The numbers on your blood test aren't discoveries. They're decisions. That applies to mine too.
In November 2017, a committee convened by the American College of Cardiology and the American Heart Association moved the line for high blood pressure from 140/90 to 130/80. Overnight, the share of American adults with "hypertension" rose from 31.9% to 45.6%. Nobody's arteries changed. The definition did.
That story, and others like it, is doing the rounds again as part of a wider argument: that medicine's diagnostic thresholds are set by committees tied to the companies that sell the treatments. Much of that argument is right, and I'll go through where. But some versions of it stand on shoogly pegs of their own. And the argument points straight back at people like me, who sell testing and use "optimal ranges." So I'm going to follow it all the way.
The numbers on a lab report aren't all the same kind of thing.
A reference range is a statistical description. A lab tests a large group of people it assumes are healthy and reports the middle 95% of results. By definition, 1 in 20 healthy people falls outside it. A reference range tells you where you sit compared with other people. It doesn't tell you whether you're well.
A diagnostic threshold or treatment cut-off is a decision made by a committee: above this number you have a condition, or you should be offered a drug. That's a judgement about risk, benefit and cost, and it's where money can get in.
A functional or "optimal" range is usually narrower than the reference range and is what practitioners like me use. It's also a decision, made by someone. More on that below, because it's the part my field tends to skip.
Most risk markers (blood pressure, glucose, cholesterol, bone density) are continuous. Risk rises smoothly with the number. There's no cliff edge at 130 or 140, so wherever the line is drawn is a choice. The honest questions are who drew it, on what evidence, and who benefits.
Blood pressure. The 2017 US guideline leaned heavily on the SPRINT trial, which measured blood pressure with an automated device, often with the patient alone in the room. That reads lower than a typical clinic measurement, and the trial was stopped early for benefit, which tends to inflate effect sizes. But there's a detail most critics leave out: the share of American adults recommended medication barely moved, from 34.3% to 36.2%. Most of the newly labelled were advised to change their lifestyle, not to take a tablet. The UK never followed anyway. NICE still uses 140/90 in clinic, confirmed by home or ambulatory readings.
Body weight. In 1998 the US National Institutes of Health lowered the "overweight" threshold from a BMI of 27.8 for men and 27.3 for women to 25 for everyone. About 29 million Americans who had been classed as normal weight no longer were. The stated reason was alignment with the World Health Organization.
Blood sugar. In 2003 the American Diabetes Association lowered the start of "impaired fasting glucose" from 6.1 to 5.6 mmol/L (110 to 100 mg/dL), creating tens of millions of new "prediabetics." To be fair to them, their stated reason was to make the fasting test pick up similar numbers of people to the glucose tolerance test, and to predict diabetes equally well. The World Health Organization looked at the same evidence and kept 6.1, and so does the UK. So a fasting glucose of 5.8 is prediabetes in Ohio and normal in Edinburgh.
Kidneys. In 2002 a kidney working group defined chronic kidney disease stage 3 as an eGFR (a calculated estimate of kidney filtering) below 60, with no other finding needed and no adjustment for age. Kidney filtration declines with age in nearly everyone, so a large share of healthy older people acquire a disease label. Nephrologists have argued for age-adjusted thresholds ever since. This is the same point I made in Your Scan Is Probably Normal For Your Age: ageing isn't a diagnosis.
Bone. In 1994 a WHO study group defined osteoporosis by DEXA scan and created a new category, "osteopenia," for bone density between normal and osteoporotic. The T-score compares you with a healthy 30-year-old, not with people your own age, so most older women fall short. As NPR's Alix Spiegel reported in 2009, drug-company marketing then helped turn a research category into a diagnosis handed to millions of women.
Cholesterol. US cholesterol thresholds were lowered in stages from 1988 onwards. In 2004 it emerged that eight of the nine authors of that year's update had not disclosed financial ties to the makers of cholesterol-lowering drugs. The UK has its own version. In 2014 NICE lowered the point at which statins are offered in England from a 20% ten-year risk to 10%, making an extra 4.5 million people eligible. In 2023 it said people below 10% shouldn't automatically be ruled out either.
Scotland is a useful comparison. The Scottish guideline, SIGN 149, kept a 20% threshold on Scotland's own risk calculator, ASSIGN, for a decade after England moved to 10%. When ASSIGN was recalibrated in 2024, its high-risk threshold became 10%, and the Scottish guideline is being updated to match. Same evidence, different committees, different lines, and a decade in which a borderline patient in Edinburgh and one in Newcastle were treated differently.
Thyroid. In the early 2000s some bodies proposed narrowing the upper limit of normal for TSH from around 5.0 to 3.0 mIU/L. At the Mayo Clinic, researchers looked at 75,882 patients without thyroid disease and found the change would quadruple the number classed as having raised TSH, from 5% to 20%. The TRUST trial (NEJM, 2017) then gave levothyroxine to adults over 65 with TSH between 4.6 and 20 and found no improvement in symptoms or quality of life compared with placebo. In fairness to the other side: people with TSH between 3.0 and 5.0 are more likely to carry thyroid antibodies and to develop thyroid disease later. That's an argument for watching them, not for treating them.
Prostate. The PSA cut-off of 4.0 ng/mL that triggers a biopsy was never grounded in solid evidence, and Richard Ablin, who discovered PSA, publicly called mass screening a public health disaster. The large European screening trial did find a modest reduction in prostate cancer deaths, but at the cost of substantial overdiagnosis and overtreatment. The UK has never introduced population PSA screening.
The pattern is real. Lines tend to move down, the eligible population grows, and the benefit per person shrinks, because each newly captured person is at lower risk than the last. Committee members often have financial ties to companies whose products will be used. None of that proves any individual threshold is wrong. It does mean every threshold deserves the question: on what evidence?
Vitamin D. The critics say the Endocrine Society's 2011 threshold of 30 ng/mL (75 nmol/L) created a fake deficiency epidemic and "remains where it was placed." The first half is a fair point. The same year, the US Institute of Medicine looked at the same evidence and said 20 ng/mL (50 nmol/L) was enough for bone health in nearly everyone. But the second half is wrong. In 2024 the Endocrine Society replaced that guideline. It no longer endorses any specific level as defining sufficiency or deficiency, and it advises against routine vitamin D testing in healthy people. That's a correction, and a fair critic reports it. The UK's advisory committee, SACN, uses 25 nmol/L as its cut-off for deficiency risk.
The critics also say supplementation has been shown to do nothing. The big US trial, VITAL, found no reduction in cancer or cardiovascular events overall, which is a fair summary. But the same trial found 22% fewer autoimmune diagnoses in the vitamin D group. You don't get to cite a trial for the findings you like and leave out the one you don't. The rule cuts both ways, though: in absolute terms that was 0.95% versus 1.19% of people over five years, and it only just reached statistical significance. If I criticise relative-risk marketing for drugs, I have to apply the same arithmetic to supplements.
Bone drugs. Bisphosphonates do slow bone breakdown rather than build new bone, and long-term use is linked to rare atypical thigh fractures and, even more rarely, jaw problems. But "the drug produced the disease it claimed to prevent" isn't what the evidence shows. In women with genuine osteoporosis, especially those who've already had a fracture, trials show clear reductions in spinal and hip fractures. The honest criticism is about treating osteopenia, where the benefit is small, not about the drugs themselves. And the same WHO group behind the T-score later developed FRAX, which estimates actual fracture risk from age, prior fractures and other factors rather than from the scan alone. The field partly corrected itself, and a fair account says so.
Lower isn't always a conspiracy. Risk is continuous. A blood pressure of 135/85 carries more risk than 115/75, and in higher-risk people SPRINT did show fewer events and deaths. The real question isn't whether lower numbers carry less risk. It's whether labelling and medicating a low-risk person at the margin does more good than harm. Often it doesn't, but that claim needs evidence case by case, not a slogan.
Check their references too. If you read essays making this case, check the sources they cite, the same way you would check a guideline. If the whole argument is "check the evidence," the argument has to survive being checked.
If thresholds set by conflicted committees deserve suspicion, so do the ranges used in my field, including mine.
Take my own ferritin target. I work to 50–100 µg/L, for men and women. Below 30 is commonly used to confirm iron deficiency. The lower end of my range has trial evidence behind it: in a randomised trial of non-anaemic women with unexplained fatigue, iron improved fatigue only when ferritin was 50 or below (Verdon, BMJ 2003). The upper end is my choice, not a validated optimum. There's no evidence of benefit above it, and observational data link higher ferritin to type 2 diabetes risk. So one end of my range stands on a trial and the other stands on judgement, and you're entitled to know which is which.
The same scrutiny applies to other popular functional targets. A TSH of 1.0–2.0 is simply the middle of the healthy population, not a target derived from outcomes; in one large Dutch cohort, people with higher TSH within the normal range actually lived longer. And the "optimal" vitamin D band of 125–175 nmol/L that appeared in my own material had no authoritative source. Mainstream guidance starts flagging potential harm from around 125–150 nmol/L, so I've corrected it.
Many functional ranges come from practitioner consensus, from narrowing the reference range towards its middle, or from a respected teacher's experience passed down through courses. Some have reasonable support. Many have none. Experience matters, but it isn't outcome data. I sell testing, and I've recommended supplements. I have exactly the kind of conflict I'd criticise in a guideline panel, and you should weigh what I say with that in mind.
A functional range doesn't become valid because the mainstream one was set by people with drug-company money. Two things can be true at once: a threshold may have been moved for commercial reasons, and the alternative may also lack evidence. "We don't have the trials" is sometimes a fair thing to say, because the trials that would settle it often aren't funded when there's nothing to patent. But it's still an absence of evidence, and it should be stated as one.
Corruption is corruption and nonsense is nonsense, whichever side of the argument it's on.
A number is one piece of evidence, not the verdict. The TDG Five-Test Programme looks at hormones, gut function, organic acids, food sensitivity and blood chemistry together, and builds the picture around you — with every range labelled for what it is.
See the TDG Programme →