HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Roblin, D. W. Articles by Joski, P. J. Search for Related Content PubMed Articles by Roblin, D. W. Articles by Joski, P. J.

Abstract C-B3-03: Applications of Mediation Analysis in Behavioral Medicine Research

Douglas W. Roblin, PhD, The Center for Health Research/Southeast, Kaiser Permanente Georgia; Jeroan J. Allison, MD, MSc, School of Medicine, University of Alabama at Birmingham; Edmund R. Becker, PhD, Rollins School of Public Health, Emory University; Peter J. Joski, MSPH, The Center for Health Research/Southeast, Kaiser Permanente Georgia

Abstract

Background: In a typical regression model, the total effect of an independent variable (e.g., social support or strain) on a dependent variable (e.g., health status) is estimated. In mediation analysis, a regression model accounts for the possibility that the total effect consists of a direct effect and an effect mediated through a third variable (e.g., patient activation). We examined the potential mediating effects of patient activation between various psychosocial circumstances and health status, dietary intake, exercise, and obesity among a sample of adults in a group-model managed care organization (MCO).

Methods: A mixed-mode survey (written and Internet options) was administered from October through December 2005 to 5309 MCO enrollees, 25–59 years of age, who were randomly sampled from 3 condition cohorts:

1. adults with type 2 diabetes,
   
2. adults with elevated lipids but no acute CVD history, and
   
3. adults with no major morbidities (low risk).
   

The survey included previously validated measures of psychosocial circumstances that measured support or strain between patients and physicians (care delivery climate), among friends and family members (social climate), and among coworkers (work climate); patient activation (PAM-13); healthy behaviors (e.g., exercise, dietary intake); and physical and emotional health (SF-12). Using the approach of Baron and Kenney, linear and logistic regression models were estimated to compute the mediating effect of patient activation. The mediating effect was tested for significance using the Sobel test.

Results: Twenty-two hundred twenty-four completed the survey (42% response rate). Forty-five percent of respondents were black; 60% were female. Adjusted for condition cohort and other patient covariates, psychosocial circumstances tended to have significant (P<0.05) total and direct effects on health status (better with more supportive climates), diet and exercise (better with more supportive climates) and BMI (lower with more supportive climates). Patient activation was a significant mediator (Sobel P<0.05) of the observed relationships between psychosocial circumstances and each of the three outcomes (health status, health behavior, and BMI) for all cohorts.

Conclusions: Patient activation has an important mediating role between psychosocial circumstances and health and healthy behaviors, particularly with respect to social climate.