Commentary: The epidemic of obesity—what's in a name?

Increases in body mass index and in the prevalence of obesity have been observed in many parts of the world. Campos et al.¹ criticize the description of obesity as ‘epidemic.’ Leaving aside for the moment the use of the word epidemic to describe obesity, they may be dismissing this phenomenon too hastily. Nationally representative data on measured heights and weights in the United States from the National Health and Nutrition Examination Survey (NHANES) programme, show that from the first Health Examination Survey in 1960–62 through the second NHANES in 1976–80, there was little change in the population.² Mean body mass index for men varied from 25.1 in 1960–62 to 25.6 in 1976–80 and for women from 24.9 to 25.3. These changes in BMI would be equivalent to a change in weight of ∼1.5 kg for men of average height and of a little over 1 kg for women.

After this 20 year period of relative stability through 1980, data from the subsequent survey (NHANES III; 1988–94) showed unanticipated increases in mean BMI and in the prevalence of overweight. As one who participated in the events leading up to the initial publication in 1994³ of these findings, I can attest that we were initially surprised by these results. We were concerned that we might have overlooked some error in data collection or processing, but no errors were found. The increases that were first noted in NHANES III have continued. Age-adjusted mean BMI for men increased to 26.8 in 1988–94 and again to 27.9 in 1999–2002. Comparable figures for women are 26.6 and 28.2. These BMI changes are equivalent to a >7 kg increase in weight since 1976–80 for both men and women of average heights.

Our investigations show that in fact the entire distribution of BMI has shifted slightly to the right in the US population, as well as becoming more skewed.⁴ Even the lowest percentiles of BMI in NHANES III were slightly above the corresponding percentiles in NHANES II and the distributions also became more skewed; thus, the prevalence of obesity is increasing faster than would be expected from the increase in mean BMI. The increasing prevalence of overweight or obesity is due to this shift in the entire distribution of BMI; it is a manifestation of underlying changes that are taking place in the entire distribution of BMI in the population.

These observations have proved difficult to explain. Clearly some kind of change in energy balance has occurred, but we have surprisingly little ability to explain exactly why this has happened. Why was body mass index relatively stable and why did it begin to increase fairly suddenly? Why is the distribution becoming more skewed? Despite the intense attention paid to obesity and the numerous hypotheses that have been put forward, we still lack explanations supported by data and we lack understanding of the mechanisms driving these changes.

Is this an epidemic?

A technical definition of ‘epidemic’ from the Dictionary of Epidemiology⁵ reads in part ‘The occurrence in a community or region of cases of an illness … or other health related events clearly in excess of normal expectancy.’ According to this definition, the crucial defining aspect is whether the cases, regardless of their number, exceed ‘normal’ expectations. A more general definition of the word epidemic can be found in any standard dictionary of the English language, for example⁶:

Adjective: 1. Affecting many individuals throughout an area at the same time (Cholera was epidemic). 2. Widely prevalent (epidemic anxiety)

Noun: 1. A rapidly spreading outbreak of contagious disease. 2. A rapid spread, growth, or development (an epidemic of robberies).

As can be seen from this definition, the word epidemic may be used in metaphorical ways not related to health and disease (anxiety, robberies). This definition also highlights two different aspects of this concept—one is high prevalence, the other is rapid increase.

The ‘obesity epidemic’ clearly fits the general definition, in that the prevalence of obesity is high and the increase over the past two decades has been rapid compared at least with the immediately preceding decades. Although the technical definition ‘clearly in excess of normal expectancy’, is qualitative, not quantitative, the increases seen in NHANES III were well in excess of what would have been expected based on previous surveys.

Epidemics of infectious disease have had profound impacts on human society and have attracted interest from historians at least as far back as Thucydides, who described the plague of Athens in 430 bc.⁷ Contemporary historians and medical sociologists have studied past and present epidemics, with additional importance being given to this topic by the AIDS epidemic.^8–13 As described by Rosenberg⁸ a classic epidemic is not only an infectious disease but a social phenomenon that historically displays several distinct stages. First, there is a reluctance to accept and acknowledge the epidemic, until admission of its presence is unavoidable. Once the epidemic has been recognized, an explanatory framework is created in which to understand it, which often expresses and legitimizes moral and social assumptions, reaffirming social values, and blaming victims. In the third stage, a public response is created. In fact one of the defining characteristics of an epidemic is the pressure it generates for decisive and visible community response. Finally, according to Rosenberg, an epidemic usually ends with a whimper not with a bang, as incidence returns to the former levels.

A search for the word ‘epidemic’ in Medline using PubMed showed that from 1995 through mid-2005, >4000 entries included the word epidemic in the title, of which 3600 referred to humans rather than to animals. The largest proportion of these articles referred to infectious diseases. Other common topics were poisons and toxins, cancer, cardiovascular disease, diabetes, obesity, violence and injuries, illegal drug use, asthma, reproductive health, multiple pregnancies, osteoporosis, and arthritis. The word was also used for a widely varied number of other topics, including dieting, domestic violence, obsessive-compulsive disorder, medical errors, disability/absenteeism, and false memory syndrome.

These uses of the word not just in the press and in popular books, but also in scientific journals as indexed in Medline, often go far beyond the historical meaning of the word epidemic in the sense of an outbreak of infectious disease. Rosenberg spells this out clearly.

We use the term ‘epidemic’ in a variety of ways—most of them metaphorical, moving it further and further from its emotional roots in specific past events. Even in relation to health, we employ the word in contexts decreasingly related to its historical origins. Medical historians speak of an epidemic of tuberculosis in Europe between 1700 and 1870 and of an epidemic of rheumatic fever in the century and a quarter after 1800. In the mass media every day, we hear of ‘epidemics’ of alcoholism, drug addiction, and automobile accidents. These cliched usages are disembodied but at the same time tied to specific rhetorical and policy goals. The intent is clear enough: to clothe certain undesirable yet blandly tolerated social phenomena in the emotional urgency associated with a ‘real’ epidemic.¹³

In a footnote to this paragraph, Rosenberg points out that the listed conditions are endemic, not epidemic phenomena, and that a fundamental aspect of the meaning of epidemic is precisely in its contrast with such phenomena. ‘A true epidemic is an event, not a trend.’

A high prevalence of overweight is not a new observation in our society. In 1988–94, more than half the US adult population (56%) was considered overweight (defined as a BMI ≥ 25) and 24% as obese (defined as a BMI ≥ 30). When the same definitions are applied to older data, however, it is clear that these prevalences have been high for many years. The earliest US national survey data available show that even in 1960–62, 45% of the population was overweight (BMI ≥ 25), including 50% of men and 40% of women. In the 1970s >22% of male adolescents ages 18 and 19 years had BMIs of 25 or above.

As noted by Rosenberg, the most obvious defining characteristics of an infectious disease epidemic are fear and sudden widespread death. Discussions of the effects of increases in overweight and obesity often refer to adverse health events that are predicted to occur in the future (e.g.¹⁴). However to date it is not clear what the major health effects of the trends in obesity and overweight have been. In the US, life expectancy continues to increase and death rates from ischaemic heart disease continue to decline.¹⁵ Cardiovascular risk factors, with the exception of diabetes, have dropped since 1960 at all BMI levels.¹⁶ More recent mortality follow-up from national surveys suggest a possible decrease in the relative risk of mortality associated with obesity.¹⁷

Summary and conclusions

The word ‘epidemic’ has some drawbacks as a descriptor. Because it has no quantitative definition, there is no precise way to determine whether something is an epidemic or not, and opinions may differ. The word also has many social and political overtones and is often used more in a metaphorical than a technical sense, even for health conditions. The word epidemic masks some of the characteristics of obesity—the endemic character of overweight, the sustained upward trends in weight associated with economic development, the shift in the entire distribution of BMI, the uncertainty about the long-term health effects associated with recent increases, and the difficulty of defining or accomplishing an end to the epidemic. Thus, the implications that might be drawn from an infectious disease epidemic do not necessarily apply to obesity.

Nonetheless, obesity does have some characteristics of an epidemic. Although it is difficult to say exactly what the ‘normal’ level of obesity is, it is clear that a surprising and unexpected increase in obesity occurred in the US in the 1980s and that the increases are continuing. Given the lack of a quantitative definition of epidemic, we cannot appeal to numbers to decide the question. Regardless of how these increases are labelled, however, it is important to understand these changes and their potential impact on health.

Disclaimer

The findings and conclusions in this report are those of the author and do not necessarily represent the views of the funding agency.

References