Referee 2's review:

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.
**********

This manuscript addresses the critically important topic of integrating prevention into the clinical services of persons living with HIV. The authors have developed a computer-based intervention that utilized a video doctor and personalized feedback to facilitate behavior change among HIV-positive patients receiving services at 5 San Francisco area outpatient HIV clinics. The development of an efficacious intervention would be highly significant and relevant to the care of this population.

A number of issues in the manuscript require clarification to assist in the reader's evaluation of the intervention's efficacy. A primary issue is lack of clarity with regards to the outcome variable under investigation, specifically the sexual risk variables. It appears from the data analysis that patients reported their condom use on each of up to 5 casual partners. This should be clarified in the Methods section. The analysis also reports on the number of casual sex partners. It's not clear in the Methods section if participants are queried regarding the number of partners. It appears in that section that participants report on the number of casual partners up to five, instituting a ceiling on the number of partners.

The authors indicate that participant randomization was stratified by risk profile. Additional detail regarding this stratification would be useful. The authors also indicate that participants and their providers were blinded to group assignment. It is not clear how this is possible given the intervention group participants receive printed feedback ("Cueing sheet") that is then given to their provider, whereas the control group does not. It would appear that any patient who receives the printed feedback would be recognized as being in the intervention group. In addition, the graphic and content of the Cueing sheet is unclear.

The statistical analysis is not strong. The investigators report on change scores over 6 months on dichotomous variables. This strategy results in a significant reduction of data, doesn't provide information about the rate of change over time, and doesn't fully explicate the interaction between time and intervention condition. It is also not clear why the authors did not include unprotected sex with HIV-negative partners, given they had the data and in those circumstances there is significant risk of further transmission of the virus.

The discussion overstates the findings of the current study. The results presented indicate no change in condom behavior, no change in risky drinking, and decrease in number of casual partners. It is inappropriate to conclude that the intervention "achieved significant cessation of drug use and unprotected sex." The authors also indicate that the intervention "overcame providers' traditional barriers to consistent risk screening and counseling." However, the authors did not discuss what those barriers are or provide data to indicate that all the providers' barriers were addressed. Additional comment on some of the findings is also warranted. For example, why might there be changes at 6 months on the outcomes, but not at 3 months? Any indication that the counseling by the provider made a difference? Why might the intervention be more successful for substance use?