- Email: [email protected]
Standards and classification: A perspective on the ‘obesity epidemic’
Social Science & Medicine 87 (2013) 9e15
Contents lists available at SciVerse ScienceDirect
Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed
Standards and classification: A perspective on the ‘obesity epidemic’ Stuart G. Nicholls* Department of Epidemiology and Community Medicine, University of Ottawa, 451 Smyth Road, Ottawa, Canada K1H 8M5
a r t i c l e i n f o
a b s t r a c t
Article history: Available online 18 March 2013
In this paper I critique the increasing standardization of obesity. Specifically, I consider two ‘definitional turns’: the way language has been standardized to such an extent that it obscures uncertainty and variation, and the appearance of objectivity through quantification and standardized measurement. These, I suggest, have fostered a simplified picture of obesity, promoting the classification of weight and thereby facilitating the emergence of the ‘obesity epidemic’. These definitional turns fail to acknowledge the distinctions between fat and mass and intraclass variation within weight categories. A consequence of this process of simplification has been the erroneous application of population level information to individuals in a clinical context, with potentially harmful results. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: BMI Standards Epidemiological methods Variation
“ -But let us not forget this either: it is enough to create new names and estimations and probabilities in order to create in the long run new ‘things.’”(Nietzsche, 1974 122, aphorism 58)
Introduction There are now a multitude of studies reporting dramatically increasing levels of obesity over the last twenty to thirty years (Ahrens, Moreno, & Pigeot, 2011; Manios & Costarelli, 2011). These studies not only indicate that the number of obese individuals is increasing, and is as high as 33% in some countries (Flegal, Carroll, Ogden, & Curtin, 2011), but that average weight is also increasing (Finucane et al., 2011). A global analysis of data estimated that in 2008 over 205 million men and 297 million women over the age of 20 were obese (Finucane et al., 2011). Moreover, this includes increasing numbers of ‘morbidly obese’ individuals, skewing the distribution of weights towards the upper extreme (Yanovski & Yanovski, 2011). The increase in obesity would not be so concerning if it were not for the increasing number of adverse health effects associated with it. To date studies have indicated relationships between obesity and a range of conditions including type 2 diabetes mellitus, fatty liver disease, endocrine and orthopaedic disorders and most of the major cardiovascular risk factors (Lobstein & Baur, 2005; Manios & Costarelli, 2011; Reilly et al., 2003). The increasing prevalence of obesity together with the indicated negative health effects have led * Tel.: þ1 613 562 5800x8288. E-mail address: [email protected]. 0277-9536/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.socscimed.2013.03.009
some authors to define the current situation as an ‘obesity epidemic’ (Flegal, 2006). Epidemiological data is often presented to underwrite these claims. Much of the data on which the estimates are based comes from national surveys using the Body Mass Index (BMI). The BMI derives from “Quetelet’s index” (Smalley, Knerr, Kendrick, Colliver, & Owen, 1990) which was developed in the 1800’s to chart the range of heights and weights of army conscripts (Oliver, 2006). In this original conception Quetelet noted a Gaussian (normal) distribution of weight to height ratios within the population, allowing for the description of the statistically average man (Oliver, 2006). Today the BMI calculated as weight (in kg)/height (in metres squared), is used to provide an estimate of body composition. Leaving aside the self-reported nature of much of the available survey data (Manson et al., 1995; Strauss, 1999; Yanovski & Yanovski, 2011), a question remains regarding the interpretation of changes in BMI. What does an increase of one BMI represent? Is there a linear trend with increasing weight, or is it a more complicated relationship such as a normal distribution or U-shaped relationship? Does each BMI increase of one have the same effect size on the specified outcome? Continuous traits, such as weight or BMI, are not amenable to straightforward assessments in the same way as grouped data. Far easier is the assignment of risk to discrete classes or categories. Sexbased risk, for example, has an altogether simpler interpretation; a man may have one risk, a woman another. The creation and use of categories for underweight, normal weight, overweight, and obese have been central to the analysis and presentation of risk estimates, and indeed goes to the very core of data purporting that an obesity epidemic has emerged.
10
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
To date, while there has been much said of the shortcomings of the BMI, there has been little discussion of the way in which the BMI has been applied nor the processes through which the BMI has become the dominant tool on which obesity prevalence and risk have been determined. In this paper, I consider both the standardization and classification of obesity and the roles these have played within the ‘obesity epidemic’. In doing so I engage with the hitherto under explored processes that have reified the BMI as the measure of obesity. Specifically I argue that the language of obesity has been standardized to such an extent that it obscures uncertainty and variation in the assessment of obesity, and the quantification and standardized measurement of obesity has furthered this simplification and has facilitated the perception of an obesity epidemic, obscuring the nuances of the data collected. This, I contend, has important, and potentially harmful, effects when it is misapplied within the clinical context. Classification: making up obese people But what does it mean to talk of classification? Indeed what do we mean by classification? To answer these questions I draw on Bowker and Starr (1999) who define a classification system as: “a set of boxes (metaphorical or literal) into which things can be put to then do some kind of work e bureaucratic or knowledge production.” (Bowker & Starr, 1999, p.10) In order for classification to occur the categories must be consistent, mutually exclusive, and complete e that is, no object from the same population may exist outside of the proposed categories. The role of classification By classifying, we group items based on some characteristics and in some way that we identify them as similar. Equally, we identify others as dissimilar in some other important or functional way. While there may be no single aim to the classification of things - and authors have postulated a range of possibilities (Caplan, 1997; Dupré, 2006; Jutel, 2006, 2011; Zerubavel, 1996) e I will follow Hacking (1988) by emphasizing two main purposes: cognitive and bureaucratic functioning. From a cognitive perspective one role for classification is to simplify the world, to reduce disorder to order (Jutel, 2011). In classifying things, we are able to streamline our perceptions and memories. We are also able to facilitate the production and determination of relationships between objects or actors, to develop explanations or model interactions in order to create predictions (Hacking, 1988; Jutel, 2011). Yet classification can also serve (but is not limited to) functional or bureaucratic purposes. One of these purposes is the ability to count. We may count in order to find out how many there are of something. More likely, as Jutel notes, counting is undertaken to assist in the answering of questions (Jutel, 2011). In the context of healthcare, we may want to know how many people are sick so that we can inform treatment protocols, health service planning, or budgeting (Jutel, 2009). The effect of classification As suggested, the process of classification requires one to group items based on characteristics that identify them as similar. A consequence of this is that the differences between individuals within a category or class e intraclass differences e are downplayed, while the differences between groups e the interclass differences e are overstated (Tajfel, 1981; Zerubavel, 1996). Put
differently, the effect of classifying individuals into groups is that grouped items are perceived as being more similar than items that exist outside of the group. Zerubavel gives the example of boxing weight categories: “[.] we perceive the metrically negligible “distances” between 119-pound (“bantamweight”) and 120-pound (“featherweight”) boxers [.] as greater than those between 120-pound and 125pound (both “featherweight”) boxers.” (Zerubavel, 1996, p.425) To the same extent that weight is manifest as discrete boxing categories, BMI is categorized into discrete classes of underweight, normal weight, overweight or obese. This serves to minimize the differences within these weight categories and introduce perceptions of significant differences between classes. This is despite the fact that the difference between normal weight and overweight is potentially the difference between a BMI of 24.9 and 25.0 while the within group variation of normal weight can be as much as the difference between a BMI of 18.0 and 24.9. To put this more starkly, for an individual who is 1.75 m tall the variation within the “normal” range constitutes 20 kg (Mascie-Taylor & Goto, 2007), while the between group variation between normal weight and overweight can be as little as 0.1 kg. It is, therefore, naïve to treat all those within a category as a homogeneous group when there may be substantial differences within a group, and minimal differences between groups at the boundaries. Grouping does not just classify weight, it classifies people. An often neglected effect of classification is that it can affect how we see those so classified in ways that are more substantial. For example, the classification of people as being of normal or ill health often affects how we respond to them, both in terms of resource allocation (as mentioned above in the pre-emptive role of classification), but also in a personal sense; it may affect whether we wish to associate with the individual or how we do so. This latter sense may be an outcome of stereotypes which, themselves, involve a process of classification: population level generalizations are applied to individuals, removing the complexity of variation and neglecting individual differences within groups (Tajfel, 1981). In this way, the classification of individuals as normal weight or obese may affect our assessment of the individual before us. For instance, studies have found that moralistic terms such as ‘lazy’ and ‘gluttonous’ are used to describe individuals perceived as obese, purely on the basis of their weight (Puhl & Latner, 2007; Schwartz & Puhl, 2003; Tiggeman & Wilson-Barrett, 1998). Ominously, studies indicate that these negative attributions are also held by those in a position to help obese individuals. In a French study of General Practitioners, 30% of those surveyed agreed to some degree that ‘Obese people are lazier and more self-indulgent than normal weight people’, with 28% indicating the same attitude for overweight people (Bocquier et al., 2005). The potential clinical implications are highlighted in a study by Hebl & Xu (2001) in which physicians who reviewed case studies for average weight, overweight, and obese individuals, indicated they would spend less time with obese individuals, would have less patience with them, and had significantly less desire to help the patient. These assessments paper over, if they even acknowledge, the in-group variation as well as illustrating the way in which stereotypes bring with them additional characteristics, in this instance perceptions of moral character. The effect of classifying individuals based on their BMI may serve to propogate such stereotypes by ‘revealing’ an individual to be overweight or obese when they physically appear ‘normal’. In a study of a school based weighing and measurement programme it was noted that while the majority of children did not express concerns, a minority disliked or hated the process, with negative
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
comments from participants indicating the potential for teasing if weight status was known (Grimmett, Croker, Carnell, & Wardle, 2008). Moreover, the child’s measured weight did not always concur with parental perceptions of weight status. In some cases the revelation of weight status was found to affect parental behaviours; there was a statistically significant increase in parental restriction of access to foods in girls who were identified as being overweight. More concerning though are findings from studies of school based screening programmes which indicate that some parents of children of normal weight reported introducing dietrelated activities and seeking medical advice (Chomitz, Collins, Kim, Kramer, & McGowan, 2003; Maclean et al., 2010). The revelation of an individuals’ BMI can in itself create a perceived need for intervention where the child has been assumed to be healthy previously. Consequently, classification may serve multiple purposes and have multiple effects. The effects of this classification can be both intended and unintended. Classifications can have the (intended) effects of restructuring healthcare services that are responsive to actual patterns of need, but also the (unintended) effect of introducing stereotypes and stigmatization. Facilitating classification As suggested, an important aspect of classification is the ability to identify similarities upon which to group items (or in the case of obesity, individuals). However, while there has been a great deal of discussion regarding the BMI levels associated with obesity, there has been little discussion about the establishment of the categories themselves and the implications of these categories and how they have facilitated the perception of obesity epidemic. In the remainder of this paper I consider two distinct ‘definitional turns’ that have served to create a simplified picture of obesity, both of which have revolved around increasing standardization of obesity definitions and have facilitated the classification of weight. I suggest that such definitional turns act through (1) a standardization of language that obscures uncertainty and variation, and (2) the appearance of objectivity through quantification and standardized measurement. The importance of this, I contend, is that these have served to present an overly simplistic picture of obesity, have facilitated the production of the obesity epidemic and, in part, have led to development of inappropriate interventions. This is not to suggest that the number of individuals e adults and children e at increased weight has not grown, there is ample data to support this (Finucane et al., 2011). Rather, my claim is that the current description and debate regarding the obesity epidemic lacks the nuanced discourse that acknowledges the distinct limitations of using standardized categories based on the BMI (and the uncertainties relating to obesity itself) and that this is promulgated by the language and approaches employed. The language of obesity: standards and certainties The first of the aforementioned definitional turns has been the increasingly standardized language around obesity and the way this has gradually minimized the variation in the assessment of weight or assignment of weight classes. This is particularly acute in the case of obesity. Early definitions referred to overweight as “increased bulk of the body, beyond what is sightly and healthy,” (Herrick, 1889 quoted in Jutel, 2006, p.2270), thus employing a subjective analysis of bulk (generally) with regard to what might or might not be deemed as ‘sightly’. In later definitions, overweight was typically defined in terms of body mass or relative weight, while obesity was a condition characterized by excessive body fat (Himes & Dietz, 1994).
11
Overweight, therefore, related to mass, and obesity to fat. Consequently, they represented qualitatively different constructs. Yet weight (or mass) reflects both weight contributed by fat, but also other tissues such as muscle, or bone. As such, fat is a component of mass, but mass will only be partly concerned with fat. This distinction is not only conceptually important, but also has clinical implications. If obesity is defined in terms of fat, this requires a method of assessment that gauges body fat percentage. Methods such as tricep skinfold thickness, whole body plethysmography or dual x-ray absorptiometry may be more appropriate (Finer, 2012). However, these methods not only require more time, they are substantially more expensive than the assessment of weight and calculation of BMI. The underlying problem is that while BMI is (partly) correlated with percentage of body fat (Smalley et al., 1990; Widhalm, Schönegger, Huemer, & Auterith, 2001), it is also correlated with bone density and mass more generally. However, more recent definitions have sought to standardize the constituent elements of overweight and obesity, principally by relating both to the BMI. Overweight has become a term used to describe unhealthy amounts of body fat, not body mass. An example of this occurred in 2007 when an expert committee established by the US Health Resources and Service Administration and the Centers for Disease Control and Prevention (CDC) endorsed the earlier Institute of Medicine (IOM) decision to depart with existing terminology, and change the category of ‘at risk for overweight’ to ‘overweight’ and the replacement of the original category of ‘overweight’ with ‘obese’ (Barlow, 2007). As reported by Himes (2009), an impetus for these changes was the perceived need to “express this concern [about increased weight] by using the term “obese”” (Himes, 2009, p.S8) to replace the previously used term “overweight”. This standardization of terms aligns both overweight and obese onto a single construct; both overweight and obese should refer to differing levels of BMI (Krebs et al., 2007). As such, both overweight and obesity, quantified by BMI, have become ‘measures’ of body fat (Hubbard, 2000). This change is now widespread; the recent UK National Health Service (NHS) report, Statistics on obesity, physical activity and diet : England, 2010, states that “Overweight and obesity are terms that refer to an excess of body fat and they usually relate to increased weight-for-height” (The NHS Information Centre, 2010 8, emphasis added). This definitional slide now means that the common use of the term obese (often in relation to BMI levels) is, in effect, an extension of the overweight category but demarcating those at the extreme end of the scale. As such, there has been a move from different but associated concepts e mass and fat e to a situation whereby both terms operate on a single construct. This obfuscation of the distinction has important methodological and interpretive implications. Firstly, how one defines obesity operationalizes how one measures or evaluates the level of overweight and obesity. If one defines obesity as the bodily composition of fat, but overweight as mass, then different measures may be used. Aligning the terms simplifies the measurement process. The importance of this difference is illustrated by two papers, both of which evaluated the same three waves of the US National Health and Nutrition Examination Survey (NHANES), but which came to different conclusions regarding obesity levels and trends (Flegal, 1993). The first paper, by a team led by Steven Gortmaker (Gortmaker, Dietz, Sobol, & Wehler, 1987) assessed tricep skinfold thickness in order to assess percentage of body fat. In contrast, Harlan, Landis, Flegal, Davis, & Miller (1988) used BMI. The studies came to diametrically opposed conclusions regarding trends in obesity for children. Moreover, in a comparison of prevalence estimates of overweight, the estimates produced using the BMI were
12
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
consistently higher, in the region of 10% higher, than those from measurements of tricep skinfold thickness (see Flegal, 1993 310). A simplified standard, in which BMI refers to both overweight and obesity, not only facilitates the removal of conceptual differences e both overweight and obesity now refer to the same concept and so can be seen on a single metric e but in doing so it demarcates overweight as the stage prior to obesity. This turns overweight into what Rosenberg has termed a ‘proto-disease’, that is into a pre-disease state, that: “Once articulated and disseminated in practice and the culture generally [.] become emotional and clinical realities, occupying a position somewhere between warning signal and pathology.” (Rosenberg, 2002, p.254) Hence overweight has become the pre-disease state for obesity meaning that overweight is now a legitimate point of intervention as it is a precursor to the unhealthy disease state of obesity. Moreover, the complexity of using terms that refer to different constituents is removed with the underlying elements aggregated. This in turn facilitates the classification of a single scalar concept into a limited number of categories as opposed to the more complex and uncertain situation of trying to assimilate two distinct nominal categories. Standardization and the appearance of objectivity A second notable trend, as alluded to earlier, has been the migration from more qualitative descriptions of obesity to descriptions defined by measurement and the subsequent standardization of these measures (Jutel, 2006). Such a trend, I suggest, has acted as a catalyst for the generation of the obesity epidemic by creating an appearance of objectivity and allowing international comparisons. This not only contributes to the ability to construct an ‘obesity epidemic’ but also an ‘obesity pandemic’. This quantification and standardization is most clearly seen through the almost universal adoption of the BMI as a measure of obesity. Indeed, the application of the BMI as an obesity measure is itself one of simplification. While several studies have suggested that BMI is correlated with percentage of body fat (Smalley et al., 1990; Widhalm et al., 2001), it is also correlated with bone density and mass more generally (Colls & Evans, 2010). A major driver in the adoption has been the relative simplicity with which it can be applied (Evans, 2009; Widhalm et al., 2001). Since the general adoption of the BMI, there has been a gradual move towards a universal standard of categories for overweight and obesity (Ebbeling & Ludwig, 2008). Thus, prior to 1999, many experts considered a BMI of 29 kg/m2 to be overweight. Later this was reduced to 27 kg/m2 and today a BMI above 25 kg/m2 is the standard used to indicate overweight (in adults) (Ross, 2005). This gradual reduction and standardization of categories simplifies and increases the appearance of certainty. The setting of a universal standard not only facilitates comparisons between populations but also presents an air of objectivity in the way it papers over potential differences between groups or populations. Yet even small changes can have a substantial impact on estimates of prevalence levels. McKay (2002), for example, argues that the 1998 reduction in the BMI threshold for overweight to 25 kg/m2 “made an extra 30 million Americans ‘overweight’ overnight.” Furthermore, studies suggest that different ethnic populations may have the same percentage of body fat at different levels of BMI or, put differently different levels of fat at the same BMI (Freedman et al., 2008; Luke, 2009; Rush, Freitas, & Plank, 2009; Wulan, Westerterp, & Plasqui, 2010). A universal standard will likely underestimate the potential health effects in some populations with higher percentage body fat to BMI ratios (for example Asian
populations) and overestimate the effect in other populations with lower percentage body fat but higher muscle mass (such as Pacific Island populations) (Luke, 2009). Indeed, variance may already be creeping into assessments as we have begun to see the emergence of what Hacking refers to as a ‘looping effect’ in which the actual classification feeds back to change the category due to the way in which those so classified perceive themselves (Hacking, 1999). The production of population specific BMI cutoffs is an example of this, with specific models being developed for Asian and Thai populations (Fu et al., 2003; Shiwaku, Anuurad, Enkhmaa, Kitajima, & Yamane, 2004; Thaikruea, Seetamanotch, & Seetmanaotch, 2006). Consequently, we are seeing what Hacking describes as “the rebellions of the sorted” (Hacking, 1999 131). This, as noted by Timmermans and Almeling (2009), is an observed process of standardization in which a level of disorder re-emerges often in response to a proliferation of standardization (Timmermans and Almeling, 2009). As such, the appearance of population specific cutoffs indicates the non-standard nature of differing populations and fallacy of attempting to produce international standards that are inappropriate to the context to which they are being applied. Implementing standards: a cautionary note Identifying the process of standardization is important, for alongside the increasing standardization there has been a creep of overly simplified usage of obesity classification and the not un-problematic application of population level standards to individuals. Again, this move has been gradual and the current individual application of the BMI differs substantially to its origins as ‘Quetelet’s index’ where the purpose was to determine population averages and ranges, rather than using it for a clinical purpose at the individual level. Definitions of overweight using the BMI, provides only a crude population-level measure, and while valuable for its convenience and simplicity in public health surveillance, screening, and similar purposes it lacks the sensitivity or specificity to be used as a diagnostic tool. Of course an appeal of the BMI is that it is fast, convenient, and noninvasive compared to the time consuming measures mentioned earlier e features that make it both eminently practical for use in large prevalence studies (Moffat, 2010), but also appealing to employment within the clinic as it requires no specialist equipment. This should not, however, blind those who use it to its shortcomings in a clinical setting. As a clinical measure, therefore, it lacks utility; primarily because individuals with the same BMI may not have the same health risks (World Health Organization, 1998) and so it “[does] not necessarily identify physiological states per se” (Flegal, Tabak, & Ogden, 2006). Thus, some have cautioned that the BMI should not be used diagnostically (Kuczmarski & Flegal, 2000). Evidence suggests, for example, that individuals at the lower end of the ‘overweight’ category do not necessarily have worse overall health outcomes (e.g. Calle, Rodriguez, Walker-Thurmond, & Thun, 2003; Carnethon et al., 2012; Uretsky et al., 2007), and may be appear to be ‘metabolically healthy’ when compared to other ‘normal weight’ individuals (Carnethon et al., 2012). This not only problematizes the imposition of overweight as a precursor to obesity (and associated ill health) but also suggests that lifestyle factors, such as diet, physical activity levels, or environmental exposures may, therefore, be more relevant to clinical care as opposed to a focus on weight per se. Despite these cautions, growth curve charts for paediatric overweight and obesity have now been used in the clinic (Voss, Metcalf, Jeffery, & Wilkin, 2006). Children in England now have their BMI assessed as part of the National Child Measurement Programme. While the scheme aims to provide a national picture, and so collect population prevalence, it also uses the child’s BMI to
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
provide individual assessments. Programme materials tell parents that: “You can also benefit, as you can receive your child’s individual results. This will help you to know if your child is in the healthy weight range. If your child is overweight, further support is available from your local NHS.” (NHS Choices, 2010) The invocation of the National Health Service (NHS) implies that the overweight status is inherently problematic. Indeed, admissions from the Department of Health that children who are obese and overweight are to be proactively followed up indicates its use as a clinical tool (Wheeler & Twist, 2010). Yet BMI category, in isolation of all other information, represents an extremely simplistic assessment; there is no consideration of the general health of the child. The programme itself runs counter to the advice received from its own National Screening Committee, who advised that the screening programme should not go ahead as it could not guarantee doing more good than harm (Colls & Evans, 2010). However, this is not an isolated case. The Australian National Health and Medical Research Council (NMHRC), in their 2003 document Clinical Practice Guidelines for the Management of Overweight and Obesity in Children and Adolescents, explicitly state that: “BMI-for-age percentile charts should be used in clinical practice [.] a BMI above the 85th percentile being indicative of overweight and a BMI above the 95th percentile being indicative of obesity.” (Steinbeck, 2003, p.14, emphasis added) BMI has also been introduced into other problematic areas, such as recruitment for the military and has been reported as a trigger to initiate retraining or as grounds for dismissal if a certain BMI threshold is crossed (Prentice & Jebb, 2001). In a further example from Australia, Queensland has introduced BMI thresholds as part of their assessment criteria for adoptive parents (House of Representatives Standing Committee on Family and Human Services, 2005). Both of these examples are indicative of the way in which the BMI has become institutionalized and implemented without acknowledgement of the uncertainty that is inherent in the simple height-to-weight metric. Yet to apply population characteristics within an individual clinical encounter, solely on the basis of individual BMI, obscures the variance contained within population estimates and inappropriately simplifies the relationship between health effects and overweight or obesity. As Ross (2005) has noted, population comparisons that indicate a 2e3 times greater relative risk of dying in a population of individuals who are obese compared to a population who are not: “[.] means that in a group of people with a BMI 30 kg/m2 it is probable that 2e3 people will die for every 1 person who dies in a comparable group of people who have a BMI 29.9 kg/m2. It does not mean that any given person with a BMI 30 kg/m2 is 2e3 times more likely to die than an individual with a BMI 29.9 kg/m2." (Ross, 2005, p.101) Applying population level associations to individuals has the potential to lead one to commit the “ecological fallacy”. The ecological fallacy refers to the case where relationships found in aggregate level data are assumed to apply at the individual level. For example: “[.] average weight gain at the population level does not necessarily equate with weight gain at the individual level. The actual weight change of a population comprises a wide range of distributions. There may be a portion of the population who lose
13
or do not change weight over time that would pull the mean weight change of the population down, which would make rather significant weight gains in some other groups of the population look trivial; ignoring these dynamics within a population may mask true changes that are happening among individuals.” (Kim & Popkin, 2006, p.62) It need not, therefore, be the case that just because an affluent country has a high prevalence of obesity then affluent individuals will also be obese. Even if data is collected at the individual level, associations are based on group averages and it is an error to assume that these averages of the group are applicable to specific individuals within the group. This is because the assumption ignores the intraclass variation discussed earlier, and represents what Conrad refers to as the individualization of social problems (Conrad, 2007) and what Tajfel (1981) refers to as “personalization”: “If there is to be an explanation in terms of the characteristics of a group, these must be characteristics which are relevant to the situation and common to the whole group, with a corresponding neglect of individual differences between the members of a group.” (Tajfel, 1981, p.138) As such, one cannot assume that because certain health conditions are associated with obesity at a population level, and that the individual in question is obese, that those associations are applicable to the individual. Conclusion The BMI is now the dominant tool on which data purporting the obesity epidemic is based. While there has been discussion regarding the limits of the BMI, and the setting of thresholds for categories, to date there has been a paucity of debate regarding the limitations imposed by neither the categorization itself nor the ongoing standardization of these categories. In this paper, I have addressed this deficit, and have suggested that the act of classifying individuals and populations as overweight and obese has been central to the emergence of the ‘obesity epidemic’. This classification, I have argued, has been accelerated by increasingly standardized definitions of overweight and obesity. This has been characterized by increasingly quantified standards, which carry an air of certainty and objectivity. In doing so I have argued that such trends serve to hide the complexity of overweight and obesity classifications, and have led to the overly simplistic implementation of the BMI. The almost ubiquitous use of BMI as the basis of obesity definitions has facilitated the categorization of weight and the classification of populations and individuals as overweight or obese. This assessment has important implications because of the way it minimizes the intraclass differences and over-emphasizes between class differences. A serious concern raised by this is the failure to acknowledge the effects of such simplification for individuals. This not only relates to the ascription of negative stereotypes associated with overweight and obesity, but also in terms of treatment interventions that may have differential effects if the individuals’ specific circumstances are not considered. An alternative and more transparent approach would be to seek to re-introduce the uncertainty within studies. This could be achieved, in part, through the use of multiple assessments of adiposity via different methods allowing for the comparison of methods and the implications these have for assessment. In addition, the analysis of data on a continuous scale, in addition to the current categorization, would not only facilitate the production of confidence intervals around estimates of adiposity, but would allow for a greater exploration of the relationship between increasing levels of
14
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
adiposity and negative health implications. This approach has recently been taken in a study synthesizing data from 960 country years of data and 9.1 million participants (Finucane et al., 2011). In their analyses, the authors specifically incorporated non-linear associations with age, time trends and stratification by sex. Moreover, their approach allows for the production of uncertainty levels, with greater indicators of uncertainty when data is sparse or inconsistent, producing a more nuanced assessment of the results. While such an approach is undoubtedly time consuming and costly, the examination of non-linear associations, and inclusion of variation, could be adopted by prospective studies as a way of illustrating the uncertainty around estimates of obesity levels and health effects. Acknowledgements This work has benefited from participation within the Identification and prevention of Dietary- and lifestyle-induced health EFfects In Children and infantS (IDEFICS) study. The IDEFICS study is funded through the European Commission Sixth Framework Programme. Area 2: Epidemiology of food-related diseases and allergies. Topic 5.4.2.1.: Influence of diet and lifestyle on children’s health (Integrated Project). Contract n 016181 (FOOD). I would also like to thank Dr Garrath Williams and Dr Kristin Voigt for comments on previous versions of this paper, although I take full responsibility for the arguments put forward. References Ahrens, W., Moreno, L. A., & Pigeot, I. (2011). Childhood obesity: prevalence worldwide e synthesis part I. In L. A. Moreno, I. Pigeot, & W. Ahrens (Eds.), Epidemiology of obesity in children and adolescents (pp. 219e235). New York: Springer. Barlow, S. E. (2007). Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity: summary report. Pediatrics, 120(Suppl 4), S164eS192. Bocquier, A., Verger, P., Basdevant, A., Andreotti, G., Baretge, J., Villani, P., et al. (2005). Overweight and obesity: knowledge, attitudes and practices of general practitioners in France. Obesity Research, 13(4), 787e795. Bowker, G. C., & Starr, S. L. (1999). Sorting things out. Classification and its consequences. Cambridge, Massachusetts: MIT Press. Calle, E. E., Rodriguez, C., Walker-Thurmond, K., & Thun, M. J. (2003). Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. Adults. New England Journal of Medicine, 348, 1625e1638. Caplan, A. L. (1997). The concepts of health, illness, and disease. In R. M. Veatch (Ed.), Medical ethics). Boston: Jones and Bartlett. Carnethon, M. R., De Chavez, P. J. D., Biggs, M. L., Lewis, C. E., Pankow, J. P., Bertoni, A. G., et al. (2012). Association of weight status with mortality in adults with incident diabetes. JAMA, 308(6), 581e590. Chomitz, V. R., Collins, J., Kim, J., Kramer, E., & McGowan, R. (2003). Promoting healthy weight among elementary school children via a health report card approach. Archives in Pediatric and Adolescent Medicine, 157, 765e772. Colls, R., & Evans, B. (2010). Re-thinking ‘the obesity problem’. Geography, 95(2), 99e105. Conrad, P. (2007). The medicalization of society. on the transformation of human conditions into treatable disorders. Baltimore: The Johns Hopkins University Press. Dupré, J. (2006). Scientific classification. Theory, Culture & Society, 23, 30e32. Ebbeling, C. B., & Ludwig, D. S. (2008). Tracking pediatric obesity. An index of uncertainty? Journal of the American Medical Association, 299, 2442e2443. Evans, B. (2009). Measuring fatness, governing bodies: the spatialities of the body mass index (BMI) in anti-obesity politics. Antipode, 41, 1051e1083. Finer, N. (2012). Better measures of fat mass e beyond BMI. Clinical Obesity, 65. Finucane, M. M., Stevens, G. A., Cowan, M. J., Danael, G., Lin, J. K., Paciorek, C. J., et al. (2011). National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9.1 million participants. The Lancet, 377, 557e567. Flegal, K. M. (1993). Defining obesity in children and adolescents: epidemiologic approaches. Critical Reviews In Food Science and Nutrition, 33(4e5), 307e312. Flegal, K. M. (2006). Commentary: the epidemic of obesity d what’s in a name? International Journal of Obesity and Related Metabolic Disorders, 72e74. Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2011). Prevalence and trends in obesity among US adults, 1999e2008. Journal of the American Medical Association, 303(3), 235e241. Flegal, K. M., Tabak, C. J., & Ogden, C. L. (2006). Overweight in children: definitions and interpretation. Health Education Research, 21(6), 755e760.
Freedman, D. S., Wang, J., Thornton, J. C., Mei, Z., Pierson, R. N., Jr., Dietz, W. H., et al. (2008). Racial/ethnic differences in body fatness among children and adolescents. Obesity (Silver Spring), 16(5), 1105e1111. Fu, W. P., Lee, H. C., Ng, C. J., Tay, Y. K., Kau, C. Y., Seow, C. J., et al. (2003). Screening for childhood obesity: international vs population-specific definitions. Which is more appropriate? International Journal of Obesity and Related Metabolic Disorders, 27(9), 1121e1126. Gortmaker, S. L., Dietz, W. H., Jr., Sobol, A. M., & Wehler, C. A. (1987). Increasing pediatric obesity in the United States. American Journal of Diseases of Children, 141, 535e540. Grimmett, C., Croker, H., Carnell, S., & Wardle, J. (2008). Telling parents their child’s weight status: psychological impact of a weight-screening program. Pediatrics, 122(3), e682e688. Hacking, I. (1988). Symposium papers, comments and an abstract: the sociology of knowledge about child abuse. Noûs, 22, 53e63. Hacking, I. (1999). The social construction of what? Cambridge, Massachusetts: Harvard University Press. Harlan, W. R., Landis, J. R., Flegal, K. M., Davis, C. S., & Miller, M. E. (1988). Secular trends in body mass in the United States, 1960e1980. American Journal of Epidemiology, 128(5), 1065e1074. Hebl, M. R., & Xu, J. (2001). Weighing the care: physicians’ reactions to the size of a patient. International Journal of Obesity, 25, 1246e1252. Herrick, S. S. (1889). Obesity. In A. H. Buck (Ed.), A reference handbook of the medical sciences. Edinburgh: Pentland. Himes, J. H. (2009). Challenges of accurately measuring and using BMI and other indicators of obesity in children. Pediatrics, 124, S3eS22. Himes, J. H., & Dietz, W. H. (1994). Guidelines for overweight in adolescent preventive services: recommendations from an expert committee. The American Journal of Clinical Nutrition, 59, 307e316. House of Representatives Standing Committee on Family and Human Services. (2005). Overseas adoption in Australia. Report on the inquiry into adoption of children from overseas. Canberra, Australia. Hubbard, V. S. (2000). Defining overweight and obesity: what are the issues? American Journal of Clinical Nutrition, 72, 1067e1068. Jutel, A. (2006). The emergence of overweight as a disease entity: Measuring up normality. Social Science & Medicine, 63, 2268e2276. Jutel, A. (2009). Sociology of diagnosis: a preliminary review. Sociology of Health & Illness, 31, 278e299. Jutel, A. (2011). Classification, disease, and diagnosis. Perspectives in Biology and Medicine, 54, 189e205. Kim, S., & Popkin, B. M. (2006). Commentary: understanding the epidemiology of overweight and obesityea real global public health concern. International Journal of Epidemiology, 35(1), 60e67. discussion 81e62. Krebs, N. F., Himes, J. H., Jacobson, D., Nicklas, T. A., Guilday, P., & Styne, D. (2007). Assessment of child and adolescent overweight and obesity. Pediatrics, 120(Suppl4), S193eS228. Kuczmarski, R. J., & Flegal, K. M. (2000). Criteria for definition of overweight in transition: background and recommendations for the United States. American Journal of Clinical Nutrition, 1074e1081. Lobstein, T., & Baur, L. (2005). Policies to prevent childhood obesity in the European Union. European Journal Of Public Health, 15, 576e579. Luke, A. (2009). Ethnicity and the BMI-body fat relationship. British Journal of Nutrition, 102(4), 485e487. McKay, B. (2002). Government’s standard Lumps Hunks, Athletes with Truly obese. Wall Street Journal (online) http://online.wsj.com/article/0,,SB1027369796834182760. djm,00.html. Maclean, L. M., Meyer, M., Walsh, A., Clinton, K., Ashley, L., Donovan, S., et al. (2010). Stigma and BMI screening in schools, or ‘Mom, I hate it when they weigh me’. In J. A. O’Dea, & M. Eriksen (Eds.), Childhood obesity prevention. International research, controversies, and interventions (pp. 17e30). Oxford: Oxford University Press. Manios, Y., & Costarelli, V. (2011). Childhood obesity in the WHO European region. In L. A. Moreno, I. Pigeot, & W. Ahrens (Eds.), Epidemiology of obesity in children and adolescents (pp. 43e68). New York: Springer. Manson, J. E., Willett, W. C., Stampfer, M. J., Colditz, G. A., Hunter, D. J., Hankinson, S. E., et al. (1995). Body weight and mortality among women. The New England Journal of Medicine, 333, 677e685. Mascie-Taylor, C. G. N., & Goto, R. (2007). Human variation and body mass index: a review of the universality of BMI cut-offs, gender and urban-rural differences, and secular changes. Journal of Physiological Anthropology, 26(2), 109e112. Moffat, T. (2010). The “Childhood obesity epidemic”. Medical Anthropology Quarterly, 24(1), 1e21. NHS Choices. (2010). The national child measurement programme. NHS Choices. Nietzsche, F. (1974). The gay science; with a prelude in rhymes and an appendix of songs. New York: Vintage Books. Oliver, J. E. (2006). Fat politics. Oxford: Oxford University Press. Prentice, A. M., & Jebb, S. A. (2001). Beyond body mass index. Obesity Reviews, 2,141e147. Puhl, R. M., & Latner, J. D. (2007). Stigma, Obesity, and the Health of the Nation’s Children. Psychological Bulletin, 133, 557e580. Reilly, J. J., Methven, E., Mcdowell, Z. C., Hacking, B., Alexander, D., Stewart, L., et al. (2003). Health consequences of obesity. Archives of Disease in Childhood, 88, 748e752. Rosenberg, C. E. (2002). The tyranny of diagnosis: specific entities and individual experience. The Milbank Quarterly, 80, 237e260. Ross, B. (2005). Fat or fiction. Weighing the ‘obesity epidemic’. In M. Gard, & J. Wright (Eds.), The obesity epidemic. Science, morality and ideology (pp. 86e 106). London: Routledge.
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15 Rush, E. C., Freitas, I., & Plank, L. D. (2009). Body size, body composition and fat distribution: comparative analysis of European, Maori, Pacific Island and Asian Indian adults. British Journal of Nutrition, 102(4), 632e641. Schwartz, M. B., & Puhl, R. (2003). Childhood obesity: a societal problem to solve. Obesity Reviews, 4, 57e71. Shiwaku, K., Anuurad, E., Enkhmaa, B., Kitajima, K., & Yamane, Y. (2004). Appropriate BMI for Asian populations. The Lancet, 363(9414), 1077. Smalley, K. J., Knerr, A. N., Kendrick, Z. V., Colliver, J. A., & Owen, O. E. (1990). Reassessment of body mass indices. American Journal of Clinical Nutrition, 52. Steinbeck, K. (2003). Clinical practice guidelines for the management of overweight and obesity in children and adolescents. Canberra: National Health & Medical Research Council. Strauss, R. S. (1999). Comparison of measured and self-reported weight and height in a cross-sectional sample of young adolescents. International Journal of Obesity, 23, 904e908. Tajfel, H. (1981). Human groups and social categories. Cambridge: Cambridge University Press. Thaikruea, L., Seetamanotch, W., & Seetmanaotch, S. (2006). Appropriate cut-off level of BMI for screening in Thai adults. Journal of The Medical Association of Thailand, 89(12), 2123e2128. The NHS Information Centre. (2010). Statistics on obesity, physical activity and diet: England, 2010. London. Tiggeman, M., & Wilson-Barrett, E. (1998). Children’s figure ratings: relationship to self-esteem and negative stereotyping. International Journal of Eating Disorders, 23(1), 83e88.
15
Timmermans, S., & Almeling, R. (2009). Objectification, standardization, and commodification in health care: a conceptual readjustment. Social Science & Medicine, 69(1), 21e27. Uretsky, S., Messerli, F. H., Bangalore, S., Champion, A., Cooper-Dehoff, R. M., Zhou, Q., et al. (2007). Obesity paradox in patients with hypertension and coronary artery disease. American Journal of Medicine, 120(10), 863e 870. Voss, L. D., Metcalf, B. S., Jeffery, A. N., & Wilkin, T. J. (2006). IOTF thresholds for overweight and obesity and their relation to metabolic risk in children ( EarlyBird 20). International Journal of Obesity, 30, 606e609. Wheeler, S., & Twist, C. (2010). Methods of assessing body fatness among children: implications for the national child measurement programme. European Physical Education Review, 16(1), 81e93. Widhalm, K., Schönegger, K., Huemer, C., & Auterith, A. (2001). Does the BMI reflect body fat in obese children and adolescents? A study using the TOBEC method. International Journal of Obesity, 25, 279e285. World Health Organization. (1998). Obesity. Preventing and managing the global epidemic. Geneva: World Health Organization. Wulan, S. N., Westerterp, K. R., & Plasqui, G. (2010). Ethnic differences in body composition and the associated metabolic profile: a comparative study between Asians and Caucasians. Maturitas, 65(4), 315e319. Yanovski, S. Z., & Yanovski, J. A. (2011). Obesity prevalence in the United States e up, down, or sideways? New England Journal of Medicine, 364(11), 987e989. Zerubavel, E. (1996). Lumping and splitting: notes on social classification. Sociological Forum, 11, 421e433.
Contents lists available at SciVerse ScienceDirect
Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed
Standards and classification: A perspective on the ‘obesity epidemic’ Stuart G. Nicholls* Department of Epidemiology and Community Medicine, University of Ottawa, 451 Smyth Road, Ottawa, Canada K1H 8M5
a r t i c l e i n f o
a b s t r a c t
Article history: Available online 18 March 2013
In this paper I critique the increasing standardization of obesity. Specifically, I consider two ‘definitional turns’: the way language has been standardized to such an extent that it obscures uncertainty and variation, and the appearance of objectivity through quantification and standardized measurement. These, I suggest, have fostered a simplified picture of obesity, promoting the classification of weight and thereby facilitating the emergence of the ‘obesity epidemic’. These definitional turns fail to acknowledge the distinctions between fat and mass and intraclass variation within weight categories. A consequence of this process of simplification has been the erroneous application of population level information to individuals in a clinical context, with potentially harmful results. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: BMI Standards Epidemiological methods Variation
“ -But let us not forget this either: it is enough to create new names and estimations and probabilities in order to create in the long run new ‘things.’”(Nietzsche, 1974 122, aphorism 58)
Introduction There are now a multitude of studies reporting dramatically increasing levels of obesity over the last twenty to thirty years (Ahrens, Moreno, & Pigeot, 2011; Manios & Costarelli, 2011). These studies not only indicate that the number of obese individuals is increasing, and is as high as 33% in some countries (Flegal, Carroll, Ogden, & Curtin, 2011), but that average weight is also increasing (Finucane et al., 2011). A global analysis of data estimated that in 2008 over 205 million men and 297 million women over the age of 20 were obese (Finucane et al., 2011). Moreover, this includes increasing numbers of ‘morbidly obese’ individuals, skewing the distribution of weights towards the upper extreme (Yanovski & Yanovski, 2011). The increase in obesity would not be so concerning if it were not for the increasing number of adverse health effects associated with it. To date studies have indicated relationships between obesity and a range of conditions including type 2 diabetes mellitus, fatty liver disease, endocrine and orthopaedic disorders and most of the major cardiovascular risk factors (Lobstein & Baur, 2005; Manios & Costarelli, 2011; Reilly et al., 2003). The increasing prevalence of obesity together with the indicated negative health effects have led * Tel.: þ1 613 562 5800x8288. E-mail address: [email protected]. 0277-9536/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.socscimed.2013.03.009
some authors to define the current situation as an ‘obesity epidemic’ (Flegal, 2006). Epidemiological data is often presented to underwrite these claims. Much of the data on which the estimates are based comes from national surveys using the Body Mass Index (BMI). The BMI derives from “Quetelet’s index” (Smalley, Knerr, Kendrick, Colliver, & Owen, 1990) which was developed in the 1800’s to chart the range of heights and weights of army conscripts (Oliver, 2006). In this original conception Quetelet noted a Gaussian (normal) distribution of weight to height ratios within the population, allowing for the description of the statistically average man (Oliver, 2006). Today the BMI calculated as weight (in kg)/height (in metres squared), is used to provide an estimate of body composition. Leaving aside the self-reported nature of much of the available survey data (Manson et al., 1995; Strauss, 1999; Yanovski & Yanovski, 2011), a question remains regarding the interpretation of changes in BMI. What does an increase of one BMI represent? Is there a linear trend with increasing weight, or is it a more complicated relationship such as a normal distribution or U-shaped relationship? Does each BMI increase of one have the same effect size on the specified outcome? Continuous traits, such as weight or BMI, are not amenable to straightforward assessments in the same way as grouped data. Far easier is the assignment of risk to discrete classes or categories. Sexbased risk, for example, has an altogether simpler interpretation; a man may have one risk, a woman another. The creation and use of categories for underweight, normal weight, overweight, and obese have been central to the analysis and presentation of risk estimates, and indeed goes to the very core of data purporting that an obesity epidemic has emerged.
10
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
To date, while there has been much said of the shortcomings of the BMI, there has been little discussion of the way in which the BMI has been applied nor the processes through which the BMI has become the dominant tool on which obesity prevalence and risk have been determined. In this paper, I consider both the standardization and classification of obesity and the roles these have played within the ‘obesity epidemic’. In doing so I engage with the hitherto under explored processes that have reified the BMI as the measure of obesity. Specifically I argue that the language of obesity has been standardized to such an extent that it obscures uncertainty and variation in the assessment of obesity, and the quantification and standardized measurement of obesity has furthered this simplification and has facilitated the perception of an obesity epidemic, obscuring the nuances of the data collected. This, I contend, has important, and potentially harmful, effects when it is misapplied within the clinical context. Classification: making up obese people But what does it mean to talk of classification? Indeed what do we mean by classification? To answer these questions I draw on Bowker and Starr (1999) who define a classification system as: “a set of boxes (metaphorical or literal) into which things can be put to then do some kind of work e bureaucratic or knowledge production.” (Bowker & Starr, 1999, p.10) In order for classification to occur the categories must be consistent, mutually exclusive, and complete e that is, no object from the same population may exist outside of the proposed categories. The role of classification By classifying, we group items based on some characteristics and in some way that we identify them as similar. Equally, we identify others as dissimilar in some other important or functional way. While there may be no single aim to the classification of things - and authors have postulated a range of possibilities (Caplan, 1997; Dupré, 2006; Jutel, 2006, 2011; Zerubavel, 1996) e I will follow Hacking (1988) by emphasizing two main purposes: cognitive and bureaucratic functioning. From a cognitive perspective one role for classification is to simplify the world, to reduce disorder to order (Jutel, 2011). In classifying things, we are able to streamline our perceptions and memories. We are also able to facilitate the production and determination of relationships between objects or actors, to develop explanations or model interactions in order to create predictions (Hacking, 1988; Jutel, 2011). Yet classification can also serve (but is not limited to) functional or bureaucratic purposes. One of these purposes is the ability to count. We may count in order to find out how many there are of something. More likely, as Jutel notes, counting is undertaken to assist in the answering of questions (Jutel, 2011). In the context of healthcare, we may want to know how many people are sick so that we can inform treatment protocols, health service planning, or budgeting (Jutel, 2009). The effect of classification As suggested, the process of classification requires one to group items based on characteristics that identify them as similar. A consequence of this is that the differences between individuals within a category or class e intraclass differences e are downplayed, while the differences between groups e the interclass differences e are overstated (Tajfel, 1981; Zerubavel, 1996). Put
differently, the effect of classifying individuals into groups is that grouped items are perceived as being more similar than items that exist outside of the group. Zerubavel gives the example of boxing weight categories: “[.] we perceive the metrically negligible “distances” between 119-pound (“bantamweight”) and 120-pound (“featherweight”) boxers [.] as greater than those between 120-pound and 125pound (both “featherweight”) boxers.” (Zerubavel, 1996, p.425) To the same extent that weight is manifest as discrete boxing categories, BMI is categorized into discrete classes of underweight, normal weight, overweight or obese. This serves to minimize the differences within these weight categories and introduce perceptions of significant differences between classes. This is despite the fact that the difference between normal weight and overweight is potentially the difference between a BMI of 24.9 and 25.0 while the within group variation of normal weight can be as much as the difference between a BMI of 18.0 and 24.9. To put this more starkly, for an individual who is 1.75 m tall the variation within the “normal” range constitutes 20 kg (Mascie-Taylor & Goto, 2007), while the between group variation between normal weight and overweight can be as little as 0.1 kg. It is, therefore, naïve to treat all those within a category as a homogeneous group when there may be substantial differences within a group, and minimal differences between groups at the boundaries. Grouping does not just classify weight, it classifies people. An often neglected effect of classification is that it can affect how we see those so classified in ways that are more substantial. For example, the classification of people as being of normal or ill health often affects how we respond to them, both in terms of resource allocation (as mentioned above in the pre-emptive role of classification), but also in a personal sense; it may affect whether we wish to associate with the individual or how we do so. This latter sense may be an outcome of stereotypes which, themselves, involve a process of classification: population level generalizations are applied to individuals, removing the complexity of variation and neglecting individual differences within groups (Tajfel, 1981). In this way, the classification of individuals as normal weight or obese may affect our assessment of the individual before us. For instance, studies have found that moralistic terms such as ‘lazy’ and ‘gluttonous’ are used to describe individuals perceived as obese, purely on the basis of their weight (Puhl & Latner, 2007; Schwartz & Puhl, 2003; Tiggeman & Wilson-Barrett, 1998). Ominously, studies indicate that these negative attributions are also held by those in a position to help obese individuals. In a French study of General Practitioners, 30% of those surveyed agreed to some degree that ‘Obese people are lazier and more self-indulgent than normal weight people’, with 28% indicating the same attitude for overweight people (Bocquier et al., 2005). The potential clinical implications are highlighted in a study by Hebl & Xu (2001) in which physicians who reviewed case studies for average weight, overweight, and obese individuals, indicated they would spend less time with obese individuals, would have less patience with them, and had significantly less desire to help the patient. These assessments paper over, if they even acknowledge, the in-group variation as well as illustrating the way in which stereotypes bring with them additional characteristics, in this instance perceptions of moral character. The effect of classifying individuals based on their BMI may serve to propogate such stereotypes by ‘revealing’ an individual to be overweight or obese when they physically appear ‘normal’. In a study of a school based weighing and measurement programme it was noted that while the majority of children did not express concerns, a minority disliked or hated the process, with negative
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
comments from participants indicating the potential for teasing if weight status was known (Grimmett, Croker, Carnell, & Wardle, 2008). Moreover, the child’s measured weight did not always concur with parental perceptions of weight status. In some cases the revelation of weight status was found to affect parental behaviours; there was a statistically significant increase in parental restriction of access to foods in girls who were identified as being overweight. More concerning though are findings from studies of school based screening programmes which indicate that some parents of children of normal weight reported introducing dietrelated activities and seeking medical advice (Chomitz, Collins, Kim, Kramer, & McGowan, 2003; Maclean et al., 2010). The revelation of an individuals’ BMI can in itself create a perceived need for intervention where the child has been assumed to be healthy previously. Consequently, classification may serve multiple purposes and have multiple effects. The effects of this classification can be both intended and unintended. Classifications can have the (intended) effects of restructuring healthcare services that are responsive to actual patterns of need, but also the (unintended) effect of introducing stereotypes and stigmatization. Facilitating classification As suggested, an important aspect of classification is the ability to identify similarities upon which to group items (or in the case of obesity, individuals). However, while there has been a great deal of discussion regarding the BMI levels associated with obesity, there has been little discussion about the establishment of the categories themselves and the implications of these categories and how they have facilitated the perception of obesity epidemic. In the remainder of this paper I consider two distinct ‘definitional turns’ that have served to create a simplified picture of obesity, both of which have revolved around increasing standardization of obesity definitions and have facilitated the classification of weight. I suggest that such definitional turns act through (1) a standardization of language that obscures uncertainty and variation, and (2) the appearance of objectivity through quantification and standardized measurement. The importance of this, I contend, is that these have served to present an overly simplistic picture of obesity, have facilitated the production of the obesity epidemic and, in part, have led to development of inappropriate interventions. This is not to suggest that the number of individuals e adults and children e at increased weight has not grown, there is ample data to support this (Finucane et al., 2011). Rather, my claim is that the current description and debate regarding the obesity epidemic lacks the nuanced discourse that acknowledges the distinct limitations of using standardized categories based on the BMI (and the uncertainties relating to obesity itself) and that this is promulgated by the language and approaches employed. The language of obesity: standards and certainties The first of the aforementioned definitional turns has been the increasingly standardized language around obesity and the way this has gradually minimized the variation in the assessment of weight or assignment of weight classes. This is particularly acute in the case of obesity. Early definitions referred to overweight as “increased bulk of the body, beyond what is sightly and healthy,” (Herrick, 1889 quoted in Jutel, 2006, p.2270), thus employing a subjective analysis of bulk (generally) with regard to what might or might not be deemed as ‘sightly’. In later definitions, overweight was typically defined in terms of body mass or relative weight, while obesity was a condition characterized by excessive body fat (Himes & Dietz, 1994).
11
Overweight, therefore, related to mass, and obesity to fat. Consequently, they represented qualitatively different constructs. Yet weight (or mass) reflects both weight contributed by fat, but also other tissues such as muscle, or bone. As such, fat is a component of mass, but mass will only be partly concerned with fat. This distinction is not only conceptually important, but also has clinical implications. If obesity is defined in terms of fat, this requires a method of assessment that gauges body fat percentage. Methods such as tricep skinfold thickness, whole body plethysmography or dual x-ray absorptiometry may be more appropriate (Finer, 2012). However, these methods not only require more time, they are substantially more expensive than the assessment of weight and calculation of BMI. The underlying problem is that while BMI is (partly) correlated with percentage of body fat (Smalley et al., 1990; Widhalm, Schönegger, Huemer, & Auterith, 2001), it is also correlated with bone density and mass more generally. However, more recent definitions have sought to standardize the constituent elements of overweight and obesity, principally by relating both to the BMI. Overweight has become a term used to describe unhealthy amounts of body fat, not body mass. An example of this occurred in 2007 when an expert committee established by the US Health Resources and Service Administration and the Centers for Disease Control and Prevention (CDC) endorsed the earlier Institute of Medicine (IOM) decision to depart with existing terminology, and change the category of ‘at risk for overweight’ to ‘overweight’ and the replacement of the original category of ‘overweight’ with ‘obese’ (Barlow, 2007). As reported by Himes (2009), an impetus for these changes was the perceived need to “express this concern [about increased weight] by using the term “obese”” (Himes, 2009, p.S8) to replace the previously used term “overweight”. This standardization of terms aligns both overweight and obese onto a single construct; both overweight and obese should refer to differing levels of BMI (Krebs et al., 2007). As such, both overweight and obesity, quantified by BMI, have become ‘measures’ of body fat (Hubbard, 2000). This change is now widespread; the recent UK National Health Service (NHS) report, Statistics on obesity, physical activity and diet : England, 2010, states that “Overweight and obesity are terms that refer to an excess of body fat and they usually relate to increased weight-for-height” (The NHS Information Centre, 2010 8, emphasis added). This definitional slide now means that the common use of the term obese (often in relation to BMI levels) is, in effect, an extension of the overweight category but demarcating those at the extreme end of the scale. As such, there has been a move from different but associated concepts e mass and fat e to a situation whereby both terms operate on a single construct. This obfuscation of the distinction has important methodological and interpretive implications. Firstly, how one defines obesity operationalizes how one measures or evaluates the level of overweight and obesity. If one defines obesity as the bodily composition of fat, but overweight as mass, then different measures may be used. Aligning the terms simplifies the measurement process. The importance of this difference is illustrated by two papers, both of which evaluated the same three waves of the US National Health and Nutrition Examination Survey (NHANES), but which came to different conclusions regarding obesity levels and trends (Flegal, 1993). The first paper, by a team led by Steven Gortmaker (Gortmaker, Dietz, Sobol, & Wehler, 1987) assessed tricep skinfold thickness in order to assess percentage of body fat. In contrast, Harlan, Landis, Flegal, Davis, & Miller (1988) used BMI. The studies came to diametrically opposed conclusions regarding trends in obesity for children. Moreover, in a comparison of prevalence estimates of overweight, the estimates produced using the BMI were
12
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
consistently higher, in the region of 10% higher, than those from measurements of tricep skinfold thickness (see Flegal, 1993 310). A simplified standard, in which BMI refers to both overweight and obesity, not only facilitates the removal of conceptual differences e both overweight and obesity now refer to the same concept and so can be seen on a single metric e but in doing so it demarcates overweight as the stage prior to obesity. This turns overweight into what Rosenberg has termed a ‘proto-disease’, that is into a pre-disease state, that: “Once articulated and disseminated in practice and the culture generally [.] become emotional and clinical realities, occupying a position somewhere between warning signal and pathology.” (Rosenberg, 2002, p.254) Hence overweight has become the pre-disease state for obesity meaning that overweight is now a legitimate point of intervention as it is a precursor to the unhealthy disease state of obesity. Moreover, the complexity of using terms that refer to different constituents is removed with the underlying elements aggregated. This in turn facilitates the classification of a single scalar concept into a limited number of categories as opposed to the more complex and uncertain situation of trying to assimilate two distinct nominal categories. Standardization and the appearance of objectivity A second notable trend, as alluded to earlier, has been the migration from more qualitative descriptions of obesity to descriptions defined by measurement and the subsequent standardization of these measures (Jutel, 2006). Such a trend, I suggest, has acted as a catalyst for the generation of the obesity epidemic by creating an appearance of objectivity and allowing international comparisons. This not only contributes to the ability to construct an ‘obesity epidemic’ but also an ‘obesity pandemic’. This quantification and standardization is most clearly seen through the almost universal adoption of the BMI as a measure of obesity. Indeed, the application of the BMI as an obesity measure is itself one of simplification. While several studies have suggested that BMI is correlated with percentage of body fat (Smalley et al., 1990; Widhalm et al., 2001), it is also correlated with bone density and mass more generally (Colls & Evans, 2010). A major driver in the adoption has been the relative simplicity with which it can be applied (Evans, 2009; Widhalm et al., 2001). Since the general adoption of the BMI, there has been a gradual move towards a universal standard of categories for overweight and obesity (Ebbeling & Ludwig, 2008). Thus, prior to 1999, many experts considered a BMI of 29 kg/m2 to be overweight. Later this was reduced to 27 kg/m2 and today a BMI above 25 kg/m2 is the standard used to indicate overweight (in adults) (Ross, 2005). This gradual reduction and standardization of categories simplifies and increases the appearance of certainty. The setting of a universal standard not only facilitates comparisons between populations but also presents an air of objectivity in the way it papers over potential differences between groups or populations. Yet even small changes can have a substantial impact on estimates of prevalence levels. McKay (2002), for example, argues that the 1998 reduction in the BMI threshold for overweight to 25 kg/m2 “made an extra 30 million Americans ‘overweight’ overnight.” Furthermore, studies suggest that different ethnic populations may have the same percentage of body fat at different levels of BMI or, put differently different levels of fat at the same BMI (Freedman et al., 2008; Luke, 2009; Rush, Freitas, & Plank, 2009; Wulan, Westerterp, & Plasqui, 2010). A universal standard will likely underestimate the potential health effects in some populations with higher percentage body fat to BMI ratios (for example Asian
populations) and overestimate the effect in other populations with lower percentage body fat but higher muscle mass (such as Pacific Island populations) (Luke, 2009). Indeed, variance may already be creeping into assessments as we have begun to see the emergence of what Hacking refers to as a ‘looping effect’ in which the actual classification feeds back to change the category due to the way in which those so classified perceive themselves (Hacking, 1999). The production of population specific BMI cutoffs is an example of this, with specific models being developed for Asian and Thai populations (Fu et al., 2003; Shiwaku, Anuurad, Enkhmaa, Kitajima, & Yamane, 2004; Thaikruea, Seetamanotch, & Seetmanaotch, 2006). Consequently, we are seeing what Hacking describes as “the rebellions of the sorted” (Hacking, 1999 131). This, as noted by Timmermans and Almeling (2009), is an observed process of standardization in which a level of disorder re-emerges often in response to a proliferation of standardization (Timmermans and Almeling, 2009). As such, the appearance of population specific cutoffs indicates the non-standard nature of differing populations and fallacy of attempting to produce international standards that are inappropriate to the context to which they are being applied. Implementing standards: a cautionary note Identifying the process of standardization is important, for alongside the increasing standardization there has been a creep of overly simplified usage of obesity classification and the not un-problematic application of population level standards to individuals. Again, this move has been gradual and the current individual application of the BMI differs substantially to its origins as ‘Quetelet’s index’ where the purpose was to determine population averages and ranges, rather than using it for a clinical purpose at the individual level. Definitions of overweight using the BMI, provides only a crude population-level measure, and while valuable for its convenience and simplicity in public health surveillance, screening, and similar purposes it lacks the sensitivity or specificity to be used as a diagnostic tool. Of course an appeal of the BMI is that it is fast, convenient, and noninvasive compared to the time consuming measures mentioned earlier e features that make it both eminently practical for use in large prevalence studies (Moffat, 2010), but also appealing to employment within the clinic as it requires no specialist equipment. This should not, however, blind those who use it to its shortcomings in a clinical setting. As a clinical measure, therefore, it lacks utility; primarily because individuals with the same BMI may not have the same health risks (World Health Organization, 1998) and so it “[does] not necessarily identify physiological states per se” (Flegal, Tabak, & Ogden, 2006). Thus, some have cautioned that the BMI should not be used diagnostically (Kuczmarski & Flegal, 2000). Evidence suggests, for example, that individuals at the lower end of the ‘overweight’ category do not necessarily have worse overall health outcomes (e.g. Calle, Rodriguez, Walker-Thurmond, & Thun, 2003; Carnethon et al., 2012; Uretsky et al., 2007), and may be appear to be ‘metabolically healthy’ when compared to other ‘normal weight’ individuals (Carnethon et al., 2012). This not only problematizes the imposition of overweight as a precursor to obesity (and associated ill health) but also suggests that lifestyle factors, such as diet, physical activity levels, or environmental exposures may, therefore, be more relevant to clinical care as opposed to a focus on weight per se. Despite these cautions, growth curve charts for paediatric overweight and obesity have now been used in the clinic (Voss, Metcalf, Jeffery, & Wilkin, 2006). Children in England now have their BMI assessed as part of the National Child Measurement Programme. While the scheme aims to provide a national picture, and so collect population prevalence, it also uses the child’s BMI to
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
provide individual assessments. Programme materials tell parents that: “You can also benefit, as you can receive your child’s individual results. This will help you to know if your child is in the healthy weight range. If your child is overweight, further support is available from your local NHS.” (NHS Choices, 2010) The invocation of the National Health Service (NHS) implies that the overweight status is inherently problematic. Indeed, admissions from the Department of Health that children who are obese and overweight are to be proactively followed up indicates its use as a clinical tool (Wheeler & Twist, 2010). Yet BMI category, in isolation of all other information, represents an extremely simplistic assessment; there is no consideration of the general health of the child. The programme itself runs counter to the advice received from its own National Screening Committee, who advised that the screening programme should not go ahead as it could not guarantee doing more good than harm (Colls & Evans, 2010). However, this is not an isolated case. The Australian National Health and Medical Research Council (NMHRC), in their 2003 document Clinical Practice Guidelines for the Management of Overweight and Obesity in Children and Adolescents, explicitly state that: “BMI-for-age percentile charts should be used in clinical practice [.] a BMI above the 85th percentile being indicative of overweight and a BMI above the 95th percentile being indicative of obesity.” (Steinbeck, 2003, p.14, emphasis added) BMI has also been introduced into other problematic areas, such as recruitment for the military and has been reported as a trigger to initiate retraining or as grounds for dismissal if a certain BMI threshold is crossed (Prentice & Jebb, 2001). In a further example from Australia, Queensland has introduced BMI thresholds as part of their assessment criteria for adoptive parents (House of Representatives Standing Committee on Family and Human Services, 2005). Both of these examples are indicative of the way in which the BMI has become institutionalized and implemented without acknowledgement of the uncertainty that is inherent in the simple height-to-weight metric. Yet to apply population characteristics within an individual clinical encounter, solely on the basis of individual BMI, obscures the variance contained within population estimates and inappropriately simplifies the relationship between health effects and overweight or obesity. As Ross (2005) has noted, population comparisons that indicate a 2e3 times greater relative risk of dying in a population of individuals who are obese compared to a population who are not: “[.] means that in a group of people with a BMI 30 kg/m2 it is probable that 2e3 people will die for every 1 person who dies in a comparable group of people who have a BMI 29.9 kg/m2. It does not mean that any given person with a BMI 30 kg/m2 is 2e3 times more likely to die than an individual with a BMI 29.9 kg/m2." (Ross, 2005, p.101) Applying population level associations to individuals has the potential to lead one to commit the “ecological fallacy”. The ecological fallacy refers to the case where relationships found in aggregate level data are assumed to apply at the individual level. For example: “[.] average weight gain at the population level does not necessarily equate with weight gain at the individual level. The actual weight change of a population comprises a wide range of distributions. There may be a portion of the population who lose
13
or do not change weight over time that would pull the mean weight change of the population down, which would make rather significant weight gains in some other groups of the population look trivial; ignoring these dynamics within a population may mask true changes that are happening among individuals.” (Kim & Popkin, 2006, p.62) It need not, therefore, be the case that just because an affluent country has a high prevalence of obesity then affluent individuals will also be obese. Even if data is collected at the individual level, associations are based on group averages and it is an error to assume that these averages of the group are applicable to specific individuals within the group. This is because the assumption ignores the intraclass variation discussed earlier, and represents what Conrad refers to as the individualization of social problems (Conrad, 2007) and what Tajfel (1981) refers to as “personalization”: “If there is to be an explanation in terms of the characteristics of a group, these must be characteristics which are relevant to the situation and common to the whole group, with a corresponding neglect of individual differences between the members of a group.” (Tajfel, 1981, p.138) As such, one cannot assume that because certain health conditions are associated with obesity at a population level, and that the individual in question is obese, that those associations are applicable to the individual. Conclusion The BMI is now the dominant tool on which data purporting the obesity epidemic is based. While there has been discussion regarding the limits of the BMI, and the setting of thresholds for categories, to date there has been a paucity of debate regarding the limitations imposed by neither the categorization itself nor the ongoing standardization of these categories. In this paper, I have addressed this deficit, and have suggested that the act of classifying individuals and populations as overweight and obese has been central to the emergence of the ‘obesity epidemic’. This classification, I have argued, has been accelerated by increasingly standardized definitions of overweight and obesity. This has been characterized by increasingly quantified standards, which carry an air of certainty and objectivity. In doing so I have argued that such trends serve to hide the complexity of overweight and obesity classifications, and have led to the overly simplistic implementation of the BMI. The almost ubiquitous use of BMI as the basis of obesity definitions has facilitated the categorization of weight and the classification of populations and individuals as overweight or obese. This assessment has important implications because of the way it minimizes the intraclass differences and over-emphasizes between class differences. A serious concern raised by this is the failure to acknowledge the effects of such simplification for individuals. This not only relates to the ascription of negative stereotypes associated with overweight and obesity, but also in terms of treatment interventions that may have differential effects if the individuals’ specific circumstances are not considered. An alternative and more transparent approach would be to seek to re-introduce the uncertainty within studies. This could be achieved, in part, through the use of multiple assessments of adiposity via different methods allowing for the comparison of methods and the implications these have for assessment. In addition, the analysis of data on a continuous scale, in addition to the current categorization, would not only facilitate the production of confidence intervals around estimates of adiposity, but would allow for a greater exploration of the relationship between increasing levels of
14
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15
adiposity and negative health implications. This approach has recently been taken in a study synthesizing data from 960 country years of data and 9.1 million participants (Finucane et al., 2011). In their analyses, the authors specifically incorporated non-linear associations with age, time trends and stratification by sex. Moreover, their approach allows for the production of uncertainty levels, with greater indicators of uncertainty when data is sparse or inconsistent, producing a more nuanced assessment of the results. While such an approach is undoubtedly time consuming and costly, the examination of non-linear associations, and inclusion of variation, could be adopted by prospective studies as a way of illustrating the uncertainty around estimates of obesity levels and health effects. Acknowledgements This work has benefited from participation within the Identification and prevention of Dietary- and lifestyle-induced health EFfects In Children and infantS (IDEFICS) study. The IDEFICS study is funded through the European Commission Sixth Framework Programme. Area 2: Epidemiology of food-related diseases and allergies. Topic 5.4.2.1.: Influence of diet and lifestyle on children’s health (Integrated Project). Contract n 016181 (FOOD). I would also like to thank Dr Garrath Williams and Dr Kristin Voigt for comments on previous versions of this paper, although I take full responsibility for the arguments put forward. References Ahrens, W., Moreno, L. A., & Pigeot, I. (2011). Childhood obesity: prevalence worldwide e synthesis part I. In L. A. Moreno, I. Pigeot, & W. Ahrens (Eds.), Epidemiology of obesity in children and adolescents (pp. 219e235). New York: Springer. Barlow, S. E. (2007). Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity: summary report. Pediatrics, 120(Suppl 4), S164eS192. Bocquier, A., Verger, P., Basdevant, A., Andreotti, G., Baretge, J., Villani, P., et al. (2005). Overweight and obesity: knowledge, attitudes and practices of general practitioners in France. Obesity Research, 13(4), 787e795. Bowker, G. C., & Starr, S. L. (1999). Sorting things out. Classification and its consequences. Cambridge, Massachusetts: MIT Press. Calle, E. E., Rodriguez, C., Walker-Thurmond, K., & Thun, M. J. (2003). Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. Adults. New England Journal of Medicine, 348, 1625e1638. Caplan, A. L. (1997). The concepts of health, illness, and disease. In R. M. Veatch (Ed.), Medical ethics). Boston: Jones and Bartlett. Carnethon, M. R., De Chavez, P. J. D., Biggs, M. L., Lewis, C. E., Pankow, J. P., Bertoni, A. G., et al. (2012). Association of weight status with mortality in adults with incident diabetes. JAMA, 308(6), 581e590. Chomitz, V. R., Collins, J., Kim, J., Kramer, E., & McGowan, R. (2003). Promoting healthy weight among elementary school children via a health report card approach. Archives in Pediatric and Adolescent Medicine, 157, 765e772. Colls, R., & Evans, B. (2010). Re-thinking ‘the obesity problem’. Geography, 95(2), 99e105. Conrad, P. (2007). The medicalization of society. on the transformation of human conditions into treatable disorders. Baltimore: The Johns Hopkins University Press. Dupré, J. (2006). Scientific classification. Theory, Culture & Society, 23, 30e32. Ebbeling, C. B., & Ludwig, D. S. (2008). Tracking pediatric obesity. An index of uncertainty? Journal of the American Medical Association, 299, 2442e2443. Evans, B. (2009). Measuring fatness, governing bodies: the spatialities of the body mass index (BMI) in anti-obesity politics. Antipode, 41, 1051e1083. Finer, N. (2012). Better measures of fat mass e beyond BMI. Clinical Obesity, 65. Finucane, M. M., Stevens, G. A., Cowan, M. J., Danael, G., Lin, J. K., Paciorek, C. J., et al. (2011). National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9.1 million participants. The Lancet, 377, 557e567. Flegal, K. M. (1993). Defining obesity in children and adolescents: epidemiologic approaches. Critical Reviews In Food Science and Nutrition, 33(4e5), 307e312. Flegal, K. M. (2006). Commentary: the epidemic of obesity d what’s in a name? International Journal of Obesity and Related Metabolic Disorders, 72e74. Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2011). Prevalence and trends in obesity among US adults, 1999e2008. Journal of the American Medical Association, 303(3), 235e241. Flegal, K. M., Tabak, C. J., & Ogden, C. L. (2006). Overweight in children: definitions and interpretation. Health Education Research, 21(6), 755e760.
Freedman, D. S., Wang, J., Thornton, J. C., Mei, Z., Pierson, R. N., Jr., Dietz, W. H., et al. (2008). Racial/ethnic differences in body fatness among children and adolescents. Obesity (Silver Spring), 16(5), 1105e1111. Fu, W. P., Lee, H. C., Ng, C. J., Tay, Y. K., Kau, C. Y., Seow, C. J., et al. (2003). Screening for childhood obesity: international vs population-specific definitions. Which is more appropriate? International Journal of Obesity and Related Metabolic Disorders, 27(9), 1121e1126. Gortmaker, S. L., Dietz, W. H., Jr., Sobol, A. M., & Wehler, C. A. (1987). Increasing pediatric obesity in the United States. American Journal of Diseases of Children, 141, 535e540. Grimmett, C., Croker, H., Carnell, S., & Wardle, J. (2008). Telling parents their child’s weight status: psychological impact of a weight-screening program. Pediatrics, 122(3), e682e688. Hacking, I. (1988). Symposium papers, comments and an abstract: the sociology of knowledge about child abuse. Noûs, 22, 53e63. Hacking, I. (1999). The social construction of what? Cambridge, Massachusetts: Harvard University Press. Harlan, W. R., Landis, J. R., Flegal, K. M., Davis, C. S., & Miller, M. E. (1988). Secular trends in body mass in the United States, 1960e1980. American Journal of Epidemiology, 128(5), 1065e1074. Hebl, M. R., & Xu, J. (2001). Weighing the care: physicians’ reactions to the size of a patient. International Journal of Obesity, 25, 1246e1252. Herrick, S. S. (1889). Obesity. In A. H. Buck (Ed.), A reference handbook of the medical sciences. Edinburgh: Pentland. Himes, J. H. (2009). Challenges of accurately measuring and using BMI and other indicators of obesity in children. Pediatrics, 124, S3eS22. Himes, J. H., & Dietz, W. H. (1994). Guidelines for overweight in adolescent preventive services: recommendations from an expert committee. The American Journal of Clinical Nutrition, 59, 307e316. House of Representatives Standing Committee on Family and Human Services. (2005). Overseas adoption in Australia. Report on the inquiry into adoption of children from overseas. Canberra, Australia. Hubbard, V. S. (2000). Defining overweight and obesity: what are the issues? American Journal of Clinical Nutrition, 72, 1067e1068. Jutel, A. (2006). The emergence of overweight as a disease entity: Measuring up normality. Social Science & Medicine, 63, 2268e2276. Jutel, A. (2009). Sociology of diagnosis: a preliminary review. Sociology of Health & Illness, 31, 278e299. Jutel, A. (2011). Classification, disease, and diagnosis. Perspectives in Biology and Medicine, 54, 189e205. Kim, S., & Popkin, B. M. (2006). Commentary: understanding the epidemiology of overweight and obesityea real global public health concern. International Journal of Epidemiology, 35(1), 60e67. discussion 81e62. Krebs, N. F., Himes, J. H., Jacobson, D., Nicklas, T. A., Guilday, P., & Styne, D. (2007). Assessment of child and adolescent overweight and obesity. Pediatrics, 120(Suppl4), S193eS228. Kuczmarski, R. J., & Flegal, K. M. (2000). Criteria for definition of overweight in transition: background and recommendations for the United States. American Journal of Clinical Nutrition, 1074e1081. Lobstein, T., & Baur, L. (2005). Policies to prevent childhood obesity in the European Union. European Journal Of Public Health, 15, 576e579. Luke, A. (2009). Ethnicity and the BMI-body fat relationship. British Journal of Nutrition, 102(4), 485e487. McKay, B. (2002). Government’s standard Lumps Hunks, Athletes with Truly obese. Wall Street Journal (online) http://online.wsj.com/article/0,,SB1027369796834182760. djm,00.html. Maclean, L. M., Meyer, M., Walsh, A., Clinton, K., Ashley, L., Donovan, S., et al. (2010). Stigma and BMI screening in schools, or ‘Mom, I hate it when they weigh me’. In J. A. O’Dea, & M. Eriksen (Eds.), Childhood obesity prevention. International research, controversies, and interventions (pp. 17e30). Oxford: Oxford University Press. Manios, Y., & Costarelli, V. (2011). Childhood obesity in the WHO European region. In L. A. Moreno, I. Pigeot, & W. Ahrens (Eds.), Epidemiology of obesity in children and adolescents (pp. 43e68). New York: Springer. Manson, J. E., Willett, W. C., Stampfer, M. J., Colditz, G. A., Hunter, D. J., Hankinson, S. E., et al. (1995). Body weight and mortality among women. The New England Journal of Medicine, 333, 677e685. Mascie-Taylor, C. G. N., & Goto, R. (2007). Human variation and body mass index: a review of the universality of BMI cut-offs, gender and urban-rural differences, and secular changes. Journal of Physiological Anthropology, 26(2), 109e112. Moffat, T. (2010). The “Childhood obesity epidemic”. Medical Anthropology Quarterly, 24(1), 1e21. NHS Choices. (2010). The national child measurement programme. NHS Choices. Nietzsche, F. (1974). The gay science; with a prelude in rhymes and an appendix of songs. New York: Vintage Books. Oliver, J. E. (2006). Fat politics. Oxford: Oxford University Press. Prentice, A. M., & Jebb, S. A. (2001). Beyond body mass index. Obesity Reviews, 2,141e147. Puhl, R. M., & Latner, J. D. (2007). Stigma, Obesity, and the Health of the Nation’s Children. Psychological Bulletin, 133, 557e580. Reilly, J. J., Methven, E., Mcdowell, Z. C., Hacking, B., Alexander, D., Stewart, L., et al. (2003). Health consequences of obesity. Archives of Disease in Childhood, 88, 748e752. Rosenberg, C. E. (2002). The tyranny of diagnosis: specific entities and individual experience. The Milbank Quarterly, 80, 237e260. Ross, B. (2005). Fat or fiction. Weighing the ‘obesity epidemic’. In M. Gard, & J. Wright (Eds.), The obesity epidemic. Science, morality and ideology (pp. 86e 106). London: Routledge.
S.G. Nicholls / Social Science & Medicine 87 (2013) 9e15 Rush, E. C., Freitas, I., & Plank, L. D. (2009). Body size, body composition and fat distribution: comparative analysis of European, Maori, Pacific Island and Asian Indian adults. British Journal of Nutrition, 102(4), 632e641. Schwartz, M. B., & Puhl, R. (2003). Childhood obesity: a societal problem to solve. Obesity Reviews, 4, 57e71. Shiwaku, K., Anuurad, E., Enkhmaa, B., Kitajima, K., & Yamane, Y. (2004). Appropriate BMI for Asian populations. The Lancet, 363(9414), 1077. Smalley, K. J., Knerr, A. N., Kendrick, Z. V., Colliver, J. A., & Owen, O. E. (1990). Reassessment of body mass indices. American Journal of Clinical Nutrition, 52. Steinbeck, K. (2003). Clinical practice guidelines for the management of overweight and obesity in children and adolescents. Canberra: National Health & Medical Research Council. Strauss, R. S. (1999). Comparison of measured and self-reported weight and height in a cross-sectional sample of young adolescents. International Journal of Obesity, 23, 904e908. Tajfel, H. (1981). Human groups and social categories. Cambridge: Cambridge University Press. Thaikruea, L., Seetamanotch, W., & Seetmanaotch, S. (2006). Appropriate cut-off level of BMI for screening in Thai adults. Journal of The Medical Association of Thailand, 89(12), 2123e2128. The NHS Information Centre. (2010). Statistics on obesity, physical activity and diet: England, 2010. London. Tiggeman, M., & Wilson-Barrett, E. (1998). Children’s figure ratings: relationship to self-esteem and negative stereotyping. International Journal of Eating Disorders, 23(1), 83e88.
15
Timmermans, S., & Almeling, R. (2009). Objectification, standardization, and commodification in health care: a conceptual readjustment. Social Science & Medicine, 69(1), 21e27. Uretsky, S., Messerli, F. H., Bangalore, S., Champion, A., Cooper-Dehoff, R. M., Zhou, Q., et al. (2007). Obesity paradox in patients with hypertension and coronary artery disease. American Journal of Medicine, 120(10), 863e 870. Voss, L. D., Metcalf, B. S., Jeffery, A. N., & Wilkin, T. J. (2006). IOTF thresholds for overweight and obesity and their relation to metabolic risk in children ( EarlyBird 20). International Journal of Obesity, 30, 606e609. Wheeler, S., & Twist, C. (2010). Methods of assessing body fatness among children: implications for the national child measurement programme. European Physical Education Review, 16(1), 81e93. Widhalm, K., Schönegger, K., Huemer, C., & Auterith, A. (2001). Does the BMI reflect body fat in obese children and adolescents? A study using the TOBEC method. International Journal of Obesity, 25, 279e285. World Health Organization. (1998). Obesity. Preventing and managing the global epidemic. Geneva: World Health Organization. Wulan, S. N., Westerterp, K. R., & Plasqui, G. (2010). Ethnic differences in body composition and the associated metabolic profile: a comparative study between Asians and Caucasians. Maturitas, 65(4), 315e319. Yanovski, S. Z., & Yanovski, J. A. (2011). Obesity prevalence in the United States e up, down, or sideways? New England Journal of Medicine, 364(11), 987e989. Zerubavel, E. (1996). Lumping and splitting: notes on social classification. Sociological Forum, 11, 421e433.