the sexes and in terms of socioeconomic differences. Many Latin American and Middle Eastern countries are at this stage.
Stage 3. In this stage, a swing occurs with those of lower socioeconomic status now having a higher obesity prevalence, but the more affluent women and children do not show any further secular increase. These features are evident in Europe.
Stage 4. This stage is where obesity prevalence actually declines but is a phenomenon we have yet to observe.
In within‐country analyses, higher rates of obesity are traditionally associated with urban environments, but Ezzati and colleagues have highlighted that more recently there has also been a marked increase in rural obesity [24]. This implies that the drivers of obesity were originally most evident in urban areas, but as the world has developed, the factors promoting obesity have penetrated the rural communities and/or the rural environment has lost some of the factors which limited the development of obesity.
Abdominal obesity
So far, we have only been considering analyses of obesity epidemiology in terms of BMI, but as previously mentioned, Asian physicians have long been concerned by the onset of obesity’s comorbidities at much lower BMIs than in Western Europe or North America. The long‐standing clinical observation that abdominal obesity was particularly hazardous had led to Vague’s identifying decades ago that there was a particular risk if a patient had an android phenotype as distinct from a gynecoid pattern of fat distribution with a small waist/hip (W/H) ratio [25]. The likelihood that abdominal obesity was associated with additional risk was widely accepted, but the definition of what constituted abdominal obesity was uncertain except in terms of a high W/H ratio (>1). The Scottish Royal Colleges of Physicians were preparing a simple guide for general practitioners in 1993 on how to tackle obesity and sought to have the assessment and management summarized on a single page chart which could be placed on a GP’s consulting room table. This necessitated a numerical definition of what normal waist measurements were and what the cut‐offs should be for specifying abdominal obesity. Based on the data from a population survey of adults in the Netherlands, Mike Lean and Jaap Seidel decided to simply take the waist measurements of 94 and 102 cm in men and 80 and 88 cm in women since these values corresponded epidemiologically to BMIs of 25 and 30 in the Dutch population [26]. They also noted that values exceeding these waist circumferences were associated with higher cardiovascular risks. These values were therefore taken as the cut‐off points for abdominal obesity in the Scottish Intercollegiate Guidelines Network (SIGN) guidelines [27] and were then promptly incorporated by the National Institute of Health (NIH) group into their US guidelines for tackling obesity [28] as well as being used in the draft IOTF report for WHO, and then incorporated as the cut‐off points in the report of the first WHO Expert Technical Committee to deal with obesity [5].
Figure 1.1 Obesity prevalences in different regions of the world. Data for males are in the left block of data and females in the right‐hand series. Within each data set, the prevalences for 1980 are on the left and for 2008 on the right.
(Source: Reproduced from Finucane et al. [16].)
The INTERHEART international study later showed that waist size and W/H ratios were better indices of the risk of CHD than BMI [29]. W/H values showed marginally better statistical discrimination probably because higher hip values seem to be protective, perhaps relating to the body’s ability to safely store fat subcutaneously – on the hips. The importance of abdominal obesity as a predictor of morbidity has been shown many times, but the waist/height ratio expressed in metric units with a simple ratio cut‐off of 0.5, rather than hip circumference, seems a better and clinically more practical predictor of disease risk factors, e.g. dyslipidemia, increased blood glucose levels, or higher blood pressures [30] especially in children [31].
Different regional societal burdens of obesity with abdominal obesity
McKeigue et al. described in 1991 the propensity to abdominal obesity and thicker truncal skinfolds as being greater in South Asians than in British adults, and for each increment in waist/hip ratios, there was a greater increase in glucose intolerance, plasma insulin levels, diabetes, hypertension, and plasma triglyceride levels in the South Asians than in the British Caucasians [32]. These findings mirrored the concerns expressed by clinical researchers such as Misra et al. [33] from India, who highlighted the problem of abdominal obesity and its associated dyslipidemia at low BMIs in Indian slum dwellers. This was also noted by the subsequent WHO Singapore meeting [8] and then led the IOTF to assess whether these propensities to diabetes and hypertension were evident on a population basis by comparing Australasian data with large data sets derived from a series of studies across Asia. Iranian data were included in the reference data set from Australia and New Zealand as they were considered for practical purposes Caucasian. This addition may have been a mistake (see below), but their inclusion did not affect the overall conclusion, based on the analysis of 21 population groups with about 263,000 individuals, that those with abdominal obesity had a greater propensity to diabetes but not to hypertension, and that the Asian community was particularly prone to abdominal obesity and its hazardous consequences [30].
Genetic susceptibility could account for this Asian propensity to abdominal obesity and indeed to excess diabetes and lipid disorders at each increment of waist enlargement, or it could be attributed to other factors. A genetic basis is supported by the relatively new analyses of human evolution that have shown the different patterns of genetics as the human race evolved out of East Africa and then left Africa to evolve through the Middle East into Europe, Asia, and then across the Siberian/Alaskan link (now the Behring straits) into North America and down into Central and South America [34]. The most significant changes are those that involve the much more unstable mitochondrial DNA with its faster rate of mutation than nuclear DNA. There are very clear patterns of mitochondrial change, designated by different haplotypes, with a subset leaving Africa and subsequently evolving down multiple haplotype pathways and with some interbreeding with early hominids, the Neanderthals and Denisovans. There are mutations of mitochondrial DNA that are associated with diabetes, but as yet, there are no extensive population studies of genetics that also test for the prevalence of glucose intolerance and diabetes in populations across the world, as well as associated haplotype analyses.
This seemingly special Asian propensity to abdominal obesity and diabetes was then shown to be a more general feature when analyses of the 2006 national survey of Mexican adults found the same features – both a greater propensity to abdominal obesity and a greater prevalence of diabetes and hypertension at each increment of abdominal expansion [35] as shown in Figure 1.2. Furthermore, US studies showed that when ethnic differences were considered, Japanese Americans, as well as Hispanic Americans, were both more likely to have abdominal obesity and greater rates of diabetes at each increment of abdominal expansion than American Whites [36]. African Americans have higher BMIs than Whites or Hispanics, but their diabetes rates are even higher than one would expect for their greater size. However, attempts to identify a genetic basis for this excess diabetes in Africans have so far been unsuccessful [37], with studies of the African diaspora showing marked differences in glucose metabolism in different communities with different BMIs and degrees of abdominal obesity enhancing glucose intolerance but they were also eating different diets and with objectively measured differences in physical activity [38]. So there was no suggestion that dilution of the African genome in Jamaica and the United States had distinct weight independent effects as their results did not differ from those observed in Ghana and South Africa.