Dear Editor,The systematic review and meta-analysis of cardiovasculardisease (CVD) in HIV-infected individuals was an importantcontribution to the growing field of research in aging withHIV infection [1]. While the analysis appears of highquality, there were three areas that we believe were notgiven critical attention: the HIV-uninfected comparisongroup, the presence of residual confounding, and the needfor a competing risks analysis.First, although the research question was succinctlystated as whether there was a difference in CVD riskin HIV-infected compared with HIV-uninfected adults,the HIV-uninfected comparison groups were not givenadequate scrutiny. The ideal comparison group would be anHIV-uninfected group that would have a similar prevalenceof risk factors for CVD compared with HIV-infected indi-viduals. Reflecting the paucity of data in this area, onlythree of the 23 studies investigated had an HIV-uninfectedcomparison group. Two of these three studies made acomparison to the general population [2,3]. The generalpopulation is an inappropriate control group because HIV-infected populations have an increased prevalence of tra-ditional risk factors for CVD such as cigarette smoking[4,5], high blood pressure [5], and dyslipidaemia [5,6] com-pared with the general population. The third study usedHIV-uninfected adults seeking care at the same affiliatedcare facilities as the HIV-infected adults [7]. However,HIV-infected participants had a greater proportion of maleand Black individuals as compared with HIV-uninfectedparticipants [7]. It is likely that the differences in theHIV-infected and -uninfected groups do not end withsociodemographic characteristics. The HIV-uninfectedgroup probably had prevalences of risk factors that weresimilar to those of the general population, as comparedwith the HIV-infected population; however, these data werenot shown.Secondly, to best isolate the causal effect of HIV, con-founders must be taken into account in the analyses. Thestudy by Lang and colleagues reported an age- and sex-standardized morbidity ratio, but did not account for otherconfounders. In the study by Obel and colleagues, HIV-uninfected participants were matched on age, sex andmunicipality of residence; they were also able to accountfor diabetes, alcoholism, hypertension, liver disease, andkidney disease. Finally, Triant and colleagues were able toadjust for age, sex, race, hypertension, diabetes, and dys-lipidaemia. Smoking data were not available for all studyparticipants; Triant and colleagues concluded “It is possiblethat increased smoking rates in the HIV group contribute tothe increased rate of AMI in the HIV group” [7], as has beendemonstrated elsewhere [4]. In addition to smoking, noneof these three studies were able to adjust for family historyof CVD and substance abuse, as the authors correctly pointout. This inability to control for these confounders amongHIV-infected compared with HIV-uninfected participantsundermines the internal validity of these studies.Finally, the authors missed the opportunity to discussa critical element missing from all the studies to date:the impact of competing risks. Prior to the advent ofeffective therapy, CVD was not a major contributor to deathamong those with HIV infection – AIDS death was CVD’smain competitor. Without utilizing methods that properlyaccount for competing risks [8], these studies are unable toproperly answer the question: Is there truly an increase inthe risk of CVD attributable to HIV, or is the CVD burdenamong people with HIV being unmasked by the increasedlife expectancy resulting from effective treatment?Although the methods of the systematic review andmeta-analysis were impeccable, the study would have beenstrengthened by a critical epidemiological examination ofthe included studies.