self-dated 「二零二六年一月一日」) sets out the ordinance's exception itself, and in more readable terms than the ordinance: the Hospital Authority must comply within 40 days; and where it cannot comply within 40 days, it must notify the applicant in writing within that 40-day limit of the position and the reasons, comply to the extent it is able first, and thereafter comply, or comply fully, as soon as practicable. [Note 54]

Its formulation corresponds exactly to section 19(2) of the Personal Data (Privacy) Ordinance: where a data user is unable to comply within the specified period, they must before the expiry of the period inform the requestor in writing of that inability and the reasons, comply to the extent (if any) that they are able, and thereafter comply, or comply fully, as soon as practicable. [Note 55]

What that means for you in practice: if the record has not arrived on day 40, the data user is not simply in breach — the ordinance allows them to fail, but requires them to notify you in writing before the 40 days expire, giving the reasons, and to supply first whatever part they can, completing it as soon as practicable afterwards. So on day 41 you should have something in hand, and there are four lawful outcomes in all: the record itself; a notice of delay giving the reasons (together with the part that can be supplied); a written notice that they do not hold the record (section 19(1)(b)); or a written notice refusing to comply under section 20. Only if none of the four arrives has the ordinance not been complied with.

The same passage has one further practical point that few people know: for the purposes of a private doctor's consultation, a patient may authorise their private doctor to contact the Hospital Authority doctor responsible in order to obtain the patient's medical record. Which is to say that if you have moved to a private doctor who wants your public hospital record, there is, besides filing for a copy yourself, a doctor-to-doctor route on your authorisation. That is not a data access request; it has no 100-dollar handling fee and is not governed by the 40-day line — but it is for your private doctor to walk.

As for how to make one: a data access request may be made in writing or on a data access request form. [Note 56]

Two things that must not be run together (the difference will cost you ten times over)

• A data access request: getting back a copy of your own records as they already exist — laboratory reports, image discs, case notes. The basic handling fee is 100 dollars.
• A medical report: a doctor writing you a new report (usually for insurance or a claim). That is not a data access request but a different product at a different price.

The fee notes of Queen Mary Hospital's Data Access Request form (form number HIRO-003, version 20260611) state: a data enquiry request is to establish whether the hospital holds the data subject's personal data or medical records, and no charge is made for such an enquiry; the handling fee is 100 dollars per request (covering the reproduction of no more than ten pages and the postage); and the reproduction charge for X-ray films, computed tomography films, electrocardiograms, electroencephalograms and the like is 300 dollars per imaging modality per disc and 300 dollars per film. [Note 57]

Two practical things, both from the original quoted above:

1. Asking "do you have my record" is free.
2. The charging unit for imaging is per modality per disc — which is to say that one CT plus one MRI is two 300-dollar charges, not one.

As for whether the handling fee is refunded if the record cannot be found — the same form says so twice, and says two different things. The first page reads 「若有關查詢資料要求未能跟進，上述處理費將予以退還」, and its English column reads that the handling fee is refundable where the medical record cannot be found or obtained; while the declaration on page 7 that the applicant has to sign reads 「初步處理費是不予退還的」. [Note 58]

⚠️ The two divergences have to be spelled out, because there is more than one layer. The first is refund or no refund: the first page says it will be refunded and page 7 says it will not, and page 7's sentence leaves no exception for the first page's situation — and page 7's is precisely the passage you have to sign. The second is that the refund condition differs between the Chinese and the English: the English is that the record cannot be found or obtained; the Chinese is 「若有關查詢資料要求未能跟進」 — the two are not the same condition, and that Chinese 「查詢資料要求」 is the name of a different kind of request, one the same page has stated two paragraphs earlier is free of charge. So this article will not tell you that it is refunded if the record cannot be found. That is a point to put to the receiving hospital before filing. And the above is the content of Queen Mary Hospital's form; the Hospital Authority's own page also states that individual hospitals or institutions may have forms of their own, so these details cannot be treated as uniform across all public hospitals.

The price on the medical report side is not what the fee table prints. Queen Mary Hospital's Application for Medical Report form (HIRO-002, version 20260611) states: the basic cost of each medical report issued, or insurance claim form completed, by an individual clinical department is 1,100 Hong Kong dollars, and depending on whether the report requires special professional handling, the maximum charge is 4,400 Hong Kong dollars. [Note 59]

⚠️ The Hospital Authority's fee table row reads 「證明書及醫療報告 (最多收取4份報告收費) — 每份報告 1,100 元」. The form quoted above shows that 1,100 dollars is the starting price per specialty and not a ceiling; the ceiling is 4,400 dollars. The fee table's "at most 4" caps the number of reports, not the price. Those two things are easily confused.

#Why does getting a copy of a medical record cost so much?

A footnote in the Privacy Commissioner for Personal Data's Proper Handling of Data Access Requests and Charging of Data Access Request Fees by Data Users (the document self-dated July 2020) has a specific explanation for medical records: a doctor's review of the case notes is a cost directly related to and necessary for complying with the data access request, because case notes contain sensitive personal data deserving special protection, and it is necessary for a doctor to review them before they are released to the requestor. [Note 60]

⚠️ That footnote cites two different administrative appeal case numbers in its Chinese and English versions, and the number cited in the Chinese version is the same as the one cited in another footnote of the same document. One of them must be wrong, so this article quotes only its substance and cites no case number.

The summary of the same document states: a data user may impose a fee that is not excessive for complying with a data access request, and should inform the requestor clearly and as soon as practicable within 40 days how much is to be charged. [Note 61]

And Exercising the Right of Data Access under the Ordinance (Privacy Commissioner for Personal Data, the document self-dated June 2016) says what to do about an excessive charge: if you consider the fee excessive, you may raise it with the data user; and if you are not satisfied with the explanation, you may complain to the office. The same document also names medical records as a typical example — common examples include an employee obtaining a copy of a performance appraisal report, a patient obtaining a copy of a medical record, and a consumer obtaining a copy of a service application form. [Note 62]

#"If I specify Chinese, must they give me Chinese?" The ordinance answers itself

No. The ordinance states that where the record is itself held in English and what you are getting is a true copy, that copy is to be in English — which is to say that specifying a language cannot be used to require a translation of a true copy.

Item 4 of Queen Mary Hospital's Application for Medical Report form (HIRO-002, version 20260611) states: all medical reports and patient information are written in English, and the hospital does not provide a translation service. [Note 63]

That sentence is the institutional answer to why so many people cannot understand their report, and it does not conflict with the ordinance. The opening words of section 19(3)(c)(iii) quoted above are 「除(B)分節另有規定外，須採用該項要求所指明的語文」 — and those first seven characters are the signpost; sub-subparagraph (B) states that where the language in which the data is held is not the language you specified and the copy is a true copy of the document containing the data, the copy must be in a language other than the one you specified. In other words, the true copy of a case note written in English is one that ought to come to you in English.

Subparagraph (i) of the same paragraph carries a parallel qualification: the copy must be "intelligible", except where it is a true copy of a document that is on its face not intelligible. A photocopy of an original case note covered in clinical abbreviations is exactly what that exception describes.

In one sentence: the general rule the ordinance protects is that you receive a copy in the language you specified and that is intelligible; a "true copy" is not a separate guarantee on top of that protection but a narrow exception — only where the original document is itself in another language, or is on its face not intelligible, is the data user relieved of translating or rewriting it for you and free to supply the original as it stands. For an explanation you can understand, you need a doctor, not a tick in the "Chinese" box on an application form.

⚠️ There is one subparagraph the above exception does not cover, which this article reports without drawing a legal conclusion. Subparagraph 19(3)(c)(ii) requires a copy so far as practicable to be comprehensible where any coding used by the data user has been sufficiently explained — and that subparagraph carries no true-copy exception of its own. Whether the clinical abbreviations in a case note count as "coding" within it is something no case or Privacy Commissioner decision among the material cited here has addressed. And the above is a statement in Queen Mary Hospital's form, not a Hospital Authority-wide policy statement.

#On the eHealth side: a different data user, a different fee schedule

The statutory data access request route covers the whole record, while eHealth's Electronic Health Record Application and Enquiry Centre is, under the same ordinance, a data user independent of the Hospital Authority. The eHealth "Obtaining a Copy of Your Electronic Health Record" page (the page self-noting 「查閱資料要求費用最後更新日期：2024年10月17日」) states: the centre will provide the applicant with the report within a reasonably practicable time (within 40 days); the handling fee is 109 Hong Kong dollars, plus a materials charge by format — 33.9 Hong Kong dollars per USB memory stick (32GB), or 2.5 Hong Kong dollars per DVD (4.7GB), or 0.1 Hong Kong dollars per A4 sheet for paper printing (whether single- or double-sided); where the data access request includes radiology image data, the report is issued on a USB memory stick only; and all administrative fees paid are non-refundable. [Note 64]

⚠️ Remember that last sentence. eHealth states in terms that all administrative fees paid are non-refundable, and that is the last sentence of the whole fee passage on that page, running the opposite way from the first page of the Queen Mary Hospital form above — so this varies by data user and has to be put to the receiving party before filing. Note also that the 109 dollars and the Hospital Authority's 100 dollars above are two separate fee schedules and cannot be used interchangeably: eHealth's Electronic Health Record Application and Enquiry Centre and the Hospital Authority are two different data users under the same ordinance, each with a fee schedule of its own.

Who this section is most use to: anyone changing doctor, emigrating with their records, applying for insurance, or simply wanting to keep their own reports over the years. And two plain statements: first, many people wanting a laboratory or radiology report can see it by opening the eHealth app, and without paying — provided they have registered for eHealth, hold a valid account and are aged 16 or over; and second, a complete copy is a right and not a favour — the ordinance says 40 days (section 19(1), subject to sections 19(2), 20 and 28(5)), and asking "do you have my record" is free.

---

#What to do next

This section will not tell you what your report means — no article can. Each item below is a position taken by a source already cited above.

1. Measure against the range in the column on the right of your own report, and not against a range from anywhere else. The reasons are in sections one and two: between ten Hong Kong laboratories, the upper limit of the AST reference interval differed by as much as 47% (HKMJ 2019).
2. Normal or abnormal, take it to a doctor. The Consumer Council's formulation (14 July 2016), after seeking the views of the Department of Health, the Hong Kong College of Pathologists and the Hong Kong Medical Association, is 「無論化驗結果正常與否亦應由醫生解讀」; and eHealth's is that a patient should seek medical advice to interpret health information about themselves. This article has said that healthy people commonly have a row or two flagged abnormal, but the same sources also say that abnormal is not meaningless — summary statement 2 of the American College of Gastroenterology in 2017 and EASL 2024 both state that transaminases above the upper limit are associated with increased liver-related mortality. Both directions point to the same action: take it to a doctor.
3. If you have symptoms, say so, and do not use a normal report to cancel them out. The "normal does not mean nothing is wrong" section above sets out the express statements of the WHO, the ACG, EASL, KDIGO and NICE on this — of which the American College of Gastroenterology's 2017 sentence is the most direct: a normal ALT may not exclude significant liver disease.
4. Keep your reports over the years, because "change relative to yourself" is the one thing a report cannot show. Australian Prescriber (1 April 2009) states: because reference ranges are population-based, a patient's results may sit consistently near the top or the bottom of the normal range; so a clinically significant change can occur while the results never leave the population reference range. An older person whose plasma creatinine has always sat near the lower limit of the reference range, for instance, and which then rises towards the upper limit, may have had a significant deterioration in kidney function.
5. To see a laboratory or radiology report again, open the eHealth app first. In 2025 the government added laboratory reports (from 9 January) and radiology reports (from 3 April) to the app's "investigation records" function, and they can generally be seen for yourself 14 days after the report is issued, without paying. Provided you have registered for eHealth, hold a valid account and are aged 16 or over. But microbiology (culture and antibiotic sensitivity testing) and anatomical pathology records are not displayed at this stage, radiology images cover only examinations on or after 1 January 2025, and what the app displays is not your complete health record.
6. For a paper copy, a disc, or what the app does not show, use a data access request. 40 days (section 19(1), subject to sections 19(2), 20 and 28(5)); the Hospital Authority's handling fee of 100 dollars per request; and 300 dollars per imaging modality per disc; while asking "do you have my record" is free. If the record has not arrived on day 40, the ordinance requires the other side to give the reasons in writing before the limit expires and to supply whatever part it can; if they do not hold your record at all they must tell you so in writing; and if they are refusing to comply under section 20, the reasons for refusal must be given in writing too. And if it is for the purposes of your private doctor's consultation, that Hospital Authority page states that a patient may authorise their private doctor to contact the Hospital Authority doctor responsible directly for the record — a route that does not require you to file anything.
7. Whether and when to repeat a test is for a doctor. Among the sources cited here, none gives a general repeat interval other than the European Thyroid Association's 「兩至三個月」 for the one specific situation of subclinical hypothyroidism.

---

#Frequently asked questions

Why does my report have a row or two marked "abnormal" when there is nothing wrong with me?

Because a reference range is set at the middle 95% of a group of reference individuals, so by design healthy people fall outside it. Australian Prescriber (1 April 2009) did the arithmetic on its own formula: on a panel of 20 items, the chance that a healthy person has all 20 inside the ranges is only 36%, so 64% have at least one flagged abnormal. That source also states that the formula rests on two assumptions it itself identifies as not holding (independence and normal distribution), so it is an indication of order of magnitude rather than a precise prediction. But that does not mean an asterisk can be treated as unseen. The same article goes straight on to say that the more extreme the deviation and the more related tests are abnormal at once, the greater the chance the abnormality has clinical meaning; and summary statement 2 of the American College of Gastroenterology in 2017 states that in a population without identifiable risk factors, an ALT or AST above the upper limit of normal is associated with increased liver-related mortality. This section explains why there is an asterisk, not that the asterisk can be ignored.

Why did two laboratories print different "normal ranges" for the same item last time and this time?

Because a reference range is set by the laboratory issuing the report. The measurement of ten Hong Kong chemical pathology laboratories (Hong Kong Med J 2019;25(4):295–304) shows: the upper limits of the AST reference intervals differed by as much as 47% (an absolute difference of 16 U/L), urea by 29% and potassium by 18%; and even on the same analytical platform, the same method and the same reagents, of the three laboratories using the Abbott platform eleven of fourteen analytes had different reference intervals, of the four using Roche eleven likewise, and of the two using Beckman three. The same paper also reports that for ten of those analytes the agreement between laboratories and between platforms was adequate — the divergence is in the ranges each sets, not in the measurement, which is why that paper calls those differences "redundant". The American College of Gastroenterology's guideline of 2017 also states that the range of the upper limit of normal for AST and ALT can differ between laboratories. So comparing results from different years means measuring against the range each report itself sets out.

All the figures are normal — can I relax?

Not one of the authoritative sources cited here says so. The American College of Gastroenterology (2017), summary statement 4: a normal ALT may not exclude significant liver disease. The WHO (2011): a glycated haemoglobin below 6.5% does not exclude diabetes diagnosed using glucose tests. EASL–EASD–EASO (2024): people with MASLD and normal transaminases can still have significant steatohepatitis and progress to advanced fibrosis or cirrhosis. KDIGO (2024): chronic kidney disease can be established by a marker of kidney damage as well as by a reduced glomerular filtration rate, and KDIGO says of itself that this is often overlooked by healthcare staff and students. NICE CG95: acute coronary syndrome cannot be excluded on a normal resting twelve-lead electrocardiogram. With symptoms, see a doctor, and do not use a normal report to cancel a symptom out.

My electrocardiogram report says "Abnormal ECG" — what should I do?

Take it to a doctor; do not interpret it yourself, and do not set it aside because you have no symptoms. The conclusion the machine prints is a first draft — a study reviewing 159 630 electrocardiograms (Eur Heart J Digit Health, January 2026) found that across 2011 to 2023, 31.3% of automated electrocardiogram reports were amended after review by a doctor; the study reports a decline established by a trend test (P < 0.001), with 2023 at 18.4%, still close to a fifth — and the period-by-period figures in its table rose twice along the way rather than falling throughout. Of the automated reports carrying "inferior infarct" and "anterior infarct", 32.0% and 44.6% respectively had to be deleted. NICE CG95 recommendation 1.2.2.7 also requires a resting twelve-lead electrocardiogram to be reviewed, in addition to any automated interpretation, by a healthcare professional qualified to interpret it. This article will not reproduce any chest pain scoring rule written for clinicians, for the reasons given in the electrocardiogram section. If you have chest pain now, do not wait — NICE CG95 recommendation 1.2.1.7 states that suspected acute coronary syndrome with current chest pain should be referred as an emergency admission; and recommendation 1.2.1.9 states that even where the pain has resolved, signs of a complication (its example being pulmonary oedema) call for referral for hospital assessment.

Is an MRI necessarily safer than a CT?

On ionising radiation, the Chinese pages of the Cancer Online Resource Hub state that magnetic resonance involves no radiation and that ultrasound carries no radiation and is harmless, while computed tomography forms its image with a rotating X-ray gantry. But magnetic resonance is not without risk: the Chinese page on the same site states that medical science does not yet know the long-term effects of a magnetic resonance examination on the human body or on a fetus; the general risks it lists itself include noise of up to one hundred and thirty decibels or more during the examination, nerve stimulation producing transient numbness or muscle twitching, and an anxiety reaction from the enclosed space; and with a gadolinium-containing contrast agent about 1% of patients have a mild allergic reaction, the risk of death is 「四十萬分之一」 (one in a little under half a million), a patient in renal failure has a 1% to seven per cent chance of nephrogenic systemic fibrosis, and metal devices in the body may move or heat up. Nothing in the material cited here compares the sensitivity of the two for any particular disease, so this article makes no such comparison; the choice of examination is decided by a doctor on the clinical question.

Can I get my own laboratory report or image disc back? What does it cost, how long does it take, and will it necessarily be in English?

Open the eHealth app first — with an eHealth account, in many cases you can see it without paying. The government added laboratory reports on 9 January 2025 and radiology reports on 3 April 2025 to the eHealth mobile app's "investigation records" function, and they can generally be viewed for yourself 14 days after the report is issued; the eHealth frequently asked questions state that participation in and use of eHealth are voluntary and free of charge. But that route has a precondition: you must have registered for eHealth, hold a valid eHealth account and be aged 16 or over — for a child under 16, a person with parental responsibility manages the account as a "carer". Three boundaries, all in the government's own words: microbiology (culture and antibiotic sensitivity testing) and anatomical pathology records are not displayed in the app at this stage; on radiology the app provides the radiology report PDF, while radiology images cover only examinations on or after 1 January 2025; and private providers' records depend on whether you have given a sharing consent and whether that provider has deposited them. eHealth also states of itself: this is not your complete health record and cannot replace the records kept for you by healthcare providers.

For a complete copy, a paper copy or a disc, or for the two classes of laboratory record the app does not display, the right is under section 18 of the Personal Data (Privacy) Ordinance (Chapter 486). Section 19(1) provides that a data user must comply within 40 days of receiving a data access request, and the same subsection opens with a qualification — 「在符合第(2)款及第20及28(5)條的規定下」: section 19(2) allows the other side to fail, but requires them to notify you in writing with the reasons before the 40 days expire and to supply whatever part they can, completing it as soon as practicable; and if they do not hold your record at all, section 19(1)(b) requires them to tell you so in writing. The Hospital Authority's published charges (effective 1 January 2026): a data access request handling fee of 100 dollars per application (covering 10 pages and the postage), and 300 dollars per disc or film. The Queen Mary Hospital form (version 20260611) also states that enquiring whether the hospital holds your record is free of charge, and that the imaging reproduction charge is 300 dollars per modality per disc. Whether the handling fee is refunded is something this article will not answer for you — the first page and page 7 of that same form say opposite things, and eHealth states in terms that all administrative fees paid are non-refundable, so it has to be put to the receiving party. Note that eHealth's Electronic Health Record Application and Enquiry Centre is a different data user, with a handling fee of 109 dollars plus materials (that page self-noting a fee update date of 17 October 2024).

On language: section 19(3)(c)(iii)(A) of Chapter 486 provides that the copy must, 「除(B)分節另有規定外，須採用該項要求所指明的語文(可為中文或英文)」, and sub-subparagraph (B) provides that where the data is held in another language and the copy is a true copy, the copy must be in a language other than the one you specified. Which is to say that the true copy of a case note written in English ought to come to you in English — and Queen Mary Hospital's Application for Medical Report form (version 20260611), stating that all medical reports and patient information are in English and that the hospital has no translation service, does not conflict with the ordinance. The "must be intelligible" of subparagraph (i) of the same paragraph carries a comparable exception, expressly exempting a true copy that is on its face not intelligible. What remains uncovered by any exception is subparagraph (ii) — the copy must be comprehensible where any coding used by the data user has been sufficiently explained; and whether the clinical abbreviations in a case note count as "coding" is something no case or Privacy Commissioner decision among the material cited here has addressed. For an explanation you can understand, you need a doctor.

#Notes: the official originals

[Note 1] The local survey by Chan Chun-hei and colleagues in the Hong Kong Medical Journal in 2019: the definition of a reference interval, and the two routes by which a laboratory establishes one.

「Reference intervals are commonly defined as limiting values, usually upper and lower limits, between which a prespecified percentage (usually 95%) of results would fall.」

「A laboratory may establish a new set of RIs by conducting an RI study with at least 120 reference individuals from each subgroup stratified by sex, age, and other parameters as appropriate. Conducting an RI study is challenging, as enormous efforts of human and financial resources are needed. … Alternatively, a laboratory may adopt the RIs established by other sources such as manufacturers or the literature and validate them with at least 20 reference individuals' results.」

[Note 2] Australian Prescriber (1 April 2009): the 95% is only a tradition and not a law; and the counter-example of serum insulin.

「By tradition, laboratories quote a reference range including 95% of the reference population.」

「For example, the reference range commonly quoted for serum insulin may only include results within one standard deviation above and one standard deviation below the mean value. This includes 68% of the reference population. In this case, 16% of normal people will have 'abnormal' high insulin and 16% will have 'abnormal' low insulin according to the quoted reference range.」

[Note 3] The Chinese abstract of that 2019 paper as printed on the second page of the PDF (including the 「百份比」 and the full-width percent signs of the original).

「然而，所有化驗室採用的參考區間皆有差別。在不同化驗室之間，AST參考區間上限的百份比差異可達47%（絕對差異16 U/L），尿素為29%（1.8 mmol/L），鉀為18%（0.8 mmol/L）。尿素參考區間下限的百份比差異可達24﹪（0.6 mmol/L），磷酸鹽為22%（0.16 mmol/L），總蛋白質為8%（5g/L）。」 (The above is the original of that paper's Chinese abstract, transcribed as printed, including its 「百份比」 and its full-width percent signs.)

[Note 4] The English body of the same paper: different reference intervals despite the same platform, method and reagents; and the concrete example of potassium.

「The investigated laboratories used different RIs despite employing the same analytical platforms, methods and reagents, for 11 out of the 14 analytes among those using the Abbott platform (labs 1-3), 11 out of 14 of analytes among those using Roche platforms (labs 6-9), and three out of the 14 of analytes among those using the Beckman platforms (labs 4-5).」

「For example, among users of the Abbott platform, the potassium RIs of labs 1 and 2 were 3.6 to 5.2 mmol/L for samples of both sexes, while that of lab 3 was 3.5 to 4.5 mmol/L for male and 3.4 to 4.4 mmol/L for female samples.」

[Note 5] The first sentence of the "results" of that paper's English abstract: adequate agreement between laboratories and platforms on ten analytes, but reference intervals differing throughout.

「Sufficient inter-laboratory and inter-platform agreement regarding the 10 analytes (albumin, alanine aminotransferase, aspartate aminotransferase, chloride, gamma-glutamyl transferase, phosphate, potassium, sodium, total protein, and urea) were demonstrated. However, the RIs were heterogeneous across all laboratories」

[Note 6] That paper's conclusion: a recommendation that Hong Kong adopt common reference intervals for general chemistry analytes.

「We recommend the use of common RIs in Hong Kong for general chemistry analytes to reduce redundant variation across laboratories.」

[Note 7] The limitations that paper states for itself, together with the "despite this" that immediately follows.

「The relatively small number and choice of QAP specimens for retrospective methodological comparisons represent a major limitation of our survey. Artificial materials used in QAP specimens generally gave rise to more variable and method-dependent results due to matrix effects. Despite this, our survey demonstrated that methodological bias would not prevent the use of common RIs for seven general chemistry analytes.」

[Note 8] The Australian Prescriber section "Why normal people often have abnormal results", together with the complete original of that article's Box 1.

「Why normal people often have abnormal results

A multiple biochemical analysis can be performed by one machine and produce 20 results. Assuming these results were all independent of each other (which they are not) and that results from the reference population are normally distributed (which they may not be), only 36% of normal people will have all 20 results in the reference range. There will be 64% with at least one abnormal result (Box 1). However, the more abnormal the result and the more related tests are abnormal, the more likely the abnormality is clinically significant.」

「Box 1 Normal results in normal people If the reference range covers 95% of results for a normal population, the chance of a healthy individual having a certain number of normal tests is: Two out of two tests 90% (0.95 x 0.95 = 0.90) All 20 of 20 tests 36% (0.95^20)」

[Note 9] The same article's explanation of regression to the mean.

「The initial results at the extremes are the result of extreme random variability in one direction or the other. The same amount and direction of variability is unlikely to occur on the second measurement in the same individual. Subsequent measurements will therefore move closer to the middle (or 'regress to the mean').」

[Note 10] The boxed recommendation on page 6 of the World Health Organization's 2011 report, with its GRADE ratings.

「HbA1c can be used as a diagnostic test for diabetes providing that stringent quality assurance tests are in place and assays are standardised to criteria aligned to the international reference values, and there are no conditions present which preclude its accurate measurement. An HbA1c of 6.5% is recommended as the cut point for diagnosing diabetes. A value of less than 6.5% does not exclude diabetes diagnosed using glucose tests. Quality of evidence assessed by GRADE: moderate Strength of recommendation based on GRADE criteria: conditional」

[Note 11] Summary statements 4, 6 and 2 of the "What are Truly Normal Liver Chemistry Tests?" section of the American College of Gastroenterology (ACG) guideline summary of 2017.

「4. A normal ALT level may not exclude significant liver disease.」

「6. AST and ALT ULN ranges can vary between different labs.」

「2. Elevated ALT or AST above the upper limit of normal (ULN) in a population without identifiable risk factors is associated with increased liver-related mortality.」

[Note 12] The EASL–EASD–EASO MASLD guideline of 2024: one passage giving first the relation between raised transaminases and liver-related mortality, and then that normal transaminases can still accompany significant steatohepatitis.

「Elevation of liver enzymes, namely aminotransferases, is associated with increased liver-related mortality.」

「However, individuals with MASLD and normal aminotransferase levels can still have significant steatohepatitis and develop advanced fibrosis or cirrhosis [ 22 ] and the risk of liver-related outcomes, i.e. mortality, hospitalisation, and HCC is increased with worsening of liver fibrosis [ 23 ].」

[Note 13] Page S137 of the KDIGO 2024 chronic kidney disease guideline.

「Note that while the definition of CKD includes many different markers of kidney damage and is not confined to decreased GFR and albumin-to-creatinine ratio (ACR) >30 mg/g [>3 mg/mmol], the classification system is based on the 2 dimensions of GFR and degree of albuminuria (Tables 2 and 3). This nuance is often missed by healthcare providers and students.」

[Note 14] Six relevant recommendations of the UK NICE clinical guideline CG95: 1.2.2.5, 1.2.2.4 and 1.2.2.6, together with the three adjacent referral recommendations 1.2.1.8, 1.2.1.7 and 1.2.1.9.

「1.2.2.5 Do not exclude an ACS when people have a normal resting 12‑lead ECG. [2010]」

「1.2.2.4 Even in the absence of ST‑segment changes, have an increased suspicion of an ACS if there are other changes in the resting 12‑lead ECG, specifically Q waves and T wave changes. … [2010]」

「1.2.2.6 If a diagnosis of ACS is in doubt, consider: taking serial resting 12‑lead ECGs; reviewing previous resting 12‑lead ECGs; recording additional ECG leads. Use clinical judgement to decide how often this should be done. Note that the results may not be conclusive. [2010]」

「1.2.1.8 If an ACS is suspected (see recommendation 1.2.1.3) and there are no reasons for emergency referral, refer people for urgent same-day assessment if: they had chest pain in the last 12 hours, but are now pain free with a normal resting 12‑lead ECG or the last episode of pain was 12 to 72 hours ago. [2010]」

「1.2.1.7 Refer people to hospital as an emergency if an ACS is suspected (see recommendation 1.2.1.3) and: they currently have chest pain or they are currently pain free, but had chest pain in the last 12 hours, and a resting 12‑lead ECG is abnormal or not available. [2010]」

「1.2.1.9 Refer people for assessment in hospital if an ACS is suspected (see recommendation 1.2.1.3) and: the pain has resolved and there are signs of complications such as pulmonary oedema. Use clinical judgement to decide whether referral should be as an emergency or urgent same-day assessment. [2010]」

[Note 15] Bradley SH and colleagues' systematic review (Br J Gen Pract 2019) of the sensitivity of chest X-ray for symptomatic lung cancer.

「Though there is a paucity of evidence, the highest-quality studies suggest that the sensitivity of chest X-ray for symptomatic lung cancer is only 77% to 80%. GPs should consider if further investigation is necessary in high-risk patients who have had a negative chest X-ray.」

[Note 16] The blood tests page of the Hong Kong government's Cancer Online Resource Hub, and one sentence from the eHealth frequently asked questions.

「不過，即使擁有先進的技術，每項血液檢查也有其局限性，例如假陽性或假陰性的結果。」

「病人應尋求醫學意見以解讀有關他／她的健康資料。」

[Note 17] The Consumer Council press release of 14 July 2016: its heading sentence, its core sentence, the expert views on individual examinations, the passage on blood tests, and the passage on tumour markers.

「消委會就健康檢查徵詢衞生署、香港病理學專科學院及香港醫學會的專業意見」

「故無論化驗結果正常與否亦應由醫生解讀，以免錯誤理解相關檢驗結果」

「在沒有臨床判斷和任何徵狀的情況下，很多常見的體檢項目檢測隱疾的成效不高。以｢靜態或運動心電圖｣作為檢測冠狀動脈疾病的檢查，在發病率低和沒有任何徵狀的人士身上，出現假陽性檢查結果的比率高。以｢胸肺X光｣作為篩查肺癌的工具，其敏感性不足，出現假陰性檢查結果的機率較高。」

「至於血液化驗，即使｢全血像/全血球計算｣和｢紅血球沉降率｣的化驗結果超出正常範圍，如驗身者本身沒有任何徵狀，單以這化驗亦不足以判斷該名人士是否患病。而以「血脂測試」用來評估心血管疾病的風險時，還須同時考慮其他因素，包括年齡、血壓水平和是否有吸煙習慣等。」

「消費者必須留意，「腫瘤標記」是血液中一些特殊物質，這些物質或會在患上某類癌症時增加，醫生一般用來輔助診斷及監察癌症康復進度，並非篩檢患上某類癌症的理想工具。而且，目前並沒有任何｢腫瘤標記｣測試具備足夠的敏感性和特異性，可單獨用作癌症篩查。」

[Note 18] StatPearls' Normal and Abnormal Complete Blood Count With Differential (NBK604207, revised 8 June 2024): spurious leucopenia, pseudo-thrombocytopenia, and that chapter's concluding passage.

「A spurious leukopenia may occur due to white blood cell clumping.」

「Platelet counts may appear low due to platelet clumping in the presence of ethylenediaminetetraacetic acid (EDTA) in blood specimen collection tubes, causing pseudo thrombocytopenia. Repeating the platelet count is recommended when a specimen is collected in a tube containing a different anticoagulant, such as citrate or oxalate.」

「Results must be interpreted in the clinical context and according to local laboratory guidelines, as reference intervals are specific to each lab. An examination of a peripheral blood smear is essential in assessing any abnormal CBC result and is recommended.」

[Note 19] Tefferi A and colleagues in Mayo Clinic Proceedings, 2005: variation from a personal baseline can occur entirely within the reference range.

「Likewise, an individual may display a substantial change from his or her baseline (ie, personal normal) without violating the "normal" reference range.」

[Note 20] The British Society of Gastroenterology's iron deficiency anaemia guideline of 2021 (Gut 2021): the definition of anaemia, and a "false-normal" ferritin.

「We recommend that anaemia is defined as a haemoglobin (Hb) concentration below the lower limit of normal for the relevant population and laboratory performing the test」

「Serum ferritin is the single most useful marker of IDA, but other blood tests (eg, transferrin saturation) can be helpful if a false-normal ferritin is suspected」

[Note 21] The American College of Gastroenterology's guideline summary of 2017: the problem with the name "liver function", and the definitions of hepatocellular and cholestatic injury.

「The most common liver chemistries ordered are serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase and bilirubin. These tests should be termed liver chemistries or liver tests.」

「Hepatocellular injury is defined as disproportionate elevation of AST and ALT levels compared with alkaline phosphatase levels. Cholestatic injury is defined as disproportionate elevation of alkaline phosphatase level as compared with AST and ALT levels.」

The two passages quoted beneath the table in the body, verbatim. ACG 2017 summary statement 1:

「A true healthy normal ALT level in prospectively-studied populations without identifiable risk factors for liver disease ranges from 29-33 IU/L for males and 19-25 IU/L for females, and levels above this should be assessed by physicians.」

The same passage of AASLD 2018 (the square brackets being that guideline's own reference markers):

「The upper limits of normal (ULN) for ALT in healthy adults are reported to be 29 to 33 U/L for males and 19 to 25 U/L for females … For purposes of guiding management of CHB, an upper limit of normal for ALT of 35 U/L for males and 25 U/L for females is recommended (Figure 1), though differences in repeat testing of the same sample have been described [ 97 , 98 ]. This might prompt clinicians to repeat testing when a single ALT elevation is near the cutoff for treatment.」

[Note 22] Pages S148 and S147 of the KDIGO 2024 guideline: the meaning of a very low serum creatinine, and the respective limitations of creatinine and cystatin C.

「Very low levels of SCr often represent poor health status, such as frailty or sarcopenia, which limits the production of creatinine.」

「creatinine, which is directly linked to muscle mass, may be misleading at extremes of body habitus, or in specific conditions (spinal cord injuries and sarcopenia), and that cystatin C is impacted by different variables (steroid use, thyroid disease, and cancer).」

[Note 23] Annex one (page 16) of the World Health Organization's 2011 report: its heading, its attribution line, and the complete original of the five classes of factor affecting glycated haemoglobin.

「Some of the factors that influence HbA1c and its measurement*. Adapted from Gallagher et al (24)」 (Reference 24 of that report: Gallagher EJ, Bloomgarden ZT, Le Roith D. Review of hemoglobin A1c in the management of diabetes. J Diabetes 2009; 1(1):9-17.)

「1. Erythropoiesis Increased HbA1c: iron, vitamin B12 deficiency, decreased erythropoiesis. Decreased HbA1c: administration of erythropoietin, iron, vitamin B12, reticulocytosis, chronic liver disease. 2. Altered Haemoglobin Genetic or chemical alterations in haemoglobin: haemoglobinopathies, HbF, methaemoglobin, may increase or decrease HbA1c. 3. Glycation Increased HbA1c: alcoholism, chronic renal failure, decreased intra-erythrocyte pH. Decreased HbA1c: aspirin, vitamin C and E, certain haemoglobinopathies, increased intra-erythrocyte pH. Variable HbA1c: genetic determinants. 4. Erythrocyte destruction Increased HbA1c: increased erythrocyte life span: Splenectomy. Decreased A1c: decreased erythrocyte life span: haemoglobinopathies, splenomegaly, rheumatoid arthritis or drugs such as antiretrovirals, ribavirin and dapsone. 5. Assays Increased HbA1c: hyperbilirubinaemia, carbamylated haemoglobin, alcoholism, large doses of aspirin, chronic opiate use. Variable HbA1c: haemoglobinopathies. Decreased HbA1c: hypertriglyceridaemia. * Some of the above interfering factors are "invisible" in certain of the available assays」 (In substance: some of the factors that influence HbA1c and its measurement, in five classes — erythropoiesis, altered haemoglobin, glycation, erythrocyte destruction and the assay itself. Iron and vitamin B12 deficiency push it up; erythropoietin, iron, B12, reticulocytosis and chronic liver disease pull it down; haemoglobinopathies, fetal haemoglobin and methaemoglobin can move it either way; alcoholism and chronic renal failure push it up; aspirin and vitamins C and E pull it down; splenectomy pushes it up; splenomegaly, rheumatoid arthritis and certain drugs pull it down; hyperbilirubinaemia pushes it up; and hypertriglyceridaemia pulls it down. The asterisked note: some of the above interfering factors are "invisible" in certain of the available assays.)

[Note 24] One sentence from the executive summary on page 3 of the same report, and the two passages on page 8 — that the diagnosis must be confirmed by a repeat test, and the meaning of the band below 6.5%.

「The expert group concluded that there is currently insufficient evidence to make any formal recommendation on the interpretation of HbA1c levels below 6.5%.」

「A report published in 2009 by an International Expert Committee on the role of HbA1c in the diagnosis of diabetes recommended that HbA1c can be used to diagnose diabetes and that the diagnosis can be made if the HbA1c level is ≥6.5%(16). Diagnosis should be confirmed with a repeat HbA1c test, unless clinical symptoms and plasma glucose levels >11.1mmol/l (200 mg/dl) are present in which case further testing is not required.」

「Levels of HbA1c just below 6.5% may indicate the presence of intermediate hyperglycaemia. The precise lower cut-off point for this has yet to be defined, although the ADA has suggested 5.7 – 6.4% as the high risk range (29). While recognizing the continuum of risk that may be captured by the HbA1c assay, the International Expert Committee recommended that persons with a HbA1c level between 6.0 and 6.5% were at particularly high risk and might be considered for diabetes prevention interventions.」

[Note 25] StatPearls' Hypothyroidism (NBK519536, revised 18 February 2024): central hypothyroidism should be diagnosed on free T4.

「Central hypothyroidism may be either of pituitary or of hypothalamic origin. In these cases, the TSH produced can be biologically inactive and affect the levels of bioactive TSH; therefore, the diagnosis of central hypothyroidism should be based on free T4 rather than TSH.」

[Note 26] The European Thyroid Association's 2013 guideline on subclinical hypothyroidism: the definition, and the grading and repeat testing arrangement.

「Subclinical hypothyroidism (SCH) is defined as a state of increased serum thyroid-stimulating hormone (TSH) levels, with circulating thyroxine (T 4 ) and tri-iodothyro-nine (T 3 ) concentrations within the population reference range.」

⚠ In the quotation above, the space in 「T 4」 and the hyphenation in 「tri-iodothyro-nine」 are artefacts of PDF extraction; the printed source reads 「T 4」 with a subscript and 「tri-iodothyronine」 unbroken.

「Subclinical hypothyroidism (SCH) should be considered in two categories according to the elevation in serum thyroid-stimulating hormone (TSH) level: mildly increased TSH levels (4.0–10.0 mU/l) and more severely increased TSH value (>10 mU/l). An initially raised serum TSH, with FT 4 within reference range, should be investigated with a repeat measurement of both serum TSH and FT 4 , along with thyroid peroxidase antibodies, preferably after a 2- to 3-month interval.」

[Note 27] The definitions of the terms in the original words of the StatPearls electrocardiogram chapters, item by item.

• Sinus rhythm — Electrocardiogram (NBK549803, revised 5 June 2023): 「A regular rhythm ECG has regular P waves preceding a QRS complex in a regular rhythm. Also, normal sinus rhythm demonstrates positive P waves in leads I, II, and aVF, suggesting a downward propagation of atrial activation from the SA node.」
• Sinus bradycardia — Sinus Bradycardia (NBK493201, revised 7 August 2023): 「Sinus bradycardia is a cardiac rhythm with appropriate cardiac muscular depolarization initiating from the sinus node generating less than 60 beats per minute (bpm).」
• Sinus tachycardia — Sinus Tachycardia (NBK553128, revised 5 March 2023): 「The normal resting heart rate for adults is between 60 and 100 beats per minute, which varies with fitness level and the presence of comorbidities.」 The chapter also notes: 「While sinus tachycardia is common as a compensatory response to exercise or stress, it is concerning when it occurs at rest.」
• Premature atrial contractions — Premature Atrial Contractions (NBK559204, revised 8 August 2023): 「Premature atrial contractions (PACs) are contractions of the atria that are triggered by the atrial myocardium but have not originated from the sinoatrial node (SA node).」
• Premature ventricular complexes / contractions — Premature Ventricular Complex (NBK547713, revised 16 February 2025): 「Premature ventricular complexes (PVCs), or premature ventricular contractions, are ectopic beats that arise from within the ventricles. Also known as premature ventricular contractions, the American Heart Association (AHA) and Heart Rhythm Society prefer the term "complexes" as electric activity does not always result in cardiac muscle contraction.」 ⚠ "Premature beats" on a report can mean either atrial or ventricular, which have different origins and cannot be treated as the same thing. This article gives only the definitions of those two terms, and says nothing whatever about what either kind means, how often it occurs or whether it needs following up — those are for a doctor.
• Axis deviation — Electrical Right and Left Axis Deviation (NBK470532, revised 8 January 2024): 「Determining the electrical axis can provide insight into underlying disease states and help steer the differential diagnosis towards or away from certain diagnoses. Axis determination is also important to correctly identify certain life-threatening arrhythmias that have implications for definitive management strategy.」
• ST segment — ST Segment (NBK459364, revised 14 August 2023): 「The ST segment on an electrocardiogram (ECG) normally represents an electrically neutral interval between ventricular depolarization (the QRS complex) and repolarization (the T wave). However, variations in ST-segment morphology may indicate a benign finding or clinically significant myocardial injury.」
• T wave — ECG T Wave (NBK538264, revised 7 November 2025): 「Normal T-wave findings are characterized by positive deflections in all leads except aVR and V1; however, their morphology may be altered by various physiologic and pathologic processes」

[Note 28] Chiu IM and colleagues (Eur Heart J Digit Health, January 2026): the 31.3% amendment rate, and the sentence on the downward trend that immediately follows.

「Our analysis revealed that 31.3% of all ECG reports underwent some forms of modification by physicians. … At the term-level, diagnoses such as 'prolonged QT interval' (newly added from 5.6% of original reports) and 'electronic ventricular pacemaker' (newly added from 3.6% of original reports) were frequently added by physicians, while diagnoses like 'inferior infarct' and 'anterior infarct' were frequently deleted from automated ECG reports (32.0% and 44.6% automated reports with these terms required removals).」

「More recent ECGs were processed using updated version of the Marquette™ 12SL ECG analysis programme, showing a substantial decline in physician modification rate (42.2% in 2011–12 to 25.6% in 2021–22 and 18.4% in 2023, P for trend < 0.001) (Table 1).」

[Note 29] NICE CG95 recommendation 1.2.2.7: review by a qualified healthcare professional in addition to automated interpretation.

「1.2.2.7 Obtain a review of resting 12‑lead ECGs by a healthcare professional qualified to interpret them as well as taking into account automated interpretation. [2010]」

[Note 30] The Hong Kong government Cancer Online Resource Hub: the Chinese originals of the X-ray imaging, ultrasound (with its blind spot on air and bone), computed tomography, magnetic resonance (with the sentence that the long-term effects are not yet known) and positron emission tomography pages.

「X光是一種人眼看不見的電磁輻射，它們穿過身體時，病人並不會有感覺。……接受診斷時，X光會穿過身體並擊中放置在另一側的感光螢幕以產生 X 光影像。 X光造影術通常用於檢測相應的身體部分的輪廓、腫塊或病變。」

「超聲波是一種高頻音波，它的頻率遠高於人類可以聽到的範圍。……在正常情況下使用超聲波掃描是不含輻射及無害的，亦不會引起痛楚。」

「超聲波遇上空氣及骨時會被反射，令其不能穿透體內組織。所以檢查時會用水溶軟膏作超聲波的介質，使超聲波可以穿透組織。」

「電腦掃瞄是一個安全的影像檢查方法，利用電腦科技及X光旋轉器，描繪出精確的橫切圖像，以診斷體內的疾病。」

「磁力共振掃描造影（簡稱磁力共振）是一種安全、準確、並無輻射或痛楚的造影檢查。此項檢查是利用強力的磁場，無線電波及精密的電腦科技，清楚地透視人體內部組織」

「醫學界現時仍未確知磁力共振檢查對人體或胎兒所産生之長遠影響。」

「正電子電腦掃描是一種先進的同位素影像科技，用以偵測癌症有否擴散到身體其他部位。」

「以下個別香港常見癌症均可能用到正電子電腦掃描：」 (followed by links to three pages: breast cancer, prostate cancer and oesophageal cancer.)

[Note 31] The magnetic resonance page on the same site: the complete list of things to tell the radiology staff before the examination, transcribed verbatim.

「以下狀況均可能對病人產生危險及干擾檢查所得的影像，如病人有下列任何一種狀況，請於檢查前通知放射科職員： 體內置有心臟起博器或人工心瓣等。 身體內裝有、植入或配戴任何金屬儀器。 現在或以前曾經從事五金行業。 曾經被金屬碎片或子彈所傷。 現在或曾經有紋身或紋眉。 有腎衰竭，腎病或以前做過腎臟手術。 已經或可能懷孕。 患有幽閉恐懼症。」

[Note 32] The computed tomography page on the same site: the complete list of preparations before the examination, transcribed verbatim.

「如曾患哮喘、風疹、濕疹、過敏性鼻炎或對某些食物、藥物或X光造影劑有過敏反應的病歷，請事先知會醫生。 如懷疑或已經懷孕，切記在檢查前通知放射科職員。 糖尿病人請遵照醫生指示調節藥份。 檢查前四小時請勿飲食。 大部份腹部檢查需要在檢查前數小時口服造影劑或水。 住院病人可事先在病房內口服造影劑或水。 如需要靜脈注射造影劑，病人要簽署同意書。」

[Note 33] The complete original of the "risks or complications" section of the computed tomography page on the same site (including its 「二十五萬分之一」 and its typographical slip 「急情腎衰竭」), and the wording at the corresponding place in the English version.

「一般而言，顯影劑是很安全的，但偶有併發症出現。 一般風險包括： 輕微反應：少許發熱、痕癢、作嘔、嘔吐、手痛、打噴嚏、咳嗽及少許呼吸困難。部分病人會較遲有反應，通常會24小時內感到注射顯影劑的靜脈注射部位痛楚、痕癢、紅疹或唾液腺有腫脹。這些症狀屬暫時性，不需或只需少許治療。 中度反應：反應症狀比較嚴重及維持比較長時間。病人會有紅疹、發燒、寒顫、血壓上升或下降或心悸。此時需要特別治療及監察。 嚴重反應：氣促、心率不齊、心絞痛、急情腎衰竭、抽搐及昏迷。如有這些反應，必須立即治療。 特別風險包括︰當注射顯影劑時，有機會發生顯影劑外滲。病人會感到靜脈注射位不適或痛楚。至於病人因對顯影劑敏感而導致死亡，機會約二十五萬分之一。」

「In extremely rare situation, the contrast medium may cause death due to severe allergic reactions.」

(The 「急情腎衰竭」 quoted above is that government page's own typographical slip; the corresponding place in the English version reads "severe kidney failure". This article transcribes the original without correcting it.)

[Note 34] The complete originals of the "general risks or complications" and "important risks or complications" sections of the magnetic resonance page on the same site.

「檢查期間儀器會發出大量而不可避免的噪音，如敲擊聲、敲門聲或鳴叫聲。而這些噪音可高達一百三十分貝或以上。神經線可能會受到刺激而産生短暫的麻痺、肌肉抽搐、發熱或痛楚。 檢查期間可能因密室空間而引致焦慮反應，過度恐懼或驚恐發作。 含釓的磁力共振造影劑是一種非常安全的藥物，産生敏感反應的情況並不常見。如有亦通常非常輕微，約百分之一的病人會有輕微敏感反應如頭痛丶暈眩丶噁心或嘔吐。另外造影劑外滲亦可引致注射位置不適，腫脹或痛楚，而這些症狀將會於一至二天後消失。」

「在罕見的情況下，含釓的磁力共振造影劑可能引致嚴重的反應，如抽搐丶嚴重過敏性休克以致死亡。導致死亡的風險是四十萬分之一。 身體內的不兼容金屬裝置或儀器有可能出現移動，發熱導致燒傷，故障或依附著磁力共振主機。 患腎功能衰竭的病人在接受含釓的磁力共振造影劑後，會有百分之一至七的機會出現系統性腎臟纖維化病變。受影響的患者會出現行動不便，關節攣縮，多個器官功能衰竭。該反應罕有地會引致死亡。」

(The 「丶」 in the quotations above is the character that page itself uses, in several places, in place of the enumeration comma 「、」. This article transcribes it as printed.)

[Note 35] The United States Food and Drug Administration's definition of effective dose in What are the Radiation Risks from CT?, and the Department of Health Radiation Health Unit's explanation of the unit in Understanding Radiation.

「The quantity most relevant for assessing the risk of cancer detriment from a CT procedure is the 'effective dose'. The unit of measurement for effective dose is millisieverts (abbreviated mSv). Effective dose allows for comparison of the risk estimates associated with partial or whole-body radiation exposures.」

「量度輻射劑量的單位是希沃特(Sv)。一毫希沃特(mSv)等於一千分之一希沃特。」

[Note 36] The FDA's two qualifications attached to Table 1 — that the actual dose can be two or three times the estimate, and that estimates can differ by a factor of ten or more.

「Radiation dose from CT procedures varies from patient to patient. The particular radiation dose will depend on the size of the body part examined, the type of procedure, and the type of CT equipment and its operation. Typical values cited for radiation dose should be considered as estimates that cannot be precisely associated with any individual patient, examination, or type of CT system. The actual dose from a procedure could be two or three times larger or smaller than the estimates.」

「Estimates of the effective dose from a diagnostic CT procedure can vary by a factor of 10 or more depending on the type of CT procedure, patient size and the CT system and its operating technique.」

[Note 37] The opening passage of the FDA's "risk estimates" section (the comparison with the lowest doses of the Japanese atomic bomb survivors).

「The effective doses from diagnostic CT procedures are typically estimated to be in the range of 1 to 10 mSv. This range is not much less than the lowest doses of 5 to 20 mSv received by some of the Japanese survivors of the atomic bombs. These survivors, who are estimated to have experienced doses only slightly larger than those encountered in CT, have demonstrated a small but increased radiation-related excess relative risk for cancer mortality.」

[Note 38] The FDA passage translating dose into a risk figure, with its own opening qualification and its closing public health warning.

「The risk of developing cancer as a result of exposure to radiation depends on the part of the body exposed, the individual's age at exposure, and the individual's sex. … A CT examination with an effective dose of 10 millisieverts … may be associated with an increase in the possibility of fatal cancer of approximately 1 chance in 2000. This increase in the possibility of a fatal cancer from radiation can be compared to the natural incidence of fatal cancer in the U.S. population, about 1 chance in 5 (equal to 400 chances in 2000). … the total risk may increase from 400 chances in 2000 to 401 chances in 2000. Nevertheless, this small increase in radiation-associated cancer risk for an individual can become a public health concern if large numbers of people undergo increased numbers of CT screening procedures of uncertain benefit.」

(In plain terms: 10 millisieverts adds about one chance in 2000 of a fatal cancer; the background rate is about one in five, which is 400 chances in 2000; so the total goes from 400 chances in 2000 to four hundred and one chances in 2000.)

[Note 39] The FDA's statement about the uncertainty of low-dose risk estimates.

「There is considerable uncertainty regarding the risk estimates for low levels of radiation exposure as commonly experienced in diagnostic radiology procedures… Some scientists believe that low doses of radiation do not increase the risk of developing cancer at all, but this is a minority view.」

[Note 40] The 80% in the Department of Health Radiation Health Unit's Understanding Radiation, and the proportion table on the Hong Kong Observatory's "Where does radiation come from?".

「天然輻射約佔我們所受輻射劑量的80%，而不同地區的輻射水平會有差異。」

「天然的比例 85.5%／氡 42.5%／γ射線 18%／宇宙射線 14%／體內放射性元素 11% 人為的比例 14.5%／醫療 14%／放射性沉降物 0.2%／其他 0.2%／核工業排放 < 0.1%」 (The English version of that page lists the same figures in the same table. Note that the "medical 14%" is a proportion of the total dose and not 14% of the artificial 14.5% — which is to say that almost all artificial radiation comes from medicine.)

[Note 41] The passage on the thresholds of acute radiation injury in the same Department of Health booklet, together with the qualification that precedes it.

「地球上的生物不斷受到一定水平的天然輻射照射。輻射雖然可能對身體細胞和組織造成損害，不過，除非所受輻射劑量極高，否則對健康的影響可謂微不足道。輻射的影響，視乎輻射的強度、接觸時間的長短及受影響身體細胞的種類而定。 人體若突然受到大量輻照(超過1 000毫希沃特)，會引致急性輻射傷害，並產生短期症狀如作悶、嘔吐、極度疲倦和脫髮等現象。如全身所受輻射劑量達到10 000毫希沃特或以上即使有適當治療，還是很大可能有生命危險。」

[Note 42] NICE CG95 recommendation 1.1.1.5: information about the risks of investigation, including radiation exposure, is to be offered.

「1.1.1.5 Offer information about the risks of diagnostic testing, including any radiation exposure. [2010]」

[Note 43] The two main risks the FDA lists for computed tomography, with incidental findings first.

「The main risks are those associated with test results that demonstrate a benign or incidental finding, leading to unneeded, possibly invasive, follow-up tests that may present additional risks and the increased possibility of cancer induction from x-ray radiation exposure.」

[Note 44] O'Sullivan JW and colleagues' umbrella review (BMJ 2018;361:k2387): the definition of an incidental imaging finding, the prevalence for each examination, the gap on plain films and ultrasound, and the authors' passage on the clinical implications.

「An incidental imaging finding was defined as an imaging abnormality in a healthy, asymptomatic patient or an imaging abnormality in a symptomatic patient, where the abnormality was not apparently related to the patient's symptoms.」

「The percentage was highest in chest computed tomography (45%, 95% confidence interval 36% to 55%), followed by computed tomography colonoscopy (38%, 21% to 57%) and cardiac MRI (34%, 22% to 46%). It was lowest in whole body PET or PET/computed tomography (2%, 1% to 4%) and chest computed tomography for incidental pulmonary embolism (2%, 1% to 4%). Intermediate rates of incidentalomas occurred in MRI of the spine (22%, 19% to 26%) and MRI of the brain (22%, 14% to 31%).」

「We did not identify any eligible systematic reviews or meta-analyses quantifying the prevalence of incidentalomas on any radiography or ultrasonography test.」

「Incidentalomas can cause patient anxiety and can lead to further investigation and treatment, some of which may cause more harm than good. Incidentalomas also have financial consequences; there are costs associated with further patient management, but also the potential change to a patient's insurance status and premiums.」

[Note 45] eHealth frequently asked questions: the four conditions for managing an account through the mobile app.

「想透過醫健通流動應用程式管理醫健通帳戶，你必須： 登記參加醫健通並持有有效的醫健通帳戶； 年滿16歲或以上； 有能力處理自己的事務及管理有關參與醫健通的事宜； 提供手提電話號碼或電郵地址以收取有關醫健通的通知。」

[Note 46] The complete scope listed on the eHealth "Sharable Data" page.

「個人資料／敏感及藥物不良反應／診斷、手術及其他醫療程序、藥物／住院、到診及預約資料／臨床摘要／出生及防疫接種紀錄／化驗及放射報告／其他檢查報告／醫療轉介資料／觀察及生活方式紀錄／醫療證明書」

[Note 47] The whole of the eHealth frequently asked questions' answer on what a patient can see.

「除專業醫護人員可取覽醫健通內的資料外，病人亦可以透過醫健通流動應用程式，取覽其於醫健通內的部份健康紀錄，例如藥物紀錄、預約紀錄、敏感及疫苗紀錄。如欲了解更多有關醫健通流動應用程式的資訊，請瀏覽以下連結（app.ehealth.gov.hk）。根據《個人資料（私隱）條例》（第486章）（《私隱條例》），病人（資料當事人或代表他／她的相關人士）可透過提出查閱資料要求，以取得他們載於醫健通內的個人資料的副本。病人應尋求醫學意見以解讀有關他／她的健康資料。」

[Note 48] Questions 5 and 6 of the eHealth thematic website (app.ehealth.gov.hk) frequently asked questions: the boundaries of laboratory and radiology records.

「5. 為甚麼我在「檢查紀錄」中找不到我部份的化驗紀錄？ 市民可透過醫健通eHealth流動應用程式查閱由私營醫護機構、醫院管理局以及衞生署存入的一般常規檢查診斷化驗紀錄。現階段，微生物（培養及抗生素敏感性測試）及解剖病理紀錄並不會在流動應用程式內顯示。在一般情況下，化驗紀錄會在結果發出後14天顯示在流動應用程式中。如你對相關的化驗紀錄有任何疑問，請向你的醫護機構查詢。」

「6. 為甚麼我在「檢查紀錄」中找不到我部份的放射紀錄？ 市民可透過醫健通流動應用程式查閱放射報告PDF, 以及在2025年1月1日或以後檢查的放射影像。醫健通流動應用程式提供查閱放射紀錄涵蓋的範圍將會逐步擴展。在一般情況下，醫健通流動應用程式放射紀錄會於放射報告發出後14天提供市民查閱。而所提供查閱的放射紀錄是按照醫護機構存放到醫健通的狀況而定，如對放射紀錄有任何疑問，請向您的醫護人員或醫護機構查詢。」

(The half-width comma in 「放射報告PDF, 以及」 in question 6 above is the one that page itself uses; this article transcribes it as printed.)

[Note 49] The Health Bureau's press releases of 9 January 2025 and 3 April 2025, and the nine categories of electronic health record updated in the latter.

「醫健通用戶現可透過醫健通流動應用程式的「檢查紀錄」功能，查閱由醫院管理局和衞生署存入其醫健通戶口的一般常規檢查化驗報告，方便市民掌握及管理自己的健康。用戶一般可在化驗報告發出14日後於應用程式查閱，應用程式的資訊中心亦會發出相關提示。」

「醫健通用戶現可透過醫健通流動應用程式查閱由醫院管理局、衞生署和及私營醫護機構存入其個人醫健通戶口的放射報告，方便市民掌握及管理自己的健康。」

「現時，醫健通用戶可查閱九類電子健康紀錄，分別是個人資料、敏感及藥物不良反應、住院、到診及預約資料、防疫接種紀錄、藥物紀錄、化驗及放射報告、醫療轉介資料、觀察及生活方式紀錄，以及醫療證明書。」

(The 「查閲」 in the heading of the press release of 3 April 2025 and the 「和及」 in its body are as printed in that government press release; this article transcribes them without correction.)

[Note 50] The Chinese text of section 18 of the Personal Data (Privacy) Ordinance (Chapter 486).

「(1) 任何個人或代表一名個人的有關人士可提出內容如下的要求 ——(a) 要求資料使用者告知他該使用者是否持有該名個人屬其資料當事人的個人資料；(b) 如該資料使用者持有該資料，要求該使用者提供一份該資料的複本。」

[Note 51] The Chinese text of section 19(1) of the same ordinance, both limbs in full.

「在符合第(2)款及第20及28(5)條的規定下，資料使用者須在收到由某人提出的查閱資料要求後的40日內，以下列方式，依從該項要求 ——(a) 如該資料使用者持有屬該項要求的標的之個人資料 ——(i) 以書面告知提出要求者該資料使用者持有該資料；及(ii) 提供一份該資料的複本；或(b) 如該資料使用者並無持有屬該項要求的標的之個人資料，以書面告知提出要求者該資料使用者並無持有該資料。」

[Note 52] The Chinese text of section 19(3)(c) of the same ordinance, in full.

「(c) 在切實可行範圍內 ——(i) 須是清楚易明的，但如該複本是一份文件的真實複本，而該份文件 ——(A) 載有該資料；而 (B) 在表面上不是清楚易明的，則屬例外；(ii) 在該資料使用者所使用的編碼已獲充分解說的情況下，須是容易理解的；及(iii)(A) 除(B)分節另有規定外，須採用該項要求所指明的語文(可為中文或英文)；如無如此指明語文，則可採用提出該項要求所採用的語文(可為中文或英文)；(B) 如(但只有在以下情況下) ——(I) 持有該資料所採用的語文，不是在該項要求所指明的語文或(如沒有如此指明語文)提出該項要求所採用的語文(視屬何情況而定)；而 (II)(在符合第20(2)(b)條的規定下)該複本是載有該資料的文件的真實複本，須採用該項要求所指明的語文以外的語文或(如沒有如此指明語文)提出該項要求所採用的語文(視屬何情況而定)以外的語文。」

[Note 53] The Chinese text of section 28(3) of the same ordinance.

「為依從查閱資料要求而徵收的費用不得超乎適度。」

[Note 54] The Chinese original of the Hospital Authority's "Data Access Request" page.

「醫管局必須根據個人資料(私隱)條例的規定，在收到查閱資料要求後的40日內，依從該項要求。如醫管局不能於40日內依從該項查閱資料要求，醫管局必須在40日的期限內以書面通知該查閱資料要求者有關情況及原因，並在醫管局能依從該項查閱資料要求的範圍內，依從該項查閱資料要求。醫管局其後必須在切實可行的範圍內盡快依從或盡快完全依從該項查閱資料要求。因應私家醫生診症需要，病人可授權其私家醫生聯絡醫管局的負責醫生以取得病人的病歷資料。 你需要就查閱資料要求繳付費用，詳情可參閱收費表。」

[Note 55] The Chinese text of section 19(2) of the Personal Data (Privacy) Ordinance, verbatim.

「(2) 凡資料使用者不能在第(1)或(1A)款指明的期間內依從查閱資料要求，他 ——(a) 須在該期間屆滿前 ——(i) 藉書面通知告知提出要求者他不能如此依從該項要求，以及其理由；及(ii) 在他能依從該項要求的範圍(如有的話)內，依從該項要求；及(b) 須在該期間屆滿後，在切實可行的範圍內盡快依從或盡快完全依從(視屬何情況而定)該項要求。」

[Note 56] One sentence on how to make a request, from the same Hospital Authority page.

「查閱資料要求可以書面或查閱資料要求表格提出。」

[Note 57] The fee notes of Queen Mary Hospital's Data Access Request form (HIRO-003, version 20260611), a single bilingual document.

「(i) Charges for Data Enquiry Request 查詢資料要求的收費 This request is only for ascertaining whether this hospital holds the Data Subject's Personal Data or Medical Records. The enquiry is not to be charged. 查詢資料要求旨在確定本院是否持有資料當事人的個人資料或醫療記錄，有關查詢免收費。 … Processing Fee 處理費 : HK$100 per request (inclusive of reproduction charge for not more than 10 pages and postage) 每次100元 (已包含不多於十頁的複製費及郵費) … Reproduction charge of radiological images e.g. X-ray ECG/EEC/C.T. Scan X光片、電腦掃描片、心電圖、腦電圖等複製費 : HK$300 per modality per disc 每種造影 每張光碟 300元 / HK$300 per film 每張底片300元 … 申請經初步處理後，本部門會以書面通知申請人所需繳付的複製費用餘額，才能領取資料複本。若有關查詢資料要求未能跟進，上述處理費將予以退還。」

(The "EEC" quoted above is that form's own typographical slip; the corresponding Chinese column reads 「腦電圖」, an EEG. This article transcribes it as printed.)

[Note 58] The two statements about refunding the handling fee that run in opposite directions in the same form — on the first page and on page 7.

「若有關查詢資料要求未能跟進，上述處理費將予以退還。」 (The English column at the same place: 「The processing fee is subject to refund for unsuccessful location or retrieval of the medical records.」)

「資料當事人及有關人士(如適用者)明瞭初步處理費是不予退還的﹐而有關的餘款亦須付清才可領取要求的資料。」 (The English column at the same place: 「The Data Subject and (where applicable) the Relevant Person understand that the initial processing fee for Data Access Request is non-refundable and the fees for the copy of Personal Data under the Copy Data Request have to be paid prior to the collection of the data.」)

[Note 59] The original of the medical report charges in Queen Mary Hospital's Application for Medical Report form (HIRO-002, version 20260611).

「(a) Medical Report / Claim Form: A minimum of HK$1,100 per specialty, special charges may be charged for reports requiring special professional input subject to a maximum of HKD4,400. 醫療報告/保險賠償表格: 每份由個別診療部門發出的醫療報告或填寫的保險賠償表格基本所需費用為港幣1,100元，視乎該報告是否需要特別專業處理，報告最高收費為港幣4,400元。」

[Note 60] The footnote in the Office of the Privacy Commissioner for Personal Data's Proper Handling of Data Access Requests and Charging of Data Access Request Fees by Data Users (July 2020).

「醫生審核病歷記錄是與依從查閱資料要求直接有關及屬必需的成本。原因是病歷記錄載有敏感的個人資料，而應處以特別保護。在發放病歷記錄予查閱資料要求者前，由醫生審核有關病歷記錄屬必需的。」

[Note 61] The summary of the same document.

「資料使用者可為依從查閱資料要求而徵收不超乎適度的費用，並盡快在40 日內清楚告知查閱資料要求者需收取多少費用。」

[Note 62] The Office of the Privacy Commissioner for Personal Data's Exercising the Right of Data Access under the Ordinance (June 2016): what to do about an excessive fee, and medical records as a typical example.

「如你認為查閱資料要求的收費超乎適度，你可向資料使用者反映你的意見。如你不滿意它們的解釋，你可向公署投訴。」

「查閱資料要求的常見例子包括僱員要求索取他們的工作表現評核報告複本，病人要求索取他們的醫療紀錄複本，以及消費者要求索取他們的服務申請表複本。」

[Note 63] The original of item 4 of Queen Mary Hospital's Application for Medical Report form.

「4. All medical report / patient information will be written in English. The hospital does not provide translation service. 所有醫療報告/病人資料均用英文書寫。本院並無翻譯服務。」

[Note 64] The complete fee notes of the eHealth "Obtaining a Copy of Your Electronic Health Record" page.

「中心會於合理可行的時間內向申請人提供有關報告（於40天內）。……我們會按有關查閱資料的要求，收取處理費用港幣109元，及根據所索取有關報告的格式，收取適度的材料費用： 每支USB記憶棒（容量32GB）：港幣33.9元；或 每張DVD光碟（容量4.7GB）：港幣2.5元；或 紙張打印：每張A4紙港幣0.1元（不論單面或雙面打印）。 … 如查閲資料要求包含放射圖像資料，有關報告只限以USB 記憶棒形式發放。 … 所有已繳付的行政費用將不獲發還。」

(The variant character 「閲」 in 「如查閲資料要求」 above is the one that government page itself uses, while other places in the same passage use 「閱」. This article transcribes it as printed.)

#What this article does not state

The scope of this section is stated once and not repeated for each item: every item below means that there is no support in the sources cited here (see "Sources" at the end), and not that none exists in the world. They are itemised because many of them are common in popular versions.

One: all "normal range" value tables

• Reference ranges for the complete blood count (white cells, red cells, haemoglobin, haematocrit, platelets). No Hong Kong source has published them. Sections one and two above explain why this article does not reproduce overseas values either. Hong Kong government websites (*.gov.hk, 2 August 2026) publish no reference range table for the complete blood count or thyroid function — that is a bounded statement, and the boundary is Hong Kong government websites.
• A single "normal range" for the liver tests (ALT, AST, alkaline phosphatase, total bilirubin, albumin, globulin). The seven published sets for ALT are in the body precisely because they belong to three different kinds of line and span more than threefold. Globulin does not appear at all.
• Reference ranges for kidney function (creatinine, urea, uric acid), and the eGFR staging table as a basis for diagnosis. Uric acid does not appear. eGFR staging is a clinical decision limit rather than a reference range, and on KDIGO 2024 the eGFR alone does not constitute a complete diagnostic criterion for chronic kidney disease.
• Reference ranges for thyroid function (TSH, free T3, free T4). That HKMJ 2019 paper cites a United Kingdom study finding significant differences in thyroid hormone reference intervals between four analytical platforms; this article publishes no single set of values.
• The "impaired fasting glucose" band of 6.1–6.9 mmol/L. No source carries that band.
• Attributing "HbA1c 5.7%–6.4% = prediabetes" to the Hong Kong Department of Health. No Department of Health document defines that band. ⚠ This item must not be run together with the WHO sentence: the WHO (2011) says there is currently insufficient evidence to make any formal recommendation on the interpretation of levels below 6.5% (its own words, "insufficient evidence to make any formal recommendation", quoted in full in the body) — the point being that its restraint is about pronouncing, not that the band is meaningless — and page 8 of the same report itself sets out the American Diabetes Association's 5.7–6.4% and the International Expert Committee's 6.0–6.5% (both quoted in the body). What this article does not state is only the sentence that the band is the Hong Kong Department of Health's definition.
• HDL cholesterol at "men > 1.0, women > 1.3", and triglycerides at "< 1.7" as a target. The ideal level table of the Centre for Health Protection's Cholesterol leaflet (revised 2025) has no triglyceride row, and its ideal HDL level is the same for both sexes, 「高於 1.0（1.6 或以上更為理想）」. The same set of errors is dealt with in our article on high cholesterol and blood lipids.
• The risk stratification framework of "LDL-C < 3.4 (low risk)". That Centre for Health Protection table has no risk stratification at all.

Two: radiation and imaging

• "An abdominal CT is about 10 mSv". The FDA's table prints 8. (The 10 mSv is the example dose in another sentence of the FDA's risk passage, not its abdominal CT row.)
• Any radiation dose value for positron emission tomography (PET). No source provides one; and the Hong Kong government's own PET page (retrieved 3 August 2026) publishes no dose, preparation or risk either.
• "Natural background radiation is about 2–3 mSv a year", and every conversion of the form "equivalent to X days or X years of background radiation". What the Department of Health and the Observatory each publish is the proportion natural radiation makes up (80% and 85.5%); no Hong Kong source has published a per capita annual background dose in millisieverts, so the conversion cannot be made at all.
• Private market charges in Hong Kong for X-ray, ultrasound, CT, MRI and PET. No source.
• Which of the "basic / intermediate / advanced" tiers a given Hospital Authority test falls into. ⚠ What is unpublished is which test is in which tier, not the prices — the prices are published, and this article prints them. The Hospital Authority fee table (effective 1 January 2026) provides for eligible persons: pathology testing services free for basic items, 50 dollars per service for intermediate items and 200 dollars per service for advanced items; and non-urgent radiology services free for basic items, 250 dollars per service for intermediate items and 500 dollars per service for advanced items. (That table has a separate set of charges for non-eligible persons.) But the fee table itself does not list which tier each test falls into, and that page links to no such classification list, so this article will not tell you which cell your examination is in.
• Comparisons of the sensitivity of CT against MRI for fracture, acute haemorrhage or lung lesions. No comparative data.
• The prevalence of incidental findings on plain X-ray and ultrasound. That BMJ 2018 umbrella review states in terms that it found no eligible systematic review quantifying it — an unknown, not a zero.
• Whether Hong Kong has diagnostic reference levels. The public web pages of the Department of Health Radiation Health Unit (2 August 2026) publish none. That is a bounded statement, and the boundary is that unit's public web pages.
• Organ-by-organ malignancy rates for incidental findings. The footnotes to table 4 of that review state that the analyses for some organs include patients known or suspected to have cancer, and that two versions of the value exist for the thyroid; a percentage detached from its footnote would misdescribe its population, so none is reproduced.

Three: the electrocardiogram

• "A slow sinus rhythm is normal in athletes or during sleep". Neither the bradycardia nor the tachycardia chapter cited here contains that statement. It is a reassuring statement about a slow heart rate, and without a source it is not made.
• "Occasional premature beats have no clinical significance", "mild axis deviation is usually meaningless", "this is a normal result", and every other reassuring conclusion. The definitions have sources; the reassuring conclusions do not. And the axis chapter's emphasis runs precisely the other way — axis determination is important in identifying life-threatening arrhythmias. That applies equally to atrial and to ventricular premature beats: the body carries only the definitions of both, and carries — and will carry — no conclusion about how much either matters.
• Formulations of the kind "an Abnormal ECG without symptoms is nothing to worry about". That is an instruction to a reader with an abnormal cardiac result to stop attending to it, running opposite to NICE CG95 recommendation 1.2.2.5, and this article states it in no form whatever.
• Any chest pain or electrocardiogram scoring rule written for clinicians. The reason is in the electrocardiogram section of the body: such rules presuppose a clinical assessment the reader has not had.
• Every "value plus symptom" threshold list for seeking urgent care (specific figures for haemoglobin, platelets, potassium, glucose, ALT or creatinine paired with particular symptoms). Not one such threshold appears in the sources cited here; and every item on such a list is conditioned on a symptom, which would leave a reader with an abnormal value but without that symptom thinking nothing was wrong. This article states none of them. With any acute symptom, seek care or go to accident and emergency directly, without first matching yourself against figures.

Four: others

• "The commonest cause of a mildly raised ALT in Hong Kong is non-alcoholic fatty liver disease", and every ranking of local Hong Kong epidemiology. No Hong Kong source supports it.
• Effect sizes such as "losing 5–10% of body weight normalises ALT" or "HbA1c usually returns below 5.6% within 6 months". No source.
• Action thresholds such as "a persistent ALT above 100 U/L should be seen". No source.
• "Abnormal but asymptomatic — repeat the test in 3–6 months". No source gives that interval. The European Thyroid Association's 「兩至三個月」 applies only to the one specific situation of subclinical hypothyroidism.
• A package list of "what a basic body check includes". No source defines a "typical basic package"; and the Consumer Council's position (2016) is precisely that one should not start from a list.
• Whether a "liver cyst" or a "kidney stone" needs surgery. No source.
• The figure "below 20%" in the Consumer Council's press release (2016). The Chinese and English versions of that press release describe two opposite statistics in the same sentence: the English writes the proportion of positive results that are truly diseased (the positive predictive value), and the Chinese the proportion of diseased people correctly detected (the sensitivity or detection rate). The two mean opposite things and imply opposite follow-up, and the underlying CHOICE magazine report was not obtained, so this article does not cite that figure at all.
• That the Consumer Council press release is a tripartite co-signed or joint position document. It is not. The original says the Council sought the professional views of the three bodies and then grouped them into a report itself; and neither the Chinese nor the English version contains 「聯署」, joint or co-signed. And the Department of Health is a government department rather than a professional body, so the three cannot be called "three professional bodies" collectively either.
• Broad statements of the kind "a normal complete blood count does not exclude cancer". No guideline or review makes that statement; only a single case report does, which is too weak a level of evidence to use.
• "A normal ALT does not exclude significant fibrosis in hepatitis B". ⚠ This item has now been narrowed, but not closed. The general statement does have a source and is in the body: summary statement 4 of the American College of Gastroenterology's 2017 guideline, 「A normal ALT level may not exclude significant liver disease.」 But no hepatitis B guideline among the sources cited here makes the express statement about hepatitis B and about fibrosis — the EASL 2024 sentence is about MASLD. Hepatitis B is an important local cause in Hong Kong, so this remains a substantive gap.
• Whether a patient can view their own laboratory or imaging report within HA Go. No source cited here explains it, so this article makes no statement. ⚠ That item is about HA Go only. The eHealth mobile app side is answered in the body — the government has published that it can be done, generally 14 days after the report is issued.
• Whether Hong Kong has since adopted common reference intervals. That HKMJ 2019 paper is a recommendation; no source shows that it has been implemented.
• The fine level under the Personal Data (Privacy) Ordinance at "level three (currently HK$10,000)". That amount comes from a Privacy Commissioner publication of 2016; the levels of fines are set by Schedule 8 of Chapter 221 and are amended, and this article has not checked the current amount, so it is not stated.
• Which committee defines the scope of eHealth's sharable data. The Chinese and English versions of that page name two different committees, and this article cites neither.
• The administrative appeal case number cited in that Privacy Commissioner guidance footnote. The Chinese and English versions cite different numbers, and this article quotes only its substance.
• Whether the data access request handling fee is refunded where the record cannot be found. The first page and page 7 of the same Queen Mary Hospital form say opposite things, and the Chinese and English versions of that form describe the refund condition differently; while eHealth states in terms that all administrative fees paid are non-refundable. This article reproduces each original without concluding which governs.

#Sources

• Chan TCH, Mak CM, Chen SPL, et al. Survey on common reference intervals for general chemistry analytes in Hong Kong. Hong Kong Med J 2019;25(4):295-304 (DOI 10.12809/hkmj197943): https://www.hkmj.org/abstracts/v25n4/295.htm / PDF https://www.hkmj.org/system/files/hkmj197943.pdf (retrieved 2 August 2026) — the definition of a reference interval and the ways of establishing one; the measurement of ten Hong Kong laboratories, fourteen analytes and four platforms; the first sentence of the "results" of the English abstract: adequate inter-laboratory and inter-platform agreement on ten analytes, but reference intervals differing at every laboratory; AST upper limits differing by as much as 47% (16 U/L), urea 29% and potassium 18%; eleven of fourteen differing on the same platform with the same reagents; the potassium interval example from laboratories 1 to 3; the recommendation that Hong Kong adopt common reference intervals; the survey's self-declared limitations together with the two sentences that follow (matrix effects, and 「Despite this…would not prevent the use of common RIs for seven general chemistry analytes」); table 2 (PDF page 8, journal page 302) with 18 rows of data covering 14 analytes, listing the published common reference intervals of the United Kingdom, Australasia, Japan and the Nordic countries together with a 「Hong Kong (proposed by this study)」 column, of which ALT is 5-40 / 10-42 / 10-70 / <54 in men and 5-35 / 7-23 / 10-45 / <40 in women (the United Kingdom not listing ALT); and the Chinese abstract (the second page of the PDF)
• Phillips P. Pitfalls in interpreting laboratory results. Aust Prescr 2009;32:43-6 (DOI 10.18773/austprescr.2009.022, the page self-dated 1 April 2009): https://australianprescriber.tg.org.au/articles/pitfalls-in-interpreting-laboratory-results.html (retrieved 2 August 2026) — the 95% convention and its "by tradition" character; the 68% serum insulin counter-example; Box 1's 0.95² = 90% and 0.95²⁰ = 36% / 64%; regression to the mean; and change from a personal baseline occurring entirely within the reference range (the creatinine example)
• Consumer Council press release 「驗身諮詢醫生意見 瞭解局限免誤診」 (the page self-dated 2016.07.14): https://www.consumer.org.hk/ws_chi/news/press/477/checkup-packages.html / English version https://www.consumer.org.hk/ws_en/news/press/477/checkup-packages.html (retrieved 3 August 2026) — 「消委會就健康檢查徵詢衞生署、香港病理學專科學院及香港醫學會的專業意見」 / the English 「The Council sought expert advice from the Department of Health (DH), the Hong Kong College of Pathologists (HKCPath), and the Hong Kong Medical Association (HKMA)」, after which the Council grouped them into a report itself; neither version containing 「聯署」, joint or co-signed; 「無論化驗結果正常與否亦應由醫生解讀」; a high false positive rate for resting or exercise electrocardiography in asymptomatic people and insufficient sensitivity for chest X-ray screening for lung cancer; and a complete blood count and erythrocyte sedimentation rate alone being insufficient to determine whether a person has a disease
• World Health Organization. Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus (WHO/NMH/CHP/CPM/11.1, 2011): https://iris.who.int/server/api/core/bitstreams/db9b9d3d-f95e-4797-9d2b-c78dcef0133f/content (retrieved 3 August 2026) — the boxed recommendation on page 6, the 6.5% cut point and 「A value of less than 6.5% does not exclude diabetes diagnosed using glucose tests.」 with its GRADE ratings; the executive summary's statement on page 3 that the evidence is insufficient for the range below 6.5%; the passage on page 8 restating the 2009 International Expert Committee report, that 6.5% or above is diagnostic and that 「Diagnosis should be confirmed with a repeat HbA1c test, unless clinical symptoms and plasma glucose levels >11.1mmol/l (200 mg/dl) are present」; the passage immediately following on page 8, that levels just below 6.5% may indicate intermediate hyperglycaemia, that the American Diabetes Association suggests 5.7–6.4% as the high-risk range, and that the International Expert Committee suggests 6.0–6.5% is particularly high risk; and the complete list of some of the factors influencing HbA1c in annex one on page 16, its asterisked note, and the attribution line above the table 「Adapted from Gallagher et al (24)」 (reference 24 of that report: Gallagher EJ, Bloomgarden ZT, Le Roith D. J Diabetes 2009;1(1):9-17)
• EASL–EASD–EASO. Clinical Practice Guidelines on the Management of Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD). Obes Facts 2024;17(4):374-444 (PMID 38852583; PMCID PMC11299976) (retrieved 2 August 2026) — one passage writing first 「Elevation of liver enzymes, namely aminotransferases, is associated with increased liver-related mortality.」, then setting ALT above 33 U/L in men and above 25 U/L in women as "elevated", and then writing that normal transaminases can still accompany significant steatohepatitis and progression to advanced fibrosis or cirrhosis
• KDIGO CKD Work Group. KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney Int 2024;105(4S):S117-S314 (PMID 38490803; DOI 10.1016/j.kint.2023.10.018): https://kdigo.org/wp-content/uploads/2026/04/KDIGO-2024-CKD-Guideline.pdf (retrieved 2 August 2026) — the diagnostic criteria for chronic kidney disease (a reduced glomerular filtration rate or a marker of kidney damage, persisting at least three months); 「This nuance is often missed by healthcare providers and students」; a very low serum creatinine reflecting frailty or sarcopenia; and creatinine against muscle mass and cystatin C against other variables
• NICE Clinical guideline CG95, Recent-onset chest pain of suspected cardiac origin: assessment and diagnosis (published 24 March 2010, last updated 30 November 2016; read from a timed Wayback Machine snapshot: https://web.archive.org/web/20260119152502/https://www.nice.org.uk/guidance/cg95/chapter/Recommendations , retrieved 2 August 2026) — recommendation 1.2.2.5 (do not exclude acute coronary syndrome on a normal electrocardiogram), 1.2.2.4 (Q wave and T wave changes), 1.2.1.7 (emergency admission for suspected acute coronary syndrome with current chest pain, or now pain free but with chest pain within the past 12 hours and an abnormal or unavailable electrocardiogram), 1.2.1.8 (where there is no reason for emergency referral, a normal electrocardiogram can still lead to same-day emergency assessment; that recommendation directly adjoins 1.2.1.7, with no other recommendation between them), 1.2.1.9 (where the pain has resolved but there are signs of a complication such as pulmonary oedema, refer for hospital assessment, the urgency being a matter of clinical judgment; immediately after 1.2.1.8), 1.2.2.6 (where the diagnosis is in doubt, consider serial electrocardiograms, reviewing old ones and adding leads, with its own qualification 「Note that the results may not be conclusive」; between 1.2.2.5 and 1.2.2.7), 1.2.2.7 (review by a qualified healthcare professional in addition to automated interpretation) and 1.1.1.5 (offer information about the risks of investigation, including radiation exposure); and section 1.3.3 "Making a diagnosis based on clinical assessment", which carries the clinical scoring rule this article expressly does not reproduce
• Bradley SH, Abraham S, Callister ME, et al. Sensitivity of chest X-ray for detecting lung cancer in people presenting with symptoms: a systematic review. Br J Gen Pract 2019;69(689):e827-e835 (PMID 31636130; DOI 10.3399/bjgp19X706853) (retrieved 2 August 2026; abstract only) — a chest X-ray sensitivity of 77%–80% for symptomatic lung cancer, and that review's own statement that almost all the studies were of poor quality
• Cancer Online Resource Hub (Hong Kong government): blood tests https://www.cancer.gov.hk/tc/cancer_diagnosis/laboratory_test/blood_test.html ; magnetic resonance imaging https://www.cancer.gov.hk/tc/cancer_diagnosis/scan_angiography/magnetic_resonance_scan.html ; computed tomography https://www.cancer.gov.hk/tc/cancer_diagnosis/scan_angiography/computer_scan.html ; ultrasound https://www.cancer.gov.hk/tc/cancer_diagnosis/scan_angiography/ultrasound.html ; X-ray imaging https://www.cancer.gov.hk/tc/cancer_diagnosis/scan_angiography/xray_imaging.html ; positron emission tomography https://www.cancer.gov.hk/tc/cancer_diagnosis/scan_angiography/positron_tomography.html (none of the pages carries a revision date; retrieved 2 August 2026, and the positron emission tomography page 3 August 2026) — every blood test having its limitations and false positives and negatives; the principle and uses of each imaging examination; ultrasound being reflected by air and bone; magnetic resonance having no radiation and its long-term effects not being known; the complete list of things to tell radiology staff before a magnetic resonance examination; the complete list of preparations before a computed tomography examination; the graded contrast reactions of computed tomography and the 「二十五萬分之一」 death risk (found only in the Chinese version; the English version https://www.cancer.gov.hk/en/cancer_diagnosis/scan_angiography/computer_scan.html has no figure at the corresponding place); about 1% mild reactions to a gadolinium-containing agent, a death risk of 「四十萬分之一」, and a 1% to seven per cent chance of nephrogenic systemic fibrosis in patients in renal failure; the "general risks or complications" section of the magnetic resonance page listing noise of up to one hundred and thirty decibels or more during the examination, nerve stimulation causing transient numbness, muscle twitching, warmth or pain, an anxiety reaction from the enclosed space, and contrast extravasation; the framing words of the computed tomography contrast reaction list being 「一般風險包括：」 and 「特別風險包括︰」, with neither version claiming the list is exhaustive; and the Hong Kong government PET page (in both languages) containing nothing beyond one introductory sentence and three links to common cancers, and publishing no dose, preparation or risk
• U.S. Food and Drug Administration. What are the Radiation Risks from CT? (the page self-dated 5 December 2017): https://www.fda.gov/radiation-emitting-products/medical-x-ray-imaging/what-are-radiation-risks-ct (retrieved 3 August 2026) — the definition and unit of effective dose; the opening passage of the "risk estimates" section, that the effective doses of diagnostic computed tomography are typically 1 to 10 mSv, a range "not much less than" the lowest doses of 5 to 20 mSv estimated to have been received by some Japanese atomic bomb survivors, who show a small but definite radiation-related excess relative risk of increased cancer mortality; the ten rows of typical effective doses in Table 1 (including abdominal computed tomography at 8 mSv); dose varying from patient to patient and the actual dose being two or three times the estimate; the sentence immediately before Table 1, that estimates can vary by "a factor of 10 or more"; the first sentence of the risk passage, that the risk depends on the part exposed and on age and sex; 10 mSv corresponding to an increase of about one chance in 2000 in fatal cancer, set against the United States population's natural incidence of about one in five, and 400 becoming 401; the last sentence of the same passage, that when large numbers of people undergo computed tomography screening of uncertain benefit that small increase can become a public health problem; the uncertainty of low-dose risk estimates; and incidental findings being listed first of the two main risks of computed tomography. The values in Table 1 are noted as coming from McCollough CH, Bushberg JT, Fletcher JG, Eckel LJ. Mayo Clin Proc. 2015;90(10):1380-92
• Department of Health Radiation Health Unit, Understanding Radiation: https://www.rhd.gov.hk/tc/RadiationHealth/understand_rad.html / English version https://www.rhd.gov.hk/en/RadiationHealth/understand_rad.html (retrieved 2 August 2026) — the relation between the sievert and the millisievert; natural radiation accounting for about 80% of the dose received; 「除非所受輻射劑量極高，否則對健康的影響可謂微不足道」; acute radiation injury above 1 000 millisieverts and danger to life at 10 000 millisieverts; and exposure to radiation increasing the chance of cancer. ⚠ The material cited in that booklet dates from the late 1990s, and this article takes only its qualitative statements
• Hong Kong Observatory, "Where does radiation come from?": https://www.hko.gov.hk/tc/radiation/monitoring/where_radiation_comes_from.html (English version https://www.hko.gov.hk/en/radiation/monitoring/where_radiation_comes_from.html ; retrieved 3 August 2026) — the table of the proportions of the radiation dose we absorb daily: natural 85.5% (radon 42.5%, gamma rays 18%, cosmic rays 14%, radioactive elements within the body 11%) and artificial 14.5% (medical 14%, radioactive fallout 0.2%, others 0.2%, nuclear industry discharges < 0.1%); the Chinese and English versions agreeing on the figures
• O'Sullivan JW, Muntinga T, Grigg S, Ioannidis JPA. Prevalence and outcomes of incidental imaging findings: umbrella review. BMJ 2018;361:k2387 (PMID 29914908; PMCID PMC6283350; DOI 10.1136/bmj.k2387) (retrieved 2 August 2026) — the definition of an incidental imaging finding; 20 systematic reviews and 240 primary studies; the prevalence for each examination with confidence intervals in table 3; 15 of 20 meta-analyses with an I² above 50%; finding no review quantifying it for plain X-ray or ultrasound; and incidental findings leading to anxiety, further investigation and treatment, cost and insurance consequences
• El Brihi J, Pathak S. Normal and Abnormal Complete Blood Count With Differential. StatPearls (NBK604207, the source self-dated as revised 8 June 2024): https://www.ncbi.nlm.nih.gov/books/NBK604207/ (retrieved 2 August 2026) — pseudoleukopenia from white cell clumping; EDTA-induced pseudothrombocytopenia and the recommendation to repeat with a different anticoagulant; results having to be interpreted against the clinical context and local laboratory guidance, reference intervals belonging to each laboratory; and abnormal results having to be assessed with a peripheral blood smear
• Tefferi A, Hanson CA, Inwards DJ. How to Interpret and Pursue an Abnormal Complete Blood Cell Count in Adults. Mayo Clin Proc 2005;80(7):923-936 (PMID 16007898; PMCID PMC7127472) (retrieved 2 August 2026) — a person can change considerably relative to their own baseline without leaving the reference range
• Snook J, Bhala N, Beales ILP, et al. British Society of Gastroenterology guidelines for the management of iron deficiency anaemia in adults. Gut 2021;70(11):2030-2051 (PMID 34497146; DOI 10.1136/gutjnl-2021-325210) (retrieved 2 August 2026) — anaemia defined as below the lower limit of normal for the relevant population and the laboratory doing the test; and "falsely normal" ferritin
• Kwo PY, Cohen SM, Lim JK. ACG Clinical Guideline: Evaluation of Abnormal Liver Chemistries. Am J Gastroenterol 2017;112(1):18-35 (PMID 27995906; DOI 10.1038/ajg.2016.517); the summary statements cited in this article are taken from the 16-page Guideline Summary published on the American College of Gastroenterology's own website (carrying the summary statements and recommendations only, without references and without the discussion text of the journal article): https://acgcdn.gi.org/wp-content/uploads/2018/04/ACG-Abnormal-Liver-Chemistries-Guideline-Summary.pdf (retrieved 3 August 2026) — that they should be called liver chemistries rather than liver function; the definitions of hepatocellular and cholestatic injury; and the summary statements of the section "What are Truly Normal Liver Chemistry Tests?": statement 1, a "true healthy normal" ALT of 29-33 IU/L in men and 19-25 IU/L in women (with its in-sentence limiting words 「in prospectively-studied populations without identifiable risk factors for liver disease」 and its closing 「levels above this should be assessed by physicians」); statement 2, 「Elevated ALT or AST above the upper limit of normal (ULN) in a population without identifiable risk factors is associated with increased liver-related mortality.」; statement 4, 「A normal ALT level may not exclude significant liver disease.」; and statement 6, 「AST and ALT ULN ranges can vary between different labs.」
• Terrault NA, Lok ASF, McMahon BJ, et al. Update on Prevention, Diagnosis, and Treatment of Chronic Hepatitis B: AASLD 2018 Hepatitis B Guidance. Hepatology 2018 (PMID 29405329; PMCID PMC5975958) (retrieved 3 August 2026) — two sets of figures with different purposes in the same passage: an upper limit of normal ALT in healthy adults of 「29 to 33 U/L for males and 19 to 25 U/L for females」 (the same as ACG 2017), and, 「For purposes of guiding management of CHB」, upper limits of 35 U/L in men and 25 U/L in women, together with the qualification immediately following that sentence, 「though differences in repeat testing of the same sample have been described … This might prompt clinicians to repeat testing when a single ALT elevation is near the cutoff for treatment.」
• Pearce SHS, Brabant G, Duntas LH, et al. 2013 ETA Guideline: Management of Subclinical Hypothyroidism. Eur Thyroid J 2013;2:215-228 (DOI 10.1159/000356507): https://www.eurothyroid.com/files/download/ETA-Guideline-Management-of-Subclinical-Hypothyroidism.pdf (retrieved 2 August 2026) — the definition of subclinical hypothyroidism; the two categories of mildly raised 4.0–10.0 mU/l and severely raised >10 mU/l; and repeating TSH and FT₄ together with thyroid peroxidase antibodies, preferably two to three months apart
• Patil N, Rehman A, Anastasopoulou C, et al. Hypothyroidism. StatPearls (NBK519536, the source self-dated as revised 18 February 2024) (retrieved 2 August 2026) — central hypothyroidism should be diagnosed on free T4 rather than TSH, and the TSH produced may be biologically inactive
• The StatPearls chapters for the electrocardiogram terms (all retrieved 2 August 2026): Sattar Y, Chhabra L. Electrocardiogram (NBK549803, revised 5 June 2023); Hafeez Y, Grossman SA. Sinus Bradycardia (NBK493201, revised 7 August 2023); Henning A, Krawiec C. Sinus Tachycardia (NBK553128, revised 5 March 2023); Heaton J, Yandrapalli S. Premature Atrial Contractions (NBK559204, revised 8 August 2023); Shams P, Chhabra L. Electrical Right and Left Axis Deviation (NBK470532, revised 8 January 2024); Kashou AH, Basit H, Malik A. ST Segment (NBK459364, revised 14 August 2023); McMahon K, Ahmed I. ECG T Wave (NBK538264, revised 7 November 2025); Sattar Y, Hashmi MF. Premature Ventricular Complex (NBK547713, revised 16 February 2025) — the definitions of each term
• Chiu IM, et al. Factors associated with physician modifications to automated ECG interpretations. Eur Heart J Digit Health 2026 (PMID 41574034; PMCID PMC12821064; DOI 10.1093/ehjdh/ztaf119) (retrieved 3 August 2026) — 159 630 electrocardiograms (the original using a space as the thousands separator); 31.3% of automated reports modified by a physician across the whole 2011–2023 period, the modification rate falling from 42.2% in 2011–12 to 25.6% in 2021–22 and 18.4% in 2023 (test for trend P < 0.001); the period-by-period values in table 1 being 42.23, 30.86, 47.57, 39.38, 21.72, 25.59 and 18.43, rising twice in between, the paper reporting only a tested trend and claiming no period-by-period monotonic fall; outlier physicians with very high modification rates excluded before inclusion (the example given being those modifying more than 95% of reports), a step that biases the 31.3% downwards; "inferior infarct" deleted in 32.0% and "anterior infarct" in 44.6%; a single centre (Cedars-Sinai Medical Center) and a GE Marquette™ 12SL; and the study's own statement that the 159 630 are not a census of the 769 088, including a random sample of 10 000 each from eight very-high-volume physicians, which the authors list as a selection bias limitation
• Personal Data (Privacy) Ordinance (Chapter 486 of the Laws of Hong Kong), sections 18, 19, 20 and 28, in both the Chinese and English texts (the Department of Justice legislation database bulk data sets hkel_c_leg_cap_301_cap_600_{zh-Hant,en}.zip, consolidated to 1 October 2022; retrieved 3 August 2026; both language texts read against each other, with no substantive divergence on the provisions cited here) — section 18, the right of access to data; section 19(1), the 40-day limit together with its opening 「Subject to subsection (2) and sections 20 and 28(5)」 and limb (b) (where the data is not held, the requestor must be so informed in writing); section 19(2), written notice where the limit cannot be met, partial compliance, and full compliance as soon as practicable afterwards; the whole of section 19(3)(c): (i) intelligible form with its true copy exception, (ii) codes to be sufficiently explained (this limb having no true copy exception), and (iii)(A) opening 「Subject to sub-subparagraph (B)」 with sub-subparagraph (B) (where the data is held in another language and the copy is a true copy, a language other than the specified language is to be used); section 20, the circumstances of refusal; and section 28(3), that the fee shall not be excessive
• Hospital Authority "Data Access Request" page (the footer self-dated the first of January 2026): https://www.ha.org.hk/visitor/ha_view_content.asp?Content_ID=201869&Lang=CHIB5&Dimension=100&Ver=HTML (English version Lang=ENG; retrieved 3 August 2026) — that the Authority must comply within 40 days; the same passage also stating that where it cannot comply within 40 days it must give written notice of the circumstances and reasons within the same 40-day limit and first comply so far as it can, complying as soon as practicable or in full afterwards; and that where a private doctor's consultation requires it, a patient may authorise their private doctor to contact the responsible Authority doctor to obtain the patient's medical records; that the request may be made in writing or on a form; and that individual hospitals may have forms of their own
• Hospital Authority fee table (the page stating that the relevant charges took effect on 1 January 2026): https://www.ha.org.hk/visitor/ha_view_content.asp?Content_ID=10045&Lang=CHIB5&Dimension=100&Ver=HTML (English version Lang=ENG; retrieved 3 August 2026) — section (F), charges for administrative items at public hospitals, the whole section divided into six groups of service across fourteen rows (data access request processing fee 100 dollars per request, 1.5 dollars per page after 10 pages, 300 dollars per disc or film, 1,100 dollars per certificate or medical report, 2,300 dollars per autopsy report, 1,100 dollars per physical examination certificate, 1,400 dollars per medical assessment, 300 dollars per record reissued, 65 dollars per prescription for drug delivery, 130 dollars per item as the administrative fee for self-financed drugs, mortuary service free for the first 28 days, 200 dollars a day from day 29 and 550 dollars a day from day 36); and the three-tier basic / intermediate / advanced structure for pathology testing services and non-urgent radiology services and its charges (for eligible persons: pathology free / 50 dollars / 200 dollars; non-urgent radiology free / 250 dollars / 500 dollars), that table not listing which tier an individual test belongs to
• Queen Mary Hospital Data Access Request form (HIRO-003, version 20260611): https://www.ha.org.hk/haho/ho/cs/DataRequest_QMH_en.pdf (one bilingual document; retrieved 3 August 2026) — no charge for an enquiry about data; a processing fee of 100 dollars per request (covering not more than ten pages, and postage); and radiological images at 300 dollars per disc and 300 dollars per film for each modality. ⚠ That form makes two statements running in opposite directions about refund of the processing fee: on the first page, 「若有關查詢資料要求未能跟進，上述處理費將予以退還。」 / 「The processing fee is subject to refund for unsuccessful location or retrieval of the medical records.」, while the declaration signed by the applicant on page 7 reads 「初步處理費是不予退還的」 / 「the initial processing fee for Data Access Request is non-refundable」; and the refund condition described on the first page is itself different in Chinese and in English
• Queen Mary Hospital Application for Medical Report form (HIRO-002, version 20260611): https://www.ha.org.hk/haho/ho/cs/MedicalReportRequest_QMH_en.pdf (one bilingual document; retrieved 2 August 2026) — a basic 1,100 dollars per specialty for a medical report, up to a maximum of 4,400 dollars; and 「所有醫療報告/病人資料均用英文書寫。本院並無翻譯服務。」 (all medical reports and patient information are written in English; the hospital has no translation service)
• eHealth frequently asked questions: https://www.ehealth.gov.hk/tc/faq/index.html (English version https://www.ehealth.gov.hk/en/faq/index.html ; the page carries no revision date; retrieved 3 August 2026) — that a patient can access part of their health record through the mobile app (the examples given being medication, appointment, allergy and immunisation records); the next sentence of the same answer pointing to the eHealth thematic website: 「如欲了解更多有關醫健通流動應用程式的資訊，請瀏覽以下連結（app.ehealth.gov.hk）。」; the right to make a data access request; 「病人應尋求醫學意見以解讀有關他／她的健康資料。」 (patients should seek medical advice to interpret their health data); the conditions for logging in to the mobile app, 「想透過醫健通流動應用程式管理醫健通帳戶，你必須：登記參加醫健通並持有有效的醫健通帳戶；年滿16歲或以上；有能力處理自己的事務及管理有關參與醫健通的事宜；提供手提電話號碼或電郵地址以收取有關醫健通的通知。」; that a child under 16 is managed on their behalf by a person with parental responsibility in the capacity of a 「照顧者」; and on fees, 「醫護機構及病人，參加及使用醫健通，均屬自願和免費的。」
• eHealth thematic website frequently asked questions: https://app.ehealth.gov.hk/app/ehealth-faq?lang=tc (the page carries no revision date; retrieved 3 August 2026) — item 3 of the "functions" section, the whole answer from its first sentence: 「當你登入你的醫健通帳戶後，便可以取覽你部份的電子健康紀錄，包括藥物紀錄、預約紀錄、敏感、疫苗紀錄、檢查紀錄及牙齒狀況。請留意你醫健通帳戶內的電子健康紀錄並不是你完整的健康紀錄，因此並不可以取代醫護機構為你備存的紀錄。如你對你的醫療紀錄有任何疑問，請向你的醫護機構查詢。」; item 5 on laboratory records: general routine investigation and diagnostic laboratory records deposited by private healthcare providers, the Hospital Authority and the Department of Health can be viewed, microbiology (culture and antibiotic sensitivity testing) and anatomical pathology records are not displayed at this stage, and records generally appear 14 days after the result is issued; item 6 on radiology records: radiology report PDFs can be viewed, and radiological images for examinations on or after 1 January 2025, generally available 14 days after the radiology report is issued, that item also stating 「而所提供查閱的放射紀錄是按照醫護機構存放到醫健通的狀況而定」 (the radiology records made available for viewing depend on what the healthcare provider has deposited into eHealth), the original drawing no public/private distinction
• Health Bureau press release 〈醫健通流動應用程式增強查閱醫護機構化驗報告功能〉 (9 January 2025): https://www.info.gov.hk/gia/general/202501/09/P2025010900165.htm (retrieved 3 August 2026) — that eHealth users can view general routine investigation laboratory reports deposited by the Hospital Authority and the Department of Health through the app's 「檢查紀錄」 function, generally 14 days after the laboratory report is issued
• Health Bureau press release 〈醫健通流動應用程式增添查閲放射報告功能〉 (3 April 2025): https://www.info.gov.hk/gia/general/202504/03/P2025040200721.htm (retrieved 3 August 2026) — that eHealth users can view radiology reports deposited by the Hospital Authority, the Department of Health and private healthcare providers through the app; and that the nine categories of electronic health record currently viewable include 「化驗及放射報告」 (the headline's 「查閲」 and the body's 「和及」 being the press release's own wording)
• eHealth "Sharable Records": https://www.ehealth.gov.hk/tc/whats-ehealth/sharable-records/index.html (English version at the same path under /en/ ; retrieved 2 August 2026) — the complete list of the scope of sharable data, including 「化驗及放射報告」
• eHealth "Get a copy of your electronic health record" (the page self-dating its data access request fees "last updated 17 October 2024"): https://www.ehealth.gov.hk/tc/you-and-your-family/support/get-a-copy-of-eHR.html (English version at the same path under /en/ ; retrieved 3 August 2026) — provision within 40 days; a processing fee of 109 dollars plus materials; anything containing radiological images issued only on a USB memory stick; and the last sentence of the fee note, 「所有已繳付的行政費用將不獲發還。」 (all administrative fees paid will not be refunded)
• Office of the Privacy Commissioner for Personal Data, Proper Handling of Data Access Request and Charging of Data Access Request Fee by Data Users (the document self-dated July 2020): https://www.pcpd.org.hk/tc_chi/resources_centre/publications/files/dar2020_c.pdf (English version https://www.pcpd.org.hk/english/resources_centre/publications/files/dar2020_e.pdf ; retrieved 2 August 2026) — that the fee must not be excessive and must be notified within 40 days; and that a doctor's vetting of the medical record is a directly related and necessary cost. ⚠ The Chinese and English versions of that footnote cite different administrative appeal case numbers, and this article takes only its substance
• Office of the Privacy Commissioner for Personal Data, Exercising the Right of Data Access under the Personal Data (Privacy) Ordinance (the document self-dated June 2016): https://www.pcpd.org.hk/tc_chi/resources_centre/publications/files/DAR_QnA_c.pdf (English version https://www.pcpd.org.hk/english/resources_centre/publications/files/DAR_QnA_e.pdf ; retrieved 2 August 2026) — a patient obtaining a copy of their medical record given as a common example; and that a person who considers a fee excessive may raise it with the data user and, if not satisfied, complain to the Office
• Centre for Health Protection Cholesterol leaflet (the document self-dated "revised 2025"): https://www.chp.gov.hk/files/her/exnnutp036_tc.pdf (English version https://www.chp.gov.hk/files/her/exnnutp036_en.pdf ; retrieved 2 August 2026) — the "ideal level" for total cholesterol, low-density lipoprotein cholesterol and high-density lipoprotein cholesterol; that table having no triglyceride row; and the ideal high-density lipoprotein level being the same for men and women

#Further reading

• On this platform: High cholesterol and blood lipids
• On this platform: High blood pressure
• On this platform: Reading an echocardiogram report
• On this platform: Acid reflux (GERD)