BMIs in non‐Caucasians?
The BMI limits of 20–25 for minimum mortality rates based on US life insurance data set out in the early 1970s were reaffirmed in the 1983 analysis from the London Royal College of Obesity [4] using data from the US Cancer Society that analyzed data on smokers and nonsmokers separately. Therefore, the big issue was whether these limits applied globally. It was hoped that this could be answered with the establishment of the International Obesity Task Force (IOTF) in 1996, but at that time, there was limited longitudinal data on adult weights and heights and their subsequent mortality in non‐Caucasian populations. However, when the IOTF proposed in 1997 cut‐off points of 25 and 30 to define new WHO criteria and set out policies for tackling global obesity, the Japanese delegate desired a lower upper normal BMI cut‐off point of 23.0, whereas the US delegate favored a higher normal cut point of 28.0 despite the original mortality data being derived from the United States. The Japanese argued that setting the upper normal BMI cut‐off point at 25 value did not adequately define overweight in Asian populations, but the United States felt that such a cut point would automatically mean that a large majority of Americans would be defined as overweight which was deemed embarrassing and requiring a rethink of health strategies!
The WHO expert committee decided to continue the long‐standing policy of considering the human race as one entity with ethnic differences being unimportant biologically, so they agreed on a universal upper normal BMI cut‐off point of 25 [5]. However, it was later proposed [6,7] that the BMI lower limit should be 18.5 rather than 20 as there was little evidence at that stage that mortality increased as BMIs fell below 20 and detailed analyses of ill health in Latin America, Africa, and Asia indicated that there were no health disadvantages at this lower level. However, at that stage, data on large populations examining the relationship of BMI to mortality was limited except in India, where it was clear that mortality rose sharply when BMIs were below 16.
Subsequently, because of the intense concern of many Asian physicians about the burden of ill health, especially diabetes, that arose within the supposedly acceptable BMI range of 20–25, a WHO meeting was convened in Singapore. It concluded that there were differences in the relationship between BMI and the health profile as well as body composition when comparing Western populations to data from several Asian countries. Therefore the option of considering an upper BMI limit of 23 was acceptable in Asian countries [8]. China, however, undertook their own extensive analyses when their Chinese obesity collaboration was formed and then concluded that an upper limit of BMI 24 should be suitable for the Chinese [9], but this judgment, as well as the Singapore conclusion about Asians, was geared more to morbidity relationships than to mortality data.
Morbidity burden starts at much lower BMIs
The setting of these cut points has led many clinicians to assume that a BMI between 20 and 25 is optimum as the mortality risk is minimum. However, it has been known for decades that the risk of diabetes increases progressively from a BMI of about 20, and in the United States is then 5 times higher in women before the BMI of 25 is reached [10]. Furthermore, the incidence of hypertension and increases in blood cholesterol levels, and the risk of both cardiovascular diseases and colon cancer show linear increases as the BMI moves up from a BMI of about 20. So considering a BMI in the range of 20–25 as optimum may result in neglecting all the increasing comorbidities of weight gain within this “acceptable” range. These issues will be considered in greater detail later when we consider the overall hazards of obesity.
Obesity epidemic starts in the early 1980s
Data from the United States suggests that mean population BMI was increasing consistently across the decades from the early 1900s, but the level of overweight and obesity, as defined by a BMI of 25 and 30, only began to increase rapidly in the early 1980s [11]. The Organization for Economic Cooperation and Development (OECD) report on obesity showed that in the 1970s the overall US adult obesity prevalence was already 14% for all adults, but it has risen progressively since then and continued to increase in the 2010s to above 35%. Similar patterns were seen in other developed countries. National surveys of English adults in the 1930s and 1940s reported obesity rates less than 5% in those below the age of 40 years but increasing to 10% in the 50‐year‐old men. Detailed national representative surveys in England in 1981 showed that overall obesity rates remained relatively low (6% in men aged 16–65 years and 8% in women) [12], but obesity rates had already risen to 11–12% in 40–60‐year‐olds. Similar prevalences were found in studies in Finland (albeit those data were self‐reported), and the Netherlands also reported an average adult prevalence of 5%, but measured prevalence data from Japan on average showed that only 2% adults with obesity.
From the early 1980s, it is clear that obesity prevalences were starting to increase in all Western societies, and by the late 1980s, upper‐middle‐income countries were beginning to follow the North American and European countries, with increasing obesity rates evident in both men and women from the age of 20 upwards. The pivotal importance of this early part of the 1980s in setting new trends of increasing obesity was vividly illustrated by Norton and others’ analysis of children’s weights and heights measured in community and other population surveys in Australia over a whole century [13]. By using the IOTF’s BMI criteria for childhood obesity [14] (which linked seamlessly with the adult BMI cut‐off points), Norton’s analysis showed that there was very little increase in childhood obesity until the early 1980s when a remarkable increase started to develop. These IOTF cut‐offs have subsequently been refined [15], but the overall picture and analyses are unchanged and reveal something very unusual changed in the environment from the early 1980s onward in affluent societies, with middle‐income countries revealing the beginning of a rise in BMIs a little later.
The global epidemic gets underway
The trends in obesity rates from the 1980s to 2008 were beautifully illustrated by Finucaneand others [16] in a comprehensive analysis of huge data sets from around the world, as shown in Figure 1.1. The data reveal marked differences in the prevalences of males and females. In males, it is clear that the more affluent a society in 1980 then the greater the likelihood of some obesity. Thus North America, Latin America, Australia, and Europe had overall prevalences of 10% obesity. In general, the greater the degree of regional affluence, the greater the increase in obesity rates in men over the subsequent 28 years. However, in the early 1980 data, the highest level of obesity among women was not found in North America but rather in emerging counties in Southern Africa, North Africa, the Middle East, and Central Latin America, with Asian women having the lowest levels. In the following three decades, the 1980 regional ranking in women was generally preserved with the fattest regions showing the most significant increases in obesity so that by 2008 several regions of the world were approaching a 40% prevalence of obesity in women.
Since these comprehensive analyses of obesity prevalences, there has been a series of updates both by the Ezzati and Murray groups [17] supported by the Gates Foundation and by the OECD [18] with updates [19] – as well as by a range of national expert groups [20,21]. These data stimulate the question as to why this epidemic had become so striking and seemingly resistant to change and, indeed, how might the problem be tackled? This resistance to change is evident, for example, from surveys conducted by Public Health England who showed that 40% or more of men and over 50% of women aged 25–74 years were trying to lose weight in 2016 [22] and yet the obesity rates remain high suggesting that under current circumstances individuals attempt to slim is very ineffective as a population strategy.
Using historical records, Jaacks and others [23] then examined the evolution of obesity and highlighted four phases in the chronological development of obesity:
Stage 1. Obesity is more prevalent in women than in men and is evident in more affluent groups with low prevalence rates in children. This phase is still evident in many South Asian countries and sub‐Saharan Africa.
Stage 2. In stage 2 of the transition, there has been a significant increase in the adult obesity