Role of behavioral confounding in the association between Marijuana use and BMI in us Adults

Citation

Thompson, C. & Hay, J. W. (2015). Role of behavioral confounding in the association between Marijuana use and BMI in us Adults. ISPOR 20th Annual International Meeting. Philadelphia, PA: Value in Health.

Abstract

OBJECTIVES Although the compounds in marijuana increase appetite, empirical studies have estimated that marijuana users have lower BMI and rates of obesity than non-users. Failure to account for differences in behaviors and attitudes increases the potential for confounding, since the decision to engage in marijuana use is tied to such attributes. This study investigated how estimates of the effect of marijuana use on BMI are altered by such considerations. METHODS Participants in Wave IV of the National Longitudinal Study of Adolescent Health (Add Health) were interviewed between 2008 and 2009 (age 24-32). Multivariate regressions estimated the relationship between marijuana use and BMI (kg/m2). Regressions were then stratified by behavioral variables, including age of onset of marijuana use, concurrent alcohol use, self-assessed weight status, risk-propensity, and gender. RESULTS Consistent with other empirical studies, marijuana use is associated with lower BMI in the core model (P=0.015). However, results change when models are stratified by behavioral variables. The negative association between marijuana use and BMI persists in individuals who drink more than once a week (P<0.05); however, in non-drinkers (who have tried alcohol), the magnitude and direction of the effect changes drastically (-0.996, p=0.015 vs. 7.96, p=0.007). Users who initiated use at 17 or older do not have lower BMI (P=0.937). In individuals who identified as satisfactory weight or overweight, there is no association (P=0.313 and P=0.748, respectively, vs. P = 0.005 in underweight individuals). Estimates are negative and significant in self-identified risk-takers (P=0.037), but not in non-risk-takers (P=0.549 and P=0.401). Results are not significant in females (P=0.976). CONCLUSIONS The extent to which marijuana use influences BMI may depend on behavioral factors. Investigation is required to pinpoint behavioral attributes responsible for differences in BMI among marijuana users. The results have implications for assessing the role of marijuana in strategies to address obesity.

URL

https://doi.org/10.1016/j.jval.2015.03.1710

Reference Type

Conference proceeding

Book Title

ISPOR 20th Annual International Meeting

Author(s)

Thompson, C.
Hay, J. W.

Year Published

2015

Volume Number

18

Pages

A293

Publisher

Value in Health

City of Publication

Philadelphia, PA

ISSN/ISBN

1098-3015

DOI

10.1016/j.jval.2015.03.1710

Reference ID

7748