Since the choices of regimen are still limited in most of low- and middle-income countries, well-managed first line cART is essential. Repeated investigation of the incidence of regimen modification and its determinants will help to keep patients on the first cART regimen as long as possible. Over the median follow-up time of 21 months (IQR 6–38), 62 (14.5%) of the patients in this cohort modified their initial antiretroviral regimen. This result is nearly in line with study conducted in Swaziland (7) but the proportion is somewhat lower than that reported from previous studies (10) and also studies in Ethiopia (12, 16). The probable reason for this may be explained as the previous studies had included stavudine in their first line regimen which may be the most common contributor for regimen modification due to drug toxicity but this study excluded stavudine due to its phase out during the study period.
The rate of initial regimen change among adult HIV patients on cART was found to be 7.66/100PY (95% CI 5.84, 9.50 PY). This finding is lower than a study conducted in Swiss 41.5/100PY (8), Brazil 28.3/100PY (17), multicenter study in North America and Europe 14.4/100PY (18)and Thailand 13.8/100PY (9). This might be explained by the difference in defining outcome variables, since in this study treatment discontinuation was not considered as regimen modification unless they restart with different regimen. Furthermore, limited combined antiretroviral options in our setting may limit the clinician decision on cART modification due to treatment failure. The other possible reasons might be regular monitoring of viral load for treatment response in developed countries might pick virological failure earlier which calls the need for regimen change.
Similarly, it is lower than studies done in Kenya and West Africa with a rate of 18.6/100PY and 16.2/100PY respectively (13, 19). This might be due to the difference in follow up period 10.7 and 15 months in Kenyan and West Africa study but 21 months for our study. In addition to this, our study included participants who started cART after 2012 in which WHO recommended to phase out D4T but Kenyan and West Africa studies were done before 2011 which might overestimate the rate.
The main reason for regimen modification was toxicity which accounts about 45 (72.6%) of the cases and contribute for 5.56/100 PY (95% CI 3.98–7.14) followed by treatment failure 13 (21.0%) and tuberculosis 4 (6.5%). The common types of toxicities were anemia 23 (51.1%), rash 13 (28.9%), CNS toxicity 7 (15.5%), and renal failure 2 (4.4%). This is also similar with other studies (20)done in Mekelle (12, 16, 21) and Gondar(12).
In this study the baseline WHO clinical stage III/IV at initiation, co-medication with cART, AZT based initial NRTI backbone, and baseline hemoglobin of the patient was found to be predictors for initial regimen modification.
Those patients who were WHO clinical stage III/IV at the initiation of cART were 2.39 times at higher risk of changing their initial regimen as compared to those with WHO clinical stage I/II. This finding is also supported with studies done in Swaziland (7) and two Kenyan studies (13, 20) and study done in Gondar (12). This might be due to the fact that those patients who had advanced disease are likely to be on other medications which might result in drug interaction, side effect which in turns result in regimen modification.
Regarding the initial NRTI backbone, patients initiated with AZT based cART regimen had 8.19 times greater chance of changing their initial cART regimen compared to those initiated with TDF based cART regimen. AZT based regimen was the dominant regimen to cause initial cART regimen modification due to its hematological toxicity and this finding was harmonized with a number of different studies (22, 23). This result is also in agreement with study done in Mekelle (16). This might be justified possibly with those patients initiated with AZT based regimen were at the higher risk for developing hematological adverse effects and may result in regimen modification when compared with TDF based regimen as initial NRTI backbone.
Patients who were taking other concurrent medications with cART treatment were 1.73 times at a greater risk of changing their initial regimen at any time as compared to those who did not take other medication. This result shows that not to use additional drugs up on the cART regimen have a protective effect for treatment regimen modification. This study is also in line with Swiss HIV Cohort Study (8) and studies conducted in Ethiopia at Mekelle hospital (16) and Gondar Referral hospital (12). The possible reason for this may the need of co-medication particularly for comorbidities in patients with advanced disease and concomitant treatment of opportunistic infections, may cause drug-drug interactions, leading to an increase in transaminase levels and thus treatment modification as a result of drug-drug interactions and cumulative drug toxicity which may finally leads to regimen modification. The other possible explanation might be poly pharmacy which could lead to poor adherence due to pill burden which in turn resulted in poor efficacy of treatment result in regimen modification secondary to treatment failure.
Patients having the low level of hemoglobin during the initiation of cART are at the increased risk for the modification of the regimen. Those patients with hemoglobin < 7 g/dl were 6.32 times and those having between 7-9.9 g/dl were 4.21 times at increased risk of initial regimen modification. Even though this is not a significant predictor from previous studies, the reason might be explained as those patients with lower hemoglobin level were at higher risk of developing anemia which is one of the common reason for the treatment modification.