In this article, we described the HIV and syphilis prevalence among men who have sex with men (MSM) in Chongqing City in southwestern China from a cross-sectional survey in 2009, comparing the risks of HIV infection between student MSM and non-student MSM. MSM in Chongqing City had a high HIV prevalence while had a relatively low syphilis prevalence. Student MSM were well represented in the Chongqing MSM population and had lower risks of HIV infection than non-student MSM.

In the discussion, we noted the potential impact of the arbitrary classification of student and non-student status on the outcome of HIV prevalence. MSM who were currently registered students (in high school or college) at the time of survey were classified as student MSM and all others as non-student MSM. The non-student MSM category included both college graduates and non-college graduates. College graduates, especially those who graduated near the survey year, could have similar HIV prevalence with currently registered students. Thus, such categorization of occupation status might lead to underestimation of the HIV risks and prevalence among student MSM.

After publication, we came up with a better approach to minimize the bias rising from the misclassification of occupation. We did an additional analysis including a newly derived occupational variable with three categories derived from education and occupation data: students (secondary/college), college graduates, and non-student/non-college graduates. The student (36.4%, 183/503) category was the same as in the original article while the article’s non-student category was further classified into college graduates (39%, 196/503) and non-student non-college graduates (24.7%, 124/503). In the univariate analysis, we included all the original variables plus the newly derived tripartite occupation variable, excluding the original education and occupation data. Multivariate logistic regression analysis showed that both college graduates [adjusted odds ratio (aOR): 3.5; 95% confidence interval (CI): (1.3-9.5)] and non-student college graduates (aOR: 5.7; 95%CI: 1.8-18.4) had higher risk of HIV infection than student MSM; and non-student college graduates had greater risk than college graduates compared with student MSM. Age and ethnicity were still two risk factors for HIV infection. In general, the new results were similar to those in the original paper, but we are nonetheless more satisfied by this more logical analysis. (see substitute Table 2 http://www.plosone.org/at...)

We also found out some errors in the original Table 2 due to formatting and these have been corrected in the revised Table 2. In conclusion, we have evidence that both non-student college graduates and college graduates have higher odds of being HIV-infected then do students, controlling for other key factors such as age.