Real-world adherence, persistence, and health care resource utilization among people with HIV receiving long-acting cabotegravir plus rilpivirine compared with oral antiretroviral therapy in the United States: The ABOVE Study

Abstract

BACKGROUND:

Long-acting cabotegravir + rilpivirine (CAB+RPV LA) is the first US Food and Drug Administration–approved complete long-acting regimen for people with HIV who are virologically suppressed. Long-acting CAB+RPV may alleviate adherence challenges with daily oral therapy.

OBJECTIVE:

To evaluate and compare real-world adherence, persistence, and health care resource utilization in people with HIV who switched to CAB+RPV LA vs maintained daily oral antiretroviral therapy (ART).

METHODS:

ABOVE is a retrospective US cohort study using Symphony Health Solutions Integrated Dataverse administrative claims data from January 1, 2020, to August 31, 2023. The study included people with HIV aged 12 years and older on stable oral ART and with continuous clinical activity for at least 12 months before the index date, categorized into people initiating CAB+RPV LA (LA ART cohort) and people remaining on oral ART (oral ART cohort). The index date was defined as the first injection of CAB+RPV LA or imputed for the oral ART cohort. We used standardized mortality ratio weighting based on propensity scores to balance baseline characteristics. Adherence (proportion of days covered ≥0.9 over 12 months following index) and persistence to the index regimen (days from index to earliest of treatment discontinuation or end of follow-up) were compared. A doubly robust logistic regression model estimated the adjusted odds ratio and 95% CI. Health care resource utilization including preventive care was compared using the doubly robust generalized linear model, with results reported as incidence rate ratios and corresponding 95% CIs.

RESULTS:

After applying eligibility criteria, the study included 1,245 people in the LA ART cohort (mean age, 47 years; 24.1% female). The oral ART cohort included 58,644 people (mean age, 50 years; 23.3% female). In the standardized mortality ratio–weighted analysis, the LA ART cohort had higher adherence rates (73.7% vs 30.1%) and higher persistence (median [IQR], 424 [201, 537] vs 393 [174, 431] days; P < 0.001 for all) compared with the oral ART cohort. Over 12 months, we observed significantly higher odds of adherence for the LA ART vs oral ART cohorts (adjusted odds ratio = 8.06; 95% CI = 6.62-9.81; P < 0.001). The LA ART cohort also had higher rates of vaccinations, cancer and STI screenings, and bone density testing and lower hospitalization rates.

CONCLUSIONS:

People who switched to LA ART had significantly higher 12-month adherence and persistence over the follow-up compared with people remaining on oral ART. Higher key preventive measure rates and lower hospitalization rates among people who switched to LA ART highlight increased medical engagement benefits.

Plain language summary

Adherence, or taking HIV medicine following doctors’ instructions, is important to stop HIV replication. Persistence, or how long people take medication, is also important. In this study, researchers assessed adherence, persistence, and medical visits in people with HIV taking daily pills who switched to long-acting injections vs people who kept taking pills. People switching to injections had better adherence and persistence and saw their doctor more over 1 year vs people who kept taking pills.

Implications for managed care pharmacy

People with HIV in the United States switching to long-acting cabotegravir + rilpivirine had significantly higher 12-month adherence, persistence, and preventive care rates and lower hospitalization rates compared with people remaining on oral antiretroviral therapy. Findings indicate that long-acting antiretroviral therapy may be a good option for people who would benefit from a regimen with less frequent dosing. People with HIV must obtain full HIV treatment benefits to progress toward ending the HIV epidemic.

Antiretroviral therapy (ART) prevents HIV disease progression through rapid, durable virologic suppression.¹ Over recent decades, ART has greatly increased the life expectancy of people with HIV.² Maintaining lifelong virologic suppression requires optimal ART adherence. Unfortunately, sustained, optimal adherence to daily oral ART can be difficult because of challenges related to stigma and fear of HIV status disclosure, among others.^3–6 Suboptimal adherence can lead to clinical and economic consequences for people with HIV and the health care system, such as more frequent hospitalizations and excess health care costs.^7–9

Long-acting cabotegravir + rilpivirine (CAB+RPV LA) is the first complete guideline-recommended LA ART regimen approved in the United States for maintaining virologic suppression.^6,10 In phase 3/3b studies, CAB+RPV LA administered monthly or every 2 months demonstrated noninferiority to daily oral ART for maintaining virologic suppression.^11–14 People receiving CAB+RPV LA administered monthly (pooled ATLAS [NCT02951052] and FLAIR [NCT02938520] studies) and every 2 months (SOLAR [NCT04542070]) reported significantly higher treatment satisfaction compared with people continuing daily oral ART.^13,15 Additionally, most people receiving CAB+RPV LA (89.9%-99.2%) in these studies preferred LA ART to daily oral therapy over 48 weeks.^13–15

Recent data have also demonstrated that CAB+RPV LA is beneficial in people experiencing treatment-related challenges with daily oral ART.¹⁶ Although CAB+RPV LA prescription for people with HIV who are not virologically suppressed is outside of the labeled indication,¹⁰ ongoing studies are also under way to evaluate this population.^16,17 Long-acting CAB+RPV maintained virologic suppression in a real-world cohort of people with adherence challenges while on prior oral ART, including people with viremia.^16,17 Nearly all people with viremia achieved virologic suppression with oral ART before initiating CAB+RPV LA. People with high rates of substance use and unstable housing were also able to maintain virologic suppression when receiving CAB+RPV LA. Notably, the ACTG 5359 LATITUDE study demonstrated CAB+RPV LA had superior efficacy when compared with oral ART in preventing virologic failure in people with adherence challenges. Based on these and other published data, recent recommendations from the International Antiviral Society–USA Panel indicate that CAB+RPV LA may be considered for people with viremia who are unable to consistently take oral ART and in whom other treatment options are not effective. Thus, CAB+RPV LA is an effective, less frequent treatment option for many people with HIV, including people struggling with oral ART adherence.

Real-world evidence can support clinical trial data with outcomes acquired from diverse populations being treated in clinical practice. A retrospective observational cohort study of US medical claims data conducted from July 2016 to September 2019 used proportion of days covered (PDC) to determine adherence.¹⁸ Approximately 61.3% of participants in this study demonstrated suboptimal adherence (<90% by PDC) on oral ART. Real-world data have demonstrated high CAB RPV LA adherence.

For example, in the BEYOND prospective observational cohort of people with HIV in the United States initiating CAB+RPV LA, 96.0% of injections were administered on time. Similarly, in the OPERA cohort, among 321 people receiving CAB+RPV LA, 90% of injections were administered on time in routine US clinical care.¹⁹ More data assessing LA ART adherence and persistence compared with daily oral ART will increase provider confidence in this innovative ART approach.

Studies in the United States have shown that people with HIV have higher hospitalization rates compared with people without HIV.^20,21 People with HIV may also experience an increased burden of comorbidities and associated health care resource utilization (HRU) and costs compared with people without HIV. Many of these comorbidities can be addressed through available prevention and screening measures. Clinic visits for CAB+RPV LA administration may provide additional preventive health care opportunities for people with HIV.

This US medical and prescription claims data retrospective analysis compared adherence and persistence outcomes as well as HRU between people who switched to CAB+RPV LA and those who maintained daily oral ART.

Methods

Study design and population

ABOVE was a retrospective observational cohort US study of people living with HIV using data from the Symphony Health Solutions Integrated Dataverse (ICON, Leopardstown, Dublin, Ireland) administrative claims database from January 1, 2020, to August 31, 2023 (ie, the study period). Symphony Health Solutions Integrated Dataverse is a nationally representative database comprising claims for more than 280 million US patients obtained from approximately 1.8 million prescribers.²² Diverse payer types are represented in the database, including commercial insurers, basic Medicaid, Managed Medicaid, Medicare, assistance programs, and cash payers. Data were deidentified in compliance with the Health Insurance Portability and Accountability Act. As such, ethics committee or institutional review board approval and informed consent were not required.

The study included people aged 12 years and older with at least 1 HIV-1 diagnosis (International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM: Z21, B20]) during the study period. People also needed to have stable, continuous use of the same guideline-recommended oral ART regimen for at least 3 months, with a 30-day allowable gap and 30-day grace period (for people initiating CAB+RPV LA with the optional oral lead-in [CAB 30 mg + RPV 25 mg]) (Figure 1A).

FIGURE 1.

(A) Study Design and (B) CAB+RPV LA and Oral ART Cohort Selection Flowchart

^aIncludes a 30-day allowable gap and a 30-day grace period.

^bRequired at least 1 medical claim with a diagnosis of HIV (ICD-10-CM: Z21, B20).

^cStable oral ART was defined as continuous treatment with the same guideline-recommended oral ART regimen for at least 3 months without a gap of more than 30 days.

^dThe index date for the LA ART cohort was defined as the date of first CAB+RPV LA claim. The index date for the oral ART cohort was imputed based on the distribution of time between the first claim for oral ART observed in the database and first claim for CAB+RPV LA in the LA ART cohort.

^eHIV-2 was identified by at least 1 medical claim with a diagnosis of HIV-2 (ICD-10-CM: B97.35) in the entire study period.

ART = antiretroviral therapy; CAB = cabotegravir; ICD-10-CM = International Classification of Diseases, Tenth Revision, Clinical Modification; LA = long-acting; NDC = National Drug Code; RPV = rilpivirine.

People who were stable on the same oral ART for at least 3 months were categorized into 2 mutually exclusive cohorts: (1) people initiating CAB+RPV LA injections (LA ART cohort) and (2) people remaining on oral ART (oral ART cohort). People in the LA ART cohort were identified using the National Drug Codes (NDCs) for CAB 400 mg/RPV 600 mg administered monthly (NDC: 49702025315) and CAB 600 mg/RPV 900 mg administered every 2 months (NDC: 49702024015). People in the oral ART cohort were identified using NDCs for regimens recommended by the Department of Health and Human Services (Supplementary Table 1 ^{(393.4KB, pdf)} , available in online article).⁶ The index date for the LA ART cohort was defined as the date of the first CAB+RPV LA claim after being on stable oral ART for at least 3 months. To minimize survival bias, index dates for the oral ART cohort were imputed to match the timing distribution between the first oral ART claim and the first CAB+RPV LA claim in the LA ART cohort. Proxy index dates were imputed for people in the oral ART cohort to mirror the distribution of index dates observed in the LA ART cohort, consistent with the prescription time-distribution matching approach described by Cullen et al.²³ Individuals whose treatment discontinuation occurred before their assigned index date were excluded to ensure temporal consistency between exposure and index assignment. The index window for the oral ART cohort was additionally limited to the time from first to last claims of oral ART in the database. People were required to be aged 12 years or older at the index date and have continuous clinical activity (≥1 medical or pharmacy claim) for at least 12 months before (ie, baseline period) and after the index date. People with an HIV-2 diagnosis (ICD-10-CM: B97.35) during the study period were excluded (Figure 1). People were followed from the index date to the earliest of either the end of clinical activity or the end of data availability (ie, August 31, 2023; observation period).

Study Outcomes

Adherence was evaluated over the 12-month observation period after index date using PDC, defined as the number of days a person has CAB+RPV LA or oral ART available, divided by the length of the observation period (ie, 365 days). The Pharmacy Quality Alliance (PQA) recommends the PDC as the preferred quality indicator for estimating therapy adherence for chronic diseases because it avoids overestimation of medication adherence.²⁴ To calculate PDC for CAB+RPV LA, this study used an approach reflecting the labeled dosing schedule per the US Food and Drug Administration prescribing information as accurately as possible. Long-acting CAB+RPV injection does not have formal days of supply information in claims databases. Thus, the days of supply were set to the minimum time between injections per the labeled dosing schedule (30 days for monthly use or 60 days for every-2-month use). A 7-day grace period after the target injection date was allowed as per the labeled dosing schedule. This grace period indicates that people may be administered the injection 7 or fewer days before or after the scheduled injection date. If a person missed a scheduled injection visit by more than 7 days but oral ART was used to cover between injections (as indicated by the labeled dosing schedule, which allows for oral ART for up to 2 months),²⁵ they were considered adherent during this period and days of supply of oral ART were added to the numerator of the PDC. For the oral ART cohort, PDC was measured using the days of supply information in the data. For both cohorts, PDC ranged from 0 to 1 and was dichotomized using a clinically meaningful threshold to classify adherent and nonadherent people with HIV (ie, <0.90: nonadherent vs ≥0.90: adherent).²⁶ This threshold is in line with the 2023 PQA guidelines, which defined optimal adherence for antiretroviral medications as PDC greater than or equal to 90%. As a sensitivity analysis, 0.85 and 0.80 were also used as thresholds to classify adherent and nonadherent people with HIV.

Discontinuation was defined as no subsequent claims within 60 days of exhausting the previous days of supply (ie, a 60-day grace period). The discontinuation date was set as the last days of supply of the index drug before reaching the 60-day grace period. Persistence was defined as the number of days from the index date until either the earliest of discontinuation date or end of observation period. If people received oral ART to cover missed CAB+RPV LA injections (based on claims for any oral ART between consecutive CAB+RPV LA prescriptions), they were considered to be persistent provided that a subsequent CAB+RPV LA injection occurred before the end of the 60-day grace period. Additionally, the frequency, type, and duration of oral ART used to cover missed injections was assessed. In the LA ART cohort, CAB+RPV LA treatment patterns were also evaluated.

Health care resource utilization was also compared during the on-treatment period (ie, index date to earliest of treatment discontinuation, end of clinical activity, or end of data availability). Health care resource utilization included all-cause and HIV-related hospitalizations and outpatient visits (excluding those where CAB+RPV LA was administered). Of note, the Symphony database does not include any specifications that would allow direct identification of the emergency department as the place of service. In addition, screening and preventive measures included vaccines, viral load testing, and screening for cancer, sexually transmitted infections, and bone mineral density.

Statistical Analysis

All analyses were conducted using SAS Enterprise Guide software, version 9.4 (SAS Institute, Cary, NC). The study used a standardized mortality ratio (SMR), with propensity score (PS)–based weighting estimators, to balance baseline characteristic distribution between both cohorts. The PS was calculated using a multivariable logistic regression model that included the 2 exposure groups (CAB+RPV LA vs oral ART) as the dependent variable and covariates evaluated during the 12-month baseline period (age, sex, geographical region, type of overall health insurance, year of index date, Elixhauser comorbidity score,²⁷ psychiatric comorbidities, medication use, stable oral ART regimen, baseline PDC, number of inpatient visits, and number of outpatient visits) as independent variables. Standard mortality ratio weights the oral ART cohort to better represent the LA ART cohort. This process results in an estimate of the average treatment effect among the treated population. People in the long-acting ART cohort were assigned a weight of 1, and people in the oral ART cohort were assigned a weight equal to the estimated PS ratio (1-PS). Baseline characteristic distributions were compared between weighted cohorts using standardized differences. A standardized difference greater than 10% indicated a meaningful imbalance across the cohorts.

The study used a weighted logistic regression model to compare adherence odds (ie, achieving PDC ≥ 0.9) between the LA ART and oral ART cohorts. A multivariable logistic regression model assessed adherence-associated factors among people with HIV in the LA cohort. Results for both models were reported using the odds ratio (OR) and corresponding 95% CI. Kaplan-Meier curves were generated to visualize time to discontinuation, and log-rank tests were used to compare the time-to-event distribution in both cohorts. Generalized linear models with a negative binomial distribution and log link function were used to compare screening/preventive measure rates and HRU between the cohorts. The generalized linear model was further adjusted using a doubly robust approach for covariates that remained meaningfully imbalanced after SMR weighting (standardized difference >20%), a conservative threshold applied to limit overfitting and focus adjustment on variables with the greatest residual imbalance. These covariates included index date in quarters of 2021, index date in quarters of 2022, stable ART duration, all-cause inpatient visits, HIV-related outpatient visits, viral load testing (categorical), and HIV-related health care costs. Healthcare resource utilization and cost covariates were modeled as multilevel categorical variables, with categorization determined empirically based on the observed distributions. Inpatient visits and HIV-related outpatient visits were categorized as no visits, 1 to 2 visits, or more than 2 visits during the baseline period. The number of viral load tests was categorized as no test, 1 test, 2 tests, or more than 2 tests. HIV-related health care costs were categorized as missing, less than or equal to $25,000 per patient per year (PPPY), less than or equal to $50,000 PPPY, or greater than $50,000 PPPY. Results were reported using incidence rate ratios (IRRs) and corresponding 95% CIs. For all analyses, statistical significance was indicated by P values less than 0.05.

Subgroup Analyses

Descriptive and comparative analyses for adherence and persistence were conducted in a subgroup of people who received bictegravir/emtricitabine/tenofovir alafenamide fumarate (BIC/FTC/TAF) as stable oral ART compared with the LA ART cohort. Additionally, descriptive analyses for adherence and persistence were conducted by payer type (commercial, Medicaid, Medicare, other).

Results

Characteristics of People with HIV

During the study period, 442,091 people with HIV were identified in the claims database. Of these, 304,335 had used stable ART for at least 3 months. A total of 1,245 and 58,644 eligible people were included in the unweighted LA ART cohort and oral ART cohort, respectively (Figure 1). The mean ± SD observation period was 17.2 ± 3.9 months and 13.4  ± 0.6 months for the LA ART and oral ART cohorts, respectively.

Among the people in the unweighted LA ART and oral ART cohorts, 24.1% and 23.3% were female, respectively (Table 1). The most common insurance type was commercial insurance (40.6% in the LA ART cohort, 49.6% in the unweighted oral ART cohort). Medicaid was the second most common insurance type (29.7% in the LA ART cohort, 26.7% in the unweighted oral ART cohort). Most people received an integrase strand transfer inhibitor plus 2 nucleoside reverse transcriptase inhibitors during the baseline period. The most common form of this regimen was BIC/FTC/TAF (52.2% in the LA ART cohort, 51.4% in the unweighted oral ART cohort). The LA ART cohort was younger (mean age, 46.7 vs 50.1 years) compared with the unweighted oral ART cohort. This cohort also had higher rates of depression (16.5% vs 10.0%), anxiety disorders (13.9% vs 8.9%), and obesity (14.5% vs 8.5%) during the baseline period. Imbalances between the cohorts were based on a standardized difference greater than 10%. Differences expressed between cohorts are numerical only, without testing for statistical significance.

TABLE 1.

Demographics and Baseline Characteristics

	Original (unweighted) sample			SMR-weighted samplea
	LA ART cohort (N = 1,245)	Oral ART cohort (N = 58,644)	BIC/FTC/TAF cohort (N = 30,128)b	LA ART cohort (N = 1,245)	Oral ART cohort (N = 1,275)	BIC/FTC/TAF cohort (N = 959)b
Age, mean (SD), yearsc,d	46.7 (12.6)	50.1 (13.0)	49.1 (13.3)	46.7 (12.6)	46.8 (13.5)	45.9 (13.7)
Sex, n (%)e
Female	300 (24.1)	13,683 (23.3)	6,761 (22.4)	300 (24.1)	306 (24.0)	216 (22.5)
Male	943 (75.9)	44,961 (76.7)	23,367 (77.6)	945 (75.9)	969 (76.0)	743 (77.5)
Region, n (%)e
South	547 (43.9)	26,305 (44.9)	13,406 (44.5)	547 (43.9)	535 (42.0)	390 (40.6)
Northeast	376 (30.2)	15,529 (26.5)	7,877 (26.1)	376 (30.2)	397 (31.1)	317 (33.0)
Midwestd	158 (12.7)	9,566 (16.3)	5,082 (16.9)	158 (12.7)	179 (14.1)	124 (12.9)
West	151 (12.1)	6,800 (11.6)	3,561 (11.8)	151 (12.1)	150 (11.8)	120 (12.5)
Unknown	13 (1.0)	444 (0.8)	202 (0.7)	13 (1.0)	13 (1.1)	9 (0.9)
Insurance plan type, n (%)e
Commerciald	505 (40.6)	29,095 (49.6)	14,419 (47.9)	505 (40.6)	497 (39.0)	379 (39.5)
Medicaid	370 (29.7)	15,685 (26.7)	8,761 (29.1)	370 (29.7)	373 (29.3)	287 (29.9)
Medicared	281 (22.6)	10,077 (17.2)	4,971 (16.5)	281 (22.6)	285 (22.4)	209 (21.8)
Assistance programsd	49 (3.9)	3,723 (6.3)	1,935 (6.4)	49 (3.9)	48 (3.8)	37 (3.9)
Cash paymentsd,f	40 (3.2)	64 (0.1)	42 (0.1)	40 (3.2)	71 (5.6)	47 (5.0)
Most prevalent comorbidities, n (%)g
Hypertension	322 (25.9)	14,140 (24.1)	7,236 (24.0)	322 (25.9)	293 (23.0)	209 (21.8)
Depression disordersd	205 (16.5)	5,853 (10.0)	3,212 (10.7)	205 (16.5)	202 (15.9)	159 (16.5)
Obesityd,f	180 (14.5)	5,007 (8.5)	2,520 (8.4)	180 (14.5)	116 (9.1)	82 (8.6)
Anxiety disordersd	173 (13.9)	5,230 (8.9)	2,822 (9.4)	173 (13.9)	173 (13.6)	138 (14.4)
Substance-related and addictive disorders	168 (13.5)	6,865 (11.7)	3,938 (13.1)	168 (13.5)	189 (14.8)	136 (14.2)
Most prevalent comedications, n (%)h
Antihyperlipidemicsf	388 (31.2)	18,739 (32.0)	9,099 (30.2)	388 (31.2)	340 (26.6)	239 (24.9)
Antihypertensivesf	328 (26.3)	15,509 (26.4)	7,768 (25.8)	328 (26.3)	272 (21.3)	198 (20.7)
ACE inhibitors	155 (12.4)	8,906 (15.2)	4,480 (14.9)	155 (12.4)	149 (11.6)	112 (11.7)
Antidiabetics	152 (12.2)	7,485 (12.8)	3,779 (12.5)	152 (12.2)	142 (11.2)	100 (10.4)
Stable oral ART regimen
INSTI + 2 NRTIs	942 (75.7)	46,213 (78.8)	30,128 (100)	942 (75.7)	961 (75.4)	959 (100)
BIC/FTC/TAF	650 (52.2)	30,128 (51.4)	30,128 (100)	650 (52.2)	638 (50.0)	959 (100)
NNRTI + 2 NRTIs	121 (9.7)	6,457 (11.0)	—	121 (9.7)	119 (9.4)	—
INSTI + 1 NRTId	132 (10.6)	3,704 (6.3)	—	132 (10.6)	126 (9.9)	—
Boosted PI + 2 NRTIs	50 (4.0)	2,270 (3.9)	—	50 (4.0)	69 (5.4)	—
Duration of baseline stable oral ART regimen, mean (SD), monthsd	9.4 (3.2)	9.8 (3.2)	9.7 (3.2)	9.4 (3.2)	10.0 (3.0)	10.0 (3.0)
Baseline adherence, assuming oral lead-ini
PDC, mean (SD)j	0.8 (0.2)	0.8 (0.3)	0.8 (0.3)	0.8 (0.2)	0.8 (0.3)	0.8 (0.3)
Adherent proportion (≥0.9 PDC), n (%)	795 (63.9)	36,907 (62.9)	18,549 (61.6)	795 (63.9)	807 (63.3)	615 (64.2)

^a Covariates included in the propensity score used to generate SMR weights were age (continuous), gender, type of health insurance, region, year of index date, Elixhauser overall comorbidity (categorical), substance-related and addictive disorders, depression disorders, anxiety disorders, ACE inhibitor use, stable oral ART regimen, baseline PDC assuming oral lead-in, inpatient visits (continuous), and outpatient visits (continuous).

^b The BIC/FTC/TAF cohort is a subset of the oral ART cohort.

^c Age at index date.

^d The absolute standard difference between the LA ART and oral ART cohorts for this variable was greater than 10% before weighting and was considered to be an imbalance between the cohorts.

^e Evaluated during the 12-month baseline period as close to the index date as possible.

^f The absolute standard difference between the LA ART and oral ART cohorts for this variable was greater than 10% after weighting and was considered to be an imbalance between the cohorts.

^g Evaluated during the 12-month baseline period, excluding the index date.

^h Evaluated during the 12-month baseline period. Medication use as a baseline covariate was defined as at least 1 claim during the baseline period.

ⁱ Baseline adherence in the LA ART cohort was calculated assuming people with HIV had 30 days of oral lead-in with CAB and RPV.

^j PDC was defined as the number of days for which a person had oral ART available, beginning from the date of first dispensing of the drug of interest until their latest dispensing (plus days of supply), divided by the length of the follow-up period (12 months).

ACE = angiotensin-converting enzyme; ART = antiretroviral therapy; BIC = bictegravir; CAB = cabotegravir; FTC = emtricitabine; INSTI = integrase strand transfer inhibitor; LA = long-acting; NNRTI = non-nucleoside reverse transcriptase inhibitor; NRTI = nucleoside reverse transcriptase inhibitor; PDC = proportion of days covered; PI = protease inhibitor; RPV = rilpivirine; SMR = standardized mortality ratio; TAF = tenofovir alafenamide fumarate.

The SMR-weighted sample included 1,245 people in the LA ART cohort and 1,275 people in the oral ART cohort. At the index date, people in both cohorts were receiving their baseline oral ART regimen for 9 to 10 months on average. Over the baseline period, adherence (proportion with ≥0.90 PDC) was similar across both cohorts after weighting. Adherence was 63.9% in the weighted LA ART cohort (assuming oral lead-in use), 63.3% in the weighted oral ART cohort, and 64.2% in the weighted BIC/FTC/TAF subgroup. Overall, after SMR weighting, the baseline characteristics were well balanced between the cohorts (ie, standardized mean differences did not exceed 10% for most variables).

CAB+RPV LA Treatment Patterns

During the observation period, most people in the LA ART cohort received CAB+RPV LA every 2 months (57.6%) or switched from monthly to every-2-month dosing (29.2%) (Supplementary Table 2 ^{(393.4KB, pdf)} ). A median of 8 (IQR = 5-10) CAB+RPV LA injections were administered per person. For the 258 people who received oral ART to cover missed injections, BIC/FTC/TAF (53.9%) and dolutegravir/lamivudine (15.5%) were most commonly used.

Adherence

In the SMR-weighted sample, a higher proportion in the LA ART cohort (73.7% [918/1,245]) were adherent (≥0.90 PDC) over 12 months compared with the oral ART cohort and BIC/FTC/TAF subgroup (30.1% and 29.5%, respectively; both P < 0.001) (Table 2). Higher adherence rates in the LA ART cohort compared with the oral ART cohort were also observed with greater than or equal to 0.85 PDC (76.5% vs 41.9%; P < 0.001) or greater than or equal to 0.80 PDC (78.8% vs 49.9%; P < 0.001) thresholds. Adherence was generally similar across payer types in both the unweighted LA cohort (commercial, 70.5%; Medicaid 75.7%; and Medicare, 77.9%) and the unweighted oral ART cohort (commercial, 40.1%; Medicaid, 26.5%; Medicare, 37.4%).

TABLE 2.

Descriptive Analysis of Adherence To ART Over 12-month Observation Period (SMR-Weighted Sample)

	LA ART cohort (N = 1,245)	Oral ART cohort (N = 1,275)	P valuea	BIC/FTC/TAF cohort (N = 959)	P valuea
Adherence, n (%)b	918 (73.7)	383 (30.1)	<0.001c	283 (29.5)	<0.001c
PDC, mean (SD)	0.86 (0.26)	0.69 (0.29)	<0.001c	0.68 (0.29)	<0.001c

^a P values were generated using a Wilcoxon rank sum test where medians are reported and chi-square test to compare proportions.

^b Adherence was defined as greater than or equal to 0.90 PDC. Adherence was calculated allowing oral ART coverage between LA ART injections (any oral ART qualified as oral ART coverage). In the LA ART cohort, 156 people (13%) received oral ART between subsequent LA ART injections.

^c P values less than 0.05, which were considered significant.

ART = antiretroviral therapy; BIC = bictegravir; FTC = emtricitabine; LA = long-acting; PDC = proportion of days covered; SMR = standardized mortality ratio; TAF = tenofovir alafenamide fumarate.

In the SMR-weighted logistic regression model, the LA ART cohort had 8-times higher odds of adherence compared with the oral ART cohort (adjusted OR = 8.06; 95% CI = 6.62-9.81; P < 0.001). Odds of adherence were also higher in the LA ART cohort than the BIC/FTC/TAF oral ART subgroup (8.19; 95% CI = 6.62-10.14; P < 0.001) (Table 3). Long-acting CAB+RPV adherence–associated factors (P < 0.05) included insurance plan type, substance-related and addictive disorders, and baseline adherence (Supplementary Table 3 ^{(393.4KB, pdf)} ) in the LA ART cohort.

TABLE 3.

Comparison of Adherence in the LA ART Cohort vs the Oral ART Cohort and BIC/FTC/TAF Cohort

Factor	OR (95% CI)	P valuea
Adherence (reference: oral ART cohort)b
Unadjusted	5.08 (4.47-5.77)	<0.001c
Doubly robust SMR-weightedd,e	8.06 (6.62-9.81)	<0.001c
Adherence (reference: BIC/FTC/TAF cohort)b
Unadjusted	5.01 (4.40-5.69)	<0.001c
Doubly robust SMR-weightedd,e	8.19 (6.62-10.14)	<0.001c

^a P values less than 0.05 were considered significant.

^b Adherence was defined as greater than or equal to 0.90 PDC.

^c Indicates significant P values.

^d Covariates included in the propensity score used to generate SMR weights were age (continuous), gender, type of health insurance, region, year of index date, Elixhauser overall comorbidity (categorical), substance-related and addictive disorders, depression disorders, anxiety disorders, ACE inhibitor use, stable oral ART regimen, baseline PDC assuming oral lead-in, inpatient visits (continuous), and outpatient visits (continuous).

^e The weighted logistic regression model was further adjusted for covariates with a standardized difference of greater than 20% after SMR weighting, which included index date in quarters of 2021 (categorical), index date in quarters of 2022 (categorical), stable ART duration (continuous), all-cause inpatient visits (categorical), HIV-related outpatient visits (categorical), viral load testing (categorical), and HIV-related health care costs (categorical), using a doubly robust approach.

ACE = angiotensin-converting enzyme; ART = antiretroviral therapy; BIC = bictegravir; CAB = cabotegravir; FTC = emtricitabine; OR = odds ratio; SMR = standardized mortality ratio; TAF = tenofovir alafenamide fumarate.

Persistence

In the SMR-weighted sample, the LA ART cohort had higher persistence (median [IQR], 424 [201, 537] days) compared with the oral ART cohort (393 [174, 431] days; P < 0.001). Persistence was also higher in the LA ART cohort than the BIC/FTC/TAF cohort (396 [162, 432] days; P < 0.001) (Supplementary Figure 1A ^{(393.4KB, pdf)} ). In the unweighted sample, persistence results were generally similar across payer types among the LA ART cohort and the oral ART cohort (Supplementary Figure 1B ^{(393.4KB, pdf)} ).

HCRU and Preventive Measures

In the SMR-weighted sample, the LA ART cohort had higher rates of all-cause outpatient visits (IRR = 1.08; 95% CI = 1.02-1.16; P < 0.001) and HIV-related outpatient visits (1.38; 95% CI = 1.23-1.54; P < 0.001) compared with the oral ART cohort (Supplementary Figure 2A ^{(393.4KB, pdf)} ). Conversely, the LA ART cohort had a lower incidence of all-cause hospitalizations (0.37; 95% CI = 0.22-0.63; P < 0.001) and HIV-related hospitalizations (0.40; 95% CI = 0.19-0.86; P < 0.001) compared with the oral ART cohort. The LA ART cohort was also more likely to have received vaccinations (IRR = 1.25; 95% CI = 1.18-1.33; P < 0.001) compared with the oral ART cohort. These vaccines included influenza (1.40; 95% CI = 1.25-1.56; P < 0.001) and COVID-19 (1.14; 95% CI = 1.05-1.24; P < 0.001). The LA ART cohort also had higher rates of several preventive screenings compared with the weighted oral ART cohort. This included cancer screenings (IRR = 1.17; 95% CI = 1.07-1.28; P < 0.001), sexually transmitted infection screenings (1.24; 95% CI = 1.09-1.40; P < 0.001), and bone mineral density tests (2.37; 95% CI = 1.64-3.42; P < 0.001) (Supplementary Figure 2B ^{(393.4KB, pdf)} ).

Discussion

Adherence and persistence are critical aspects of long-term HIV treatment success. This study found that while baseline adherence was similar among people on stable oral ART who did and did not switch to LA ART, switching to LA ART was associated with significantly improved adherence. Conversely, remaining on oral ART, including BIC/FTC/TAF, was associated with decreased adherence over 12 months of follow-up. Similarly, persistence over 12 months was significantly greater among people who switched to LA ART compared with people who remained on oral ART, including BIC/FTC/TAF.

People who switched to LA ART had increased all-cause and HIV-related outpatient visits and increased preventive HRU compared with people who remained on oral ART. These preventive health care measures included vaccine administration, cancer and sexually transmitted infection screenings, and bone mineral density assessments. Strikingly, people who switched to LA ART had fewer hospitalizations, despite more frequent outpatient visits, compared with people who remained on oral ART. This pattern may reflect the structure of LA ART care delivery, which requires regular clinic visits for injections, These recurring visits not only provide opportunities for medication administration but also serve as consistent touchpoints for engagement, allowing providers to address preventive health needs, monitor comorbidities, and identify issues early—potentially reducing the likelihood of hospitalizations. Moreover, prior studies have demonstrated an inverse relationship between treatment adherence and HRU, with better adherence linked to fewer emergency department visits and hospitalizations.^20,26 The higher adherence typically associated with LA ART may therefore contribute to improved overall health outcomes and reduced inpatient utilization in this population.

The high adherence and persistence that the ABOVE study exhibits reinforces LA ART benefits. The ABOVE results are also consistent with a recent systematic literature review, which included ABOVE 6-month results and 15 other studies reporting high CAB+RPV LA adherence in real-world US settings. Of these 15 other studies, 5 specifically reported CAB+RPV LA adherence outcomes in people who were virologically suppressed at baseline. These 5 studies defined adherence as injections occurring within 7 days of the target injection date, with 89% to 91% administered on time.

Adherence to ART can be influenced by a number of factors such as comorbidities, side effects, stigma, and lifestyle factors.^4,28 Long-acting ART may play an even bigger role in improving clinical outcomes and reducing health care costs among people who are experiencing adherence challenges with oral ART. For example, the phase 3 LATITUDE study reported high CAB+RPV LA adherence (93% of injections received on time) from an interim analysis among people who had experienced prior adherence challenges when receiving oral ART. These participants also had fewer virologic failures, treatment-related failures, and permanent treatment discontinuations compared with participants using standard-of-care oral ART. A recent real-world study reported findings consistent with LATITUDE, demonstrating that long-acting ART can be effective among people with HIV who have a history of adherence challenges to oral therapy.²⁹ Poor ART adherence is closely linked to poorer health outcomes and increased HRU and associated costs.³⁰ Thus, the improved adherence with CAB+RPV LA can reduce avoidable HRU, lower overall costs, and ultimately contribute to improved patient outcomes. Overall, the high adherence and persistence observed in the present study, in combination with other clinical trial and real-world adherence data, provide confidence in real-world clinical CAB+RPV LA use.

Adherence definitions and reporting in real-world CAB+RPV LA studies are highly variable.³¹ Proportion of days covered is the preferred adherence measure for claims data adopted by the PQA.²⁴ The PDC threshold represents the adherence level above which a treatment is reasonably likely to achieve the most clinical benefit. The 2023 guidelines from the PQA define a PDC threshold of greater than or equal to 90% for optimal ART adherence, which was the adherence threshold for this study. Furthermore, a systematic review found that adherence thresholds of greater than 90% and greater than 95% were similarly consistently associated with virologic suppression, whereas findings were inconsistent with adherence thresholds less than 90%.³²

We acknowledge that excluding individuals whose discontinuation preceded their imputed index date may introduce selection bias; however, the number of such exclusions was minimal and unlikely to affect comparability between cohorts. The use of index-date imputation based on the distribution of the LA ART cohort helps align follow-up start times between treatment groups and reduce survival bias in real-world comparative studies.

The data source and methodology this study used offer several strengths. ABOVE was the first study to determine CAB+RPV LA adherence and persistence using a large, nationally representative, administrative claims database with longitudinal follow-up and data across the primary payer channels (commercial, Medicare, and Medicaid).²² Robust statistical methods such as SMR weighting were used to address confounding. Furthermore, the study’s inclusion of a broad US population beyond what is typically studied in randomized control trials may provide additional CAB+RPV LA adherence information and inform routine clinical practice.

Limitations

This study has limitations that are inherent to claims database studies. Administrative claims data may be subject to inaccuracies and missing data due to errors in diagnoses coding. Additionally, the oral ART data reflect that medications were dispensed, not necessarily consumed or administered, potentially resulting in an overestimation of adherence or persistence. Furthermore, although SMR weighting was used to account for potential differences between cohorts, the possibility of unmeasured confounding cannot be excluded. The underlying reasons for switching to LA ART also cannot be identified in claims data, and these factors may influence outcomes such as adherence and persistence. Greater proportions of participants in the unweighted LA ART cohort had depression and anxiety disorders at baseline compared with the oral ART cohort. It is also unknown whether these factors influenced ART choice and adherence. Additionally, the Symphony database does not include any specifications that would allow for identification of the emergency department as a place of service for HRU. Although clinical outcomes such as viral load were not included in the present study, adherence and persistence are suitable proxies for viral suppression given the known effectiveness of these drugs.

Conclusions

These data demonstrate significantly higher 12-month adherence and persistence among people with HIV in the United States switching to CAB+RPV LA compared with people remaining on oral ART. The BIC/FTC/TAF cohort demonstrated similar adherence and persistence to that observed in the overall oral ART cohort. We also noted similar differences in adherence and persistence regardless of payer type. People switching to CAB+RPV LA also had higher rates of key preventive health care measures, more frequent outpatient visits, and lower hospitalization rates. These data suggest the benefits of increased medical engagement while receiving CAB+RPV LA. We observed adherence and persistence advantages with LA ART use. Future studies incorporating clinical endpoints are needed to confirm the broader benefits of LA ART.

Disclosures

Ms Garris is an employee of ViiV Healthcare and may own stock in GSK. Dr Teichner is an employee of ViiV Healthcare and may own stock in GSK. Dr Overton is an employee of ViiV Healthcare and may own stock in GSK. Dr DerSarkissian is an employee of Analysis Group, which was contracted by ViiV Healthcare to perform this analysis. Dr Chang is an employee of Analysis Group, which was contracted by ViiV Healthcare to perform this analysis. Ms Clear is an employee of Analysis Group, which was contracted by ViiV Healthcare to perform this analysis. Ms Chen is an employee of Analysis Group, which was contracted by ViiV Healthcare to perform this analysis. Ms Liu is an employee of Analysis Group, which was contracted by ViiV Healthcare to perform this analysis. Dr Desai is an employee of Analysis Group, which was contracted by ViiV Healthcare to perform this analysis.