Open Access Open Badges Research article

A population-based study of Kaposi Sarcoma-associated herpesvirus seropositivity in Uganda using principal components analysis

Joanne T Chang13*, Fatma M Shebl14, Ruth M Pfeiffer1, Benon Biryahwaho2, Barry I Graubard1 and Sam M Mbulaiteye1

Author Affiliations

1 Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, USA

2 Uganda Virus Research Institute, Entebbe, Uganda

3 University of Michigan, School of Public Health, Ann Arbor, MI, USA

4 Currently at Yale School of Public Health, New Haven, CT, USA

For all author emails, please log on.

Infectious Agents and Cancer 2013, 8:3  doi:10.1186/1750-9378-8-3

Published: 16 January 2013



Kaposi sarcoma-associated herpesvirus (KSHV) seropositivity is associated with sexual, environmental, and socioeconomic exposures. Whether these characteristics are independent risk factors is uncertain because of reliance on selected high-risk or hospital-based populations and incomplete adjustment for confounding. Therefore, we evaluated risk factors for KSHV seropositivity in a population-based study in Uganda using principal components analysis (PCA).


The study population comprised 2,681 individuals randomly selected from a nationally-representative population-based HIV/AIDS sero-behavioral survey conducted in 2004/05. Questionnaire and laboratory data (97 variables) were transformed into a smaller set of uncorrelated variables using PCA. Multivariable logistic regression models were fitted to estimate odds ratios and 95% confidence intervals for the association between components and KSHV seropositivity.


Data were reduced to three principal components (PCs) labeled as Sexual behavioral, Socioeconomic, and Knowledge PCs. In crude analysis, KSHV seropositivity was associated with the Knowledge (ptrend = 0.012) and Socioeconomic components (ptrend = 0.0001), but not with the Sexual-behavioral component (ptrend = 0.066). KSHV seropositivity was associated with the Socioeconomic PC (ptrend = 0.037), but not with the Sexual-behavioral and Knowledge PCs, in the models including PCs, age, gender and geographic region.


Our results fit with the view that in Uganda socioeconomic characteristic may influence KSHV seropositivity. Conversely, the results fit with the interpretation that in Uganda sexual-behavioral characteristics, if relevant, contribute minimally.

Kaposi sarcoma-associated herpesvirus; Uganda; Kaposi sarcoma; Socioeconomic; Principal Components Analysis; Human herpesvirus 8