Following approval from the Institutional ethics committee, a retrospective observational study was done. It was conducted at HIV clinic and medical records section of the tertiary care hospital.

Inclusion criteria: Adult HIV positive patients who were initiated on first line ART between January 2012 and December 2014 were included in the study.

Exclusion criteria: HIV positive treatment-naive pregnant women initiated on ART during the above specified period.

Follow up: Follow up time started at initiation of ART. The study endpoint was time until either: (i) modification; (ii) completion of 12 months of therapy whichever was earliest [4].

Data collection: The case records of the selected patients were studied. No personal identification details were collected. All collected data was kept confidential. The following data was collected - age, gender, baseline haemoglobin (Hb), weight andCD4 count, WHO stage at start of therapy, initiated ART regimen, reason for modification of treatment, coexisting diseases which caused modification in ART, HIV RNA load (if available), ADRs which resulted in ART modification, modifications made- substitution/switch to second line regimen, date of modification, and relevant laboratory data.

Independent t-test was used to compare mean baseline age, weight and haemoglobin and Mann-Whitney U test for median CD4 count values between tenofovir and non-tenofovir based regimens. Factors assessed were age at start of treatment, gender, baseline weight, haemoglobin, CD4 count, ART regimen, opportunistic infections and ADRs. Association between factors and modification was done using Chi-square test. A binomial logistic regression was used to identify independent risk factors for change in regimen.

A total of 202 case records were assessed. Out of these, two patients did not come for follow up after ART initiation. So, they were excluded from analysis. Hence, case records of 200 patients were studied, of which 152 were males and 48 were females. The mean age of patients was 44.68± 11.31 years.

Out of 200 patients, 54 (27%) had a modification in treatment regimen. Of these, 40 were males and 14 females. In 46 cases, modification involved substitution whereas in the remaining eight, there was a change of entire regimen. It was most common (30 patients) in the age group 31-45 years followed by 20 patients over 45 years of age. Rest was in the age group of less than 30 years.

The reasons for modification of ART were ADRs – 43 (79.62%); treatment failure – 5 (9.25%); physician decision due to patient non compliance- 3 (5.55%) and comorbidities (tuberculosis) –3 (5.55%).

Overall, non-tenofovir based regimens were prescribed in 108 (54%) patients and tenofovir based regimens in 92 (46%) patients [Table/Fig-1]. Non tenofovir based ART included zidovudine (AZT) or stavudine-based regimens.

Overall, the median (interquartile range, IQR; 25^th, 75^th percentile) time for modification was 173 (152.25,293.50) days. The median (IQR) time required for substitution in ART due to ADRs was 175 (170,341) days. It was 340 (172.50, 351) days due to treatment failure.

This is shown in [Table/Fig-2]. The time to modification was taken as within 6 months and between 6-12 months. Age, gender, baseline haemoglobin, weight, CD4 count and regimen did not show significant relation with time to modification.

There was no significant difference in baseline characteristics between tenofovir and non tenofovir based regimens [Table/Fig-3]. Comparison with WHO clinical stage could not be done as data was missing in lot of cases.

The ADRs causing modification in ART are shown in [Table/Fig-4]. A total of 43 ADRs resulted in modification of treatment. The suspect drugs for ADRs included Zidovudine, Stavudine, Tenofovir, Nevirapine and Efavirenz. Zidovudine induced anamia was the most common ADR. Central Nervous System (CNS) side effects of Efavirenz included sleep disturbances.

There was no mortality; 2 patients were lost to follow-up following treatment initiation and they were not considered in this study analysis.

Gender wise distribution of change in regimen was studied. No statistical difference was seen between the groups [Table/Fig-5]. Change in regimen was analysed against whether patients had ADR or not. Occurrence of ADR was significantly associated with change in regimen [Table/Fig-6]. Out of the 200 patients in the study, 184 had Opportunistic Infections (OIs) whereas16 had no OIs. Occurrence of change in regimen was analysed against whether the patients had opportunistic infections or not. There was no significant difference between the two groups [Table/Fig-7]. There was no significant difference between baselines CD4 ranges versus change in regimen [Table/Fig-8].

Non tenofovir based regimen was significantly associated with change in regimen. A change in regimen was less common with tenofovir based regimens than non tenofovir regimens [Table/Fig-9].

The effect of age, gender, OIs, baseline haemoglobin, weight, CD4 count, ADRs and first line regimen on modification of ART was evaluated using binomial logistic regression. Occurrence of ADR and non-tenofovir based first-line regimens were associated with higher likelihood of modification in regimen (p<0.05). The logistic regression model was statistically significant χ²= 53.073 and p< 0.05 in predicting likelihood of modification.

The study was carried out to assess the type and factors determining modification in ART following its initiation in treatment-naive patients in a tertiary care hospital.

The age of patients enrolled on ART in the studies in India have ranged from 30-40 years, which was similar to studies reported from African countries [4]. The mean age of patients in our study was 44.68±11.31 years, which is higher than other studies where patients were in their early thirties. This could probably reflect the delay in start of ART which could be due to delayed diagnosis or patient’s reluctance for treatment. Age was not associated with treatment modification in many studies. In a Brazil study, toxicity-related modifications within one year of initiation of ART showed a trend of increase with age, though it was not statistically significant [5]. In a Kenya study, there was an increased likelihood of modification in the older population. Another study in Nigeria reported lower risk in patients of more than 35 years [6,7]. But we did not find any relation between age and likelihood of modification in this study. This could be due to the small sample size.

Gender was not a risk factor for modification in this study. Similar findings have been reported from another study in Southern India [8]. In African countries, female to male ratio showed a slight female preponderance. This may be due to the large percentage of infected females in this region [4].

Tenofovir based regimens were commonly prescribed followed by Zidovudine and Stavudine in our study. A similar finding was observed in other studies in India [9]. Studies have shown that Stavudine and zidovudine based regimens were more frequently associated with modifications, unlike Tenofovir based regimens [10].

Most of the studies have reported ADRs to be the most common cause for ART modification. Gastrointestinal disturbances were the commonest side effect (28.9%), followed by hypersensitivity (18.3%), CNS effects and hepatotoxicity (11.5%) in an European study. Efavirenz therapy was changed due to CNS effects (44.4%) and nevirapine due to hypersensitivity reaction (40.7%). Drug-related toxicity was less with Tenofovir than Zidovudine-based regimens [11]. In this study, ADRs were the most common cause (79.62%) for modification of ART, as seen in most other studies. They included anaemia, rash, gastrointestinal intolerance, lipodystrophy, peripheral neuropathy, pancreatitis, lactic acidosis, renal dysfunction, and sleep disturbances. Anaemia was the most common ADR responsible for treatment modification. Haemoglobin levels are routinely monitored in patients on Zidovudine to detect drug toxicity and to assess clinical response. Anaemia due to Zidovudine can occur at doses used for treatment– it is reversible on stoppage of therapy [12]. Zidovudine can cause gastrointestinal intolerance and lactic acidosis. Gastrointestinal tolerance rarely requires modification but if severe, it can affect compliance and needs modification. Studies in USA and Kenya have shown ART modification due to gastrointestinal disturbances [10,13]. Nevirapine was discontinued due to development of rash and hepatotoxicity. The former accounted for 9.25% of modification, unlike a study in Kenya where it accounted for 20% change in regimen of the cases. Few patients had both rash and elevated liver enzymes, which indicate that it was a hypersensitivity reaction [14]. Stavudine has been commonly prescribed because it is cheap and efficacious. But, in the NACO guidelines, it is reserved for use in special circumstances [3]. Since the study involved patients who were started on ART before 2013, a fairly good number of patients in our study were on Stavudine. Lipodystrophy and peripheral neuropathy were the main side effects of Stavudine. In another study, incidence of peripheral neuropathy and prevalence of dystrophy were 6%-37% and 10%-80% respectively [15]. Studies have shown modification of therapy may prevent progression of lipodystrophy [10]. Adverse effects related to CNS like sleep disturbances were observed with Efavirenz-containing regimens, which resulted in treatment modification in our study. Various studies have evaluated side effects of Efavirenz and the resultant discontinuation of the drug. In these studies, psychiatric symptoms were reported in 25%-40% of patients receiving Efavirenz which was the main reason for modification of therapy [16,17]. In another study, Efavirenz induced sleep disturbances was the reason for its discontinuation in 75.2% patients [18].

Renal dysfunction due to Tenofovir counted for 5.55% of modifications in our study as compared to 12% in another study [10]. According to a meta-analysis of studies done prospectively, the renal dysfunction caused by Tenofovir may not have a significant impact on the patient’s clinical condition [19]. Contrary to this, retrospective studies have reported that renal dysfunction due to Tenofovir, can affect treatment continuation [20].

Treatment failure was the next common reason (9.25%) for modification. This was also seen in a few studies conducted in Africa [10]. Though, all patients are asked to do viral load testing, very few patients actually do so because of the costs involved. Therefore, decision is based on clinical and immunological outcome. Treatment failure as a cause for modification was less as compared to other studies in India where it was 14% [21].

Opportunistic infections increase the number of medications for a patient which could also have a potential for drug-drug interactions. Initiation of antitubercular therapy led to treatment modification in our study with Nevirapine being substituted with Efavirenz, to prevent hepatotoxicity, as the patient would also be on Rifampicin. Though, this was the reason for modification in a small (three) number of patients in our study, comorbidities have been an important cause of modification in another study in India [22]. Presence of comorbidities like tuberculosis has contributed to treatment modification in India, Africa and the West. Pregnancy and Hepatitis B were other reasons for modification in other studies [10,11,22,23].

Cost was an important reason for modification in treatment in some Indian studies [21]. In this study, this was not observed. A few patients in our study were switched over to fixed dose combinations to improve compliance.

The median time to modification ranged from 40 days in India, to 8-11.8 months in USA. It was 40,151 and 406 days due to adverse effects, cost and treatment failure respectively in an Indian study [21]. The median time to treatment modification due to ADRs was slightly longer: 173 days in our study. In developed countries, it is even earlier [10]. In a study in Kenya, it was 28 months (overall), with 5 months for tuberculosis, 20 months for pregnancy, 30 months for toxicity and 34 months for treatment failure related modifications. The long period in Kenya is probably due to an inability of the patient to understand the adverse effect or failure of diagnosis by the physician [10,11,13,21,24].

There were some limitations in our study. Data like WHO stage of disease was missing in some case records. Treatment failure was based on CD4 count (immunological failure). Risk factors for individual ADRs were not assessed. Also, the duration of follow up was only for a year.

Occurrence of ADRs and use of non-tenofovir based regimens were significantly associated with modification of initial ART. ADRs may resemble symptoms of the disease itself, which can make its identification difficult. So, the treating physician needs to be vigilant, and patients should be educated and counselled about ADRs. Moreover, increased use of tenofovir should be favoured to decrease risk of modification.

Further studies can be done to assess how modifications affect treatment outcome. A comparison can be made on durability of initial ART between patients treated at ART centre and a private tertiary hospital.