When the U.S. wrestler and actor Dwayne “The Rock” Johnson was in his prime, his Body Mass Index (BMI) was around 33 kg/m², technically ‘obese’ by World Health Organization (WHO) standards. Yet no one watching him could call him unhealthy. This paradox is one glimpse of how numbers alone can’t be used to define good human health.
For many years, the WHO has used universal BMI cut-offs: over 25 kg/m² for ‘overweight’ and 30 kg/m² for ‘obesity’. But research in Asian populations has revealed that even at lower BMI levels, people developed diabetes and heart disease more often. Recognising these ethnic differences in fat distribution, the WHO revised the ‘normal’ upper limit for many Asian countries to 23 kg/m². Thus, ‘normal’ health is not fixed or global but a shifting statistical boundary shaped by population data and biological diversity.
The reference range
Every lab report for continuous variables includes a reference range, which is often mistaken to be the normal range. Laboratories use the term ‘reference’ because the range is statistical, not absolute. This means it’s derived from the test results of a large number of individuals. When these values are plotted on a graph, they form a bell-shaped curve, with most results clustering near the average and fewer at the extremes. The reference interval is the central 95% of this distribution, meaning 95 out of 100 healthy people fall within it, while 5% lie outside it. This is why falling outside the reference range doesn’t always indicate disease — nor does falling within guarantee good health. It’s only a guide, rather than a definitive boundary of normalcy.
Diagnosing disease would be simple if health and illness were cleanly separated, the way India and Sri Lanka are separated by the Park Strait. But biology resembles the Tamil Nadu-Kerala border: fluid and gradual, without a clear point where Tamil-speaking people end and Malayalam-speaking people begin. In the border villages, both languages mingle and one shades into the other. Similarly, no single value separates health from disease with precision.
Consider diabetes: when the fasting blood glucose levels of thousands of people are plotted on a graph, the healthy and the diabetic curves merge smoothly. The diagnostic cut-off of 126 mg/dl wasn’t a mathematical discovery but a practical compromise; there’s no fixed line beyond which a person suddenly becomes diabetic. Clinicians draw a line only for practical purposes, and assume that beyond that limit, the risk of complications such as diabetic retinopathy rise sharply.
Put another way, all such medical boundaries balance the two concepts of sensitivity and specificity. A lower threshold would detect all true cases but falsely label many healthy ones, increasing the number of false positives; vice versa for a higher threshold. No test in medicine is 100% sensitive and 100% specific at the same time. Since no test is perfectly accurate, the chosen cut-off reflects medical judgment shaped by a disease’s severity, prevalence, and social consequences. For serious yet treatable conditions, clinicians prefer caution and accept more false positives; for mild or stigmatising ones, they raise the bar to avoid overdiagnosis. Every cut-off in medicine is thus a negotiation and, for that reason, they must be regularly revisited.
Mathematics of normalcy
Not all biological measures vary equally. Sodium and potassium can vary within very tight limits because even small deviations from this range can disrupt muscle or nerve function. On the other hand, cholesterol and liver enzymes can vary across a wider range of values: their smaller changes reflect only incidental changes, such as the person’s water intake on that day. Hormones are even more dynamic, fluctuating with time of day, age, sex, stress, and nutrition. Interpreting the corresponding numbers on a lab report thus demands more than just numeracy. To do so, a health worker must understand when the sample was collected, how it was tested, and what the patient’s clinical picture reveals (or has already revealed).
Every reference range is also determined using a particular mathematical process, and understanding this process can allow health workers to make more informed decisions. Among other things, the distribution of the values of a medical variable is assumed to follow the “68-95-99.7” rule. For instance, if health workers collect blood samples from a number of people and find that the average value is 70 and the standard deviation is 10, the rule says that about 68% of samples score 60-80, about 95% score 50-90, and about 99.7% score 40-100. In other words, the rule helps to visualise how ‘spread out’ the possible values are and how rare it is for a value to fall far from the mean.
The catch here is that not all biological data are symmetrical, that is, fall with equal probability on either side of the mean. For instance, most hormone-related readings are low while a few are extremely high. In such cases, labs select values from the third or the 97th percentiles. For this, they need at least 120 health samples, otherwise their tests won’t be reliable.
Local data matters
When we call something “normal,” we rarely ask, “normal for whom?”.
Laboratory reference ranges are not universal truths but reflect the populations they study. Some parameters, like height, vary widely across gender and ethnicity. A Scandinavian and a South Indian may differ greatly yet both be healthy. People’s nutritional and metabolic needs change with age, gender, race, and even occupation. ‘Normal’ haemoglobin or lipid related values depend on genetics, diet, and the environment. Men have higher red blood cell counts; Africans have lower white cell counts; and Himalayan populations have more haemoglobin. On the other hand, physiological constants such as heart rate or kidney filtration vary little worldwide.
As a result, medical interpretation in India demands that we recognise these differences rather than borrow benchmarks wholesale from the West. Unfortunately, for many common laboratory parameters, including haemoglobin, no large, authoritative Indian reference studies exist. As a result, we risk both over diagnosing and underdiagnosing disease.
Clinical correlates
The reference ranges are not natural: they are statistical constructs that people put together by testing others in particular circumstances, with particular assumptions. At the bottom of every lab report is a line written keeping this fact in mind: “Please correlate with clinical findings.” To correlate clinically means to bring a number back into the world of people, that is, to listen to a patient’s story, examine them, and interpret the number in their full context.
For example, a high BMI may signal overweight or, on the positive side, increased muscle mass, which is often the case with athletes. Similarly a borderline thyroid reading may reflect an illness or it could be a normal physiological variation. Ultimately, health workers must go in with a measure of humility that focuses as much on the term “range” as on “reference”.
Dr. C. Aravinda is an academic and public health physician. The views expressed here are personal.
Published – November 19, 2025 08:30 am IST
