Incidence Rate and Factors Associated with Antiretroviral Therapy Regimen Change among Adult HIV Patients at Ngoerah Hospital Bali, Indonesia

Abstract

Background

The human immunodeficiency virus (HIV) epidemic remains a major global health concern. While antiretroviral therapy (ART) has significantly improved survival rates, regimen changes during long-term treatment are common and may impact clinical outcomes. Data regarding the incidence rate and determinants of ART regimen changes among people living with HIV (PLHIV) in Bali, Indonesia, remain limited.

Material and Methods

This retrospective cohort study included adult PLHIV receiving ART at Ngoerah Hospital, Bali, Indonesia, from July 2004 to December 2023. A total of 265 eligible patients were selected through systematic random sampling. Multivariable Cox Regression was used to identify factors associated with regimen change.

Results

During the study period, the incidence rate of ART regimen changes was 2.22 per 100 person-years (95% confidence interval [CI], 1.760-2.802), with 3,196.7 person-years of observation, and the median follow-up duration for patients was 11.7 years (interquartile range, 11.1–13.7). The median age was 45 years, and the majority were male (64.5%). In multivariate analysis, nevirapine (NVP)-based initial regimens (adjusted hazard ratio [aHR], 1.853; 95% CI, 1.031–3.329; P=0.039) and baseline CD4 < 200 cells/µL (aHR, 1.940; 95% CI, 1.215–3.098; P=0.005) were significantly associated with regimen change. In addition, most regimen changes (87.3%) occurred after more than ten years of treatment. The primary reasons for regimen change were treatment failure (36.6%), medication stockout (23.9%), and adverse effects (21.1%).

Conclusion

NVP-based regimens and low baseline CD4 counts were key factors of regimen change. Continuous monitoring, timely detection of treatment failure, and the use of more durable regimens are essential to improve long-term treatment outcomes for the HIV patients in our outpatient clinic in Bali, Indonesia.

Graphical Abstract

INTRODUCTION

The World Health Organization (WHO) estimated that by the end of 2021, approximately 38.4 million people were living with human immunodeficiency virus (HIV). In the same year, around 650,000 people died from HIV-related causes, and 1.5 million people acquired HIV, with a global prevalence of 0.7% [1]. In Indonesia, the Ministry of Health reported 22,331 newly diagnosed people living with HIV (PLHIV) between January and June 2022, most of whom were aged 25–49 years (68.1%) [2]. In Bali Province, the cumulative number of HIV cases reached 27,114 as of March 2023 [3]. Data from the ART service clinic at Ngoerah Hospital, which has provided antiretroviral therapy (ART) since July 2004, the cumulative number of HIV cases by the end of 2023 was 2,009, with 1,769 (88.1%) individuals still regularly receiving ART.

Lifelong ART remains the mainstay of HIV management. However, regimen modification often occurs during the course of treatment. Early regimen modification has been reported to shorten the duration of viral suppression due to cross-resistance and overlapping toxicities, both within and between ART drug classes, resulting in poorer clinical outcomes and a higher risk of treatment failure [4]. The durability of first-line ART varies across regions, influenced by geography, socioeconomic status, treatment period, and drug availability [5,6,7,8,9].

In Indonesia, a total of 334,185 PLHIV had initiated ART, but by 2022, only 274,644 were still alive, and as of September 2022, only 169,767 (50.8%) remained on treatment. Among those no longer in care, 40% had died, 54% were lost to follow-up, and 6% had discontinued ART. Retention in HIV care remains a critical challenge in Indonesia [3]. Despite these national data, the Ministry of Health has not reported the proportion of PLHIV experiencing regimen changes, nor the frequency of such changes during ART. Furthermore, electronic searches revealed a paucity of publications examining the rate, causes, and associated factors of ART regimen modifications in Bali, Indonesia. Therefore, this study was conducted as a 19-year retrospective cohort analysis at Ngoerah Hospital to evaluate patterns of ART regimen changes among adult PLHIV.

MATERIALS AND METHODS

1. Study design and setting

This study employed an analytical observational design with a retrospective cohort approach. It was conducted at the outpatient “Voluntary Counselling and Testing (VCT)” Clinic at Ngoerah Hospital, the central referral hospital of Bali, which has been providing ART therapy since 2004. The study population included adults living with HIV who received ART between July 1, 2004, and December 31, 2023.

2. Ethics statement

This study was approved by the Institutional Review Board (Ethics Committee) with document No: 1582/UN14.2.2.VII.14/LT/2024 and research permit No: DP.04.03/D.XVII.2.2.2/32107/2024. This study was also conducted in accordance with the Helsinki Declaration of 1975, as revised in 2024.

3. Sample size and sampling technique

The sample size was calculated using a single-population proportion formula, with the following assumptions: marginal error (d)=0.05, Z value at the 95% confidence level=1.96, and prevalence of regimen change (p)=0.22, based on a previous study in Africa [10]. The minimum sample size required was 265 patients. Systematic random sampling was applied.

The total number of patients visiting our clinic in one month was 1,769, with an average of 70-80 visits per day. Patients who arrived with the number 1 and multiples of 7 were selected as subjects, and those chosen as samples were then traced through their medical records until the minimum required sample size was reached.

4. Inclusion and exclusion criteria

Inclusion criteria:

• 1) Adult HIV patients aged ≥18 years.

• 2) Receiving combination ART.

• 3) Routinely attending follow-up visits at the VCT clinic between 2004 and 2023.

• 4) Have at least one follow-up visit after ART initiation

Exclusion criteria:

• 1) Incomplete medical records,

• 2) Patients who have moved from another health facility and have been on ART for more than 3 months

5. Study variables and operational definition

Independent variables include sociodemographic characteristics, clinical history, baseline CD4 count, ART regimen, occurrence of regimen change, adverse drug reactions, tuberculosis coinfection, and treatment outcomes. Dependent variable including the incidence rate of regimen change and time from initiation to first regimen change.

• ● Regimen change: Any substitution/switch or discontinuation of at least one antiretroviral drug from a triple ART regimen.

• ● Time to first regimen change: the first adjustment to the ARV regimen recorded during the follow-up period since the start of highly active antiretroviral therapy (HAART).

Cohort entry point: day 1 of first ART initiation. Start of follow-up: The date of ART initiation (time zero). Censoring criteria: Patients were censored at the end of the observation period.

6. Data collection

Data were extracted from patient medical records, three healthcare professionals (one internist - infectious disease consultant, one pharmacist, and one nurse) working in the hospital, and two research assistants were recruited and trained on the aims and methods of data collection before the actual study.

7. Statistical analyses

Descriptive statistics were used to summarize baseline characteristics. The incidence rate of regimen change was calculated per 100 person-years of observation. Factors associated with regimen change were analysed using cox proportional hazards regression. Hazard ratios (HR) and adjusted hazard ratios (aHR) with 95% confidence intervals (CI) were reported. A P-value <0.05 was considered statistically significant.

RESULTS

1. Patient characteristics

Out of 300 recruited patients, 265 met the eligibility criteria and were included in the final analysis. The majority of the study population was male (64.5%), with a median age of 45 (41-50) years. Most participants were aged 18–50 years (78.5%), resided in urban areas (52.1%), and initiated ART with a nevirapine (NVP)-based regimen (69.4%).

At treatment initiation, the median baseline CD4 count was 245 cells/µL, and 39.3% of patients had a CD4 count <200 cells/µL. The majority were at WHO clinical stage IV (70.9%) at baseline. Details of sociodemographic and clinical characteristics are presented in Table 1.

Table 1. Characteristics of adult HIV patients at Ngoerah hospital 2004–2023 (n=265).

Characteristic		n (%)
Sex
	Male	171 (64.5)
	Female	94 (35.5)
Age, median (IQR)		45 (41–50)
	18–50 years	208 (78.5)
	>50 years	57 (21.5)
Residence
	Urban	138 (52.1)
	Rural	127 (47.9)
Initial ART regimen
	Efavirenz-based	80 (30.2)
	Nevirapine-based	184 (69.4)
	PI-based	1 (0.4)
	INSTI-based	0
Number of ART tablets
	1 tablet	19 (7.2)
	>1 tablet	246 (92.8)
History of regimen change
	Yes	71 (26.8)
	No	194 (73.2)
History of pulmonary TB
	Yes	10 (3.8)
	No	255 (96.2)
Baseline CD4 count
	≥200 cells/µL	161 (60.8)
	<200 cells/µL	104 (39.3)
HIV clinical stage (WHO)
	Stage I	32 (12.1)
	Stage II	18 (6.8)
	Stage III	27 (10.2)
	Stage IV	188 (70.9)
Adverse effects
	None	238 (89.8)
	Present	27 (10.2)
Treatment failure
	Yes	34 (12.8)
	No	231 (87.2)
Body weight at ART initiation
	56–90 kg	115 (43.4)
	39–55 kg	150 (56.6)
Duration of HIV at ART initiation
	<5 years	252 (95.1)
	≥5 years	13 (4.9)
ART regimen change
	Yes	71 (26.8)
	No	194 (73.2)

HIV, Human Immunodeficiency Virus; ART, antiretroviral therapy; PI, protease inhibitor; INSTI, integrase strand transfer inhibitor; TB, tuberculosis; WHO, World Health Organization.

2. Rate of regimen change

Overall, 71 patients (26.8%) experienced at least one change in ART regimen during the observation period. The total incidence rate of regimen changes in this study was 2.221 per 100 person-years (95% CI, 1.760-2.802) with a total observation of 3,196.7 person-years, the median duration of patient follow-up was 11.7 years (interquartile range [IQR], 11.1–13.7), with the shortest duration being 0.2 years and the longest being 18.9 years.

A time-to-event analysis of 265 participants (Fig. 1) revealed a relatively low and stable incidence rate in the first 10 years. In the 10-15 year period, an exponential risk elevation was observed, with the highest incidence burden was 62.0% (44 of 71 events) with an incidence rate of 7,248 per 100 person-years. Given the median follow-up of 11.7 years (IQR, 11.1–13.7), the risk spike in the second decade represents a clear clinical trend rather than a data anomaly. Meanwhile, in the >15 year period, the incidence rate peaked at 7,627 per 100 person-years, based on only 3 events in 39.3 person-years of observation with a very wide confidence interval (95% CI, 2,460–23,648), indicating data instability due to reduction of sample size toward the end of the follow-up period.

Figure 1. Cumulative incidence curve of ART regimen changes based on time since ART initiation.

ART, antiretroviral therapy.

3. Incidence by clinical characteristic factors

Incidence rates and cumulative rate of regimen change varied across subgroups. Higher cumulative rates were observed in patients residing in urban areas (65.5%) and those with baseline CD4 <200 cells/µL (63.2%) (Table 2).

Table 2. Incidence rate and cumulative rate of ART regimen change among adult HIV patients at Ngoerah hospital 2004–2023.

Characteristic		Number of events	Time at riska	Incidence rateb	95% CI (for IR)	Cumulative incidencec (%)
Total		71	3,196.7	2.221	1.760–2.802	51.2
Gender
	Male	47	2,099.2	2.239	1.682–2,980	54.7
	Female	24	1,097.5	2.187	1.466–3.262	47.1
Age
	18–50 years	58	2,513.2	2.308	1.784–2.985	57.3
	>50 years	13	683.4	1.902	1.104–3.276	25.7
Residence
	Urban	35	1,666.4	2.100	1.508–2.925	65.5
	Rural	36	1,530.2	2.353	1.697–3.261	39.5
Initial ARV regimen
	Efavirenz-based	14	980.2	1.428	0.846–2.412	16.5
	Nevirapine-based	57	2,203.2	2.587	1.996–3.354	31.8
	PI based	0	13.2
ARV Pill Burden
	1 tablet	5	225	2.222	0.925–5.339	64.2
	>1 tablet	66	2,971.7	2.221	1.745–2.827	50.4
History of pulmonary TB
	Never	68	3,086.2	2.203	1.737–2.795	24.6
	Ever	3	110.5	2.715	0.876–8.418	20.0
Baseline CD4 Count
	CD4 ≥200 cells/µL	35	1,988.7	1.760	1.264–2.451	45.8
	CD4 < 200 cells/µL	36	1,207.9	2.980	2.150–4.132	63.2
WHO HIV clinical stage
	Stage I	9	347.9	2.587	1.346–4.972	34.9
	Stage II	7	216.9	3.227	1.538–6.769	36.8
	Stage III	4	342.8	1.167	0.438–3.109	33.3
	Stage IV	51	2,289.0	2.228	1.693–2.932	43.9
Body weight
	39–55 kg	39	1,816.7	2.147	1.568–2.938	40.2
	56–90 kg	32	1,380.0	2.319	1.640–3.280	53.0
Duration of HIV Infection
	<5 Years	66	3,048.4	2.165	1.701–2.756	20.8
	≥5Years	5	148.2	3.373	1.404–8.103	24.8

^aPer 100 person-years.

^bPer 100 person-years.

^cCalculated based on Kaplan–Meier estimates at the end of the observation period.

ARV, antiretroviral; HIV, human immunodeficiency virus; CI, confidence interval; IR, incidence rate; ARV, antiretroviral; PI, protease inhibitor; WHO, World Health Organization.

4. Factors associated with regimen change

To identify the determinants of ART regimen change, a multivariable analysis was performed using the Cox proportional hazards model (backward stepwise method). The proportional hazards assumption, based on Schoenfeld residuals, was met globally (P=0.507) and for each covariate in the model (all P>0.050). Therefore, the hazard ratios generated from the multivariate Cox regression model can be interpreted validly.

Based on the final model, the factors significantly associated with a higher risk of regimen change were the use of a NVP-based initial regimen and a baseline CD4 count of <200 cells/µL (Table 3).

Table 3. Factors associated with regimen change in adult HIV patients at Ngoerah hospital 2004–2023.

Characteristic		HR	95% CI		P-value
			Lower	Upper
Gender
	Male	Ref
	Female	0.950	0.581	1.554	0.839
Age
	18–50 years	Ref
	>50 years	0.840	0.460	1.533	0.570
Residence
	Urban	Ref
	Rural	1.108	0.695	1.766	0.665
Initial ARV regimen
	Efavirenz-based	Ref
	Nevirapine-based	1.875	1.043	3.369	0.036a
	PI-based	9.621	0.000	NE	>0.999
	INSTI-based	NE	NE	NE	NE
ARV pill burden
	1 tablet	Ref
	>1 tablet	1.039	0.417	2.587	0.934
History of pulmonary TB
	Never	Ref
	Ever	1.552	0.458	4.968	0.458
Baseline CD4 count
	CD4 ≥200 cells/µL	Ref
	CD4 < 200 cells/µL	1.931	1.210	3.082	0.006a
WHO HIV clinical stage
	Stage I	Ref
	Stage II	1.187	0.440	3.200	0.734
	Stage III	0.365	0.111	1.198	0.096
	Stage IV	0.786	0.386	1.599	0.506
Body Weight
	39–55 kg	0.812	0.506	1.302	0.388
	56–90 kg	Ref
Duration of HIV infection
	<5 Years	Ref
	≥5 Years	1.983	0.789	4.982	0.145

Variables with a P-value <0.250 are eligible for inclusion in the initial multivariable Cox regression model.

^aCox Regression analysis is significant at P<0.050.

HIV, human immunodeficiency virus; HR, hazard ratio; CI, confidence interval; Ref, reference; ARV, antiretroviral; PI, protease inhibitor; INSTI, integrase strand transfer inhibitor; NE, not estimable; TB, tuberculosis.

In the multivariable analysis, patients initiated on NVP-based regimens were significantly more likely to experience a change than those on other regimens (aHR, 1.853; 95% CI, 1.031–3.329; P=0.039). Similarly, a baseline CD4 count of <200 cells/µL was a strong predictor of regimen modification (aHR, 1.940; 95% CI, 1.215–3.098; P=0.005) (Table 4).

Table 4. Factors associated with changes among adult HIV patients at Ngoerah hospital 2004 –2023.

Characteristic		Initial model				Final model
		HR	95% CI		P-value	aHR	95% CI		P-value
			Lower	Upper			Lower	Upper
Initial ARV regimen
	Efavirenz-based	Ref				Ref
	Nevirapine-based	1.679	0.926	3.045	0.088	1.853	1.031	3.329	0.039a
	PI-based	NE	0	NE	>0.999	NE	NE	NE	NE
Baseline CD4 count
	CD4 ≥200 cells/µL	Ref				Ref
	CD4 < 200 cells/µL	2.022	1.252	3.262	0.004a	1.940	1.215	3.098	0.005a
WHO HIV clinical stage
	Stage I	Ref
	Stage II	1.235	0.448	3.400	0.683
	Stage III	0.449	0.136	1.486	0.190
	Stage IV	0.856	0.412	1.780	0.677
Duration of HIV infection
	<5 Years	Ref
	≥5 Years	2.548	0.970	6.688	0.058

^aStatistically significant at P<0.050.

HIV, human immunodeficiency virus; HR, hazard ratio; CI, confidence interval; aHR, adjusted hazard ratio; Ref, reference group; NE, not estimable.

5. Characteristics of patients with regimen change

Among the 71 patients who changed regimens, most (87.3%) did so after more than five years on ART, and the majority had only one or two regimen changes (87.3%) during the study period. The primary reasons for regimen modification included treatment failure (36.6%), medication stock outs (23.9%), and adverse drug effects (21.1%), while the availability of newer drug classes accounted for 14.08% of the changes. Regarding the subsequent therapy, most patients transitioned to protease inhibitor (PI)-based (38.0%) or integrase strand transfer inhibitor (INSTI)-based (29.6%) regimens (Table 5).

Table 5. Characteristics of antiretroviral regimen changes.

Characteristic		n (%)
Time to regimen change
	<3 months	4 (5.6)
	3–6 months	2 (2.8)
	7–12 months	2 (2.8)
	1–5 years	1 (1.4)
	> 5 years	62 (87.3)
Frequency of changes
	1–2 times	62 (87.3)
	3–5 times	9 (12.7)
Regimen switching patterns
	Efavirenz to INSTI	5 (7.0)
	Efavirenz to nevirapine	2 (2.8)
	Efavirenz to PI	7 (9.9)
	Nevirapine to efavirenz	21 (29.6)
	Nevirapine to INSTI	16 (22.5)
	Nevirapine to PI	20 (28.2)
Reason for change
	Treatment failure	26 (36.6)
	Adverse effects	15 (21.1)
	Medication stockout	17 (23.9)
	Availability of new regimen	10 (14.1)
	Presence of comorbidities	3 (4.2)

This table only includes patients who experienced a change in antiretroviral regimen during the observation period (n=71).

INSTI, integrase strand transfer inhibitor; PI, protease inhibitor.

DISCUSSION

In this study, the incidence rate of ART regimen changes was 2.22 per 100 person-years (PY). This rate is higher than those reported in a multicentre study in Asia, which found a rate of 1.44 per 100 PY (95% CI, 1.35–1.54) and a study in Yemen reporting 1.8 per 100 PY [11,12]. However, our findings indicate a significantly lower incidence of regimen change than in studies from other regions. For instance, our finding is lower than studies in Ethiopia (10.11 per 100 PY; 95% CI, 8.29-12.6) [6], Kenya (18.6 per 100 PY; 95% CI, 16.2–21.8) [7], multi-centre studies in North America and Europe 14.4 per 100 PY [13], Switzerland 41.5 per 100 PY [14] and Brazil 28.3 per 100 PY [15].

Time-to-event analysis in this study shows a non-linear risk pattern, where the risk of ART regimen changes is not evenly distributed during the therapy period, there is a relatively long "safe" period in the first 10 years, followed by an exponential increase in risk after that duration, namely in years 10 to 15. Therefore, patients who receive ART for >10 years can be considered a high-risk group that requires monitoring of viral load, side effects, and support in maintaining more optimal adherence. This finding is similar to a study in Japan, in which the median duration of switching from first-line to second-line therapy was 8 years [16]. This is also in accordance with a study in Asia Pacific, which showed that retention on first-line ART in the first year of ART use before the regimen change was 87.9% in 2010–2013 [17]. However, this pattern is different from that reported by Sisay M et al., where the switch peaked in the 5th year of ART treatment [18].

On the other hand, the relatively low incidence rate and the delayed timing of regimen changes in this cohort are likely attributable to limited access to viral load (VL) testing. Consequently, regimen changes were primarily driven by clinical failure and the emergence of drug side effects. In addition, the choice of second-line drugs available is also limited; the limited availability of second-line drug options may lead clinicians to be more cautious when deciding to switch treatment regimens.

Our analysis showed that NVP-based regimens were significantly associated with higher rates of regimen change. This finding is similar to a study in Jakarta, in which ART retention was higher among patients receiving an efavirenz (EFV)-based regimen [19]. The increased risk of ART changes in NVP-based therapy is likely due to NVP side effects such as hepatotoxicity, hypersensitivity reactions, and virological failure [10,20]. Interestingly, a study in Africa found that although NVP was frequently "switched," patients taking it experienced less attrition (loss to follow-up or death) compared to those on other regimens. This indicates that the profile of patients using NVP in this study had good engagement with healthcare services [21]. In addition to these factors, the change in NVP use in our clinic is also due to the following changes in national and global policies [22,23,24,25,26]. NVP and EFV were introduced in 2004 and is the first-line therapy of choice. If first-line therapy fails, if the first line fails, the second-line guideline is lopinavir boosted ritonavir (LPV-r) [24]. In 2022, the Ministry of Health began introducing the TLD regimen (tenofovir disoproxil fumarate, lamivudine, dolutegravir) as an fixed-dose combination, and the massive transition from NVP to dolutegravir (DTG) occurred after March 2023 [26]. Meanwhile, the ART backbone that is often used includes zidovudine-lamivudine, tenofovir disoproxil fumarate-lamivudine, and abacavir-lamivudine [24,26]

This study found that low baseline CD4 levels increase the risk of changing therapy regimens. This is related to low baseline CD4 levels being significantly associated with the risk of virologic failure [27], which results in the need for changes in therapy regimens. This is aligned with research in Mexico, where ART initiation when CD4 levels are still high is more likely to achieve and maintain viral suppression [28]. On the other hand, a meta-analysis found that patients who started ART with higher baseline CD4 levels tended to have a greater risk of non-adherence than those with low CD4. Non-adherence triggered by high baseline CD4 can lead to failure of virologic suppression, even though patients started therapy in a relatively good immune condition. So the approach to this group requires optimal support for adherence [29].

The main reasons for regimen changes in this study were treatment failure and drug side effects. These results differ from other studies in Thailand and various African cohorts [7,10,18], where the dominant reason was side effects following therapy failure. The definition of treatment failure in this study is based on the clinical judgment of the internal medicine specialist at the time. When the determination is highly unique and dynamic, clinically and laboratory-supporting examinations are available at that time. The definition of failure is determined based on clinical and/or immunological and/or virological, namely; clinical failure if the emergence of new or recurrent OI (WHO clinical stage 4), or immunological failure if CD4 ≤250 cells/µL accompanied by clinical failure or persistent CD4 <100 cells/µL and or virological failure if the HIV RNA level is >1,000 copies/mL based on two HIV RNA examinations with a distance of 3–6 months [24].

Treatment failure in the study was likely related to poor adherence and limited access to laboratory testing for early detection. Laboratory access at our hospital was highly variable. During the 2013–2018 CD4 testing dominance era, access to VL testing was scarce, expensive, and required referral to a central laboratory. Selective VL testing was only performed if there was a clinical suspicion of therapy failure or immunological failure. Meanwhile, most of our patients were vulnerable and relied on government health insurance, which did not cover routine VL testing at that time.

In 2019, there was a turning point with the update of the National Guidelines for HIV Services, which shifted the focus from CD4 monitoring to routine VL monitoring. VL was established as the gold standard for monitoring treatment success. However, implementation in the field has not been optimal due to the coronavirus disease 2019 (COVID-19) pandemic, in which access was disrupted in 2020-2021 because PCR machines (such as GeneXpert) normally used for VL examinations were diverted to COVID-19 testing [22]. In 2023, the Ministry of Health required routine VL examinations at 6 and 12 months after starting ARVs, then annually for those who are stable [26]. Based on the implementation of the VL monitoring program, we hope to accelerate the development of resistance testing in cases that have virologically failed, enabling targeted regimen changes based on susceptibility test results. Therefore, the availability of resistance and susceptibility testing for practical services is both a challenge and a hope for improving service standards in the future.

Furthermore, close monitoring for toxicities and the prompt transition to more durable second-line regimens are essential to maintain long-term virologic suppression. Beyond clinical factors, systemic issues significantly influenced the regimen, including medication stockouts and the introduction of new regimens. Shortages of certain types of drugs occurred at the start of the COVID-19 pandemic in 2020, when ART distribution from the central government to local governments and ART clinics was disrupted. Consequently, many patients were switched to alternative regimens solely on the basis of local availability. This significantly contributed to regimen changes (23.9%) to regimens that were still available at the time.

Additionally, the introduction of the DTG-based fixed-dose combination (tenofovir disoproxil fumarate/lamivudine/dolutegravir) in 2022 contributed to 14.1% of the changes. These findings highlight that health system factors beyond patient characteristics are essential, underscoring the need for a reliable drug supply chain and the timely integration of newer ART options into national programs. The impact of limited drug stocks and the availability of new drugs occurred almost the same in the patient groups with CD4 <200 cells/µL and CD4 >200 cells/µL, specifically (52.9% vs. 47.1%) and (50.0% vs. 50.0%), respectively.

Following a regimen change, the most frequently prescribed drug classes in this study were PIs, EFV, and INSTIs. This is in accordance with the 2013 WHO Guidelines and national guidelines recommending the combination of atazanavir/ritonavir or lopinavir/ritonavir paired with two NRTIs as the preferred second-line strategy following the failure of a non-nucleoside reverse transcriptase inhibitor based regimen [22,23]. However, following the 2022 updates to the WHO and national recommendations, which recommended INSTIs as the first choice for second-line therapy [25,26], the use of INSTIs as a switch regimen then displaced PIs. INSTI-based medications, such as DTG, have proven highly effective in suppressing the virus with minimal side effects (well tolerated) and rarely cause treatment failure [16,17].

This study provides critical insights into the trends of ART regimen change in our outpatient clinic, from which key implications emerge. First, strengthening early HIV diagnosis and ensuring timely ART initiation may reduce the risk of regimen change linked to advanced immunosuppression. Second, the expansion of access to integrase inhibitor–based regimens, which have demonstrated superior durability and tolerability, should be prioritized. Third, health policymakers should address systemic issues, such as medication stock outs, which remain an avoidable cause of regimen changes.

The strength of this study is its comprehensive evaluation of ART regimen changes over a 19-year period. However, some limitations must be acknowledged. As with many retrospective cohort studies, the analysis relied on medical records, which resulted in non-standardized data for certain variables. Furthermore, technically, the incidence rates reported in this study represent only the sampled cohort, but are intended to estimate parameters of the clinical population. With properly conducted systematic sampling (consistent intervals and random starting points), these cohort incidence rates are unbiased estimators of our clinical population, but they are not representative of the general population (broader community).

In conclusion, this 19-year cohort retrospective study of adult PLHIV at Ngoerah hospital revealed a low incidence of ART regimen change (2.22 per 100 PY), most commonly after ten years of therapy. The use of NVP-based initial regimens and low baseline CD4 counts were independent predictors of regimen change. While treatment failure and adverse drug effects were the leading causes of regimen change, these findings underscore the importance of early ART initiation, sustained viral load monitoring, and timely switching to more durable regimens, such as INSTI-based therapy. Strengthening drug availability and ensuring continuity of care are essential to improve long-term treatment outcomes and retention in HIV care in our hospital in Bali, Indonesia.