Healthcare Utilization and Costs in People Living with HIV and Neuropsychiatric Disorders in the United States

Abstract

Introduction

People living with HIV (PLWH) face an increased prevalence and risk of neuropsychiatric disorders than individuals without HIV. This study compared incremental all-cause healthcare resource utilization (HCRU) and costs among US antiretroviral therapy (ART)-treated PLWH with versus without neuropsychiatric disorders.

Methods

A retrospective claims analysis was conducted using Optum’s de-identified Clinformatics^® Data Mart Database (Jan 2020–Dec 2022). Adults (≥ 18 years) with HIV and ≥ 1 pharmacy claim for an anchor ART agent [integrase strand transfer inhibitor (INSTI), non-nucleoside reverse transcriptase inhibitor (NNRTI), or protease inhibitor (PI)] in 2021 were included (index date: earliest anchor ART claim). PLWH were followed up to 12 months or end of continuous enrollment. Neuropsychiatric disorders were identified during the baseline period (12 months pre-index) using ICD-10 diagnosis codes, and cohorts were defined based on their presence (yes/no). Differences between neuropsychiatric groups in per-patient-per-month (PPPM) HCRU and costs (converted to 2023 USD) were estimated using multivariable generalized linear models (negative binomial/Poisson for HCRU; gamma for costs), adjusting for baseline characteristics.

Results

Among 22,402 PLWH, 9506 (42.4%) had neuropsychiatric disorders. PLWH with versus without neuropsychiatric disorders were older (mean age 56 vs. 54 years), more often female (21.0% vs. 17.5%), white (51.0% vs. 44.0%), had higher comorbidity (mean Quan-Charlson 1.82 vs. 0.92), and baseline total costs ($4579 vs. $3416); all P < 0.001. Unadjusted all-cause PPPM HCRU and costs were significantly higher among PLWH with neuropsychiatric disorders than among those without; all P < 0.001. PLWH with neuropsychiatric disorders generally had higher adjusted all-cause PPPM HCRU (including 28% more office visits) and higher total costs; outpatient emergency department and outpatient other visit and costs were not significantly different.

Conclusions

PLWH with neuropsychiatric disorders have higher all-cause HCRU and costs than those without these conditions. Further research on the role of HIV and ART in the development of neuropsychiatric disorders may support more individualized care and help reduce excess HCRU and costs.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s12325-026-03705-1.

Key Summary Points

Why carry out this study?

People living with HIV (PLWH) face an increased risk of neuropsychiatric disorders compared to individuals without HIV, and antiretroviral therapy (ART) may precipitate or worsen these conditions. This risk may increase with age and long-term exposure to ART.

This retrospective cohort study described the current economic burden of neuropsychiatric disorders in the ART-treated PLWH with versus without these comorbidities using a real-world cohort (2020–2022).

What was learned from the study?

PLWH with neuropsychiatric disorders had significantly greater all-cause healthcare resource utilization (HCRU) and costs than those without these comorbidities, including higher adjusted all-cause HCRU across all service types (22% more inpatient hospitalizations, 28% more office visits, and 28% more pharmacy fills).

Adjusted all-cause total costs were 7% higher (+ $504 per patient per month) among PLWH with versus without neuropsychiatric disorders

These findings highlight the importance of early identification and management of neuropsychiatric disorders, as well as the continued development of novel therapies that minimize the risk of neuropsychiatric disorders to improve long-term health of PLWH.

Introduction

People living with HIV (PLWH) have increased prevalence and risk of neuropsychiatric disorders compared to people without HIV [1–4]. Aging PLWH may be more at risk of HIV-related changes in brain metabolism[5, 6] and antiretroviral therapy (ART) may precipitate or worsen neuropsychiatric disorders in this population [4]. Multimorbidity is common among older PLWH [2, 3], and chronic inflammatory conditions such as diabetes and cardiovascular disease may also contribute to the risk of neuropsychiatric disorders [4].

While the exact causes of the increased prevalence of neuropsychiatric disorders in PLWH are yet to be understood, these conditions are associated with increased healthcare costs. Moreover, neuropsychiatric disorders may negatively affect ART adherence and virologic suppression, while increasing the need for outpatient services, emergency department visits, hospitalizations, behavioral health services, and pharmacologic treatment [8, 9].

Prior studies have demonstrated that neuropsychiatric disorders contribute to increased healthcare expenditures among PLWH. Higher medical costs have been reported among PLWH with neuropsychiatric disorders compared with those without [7]. Similarly, individuals with HIV and co-occurring mental and substance use disorders have been shown to incur substantially greater healthcare costs than those without these comorbidities[10]. More recently, analyses of MarketScan Multi-State Medicaid claims demonstrated that neuropsychiatric disorders remain common among ART-treated PLWH and are associated with considerable HCRU and costs[11]. Collectively, these findings suggest that neuropsychiatric disorders are one of the important drivers of healthcare burden in PLWH.

However, much of the existing evidence is based on earlier ART treatment eras, focuses on specific neuropsychiatric conditions rather than broader neuropsychiatric disorder groups, and may not fully reflect current real-world economic burden in aging PLWH populations. In particular, much of the literature examining the economic burden of neuropsychiatric conditions among PLWH remains outdated [7, 10] while broader studies of HIV-related costs do not specifically evaluate the incremental burden associated with neuropsychiatric disorders[12]. As HIV care has evolved substantially over the past two decades, updated estimates are needed to better characterize the economic burden associated with neuropsychiatric disorders in PLWH.

Understanding the contemporary economic burden of neuropsychiatric disorders in PLWH could help assist policy-making, planning, and management of healthcare resource utilization (HCRU) and costs in the HIV population. The objective of this study was to assess incremental all-cause HCRU and healthcare costs among ART-treated PLWH with and without neuropsychiatric disorders in the United States.

Methods

Study Design and Data Source

A retrospective observational cohort study was conducted using administrative claims data from Optum’s de-identified Clinformatics^® Data Mart Database (Optum^® CDM) from January 1, 2020, through December 31, 2022 (Fig. 1). The database contains detailed longitudinal information on medical and pharmacy claims, inpatient confinement claims, member enrollment, and linked socioeconomic data for members of large US commercial and Medicare Advantage plans, including more than 77 million people across all 50 US states. All data were statistically certified as de-identified for research purposes in accordance with the Health Insurance Portability and Accountability Act of 1996 and no identifiable protected health information was extracted, so institutional review board approval and informed consent were not required.

Fig. 1.

Study design. ART antiretroviral therapy, INSTI integrase strand transfer inhibitor, NNRTI non-nucleoside reverse transcriptase inhibitor, PI protease inhibitor

Study Sample

The study population included all adults (≥ 18 years) living with HIV identified in Optum^® CDM who had ≥ 1 medical or pharmacy claim for an anchor ART filled from January 1 to December 31, 2021. The index date was the date of the first prescription claim for an anchor ART during this period. Anchor agents included integrase strand transfer inhibitor (INSTI), non-nucleoside reverse transcriptase inhibitor (NNRTI), or protease inhibitor (PI) (Supplemental Table 1), and index regimens were identified based on the ART prescription (anchor, backbone, and booster) filled within the first 15 days post-index date. Included PLWH were also required to have ≥ 1 inpatient or outpatient HIV diagnosis claim (ICD-10-CM codes B20, B97.35, O98.711–713, O98.719, O98.72–73, Z21) during the 12-month baseline period and ≤ 30 days after the index date. Continuous health plan enrollment was required for the 12-month baseline period and ≥ 30 days after index date.

PLWH with an anchor ART treatment on the index date who received pre- or post-exposure prophylaxis during the baseline or follow-up period were excluded. Pre-exposure prophylaxis was defined as ≥ 1 pharmacy claim for tenofovir disoproxil fumarate or tenofovir alafenamide plus emtricitabine during the baseline or follow-up period with no pharmacy claim for another ART medication. Post-exposure prophylaxis was defined as ≥ 1 claim with ICD-10-CM codes Z20.6 or Z11.4 during the baseline or follow-up periods.

The study sample was stratified into two groups based on presence of neuropsychiatric disorders (yes/no) within the baseline period, which was identified using ICD-10 diagnosis codes from medical claims (≥ 1 inpatient or ≥ 2 outpatient claims; Supplemental Table 2) [13]. A single outpatient diagnosis code may reflect a rule-out or probable diagnosis; therefore, requiring at least two outpatient claims increases the likelihood of a true clinical diagnosis. Neuropsychiatric disorders included depression, anxiety, suicidal ideation, bipolar/manic disorders, insomnia/sleep-related disorders, cognitive impairment/poor concentration, amnestic disorders, dementia-type disorders, dissociative and conversion disorders, intellectual disorders, pervasive developmental disorder, psychosis, and schizophrenia. PLWH were followed to the earliest of 12 months, end of the study period, or end of continuous enrollment.

In this study, the ART-based index date was used solely as a methodological criterion for cohort identification. The study does not assume or evaluate the index ART prescription as a trigger for the onset of neuropsychiatric conditions or related cost accrual. The objective of this study is to describe the costs and HCRU burden associated with neuropsychiatric disorders among ART-treated PLWH, rather than to assess any causal relationship between ART exposure and neuropsychiatric outcomes.

Study Measures

Sociodemographic characteristics were assessed at the index date and included age at index date, sex, race/ethnicity, geographical region, income, insurance type, and commercial plan type. Clinical characteristics were measured during the baseline period and included Quan–Charlson Comorbidity Index (QCCI) score (based on diagnosis codes on medical claims for 12 condition categories, excluding HIV/AIDS [14]), other individual comorbidities, prior ART use, and use of concomitant medications. Selected comorbid conditions at baseline were identified using diagnosis codes on medical claims (≥ 1 inpatient admission or ≥ 2 outpatient claims). The use of concomitant medications (antidiabetic, neuropsychiatric, or cardiovascular disease) was identified by the presence of ≥ 1 prescription claim for the respective medications during the baseline period.

The main outcome variables were all-cause HCRU and healthcare costs, which were defined using medical and pharmacy claims associated with any condition, and measured during the follow-up period for ≥ 1 month and a maximum of 12 months post-index date. All-cause HCRU was defined as the number of healthcare visits and included the following service types: inpatient hospitalizations, inpatient hospital length of stay (LOS), physician (office-based) visits, outpatient hospital visits, outpatient emergency department (ED) visits, and pharmacy fills. These were identified using Current Procedural Terminology codes, place of service, and revenue codes (Supplemental Table 3). All-cause healthcare costs included both health plan- and enrollee-paid amounts and included costs for each HCRU service type as well as total healthcare costs, which was computed as the sum of direct healthcare costs incurred from medical (inpatient and outpatient) and pharmacy claims.

All-cause HCRU and costs were measured per-patient-per-month (PPPM) and adjusted to 2023 USD using the medical care component of the Consumer Price Index [15]. In addition, HCRU and costs specific to HIV (i.e., associated with HIV diagnosis and ART pharmacy claims) and specific to neuropsychiatric disorders (i.e., associated with diagnosis and pharmacy claims for neuropsychiatric disorders) were also examined.

Statistical Analyses

Baseline demographic and clinical characteristics were summarized for each cohort using descriptive statistics, including frequency and percentage for categorical variables or mean and standard deviation (SD) for continuous variables. Differences between PLWH with versus without neuropsychiatric disorders were evaluated using the chi-square test or Fisher exact test (when cell size < 5) for baseline characteristics and the Wilcoxon rank-sum test for HCRU and cost outcomes.

We estimated differences in all-cause PPPM HCRU and costs between PLWH who have neuropsychiatric disorders compared to those without neuropsychiatric disorders, adjusting for baseline characteristics. We used multivariable generalized linear models with negative binomial or Poisson distribution with log link function for HCRU and gamma distribution with log link function for costs, adjusted for baseline characteristics. Gamma distribution for costs was selected based on implementing a modified Park test. Because Gamma regression requires strictly positive outcomes, a value of $1 was assigned to observations with zero costs to enable model estimation while preserving the full analytic sample. The final multivariable regression model was based on the Akaike information criterion or Bayesian information criterion, with preference given to the model with lower values (Supplemental Table 4). The adjusted least squares (LS) mean incremental PPPM HCRU and costs among PLWH with versus without neuropsychiatric disorders were presented along with the rate ratio (RR) and corresponding 95% confidence interval (CI).

Variables identified to be statistically significant or clinically relevant to the study objectives were used in the multivariable regression analysis after examining for multicollinearity (diagnostic threshold: variance inflation factor ≥ 5) [16]. The regression models adjusted for the following covariates: age at index date, sex, race and ethnicity, US region, income, insurance, commercial plan type, anchor ART agent, hyperlipidemia, peripheral vascular disease, liver disease/cirrhosis, any malignancy, substance use disorders (opioid use/drug use/alcohol use/tobacco use), chronic obstructive pulmonary disease, asthma, rheumatoid arthritis, osteoporosis, hypothyroidism, gastrointestinal disorders, prior ART medication, and cardiovascular disease medications.

Subgroup analyses were conducted to examine differences in adjusted all-cause PPPM pharmacy fills and total healthcare cost estimates between groups when stratified by factors based on previous literature: age, sex, race/ethnicity, and geographic region [17–19]. Sensitivity analyses were conducted to examine all-cause total costs [PPPM and per patient per year (PPPY)] for a fixed follow-up period of 12 months among PLWH who were continuously enrolled for at least 12 months post-index date. In addition, all-cause total costs were assessed based on the initiation of ART medication prior to the index date. Treatment status was identified using claims for any ART medication during the baseline period: PLWH with ART claims were classified as treatment-experienced and those without ART claims were classified as treatment-naïve. The adjusted LS mean PPPM (for subgroup analysis) and PPPM and PPPY (for sensitivity analysis), along with RR and corresponding 95% CI, were reported between cohorts with and without neuropsychiatric disorders.

All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC, USA). A two-sided p value of < 0.05 was considered statistically significant.

Results

Study Population

Among the 22,402 PLWH identified (Fig. 2), 9506 (42.4%) had neuropsychiatric disorders. The median length of follow-up was 12 months in both cohorts. The mean follow-up was statistically significantly longer in PLWH with neuropsychiatric disorders, though the difference was small (10.96 vs. 10.92 months; p = 0.037). In both cohorts, most PLWH were male, white, and from the US South (Table 1).

Fig. 2.

Attrition diagram. ^aIndex ART regimen is identified based on the ART prescription (anchor, backbone, booster) filled within the first 15 days after the index date, which is defined as the first prescription date for an anchor ART medication during the study identification period. ART antiretroviral therapy, PEP post-exposure prophylaxis, PLWH people living with HIV, PrEP pre-exposure prophylaxis

Table 1.

Baseline characteristics of PLWH with and without neuropsychiatric disorders

	With NPD (N = 9506)	Without NPD (N = 12,896)	p value
Age at index, years, mean (SD)	55.78 (12.61)	53.99 (13.81)	< 0.001
Age group, n (%)
< 50 years	2598 (27.3)	4443 (34.4)	< 0.001
≥ 50 years	6908 (72.7)	8453 (65.6)
Sex, n (%)a
Male	7513 (79.0)	> 10,632 (> 82.4)	< 0.001
Female	1993 (21.0)	2253 (17.5)
Unknown	0 (0.0)	< 11 (< 0.1)
Race/ethnicity, n (%)
White	4844 (51.0)	5674 (44.0)	< 0.001
Black	2599 (27.3)	4148 (32.2)
Hispanic	1348 (14.2)	1931 (15.0)
Asian	147 (1.6)	265 (2.1)
Unknown	568 (6.0)	878 (6.8)
Region, n (%)a
Northeast	1042 (11.0)	1543 (12.0)	< 0.001
Midwest	1155 (12.2)	1795 (13.9)
South	> 5445 (> 57.3)	7391 (57.3)
West	1853 (19.5)	2154 (16.7)
Unknown	< 11 (< 0.1)	13 (0.1)
Insurance, n (%)
Commercial	4136 (43.5)	8073 (62.6)	< 0.001
Medicare Advantage	5370 (56.5)	4823 (37.4)
Anchor agent on index date, n (%)b
INSTI	7119 (74.9)	9794 (75.9)	< 0.001
PI	1193 (12.6)	1328 (10.3)
NNRTI	1194 (12.6)	1774 (13.8)
QCCI score, mean (SD)	1.82 (2.25)	0.92 (1.52)	< 0.001
ART use prior to index date, n (%)	9166 (96.4)	12,247 (95.0)	< 0.001
Select comorbidities, n (%)c
Hyperlipidemia	5664 (59.6)	6090 (47.2)	< 0.001
Hypertension	5409 (56.9)	5469 (42.4)	< 0.001
Cardiovascular disease	2907 (30.6)	2010 (15.6)	< 0.001
Type 2 diabetes mellitus	2334 (24.6)	2228 (17.3)	< 0.001
Concomitant medications, n (%)
Antidiabetic medication	1852 (19.5)	1811 (14.0)	< 0.001
Neuropsychiatric medication	6513 (68.5)	1963 (15.2)	< 0.001
Cardiovascular medication	6787 (71.4)	7287 (56.5)	< 0.001
Baseline all-cause total healthcare costs PPPM, mean (SD)d	$4578.75 ($4821.32)	$3416.22 ($2685.61)	< 0.001

ART antiretroviral therapy, ICD-10 International Classification of Disease, 10th Edition, INSTI integrase strand transfer inhibitor, NNRTI non-nucleoside reverse transcriptase inhibitor, NPD neuropsychiatric disorders, PI protease inhibitor, PPPM per patient per month, QCCI Quan–Charlson Comorbidity Index, US United States

^aUnknown for n < 11 individuals. Data were coarsened for other categories to preserve confidentiality and designated with > for the largest category

^bTreatment hierarchy: ART regimens containing any INSTI with or without NRTI were classified as INSTI; those with any PI were classified as PI; those with NNRTI and no INSTI and no PI were classified as NNRTI

^cIdentified using ICD-10 diagnosis codes in the presence of ≥ 1 inpatient or ≥ 2 outpatient claims during baseline period

^dTotal healthcare costs: sum of medical and pharmacy PPPM costs. Adjusted to 2023 USD using Consumer Price Index

People living with HIV and neuropsychiatric disorders were statistically significantly older, a significantly larger proportion were men and white, and they had significantly higher mean QCCI scores and total healthcare costs at baseline compared to those without neuropsychiatric disorders (all p < 0.001; Table 1). All comorbidities that were examined were statistically significantly more prevalent in PLWH with versus without neuropsychiatric disorders (Supplemental Table 5). For example, hypertension (56.9% vs. 42.4%), cardiovascular disease (30.6% vs. 15.6%), and type 2 diabetes mellitus (24.6% vs. 17.3%) were significantly more prevalent in PLWH with neuropsychiatric disorders (all p < 0.001; Table 1). The use of concomitant medications was also statistically significantly higher in PLWH with neuropsychiatric disorders (all p < 0.001). PLWH with neuropsychiatric disorders were significantly more likely to have an annual household income lower than $40,000 compared with those without neuropsychiatric disorders (40.1% vs. 35.5%; p < 0.001; Supplemental Table 5). PLWH with neuropsychiatric disorders were more likely to have Medicare Advantage insurance (56.5% vs. 37.4%; p < 0.001; Table 1) and those with commercial insurance were more likely to have an HMO plan (24.0% vs. 17.2%; p < 0.001; Supplemental Table 5). The distribution of INSTI-, PI-, and NNRTI-based regimens between PLWH with and without neuropsychiatric disorders was similar in magnitude.

HCRU and Costs

Mean unadjusted all-cause PPPM HCRU and costs were statistically significantly higher among PLWH with versus without neuropsychiatric disorders across all categories (all p < 0.001; Table 2). After adjusting for baseline characteristics, most of these statistically significant differences in all-cause PPPM HCRU and costs persisted. For example, PLWH with neuropsychiatric disorders had significantly higher adjusted all-cause HCRU PPPM across all service types—including 22% more inpatient hospitalizations, a 30% longer inpatient hospital LOS, 28% more office visits, and 28% more pharmacy fills (all p < 0.02)—except outpatient ED visits, which were numerically but not significantly higher (Fig. 3A). Adjusted all-cause total costs were 7% higher (+ $504 PPPM) in PLWH with versus without neuropsychiatric disorders (p < 0.001; Fig. 3B). Most individual cost types were statistically significantly higher in PLWH with neuropsychiatric disorders, including inpatient costs (11% higher, + $70 PPPM; p < 0.01), office visit costs (36% higher, + $4 PPPM; p < 0.001) and pharmacy costs (6% higher, + $353 PPPM; p < 0.001). However, differences in outpatient ED and outpatient other costs were numerically, but not statistically significantly, higher for PLWH with versus without neuropsychiatric disorders.

Table 2.

Unadjusted all-cause HCRU and costs PPPM among PLWH with and without neuropsychiatric disorders

	Mean (95% CI)		p value
	With NPD (N = 9506)	Without NPD (N = 12,896)
HCRU
Inpatient hospitalization	0.05 (0.05–0.05)	0.02 (0.02–0.02)	< 0.001
Inpatient length of stay, days	0.40 (0.37–0.42)	0.18 (0.15–0.20)	< 0.001
Office visits	0.10 (0.10–0.11)	0.06 (0.06–0.07)	< 0.001
Outpatient hospital	0.12 (0.11–0.12)	0.06 (0.05–0.07)	< 0.001
Outpatient ED	0.02 (0.01–0.02)	0.01 (0.01–0.01)	< 0.001
Outpatient other	0.05 (0.05–0.05)	0.03 (0.03–0.03)	< 0.001
Pharmacy fills	4.93 (4.85–4.99)	3.13 (3.07–3.19)	< 0.001
Healthcare costsa
Total costs	$8530 ($8376–$8683)	$6883 ($6751–$7015)	< 0.001
Medical costsb	$1683 ($1578–$1789)	$858 ($768–$950)	< 0.001
Inpatient costsc	$1531 ($1430–$1633)	$779 ($693–$866)	< 0.001
Office visit costs	$21 ($20–$23)	$12 ($11–$14)	< 0.001
Outpatient hospital costs	$111 ($97–$126)	$54 ($43–$68)	< 0.001
Outpatient ED costs	$9 ($8–$12)	$7 ($6–$9)	< 0.001
Outpatient other costsd	$11 ($10–14)	$7 ($6–$9)	< 0.001
Pharmacy costse	$6847 ($6744–$6951)	$6024 ($5935–$6113)	< 0.001

ED emergency department, HCRU healthcare resource utilization, NE not estimable, NPD neuropsychiatric disorders, PLWH people living with HIV, PPPM per patient per month, US United States

^aCosts adjusted to 2023 USD using Consumer Price Index

^bSum of inpatient and outpatient PPPM costs

^cPPPM costs for inpatient stays

^dHospice, independent laboratory, group home, mobile unit, and home health costs identified from outpatient claims file

^ePrescription-related PPPM costs

Fig. 3.

Adjusted all-cause A HCRU and B healthcare costs PPPM in PLWH with and without neuropsychiatric disorders. ^aAdjusted for the following covariates: age, sex, race/ethnicity, region, income, insurance, commercial plan type, anchor agent, cardiovascular disease, hypertension, type 2 diabetes mellitus, chronic kidney disease, hyperlipidemia, peripheral vascular disease, liver disease/cirrhosis, any malignancy, substance use disorders, chronic obstructive pulmonary disease, asthma, rheumatoid arthritis, osteoporosis, hypothyroidism, prior antiretroviral therapy, cardiovascular medications, and gastrointestinal disorders. ^bCosts adjusted to 2023 USD using Consumer Price Index. ^cOutpatient other costs include hospice, independent laboratory, group home, mobile unit, and home health costs identified from outpatient claims file. ^dMedical costs: sum of inpatient and outpatient PPPM costs. ^eInpatient costs: PPPM costs for inpatient stays. ^fPharmacy costs: prescription-related PPPM costs. ED emergency department, HCRU healthcare resource utilization, NE not estimable, NPD neuropsychiatric disorders, PLWH people living with HIV, PPPM per patient per month

While the study aimed to investigate HIV-specific and comorbidity-specific HCRU and costs as well, the model either did not converge or converged with a warning, primarily due to sparse data points. The results are not reported herein because the questionable model fit limited our ability to interpret the estimates accurately.

Subgroup Analyses

Adjusted all-cause total healthcare costs PPPM were statistically significantly higher for PLWH with versus without neuropsychiatric disorders across all subgroups except Asian PLWH (Fig. 4A). All-cause pharmacy fills PPPM were statistically significantly higher in PLWH with neuropsychiatric disorders across all age, sex, racial/ethnic, and geographical subgroups (Fig. 4B).

Fig. 4.

Average adjusted A total all-cause healthcare costs and B pharmacy fills PPPM in PLWH with and without neuropsychiatric disorders by subgroups. Δ% = percentage increase in PPPM costs for PLWH with neuropsychiatric disorders versus without neuropsychiatric disorders. Bars 95% confidence intervals. ^aDifferences within subgroups are statistically significant (p < 0.05). ^bIndividuals of unknown sex or region were excluded from the subgroup analysis. NPD neuropsychiatric disorders, PLWH people living with HIV, PPPM per patient per month

Sensitivity Analyses

In the ART initiation sensitivity analysis, adjusted all-cause total costs PPPM were statistically significantly higher in ART treatment-experienced PLWH with versus without neuropsychiatric disorders (7% higher, + $513 PPPM; p < 0.001; Table 3), whereas the difference was not statistically significant in ART treatment-naïve PLWH. Results from the sensitivity analysis limited to PLWH with 12 months of follow-up were also similar to the overall cohort results. The adjusted all-cause total costs PPPY were 8% higher (+ $7265, p < 0.001) in PLWH with neuropsychiatric disorders compared with those without neuropsychiatric disorders.

Table 3.

Sensitivity analysis: all-cause total costs in PLWH by ART initiation and among those with 12 months follow-up^a

	Unadjusted, mean				Adjusted, least squares meansb
	With NPD	Without NPD	Δ%c	p valued	With NPD	Without NPD	Δ%c	p value
ART initiation
Treatment-naïve	n = 340	n = 649			n = 340	n = 649
PPPM	$7696.31	$5775.79	33.25%	0.013	$6188.77	$5797.24	6.75%	0.239
Treatment-experienced	n = 9166	n = 12,247			n = 9166	n = 12,247
PPPM	$8560.89	$6941.35	23.33%	< 0.001	$7661.78	$7148.72	7.18%	< 0.001
PLWH with 12 months of follow-up	n = 7198	n = 9595			n = 7198	n = 9595
PPPY	$105,666.20	$86,819.62	21.71%	< 0.001	$96,803.00	$89,538.00	8.11%	< 0.001
PPPM	$8805.53	$7234.98	21.71%	< 0.001	$8066.96	$7461.48	8.11%	< 0.001

ART antiretroviral therapy, NPD neuropsychiatric disorders, PLWH people living with HIV, PPPM per patient per month, PPPY per patient per year

^aCosts adjusted to 2023 USD using the Consumer Price Index

^bLeast square means obtained from generalized linear model after adjusting for covariates

^cΔ% = percentage increase in costs for PLWH with NPD compared with PLWH without NPD

^dp values obtained from Wilcoxon rank-sum test

Discussion

With increased life expectancy and continued exposure to ARTs, PLWH face a higher burden of comorbidities as they age and exhibit higher healthcare costs due to increased HCRU. Using administrative claims data from commercially insured and Medicare Advantage plan members, this study found that PLWH with neuropsychiatric disorders had higher all-cause office visits, pharmacy fills, and medical costs as compared to PLWH who did not have these comorbidities. This study provides insights from a real-world cohort of ART-treated PLWH during 2020–2022 who were followed for a median of 12 months.

Overall, PLWH with neuropsychiatric disorders were older, enrolled in Medicare Advantage plans, had a higher baseline QCCI, and had a greater prevalence of hyperlipidemia, hypertension, and cardiovascular disease compared to PLWH without neuropsychiatric disorders. Among the neuropsychiatric disorders examined, depression and anxiety were the most common, consistent with prior literature documenting their higher prevalence in PLWH compared with the general population [20–22]. Increased risk of mental health disorders has also been reported among older adults with HIV, potentially related to prolonged ART exposure or chronic HIV infection [4].

Incremental healthcare costs associated with neuropsychiatric conditions were substantial, suggesting that these conditions may be more severe or clinically complex among PLWH. A retrospective study from California including Medicare and Medicaid enrollees reported that mental health–related treatment costs in PLWH accounted for 15.2% of total annual inpatient and outpatient costs. In our study, PLWH with neuropsychiatric disorders such as depression (22.7%) and anxiety (21.2%) reported 28% higher office visits than those without neuropsychiatric disorders. This finding aligns with a prior study conducted in a community setting, which reported increased medical costs and higher numbers of clinic and physician visits among PLWH diagnosed with neuropsychiatric disorders compared to those without neuropsychiatric disorders [7]. One potential explanation for the higher prevalence of depression and anxiety observed during the study period is the unprecedented stress associated with the COVID-19 pandemic, which may also have contributed to increased office visits and related healthcare costs [23].

It is well documented that untreated neuropsychiatric disorders can negatively impact quality of life and ART adherence among PLWH [24]. In addition to increased HCRU and costs, neuropsychiatric disorders may introduce challenges across the HIV continuum of care. Certain ARTs are likely to be associated with worsening neuropsychiatric conditions [25–27], which should be considered during ART selection or switching. Given the substantial all-cause HCRU and costs observed among PLWH with neuropsychiatric disorders, these findings underscore the importance of mental health screening and management within HIV care, including consideration of risk-modification strategies that account for age, sex, and other patient characteristics [17, 28]. Such tailored approaches may improve the holistic management of PLWH who are at risk for, or living with, neuropsychiatric conditions. Healthcare policymakers should recognize and address barriers to mental health services among the most vulnerable groups (e.g., older PLWH), and proactively identify individuals at highest risk to mitigate adverse clinical outcomes and unnecessary HCRU. There is a critical need for clinical programs that incorporate the influence of social determinants of health, particularly those that hinder access to mental health and subspecialty care.

Interpretation of our study results is subject to certain limitations. Administrative claims data are not designed for research purposes and thus are subject to data coding and entry errors. To minimize misclassification bias, the study cohorts were defined using validated algorithms [13]. However, it is possible that certain clinical conditions or disease severity were not fully captured by the ICD-10 codes used in our analysis [29]; this may have contributed to the model instability and non-convergence issues observed for the HIV-specific and comorbidity-specific analyses, which were not reported due to questionable model fit. Indirect costs were not captured in this study due to limited data availability. The follow-up duration was similar between the two cohorts; however, the possibility of informative censoring due to differential mortality or disenrollment cannot be fully ruled out. Because laboratory data were not available, this study was unable to account for virologic suppression, which may influence differences in HCRU and costs between the two cohorts. We also did not account for the potential development of neuropsychiatric disorders during the follow-up period among PLWH without neuropsychiatric disorders, which may have led to an underestimation of differences in HCRU and costs between the two groups. Although we adjusted for available covariates, the possibility of unmeasured or residual confounding remains. Additionally, covariates were selected a priori based on clinical relevance and consistency across related comorbidity analyses; however, some adjusted comorbid conditions may partially mediate the relationship between neuropsychiatric disorders, and HCRU and costs, potentially attenuating the observed associations. As such, adjusted estimates may approximate a direct association, although causal interpretation is limited. The study was conducted in a US-based population covered by commercial or Medicare Advantage plans, so results may not be generalizable to other populations such as PLWH who are insured by Medicaid or Medicare Fee-for-Service, enrolled in Veterans Health Administration care or an AIDS Drug Assistance Program, or uninsured PLWH.

Conclusions

PLWH with neuropsychiatric disorders experienced a greater all-cause HCRU and cost burden than those without neuropsychiatric disorders. Early intervention and management of neuropsychiatric disorders could help reduce HCRU and healthcare costs while improving the long-term health of PLWH. Further investigation of the role of HIV, including ART, in the development of neuropsychiatric disorders may also help improve individualization of care and prevent excess HCRU and costs.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Shweta Kamat, Girish Prajapati, Viktor Chirikov, and Mark Bounthavong contributed to the study concept and design. Shweta Kamat, Girish Prajapati, Viktor Chirikov, and Wenying Quan contributed to the collection and assembly of data. Sean P. Fleming, Shweta Kamat, Viktor Chirikov, Wenying Quan, and Mark Bounthavong provided data analysis and interpretation. Sean P. Fleming, Shweta Kamat, and Viktor Chirikov contributed to manuscript writing and all authors approved the final manuscript.

Funding

The study was funded by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA, who also funded the journal’s Rapid Service and Open Access fees.

Data Availability

The data that support the findings of this study are available from Optum’s de-identified Clinformatics^® Data Mart Database. Restrictions apply to the availability of these data, which were used under license for this study, such that the data used to generate these results cannot be disclosed publicly. Additional detailed results may be available upon request to the authors.

Declarations

Conflict of Interest

Sean P. Fleming and Girish Prajapati are employees of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA and shareholders of Merck & Co., Inc., Rahway, NJ, USA. Shweta Kamat, Viktor Chirikov, and Wenying Quan are employees of OPEN Health Group, which was paid to conduct the study under contract with Merck & Co., Inc., Rahway, NJ, USA. Mark Bounthavong is an employee of the University of California San Diego and a consultant for Merck & Co., Inc., Rahway, NJ, USA.

Ethical Approval

This study used de-identified data and therefore does not meet the National Institute of Health (NIH) guidelines for clinical research and thus, informed consent and Institutional Review Board/Ethics Committee review were not required.