In this study, TB treatment in TB-HIV co-infected patients had no significant effect on CD4+ cell count. The evolution of CD4+ cell count is mainly driven by ART use. The average CD4+ cell count for patients not on ART remained below the cut-off for initiating ART of 350 cells/mm3 during the first 12 weeks of TB treatment. Therefore, a delayed assessment of the first CD4+ cell count in itself would probably not lead to missing an opportunity to start ART based on the cut off of 350 cells/mm3. This finding is relevant in the Mozambican setting where not all ART treatment facilities are capable of performing CD4+ cell counts and prescription of ART is prioritized. Health facilities sent blood samples for CD4+ cell count to another laboratory daily or weekly, depending on the distance between the sending and receiving facilities.
This study showed a small increase in CD4+ cell count during TB treatment in both patients on ART and patients not on ART as has been described in non-immune compromised TB patients [10
]. In other studies the CD4+ cell count in HIV-infected TB patients not on ART did not increases [12
]. It seems that the immune response in HIV-infected TB patients not on ART is variable.
This was a retrospective study based on routine data and as such has several limitations. First, bias may have occurred by not including patients who received HIV treatment at another health facility. These patients may have been treated differently or adherence may have been different. This would have influenced treatment outcome and CD4+ cell count. All health care facilities in Mozambique follow the same national guidelines and as such the chance of a difference in treatment strategy is unlikely. Furthermore, the characteristics for patients whose clinical record was identified were very similar with those from patients without an HIV record (Table ).
Second, not all TB patients in this study had an HIV-test or the result recorded in the TB register. Therefore, not all HIV-infected TB patients were included. Given the high testing rate of more than 80%, it is unlikely that the non-availability of the HIV-test result would markedly bias the results of the study.
Third, this study took place in three health facilities in a single province of Mozambique. The results may be different in other areas in Mozambique. However, we believe that the situation in Manica does not differ much from that in other provinces in the country at the time of the study, apart from the larger cities where more ART facilities are available. There is also more specialist care available in the larger cities.
Fourth, despite a considerable amount of patients, the number of available CD4+ cell counts per patient was small, reflecting the indications in the national guidelines as to when to perform this test. The use of the mixed model allowed us to use all available data and was therefore the recommended methodology for our data set.
Fifth, about one third of the patients that used ART during TB treatment started their ART before the start of TB treatment and potentially had incident TB while using ART. Emerging evidence shows that CD4+ cell count response is smaller in these patients [17
]. However, earlier evidence showed a similar CD4+ cell count response in both patients with prevalent and those with incident TB compared to patients on ART without TB [18
]. We cannot completely rule out a potential underestimation of the effect of TB treatment in our study. However, the majority of patients did not have this incident TB and we are confident that our results are valid.
Despite these limitations, we consider the results relevant and important because limited data are available on CD4+ cell count response in cohorts of TB patients.
The presently used HIV treatment guidelines in Mozambique are not yet in line with the WHO recommendation to initiate ART in HIV-infected TB patients as soon as possible after the start of TB treatment irrespective of the CD4+ cell count [8
]. The present study supports this recommendation as patients on ART had a much better immune restoration than those not on ART. However, like in Mozambique, these WHO guidelines have not yet been implemented everywhere.
Also, many countries with a high burden of HIV struggle to maintain all HIV-infected patients on ART. Lack of funding may lead to stock outs of antiretroviral drugs at facility level [19
]. Therefore countries may wish to prioritize new initiations of ART to those most in need. For HIV-infected TB patients, the CD4+ cell count provides a tool to prioritize. This study shows that obtaining a sample for CD4+ cell count assessment in the first 12 weeks of TB treatment will be a reliable indicator for the need to initiate ART, since this measurement is not influenced by concurrent TB treatment. The opportunity for identifying the HIV-infected TB patients most in need of ART is unlikely to be missed.