Survival of people on antiretroviral treatment in Zambia: a retrospective cohort analysis of HIV clients on ART

Introduction Provision of free anti-retroviral therapy in Zambia started in June 2004. There were only 15,000 people on treatment as at December that year, mainly due to lack of access. This number rose to 580,000 people as at December 2013. The general objective of this study was to determine survival of people on ART and to examine associated predictors for survival. Methods The study included ART patients enrolled between the year 2002 and 2013 (n=10,395) in 285 health facilities in Zambia. Patient files were analyzed retrospectively. The study used Kaplan Meier and Cox-proportional hazard models to describe the relationship between lost to follow up and age, sex, baseline CD4 cell count and weight. Results Results showed that lost to follow up accounted for 90% of the clients that had dropped out, while 10% was to deaths. Low baseline CD4 count (p-value 0.001, HR 0.9994, (95% CI 0.9993, 0.9996) at initiation was associated with lost to follow up together with weight at initiation (p-value 0.031, HR 0.9987 at 95% CI (0.9975, 0.9998)) of ART. Conclusion This study has demonstrated that lost to follow up is a substantial contributing factor to drop outs among HIV patients on treatment. Strengthening of community treatment supporters especially immediate family members in emphasizing to the client the need to continue treatment is necessary. The health facility could do more in emphasizing the importance of treatment especially in the initial stages. Further, in order to reduce opportunistic infections and probable deaths during treatment, cotrimoxazole prophylaxis should be maintained so as to raise the CD4 levels. Improved nutritional assessment and counseling to boost the nutritional status of the clients throughout should be encouraged.


Introduction
The beginning of Anti-Retroviral Treatment (ART) in 1996 led to a revolution in the care of patients with AIDS in the developed world.
Although this treatment does not cure and also presents new challenges with respect to side effects and drug resistance, it has dramatically reduced rates of morbidity and mortality, has also improved the quality of life of people with HIV/AIDS and has revitalized communities [1]. Zambia has one of the most devastating HIV and AIDS epidemics. Lusaka posting over 19% [3]. In the year 2002, the Zambian government took a policy decision to make ART available to everyone needing treatment and allocated 3 million US dollars to purchase ARV drugs for 10,000 people, to be provided through the public service [4]. The provision of ART was recognized as an integral component of the multi-sectoral response to HIV/AIDS. This was provided on a cost-sharing basis. In June 2004, the Ministry of Health (MoH) started providing free treatment to people eligible for ART. Since then, the ART program has been scaled up from the two initial pilot hospitals to over 550 of the 2000 public and private health facilities across the country [5,6]. The new policy includes free drugs, basic laboratory tests and CD4 counts. Over 80% of those eligible for ART are actually on it [5] and focus has shifted to maintaining these people on treatment thereby increasing the rate of survival on treatment. It is estimated that in Sub-Saharan Africa between 8% and 26% of patients die in the first year of ART, with most deaths occurring in the first few months [7]. Mortality rates are likely to depend not only on the care delivered by ART program, but more fundamentally on how advanced disease is at program enrolment and the quality of preceding health-care According to the Time intervals at which these events occurred were recorded as they were essential to the analyses. A patient is declared LTFU when they miss an appointment and efforts to trace the client for 90 days have proved futile. A patient is declared dead if they die in the hospital and the information reaches the ART clinic or if during the follow up time, the ART clinic has been informed that its client died in the hospital or community without the information reaching the  Table 1 also shows that the ART program lost over 37% of its clients due to LTFU and deaths within the first 12 months, and lost as many as 46% and above 50% of clients after 24 months and 36 The differences between year of registration and whether one is lost to follow up or not at the end of 2013 were found to be significant with p-value less than 0.0001 at 95% confidence level.
Determinants of survival and mortality: About 70% of the patients had their baseline CD4 count taken at enrollment into the ART program. Figure 1 below shows the median CD4 count distribution over the years at enrollment into the ART program. As shown in Figure 1, the median CD4 count increased from as low as  Table   3below showing the life table for the people that were LTFU.

Discussion
This 11 year retrospective cohort survival analysis gives insights into the length of time the ART clients stay on treatment. The average age of clients was found to be 31 years which compares well with similar studies across Africa. A similar study done in Ethiopia had a mean of 33 years [10] and another done in Cameroun had mean of 31 years [11]. The earliest patient was enrolled in 2002 and the latest in 2013. As expected, the earlier one is initiated, the less chances they had of being alive and on treatment in 2013. The cumulative dropout was 58% as the program could only maintain 42% of clients in the study. This may compare well with a 5 year retrospective cohort study done in India which reported a cumulative dropout of 37.7% [12]. The LTFU was found to account for 90% of the cumulative drop outs in this study. Further comparison with the factors associated with LTFU found similar results. Having a CD4 count less than 100 cells/µl was found to be associated with LTFU in this study and this was consistent with the results found in the 5 year cohort India study which also had CD4 <100 cells/µl associated with LTFU [12]. The cumulative deaths for this study were found to be 10%, consistent with a 5 year cohort study done in Ethiopia which also had a cumulative deaths at 10.4% [13], although other studies done in other African settings posted much higher deaths, i.e 29.7% in a study in Tanzania [14] and 23% in Cameroun [11]. This could be attributed to the fact that most of the deaths could have occurred unregistered by the program in Zambia. A large part of the LTFU in this study may have died but the program did not have that information hence the clients were classified as such. One of the major findings for Sample Vital Registration on Verbal autopsy [15], a study on the causes of community deaths in Zambia revealed that 20% of community deaths were due to HIV. It is thus expected that the large number of the LTFU clients may have died in the community without the information reaching the respective ART clinic of a health center.
This study did not follow up the clients in the community to establish if they had died or not.
The low rates of survival seen in the clients in the first year of being on ART as shown in the life table (Table 3 below) could easily be attributed to community deaths. This compares well with high mortality in other studies 29.7% and 23% in Tanzania [14] and Cameroun [11] respectively occurring in the first 12 months post commencement of ART. The two studies actually followed up the deaths in the communities which this study did not do. The Kaplan-Meier estimate also confirms these results as shown in Figure   2 below. In comparison, sex and age were found to have no significant bearing on whether the patient is LTFU or not. This finding is also similar to other studies in African patients. A study in Ethiopia also found no correlation between sex and the chances of being LTFU [10]. The striking finding was the relationship with CD4 count at the time of presentation. The lower the CD4 count at initiation, the higher the chances of LTFU and the relationship was found to be significant. For those that had CD4 count less than 200 cells/µl, their chances of survival were lower than the category with CD4 count above 200 cells/µl as shown in Figure 2 (Survival estimate for LTFU clients by baseline CD4 count). This may confirm that most of the patients deemed LTFU may have actually died out in the community without the health center in question having the information. This finding is consistent with studies across the world, more strikingly studies in the African setting of Tanzania [14] and Cameroun [11] and Ethiopia [13]. With CD4 being one of the major contributors of LTFU, the lack of increase in the median poses some major challenges to the ART program. Weight at initiation was also critical as observed in the results above. The lower the weight at initiation intuitively may suggest that the client is weaker and sicker. As it increases the chances of LTFU, it also increases the chances that client dies off in the community or in other departments of the hospital without ART clinic having the information. As observed above, these results are consistent with program data that suggest that 20% of clients are lost from the program within 12 months of initiation [18]. The findings from the study suggest the Ministry of Health lose an even higher number of clients at 37% in the first 12 months of initiation than the earlier suggested 20%. This result was consistent with a cohort study done in Ethiopia which posted as high as 38.6% LTFU within the first 12 months [13]. The ministry further loses 46% and above 50% in 24 and 36 months respectively. A sample vital registration with verbal autopsy (SAVVY) conducted in Zambia (2010)(2011)(2012) showed that over 20% of community deaths are due to HIV [15]. This can also support the study assertions that the majority of the people that are LTFU end up dead without the information reaching the health facility. The study indicated that about 10% of the clients had died. The cause of deaths was not studied in this study. The strength of this study is due to the fact that it covers the whole country with diverse scope in terms of the health centers involved. All levels of health care were part of the sample and thus can be replicated with a larger sample within the country and in other countries as well. Some weaknesses of the study included lack of more background characteristics of the patients such as economic status to link to the possible cause of low BMI and baseline CD4 count at initiation of therapy. It also did not include the interviews with the patients currently on ART to determine some challenges they may be facing while taking medication which may lead to LTFU.

Conclusion
The study has shown high rates of LTFU within the first 12 months of initiation antiretroviral of therapy in Zambia. Other studies on the same had observed high mortality in the same period thus, the majority could have died off. The rate was particularly high for those that had low CD4 count at commencement of ARVs. This was also true for those that had low weight at commencement of ART.

Competing interests
The authors had no competing interest.