Changing body mass index: The need for a more measured approach?

Measure what can be measured and make measurable what cannot be measured

Galileo (1564–1642)

The term BMI, which denotes body mass index, seems ubiquitous both in clinical practice and day-to-day vernacular. It is based on a ratio of weight to height, and as clinicians, we are increasingly using this index to gauge the physical health of our patients (see Figure 1). But is it a reliable and accurate measure of body habitus? There are several immediate problems. First, it does not account for differential density of tissues – for example, muscle and bone are generally heavier than other tissues and simply measuring the overall weight of an individual provides little information regarding tissue composition. Second, the BMI does not accommodate gender and age differences, which can be substantial given the natural variance in the distribution of fat within the body in men and women of all ages. Third, even within tissues, there are important differences. For instance, fat within the body can be visceral or subcutaneous, with important implications for overall health – and again the BMI is unable to capture such detail. So why is the BMI used at all? To consider this, we need to review its origins.

Figure 1.

Different measures of body habitus. In addition to simple measures of body habitus based on height, weight, hip and waist circumference, there are more sophisticated measures on the horizon that are likely to be relevant to health outcomes and soon be used more broadly in clinical practice.

Lambert Adolphe Jacque Quetelet (1796–1874), an aptly named Belgian polymath who contributed to fields as diverse as astronomy, mathematics and sociology, but was perhaps best recognised as the patriarch of statisticians, devised the BMI for the purpose of defining the ‘average man’. Interestingly, this was possibly one of the very first direct applications of mathematics to human characteristics. Hence, initially, when first devised in 1832, the BMI was referred to as the Quetelet Index; it was almost a century and a half later, in 1972, that the American scientist Ancel Keys coined the term BMI and introduced it into our lexicon.

A normal BMI is considered to lie between 18.5 and 25.0, overweight between 25.0–30.0 and obesity beyond 30.0. The latter is further subdivided into Class ¹ I obesity a BMI of 30.0–34.9; Class II obesity a BMI of 35–39.9 and Class III denotes obesity that extends beyond 40.0. Of note, the higher the BMI, the greater the risk of type 2 diabetes, cancer, sleep apnoea and cardiovascular illnesses. Interestingly, prior to 1998, Americans in the United States had a BMI cut off that was 2 BMI points above the World Health Organization guidelines for normal and overweight categories – but this has since been brought into line.

Critics of the BMI suggest that simply squaring height (m²) does not accommodate sufficiently for the natural variations in habitus observed at the extremes of height. For example, in short people, the BMI creates a disproportionately large denominator, whereas in tall people it is relatively modest. Consequently, tall people are prone to have their weight overestimated as compared to short people who tend to have it underestimated. This occurs simply because people are three-dimensional, not two, and healthy bodies grow at different rates – ultimately achieving diverse shapes and sizes. To address this, Nick Trefethen (2013), a University of Oxford Professor of mathematics, who questioned the usefulness of the BMI formula, described it as a ‘bizarre measure’, and proposed an alternative: the ‘new BMI’ (see Figure 1)

Despite having devised a new measure – purported to be more accurate – Trefethen maintained that any calculation that assigns one number to a person is doomed to be imperfect, because the shape and composition of humans is fundamentally too complex to be described by a single figure. Nevertheless, his new calculation of human shape and size was thought to better approximate reality than the traditional BMI.

In addition to the technicalities of the BMI, recent studies have cast doubt over the assumption that a higher BMI automatically denotes a health risk and suggest that especially from middle age onwards, a slightly higher BMI may actually protect against death from all causes of mortality and in particular from death due to illnesses normally associated with an overweight body habitus, such as stroke, heart failure and diabetes. This is the ‘obesity–mortality paradox’ that describes an inverse relationship between BMI and mortality. Several studies have added evidence to this somewhat counterintuitive finding. For instance, Flegel et al. (2013), who studied the association of BMI and mortality in 97 studies to create a combined sample of 2.88 million people with 270,000 deaths, found that obesity grades II and III were associated with higher all-cause mortality, but simply being overweight or having grade I obesity was not. In fact, the latter two classes were associated with significantly reduced mortality.

So, how can the survival advantage of being overweight or obese be explained? Some have posited that while excessive visceral fat predisposes to the metabolic syndrome, excessive peripheral fat is not only metabolically inert but also provides a safe harbour for toxic lipids, thereby improving metabolic and cardiovascular health. This may be so because increased subcutaneous fat has been found to be associated with lower glucose and lipid levels independent of abdominal fat (Porter et al., 2009). Subcutaneous fat might also provide crucial energy reserves during illness and thereby decrease mortality. However, loss of this subcutaneous fat from key areas of the body may increase the risk of diabetes and dyslipidaemia.

It is estimated that 24% of adults in the United States with normal BMI have an unhealthy metabolic profile, even in the absence of current illness. This ‘metabolically unhealthy normal BMI’ – ‘Thin on the outside, fat on the inside’ (TOFI) relates to excessive visceral adipose tissue which accumulates around internal organs. This is more of a concern for Asians who tend to have increased visceral fat while maintaining a normal BMI and are therefore more vulnerable to developing diabetes. Visceral adipocytes produce inflammatory mediators like tumour necrosis factor-α, interleukins and other cytokines, but secrete less leptin than subcutaneous adipose tissue. Again, these crucial differences between different types of adipose tissue are not captured by the BMI.

In addition to fat, other tissues also play an important role in healthy metabolism. For instance, skeletal muscle is a primary repository for glucose and, as such, muscle loss due to inactivity or ageing can cause impairment of insulin sensitivity and affect cardiovascular health. But because muscle weighs more than fat, a decrease in weight due to muscle loss can create a false sense of security if relying solely on BMI measurement.

To address these shortcomings of the BMI, a combination of the waist-to-height ratio (WHtR) with BMI calculations to generate a ‘matrix’ that can be used to categorise health risk has been proposed (Ashwell and Gibson, 2016). The clinical recommendation is to keep waist circumference below half of the height measurement. This is thought to be a better predictor than BMI alone of the risk of type 2 diabetes and cardiovascular diseases, as borne out in a study of 1453 adults from the UK National Diet and Nutrition Survey (NDNS), in which 35% of the adult group who were judged to be at ‘no increased risk’ according to BMI measurements were found to have a WHtR > 0.5, and, if reliant solely on BMI, might not have been alerted to take action. Waist measurement captures the distribution of fat around the abdomen and therefore can be a useful indicator of the amount of visceral fat. Naturally, this can only be considered a screening tool, but as such would be remarkably cost-effective, as even a piece of string can be used to compare, albeit crudely, the relationship between waist and height.

A further measure recommended for use in conjunction with the BMI is the waist-to-hip ratio (WHR). This targets adipose tissue around the waist, buttocks and abdomen. It is measured while standing up straight and having exhaled fully. A tape measure is used to check the distance around the largest part of the waist, defined generally as the mid-point between iliac crest and the lower margin of the ribs. Then, the distance around the largest part of the hips is measured and WHR is calculated by dividing waist circumference by hip circumference. A normal measure for women is <0.85 and for men <0.90, and unlike BMI, some guidelines for WHR have been adjusted for different ethnicities. For both men and women, a measure of >1.0 is associated with an increased risk for illnesses linked to being overweight. However, the hip measurement is particularly prone to error as the optimal place for this measure is open to interpretation and muscle gain in this area can distort the WHR.

Given these difficulties in making external measurements, there is clearly a need for tools that can accurately assess body composition and other biomarkers that can be used to predict the risk of metabolic disorders and mortality related to obesity. In this regard, several methods for measuring body composition already exist. However, currently, the use of sophisticated measurement techniques like dual-energy X-ray absorptiometry (DEXA) scans or magnetic resonance imaging is largely confined to research or specialist medical facilities, but ultrasonography may soon be the tool of choice, as it is a more cost-effective and less-invasive modality that has shown to reliably delineate visceral and subcutaneous fat thicknesses with acceptable precision (Stolk et al., 2001).

Clearly, there is a burgeoning need for advances in the measurement of obesity and related factors. In the future, determining the optimal weight for an individual, by taking into account multiple factors such as age, sex, fitness, pre-existing diseases, body composition, adipose hormones, metabolic parameters and biomarkers will provide new insights into the causal role of obesity in health and disease. Looking beyond BMI and recognising the potential utility for inexpensive and non-invasive imaging techniques might help identify people at early risk of such diseases and determine the effects of clinical intervention more clearly. However, in the meantime, while noting their limitations, the BMI and other simple measures should be used widely in order to improve physical and psychiatric health of all patients.