Body Mass Trajectories and Mortality Among Older Adults
Body Mass Trajectories and Mortality Among Older Adults
Purpose of the Study. The aim of this study was to investigate heterogeneity in body weight trajectories among older adults and their association with mortality risks.
Design and Methods. Information on body mass index (BMI) and survival come from nine waves of the Health and Retirement Study, a 16-year survey of adults aged 51–61 at baseline (N = 9,703). We used a sex-stratified joint growth mixture-discrete time survival model to characterize BMI trajectory groups and their associated mortality.
Results. Three distinct classes of BMI trajectories were identified: "stable overweight," "obese gaining," and "obese losing." Relative to the stable overweight class, which comprised about 90% of the sample, the obese gaining class had approximately 50% higher mortality risk; the highest mortality was found in the obese losing category (OR > 2.7, p <.001). The results were similar for men and women.
Implications. The findings highlight substantial heterogeneity in weight trajectories of older Americans, as well as large survival differentials across the classes. The direction of weight changes appears inextricably linked to the overall BMI level in terms of predicting older adults' longevity. Weight loss is associated with particularly high mortality risk even when the typical BMI change is from obesity to overweight.
Prior studies have documented an association between body mass index (BMI) and mortality. The lowest risk of death among older adults is typically found for those with BMI considered to be in the overweight range compared with those who are underweight, obese, or in the normal weight range (Flegal, Graubard, Williamson, & Gail, 2005; Flegal, Graubard, Williamson, & Gail, 2007; Janssen, 2007; McGee, 2005; Troiano, Frongillo, Sobal, & Levitsky, 1996). However, most previous research has characterized the BMI–mortality relationship using BMI measured at only one point in time. This is a limitation for two reasons. First, observed associations between a single measure of BMI and subsequent mortality may reflect underlying diseases and health conditions that influence weight, typically causing weight loss and also increasing the risk of death (Greenberg, 2006; Robins, 2008). Reliance on a single BMI data point may thus lead researchers to invalid conclusions about the association between body weight and mortality. Second, using one-time BMI information fails to show how weight change over time predicts mortality. Body weight can change substantially over the adult life course, and weight history may provide important information about an individual's health and mortality risks.
To overcome these limitations, recent research on BMI and mortality has begun to move away from a single, static measure of BMI and toward examining weight change over time using panel data with multiple BMI data points. Although such studies vary in sample compositions, length of follow-up, and analytic approaches, there are some consistent findings. In particular, weight loss is associated with excess mortality risk (Andres, Muller, & Sorkin, 1993; Lee et al., 2011; Myrskyla & Chang, 2009; Newman et al., 2001), whereas stable body weight tends to be associated with the lowest mortality risk (Bamia et al., 2010; Lee et al., 2011; Newman et al., 2001; Strandberg et al., 2009). Additionally, some studies have found modest weight gain associated with a comparable or decreased mortality risk relative to a stable weight trajectory (Andres et al., 1993; Myrskyla & Chang, 2009; Stevens, Juhaeri, & Cai, 2001).
The recent focus on weight change is a marked improvement over analyses of a single measure of BMI. However, prior studies have tended to use a variable-based approach that focuses on an a priori categorization of BMI level and change, rather than a person-based approach that could describe the shape of BMI trajectories that are actually observed in the population. The variable-based approach includes regression models where the mortality risk is modeled as a function of the starting or average BMI level and BMI change over time. One strategy is to control for baseline BMI, effectively treating it as a nuisance parameter. Alternatively, some researchers have created multiple categories of BMI change, such as "gaining from normal weight to overweight" or "losing weight from obese to overweight range," to capture a sense of both level and change in weight (Lee et al., 2011; Myrskyla & Chang, 2009; Newman et al., 2001; Stevens et al., 2001; Strandberg et al., 2009). Although this approach allows for a more nuanced understanding of the overall level and change in BMI, the categories depend heavily on the chosen thresholds and thus make it difficult to draw conclusions about actual population patterns.
This study uses a unique approach to examine the association between BMI and mortality in older adults by applying person-centered modeling techniques to relate long-term trajectories of BMI observed in the population to mortality risk. The substantive goals of this study are to (a) identify typical BMI trajectories and describe their shapes, (b) determine the mortality risk associated with different BMI trajectory groups, and (c) account for baseline health information in order to explore how baseline health and smoking status influences membership in different BMI trajectory groups. We address these study aims by using a joint generalized growth mixture–discrete-time survival model to analyze nine waves of data collected over 16 years from a large, nationally representative, longitudinal study of older adults.
Abstract and Introduction
Abstract
Purpose of the Study. The aim of this study was to investigate heterogeneity in body weight trajectories among older adults and their association with mortality risks.
Design and Methods. Information on body mass index (BMI) and survival come from nine waves of the Health and Retirement Study, a 16-year survey of adults aged 51–61 at baseline (N = 9,703). We used a sex-stratified joint growth mixture-discrete time survival model to characterize BMI trajectory groups and their associated mortality.
Results. Three distinct classes of BMI trajectories were identified: "stable overweight," "obese gaining," and "obese losing." Relative to the stable overweight class, which comprised about 90% of the sample, the obese gaining class had approximately 50% higher mortality risk; the highest mortality was found in the obese losing category (OR > 2.7, p <.001). The results were similar for men and women.
Implications. The findings highlight substantial heterogeneity in weight trajectories of older Americans, as well as large survival differentials across the classes. The direction of weight changes appears inextricably linked to the overall BMI level in terms of predicting older adults' longevity. Weight loss is associated with particularly high mortality risk even when the typical BMI change is from obesity to overweight.
Introduction
Prior studies have documented an association between body mass index (BMI) and mortality. The lowest risk of death among older adults is typically found for those with BMI considered to be in the overweight range compared with those who are underweight, obese, or in the normal weight range (Flegal, Graubard, Williamson, & Gail, 2005; Flegal, Graubard, Williamson, & Gail, 2007; Janssen, 2007; McGee, 2005; Troiano, Frongillo, Sobal, & Levitsky, 1996). However, most previous research has characterized the BMI–mortality relationship using BMI measured at only one point in time. This is a limitation for two reasons. First, observed associations between a single measure of BMI and subsequent mortality may reflect underlying diseases and health conditions that influence weight, typically causing weight loss and also increasing the risk of death (Greenberg, 2006; Robins, 2008). Reliance on a single BMI data point may thus lead researchers to invalid conclusions about the association between body weight and mortality. Second, using one-time BMI information fails to show how weight change over time predicts mortality. Body weight can change substantially over the adult life course, and weight history may provide important information about an individual's health and mortality risks.
To overcome these limitations, recent research on BMI and mortality has begun to move away from a single, static measure of BMI and toward examining weight change over time using panel data with multiple BMI data points. Although such studies vary in sample compositions, length of follow-up, and analytic approaches, there are some consistent findings. In particular, weight loss is associated with excess mortality risk (Andres, Muller, & Sorkin, 1993; Lee et al., 2011; Myrskyla & Chang, 2009; Newman et al., 2001), whereas stable body weight tends to be associated with the lowest mortality risk (Bamia et al., 2010; Lee et al., 2011; Newman et al., 2001; Strandberg et al., 2009). Additionally, some studies have found modest weight gain associated with a comparable or decreased mortality risk relative to a stable weight trajectory (Andres et al., 1993; Myrskyla & Chang, 2009; Stevens, Juhaeri, & Cai, 2001).
The recent focus on weight change is a marked improvement over analyses of a single measure of BMI. However, prior studies have tended to use a variable-based approach that focuses on an a priori categorization of BMI level and change, rather than a person-based approach that could describe the shape of BMI trajectories that are actually observed in the population. The variable-based approach includes regression models where the mortality risk is modeled as a function of the starting or average BMI level and BMI change over time. One strategy is to control for baseline BMI, effectively treating it as a nuisance parameter. Alternatively, some researchers have created multiple categories of BMI change, such as "gaining from normal weight to overweight" or "losing weight from obese to overweight range," to capture a sense of both level and change in weight (Lee et al., 2011; Myrskyla & Chang, 2009; Newman et al., 2001; Stevens et al., 2001; Strandberg et al., 2009). Although this approach allows for a more nuanced understanding of the overall level and change in BMI, the categories depend heavily on the chosen thresholds and thus make it difficult to draw conclusions about actual population patterns.
This study uses a unique approach to examine the association between BMI and mortality in older adults by applying person-centered modeling techniques to relate long-term trajectories of BMI observed in the population to mortality risk. The substantive goals of this study are to (a) identify typical BMI trajectories and describe their shapes, (b) determine the mortality risk associated with different BMI trajectory groups, and (c) account for baseline health information in order to explore how baseline health and smoking status influences membership in different BMI trajectory groups. We address these study aims by using a joint generalized growth mixture–discrete-time survival model to analyze nine waves of data collected over 16 years from a large, nationally representative, longitudinal study of older adults.
Source...