i
Excess Deaths Associated With Underweight, Overweight, and Obesity: An Evaluation of Potential Bias
-
June 2018
-
[PDF-687.21 KB]
Details:
-
Corporate Authors:
-
Description:Objectives: Previous analyses of mortality data from the National Health and Nutrition Examination Survey published in 2005 found that relative to the normal weight category of body mass index (BMI), defined as a BMI of 18.5–24.9, underweight (BMI less than 18.5) and obesity (BMI greater than or equal to 30) were associated with excess mortality, and overweight (BMI 25–29.9) was associated with reduced mortality, after adjusting for sex, smoking status, race–ethnicity group, and alcohol consumption. Possible sources of bias in those analyses include residual confounding due to imperfect adjustment for smoking and prevalent illness at baseline. This report presents an evaluation of these sources of bias.
Methods: The relative risks associated with BMI categories were calculated both before and after exclusions based on self-reported health status, physician-evaluated health status, smoking, early death, and weight stability. Additional analyses examined the effects of using BMI measured after age 70.
Results: No systematic or large effects of exclusions on the magnitude or direction of the relative risks were observed. Underweight was associated with elevated risk and overweight with decreased risk relative to the normal weight category, even after exclusions based on smoking, health status, early death, and weight stability. Applying these exclusionary criteria simultaneously excluded up to 85% of the original sample and 92% of the deaths.
Conclusions: These analyses of the effects of exclusions and stratification by health status do not suggest that the 2005 results were affected by large or systematic biases from residual confounding due to smoking, prevalent illness at baseline, or illness-induced weight loss.
Suggested citation: Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths associated with underweight, overweight, and obesity: An evaluation of potential bias. National Center for Health Statistics. Vital Health Stat 3(42). 2018.
CS290481
-
Subjects:
-
Series:
-
Document Type:
-
Genre:
-
Place as Subject:
-
Pages in Document:vi, 13 numberd pages
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
