Weight History Links Overweight/Obesity and Mortality

Becky McCall

April 04, 2017

People who were overweight or obese in the past have a higher risk of all-cause and specific-cause mortality, compared with those who consistently maintained a healthy weight, according to a large, US-based, multicohort analysis.

By using weight history to determine body mass index (BMI), the findings add some much-needed clarity to the ongoing debate about the optimal BMI in relation to mortality. The findings reverse the previously reported paradoxical, protective association between overweight and mortality that was seen with a single-point measurement of BMI.

For the current study, the investigators drew on a 16-year weight history. Compared with people of normal BMI (18.5–24.9 kg/m2) throughout the period, participants with a maximum BMI (the highest recorded BMI in the weight-history period) in the overweight and obese categories were at greater risk for all-cause death, as well as death from cardiovascular disease (CVD) and death due to non-CVD causes, cancer, or respiratory disease.

The findings are published online April 3 in the Annals of Internal Medicine.  

The analyses showed that the lowest risk for death occurs among individuals with a maximum BMI within the normal range at all ages, regardless of sex and whether they smoked or not. Maximum BMIs in the overweight (25.0 to 29.9 kg/m2), obese I (30.0 to 34.9 kg/m2), and obese II (≥35.0 kg/m2) categories were associated with higher risks for all-cause death (HR, 1.06; 95% CI, 1.03–1.08; HR, 1.24; 95% CI, 1.20–1.29; and HR, 1.73; 95% CI, 1.66–1.80, respectively), as well as death due to CVD, cancer, and other causes.

"The key distinction between our study and prior studies [including a large meta-analysis that found an inverse association between overweight/obesity and all-cause mortality (JAMA. 2013;309:71-82)] was the examination of weight history as opposed to a snapshot of weight at a single point of time," explained Andrew Stokes, PhD, senior study author and assistant professor in the department of global health at Boston University School of Public Health, Massachusetts.

In an accompanying editorial, Jean-Pierre Després, PhD, from Laval University, Québec, highlights the importance of the type of BMI measure used, noting that the study sheds light on the apparent reverse-causality problem. "This lesson suggests that we need to go beyond a single BMI measurement to refine our risk assessment, particularly within this prevalent subgroup of overweight patients who may sometimes read in the lay press that their body weight is okay."

Clarifying Uncertainty

The researchers examined the association between excess weight and the risk of dying from all causes as well as specific causes of death. The maximum BMI measures, drawn from participants' 16-year weight histories, were related to mortality risks over a 12-year follow-up period. Data from just over 225,000 participants were included in the analysis.

Few studies have used weight-history data previously because it is difficult to secure high-quality, longitudinal data across the adult life course. Dr Stokes pointed out that a key benefit of using weight history is that it minimizes reverse causality that has biased some prior studies on the BMI-mortality relationship.

In this scenario, reverse causality would affect results if an individual had an underlying disease prior to BMI measurement that could have an impact on the outcome, rather than the BMI itself affecting mortality. Often, patients with serious disease can lose weight due to illness up to 10 years before death. In prior studies, a snapshot BMI might have been measured when a disease had already taken hold of an individual and they had already lost weight. "Using this snapshot measure of BMI is not a true reflection of their stable BMI status," explained Dr Stokes.

In this study, the 16-year weight-history period was taken prior to baseline (start of the 12-year follow-up), which was considered an adequate period of time to obtain minimally biased estimates of BMI. Longitudinal data were drawn from the US Nurses' Health Study 1 and 2, as well as the Health Professionals Follow-Up Study (HPFS), and the potential associations between different categories of BMI with all-cause mortality and cause-specific mortality were generated.

Also, in a departure from previous studies, the current study did not exclude smokers, both former and current, and anyone with a condition at the time of BMI assessment. ""With respect to smokers, in the US, approximately 50% of adults are former or current smokers, which would mean excluding a large proportion of people, compromising sample size and potentially skewing results,"" Dr Stokes pointed out. The statistical analysis stratified results for smoking, age, and sex.

In order to investigate the use of a baseline BMI value (as used in studies that showed an inverse association) as opposed to the weight-history approach, the researchers compared the association with mortality found using a weight-history BMI measurement and the association found using a baseline BMI.

These results were found to differ depending on the measure used: a significant inverse association between overweight and mortality (HR, 0.96; 95% CI, 0.94–0.99) was observed when BMI was defined using a single baseline measurement vs an HR of 1.06 observed when weight history was used, explained Dr Stokes.

In terms of cause-specific mortality, "coronary heart disease seemed to stand out as strongly associated with increased BMI [overweight: HR, 1.32; obese I: HR, 1.97; and obese II: HR, 3.34]." The overall cardiovascular mortality subgroup also showed raised risk for all categories of overweight/obesity. "In fact, in each cause-specific category we found consistent strong effects between overweight/obesity and each specific cause including cancer [HR, 1.06; 95% CI, 1.01–1.13 in the obese I category; and HR, 1.28; 95% CI, 1.19–1.38 in obese II].

Clinical Implications

Dr Stokes explained that many previous studies have looked at more severe obesity, but this study serves to stress the importance of overweight, as well as obesity, as a risk factor for disease. He noted that this was particularly relevant in the United States where one-third of adults were overweight.

"The finding that overweight is associated with an increased risk of mortality highlights the importance of this particular group. It shows that the emphasis needs to be on more than just extreme obesity because being overweight raises an individual's risk of heart disease and cancer, among other diseases."

He also emphasized that the risks for any given overweight individual might not be that large, despite the statistical significance found in the study. However, because so many people are in the overweight category, the findings will translate into many new cases of heart disease and cancer each year.

"Also, in terms of clinical implications of our results, it suggests that clinicians should consider weight at prior points in time as well as the patient's current weight," added Dr Stokes, highlighting the importance of maintaining normal weight over the life course, especially in view of the recognized creep in weight with age.

Finally, he noted that given that rates of overweight/obesity have been rising and because overweight/obesity are shown in this study to be significant risk factors for dying, "it makes sense to consider that obesity is leaving an imprint on trends in life expectancy at the population level. Our results suggest this, but we have yet to tease out this impact."

In the editorial, Dr Després recognizes the value of using the weight-history approach but asks about the potential practical application of how best to assess weight/BMI: "How could clinical practice be improved for patients in this overweight BMI category, which still has an identity crisis?"

He notes that the recent findings "support our previous proposal that adding waist circumference to BMI may be the simplest way in clinical practice to refine evaluation of the health risk of overweight when no precise record of weight history is available."

Dr Després adds that the assessment and use of ''overall nutritional quality scores would be instrumental in better discrimination of morbidity and mortality risk. The development and implementation of such tools in clinical practice should be a high-priority objective for clinical nutrition experts.''

The study was funded by the National Institutes of Health. The authors have declared no relevant financial relationships. Dr Després reports personal fees from Torrent Pharmaceuticals outside the submitted work.

For more diabetes and endocrinology news, follow us on Twitter and on Facebook.

Ann Intern Med. Published online April 4, 2017. Abstract, Editorial


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: