May 2013

Obesity and the Pension Actuary

By Sam Gutterman

Background
Today's abundant calorie-rich food often overwhelms the body's weight regulatory system to such an extent that, with many individuals unable to regulate all of this input, a massive societal (especially, but not exclusively in the United States) weight gain has occurred over the past 35 years in all population segments. The current era of obesity and inactivity is threatening the substantial progress made in the modern era in postponing illness and death.

The fundamental causes of the increase in obesity are rooted in the nature of current Western culture with its wealth, incentives to live in an increasingly sedentary manner and to consume a high-fat, energy-dense diet, while spending an ever-smaller share of income on food. While historically people were mostly occupied by simply obtaining food, they now think more about how to enjoy it.

Although it has been estimated that genetics is the source of 40 to 50 percent of obesity prevalence, the rapid upturn in weight during the last third of the twentieth century points to a major role being played by behavioral shifts. These can be influenced by ineffective personal control mechanisms and environmental, lifestyle and nutritional factors (e.g., size of servings, inadequate amount of fruits and vegetables, excessive fructose sugar-flavored drinks, lack of physical activity and dieting failures). Possibly interacting with genetic susceptibility, these behavioral effects contribute not only directly to an individual's weight, but can themselves constitute risk factors contributing to chronic disease.

The current prevalence of obesity (for adults, popularly defined as a body mass index (BMI) greater than 30; the “overweight” category is between 25 and 30 BMI) of about 35 percent of the adult population in the United States, seen in Figure 1, is a significant concern to society and to practicing actuaries. The trend in childhood obesity has also been adverse, with potential serious long-term effects.

Not only has average weight increased over the last few decades, but the percentage in excess of given weight levels have shifted the prevalence distribution. Although the overall obesity level has been stabilizing but is still increasing somewhat, in the 2000s, the population of those morbidly obese (class II and greater—a BMI of at least 40) category continues to get worse. This BMI prevalence curve shift (highlighted in Figure 2, especially in the very obese) is of significant concern, because the morbidly obese experience extremely high mortality and health care costs. This is unfortunately an area in which Americans stand out.

psn-2013-iss80-gutterman-fig1 
psn-2013-iss80-gutterman-fig2

 

Mortality
Obesity directly contributes to conditions such as diabetes and hypertension, while to others such as stroke and heart disease, indirectly. Diseases and conditions often associated with obesity include:

  • Type 2 diabetes. Associated with a doubling of the risk of heart disease and stroke as well as a leading cause of blindness, kidney failure and non-traumatic amputations. Almost 90 percent of diabetes sufferers are overweight or obese.
  • Cardiovascular and heart disease. Associated with numerous cardiac complications such as coronary heart disease, congestive heart failure and sudden death.
  • Cardiovascular risk factors, including hypertension and adverse cholesterol levels.
  • Cancers. The American Cancer Society has indicated that overweight and obesity contribute to between 14 percent and 20 percent of all cancer-related deaths.
  • Kidney and liver diseases.
  • Psychological disorders, including depression, anxiety, stress, bipolar disorder, schizophrenia, sleep apnea, sex disorders, weight stigma and dementia.
  • Others, including musculoskeletal problems, arthritis and asthma.

Although these associations generally apply to all categories of the obese, they particularly apply to the extremely obese. Compared with those in the ideal BMI range (usually assigned to those with BMIs between 22.5 and 24.9), extremely obese adults have, for example, seven times the risk of diabetes, six times the risk of hypertension, four times the risk of arthritis and three times the risk of asthma.

Nevertheless, it can be difficult to attribute premature mortality and health-related costs directly to obesity; in 2002 a quarter of the obese had six or more adverse medical conditions. This difficulty in attribution, in part as a result of the complex nature of and inter-relationships among health processes, exacerbated by the lag between cause and effect, all affect the findings of any study of the sources of mortality and health care costs.

Many studies have found either J-curve or U-curve relationships between BMI and mortality rates relative to those ratios for those in the ideal BMI group. These curves are hazard rates (the ratios of mortality rates to a benchmark rate, in this case that for those in the ideal BMI range), with higher relative mortality rates for both the underweight and the obese, the difference being the degree of additional relative mortality rates for those in the highest obese category. This relationship between BMI and mortality is usually more evident in those studies with a long observation period, reflecting the lag between the obesity condition and subsequent mortality, especially with respect to adverse cardiovascular conditions. For example, the well-known Framingham Heart Study has found that being overweight can be an independent, long-term predictor of cardiovascular disease, associated with large decreases in life expectancy and increases in premature mortality.

In contrast, other studies have reported what has been referred to as an obesity paradox, in which mortality rates for those overweight and mildly obese are lower than for those in the so-called ideal BMI category. Note that some of these studies have involved relatively short follow-up periods, considering the long lag between excess adiposity and resulting adverse mortality. For example, a wide-ranging meta-analysis of many research studies conducted by Flegal et.al. (2013) indicated that those overweight have experienced better mortality than those in the ideal BMI category, with no adverse experience observed until the higher levels of obesity.

Of particular concern to pension actuaries is the effect of obesity on mortality of those over age 65, especially in light of the seemingly continuous favorable trends in overall mortality that we have experienced over the last century that are embedded and incorporated in many of our mortality tables. Certainly, because of the lengthy lags between certain behaviors and consequential mortality, any negative effect on mortality levels from the increase in the prevalence of obesity has been overwhelmed by favorable changes that have resulted from the dramatic reductions in smoking and enhanced control of other cardiovascular risk factors over the last few decades.

On an overall basis, little, if any, additional mortality for the elderly (say, over age 75) has been observed in population studies for those overweight or at moderate obesity levels. But several observations are necessary to put this in the proper context.

  • The metric for obesity. Most studies of obesity have been based on BMI, a function of weight and height, a surrogate or indirect estimate of adiposity, used in large part because of its simplicity in measurement. However, such an aggregate measure may underestimate the adverse effects of excess adiposity (fat) tissues in older adults compared to that of younger adults. Due to the loss in muscle and bone mass that can be the result of inactivity, illness, or simply aging, the interpretation of the effect of measures based on weight and height can be problematic. Thus BMI is a less useful indicator of adiposity's effects, particularly for those at advanced ages, whose fat tends to shift from peripheral to central body sites, with a resultant increase in their waist-to-hip ratio, but with no change in BMI. As a result, waist circumference or an abdominal measure has been shown in some studies to be a better at-risk measure at these ages. Unlike younger ages, a BMI somewhat greater than 25 may be a more appropriate upper range cut-off point for an ideal class for older adults.
  • When measured. The cumulative exposure to additional weight and adiposity tissues has a greater influence on health and contributes to higher mortality rates than such exposure at a particular point in time. An example illustrates a reason for low observed correlation between current weight and current additional deaths where an insufficient follow-up period is provided for. An individual might be obese in her 40s with an onset of diabetes in her 50s, which in turn might lead to a myocardial infarction in her 60s, heart failure and weight loss at age 70 with death occurring a year later. In this case, an epidemiological study of the relationship between BMI on mortality that only measures BMI at age 70 after the weight loss occurs would be unable to identify the original cause of the premature death. So, studies with an insufficiently long follow-up experience period to properly recognize the cumulative effect of the obesity surge that began in the 1970s may not properly capture the effect of this surge.

The long lag between being obese and eventual death can lead to misleading results where a short follow-up period between body measurement and study period is provided for, as many interim adverse chronic conditions and diseases can lead to weight loss at older ages. Studies with longer follow-up periods and exclusions of pre-existing conditions (e.g., those who previously smoked or had cancer) can provide a better perspective on the relationship between obesity and mortality in the elderly than if measured concurrently. In any event, to obtain more useful results, it is desirable to segment experience by major age categories.

Studies of mortality of the aged have shown inconsistent results. For example, two meta-analyses focusing on those older than 65 showed somewhat different results—Janssen and Marks (2007), incorporating the results of 26 independent studies, found an average hazard rate (compared with those in the ideal BMI range) of 1.0 for those overweight (not obese), while for those of moderate obesity it was 1.10. In contrast Heiat (2001), incorporating 13 studies with follow-up periods of between three and 23 years, indicated a positive relationship between BMI and mortality. However, the latter meta-analysis indicated an optimal BMI of at least 27, rather than the 22.5-25.0 range typically found for those at younger ages.

The Cardiovascular Health Study, as described in the former meta-analysis, that provided a follow-up period of up to nine years, indicated a mortality hazard risk for those overweight of 11 percent less than those in the ideal BMI range. Nevertheless, a significantly higher rate of diabetes was noted in recent years at ages greater than 65. Only two studies (the Framingham Heart Study and the American Cancer Society Cancer Prevention Study) included in the second meta-analysis showed a positive relation between BMI and mortality for the obese, with its other studies showing either no or a negative relationship between mortality and BMI. In these studies, there was a U-shaped BMI mortality curve for cardiovascular disease, with BMI at the lowest mortality level not reached until it reached 31 or 32, with a less steep upward slope at higher BMI values than at younger ages.

Why such counter-intuitive results of so many studies of the aged? As indicated above, some of the results may arise from measuring BMI, a wrong indicator of adverse adiposity at those ages, while others may not contain adequate follow-up periods. In addition, some analysts have hypothesized that: (1) a higher percent of those previously obese had already died by that time, that is, the "selective survivor effect," resulting in a flattening of the weight/mortality curve, with those of higher risk having died at earlier ages, (2) those of greater weight left were relatively stronger and more healthy, (3) current excess body fat becoming less important in the aging process and may provide protective reserves against adverse health conditions, including frailty, and (4) as people age they incur far more frequent multiple health hazards that might mask underlying relationships.

It seems clear that body mass and skeletal structure due to the aging process may result in a higher ideal BMI level. And due to the lags involved, it is possible that the surge in growth of weight hasn't had time to see its cumulative effect fully felt in reported mortality studies.

Physical activity and fitness is also important for the elderly. Fitness has been found to be a significant indicator of mortality (for example, in one study of those over 65 the hazard rate for the leanest quintile of fitness was about one third compared with that of the least-lean quintile), independent of overall or abdominal adiposity. In that study, those class I obese experienced about a 30 percent higher rate of mortality, while those class II or III obese experienced 130 percent higher mortality rates, with those with a waist circumference of greater than 88 cm for women and 102 cm for men having about 30 percent higher mortality rates.

Morbidity
Morbidity can result in both human suffering and adverse financial consequences, measured by the cost of medical services, loss of income and needed assistance in performing activities of daily living (ADLs). Although the focus on obesity has often been on its effects on mortality, a possibly greater concern is its consequential impact on health care costs. Over the last several decades, a simultaneous improvement in mortality rates and increase in health care costs have occurred, in part as a result of improvements in and aggressive treatment of, for example, cardiovascular disease risk factors that can at the same time be adversely affected by obesity.

Those overweight or obese are more likely to have additional functional impairments. Obesity can independently affect the onset of strength impairment, reduced body mobility and ADL problems. An increase in dementia has also been found to be associated with higher BMI and visceral adiposity. The use of long-term care facilities and home health care by the obese may also increase in the future as those with greater weight reach old ages.

Health care costs for those obese over age 65 are significantly higher than those of normal BMI. One study reported fee-for-service Medicare charges for those severely obese about 95 percent higher, for the class I obese 50 percent higher, and for those overweight 20 percent higher than those in the ideal BMI category.

In contrast to several studies published in the 1990s that indicated that health care expenditures related to obesity were between 5 percent and 7 percent of annual U.S. health care expenditures, two recent studies have estimated that between 9.1 percent and 16.5 percent of total health care costs can be directly attributed to the effect of being overweight or obese. Even the lower percentage in this range seems frightening, with a significantly portion involving costs for the morbidly obese. It has been estimated that up to a third of the increase in overall health care costs as a percent of U.S. gross domestic product (GDP) over the last two decades has been due to the increased prevalence of obesity.

What Can Be Done
There are many individual contributors to today's high prevalence of obesity. Therefore, a program to achieve a healthier population to be tailored in a multi-faceted manner to the individual. It should focus not only on weight, but also on contributing behaviors, primarily involving nutrition and physical activity. The development of effective weight management programs, including dieting, general education campaigns and a healthier attitude among a wide range of the population, will remain a challenge, as human behavior is quite resistant to associated required inconveniences.

The future effects on mortality, morbidity and health care of those who are now overweight and obese, especially the growing population of those morbidly obese should not be ignored. In particular, the resulting health care and disability costs are shared by the public. It has taken decades of intense government and private efforts to gain modest control over smoking. It will take at least as long to win a fight against obesity and sedentary living. In a society in which food is plentiful and affordable, and exercise is no longer necessary for immediate survival, only comprehensive long-term approaches and changes in attitude at both an individual and societal level will lead toward an effective solution.

This article is adapted from a paper presented at the Society of Actuaries’ 2011 Living to 100 International Symposium, which was held Jan. 5–7 in Orlando, Fla. The original paper is included in the symposium monograph.

Sam Gutterman, FSA, MAAA, FCAS, FCA, HONFIA, CERA, is director & consulting actuary at PricewaterhouseCoopers LLP in Chicago, Illinois. He can be reached at sam.gutterman@us.pwc.com.

References

Behan, D.F. and S.H. Cox. 2010. Obesity and its Relation to Mortality and Morbidity Costs. Society of Actuaries.

Flegal, K.M., K. Brian, H. Orpana, B.I. Graubard. 2013. Association of All-Cause Mortality With Overweight and Obesity Using Standard Body Mass Index Categories, A Systematic Review and Meta-analysis. Journal of the American Medical Association: 309(1):71-82. Doi:10.1001/jama.2012.113905.

Gutterman, S. 2011. Obesity—Status and Effects. Living to 100 Symposium Monograph. Society of Actuaries.

Heiat, A., V. Vaccarino, H.M Krumholz. 2001. An Evidenced Based Assessment of Federal Guidelines for Overweight and Obesity as They Apply to Elderly Persons. Archives of Internal Medicine. 161: 1194-1203.

Janssen, I., A.E. Mark. 2007. Elevated Body Mass Index and Mortality Risk in the Elderly. Obesity Reviews. 8(1), 4a:59.