I want to tell you about a client I worked with who had been in hospital following multiple strokes. While they were there, the medical team ran a blood glucose test. The result came back elevated — significantly so. The word "diabetic" was used. A medication pathway was discussed.
I asked one question: what had the client eaten or drunk in the two hours before the blood draw?
It turned out a nurse had brought them a cup of tea — with sugar added — about twenty minutes before the sample was taken.
The result was meaningless. And yet it very nearly became a diagnosis.
As it happens, that client did have pre-diabetes. It was a significant factor that needed addressing — lifestyle, diet, targeted testing, the full picture. But it needed to be found properly. Not via a glucose reading taken twenty minutes after a sugared drink in a hospital ward. And the near-miss in that situation is not a one-off. It is, in various forms, a routine feature of NHS blood testing — not because the people running the tests are negligent, but because the system they're working within wasn't designed to catch these errors. And because nobody ever explains any of this to the patient.
Pre-analytical error — the phase nobody talks about
In laboratory medicine there's a term for everything that happens to a sample before it enters the analyser: the pre-analytical phase. It covers the conditions of the patient before the draw, the draw itself, how the sample is handled, transported, and stored. Research consistently shows that the pre-analytical phase accounts for the majority of errors in laboratory testing — estimates range from 46% to over 70% of total lab errors. The analytical phase — the actual chemistry — is by comparison extremely reliable.
In other words, the machine is rarely the problem. What happens before and around the machine is where most of the inaccuracy lives.
For clinical testing to be useful — genuinely useful, not just administrative box-ticking — it needs to meet the basic standards of good science: reproducibility, accuracy, validity. A test that produces a different result depending on whether you've eaten that morning, which arm was used, what time of day the sample was taken, or whether you walked briskly to the appointment is not a reliable test. You can't draw clinical conclusions from an unreliable result. You certainly can't miss a diagnosis because of one — or make one because of one.
The fasting question
Fasting matters for a significant number of the markers routinely included in a blood chemistry panel. Not all of them — haemoglobin, thyroid hormones, inflammatory markers, and most organ function tests are broadly unaffected by food intake. But the metabolic and lipid markers — the ones most commonly used to assess cardiovascular and metabolic risk — tell a fundamentally different story fasted versus fed.
The practical challenge in NHS practice is real and I don't dismiss it. Morning appointment slots are limited. Asking patients to fast adds friction. A significant proportion of people simply won't or can't comply. When you're running thousands of tests a week across a busy practice, chasing fasted samples for everyone is genuinely difficult. I understand that.
But where possible — where the appointment can be arranged in the morning, where the patient can be told clearly and simply what fasting means (water only, nothing else, 10–12 hours before the draw) — good practice is to encourage it. And critically, the patient should be told whether or not their sample was fasted, so the clinician interpreting it knows what they're looking at. That information is often absent from the result sheet entirely.
A client came to me after being told, four years earlier, that their blood glucose was slightly elevated at a routine check. They were advised to watch their diet and come back in six months. Six months became a year. A year became four years. They came back because they'd been told their HbA1c was now elevated and they "might be pre-diabetic."
I looked at the original result. It was a non-fasted sample, taken at 2pm following lunch. There was no indication on the result sheet that fasting status had been recorded or considered. The mildly elevated reading may have been physiologically normal for a post-lunch draw — or it may have been a genuine early signal that was never properly followed up. Four years on, we'll never know. What we do know is that the window for easy, lifestyle-based intervention had narrowed considerably.
Blood pressure — the same logic applies
Blood pressure is not a blood test, but the same principle applies: the conditions of measurement matter enormously, and they're inconsistently applied.
The evidence supports three separate readings, taken at least one minute apart, on the same arm, with a correctly sized cuff, after five minutes of seated rest, in a warm room, before any physical activity. Blood pressure varies with the time of day, emotional state, hydration, electrolyte balance, kidney function, ambient temperature, and whether you walked briskly from the car park. White coat hypertension — elevated readings in a clinical setting that normalise at home — is real, documented, and clinically significant, because treating white coat hypertension with antihypertensive medication is likely to produce harm.
A single blood pressure reading, taken by a distracted receptionist with a medium cuff on a large arm, in a cold waiting room, after a 20-minute walk from the bus stop, is not a reliable measurement. Yet it is regularly used to initiate treatment.
Blood pressure is also worth understanding as what it is: an output measure, a downstream signal of how the cardiovascular and renal systems are functioning, not a thing in itself. High blood pressure is a clue to investigate, not automatically a condition to medicate. The question is why it's high — and that question requires consistent, reproducible measurement to even begin to answer.
A different kind of testing failure
I want to be careful here, because I'm not in the business of NHS-bashing. The NHS does extraordinary things under extraordinary pressure, and acute care — genuine emergencies, surgery, infection, trauma — is something I have enormous respect for. There is a time and a place for every tool in medicine, including medications that are sometimes portrayed as villains. Statins, PPIs, antihypertensives — they all have legitimate uses in the right clinical context.
What I'm describing is something narrower: a structural gap between how routine diagnostic testing is collected and how it needs to be collected to be genuinely informative. The system was not designed with the pre-analytical phase as a priority. It was designed for throughput, accessibility, and cost-efficiency — all legitimate goals. But the downstream consequence is that a significant proportion of routine blood results are less reliable than anyone is saying out loud.
A client I saw had been assessed for DVT twice in three days — missed on the first visit, diagnosed on the second with clots in both the lower limb and the lung. He was started on anticoagulants. Three months later, nobody had followed up with D-dimer to assess whether anticoagulation was still needed. Blood results from the same period suggested possible pre-diabetes — but the sample wasn't fasted, the client wasn't advised to fast, and the result was not discussed with him. There were also markers suggesting possible fatty liver.
No single individual in this chain was negligent. Each acted within the norms of the system they were working in. But taken together, the picture shows what happens when testing is treated as administrative process rather than clinical investigation — when the question of what the result actually means is nobody's specific job.
That gap is precisely where functional medicine operates.
What this means for you
If you're having a blood test — NHS or private — and the markers include glucose, insulin, triglycerides, or lipids, ask whether you should be fasted. The answer should be yes, and the standard is water only for 10–12 hours before the draw. No coffee, no tea, no juice, nothing that your body needs to metabolise. Ideally book a morning slot and take your test before you eat anything.
If you're having blood pressure measured, sit quietly for five minutes first. Breathe. Mention if you've just walked somewhere. Ask for the reading to be taken twice. Know which arm has been used so future readings can be consistent.
And if you have results that have been dismissed as normal — results that don't feel right given how you actually feel — consider whether the conditions of collection may have contributed to the picture. Then consider whether the results were interpreted against the right standards in the first place. That's a different problem entirely, and it's what the next article in this series is about.
This article is part of a four-part series on why blood testing fails. The full clinical evidence base — covering blood chemistry functional ranges, metabolic health, HPA axis, and more — with 34+ PubMed referenced papers:
Read the TDG Evidence Base →Have results you don't fully understand?
Send me your blood test results and I'll produce a written clinical report using functional optimal ranges — identifying what's actually going on across nine body systems. Delivered within seven days.
Blood Chemistry Health Audit — £295 → Take the free gut symptom quiz