In this study we observed a direct correlation between the two different outcomes (survival and CD4+ T-cell count recovery) and adherence evaluated through our score: our data substantially confirm previous findings [25, 26] suggesting that good adherence enhances CD4+ T-cell recovery and decreases mortality.
A significant proportion of our study population was represented by women (72.8%), since CMSC is a medical center devoted to “mother and child health-care” where patients are predominantly pregnant women and their HIV-infected partners . Hence, patients who went to CMSC were usually asymptomatic at enrolment (i.e. at an early WHO clinical stage at baseline). Conversely, the other two centers mainly admit symptomatic individuals and hospitalized late presenters.
Moreover, people coming from Nanoro rural District often emigrate particularly to Ivory Coast, delaying access to HIV screening and care . The high proportion of HIV-infected migrants on HAART who work abroad may be partly responsible for the worsening of therapeutic adherence in this setting. This phenomenon was taken into account during the statistical analysis and the multivariate logistic regression model was realized after stratifying for WHO clinical stage at baseline and for calendar year.
The lack of a gold standard method to evaluate adherence brought us to use a multi-parametric strategy based on a combination of questionnaires, pills count, punctuality on clinical consultations and blood tests. This method allowed us to bypass single measurements–related biases, by taking into account several aspects of treatment adherence according to WHO recommendations . Previous studies employed a variety of methods to assess adherence, with different thresholds to define good adherence, and a wide range of adherent patients proportions were reported (from 20.7 to 93.5%) [10, 11, 29–34]. Since the methodologies of these studies widely varied from each other, they are not easily comparable to our research.
The increase in CD4+ T-cell count was similar in patients with optimal and suboptimal adherence (0–7 and 8–10 group) during the first year of follow-up, while a better adherence (8–10 group) during the second year was significantly related to a progressive increase in CD4+ T-cell count. Our findings are in line with those described in the DART trial where investigators observed a worse immunological recovery (assessed through the CD4+ T-cell count evolution) among less-adherent patients.
We also found that good adherence is strongly related to a lower risk of death in univariate and multivariate models, given that the adherence score was considered as continuous variable or as a binomial variable (8–10 vs 0–7 points).
The univariate analysis of death predictors revealed that living in urban areas was associated to a better survival, while the other socio-demographic features (education, religion, gender and increasing age) were not. In fact, living in a urban area is likely to be associated with fewer disruptions in access to medicines, which seem to facilitate adherence [35–37]. Moreover, WHO clinical stage at baseline resulted to be significantly related to death, while CD4+ T-cell count (<50 vs >50 cells/μl) did not.
The multivariate analysis showed an increased risk of death among patients who were followed at CERBA or Nanoro as compared to those followed at CMSC, as a consequence of the above-mentioned differences between patients enrolled in the three healthcare facilities.
In addition, receiving a regimen including FDCs was confirmed to be significantly related to survival, suggesting the importance of a wider access to HAART in low-income Countries with a particular attention to compact regimens .
At any rate, some limitations should b e taken into consideration while interpreting our study. First of all, it was not possible to collect full data for the whole cohort since medical records were sometimes incomplete or because patients had died, had been lost to follow-up or transferred elsewhere. Second, our adherence score has not been validated and we are not able to establish the relative weight of each component of the score. Moreover, we did not compare this score with other methods to evaluate adherence since validation of this score was not the aim of this study. Lastly, the assignation of our score is purely at the discretion of the physician who performed the visit, and it should not be interpreted as a objective measure. Although the scores assigned at each visit were equally weighted, despite the variability in the length of assessment, the patients were followed in accordance with current WHO guidelines and all data were analyzed only retrospectively.