Authors' reply: immunodeficiency at the start of combination antiretroviral therapy: steady improvement or step changes?

Abstract

We thank Ashley Olson and colleagues for their interest in our study of CD4 cell counts in over 350,000 patients who started combination antiretroviral therapy (ART) from low-, middle- and high-income countries during the years 2002 to 2010,1 based on the International epidemiological Databases to Evaluate AIDS (IeDEA)2 and the ART Cohort Collaboration (ART-CC).3 In response to their first point, we found a steady increase in CD4 cell counts rather than step changes following the change in WHO guidelines on when to start ART in 2006 and 2009. We modeled CD4 counts both using additive and linear mixed effects models but that there was virtually no difference in the fit of the two models. The effect of the change in the guidelines was thus gradual, rather than step-wise. This may be due to the understandable lag between a change in WHO guidelines and individual country-level implementation resulting in gradual rather than step-wise increases. Interestingly, our separate analysis of CD4 measures at the start of ART in children showed that trends were clearly non-linear, although there was again little evidence of step-wise increases following guideline changes.4 We agree with Olson and colleagues that late presentation for HIV care is one reason for the late start of ART, and that stratifying the analysis by the time from presentation with HIV infection to start of ART would be interesting. Unfortunately, the date of first presentation was not generally recorded in the cohorts participating the International epidemiological Databases to Evaluate AIDS. Also, where recorded, the definitions of first presentation varied. Of note, first presentation will often not have taken place at the facility where ART was subsequently initiated. Finally, we were a bit puzzled by the correspondents’ statement that “that median CD4 at initiation in this analysis is likely to overestimate the true median at initiation, as those who do not initiate are not included in this analysis.” Surely the CD4 count at initiation can only be studied in patients who initiate ART. However, as we discussed in our paper,1 CD4 cell counts at start of cART were missing in some patients, and these patients were more likely to be from lower- and middle income countries and in more advanced stages of HIV-infection than patients with CD4 counts. It is thus likely that our estimates of median CD4 cell counts at cART initiation were biased upward for these countries.