Treatment patterns, health care resource utilization, and costs of heavily treatment-experienced people with HIV in the United States

Abstract

BACKGROUND:

Although advancements in antiretroviral therapy (ART) have significantly improved outcomes for people with HIV (PWH), there remains a subset of PWH who are heavily treatment experienced (HTE) and at increased risk for poor outcomes, including AIDS-related complications and mortality.

OBJECTIVE:

To describe the clinical characteristics, treatment patterns, health care utilization, and costs associated with HTE among PWH.

METHODS:

Treatment-experienced PWH who had at least 2 ART lines of therapy (LOTs) between January 1, 2015, and December 31, 2022, were identified from the Veradigm Network electronic health record (EHR)–linked claims database. HTE PWH met at least 1 of 4 HTE-defining criteria based on ART indication and exposure, evidence of viremia, and/or ART resistance (index date = earliest criterion met). Patients were eligible for study inclusion if they were aged 18 years or older, had a prior HIV diagnosis, and had continuous claims enrollment and EHR activity for at least 6 months before and after the index date. A matched comparator group of treatment-experienced non-HTE PWH was also identified. Treatment patterns were measured for HTE PWH in a variable follow-up period lasting until the end of continuous enrollment or December 31, 2022. HIV-related clinical measures in the 6-month follow-up, and health care resource utilization and costs in the variable-length follow-up, were measured for both HTE PWH and non-HTE PWH.

RESULTS:

Among the 19,221 HTE PWH identified in the Veradigm dataset, 2,507 met all remaining study criteria. On average, HTE PWH were aged 50.6 years and predominantly male (63.3%), Black (43.6%), and insured through Medicaid (56.7%). Over a mean 3 years of follow-up, more than 50% of HTE PWH had ART on hand for at least 75% of the follow-up period (median percentage of days = 75.4%); however, treatment modifications were common, as 58.1% had at least 3 LOTs during follow-up. Although the majority of HTE PWH avoided gaps in ART, 26.5% to 32.5% of people in each LOT experienced a gap in ART of at least 45 days. Mean (SD) annualized all-cause and HIV-related health care costs for HTE PWH vs non-HTE PWH were $69,529 ($156,077) vs $51,726 ($89,179) (P < 0.001) and $40,082 ($91,703) vs $30,717 ($66,926) (P < 0.001), respectively. Pharmacy costs and outpatient costs were the largest factors contributing to increased costs for HTE PWH.

CONCLUSIONS:

HTE PWH experienced frequent changes to their ART regimens, and they had higher disease burden, health care utilization, and health care costs compared with PWH who do not meet the criteria of HTE, suggesting an unmet need in this population.

Plain language summary

This study measured medication use, health care use, and health care costs among people with HIV who have tried many different treatment options. It found that these patients change treatments often by adding or subtracting medications. Up to one-third of patients go at least 45 days without refilling their medication after running out of pills. These patients also have poorer health and higher health care costs compared with other people with HIV.

Implications for managed care pharmacy

This study found that heavily treatment-experienced people with HIV (PWH) experience frequent treatment switches and have high health care resource utilization, particularly pharmacy claims and outpatient visits. Program investment in continuous monitoring of PWH may identify opportunities to intervene and optimize their treatment regimen. The study population demographics suggest that health care decision-makers stand to benefit from investing resources in particular heavily treatment-experienced PWH subgroups: those that are Black and those that reside in the South.

Advancements in antiretroviral therapy (ART) have resulted in significant improvements in viral suppression, immune function preservation, and overall morbidity and mortality among people with HIV.^1–3 Although the development of highly effective ARTs and the introduction of preexposure prophylaxis has led to a significant decrease in the annual incidence of new HIV infections in the United States (from approximately 48,000 in 2012 to 32,000 in 2022), there remain an estimated 1.2 million people with HIV in the United States.^4–6 This results in a heavy economic burden as recent studies have estimated the annual direct health care costs of the average patient with HIV to be roughly $30,000 to $40 000, whereas the incremental discounted lifetime costs of HIV have been estimated to be $1.1 million.^7–10

Among people with HIV, there is a subset who have limited ART options owing to drug resistance, intolerance, or interactions with concomitant medications and are classified as heavily treatment-experienced (HTE) people with HIV.^2,11 Although it is difficult to estimate the true prevalence of HTE people with HIV because of the lack of a formalized definition, 2 studies evaluating a variety of algorithms for identifying HTE people with HIV based on current medication regimen and/or history of ART estimated the prevalence of HTE to range from 1.9% to 6.0% of people with HIV.^12,13 Although the introduction of newer ARTs has reduced the number of patients with limited treatment options over time,¹⁴ HTE people with HIV pose management challenges because of their individualized viral resistance profiles as well as having a higher disease burden, pill burden, risk of developing AIDS-related and non–AIDS-related complications, and increased mortality rate compared with people with HIV who are not HTE.^2,9,11,15,16

Although there are some existing data on the treatment patterns and health care costs associated with HTE people with HIV in the United States,^9,15 these prior studies relied exclusively on either electronic health records (EHRs) or administrative claims data. This study, which leveraged a linked EHR and claims dataset to identify an HTE population using prescription records and laboratory test results, sought to expand upon the existing literature by providing a breakdown of the treatment journey of HTE people with HIV across multiple lines of therapy (LOTs) and capturing health care utilization and costs in a population that included individuals with Medicaid insurance, the largest provider of insurance for people with HIV.^9,17

Methods

STUDY DESIGN AND DATA SOURCES

This was a retrospective, observational analysis using data from the Veradigm Network EHR linked to administrative claims data for the period spanning January 1, 2015, through December 31, 2023 (study period). Veradigm Network EHR aggregates clinical data from 3 major US ambulatory care EHR platforms (Allscripts, Practice Fusion, and NextGen). These data provide insight into health status and medical care received and/or recorded at contributing outpatient facilities. In addition to capturing diagnoses, medications, and procedures, EHR data contain clinically rich variables, including vital signs, laboratory test results, and medical history. The linkage to claims data provides a comprehensive view of the care received during a period of continuous enrollment in a health insurance plan because it captures care received in inpatient, outpatient, and pharmacy settings. As Practice Fusion provides its EHR software free of charge to many free and low-cost clinics, including members of the National Associated of Free and Charitable Clinics,¹⁸ the dataset is enriched with patients having Medicaid insurance. In the United States, Medicaid is the largest provider of insurance for adults with HIV, covering an estimated 40% of adults with HIV.¹⁷

The final linked dataset was created as a merge of the patient-level deidentified tokens in each individual dataset and contains no personal health information. The linked dataset contains only deidentified data as per the deidentification standard defined in Section §164.514(a) of the Health Insurance Portability and Accountability Act of 1996 (HIPAA) Privacy Rule.¹⁹ As a noninterventional, retrospective database study using data from a certified HIPAA-compliant deidentified research database, approval by an institutional review board was not required.

LINE OF THERAPY CONSTRUCTION

LOTs during the study period were constructed using only claims data for each individual with a claim for an antiretroviral drug. For each individual, the first step was to identify all ART drugs received during the study period identified by generic name and then build drug epochs during which the individual had a continuous supply of the drug on hand with a 45-day allowable gap. The second step was to overlay all drug epochs for an individual to determine the LOTs. Drugs started on the same day were considered part of the same LOT. If drug B was started after drug A, then it was considered to be the start of a new LOT. However, if there was at least a 45-day overlap in medication availability, the drugs were considered part of the same LOT. If the overlap was less than 45 days, then it was assumed to be a treatment switch and that the patient discarded the remaining supply of drug A when they filled the prescription for drug B.

STUDY COHORT

People with HIV with at least 2 ART LOTs between January 1, 2015, and December 31, 2022, were identified using claims data. Qualifying individuals were then evaluated for HTE using the following criteria: (1) they received ART specifically indicated for HTE (enfuvirtide, fostemsavir, ibalizumab, and lenacapavir) after at least 1 ART LOT, (2) they received ART suggestive of HTE after at least 1 ART LOT and had evidence of prior viremia, (3) they were resistant to at least 3 ART classes, or (4) they had a history of exposure to the 3 core ART classes (integrase strand transfer inhibitors, nonnucleoside reverse transcriptase inhibitors, and protease inhibitors).

For criterion 1, the index date was the date of their first prescription for an ART specifically indicated for HTE. For criterion 2, the index date was the date of their first new prescription suggestive of HTE after a test indicating viremia during their previous LOT. For criterion 3, the index date was the date of their third positive resistance test. For criterion 4, the index date was the later of either the earliest date that the patient had been exposed to all 3 classes or the start date of their first LOT during the study period if they had already been exposed to all 3 classes. Exposure to medication classes in criterion 4 was determined using both claims and EHR data. All other criteria used only claims data to determine medication usage. For patients who met multiple HTE criteria, the index date was the earliest first qualifying index date. Details of the HTE criteria can be found in Supplementary Table 1 ^{(365.1KB, pdf)} (available in online article).

To help contextualize the study findings, a matched comparator group of treatment-experienced non-HTE people with HIV was identified. Individuals eligible for inclusion in this non-HTE cohort were those with at least 2 ART LOTs between January 1, 2015, and December 31, 2022, and no evidence of either HTE or viral suppression (non-HTE). People with HIV in the non-HTE cohort were assigned an index date based on the start date of a randomly selected antiretroviral (ARV) LOT between January 1, 2015, and December 31, 2022.

All individuals were required to be 18 years or older on the index date, have a diagnosis of HIV any time prior to or on the index date, and have continuous enrollment in medical and pharmacy claims for at least 6 months before and after the index date. HTE people with HIV were directly matched to people with HIV in the non-HTE cohort based on characteristics at index, including age group, sex, race, ethnicity, geographic region, year of index, and payer type. HTE people with HIV who could not be directly matched to an individual in the non-HTE cohort were excluded from the health care resource and utilization analysis.

STUDY PERIODS

The 6-month baseline period consisted of the 6 months preceding and not inclusive of the index date. The 6-month follow-up period consisted of the 6 months following and inclusive of the index date. The variable follow-up period started at the index date and continued until the earliest of either the end of the patient’s continuous enrollment or the end of available data (December 31, 2022). A diagram of the study design can be found in Supplementary Figure 1 ^{(365.1KB, pdf)} .

STUDY VARIABLES

Demographic characteristics were measured on the index date and included age, age group by decade (18-70+ years), sex, race, ethnicity, geographic region, year of index, and payer type. Baseline clinical characteristics and select non-ART medications were captured in the 6-month baseline period. HIV-related laboratory tests (viral load and cluster of differentiation 4–positive [CD4+] T lymphocyte counts), adverse events (diarrhea, fatigue, nausea, and vomiting), and the presence of Centers for Disease Control and Prevention (CDC)–defined AIDS-defining conditions were measured in the 6-month follow-up period.

If multiple laboratory results for a patient were available in the 6-month follow-up period, the laboratory result closest to but before the end of the follow-up period was used. AIDS-defining conditions were defined using the CDC guidance and are reported in Supplementary Table 2. ^{(365.1KB, pdf) 20} International Classification of Diseases, Tenth Revision, Clinical Modification codes used to identify baseline clinical characteristics are reported in Supplementary Table 2 ^{(365.1KB, pdf)} .

TREATMENT PATTERNS

Treatment patterns were measured during the variable follow-up period for all HTE people with HIV. For individuals who qualified as HTE by criteria 1, 2, or 4, the index LOT was the LOT initiated on the index date. For individuals who qualified as HTE by criterion 3, the index LOT was the first LOT on or after the index date. The index LOT was not required to be the first LOT during the study period.

For each HTE person with HIV, the analysis captured the number of LOTs and the percentage of days covered (PDC) by ART during the variable follow-up period. For up to 5 LOTs, the analysis captured the duration of the LOT, the number of days between the end of a LOT and the start of the next LOT, the number of unique pills per day in the LOT, and the reason for the end of the LOT. A LOT continued until the earliest of the following: addition of at least 1 new class of medication, subtraction of at least 1 class of medication from the LOT, a gap in therapy of at least 45 days of all medications in the LOT, or the end of the variable follow-up period. The use of the pharmacokinetic boosting agents cobicistat and ritonavir were not considered when determining LOTs. Medication classes used to identify LOTs are reported in Supplementary Table 2 ^{(365.1KB, pdf)} .

An individual was considered persistent on a LOT if the LOT continued until the end of the variable follow-up period. For all other LOTs, the reason for nonpersistence was identified as one of the following: augmentation, subtraction, augmentation/subtraction, discontinuation, restart, switch with a gap in therapy, and switch without a gap in therapy.

Augmentation was defined as the addition of at least 1 new class of ART without the discontinuation of any medications in the current LOT. Subtraction was defined as the discontinuation of at least 1 class of medication in the current LOT, as well as the continuation of at least 1 class of medication in the current LOT. Augmentation/subtraction was defined as the combination of the addition of at least 1 new class of ART, the discontinuation of at least 1 class of medication in the current LOT, and the continuation of treatment with at least 1 medication in the current LOT. Discontinuation was defined as a gap in therapy of at least 45 days of all medications in the LOT and no additional LOTs for the remainder of the follow-up period. Restart was defined as a gap in therapy of at least 45 days of all medications in the LOT followed by resumption of treatment with all drugs in the previous LOT. Switch with a gap in therapy was defined as a gap in therapy of at least 45 days of all medications in the LOT followed by initiation of treatment with a new class or set of classes of ART. Switch without a gap in therapy was defined as the addition of at least 1 new class of ART with the discontinuation of all medications in the current LOT.

HEALTH CARE UTILIZATION AND COSTS

All-cause and HIV-related health care utilization and costs, including inpatient, emergency department, outpatient, and pharmacy services, were reported in the variable follow-up period for PWH in the matched cohorts. HIV-related medical claims included inpatient medical claims with a primary diagnosis of HIV, outpatient medical claims with a diagnosis of HIV in any position, HIV-related laboratory claims (HIV antibody, HIV antigen, HIV viral load, drug resistance, and CD4+ laboratory tests), and pharmacy claims for ART medications including the pharmacokinetic boosters cobicistat and ritonavir.

For medical and pharmacy services, the analysis captured the number and percent of individuals with at least 1 claim for specific service types during the follow-up period. Also captured were the average number and cost of services in the variable follow-up period. Utilization and costs were annualized per person, per year. For inpatient visits, the annualized number of hospitalized days and the length of stay per admission among individuals with at least 1 admission was captured. Utilization and costs were measured using only claims data.

DATA ANALYSIS

Study measures were reported descriptively using mean and SD for continuous variables, as well as number and percent for categorical variables. Median and IQR were reported for laboratory tests, LOT duration variables, and cost and utilization variables. P values were calculated using chi-square tests for categorical variables or the Wilcoxon rank-sum for continuous variables. Descriptive statistics were generated using SAS v9.4 (SAS Institute Inc.).

Results

Among the 304,196 people with HIV who had at least 2 ART LOTs, 19,221 (6.3%) met at least 1 criterion for HTE, and 2,507 (0.8%) met all the study criteria (Figure 1). Most individuals of these 2,507 (95.6%) qualified for the study based on exposure to the 3 core ART classes (integrase strand transfer inhibitors, nonnucleoside reverse transcriptase inhibitors, and protease inhibitors) (Table 1).

FIGURE 1.

Selection of HTE People With HIV

^aDetails of HTE inclusion criteria can be found in Supplementary Table 1 ^{(365.1KB, pdf)} .

^bIndex date was defined as the earliest date on which one of the HTE-defining criteria was met.

ART = antiretroviral therapy; EHR = electronic health record; HTE = heavily treatment experienced; LOT = line of therapy.

TABLE 1.

Demographic Characteristics of HTE People With HIV

	All HTE
	N = 2,507
Age, index date, mean (SD), years	50.6 (11.7)
Sex, n (%)
Male	1,586 (63.3)
Female	920 (36.7)
Unknown	1 (0.0)
Race, n (%)
White	780 (31.1)
Black	1,092 (43.6)
Asian	55 (2.2)
Other	403 (16.1)
Unknown/not reported	177 (7.1)
Ethnicity, n (%)
Hispanic	188 (7.5)
Non-Hispanic	1,746 (69.6)
Unknown	573 (22.9)
Geographic region, n (%)
Northeast	560 (22.3)
Midwest	342 (13.6)
South	954 (38.1)
West	616 (24.6)
Other/unknown	35 (1.4)
Payer type, n (%)
Commercial	554 (22.1)
Medicaid	1,422 (56.7)
Medicare	258 (10.3)
Other/unknown	273 (10.9)
HTE-defining criteria a
Criterion 1	77 (3.1%)
Criterion 2	139 (5.5%)
Criterion 3	1 (0.0%)
Criterion 4	2,397 (95.6%)

^a See Supplementary Table 1 ^{(365.1KB, pdf)} for definitions of HTE-defining criteria. Patients could meet more than 1 criterion.

HTE = heavily treatment experienced.

The mean age of HTE people with HIV was 50.6 years, and 63.3% were male (Table 1). These individuals were predominantly Black (43.6%), non-Hispanic (69.6%), and receiving Medicaid insurance (56.7%). Most people with HIV indexed between 2016 and 2018 (Supplementary Table 3 ^{(365.1KB, pdf)} ).

Overall, 44.3% of HTE people with HIV had a baseline diagnosis of primary hypertension, and 30.4% had a diagnosis of another cardiovascular disease (Supplementary Table 4 ^{(365.1KB, pdf)} ). Other common comorbidities included hyperlipidemia (35.5%), psychoactive substance use disorder (29.3%), and depression (25.6%). In the baseline period, the use of antihypertensives, lipid-lowering medications, and corticosteroids was documented in 44.3%, 31.2%, and 23.5% of HTE people with HIV, respectively.

In the 6-month follow-up period, 21.7% of all HTE people with HIV had a viral load result available, and 21.1% had a CD4+ T lymphocyte cell count available (Table 2). Median (IQR) viral load and CD4+ laboratory test results in the follow-up period were 1.9 (1.4-2.9) log₁₀ copies/mL and 493 (256-761) cells/mL, respectively. Nausea, diarrhea, fatigue, and vomiting were documented in the 6-month follow-up period of 10.0%, 9.9%, 8.9%, and 7.7% of HTE people with HIV, respectively. The most common AIDS-defining conditions were pneumonia (7.9%) and candidiasis (7.3%). Other conditions were present in less than 3% of HTE people with HIV.

TABLE 2.

HIV-Related Clinical Measures in the 6 Months Inclusive of and Following the Index Date

	All HTE	Matched Cohorts
		HTE	Non-HTE	P value
	N = 2,507	N = 2,317	N = 2,317
HIV-related laboratory test results a
Viral load results available, n (%)	545 (21.7)	520 (22.4)	87 (3.8)	<0.001b
Viral load, median (IQR), log10 copies/mL	1.9 (1.4-2.9)	1.9 (1.4-3.0)	1.9 (1.2-3.1)	0.476
CD4+ T lymphocyte result available, n (%)	530 (21.1)	500 (21.6)	141 (6.1)	<0.001b
CD4+ T lymphocyte cell count, median (IQR), cells/mm3	493 (256-761)	486 (254-760)	570 (319-804)	0.104
Adverse events, n (%)
Nausea	251 (10.0)	236 (10.2)	184 (7.9)	0.008b
Diarrhea	248 (9.9)	233 (10.1)	142 (6.1)	<0.001b
Fatigue	224 (8.9)	209 (9.0)	165 (7.1)	0.018b
Vomiting	193 (7.7)	181 (7.8)	135 (5.8)	0.007b
AIDS-defining conditions, n (%) c
Pneumonia	198 (7.9)	185 (8.0)	122 (5.3)	<0.001b
Candidiasis	182 (7.3)	172 (7.4)	97 (4.2)	<0.001b
Herpes simplex virus	55 (2.2)	52 (2.2)	44 (1.9)	0.409
Wasting syndrome due to HIV	53 (2.1)	49 (2.1)	31 (1.3)	0.042b
Lymphoma	48 (1.9)	44 (1.9)	21 (0.9)	0.004b
Cytomegalovirus	34 (1.4)	32 (1.4)	5 (0.2)	<0.001b
Pneumocystis pneumonia	29 (1.2)	28 (1.2)	11 (0.5)	0.006b
Encephalopathy, HIV-related	28 (1.1)	28 (1.2)	23 (1.0)	0.481

^a The value closest to the end of the follow-up period is reported.

^b Statistically significant.

^c Conditions that were present in less than 1% of the All HTE cohort are not reported.

CD4+ = cluster of differentiation 4 positive; HTE = heavily treatment experienced.

TREATMENT PATTERNS

The mean (SD) available follow-up time for HTE people with HIV was 3.2 (2.1) years (Supplementary Table 5 ^{(365.1KB, pdf)} ). During this variable follow-up period, 81.2% of HTE people with HIV had at least 4 LOTs, and 27.4% had at least 5 LOTs. The median (IQR) PDC during this period was 75.4% (49.9%-91.7%). The median (IQR) duration of LOT was 3.0 (1.0-10.2) months for the first LOT, 3.9 (1.5-12.0) for the second LOT, 3.4 (1.1-11.0) for the third LOT, 2.8 (1.0-8.2) for the fourth LOT, and 2.3 (1.0-6.1) for the fifth LOT. The median time between all LOTs was 0 days. The mean pill burden during a LOT was approximately 2.0 pills per day, and approximately 98% of patients in each LOT had fewer than 5 pills per day.

There were 49 HTE people with HIV who were excluded from the treatment patterns analysis because their ART use on or after the index date was captured only in the EHR. Among the remaining HTE people with HIV, 9.0% were persistent on their index LOT until the end of available follow-up (Figure 2). Across all other LOTs, 13.5% to 16.3% of HTE people with HIV remained persistent on therapy until the end of follow-up. The majority of people with HIV avoided gaps in ART either by remaining persistent (9.0–16.3%) on therapy or by modifying their regimen without a gap in therapy via addition and/or subtraction to a regimen, or switching to a new regimen (51.2%-64.5%). However, this left 26.5% to 32.5% of people in each LOT who experienced a gap in ART of at least 45 days. This included 8.2% to 15.9% of HTE people with HIV who discontinued treatment before the end of follow-up without evidence of restart or switch to a new ART. The top 10 most common drug combinations observed in all LOTs are documented in Supplementary Figure 2 ^{(365.1KB, pdf)} .

FIGURE 2.

Treatment Patterns^a of HTE People With HIV by LOT

^aTreatment patterns were captured using only claims data. As a result, 49 individuals were excluded from the treatment patterns analysis because their ART use on or after the index date was captured only in the EHR.

ART = antiretroviral therapy; EHR = electronic health record; HTE = heavily treatment experienced; LOT = line of therapy.

CHARACTERISTICS OF MATCHED COHORTS

After direct matching all demographic characteristics, 190 HTE people with HIV were excluded from the matched cohort because they did not have a direct match in the non-HTE cohort. Demographic characteristics after matching followed the same trends as the overall population (Supplementary Table 3 ^{(365.1KB, pdf)} ). The prevalence of baseline clinical conditions and use of non-ART medications tended to be higher in the HTE cohort than in the non-HTE cohort (Supplementary Table 4 ^{(365.1KB, pdf)} ). For example, hypertension was documented in 43.8% of the matched HTE cohort but only 39.2% (P = 0.002) of the matched non-HTE cohort. One notable exception was psychoactive substance use disorder, which was documented in 29.3% and 30.7% of the HTE and non-HTE cohorts, respectively (P = 0.290).

Clinical measures in the HTE cohort were similar before and after matching (Table 2). Significantly fewer people with HIV in the non-HTE cohort had HIV-related laboratory test results available in the dataset in the 6-month follow-up period. Where test results were available, the median (IQR) viral load (1.9 [1.4-3.0] vs 1.9 [1.2-3.1], P = 0.476) and CD4+ T lymphocyte counts (486 [254-760] vs 570 [319-504]) were similar to the HTE cohort. In general, adverse events and AIDS-defining conditions were significantly more common in the HTE cohort than in the matched non-HTE cohort. The mean (SD) duration of available follow-up in the non-HTE cohort was similar to the overall cohort at 3.0 (2.0) years.

HEALTH CARE UTILIZATION AND COSTS

HTE PWH had significantly higher all-cause and HIV-related health care utilization and costs compared with PWH who did not meet the criteria of HTE (Figure 3; Supplementary Table 6 ^{(365.1KB, pdf)} ). For example, a larger percentage of the HTE cohort had at least 1 all-cause inpatient admission (40.0% vs 33.9%, P < 0.001) or at least 1 HIV-related inpatient admission (7.1% vs 4.3%, P < 0.001) during the variable follow-up period compared with the non-HTE cohort. Mean (SD) length of stay per HIV-related hospitalization was also longer in the HTE cohort than in the non-HTE cohort (17.3 [79.0] days vs 9.1 [10.1] days, P < 0.001).

FIGURE 3.

All-Cause and HIV-Related Health Care Costs in the Variable Follow-Up Period Among HTE People With HIV and Matched Controls

HTE = heavily treatment experienced; USD = US dollars.

Mean (SD) all-cause health care costs were significantly higher for the HTE cohort than for the non-HTE cohort ($69,529 [$156,077] vs $51,726 [$89,179], P < 0.001) (Figure 3; Supplementary Table 6 ^{(365.1KB, pdf)} ). Median (IQR) all-cause costs were $43,891 ($28,714-$67,515) for the HTE cohort and $35,111 ($21,483-$53,297). Similarly, mean HIV-related health care costs were significantly higher for the HTE cohort than for the non-HTE cohort ($40,082 [$91,703] vs $30,717 [$66,926], P < 0.001). Median (IQR) HIV-related costs were $31,799 ($20,347-$42,559) for the HTE cohort and $25,734 ($14,277-$35,131).

Pharmacy costs ($31,921 [$28,368] vs $24,363 [$19,410], P < 0.001) and outpatient services costs ($24,678 [$131,434] vs $17,645 [$71,285], P < 0.001) were the largest factors contributing to total costs in all cohorts (Figure 3). Mean (SD) HIV-related pharmacy costs were $26,907 ($16,338) for the HTE cohort and $20,700 ($12,169) for the non-HTE cohort. A full breakdown of utilization and cost categories, along with median and IQR values, is reported in Supplementary Table 6 ^{(365.1KB, pdf)} .

Discussion

This study leveraged a linked EHR and claims database to identify and characterize HTE people with HIV. From 19,221 individuals who met the criteria for HTE, we identified 2,507 who met all the study criteria and 2,317 who could be directly matched to people with HIV with prior ART experience but no evidence of HTE. HTE people with HIV had a high burden of other clinical conditions, experienced frequent changes to their ART regimens, and had high all-cause and HIV-related health care utilization and costs. Although HTE people with HIV represent a small percentage of the overall people with HIV population, this analysis demonstrates they incur high medical costs and utilization compared with a matched people with HIV cohort without HTE.

Although there is no standardized definition of HTE, these people with HIV are often described as having 2 or fewer classes available with limited fully active agents to construct a suppressive regimen.¹¹ The lack of a group-specific diagnosis code or a standardized definition has made it challenging to identify this subgroup of people with HIV in real-world datasets. Prior attempts to identify this population in US data sources relied solely on medication use and evidence of resistance testing to create a composite definition for the identification of HTE people with HIV.^12,13 In this study, we refined those approaches by incorporating viral load and ART resistance test results, where available, into the composite definition.

The demographic characteristics of the HTE population identified in this study were generally consistent with prior studies, with a few notable exceptions.^9,12,14,15 Specifically, the HTE population in this study included a greater proportion of female individuals (36.4%) compared with 15% to 19% in most prior studies and in the overall CDC estimates of new HIV infections.^5,9,14,15 One exception was the study by Henegar et al, in which 28% to 35% of HTE people with HIV were female.¹² Another difference in the HTE population identified for this study was that Black individuals outnumbered White individuals. Although this aligns with CDC data,⁵ it contrasts with demographic characteristics of prior US studies.^9,12,14,15 These demographic differences in race and gender likely stem from the high percentage (57%) of our study population who received their health insurance through Medicaid, as both more women and Black individuals disproportionately receive their health insurance through Medicaid.^21,22

The ART treatment patterns observed in this study suggest that HTE people with HIV need to navigate a complicated treatment journey. For example, during a mean variable follow-up of more than 3 years, more than 80% had at least 1 LOT in addition to their index HTE LOT, and more than 27% had at least 4 LOTs in addition to their index LOT. Up to 15.9% of HTE people with HIV in each LOT discontinued treatment before the end of follow-up without evidence of restarting or switching to a new ARV, potentially increasing the risk of transmission and progression to AIDS. Furthermore, gaps in therapy were observed an estimated 27% to 33% of patients in each LOT.

Although reasons for gaps in therapy were not assessed in this study, in a previous study of social determinates of health among 15,964 people with HIV, 83% had at least 1 socioeconomic barrier to health care and a higher number of barriers was associated with a decreased likelihood of excellent ART adherence.²³ Although social determinates of health were not directly captured in the current study, 57% of patients were insured through Medicaid, suggesting that they faced socioeconomic barriers to optimal care.

Treatment interruptions in ART therapy have been associated with an increased risk of progression to AIDS or death, particularly among patients with a CD4+ count of less than 200 cells/mL.^16,24 Lower CD4+ cell counts are also associated with higher medical costs because of inpatient admissions.²⁵ Although the median CD4+ count in this study was 493 cells/mL, roughly 25% of the people with HIV population with available test results had a cell count below 256 cells/mL.

In this study, HTE people with HIV had a higher comorbidity burden, greater use of non-ART medications, a higher burden of AIDS-defining conditions, and greater use of all-cause and HIV-related health care services compared with people with HIV who did not meet the criteria for HTE. These trends align with those previously reported by Priest et al.¹³ In that study of patients with commercial or Medicare advantage insurance, median 12-month all-cause costs were $52,063 for the HTE cohort and $33,708 for the treated non-HTE cohort, whereas median annualized HIV-related costs were $45,409 and $30,246, respectively.⁹ The higher median costs reported by Priest et al compared with our study may be a result of the inclusion of patients with Medicaid insurance in this study. As Medicaid is a government insurance program restricted to patients with limited income and resources, costs are often lower and barriers to care are higher in this population.

In this study, the large SD in mean total costs was driven primarily by a high variability in spending on outpatient services other than office visits or laboratory tests, which include but are not limited to imaging services, radiology services, and infusion services. This suggests that a small number of people with HIV have highly costly procedures or other services. One possibility is that these high cost events occur after regimen changes or gaps in therapy. A recent study of patients with HIV in Spain reported short-term increases in costs after patients switched ART in response to adverse events or treatment failure.²⁶ Future research could identify periods of increased costs and see if they correlated with disruptions or changes in ART, test results indicating high viral load, adverse events, or other factors.

In an era in which cuts to Medicaid are prevalent and costs of medical care continue to rise for all plan types, managed care decision-makers are pressured to do more with less. Our study identifies a subpopulation of people with HIV who are particularly costly and provides a method for identifying this subpopulation using routinely collected data. Strategies at the provider level or benefit manager level (eg, dedicated patient navigators that specialize in HIV treatment care) that help improve adherence and stabilize treatment patterns may improve outcomes and reduce costs in this vulnerable population.

LIMITATIONS

This retrospective study used routinely collected claims and EHR data and is subject to the typical limitations of using data not gathered specifically for research purposes. This includes data entry errors, missing data, and coding specificity limitations. There is no diagnosis code specific to HTE people with HIV, so we used a 4-criteria method, using drug claims, laboratory tests, and medical records, to identify likely candidates. Additionally, the limited availability of resistance data may have led to misidentification and underreporting of the HTE population size. However, we anticipate that the majority of HTE people with HIV would be identified through the other study criteria. Furthermore, it was not possible to tell from the available data if regimen changes were a result of virologic rebound, side effects, or nonclinical factors, such as changes to insurance formularies, copayments, or provider.

Many people with HIV remain uninsured in spite of the passing of the Affordable Care Act and the expansion of Medicaid programs in many states, both of which significantly improved health care insurance coverage rates among people with HIV.²⁷ The results of our study may not apply to uninsured people with HIV as we included only insured individuals since health care insurance coverage is required for accurate treatment patterns analysis. However, the Veradigm Network EHR linked to claims includes patients insured through Medicaid, potentially making it more representative of the true HTE population and setting it apart from other datasets used in prior HTE studies. Additionally, this study represents patients within the United States and therefore may not be applicable to other countries.

Although certain conditions are considered AIDS-defining conditions, it was not possible to confirm if a specific diagnosis was attributable to HIV infection or another diagnosis. Specifically, candidiasis of any site was used in our study definition; however, only candidiasis of the esophagus, bronchi, trachea, or lungs is considered an AIDS-defining condition. This approach was taken because, although candidiasis of other sites may not be exclusive to patients with a CD4+ count of less than 200 cells/mL, it is more common among these patients and represents a clinical burden.^28–30

Finally, although we tried to capture people with HIV at the first event where they qualified as HTE, patients may have already been considered HTE for an unknown amount of time prior to their first LOT identified in this study. In addition, although HTE criteria were determined based on all available data, people with HIV were only required to have at least 6 months of continuous enrollment in claims and EHR activity before and after the index date. As a result, people with HIV, particularly those with only the minimum available data, may be miscategorized. It should be noted that the short 12 continuous enrollment requirement (6 months before and 6 months after) was used because preliminary feasibility analysis showed that insurance coverage was particularly unstable in this population. With the current approach, 80% of patients were excluded based on the enrollment/activity criterion, leading to potential selection bias for individuals with stable insurance, which would have been exacerbated by applying a longer data requirement. Furthermore, insurance instability is more common among individuals with Medicaid insurance than among those with commercial insurance, so increasing the enrollment criteria would have differentially impacted the eligibility of patients on Medicaid.^31,32

Conclusions

HTE people with HIV experienced frequent changes to their ART regimens; however, the majority of patients maintained continuous treatment between LOTs. Yet there remained 26% to 33% of patients who experienced a gap in therapy, providing an opportunity to improve patient outcomes by reducing these treatment disruptions through either removing barriers to treatment adherence or leveraging long-acting formulations that reduce the daily burden on patients. Improving adherence may reduce the higher disease burden, higher health care utilization, and higher health care costs observed in HTE people with HIV compared with people with HIV who did not meet the criteria of HTE.

Disclosures

Samir Gupta is an advisor/consultant for Gilead Sciences, Inc., and ViiV. He also received grant/research support from ViiV. Sorana Segal-Maurer was an advisor/consultant for Gilead Sciences, Inc., Janssen, Theratechnologies, and ViiV at the time of data collection and first analysis. At the time of final data analysis and manuscript preparation, Sorana Segal-Maurer was an employee and shareholder of Gilead Sciences, Inc. Machaon Bonafede is an employee and shareholder of Veradigm. Katherine Cappell and Robert Sedgley are employees of Veradigm. Veradigm received funding from Gilead Sciences, Inc., to conduct this study. Dylan Mezzio, JeanPierre Coaquira, Kwanza Price, Joshua Gruber, Seojin Park, and Soodi Navadeh are employees and shareholders of Gilead Sciences, Inc. This study was funded by Gilead Sciences, Inc.