Body Mass Index, Weight Change and Death in an Older Hypertensive Population: The SHEP Study Grant W Somes Stephen B Kritchevsky Marco Pahor Ronald I Shorr William B Applegate University of Tennessee Health Science Center, Memphis, TN
Background: Static Measures of Weight and Mortality • In older populations, most studies show that low BMI is associated with increased mortality risk • Fewer studies have shown that high BMI is associated with increased mortality risk • The relationship with low BMI is attenuated in those with long-term weight stability, non-smokers, and those without weight related health conditions (Diehr et al., 1999; Calle et al., 1999).
Background: Dynamic Measures of Weight and Mortality • On a population-level, there is a general tendency for adults to lose weight after age 60. The dynamics of this change may be intimately related to health status. • In clinical settings, acute weight loss (>5%/year) is associated with increased mortality. • Most previous studies in older populations have had only two weights, one of which is recalled from over a decade previously. • There is little data on older populations, looking at the importance of short-term weight change and mortality risk
Study Questions: In older adults... • What is the relationship between baseline BMI and mortality? • Outside of a clinical setting is weight change associated with mortality? • What are the simultaneous effects of static and dynamic measures of weight on mortality?
Systolic Hypertension in the Elderly Program (SHEP) • Randomized clinical trial of 4736 (isolated systolic) hypertensives, age 60 and older, comparing drug treatment with placebo. • Primary endpoint was stroke (fatal or non-fatal). • Vital status known for nearly 100% of enrolled subjects. (n=455 deaths) • Average duration of follow-up of 4.5 years.
Present Study • Inclusion Criteria • > 3 weight values • >1 year follow-up beyond 3rd weight value • Exclusion Criteria • Missing data related to BMI calculation at baseline (randomization visit). • Missing data related to covariates (eg. Gender, smoking, diabetes, prevalent CHD etc) • 4485/4736 subjects were included • 351 (7.83%) deaths
Methods • BMI represents the BMI recorded at randomization visit (kg/m2). • Weight measured quarterly, and at other SHEP visits. • Weight change is the average annualized weight change based on the regression of weight on time from baseline until one year before last measurement (death, loss to follow-up, or end of study). • Other baseline variables are gender, age (divided at 70), current smoking status, and risks (previous heart attack, stroke or diabetes)
Analytic Strategy • The primary outcome was all cause mortality. • BMI and Weight Change were divided into quintiles • Logistic regression was used to related mortality risk to weight change and baseline BMI compared to participants at lowest risk (23.6 to 28.0 BMI and -0.7 to 0.5 kg/year Weight Change) • Covariates • Age • Gender • Baseline smoking status • “Risk” • Prevalent CVD • Diabetes
Weight Change in BMI Groups BMI < 23.6 BMI > 31.1
Discussion • Weight loss of more than 0.7 kg/year was associated with high mortality risk--even in persons with high BMI. • Weight gain of more than 0.5 kg/year was associated with high mortality risk--even in persons with low BMI. • Findings consistent with 1-year and 2-year lag. Therefore observed relationship between weight change and mortality less likely to be ascribed to terminal illness. • Our findings consistent across major causes of death.
Conclusions • Limitations: • Setting: Clinical trial in Isolated Systolic Hypertension • Involuntary v. voluntary weight change • Dynamic weight measures are a more important predictor of mortality than static weight measures. • Weight stability regardless of BMI is associated low mortality in older adults.