Vide VIDUS

Ned Glick appears to recognize the health inequalities faced by Aboriginal people but views these inequalities as inevitable since, from his statistical viewpoint, Aboriginal ethnicity is a factor “that cannot be modified.” Not surprisingly, our Aboriginal partners in VIDUS do not want to modify their ancestry. Instead, they wish to see the social determinants of ill health improved in their communities. Although we cannot change the history of assimilationist residential school policy, we can begin to address its legacy — despair, poverty, addiction, unemployment, child prostitution and incarceration. Fortunately, these latter factors are eminently modifiable — if we have the will to confront this national shame.

Glick notes inflections and crossings in the cumulative HIV incidence curves for Aboriginal and non-Aboriginal participants. On this basis, he suggests that the Cox proportional hazard regression model is inappropriate for these data. However, in this study, Cox proportional hazard regression was not used to model the hazard ratio of Aboriginal versus non-Aboriginal participants. Rather, it was used to assess the independent effects of risk factors on time to HIV seroconversion among Aboriginal participants only. Thus, the appropriateness of the Cox model should be assessed using time-to-event data for the Aboriginal participants only. Indeed, we did assess departures from the assumption of proportional hazards using both graphic and numeric checks. For the variables of interest in these analyses, we found no evidence of significant departures from this assumption.

Glick suggests that contingency table analysis might be useful for demonstrating the difference between Aboriginal and non-Aboriginal participants with respect to the percentage of individuals who experienced seroconversion within 42 months (if known). However, this approach assumes that each participant has been followed equally for 42 months. We believe that a more appropriate method in this situation is survival analysis, which accounts for different lengths of follow-up among the participants. Nonetheless, we did carry out the analyses Glick suggested and obtained results entirely consistent with those presented in our article.¹

Although Glick found the tone of our article “overbearing,” the language we used was pre-reviewed and approved by a number of our Aboriginal partners. Understandably, these organizations advocate the need to be assertive, as their community is experiencing prevalence rates for HIV of over 40% and for hepatitis C of over 90%, along with significantly reduced life expectancies. Perhaps if Glick were witnessing similar devastation in his own community, he might feel the need to be similarly assertive.

Kevin J.P. Craib Patricia M. Spittal Evan Wood British Columbia Centre for Excellence in HIV/AIDS St. Paul's Hospital Vancouver, BC On behalf of the VIDUS investigators