HIV Prevention Research for Men Who Have Sex with Men:Meta-Analysis, Intraclass Correlation, and Transformation BetweenCount and Dichotomous Outcomes Open Access

Johnson, Wayne Douglas (2009)

Permanent URL:


Abstract HIV Prevention Research for Men Who Have Sex with Men: Meta-Analysis, Intraclass Correlation, and Transformation Between Count and Dichotomous Outcomes By Wayne D. Johnson, Jr.

This dissertation describes three methodological and content-based studies related to meta-analysis of HIV prevention research for men who have sex with men (MSM). First, we compared ANOVA models for individually- and group-randomized trials to derive the factors necessary to account for intraclass correlation (ICC) in three classic designs. For the simplest design, the factor is (n −1)VIF / (n−VIF) where n is the number of participants per condition, and VIF is the variance inflation factor 1+(m−1)ICC. Simulations confirmed our correction factors for both additive and multiplicative models.

Second, we used regression and the method of moments to identify candidate formulas for transformation between proportions (summarizing dichotomous data), and means and variances (summarizing count data). Best empirical results were obtained from regression models predicting the proportions as a function of the mean and variance (and vice versa), or by the method of moments assuming a negative binomial distribution.

Third, we applied these results in a meta-analysis of behavioral HIV prevention for MSM. We found 54 interventions with 16,224 participants. The 38 interventions that were compared to minimal or no HIV prevention controls reduced unprotected sex by 27% (95% confidence interval [CI] = 15% to 37%). The other 16 interventions reduced unprotected sex by 17% compared to standard or other HIV prevention interventions (CI = 5% to 27%). Our methodological work permitted a robust conclusion that behavioral prevention reduces self-reported unprotected sex among MSM.

HIV Prevention Research for Men Who Have Sex with Men: Meta-Analysis, Intraclass Correlation, and Transformation Between Count and Dichotomous Outcomes By Wayne D. Johnson, Jr. M.S.P.H., University of South Carolina, 1987 B.A., University of South Carolina, 1980 B.S., University of South Carolina, 1980 Advisor: William M. McClellan, M.D., M.P.H. A dissertation submitted to the Faculty of the Graduate School of Emory University in partial fulfillment of the requirements of the degree of Doctor of Philosophy in Epidemiology 2009

Table of Contents

Table of Contents 1. Overview...1 2. Background...4

2.1. Epidemiology and prevention of HIV/AIDS...4

2.1.1. Biology of HIV infection...5 2.1.2. Medical treatment and prevention efforts...6 2.1.3. Risk of transmission through sexual contact...7 2.1.4. MSM still at high risk for HIV and STD transmission...10 2.1.5. HIV prevention needs among MSM...12 2.1.6. Related risk factors...13 2.1.7. Condoms for HIV prevention...13 2.1.8. Alternatives to use of male condoms...17 2.1.9. Risk factors for progression to HIV disease and AIDS...19 2.1.10. Correlates of sexual risk...20

2.2. Previous reviews of HIV prevention for MSM...21 2.3. Accounting for the unit of assignment in group-randomized trials...26 2.4. Transformation between different types of data for meta-analysis...28

3. Objectives...30 4. Methods...31

4.1. Methods for determining the factor necessary to adjust for clustered assignment...31 4.2. Methods for determining formulas to transform between count and dichotomous outcomes...32 4.3. Methods for meta-analysis of HIV prevention research for MSM...33

4.3.1. Search strategies and eligibility criteria...33 4.3.2. Calculation of effect sizes...35 4.3.3. Statistical analysis...38

5. Study I: Adjusting for Clustered Assignment...42 6. Study II: Transformation between Count and Dichotomous Outcomes...43 7. Study III: Meta-analysis of HIV Prevention Research for MSM...44 8. Discussion...45

8.1. General conclusions...45 8.2. Strengths...46 8.3. Limitations...48 8.4. Implications and future research...49

9. References...52

Appendix A. Correcting the Variance of the Intervention Effect in Group-Randomized Trials when Only the Variance Appropriate to an Individually-Randomized Trial is Available...77 Appendix B. Transformations between Count and Dichotomous Outcome Measures for Meta-analysis: Predicting Risk Ratios from Means and Rate Ratios from Proportions...78 Appendix C. Meta-Analysis of HIV Prevention Research for Men Who Have Sex with Men...79 Appendix D. Behavioral Interventions to Reduce Risk for Sexual Transmission of HIV Among Men Who Have Sex with Men...80

About this Dissertation

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
  • English
Research field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files