Antiretroviral Therapy Switching Among People with HIV in the United States is not Uncommon Despite Virologic Suppression: An OPERA Cohort Study

Abstract

Introduction

Although antiretroviral therapy (ART) has substantially improved treatment-related outcomes among people with HIV (PWH), individuals may still undergo ART regimen switches for clinical and non-clinical reasons. While prior studies have focused on virologic outcomes, contemporary data describing ART regimen switching in routine care in the US, especially among virologically suppressed PWH, are limited. In an evolving HIV treatment landscape, including expanding switch strategies, this study estimated the incidence of ART regimen switching among virologically suppressed PWH and characterized switch patterns.

Methods

This observational cohort study used prospectively collected, routine electronic health records data from the OPERA Cohort. Adult PWH who were active in care and on a complete ART regimen between July 1, 2024 and June 30, 2025 were eligible. An ART regimen switch was defined as any change in antiretroviral agent, excluding pharmacokinetic boosting agents. Incidence of ART regimen switching was assessed during this 1-year period, overall and among PWH who were virologically suppressed. PWH who underwent a switch were characterized. Pre- and post-suppressed switch regimens, as well as patterns of suppressed switching, were outlined.

Results

Among 73,078 PWH who were eligible, 8188 (11%) experienced ≥ 1 ART regimen switch during the 1-year study period. Of 68,147 individuals who were virologically suppressed to < 200 copies/ml during the study period, 6888 (10%) experienced ≥ 1 ART regimen switch while suppressed. The rate of suppressed switching was 14.5 per 100 person-years. Integrase inhibitors predominated both pre- and post-switch regimens. Overall, 51% of suppressed switches resulted in regimen simplification (lower pill count, fewer anchor agents, and/or transition to a complete long-acting injectable regimen).

Conclusions

In this large, contemporary US cohort, ART regimen switching was somewhat common and most frequently occurred among virologically suppressed PWH. These real-world findings on the incidence and patterns of virologically suppressed switches provide insight into new trends in treatment optimization and the need for new therapeutic options.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40121-026-01334-3.

Key Summary Points

Why carry out this study?

While antiretroviral therapy (ART) has been successful in easing the burden of HIV, current switch rates and patterns have been understudied.

This study aimed to estimate the annual incidence of antiretroviral therapy regimen switching among people with HIV who are virologically suppressed and characterized patterns of ART regimen switching while virologically suppressed.

What was learned from this study?

Over a 1-year period, 11% of people on antiretroviral therapy experienced at least one regimen switch, most of whom (84%) had at least one switch while virologically suppressed (viral load < 200 copies/ml).

Approximately half (51%) of suppressed switches resulted in regimen simplification, defined as a lower pill count, fewer anchor agents, and/or a switch to a complete long-acting injectable regimen.

These findings highlight new trends in HIV treatment optimization and the need for new therapeutic options for people with HIV including virologically suppressed individuals who may benefit from treatment optimization.

Introduction

Human immunodeficiency virus (HIV) and acquired immunodeficiency syndrome (AIDS) were first recognized as clinical conditions in the late 1970s and early 1980s [1, 2]. Since then, the development of, and advances in, antiretroviral therapy (ART) have prolonged life expectancy for people with HIV (PWH) by decreasing both mortality and morbidity [3, 4]. While effective, use of initial standard of care antiretrovirals (ARVs), such as first-generation non-nucleoside reverse transcriptase inhibitors (NNRTIs) and protease inhibitors (PIs), were generally associated with toxicity concerns and resistance challenges [5–9]. In the last decade, newer treatment options, including second-generation integrase strand transfer inhibitors (INSTIs), have yielded improved safety and tolerability, a higher barrier to the development of resistance to ART, and an improved ability to both achieve and maintain virologic suppression [5, 10, 11].

Even with the broad successes of ART, regimen switching remains an important component of long-term HIV management. While treatment effectiveness can be an important driver of switching, regimens may be switched among virologically suppressed PWH for reasons related to simplification, side effects, adherence, tolerability, toxicity, drug–drug interactions, comorbidities, or cost [12–17]. United States (US) HIV treatment guidelines consider each of these as sufficient reasons for switching ART while virologically suppressed with the goal of maintaining virologic suppression without limiting future treatment options [12, 18]. Further, these guidelines note that PWH with no history of drug-resistance mutations or virologic failure can switch to a regimen with fewer nucleoside reverse transcriptase inhibitors (NRTIs) or to any ART regimen that has been shown to be effective in ART-naïve PWH; PWH with prior resistance may still require multiple anchor agents or multi-tablet regimens [19–23].

Despite these recommendations and the increasing availability of effective ART options, there is a notable lack of contemporary real-world evidence describing the frequency of ART regimen switching. Although many studies have estimated rates of achieving or maintaining virologic suppression after switching specific regimens, recent data characterizing overall switch rates, virologic status at time of switch, or patterns of pre- and post-switch regimens in routine clinical care in either the US or Europe are limited. Given the constantly evolving HIV treatment landscape, the predominance of second-generation INSTIs, and recent switch strategies aimed at maintaining virologic suppression or simplifying ART regimens [24, 25], the frequency of ART regimen switching remains unclear. To address this evidence gap, the objective of this study was to estimate ART regimen-switching rates among PWH in routine clinical care in the US and to characterize patterns of switching among PWH who are virologically suppressed. Understanding practices of current ART regimen switching may inform clinical decision-making and support the development of future therapeutic strategies.

Methods

Study Population

This study used prospectively collected, routine electronic health records (EHR) data from the Observational Pharmaco-Epidemiology Research and Analysis (OPERA^®) Cohort. In OPERA, all laboratory tests, diagnoses, comorbidities, demographic characteristics, and clinical observations are gathered from EHR data. At the time of the study, OPERA included data from 97 clinics across 21 US states and Puerto Rico, including large metropolitan health clinics, rural clinics, wellness centers, sexual health clinics, health departments, and federally qualified health centers.

Institutional review board (IRB) approval covering patient data contained in the OPERA database was received from Advarra IRB; a waiver of informed consent and authorization for the use of protected health information for patient data was granted (Pro00023648). The study was conducted in accordance with HIPAA and HITECH requirements, which expand upon the ethical principles detailed in the 1964 Declaration of Helsinki.

Adults with HIV who were active in care, defined as having ≥ 1 healthcare visit during the study period (July 1, 2024 to June 30, 2025), were identified. PWH who participated in a clinical trial during the study period were excluded. PWH were eligible for the analysis if they were on a complete ART regimen (i.e., ≥ 2 ARVs, including ≥ 1 anchor agent) between July 1, 2024 and June 30, 2025. PWH were followed until the first of the following events: (1) death, (2) loss to follow-up (12 months after last clinical contact), or (3) study end (June 30, 2025). PWH who experienced a switch were then identified from this population. PWH who were virologically suppressed to either < 200 copies/ml or < 50 copies/ml and experienced a switch were further identified and characterized. An individual was eligible to have multiple switches observed during the study period.

Study Variables

An ART regimen switch was defined as the addition or discontinuation of any ARV agent from the current ART regimen. Changes across or within a class of ARVs were considered as an ART regimen switch, but changes in pharmacokinetic boosting agents (i.e., ritonavir and cobicistat) were not. Based on the most recent viral load before or on the date of the switch, switches were characterized as suppressed to < 200 copies/ml (suppressed_<200) and/or suppressed to < 50 copies/ml (suppressed_<50). Viral loads measured up to six months before the start of the study and throughout the entire study period (i.e., between January 1, 2024 and June 30, 2025) were considered.

Switches were categorized as: (1) simplification of regimen (i.e., reduction in pill count, reduction in the number of anchor agents in the post-switch regimen compared to the pre-switch regimen, or switch to a complete two-drug, long-acting injectable regimen), (2) change in anchor agent across classes, (3) change in anchor agent within the same class, (4) addition of anchor agent, or (5) change in NRTI backbone agent; these categories were not mutually exclusive. Anchor agents included INSTIs, PIs, NNRTIs, capsid inhibitors (CIs), and entry inhibitors (EIs) or attachment inhibitors (AIs).

Statistical Analyses

The number and proportion of PWH who experienced any switch, a suppressed_<200 switch, and a suppressed_<50 switch were enumerated. The number of switches per person was described using means and standard deviations (SDs). Rates of ART regimen switching were estimated using univariate Poisson regression. The numerator for the rate was the total number of ART regimen switches experienced during the 1-year study period, and the denominator was cumulative person-time on ART during the study period. Rates of suppressed ART regimen switching were estimated the same way, although only person-time on ART following a viral load < 200 (suppressed_<200 switches) or < 50 copies/ml (suppressed_<50 switches) was considered in the denominator. Frequency and rate of switching, including switching while virologically suppressed, were described overall and stratified by legal sex.

Demographic and clinical characteristics of PWH who experienced an ART regimen switch were described at the person-level at the time of their first ART regimen switch during the study period. These were described among all PWH who experienced an ART regimen switch as well as in the subsets of PWH whose first ART regimen switch during the study period was a suppressed_<200 and a suppressed_<50 switch. Absolute numbers and proportions were provided for categorical variables; medians and interquartile ranges (IQRs) were provided for continuous variables.

Patterns of ART regimen switching were assessed at the switch-level, among virologically suppressed switches only. If an individual experienced multiple switches during the eligibility window, each of their pre- and post-switch ART regimens was counted as a separate event. The most common ARV combinations for each of pre- and post-switch ART regimens were described in addition to the type of switch.

Sensitivity Analyses

In the first sensitivity analysis, all analyses were repeated in the subset of the population who maintained suppression throughout the study period (i.e., PWH without any viral load measurement ≥ 200 copies/ml and ≥ 50 copies/ml, respectively). In the second sensitivity analysis, a more restrictive definition of switch was used. For this analysis, an ART regimen switch was defined as the addition of or discontinuation of any anchor ARV agent from the current ART regimen. Anchor agents included INSTIs, PIs, NNRTIs, capsid inhibitors (CIs), and entry inhibitors (EIs) or attachment inhibitors (AIs). For this sensitivity analysis, across- and within-class changes in anchor agents were counted as a switch, but changes in pharmacokinetic boosting agents or NRTI backbone agents were not.

Results

Study Population

In total, 156,924 adults with HIV were identified in OPERA, including 75,280 that were active in care and were not participants in a clinical trial (Fig. 1). Of these, 73,078 were on a complete ART regimen containing ≥ 2 ARVs, including ≥ 1 anchor agent, and were considered eligible to experience an ART regimen switch during the study period. Of these individuals, 68,147 (93%) and 65,475 (90%) were ever virologically suppressed to < 200 and < 50 copies/ml, respectively.

Fig. 1.

Inclusion in the study and frequency of ART regimen switch. ART antiretroviral therapy, JUL June, JUN June, n number, SD standard deviation

Frequency and Rate of ART Regimen Switches

Of the 73,078 eligible PWH, 8188 (11%) experienced ≥ 1 ART regimen switch during the 1-year study period (Fig. 1); 1442 (18%) of these individuals experienced multiple switches. There were a total of 10,329 switches and the rate of switching ART regimens was 15.3 (95% CI: 15.0, 15.6) per 100 person-years. The proportion of individuals experiencing a switch was comparable between men and women, although the rate of switching was slightly higher among women than men (Table 1).

Table 1.

Frequency and rate of ART regimen switch between July 1, 2024 and June 30, 2025, overall and while virologically suppressed

	PWH on ART N = 73,078	Men on ART N = 60,600	Women on ART N = 12,478
Experienced a switch, n (%)	8188 (11)	6679 (11)	1509 (12)
Total # switches, N	10,329	8416	1913
Experienced > 1 switch, n (%)	1442 (18)	1163 (17)	279 (18)
# switches per person, mean (SD)	1.3 (0.7)	1.3 (0.7)	1.3 (0.7)
Incidence rate, number of switches per 100 person-years on ART (95% CI)	15.3 (15.0, 15.6)	15.0 (14.7, 15.4)	16.7 (15.9, 17.4)
Virologically suppressed to < 200 copies/ml during study period, n (%)	68,147 (93)	56,534 (93)	11,613 (93)
Experienced a suppressed<200 switch, n (%)	6888 (10)	5662 (10)	1226 (11)
Total # suppressed<200 switches, N	8564	7025	1539
Experienced > 1 suppressed<200 switch, n (%)	1148 (17)	926 (16)	222 (18)
# suppressed<200 switches per person, mean (SD)	1.2 (0.7)	1.2 (0.7)	1.3 (0.6)
Incidence rate, number of suppressed<200 switches per 100 person-years on ART while virologically suppressed to < 200 copies/ml (95% CI)	14.5 (14.2, 14.8)	14.3 (14.0, 14.6)	15.4 (14.6, 16.1)
Virologically suppressed to < 50 copies/ml during study period, n (%)	65,475 (90)	54,317 (90)	11,158 (89)
Experienced a suppressed<50 switch, n (%)	6187 (9)	5101 (9)	1086 (10)
Total # suppressed<50 switches, N	7606	6258	1348
Experienced > 1 suppressed<50 switch, n (%)	996 (16)	54,317 (90)	11,158 (89)
# suppressed<50 switches per person, mean (SD)	1.2 (0.7)	1.2 (0.7)	1.2 (0.6)
Incidence rate, number of suppressed<50 switches per 100 person-years on ART while virologically suppressed to < 50 copies/ml (95% CI)	14.1 (13.8, 14.4)	14.0 (13.7, 14.4)	14.7 (13.9, 15.5)

ART antiretroviral therapy, CI confidence interval, n number, PWH people with HIV, SD standard deviation

Of the 68,147 PWH who were virologically suppressed to < 200 copies/ml at any point during the study period, 6888 (10%) experienced ≥ 1 suppressed_<200 switch (Fig. 1) and 1148 (17%) of them experienced multiple suppressed_<200 switches. There were 8564 suppressed_<200 switches and the rate was 14.5 (95% CI: 14.2, 14.8) suppressed_<200 switches per 100 person-years. Both the frequency and rate of suppressed_<200 switching were comparable between men and women (Table 1). PWH with a suppressed_<200 switch represented 84% of the total number of PWH who experienced any switch during the study period (6888/8188).

Among the 65,475 PWH who were virologically suppressed to < 50 copies/ml at any point during the study period, 6187 (9%) experienced ≥ 1 suppressed_<50 switch (Fig. 1) and 996 (16%) of them experienced multiple suppressed_<50 switches. There were 7606 suppressed_<50 switches and the rate was 14.1 (95% CI: 13.8, 14.4) suppressed_<50 switches per 100 person-years. Both the frequency and rate of suppressed_<50 switching were similar among women and men (Table 1). PWH with a suppressed_<50 switch represented 76% of the total number of PWH who experienced any switch during the study period (6187/8188).

Baseline Characteristics

Demographic and clinical characteristics at the time of the first regimen switch during the study period are presented in Table 2. The median age of PWH who experienced any ART regimen switch was 44 (IQR: 34, 57) years old. Overall, 82% were male, 46% Black and 42% white. Median time since HIV diagnosis was 8 years (IQR: 2, 17). Eighty-four percent of PWH who experienced a switch had at least one of the selected comorbidities. The most common types of comorbidities observed were endocrine disorders, mental health conditions, renal disease or impairment, and substance use. A median of five visits and three viral load measurements were recorded in the 12 months before switch. There was little variation in baseline characteristics, regardless of viral load before first switch.

Table 2.

Baseline^a characteristics of individuals who experienced an ART regimen switch, overall and by viral load at or before first switch

	PWH with any ART regimen switch N = 8188	PWH with viral load < 200 copies/ml before first ART regimen switch N = 6763b	PWH with viral load < 50 copies/ml before first ART regimen switch N = 6042c
Age, median years (IQR)	44 (34, 57)	44 (34, 57)	44 (34, 57)
Female sex, n (%)	1509 (18)	1203 (18)	1065 (18)
Race, n (%)
Asian	172 (2)	155 (2)	145 (2)
Black	3731 (46)	2950 (44)	2586 (43)
White	3468 (42)	2987 (44)	2715 (45)
Other/unknown	817 (10)	671 (10)	596 (10)
Hispanic ethnicity, n (%)	2164 (26)	1857 (27)	1695 (28)
US Geographic region, n (%)
Northeast	949 (12)	826 (12)	741 (12)
South	5136 (63)	4210 (62)	3768 (62)
Midwest	343 (4)	280 (4)	241 (4)
West	1702 (21)	1398 (21)	1251 (21)
US Territory	58 (1)	49 (1)	41 (1)
Payer, n (%)d
Medicaid	2431 (30)	1935 (29)	1686 (28)
Medicare	1297 (16)	1073 (16)	940 (16)
Commercial Insurance	5309 (65)	4485 (66)	4044 (67)
Ryan White/ADAP	1136 (14)	922 (14)	829 (14)
Cash	468 (6)	340 (5)	310 (5)
Other/unknown	3558 (43)	2982 (44)	2669 (44)
Years since HIV diagnosis, median (IQR)	8 (2, 17)	8 (2, 17)	8 (2, 17)
Months since initiation of last ART regimen before switch, median (IQR)	23 (10, 60)	23 (10, 61)	24 (10, 62)
Any select comorbidity (ever), n (%)	6996 (85)	5860 (87)	5247 (87)
Cardiovascular diseasee	986 (14)	842 (14)	740 (14)
Invasive cancerf	709 (10)	610 (10)	542 (10)
Endocrine disordersg	5410 (77)	4661 (80)	4210 (80)
Mental health conditionsh	3948 (56)	3248 (55)	2895 (55)
Liver diseasei	1287 (18)	1077 (18)	948 (18)
Bone disordersj	372 (5)	344 (6)	323 (6)
Renal disease or impairmentk	1692 (24)	1463 (25)	1310 (25)
Autoimmune disordersl	296 (4)	252 (4)	226 (4)
Substance use disorderm	1540 (22)	1221 (21)	1073 (20)
≥ 12 months of care in OPERA prior to first ART regimen switch during the study period, n (%)	6750 (82)	5629 (83)	5056 (84)
Has ≥ 1 outpatient healthcare visit 12 months before switch, n (%)	6604 (98)	5592 (99)	5026 (99)
Healthcare visits in 12 months before switch, median (IQR)	5 (3, 7)	5 (3, 8)	5 (3, 8)

ADAP AIDS Drug Assistance Program, ART antiretroviral therapy, IQR interquartile range, n number, PWH people with HIV, US United States

^aBaseline is defined as the date of first ART regimen switch during the study period

^b139 individuals with a suppressed_<200 switch during the study period were excluded because they had a viral load ≥ 200 copies/ml at their first switch

^c156 individuals with a suppressed_<50 switch during the study period were excluded because they had a viral load ≥ 50 copies/ml at their first switch

^dPayer categories are not mutually exclusive

^eDiagnosis of myocardial infarction, angina, other/unspecified coronary heart disease, occlusion/stenosis of precerebral arteries, stroke, transient ischemic attack, other cerebrovascular disease, peripheral arterial disease, or abdominal aortic aneurysm conditions

^fDiagnosis of lymphoma, myelodysplastic syndromes, non-Hodgkin’s lymphoma, lymphoma of the brain, myeloma, breast cancer, gynecologic cancer, GI cancer, lung cancer, GU cancer (prostate, adenocarcinoma, bladder cancer, malignant neoplasm, etc.), cervical cancer, Kaposi’s Sarcoma, or all other cancer

^gHyperlipidemia, type 1 diabetes mellitus (diagnosis), type 2 diabetes mellitus (diagnosis), hypothyroidism, hyperthyroidism, thyroiditis

^hDiagnosis of anxiety disorders, bipolar or manic disorders, major depressive disorder, schizophrenic disorder, dementia, or suicidality

ⁱDiagnosis of fatty liver, alcoholic liver disease, liver cirrhosis, hepatic failure or inflammatory liver disease, or viral hepatitis (diagnosis and/or lab; HBV, HCV: ever)

^jDiagnosis of osteoporosis, osteopenia, or pathologic fracture

^kLast eGFR < 60 ml/min/1.73 m², or a diagnosis of moderate/severe chronic kidney disease (stage 3+), end-stage renal disease, or acute kidney failure

^lRheumatoid arthritis, systemic lupus erythematosus (lupus), inflammatory bowel disease (IBD; Crohn’s disease, ulcerative colitis), multiple sclerosis, myasthenia gravis, Graves’ disease, Guillain–Barre, Hashimoto’s, Sjögren’s, vasculitis, psoriasis, chronic inflammatory demyelinating polyneuropathy

^mDiagnosis of dependence, abuse, or misuse of any drug or alcohol, injection drug use, alcoholism, or drug-seeking behavior

Patterns of Suppressed ART Regimen Switches

Characteristics of the ART regimens received immediately before a virologically suppressed switch are described in Table 3. Most pre-switch regimens for suppressed switches were three-drug regimens (3DRs; 68–69%, depending on if switch was a suppressed_<200 or suppressed_<50 switch), followed by oral two-drug regimens (2DRs; 14–14%), complete long-acting injectable regimens (9–10%), and regimens containing more than three ARVs (9–9%). Note that boosting agents did not count towards the number of agents when defining 2DRs, 3DRs, and regimens containing more than three ARVs. Regarding anchor agent classes, most pre-switch regimens (64–65%) contained an INSTI as the only anchor agent. Regimens containing anchor agents from multiple classes were also common (23–23%); of these regimens, the majority (94–94%) included an INSTI.

Table 3.

Summary of pre-switch ART regimens^a for suppressed switches, by last viral load measurement before switch

	Viral load < 200 copies/ml before switch N = 8564 switches	Viral load < 50 copies/ml before switch N = 7606 switches
Pre-switch ARV combination consistent with a single-tablet regimen, n (%)	6036 (70)	5421 (71)
Type of pre-switch regimenb, n (%)
Two-drug regimenc	1202 (14)	1067 (14)
Three-drug regimen	5841 (68)	5229 (69)
> 3 drugs in regimen	732 (9)	578 (8)
Complete long-acting injectable regimend	789 (9)	732 (10)
Pre-switch anchor agent classe, n (%)
INSTI	5479 (64)	4940 (65)
PI	612 (7)	507 (7)
NNRTI	470 (5)	437 (6)
CI	≤ 5f	≤ 5e
EI or AIf	≤ 5e	≤ 5e
≥ 2 anchor agent classesg	1999 (23)h	1718 (23)i

AI attachment inhibitor, ART antiretroviral therapy, CI capsid inhibitor, EI entry inhibitor, INSTI integrase strand transfer inhibitor, NNRTI non-nucleoside reverse transcriptase inhibitor, PI protease inhibitor

^aRegimen directly preceding a suppressed switch

^bPharmacokinetic boosting agents (e.g., cobicistat and ritonavir) are not counted as an agent

^cExcluding complete long-acting injectable regimens

^dAt the time of the study, cabotegravir + rilpivirine long-acting injections were the only complete long-acting injectable regimen

^eAnchor agent classes are mutually exclusive

^fIncludes enfuvirtide, fostemsavir, maraviroc, and ibalizumab

^fHIPAA regulations require the masking of any cells with a value of 1–5

^gIncludes cabotegravir + rilpivirine long-acting injections; there were 789 and 732 pre-suppressed_<200 and pre-suppressed_<50 switch regimens of cabotegravir + rilpivirine, respectively

^hOf 1999 pre-suppressed_<200 switch regimens containing multiple anchor agent classes, 1873 (94%) contained an INSTI

ⁱOf 1718 pre-suppressed_<50 switch regimens containing multiple anchor agent classes, 1622 (94%) contained an INSTI

The top ten most common pre-switch regimens for suppressed_<200 and suppressed_<50 switches can be found in Tables S1–S2. For both, the most common pre-switch regimens were bictegravir/emtricitabine/tenofovir alafenamide (BIC/FTC/TAF; 29–29%, depending on viral load before switch), dolutegravir/abacavir/lamivudine (DTG/ABC/3TC; 11–12%), dolutegravir/lamivudine (DTG/3TC; 10–11%), cabotegravir + rilpivirine (CAB + RPV; 9–10%), and elvitegravir/cobicistat/emtricitabine/tenofovir alafenamide (EVG/c/FTC/TAF; 7–7%).

Characteristics of the ART regimens received immediately after a virologically suppressed switch are described in Table 4. Post-switch regimens for suppressed switches were most frequently 3DRs (38–39%), but included a higher proportion of oral 2DRs (28–29%) and CAB + RPV (26–27%) than the pre-switch regimens. In terms of anchor agent classes, INSTIs were the most common (55–55%), followed by regimens containing multiple anchor agent classes (39–39%). Of the post-suppressed_<200 and suppressed_<50 switch regimens containing multiple anchor agent classes, 97% contained an INSTI. While switches involving an NRTI backbone were observed in 75–77% of suppressed switches, treatment simplification was also very common (51–52%), either through simplification of the oral regimen or switch to a complete long-acting injectable regimen.

Table 4.

Summary of post-switch ART regimens^a for suppressed switches, by last viral load measurement before switch

	Viral load < 200 copies/ml before switch N = 8564 switches	Viral load < 50 copies/ml before switch N = 7606 switches
Post-switch ARV combination consistent with a single-tablet regimen, n (%)	5049 (59)	4521 (59)
Type of post-switch regimenb, n (%)
Two-drug regimenc	2419 (28)	2194 (29)
Three-drug regimen	3315 (39)	2905 (38)
> 3 drugs in regimen	615 (7)	464 (6)
Complete long-acting injectable regimend	2215 (26)	2043 (27)
Post-switch anchor agent classe, n (%)
INSTI	4672 (55)	4205 (55)
PI	304 (4)	249 (3)
NNRTI	236 (3)	216 (3)
CI	≤ 5f	≤ 5f
EI or AIg	≤ 5f	≤ 5f
≥ 2 anchor agent classesh	3348 (39)i	2933 (39)j
Type of switch, n (%)	–	–
Regimen simplification	4361 (51)	3935 (52)
Oral ART regimen simplificationk	2146 (49)	1892 (48)
Switch to complete long-acting injectable regimend	2215 (51)	2043 (52)
Change in anchor agent across-classl	1021 (12)	882 (12)
Change in anchor agent within-class	2371 (28)	2116 (28)
Addition of anchor agent	2806 (33)	2489 (33)
NRTI backbone agent changem	6463 (75)	5848 (77)

AI attachment inhibitor, ART antiretroviral therapy, CI capsid inhibitor, EI entry inhibitor, INSTI integrase strand transfer inhibitor, NNRTI non-nucleoside reverse transcriptase inhibitor, PI protease inhibitor

^aRegimen directly preceding a suppressed switch

^bPharmacokinetic boosting agents (e.g., cobicistat and ritonavir) are not counted as an agent

^cExcluding complete long-acting injectable regimens

^dAt the time of the study, cabotegravir + rilpivirine long-acting injections were the only complete long-acting injectable regimen

^eAnchor agent classes are mutually exclusive

^fIncludes enfuvirtide, fostemsavir, maraviroc, and ibalizumab

^gHIPAA regulations require the masking of any cells with a value of 1–5

^hIncludes cabotegravir + rilpivirine long-acting injections; there were 2215 and 2043 post-suppressed_<200 and post-suppressed_<50 switch regimens of cabotegravir + rilpivirine, respectively

ⁱOf 3348 post-suppressed_<200 switch regimens containing multiple anchor agent classes, 3243 (97%) contained an INSTI

^jOf 2933 post-suppressed_<50 switch regimens containing multiple anchor agent classes, 2856 (97%) contained an INSTI

^kReduction in pill count or fewer anchor agents; does not include switches to cabotegravir + rilpivirine long-acting injectable

^lReplacement of an anchor agent class with a different anchor agent class

^mAny change (addition, removal, or switch) in NRTI backbone agent

The top ten most common post-switch regimens for suppressed_<200 and suppressed_<50 switches can be found in Tables S3–S4. The most common post-switch regimens were CAB + RPV (26–27%), BIC/FTC/TAF (24–25%), and DTG/3TC (24–24%). No other regimen exceeded 2% frequency. Sankey diagrams of switch patterns for each pre-suppressed_<200 switch regimen and each pre-suppressed_<50 switch regimen with ≥ 4% frequency are provided in Figs. 2 and 3.

Fig. 2.

Switch patterns^a for all pre-suppressed_<200 switch regimens exceeding 4% frequency. 3TC lamivudine, ABC abacavir, BIC bictegravir, CAB cabotegravir, c cobicistat, DOR doravirine, DRV darunavir, DTG dolutegravir, EVG elvitegravir, FTC emtricitabine, FTR fostemsavir, RPV rilpivirine, TAF tenofovir alafenamide, TDF tenofovir disoproxil fumarate. ^aPost-switch regimens were limited to the top ten most common regimens for each pre-switch regimens; only post-switch regimens with frequency > 5 were included

Fig. 3.

Switch patterns^a for all pre-suppressed_<50 switch regimens exceeding 4% frequency. 3TC lamivudine, ABC abacavir, BIC bictegravir, CAB cabotegravir, c cobicistat, DOR doravirine, DRV darunavir, DTG dolutegravir, EVG elvitegravir, FTC emtricitabine, FTR fostemsavir, RPV rilpivirine, TAF tenofovir alafenamide, TDF tenofovir disoproxil fumarate. ^aPost-switch regimens were limited to the top ten most common regimens for each pre-switch regimens; only post-switch regimens with frequency > 5 were included

Sensitivity Analyses

The first sensitivity analysis was restricted to PWH for which all viral loads during the study period were < 200 copies/ml (N = 46,199) and < 50 copies/ml (N = 37,348). Of the 46,199 individuals who were suppressed to < 200 copies/ml for the entire study period, 4822 (10%) experienced ≥ 1 suppressed_<200 switch and 857 (18%) of them experienced multiple suppressed_<200 switches. There was a total of 6103 suppressed_<200 switches in this population and the rate of suppressed_<200 switches was 12.8 per 100 person-years (95% CI: 12.5, 13.1). Of the 37,348 PWH who were suppressed to < 50 copies/ml for the entire study period, 3770 (10%) experienced ≥ 1 suppressed_<50 switch and 658 (17%) of them experienced multiple suppressed_<50 switches. There were 4736 total suppressed_<50 switches and the suppressed_<50 switch rate was 12.4 per 100 person-years (95% CI: 12.1, 12.7). Results from this sensitivity analysis can be found in Table S5.

The second sensitivity analysis used a more stringent definition of ART regimen switch, such that only a change in anchor agent could be counted as a regimen switch. Of the 73,078 individuals who were eligible, 7349 (10%) experienced an anchor agent switch and 1295 (18%) of them experienced multiple anchor agent switches. The overall anchor agent switch rate was 13.7 (95% CI: 13.4, 13.9) per 100 person-years. Among 68,147 individuals who were virologically suppressed to < 200 copies/ml at any point during the study period, 6139 (9%) experienced ≥ 1 suppressed_<200 anchor agent switch and 1025 (17%) of them experienced multiple suppressed_<200 anchor agent switches. There was a total of 7571 suppressed_<200 anchor agent switches and the rate of suppressed_<200 anchor agent switch was 12.8 (95% CI: 12.5, 13.1) per 100 person-years. Among 65,475 individuals who were virologically suppressed to < 50 copies/ml at any point during the study period, 5486 (8%) experienced ≥ 1 suppressed_<50 anchor agent switch and 889 (16%) of them experienced multiple suppressed_<50 anchor agent switches. There was a total of 6696 suppressed_<50 anchor agent switches and the rate of suppressed_<50 anchor agent switch was 12.4 (95% CI: 12.1, 12.7) per 100 person-years. Results from this sensitivity analysis can be found in Table S6.

Discussion

In this contemporary assessment of ART regimen switching in routine clinical care in the United States, switching was not uncommon despite virologic suppression and the predominance of INSTI-based therapy. During a 1-year period, 11% of adults with HIV experienced at least one ART regimen switch and 10% switched while virologically suppressed to < 200 copies/ml. Notably, the vast majority (83%) of switches observed occurred while the individual was virologically suppressed to < 200 copies/ml, with an overall rate of 14.5 suppressed_<200 switches per 100 person-years of follow-up while virologically suppressed. While the overall rate of switching was slightly higher among women than men, the rate of suppressed switching was numerically and statistically comparable between men and women. Approximately three-quarters of suppressed switches included an NRTI backbone agent change and around half resulted in treatment simplification (i.e., lower pill count, fewer anchor agents, or switch to a complete long-acting injectable regimen). These findings demonstrate that, even in the current era of highly effective ART, regimen switching remains a somewhat frequent occurrence of longitudinal HIV care, rather than a rare event.

Interestingly, 20% of switches were derived from PWH on 2DRs. INSTIs remained the most common anchor agent in post-switch regimens with a high proportion of INSTI 2DR post-switch regimens (e.g., DTG/3TC and CAB + RPV). Despite being a two-drug single-tablet regimen, very few PWH switched to DTG/RPV compared to CAB + RPV. While oral RPV is associated with food requirements and drug–drug interactions (e.g., proton pump inhibitors), that is not the case with intramuscular RPV injections. It therefore appears that NNRTI-based regimens remain an attractive option when such inconveniences are not present.

There are limited studies of overall ART switch rates in North America and Europe, and the existing literature tends to be slightly outdated given the current ART landscape. A recent study of hospitalized PWH in Romania estimated that 9% switched ART regimens in a 6-month time period in 2023, of which most were due to simplification [26]. Additionally, an older study using claims data from 2007 to 2011 in the US found that 12% of PWH with commercial insurance and 14% of PWH on Medicaid switched during a 1-year period [27]. Further, a cohort study from the UK reported a rate of 28 anchor agent changes per 100 person-years, and of 15 NRTI backbone changes per 100 person-years between 2012 and 2015 [28]. The observed switch rate in this study is broadly consistent with prior studies conducted in North America and Europe, despite differences in study populations, designs, and using a more recent time period, thus including more modern ART regimens. While there were no major ARV approvals within 3 years of the present analysis (other than lenacapavir, which generally has its use restricted to a small group of heavily treatment-experienced PWH), prior studies were conducted during periods with significant advancements in ART (e.g., the approval of each of first- and second-generation INSTIs). Importantly, the present study extends this literature by providing recent real-world evidence of switching ART, a period characterized by mature INSTI use, high levels of virologic suppression, and increased uptake of simplified regimens, including oral and injectable 2DRs. The persistence of switching at rates comparable to earlier eras suggests that regimen optimization, rather than treatment failure, is now a dominant driver of ART regimen modification.

Detailed characterization of pre- and post-suppressed switch ART regimens further supports this interpretation. Pre-switch regimens among virologically suppressed individuals were predominantly INSTI-based 3DRS, and most commonly included either bictegravir, dolutegravir, or elvitegravir. Oral 2DRs and CAB + RPV long-acting injections were also frequently observed prior to switching, underscoring that switches occurred even from modern, simplified regimens. Post-switch ART regimens reflected a clear shift toward further simplification. Compared with pre-switch regimens, a substantially higher proportion of PWH transitioned to oral 2DRs or to complete long-acting injectable therapy. Switching patterns suggest intentional optimization, rather than reactive changes due to loss of virologic control. The prominence of CAB + RPV and DTG/3TC among post-switch regimens further illustrates how guideline-supported simplification strategies are being implemented in real-world practice. The frequency and nature of observed switches underscore the dynamic nature of long-term HIV treatment and the ongoing need for flexible therapeutic options that support durable suppression while accommodating patient preferences, tolerability, comorbidities, and evolving models of care.

This study was conducted using data from the OPERA Cohort, which is a large, diverse cohort representative of HIV care in the US. At the time of the study, OPERA included EHR data from over 160,000 PWH, representing approximately 15% of PWH in the US [29]. This large database of routine clinical care data provided access to information on ART prescriptions and laboratory testing of a very large number of PWH to accomplish the goals of this study. Sensitivity analyses showed that the results in this study were robust to changes in the population assessed for suppressed switches (including or excluding PWH with virologic blips), as well as variations in the definition of switch (including or excluding changes in NRTI backbone).

This study is not without limitations. Data in the EHR were collected for the medical management and care of PWH, not for the primary purpose of research. Because of this, data in OPERA reflect prescriptions, rather than use, and adherence cannot be determined. Moreover, the quality of the data is dependent on the quality of record-keeping, which may vary greatly across providers. For example, regimen end dates are not always entered in the EHR when it is replaced by a new regimen, which may result in intervals with an overlap of two regimens. Regimen review and data processing algorithms have been implemented and tested to mitigate this issue. As a result, some true regimen switches may have been collapsed, thereby underestimating the switch rate. However, quality assurance review of a sample of the population suggests that this error is uncommon. Similarly, a change in ART regimens was only considered to be a switch if the combination of anchor agents or backbone NRTI agents changed. There were instances where the brand-name medications might have changed but no switch was observed if the ARV agents, excluding boosting agents, remained the same. While this represents a change in terms of pharmacy dispensation, it may not represent a biological change. Finally, reasons for switching regimens are inconsistently recorded in the EHR and were missing for the majority of switches in an exploratory analysis. Instead, types of changes from pre- to post-switch regimen have been provided; these relied solely on prescription records and can provide important context for the switches observed.

Conclusions

In this large, contemporary US cohort, ART regimen switching was not uncommon, with approximately 11% of PWH in routine clinical care experiencing an ART regimen switch and 10% of virologically suppressed PWH experiencing an ART regimen switch over a 1-year period. Importantly, most (83%) switches occurred while the individual was virologically suppressed and frequently resulted in regimen simplification, including transitions to oral 2DRs or complete long-acting injectable therapies. These real-world findings demonstrate that ART regimen switching remains an integral component of HIV care in an era where second-generation INSTIs are dominant. They also provide insight into new trends in treatment simplification and can assist in the identification of PWH who may benefit from new, simple therapeutic options or other forms of treatment optimization.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Michael D. Osterman, Laurence Brunet, and Jennifer S. Fusco share the responsibility for the design of this study. Karam Mounzer, Ricky K. Hsu, Anthony M. Mills, Gerald Pierone, Jr., Michael G. Sension, Jennifer S. Fusco, and Gregory P. Fusco contributed to the acquisition of data. Michael D. Osterman and Laurence Brunet are responsible for all analyses. All authors have contributed to the interpretation of results, have critically reviewed and approved the manuscript, and have participated sufficiently in the work to take public responsibility for its content.

Funding

This work, including the journal’s Rapid Service Fees, was supported by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA.

Data Availability

The datasets generated during and/or analyzed during the current study are not publicly available due to privacy concerns and the proprietary nature of the database but can be accessed upon reasonable request through the corresponding author to the OPERA Epidemiological and Clinical Advisory Board. Access to codes may be granted upon request with parties agreeing to privacy restrictions and technological specifications and requirements.

Declarations

Conflict of Interest

Karam Mounzer has received research grants from Gilead Sciences, Merck, Janssen, and GSK/ViiV Healthcare and honoraria for Speakers Bureau and Advisory Boards from Gilead Sciences, Merck, Janssen and GSK/ViiV Healthcare; and advisory board participation with Epividian. Michael D. Osterman, Laurence Brunet, Jennifer S. Fusco, and Gregory P. Fusco are employed by Epividian, Inc.; Epividian has had research funded by AIDS Healthcare Foundation, EMD Serono, Gilead Sciences, Janssen Scientific Affairs, LLC, Merck & Co., Theratechnologies Inc., and ViiV Healthcare. Ricky K. Hsu has received research grants from Gilead Sciences and ViiV Healthcare, speaker honoraria from ViiV Healthcare, Merck and Gilead Sciences, and advisory board participation with ViiV Healthcare, Gilead Sciences, and Epividian. Anthony M. Mills has received research funding from Gilead, ViiV, GSK, Abbott, Roche, and Merck; he has attended advisory boards for Gilead, ViiV, and Epividian. Gerald Pierone, Jr. has received research funding paid to his institution from GSK/ViiV Healthcare, NIH/DAIDS, Gilead Sciences and Precision for Medicine, and is a member of the Epidemiology and Clinical Advisory Board for Epividian. Michael G. Sension is on the Speakers Bureau for ViiV Healthcare and Gilead Sciences, and on the advisory board for ViiV Healthcare and Epividian. Sean P. Fleming and Girish Prajapati are employed by Merck & Co., Inc. This work was supported by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc.

Ethical Approval

Institutional review board (IRB) approval covering patient data contained in the OPERA database was received from Advarra IRB; a waiver of informed consent and authorization for the use of protected health information for patient data was granted (Pro00023648). The study was conducted in accordance with HIPAA and HITECH requirements, which expand upon the ethical principles detailed in the 1964 Declaration of Helsinki.