Article

Metaanalysis and metaregression in interpreting study variability in the impact of sexually transmitted diseases on susceptibility to HIV infection.

GLOBINF--Centre for Prevention of Global Infections, University of Oslo, Oslo, Norway.
Sex Transm Dis (Impact Factor: 2.75). 07/2005; 32(6):351-7. DOI: 10.1097/01.olq.0000154504.54686.d1
Source: PubMed

ABSTRACT Observational studies examining the effects of other sexually transmitted diseases (STDs) on HIV susceptibility differ in the populations observed and in which "other STDs" are examined. The extent to which an STD alters the risk of transmission of HIV may vary according to disease and population characteristics.
The goals of this study were to review studies examining the effect of other STDs on HIV-1 susceptibility and to correlate their effect estimates with type of "other STD", study design, and population characteristics.
Relevant studies with longitudinal design were identified through a systematic search of the PubMed database, and their evidence was critically evaluated. Metaregression techniques were then used to correlate study characteristics with corresponding effect estimates.
Of 31 studies included, 4 contained direct data on exposure to HIV-1. Three of these were inconclusive, the fourth indicating a strong relationship between STDs and transmission of HIV. Pooled effect estimates using all studies are statistically significant and indicate a 2- to 3-fold increase in risk of HIV-1 acquisition. Effect estimates corresponding some of the "other STD" categories exhibit heterogeneity, but no significant associations with study characteristics were found.
Most of the studies lack direct exposure data, lending them susceptible to exposure bias. Another problem may be measurement error about risk factors and STD status at time of HIV-1 infection. Because direct exposure data are difficult to come by (4 of 31 studies contained such data, all but 1 inconclusive), future observational studies on the influence of STDs on HIV-1 transmission should include quantitative analyses of the sensitivity of results to potential confounding and measurement error if they are to further understanding.

0 Followers
 · 
71 Views