Geriatric Conditions Associated with Nonadherence to Antiretroviral Therapy Among Older People with HIV: The Importance of Frailty

Abstract

Poor compliance with medications is a growing concern in geriatric care and is increasingly more relevant among people living with HIV (PLWH) as they age. Our goal was to understand geriatric conditions associated with antiretroviral therapy (ART) nonadherence in a Medicare population of older PLWH. We analyzed Medicare data from PLWH aged 50 years or older who were continuously enrolled in fee-for-service Medicare from January 1, 2014 to June 30, 2015. Prevalent geriatric conditions (dementia, depression, falls, hip fracture, sensory deficits, osteoporosis, orthostatic hypotension, urinary incontinence, frailty) were identified in January 1, 2014–December 31, 2014. ART nonadherence was defined as <80% proportion of days covered (PDC) by at least two ART medications in January 1, 2015–June 30, 2015. We examined geriatric condition association with nonadherence using lowest Akaike Information Criterion multi-variate logistic models, controlling for age, sex, race, census region, substance use, Medicaid eligibility, and polypharmacy. Of 8778 PLWH, 23% (n = 2042) had <80% PDC. The average age was 60 years (standard deviation ±8), and >70% were males. In adjusted models, age was not associated with nonadherence, frailty status was the only geriatric condition associated with nonadherence [robust: reference, prefrail odds ratio (OR): 0.97, confidence interval (95% CI) 0.86–1.10, frail OR: 1.34 95% CI 1.11–1.61], and odds of nonadherence were lower for polypharmacy [OR: 0.48 (0.43–0.54)]. Our findings suggest that patient-centered care plans aimed at improving ART adherence among older PLWH would benefit from long-term surveillance; a deeper understanding of the role of frailty and polypharmacy, even at chronologically younger ages in PLWH.

Introduction

High adherence with HIV antiretroviral therapy (ART) is essential if people living with HIV (PLWH) are to achieve viral suppression and realize the full benefits of ART. ART is indicated no matter the patient's age or symptoms.¹ Older age has been associated with better ART adherence,^2
–4 but it is important to consider that older PLWH are a mix of long-term survivors and people diagnosed with HIV at older ages; 20% of new infections in the United States are among people aged 50 or older.⁵ A summary of HIV cohort studies demonstrated the consistency with which older PLWH are at increased risk for comorbidities such as cardiovascular disease, dyslipidemia, and hypertension; older protease inhibitor-based ART regimens contribute to risk for cardiovascular disease⁶ and the latter potentially linked to greater risk for heart attack.⁷ Although HIV may be one of multiple factors to explain this elevated risk for comorbidities, PLWH with comorbid burden are younger as a cohort than people without HIV and more often have high risk behaviors such as smoking and misuse of drugs and alcohol that also make them more likely to have certain comorbidities. The summary is that >50% of PLWH are older than age 50, and they have a mix of comorbidities that may accumulate before an HIV diagnosis and subsequent ART initiation.^7,8

Accumulating comorbidities have been associated with ART nonadherence.^2,9 Particularly true for PLWH with severe illness, cognitive impairment, including deficits in executive function, and memory impairment;^3,4 having neurological or psychological impairment PLWH were 2.5 times more likely to be nonadherent.³ However, other studies have shown contradicting associations of nonadherence with mental health conditions.^4,10 In particular Barclay et al.⁴ showed higher likelihood of nonadherence among those with diagnosed depression but Mutchler et al.¹⁰ demonstrated lower likelihood with self-reported depression among a subgroup of older Black men who have sex with men.¹¹ Moreover, Mann et al.¹¹ showed lower ART adherence for PLWH and diabetes but not with HIV and hypertension or hyperlipidemia.¹¹ It has been shown that increasing comorbidities raise the risk of ART nonadherence by up to 25%.¹² One recent study showed that having incrementally more comorbidities there is a complimentary reduction in likelihood of adherence.⁹ However, the lack of agreement across the literature and inconsistent consideration of non-ART medication use to treat comorbidities suggest that there are nuances to these associations demanding further confirmation and exploration of other subgroups.

Geriatric conditions are a subgroup within comorbidities that are also accumulating among older PLWH. There is relative lack of literature examining geriatric conditions associated with ART nonadherence, which potentially misses an opportunity for improving health care. Geriatricians specialize in treatment of geriatric conditions and have been incorporated into select HIV clinics around the world.^13,14 Despite these programs being highly heterogeneous in their goals, care linkages, and overall implementation,¹⁴ geriatric care in HIV clinics has been well received by patients.¹³ Yet, universal implementation of geriatric assessments and treatment of geriatric conditions are not included as standard for older PLWH, particularly if they are thought to be of younger chronologic ages than more traditional geriatric patients.¹⁴ To that end, a previous study found that 23.1% of older adults have at least two geriatric conditions,¹⁵ while more than half of aged 50+ people with HIV have at least two geriatric conditions.^16
–18

Geriatric conditions, including but not limited to frailty, depression, and dementia, are distinguished from other comorbidities in that any one geriatric condition can result from multiple pathways of organ system dysregulation or impairment.¹⁹ In conjunction with comorbidities,²⁰ accumulating geriatric conditions amount to reduced physiologic reserve or frailty.^16,19,21 Frailty is a geriatric condition in its own right. One way it can be measured is using a deficit accumulation frailty index (FI), which represents the total burden of age-related health deficits and is dynamic in time.^22,23 Among PLWH, frailty has been shown to result from long-term ART, an AIDS diagnosis, and cognitive impairment.²⁴ As geriatric conditions accumulate so too do the medications to treat them resulting in polypharmacy, which is often considered a geriatric condition unto itself. Older PLWH are predisposed to polypharmacy and take approximately three more medications than age-matched non-HIV infected individuals.^12,17,25,26 Polypharmacy has been linked to nonadherence with certain medications, but not ART.²⁷

Further developing an understanding of geriatric conditions among PLWH means greater opportunities for clarifying the role of geriatrics in HIV care, including anticipating and/or preventing issues with ART adherence as the HIV population ages. To that end, geriatric conditions are more often studied as outcomes. For example, viremic HIV, long-term ART, and older age are independently associated with osteoporosis,²⁸ and polypharmacy is associated with frailty.²⁹ However, as previously noted, geriatric conditions may exist before HIV diagnoses and ART initiation; if comorbidities are associated with nonadherence it is possible that geriatric conditions are as well. Clarifying the role of geriatric conditions within the literature examining associations with nonadherence will be important for resource allocation and strategic approaches to geriatric HIV care.

The objective of this study was to understand whether geriatric conditions are associated with ART nonadherence among a Medicare population of older PLWH. We hypothesized that the presence of frailty and other geriatric conditions will be associated with ART nonadherence.

Methods

Data sources

We used an enhanced 5% sample of 2014–2016 Centers for Medicare and Medicaid Services (CMS) claims data, which ensures that anyone who enters the sample stays in the sample for longitudinal follow-up. These data include the following: chronic conditions data warehouse (CCW) files, the master beneficiary summary file, Medicare Part A inpatient claims, Medicare Part B outpatient claims, and Part D prescription claims. The CCW captures historical data on first diagnosis of chronic conditions, even that predate these data.

Population

PLWH were identified as anyone in 3 years of data having at least one of the following: (1) one diagnosis for HIV or AIDS in inpatient claims or (2) two diagnoses at least 90 days apart in outpatient claims or (3) at least two antiretroviral medications prescribed in the 6 months before the index date (n = 20,572); this enables capture of anyone without diagnosis codes. We excluded a total of 10,015; that included anyone younger than 50 (n = 5971), deceased before the index date (n = 439), hospital or nursing home inpatients at the index date (n = 322), who lived outside of the 50 US states in 2015 (n = 1332), or who were not eligible for Medicare (A, B, D) continuously from January 1, 2014 to June 30, 2015 (n = 5277). Medicare eligibility ensures longitudinal follow-up. We also excluded anyone with no ART after the index date (n = 1779) because we are defining adherence as medication implementation (definition details below).³⁰ Our final sample included 8778 PLWH on January 1, 2015, the index date.

Covariates and characteristics

The 2015 master beneficiary summary file provided baseline age, race (non-Hispanic White, non-Hispanic Black, Other), sex, US census region [Northeast, South, Midwest, West (including Hawaii and Alaska)], and if a person was dually eligible for Medicaid for at least 50% of the year before the index date; Medicaid dual eligibility is the only proxy for income available in these claims data. Multi-morbidity was an indicator for having two or more conditions (other than HIV) in the CCW with diagnosis dates before the index date. Polypharmacy was calculated as five or more anatomic therapeutic chemical third level classes of drugs (ATC3) for at least 90 days in the 6 months before the index date, excluding ART drugs. History of substance use was calculated as having at CCW date before the index date for at least one of: tobacco, alcohol, opioid use disorder, or use of other drugs.

Exposures

Geriatric conditions assessed in the 1-year preceding the index date were identified by one diagnosis in inpatient claims or two diagnoses in outpatient claims at least 90 days apart. Geriatric conditions included the following: Alzheimer's and related dementias, depression, falls, hip fracture, sensory deficits (eye/ear), osteoporosis, orthostatic hypotension, and urinary incontinence. Frailty was measured using a validated claims-based FI from the 1-year look back period. Frailty categories were determined using common cut-points to classify people as robust (FI <0.15), prefrail (0.15 ≤ FI <0.25), and frail (0.25 ≤ FI).^22,23

Outcome

Nonadherence was defined as a function of dosing history or implementation of medication.³⁰ We measured implementation as the proportion of days covered (PDC) by at least two antiretroviral medications in the 6 months following the index date, excluding anyone without ART at the index date.³¹ A PDC of 80% or better was used because of higher potency of newer medications;² sensitivity analyses use 85%; and 95% PDC were also performed. We chose 6 months to allow for prescriptions longer than 30 days.³²

Statistical analysis

Chi-squared test for independence determined multi-morbidity was significantly correlated (p < 0.01) with all variables, but sex, and so was not considered for multi-variate regression models. Polypharmacy and Medicaid eligibility were correlated (p < 0.01) with some, not all, model variables. They were stepped into regression models because they were adjudicated as clinically meaningful given that the focus of the study is medication and both represent factors associated with medication access³³ and adherence.²⁷ All geriatric conditions were assessed in logistic regression with nonadherence as the outcome. Bivariate analysis of each geriatric condition and covariates was followed by adjusted logistic regression of all geriatric conditions on nonadherence. To avoid overfitting, geriatric conditions were chosen based on lowest Akaike Information Criterion (AIC) where age, sex, race, census region, history of substance use, Medicaid eligibility, and polypharmacy were forced. This study has been approved by the Hebrew SeniorLife Institutional Review Board.

Post hoc analysis

Frailty status was the only geriatric condition in all adjusted models. To try to understand health profiles for frail and prefrail PLWH by adherence status, we calculated the difference between the proportion of nonadherent and proportion of adherent PLWH that had each of a list of geriatric and chronic conditions among the frail and among the prefrail groups.

Results

Of 8775 PLWH, the average PDC was 86% (standard deviation,³⁴ 50.1%; median, 94%; range, 0.5%–100%; Fig. 1). Across all PLWH, 51% (n = 4512) had <95% PDC, 29% (n = 2552) has <85% PDC, 23% (n = 2042) had <80% PDC, where mean PDC was 73%, 59%, and 53%, respectively (Table 1). The average age was 60.1 years (standard deviation 7.6); nonadherent groups had mean age range 59.6 (7.5) to 59.8 (7.5). More than 70% of all nonadherent groups were males (95% PDC: 73%; 85% PDC: 72%; 80% PDC: 71%) and a plurality was Black (95% PDC: 44.9%; 85% PDC: 49.1%; 80% PDC: 51.1%). Approximately 30% of nonadherent groups had a history of substance use (95% PDC: 29%; 85% PDC: 31%; 80% PDC: 33%). Figure 2 shows the distribution of geriatric conditions among nonadherence groups where depression (95% PDC: 22%; 85% PDC: 23%; 80% PDC: 22%) and frailty status (95% PDC: prefrail: 32%, frail: 11%; 85% PDC: prefrail: 33%, frail: 12%; 80% PDC: prefrail: 32%, frail: 12%) were the most common.

FIG. 1.

Distribution of the proportion of days covered across all older PLWH (n = 8778). The lines denote the mean (solid) and median (dotted) proportion of days covered: 86% (standard deviation 50.1%), 94%. PLWH, people living with HIV.

FIG. 2.

Percent of each nonadherence group having geriatric conditions.

Table 1.

Descriptive Statistics by Nonadherence Threshold

	All PLWH (n = 8778)	Nonadherent (<95% PDC ^a ) (n = 4512)	Nonadherent (<85% PDC) (n = 2552)	Nonadherent (<80% PDC) (n = 2042)
	n (%)	n (%)	n (%)	n (%)
PDC, mean (SD)	86 (50)	73 (23)	59 (22)	53 (21)
Age, mean (SD)	60.1 (7.6)	59.8 (7.5)	59.6 (7.5)	59.6 (7.5)
Sex
Female	2146 (24)	1202 (27)	709 (28)	593 (29)
Male	6632 (76)	3310 (73)	1843 (72)	1449 (71)
Race
White	3815 (43)	1749 (39)	908 (36)	681 (33)
Black	3510 (40)	2025 (45)	1254 (49)	1043 (51)
Hispanic	1221 (14)	619 (14)	311 (12)	252 (12)
Other and unknown	232 (3)	119 (3)	79 (3)	66 (3)
Census region
Midwest	1097 (12)	569 (13)	301 (12)	231 (11)
Northeast	2104 (24)	1069 (24)	607 (24)	497 (24)
South	3726 (42)	2004 (44)	1160 (46)	931 (46)
West	1851 (21)	870 (19)	484 (19)	383 (19)
Dually eligible for Medicaid	3951 (45)	1977 (44)	1121 (44)	926 (45)
Original eligibility reason
Age 65+	1398 (16)	689 (15)	371 (15)	298 (15)
Disability	7182 (82)	3707 (82)	2109 (83)	1682 (82)
Other reason	198 (3)	116 (3)	72 (3)	62 (3)
Current eligibility reason
Age 65+	2600 (30)	1277 (28)	711 (28)	571 (28)
Disability	6048 (69)	3162 (70)	1794 (70)	1431 (70)
Other reason	130 (1)	73 (2)	47 (2)	40 (2)
Polypharmacy	4549 (52)	2042 (45)	1039 (41)	796 (39)
Prevalent multi-morbidity	750 (9)	403 (9)	221 (9)	183 (9)
History of substance use	2259 (26)	1287 (29)	800 (31)	664 (33)
Geriatric conditions
Alzheimer's and related dementias	197 (2)	98 (2)	59 (2)	45 (2)
Depression	1923 (22)	989 (22)	573 (23)	450 (22)
Falls	298 (3)	154 (3)	104 (4)	86 (4)
Hip fracture	47 (1)	23 (1)	15 (1)	13 (1)
Sensory impairment (eye, ear)	344 (4)	159 (4)	84 (3)	70 (3)
Osteoporosis	393 (4)	180 (4)	101 (4)	82 (4)
Orthostatic hypotension	57 (1)	33 (1)	16 (1)	11 (1)
Incontinence	178 (2)	90 (2)	48 (2)	41 (2)
Frailty status
Robust	4947 (56)	2557 (57)	1420 (56)	1142 (56)
Prefrail	2929 (33)	1460 (32)	831 (33)	647 (32)
Frail	902 (10)	495 (11)	301 (12)	253 (12)

Multi-morbidity = having ≥2 conditions other than geriatric conditions with diagnosis dates before January 1, 2015 in the chronic condition data warehouse files.

PDC covered where nonadherence was defined as <95%, 85%, or 80% of days in the 6 months after January 1, 2015 covered by 2+ antiretroviral medications.

PDC, proportion of days covered; PLWH, people living with HIV; SD, standard deviation.

Tables 2 and 3 show the bivariate and main analyses, respectively. The main adjusted logistic regression used <80% PDC as the threshold for nonadherence. Age was marginally not associated with nonadherence, and frailty was the only geriatric condition significantly associated with nonadherence [robust: reference, prefrail odds ratio (OR): 0.97 95% confidence interval (CI) 0.86–1.10, frail OR: 1.34, 95% CI 1.11–1.61]. No other geriatric conditions were associated with nonadherence; despite that falls were included in the lowest AIC model. A history of substance use (OR 1.59, 95% CI 1.41–1.79) and races compared to White were associated with increased odds of nonadherence (Black OR: 1.77, 95% CI 1.57–2.00; Hispanic OR: 1.20, 95% CI 1.02–1.42; Other races OR: 1.85, 95% CI 1.36–2.50). These findings persisted in sensitivity analyses (Table 4).

Table 2.

Bivariate Logistic Regression Across Three Thresholds for Nonadherence

	Nonadherent (<95% PDC) (n = 4512)	Nonadherent (<85% PDC) (n = 2552)	Nonadherent (<80% PDC) (n = 2042)
	OR (95% CI)	OR (95% CI)	OR (90% CI)
Age	0.99 (0.98–1.00)	0.99 (0.98–1.00)	0.99 (0.98–1.00)
Sex
Male	0.78 (0.71–0.86)	0.78 (0.7–0.87)	0.73 (0.66, 0.82)
Female (Ref.)	—	—	—
Race
Black	1.61 (1.47–1.77)	1.78 (1.61–1.97)	1.95 (1.74–2.17)
Hispanic	1.22 (1.07–1.38)	1.09 (0.94–1.27)	1.2 (1.02–1.41)
Other and unknown	1.24 (0.95–1.62)	1.65 (1.25–2.19)	1.83 (1.36–2.46)
White (Ref.)	—	—	—
Census Region
Midwest	1.04 (0.90–1.21)	0.93 (0.79–1.10)	0.86 (0.72–1.03)
South	1.13 (1.01–1.25)	1.12 (0.99–1.25)	1.08 (0.95–1.22)
West	0.86 (0.76–0.97)	0.87 (0.76–1.01)	0.84 (0.73–0.98)
Northeast (Ref.)	—	—	—
History of substance use	1.35 (1.23–1.49)	1.49 (1.35–1.65)	1.55 (1.39–1.73)
Geriatric conditions
Polypharmacy	0.58 (0.53–0.63)	0.53 (0.48–0.58)	0.51 (0.46–0.56)
Frailty status
Prefrail	0.93 (0.85–1.02)	0.98 (0.89–1.09)	0.95 (0.85–1.05)
Frail	1.14 (0.99–1.31)	1.24 (1.07–1.45)	1.3 (1.11–1.52)
Robust (Ref.)	—	—	—
Alzheimer's and related dementias	0.93 (0.71–1.24)	1.04 (0.77–1.42)	0.98 (0.70–1.37)
Depression	1.00 (0.91–1.11)	1.05 (0.94–1.17)	1.01 (0.90–1.14)
Falls	1.01 (0.80–1.28)	1.32 (1.04–1.68)	1.35 (1.05–1.75)
Hip fracture	0.91 (0.51–1.61)	1.14 (0.62–2.12)	1.26 (0.67–2.40)
Sensory impairment (eye, ear)	0.81 (0.65–1.00)	0.78 (0.61–1.00)	0.84 (0.64–1.09)
Osteoporosis	0.79 (0.65–0.97)	0.84 (0.67–1.06)	0.86 (0.67–1.11)
Orthostatic hypotension	1.30 (0.77–2.21)	0.95 (0.53–1.70)	0.79 (0.41–1.52)
Incontinence	0.97 (0.72–1.30)	0.90 (0.64–1.26)	0.99 (0.69–1.40)

Bold = p value <0.05.

PDC, proportion of days covered; Ref., reference category.

Table 3.

Adjusted Odds Ratios for Nonadherence Using the Cutoff <80% of Days Covered

	Model 1 ^a OR (95% CI)	Model 2 ^b OR (95% CI)	Model 3 ^c OR (95%CI)	Model 4 ^d OR (95% CI)
Age	0.99 (0.99–1.00)	0.99 (0.99–1.00)	0.99 (0.99–1.00)	0.99 (0.99–1.00)
Sex
Male	0.86 (0.77–0.95)	0.88 (0.79–0.99)	0.84 (0.75–0.95)	0.84 (0.75–0.95)
Female (Ref.)	—	—	—
Race
Black	1.83 (1.63–2.06)	1.88 (1.67–2.12)	1.78 (1.58–2.01)	1.77 (1.57–2.00)
Hispanic	1.20 (1.02–1.41)	1.25 (1.06–1.48)	1.20 (1.01–1.42)	1.20 (1.02–1.42)
Other	1.91 (1.42–2.58)	1.95 (1.45–2.64)	1.84 (1.36–2.49)	1.85 (1.36–2.50)
White (Ref.)	—	—	—
Census region
Midwest	0.92 (0.76–1.10)	0.89 (0.74–1.07)	0.90 (0.75–1.08)	0.89 (0.74–1.07)
South	1.05 (0.92–1.19)	1.00 (0.88–1.14)	1.02 (0.90–1.17)	1.02 (0.90–1.17)
West	1.05 (0.90–1.23)	1.04 (0.89–1.22)	1.07 (0.91–1.26)	1.08 (0.92–1.26)
Northeast (Ref.)	—	—	—
History of substance use	1.52 (1.36–1.69)	1.55 (1.39–1.73)	1.66 (1.49–1.86)	1.59 (1.41–1.79)
Dually eligible		0.83 (0.75–0.93)	0.90 (0.80–1.00)	0.89 (0.79–0.99)
Polypharmacy			0.50 (0.45–0.56)	0.48 (0.43–0.54)
Frailty status
Prefrail	—	—	—	0.97 (0.86–1.1)
Frail	—	—	—	1.34 (1.11–1.61)
Robust (Ref.)	—	—	—	—
Alzheimer's and related dementias	—	—	—	^*
Depression	—	—	—	^*
Falls	—	—	—	1.29 (0.98–1.69)
Hip fracture	—	—	—	^*
Sensory impairment (eye, ear)	—	—	—	^*
Osteoporosis	—	—	—	^*
Orthostatic Hypotension	—	—	—	^*
Incontinence	—	—	—	^*

Bold = p value <0.05.

These variables were excluded from the lowest AIC model.

Model 1 = independent effects of demographic characteristics.

Model 2 = Model 1 + independent effects of dual eligibility.

Model 3 = Model 2 + independent effects of polypharmacy.

Model 4 = Model 3 + All 9 geriatric conditions (frailty, Alzheimer's and related dementias, depression, falls, hip fracture, sensory impairment, osteoporosis, orthostatic hypotension, urinary incontinence) selected down by lowest AIC.

AIC, Akaike Information Criterion; ART, antiretroviral therapy; CI, confidence interval; dually eligible, eligible for both Medicare and Medicaid; OR, odds ratio; PDC, proportion of days covered; Ref., reference category.

Table 4.

Sensitivity Analysis: Multi-Variate Logistic Regression Among Different Levels of Nonadherence

	<95% PDC OR (95% CI)	<85% PDC OR (95% CI)
Age	0.99 (0.99–1.00)	0.99 (0.99–1.00)
Sex
Male	0.82 (0.74–0.91)	0.85 (0.76–0.95)
Female (Ref.)	—	—
Race
Black	1.49 (1.35–1.65)	1.66 (1.48–1.85)
Hispanic	1.27 (1.11–1.46)	1.13 (0.97–1.32)
Other	1.29 (0.99–1.70)	1.70 (1.27–2.27)
White (Ref.)	—	—
Census region
Midwest	1.07 (0.92–1.25)	0.94 (0.79–1.11)
South	1.07 (0.95–1.19)	1.04 (0.92–1.18)
West	1.02 (0.90–1.16)	1.07 (0.93–1.25)
Northeast (Ref.)	—	—
History of substance use	1.35 (1.22–1.50)	1.53 (1.37–1.72)
Dually eligible	0.82 (0.74–0.90)	0.83 (0.75–0.92)
Polypharmacy	0.58 (0.53–0.63)	0.51 (0.46–0.56)
Frailty status
Prefrail	0.99 (0.89–1.09)	1.02 (0.92–1.15)
Frail	1.24 (1.06–1.45)	1.3 (1.09–1.55)
Robust (Ref.)	—	—
Alzheimer's and related dementias	—	—
Depression	—	—
Falls	—	1.27 (0.98–1.64)
Hip fracture	—	—
Sensory impairment (eye, ear)	—	—
Osteoporosis	0.83 (0.68–1.03)	—
Orthostatic hypotension	—	—
Incontinence	—	—

Bold = p value <0.05.

PDC, proportion of days covered; Ref., reference category.

Post hoc analysis is presented in Table 5; there are four conditions among prefrail PLWH that have significantly different proportions across all adherence thresholds: polypharmacy (−12, −18, −18 or 12 and 18 percentage points lower among nonadherent than the adherent in the 95%, 85%, and 80% thresholds, respectively), cataracts (−3, −5, −5), hyperlipidemia (−7, −10, −2), and hypothyroidism (−3, −3, −5). In addition to these four, there were two more conditions (rheumatoid arthritis, benign prostatic hyperplasia) which had significantly different proportions across the nonadherent and adherent prefrail group at the 95% PDC threshold, there were eight more conditions [osteoporosis, urinary incontinence, myocardial infarction (MI), diabetes, glaucoma, ischemic heart disease, rheumatoid arthritis, benign prostatic hyperplasia] among the prefrail group at the 85% PDC threshold, and there were six more conditions [MI, chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), diabetes, ischemic heart disease, asthma] among the prefrail group at the 80% PDC threshold. Only at the lowest threshold for nonadherence were there conditions with higher prevalence in the prefrail nonadherent than the adherent groups (MI, COPD, CHF).

Table 5.

Post hoc Difference Between Proportions of Nonadherent and Adherent Older People Living with HIV (PLWH) That Have Geriatric and Chronic Conditions

	<95% PDC				<85% PDC				<80% PDC
	Frail		Prefrail		Frail		Prefrail		Frail		Prefrail
	Δ Proportion	p	Δ Proportion	p	Δ	p	Δ	p	Δ	p	Δ	p
Sex
Female	9	<0.01	3	0.05	8	0.02	5	0.02	9	<0.01	6	<0.01
Male	−9		−3		−8		−5	0	−9		−6
Race/ethnicity
White	−8	0.05	−8	<0.01	−9	<0.01	−9	<0.01	−12	<0.01	−11	<0.01
Black	8		9		11		12	0	12		14
Hispanic	0		−1		−3		−4	0	−1		−4
Other	0		−1		1		1	0	1		1
Census region
Midwest	−1	0.1	0	<0.01	−1	0.86	−2	<0.01	−1	0.71	−2	<0.01
Northeast	−4		−2		0		−5		2		−3
South	8		6		2		9		2		8
West	−3		−4		−2		−3		−3		−2
Dually eligible	−8	0.01	−7	<0.01	−9	0.01	−9	<0.01	−6	0.13	−6	0.01
Prevalent conditions
Alzheimer's and related dementias	0	0.98	0	0.39	1	0.68	0	0.5	0	0.94	0	0.31
Depression	−2	0.48	0	0.99	0	0.83	0	0.87	−3	0.49	−1	0.57
Fall	−5	0.02	1	0.25	2	0.50	2	0.13	2	0.41	1	0.27
Hip fracture	1	0.27	−1	0.32	2	0.07	−1	0.16	2	0.10	−1	0.35
Sensory impairment	−1	0.64	0	0.83	0	0.93	−1	0.28	0	0.80	−1	0.36
Osteoporosis	−3	0.20	−1	0.14	−1	0.74	−2	0.03	−1	0.64	−1	0.26
Orthostatic hypotension	1	0.75	1	0.34	0	0.99	−1	0.56	−1	0.62	−1	0.26
Urinary incontinence	2	0.24	−1	0.18	1	0.50	−1	0.04	1	0.48	−1	0.08
Polypharmacy	−12	<0.01	−16	<0.01	−14	<0.01	−18	<0.01	−16	<0.01	−18	<0.01
Chronic conditions
Myocardial infarction	−1	0.35	0	0.40	1	0.90	0	0.02	0	0.99	1	<0.01
Atrial fibrillation	−3	0.11	−1	0.15	−1	0.53	0	0.37	−2	0.20	0	0.55
Cataracts	1	0.77	−3	<0.01	0	0.99	−5	<0.01	−1	0.75	−3	<0.01
Chronic kidney disease	0	0.93	−1	0.27	5	0.10	−3	0.09	5	0.17	−3	0.17
Chronic obstructive pulmonary disease	−2	0.50	1	0.60	−7	0.06	2	0.18	−6	0.09	3	0.03
Congestive heart failure	−2	0.53	1	0.65	2	0.44	2	0.11	1	0.91	5	<0.01
Diabetes	1	0.68	−3	0.06	2	0.65	−7	<0.01	3	0.46	−7	<0.01
Glaucoma	−1	0.65	−1	0.18	−1	0.72	−2	0.01	−1	0.82	−2	0.1
Ischemic heart disease	−9	<0.01	−1	0.44	−5	0.12	−5	<0.01	3	0.48	−5	<0.01
Rheumatoid arthritis	−1	0.71	−4	<0.01	0	0.94	−5	<0.01	0	0.92	0	0.67
Stroke	−3	0.23	0	0.35	1	0.72	−1	0.14	−2	0.39	0	0.51
Any cancer (colorectal, endometrial, breast, lung, prostate)	−2	0.18	0	0.41	−1	0.51	1	0.18	0	0.98	0	0.91
Anemia	−1	0.88	−2	0.5	−1	0.9	1	0.54	0	0.97	1	0.72
Asthma	−1	0.78	1	0.54	1	0.92	1	0.91	−12	<0.01	−12	<0.01
Hyperlipidemia	−8	0.01	−7	<0.01	−12	<0.01	−10	<0.01	−2	0.31	−2	0.03
Benign prostatic hyperplasia	−4	0.12	−3	<0.01	−1	0.44	−2	0.01	5	0.05	−4	0.06
Hypertension	1	0.62	−2	0.35	3	0.24	−3	0.13	−3	0.23	−3	0.05
Hypothyroidism	0	0.89	−3	<0.01	0	0.94	−3	<0.01	3	0.48	−5	<0.01

Bold = p value <0.05.

Values are absolute differences between the adherent and nonadherent proportions (Δ = %nonadherent − %adherent). Example interpretations: In the frail stratum and <95% PDC, the proportion of females was 9 percentage points higher in the nonadherent compared to the adherent group and the proportion dually eligible was 8 percentage points lower in the nonadherent compared to the adherent group.

Δ, difference; PDC, proportion of days covered; PLWH, people living with HIV.

Discussion

Accumulating comorbidities have been associated with nonadherence to certain medications, although it is less clear if the same effect is observed among older PLWH and ART.^2,9 Our goal was to understand if geriatric conditions, like other comorbidities, are associated with ART nonadherence in a Medicare population of older PLWH. We hypothesized that adherence to ART would be lower for older PLWH with frailty and other geriatric conditions. We found that frailty status was the only geriatric condition consistently associated with ART adherence at all thresholds for ART nonadherence. To that end, robust and prefrail people were similar, and being frail was associated with ∼30% higher likelihood of ART nonadherence.

Older PLWH make up 50% of the HIV population and nearly 20% of new infections in the United States.⁵ Regardless of the debate about whether PLWH experience accelerated or accentuated aging, older PLWH are accumulating geriatric conditions at younger ages than their non-HIV counterparts;^16,35 the most prevalent in our sample were depression, prefrailty, and frailty. Polypharmacy was present in more than half, although fewer of the nonadherent than adherent had polypharmacy. To that end, frailty was associated with higher likelihood of nonadherence. Frailty, in this case, was an assessment of total burden of deficit because we used a deficit accumulation FI.^22,23 The utility of knowing frailty's relationship with nonadherence is in the opportunity to understand what deficits make up the FI; having a FI indicative of frailty does not necessarily mean the same deficits have accumulated.³⁶ This is particularly true because of the younger chronologic age at which PLWH begin developing geriatric conditions and the unique experiences PLWH have with intersecting effects of HIV, treatment, and aging.^35,37 Post hoc analysis demonstrated that there was greater heterogeneity between prefrail nonadherent and adherent older PLWH than among the frail group. Heterogeneity among prefrail older adults creates challenges for identifying and intervening where frailty is more homogenous in global deficits. Research has shown that one approach may be to use latent class analysis among each frailty subgroup to identify dominant deficits for targeted intervention to maintain or improve ART adherence.³⁶

Polypharmacy was associated with lower likelihood of nonadherence. In our sample a significantly larger proportion of the adherent group had polypharmacy, which may mean that they are high utilizers of health care, seen often by providers and have many comorbidities that each come with a treatment.³⁸ This agrees with our overall finding that prefrailty had lower likelihood of nonadherence and also the nonadherent prefrail groups had lower proportions of comorbidities than the adherent groups in post hoc analyses. To detect a condition in claims data, patients must be seen by a physician for that condition. Conditions that are treated and controlled reduce opportunities for additional complications. For example, orthostatic hypotension and falls have been associated in the literature and influenced by blood pressure control medications.³⁹ Our findings also agree with the literature showing that PLWH engaged in care do better with ART adherence and viral suppression, which may be a phenomenon pertaining to more than HIV-associated outcomes. Clearly geriatric conditions exist among an older HIV population in Medicare, but this is a specific group of older PLWH who have access to health insurance. Having health insurance has been linked to better health outcomes and disease management, including treatment adherence. Our finding that nonadherence is less likely for those with dual eligibility status demonstrated this. Therefore, factors other than having geriatric conditions and related treatments seem to also contribute to the likelihood that older PLWH would be nonadherent.

The relevance of these findings to clinical practices extends beyond medical conditions and treatments to include demographic factors. Female sex, race/ethnic minorities, and having a history of substance use seemed to have highest risk for nonadherence, a finding consistent with the literature.^40,41 Brown et al. demonstrated that >50% of females and more than two thirds of minorities had suboptimal ART adherence.⁴¹ The association of race with nonadherence may reflect HIV stigma and discrimination that persists in health care,⁴² and substance use is a known risk factor for inconsistencies in adherence.^3,10 However, having secondary insurance might mitigate risk. Eligibility for Medicaid in addition to Medicare, as we had in 45% of this sample, can create greater access to health care, particularly nursing homes, which are otherwise not covered by Medicare and are increasingly relevant to the older HIV population.^35,43 In addition, Medicaid eligibility is indicative of meeting certain criteria, including low income which qualifies people for the low income subsidy and lower cost of prescriptions.³³

PLWH have differing life history exposures of stress, stigma, and comorbid exposures while they age compared to HIV uninfected.²¹ Therefore, there is high value in exploring the potential that geriatric conditions are a subset of comorbidities associated with ART nonadherence. This extends the literature beyond general comorbidities and beyond nonadherence as the presumed cause of poor health outcomes. These findings support that frailty evaluation may need to be part of the routine clinical care of older PLWH and at ages younger than people without HIV. Brothers and Rockwood and Brañas et al. emphasized this point in their work describing the utility of frailty in optimizing geriatric HIV care.^14,44 Finally, deprescribing is an important practice in geriatrics to reduce pill burden and minimize potential for drug–drug interactions and related poor health outcomes.^12,29 However, here we show that polypharmacy in PLWH may indicate appropriate engagement in care, which is different from some literature where polypharmacy is assumed to be excessive treatment often linked to poor health outcomes.²⁹ More work is needed to understand the critical role of appropriate versus inappropriate polypharmacy. As a result, long-term surveillance of frailty and polypharmacy may be a first step at refining patient-centered care aimed at improving ART adherence.

Limitations

This study has several limitations that include limitations in our ability to make causal conclusions because of the cross-sectional nature of the data; we cannot rule out that poor ART adherence precedes geriatric conditions but have done our best to simulate temporality using an index date 1 year into the data time frame. Geriatric conditions selected were suggested by a geriatrician and agreed upon by all authors as pertinent to the topic. However, other conditions commonly associated with aging and multi-system dysregulation, such as diabetes, have also been shown to be associated with ART nonadherence.¹⁶ In defining ART adherence, we used multiple threshold values of 95% commonly used in the literature,^3,31 but tried to adjust for improvements in regimens with less common thresholds of 85% and 80%. We also do not have data on key social determinants of health (other than dual eligibility for Medicaid), such as housing instability or life events such as death of a partner, which could play a role in timing of ART prescribing and how well they are able to adhere. Moreover, we did not know duration of HIV infection or treatment. We required continuous enrollment in Medicare, where longer continuous enrollment would reduce sample size, but potentially capture more prevalent conditions in 1 year look back period. Using the CCW data helped mitigate the risk of inadequate condition capture as it includes history of all conditions, even that predate the data. Finally, we could only estimate ART adherence as a function of implementation using PDC but that does not mean PLWH actually took medication.

In conclusion, adherence to ART regimens in PLWH was associated with frailty and polypharmacy. Documenting accumulated deficits in health records and claims may represent who is engaged in care and treated for geriatric conditions and HIV. These findings add to the literature calling for incorporation of frailty assessments into the care of older PLWH, even at ages younger than are traditionally associated with geriatric care.

Footnotes

Authors' Contributions

B.O.-M.: Conceptualization, Methodology, Formal analysis, Writing—Original draft preparation; I.W.: Validation, Writing—Reviewing and Editing. S.S.: Writing—Reviewing and Editing. M.M.: Writing—Reviewing and Editing; E.P.M.: Writing—Reviewing and Editing; G.O.: Formal analysis; D.H.K.: Validation, Writing—Reviewing and Editing.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

B.O.-M. and S.S. were supported by the Harvard Translational Research in Aging Training Program, T32 AG023480. D.H.K. was supported by R01AG071809.

References

U.S. Department of Health and Human Services. Guidelines for the Prevention and Treatment of Opportunistic Infections in HIV-Infected Adults and Adolescents: Recommendations from the Centers for Disease Control and Prevention. Atlanta, GA: National Institutes of Health and the HIV Medicine Association of the Infectious Disease Society of America; 2013.

Ghidei

, Simone

, Salow

, et al. Aging, antiretrovirals, and adherence: A meta analysis of adherence among older HIV-infected individuals. Drugs Aging, 2013; 30:809–819.

Hinkin

, Hardy

, Mason

, et al. Medication adherence in HIV-infected adults: Effect of patient age, cognitive status, and substance abuse. AIDS (London), 2004; 18 Suppl 1(Supplement 1):S19–S25.

Barclay

, Hinkin

, Castellon

, et al. Age-associated predictors of medication adherence in HIV-positive adults: Health beliefs, self-efficacy, and neurocognitive status. Health Psychol, 2007; 26:40–49.

Centers for Disease Control and Prevention. HIV and Older Americans. Available at: https://www.cdc.gov/hiv/group/age/olderamericans/index.html (Last accessed June 30, 2021 ).

Laurence

, Elhadad

, Ahamed

. HIV-associated cardiovascular disease: Importance of platelet activation and cardiac fibrosis in the setting of specific antiretroviral therapies. Open Heart, 2018; 5:e000823.

Sabin

, Reiss

. Epidemiology of ageing with HIV: What can we learn from cohorts?. AIDS, 2017; 31(Suppl. 2):S121–S128.

Guaraldi

, Orlando

, Zona

, et al. Premature age-related comorbidities among HIV-infected persons compared with the general population. Clin Infect Dis, 2011; 53:1120–1126.

Abara

, Adekeye

, Xu

, et al. Correlates of combination antiretroviral adherence among recently diagnosed older HIV-infected adults between 50 and 64 years. AIDS Behav, 2016; 20:2674–2681.

10.

Mutchler

, Bogart

, Klein

, et al. Age matters: Differences in correlates of self-reported HIV antiretroviral treatment adherence between older and younger Black men who have sex with men living with HIV. AIDS Care, 2019; 31:965–972.

11.

Mann

, Morrow

, Coyle

, et al. Lower cumulative antiretroviral exposure in people living with HIV and diabetes mellitus. J Acquir Immune Defic Syndromes (1999), 2020; 85:483–488.

12.

Edelman

, Gordon

, Glover

, et al. The next therapeutic challenge in HIV: Polypharmacy. Drugs Aging, 2013; 30:613–628.

13.

Siegler

, Burchett

, Glesby

. Older people with HIV are an essential part of the continuum of HIV care. J Int AIDS Soc, 2018; 21:e25188.

14.

Brañas

, Ryan

, Troya

, Sánchez-Conde

. Geriatric-HIV Medicine: The geriatrician's role. Eur Geriatr Med, 2019; 10:259–265.

15.

Lee

, Cigolle

, Blaum

. The co-occurrence of chronic diseases and geriatric syndromes: The health and retirement study. J Am Geriatr Soc, 2009; 57:511–516.

16.

Greene

, Covinsky

, Valcour

, et al. Geriatric syndromes in older HIV-infected adults. J Acquir Immune Defic Syndromes, 2015; 69:161–167.

17.

Kong

, Pozen

, Anastos

, et al. Non-HIV comorbid conditions and polypharmacy among people living with HIV age 65 or older compared with HIV-negative individuals age 65 or older in the United States: A retrospective claims-based analysis. AIDS Patient Care STDs, 2019; 33:93–103.

18.

Cesari

, Marzetti

, Canevelli

, Guaraldi

. Geriatric syndromes: How to treat. Virulence, 2017; 8:577–585.

19.

Inouye

, Studenski

, Tinetti

, Kuchel

. Geriatric syndromes: Clinical, research, and policy implications of a core geriatric concept. J Am Geriatr Soc (JAGS), 2007; 55:780–791.

20.

McMillan

, Gill

, Power

, et al. Comorbidities in older persons with controlled HIV infection: Correlations with Frailty Index Subtypes. AIDS Patient Care STDs, 2020; 34:284–294.

21.

Justice

. HIV and aging: Time for a new paradigm. Curr HIV/AIDS Rep, 2010; 7:69–76.

22.

Shi

, McCarthy

, Mitchell

, Kim

. Changes in predictive performance of a Frailty Index with availability of clinical domains. J Am Geriatr Soc (JAGS), 2020; 68:1771–1777.

23.

Searle

, Mitnitski

, Gahbauer

, et al. A standard procedure for creating a frailty index. BMC Geriatr, 2008; 8:24.

24.

Brothers

, Kirkland

, Guaraldi

, et al. Frailty in people aging with human immunodeficiency virus (HIV) infection. J Infect Dis, 2014; 210:1170–1179.

25.

Greene

, Steinman

, McNicholl

, Valcour

. Polypharmacy, drug-drug interactions, and potentially inappropriate medications in older adults with human immunodeficiency virus infection. J Am Geriatr Soc (JAGS), 2014; 62:447–453.

26.

Holtzman

, Armon

, Tedaldi

, et al. Polypharmacy and risk of antiretroviral drug interactions among the aging HIV-infected population. J Gen Internal Med, 2013; 28:1302–1310.

27.

Siefried

, Mao

, Cysique

, et al. Concomitant medication polypharmacy, interactions and imperfect adherence are common in Australian adults on suppressive antiretroviral therapy. AIDS (London), 2018; 32:35–48.

28.

Finnerty

, Walker-Bone

, Tariq

. Osteoporosis in postmenopausal women living with HIV. Maturitas, 2016; 95:50–54.

29.

Saum

, Schöttker

, Meid

, et al. Is polypharmacy associated with frailty in older people? Results from the ESTHER Cohort Study. J Am Geriatr Soc (JAGS), 2017; 65:e27–e32.

30.

De Geest

, Zullig

, Dunbar-Jacob

, et al. ESPACOMP Medication Adherence Reporting Guideline (EMERGE). Ann Intern Med, 2018; 169:30–35.

31.

DuChane

, Clark

, Hou

, et al. Impact of HIV-specialized pharmacies on adherence to medications for comorbid conditions. J Am Pharm Assoc (2003), 2014; 54:493–501.

32.

Department of Health and Human Services. Management of the Treatment-Experienced Patient: Discontinuation or Interruption of Antiretroviral therapy. Available at: https://clinicalinfo.hiv.gov/en/guidelines/adult-and-adolescent-arv/discontinuation-or-interruption-antiretroviral-therapy?view=full (Last accessed June 5, 2021 ).

33.

Olivieri-Mui

. Access to HIV medication in the community versus a nursing home for the Medicare eligible HIV population. Delaware J Public Health, 2019; 5:74–78.

34.

Langebeek

, Gisolf

, Reiss

, et al. Predictors and correlates of adherence to combination antiretroviral therapy (ART) for chronic HIV infection: A meta-analysis. BMC Med, 2014; 12:142.

35.

High

, Brennan-Ing

, Clifford

, et al. HIV and aging: State of knowledge and areas of critical need for research. A report to the NIH Office of AIDS Research by the HIV and Aging Working Group. J Acquir Immune Defic Syndr (1999), 2012; 60 Suppl 1(Supplement 1):S1–S18.

36.

Olivieri-Mui

, Shi

, McCarthy

, et al. Beyond the health deficit count: Examining deficit patterns in a deficit-accumulation Frailty Index. J Am Geriatr Soc (JAGS), 2021; 69:792–797.

37.

Cahill

, Valadez

. Growing older with HIV/AIDS: New public health challenges. Am J Public Health (1971), 2013; 103:E7–E15.

38.

Hughes

, McMurdo

MET

, Guthrie

. Guidelines for people not for diseases: The challenges of applying UK clinical guidelines to people with multimorbidity. Age Ageing, 2013; 42:62–69.

39.

Juraschek

, Simpson

, Davis

, et al. Effects of antihypertensive class on falls, syncope, and orthostatic hypotension in older adults: The ALLHAT Trial. Hypertension, 2019; 74:1033–1040.

40.

Krentz

, Gill

. The impact of non-antiretroviral polypharmacy on the continuity of antiretroviral therapy (ART) among HIV patients. AIDS Patient Care STDs, 2016; 30:11–17.

41.

Brown

, Littlewood

, Vanable

. Social-cognitive correlates of antiretroviral therapy adherence among HIV-infected individuals receiving infectious disease care in a medium-sized northeastern US city. AIDS Care, 2013; 25:1149–1158.

42.

Johnson Shen

, Freeman

, Karpiak

, et al. The intersectionality of stigmas among key populations of older adults affected by HIV: A thematic analysis. Clin Gerontol, 2019; 42:137–149.

43.

Justice

, Akgün

. What does aging with HIV mean for nursing homes?. J Am Geriatr Soc (JAGS), 2019; 67:1327–1329.

44.

Brothers

, Rockwood

. Frailty: A new vulnerability indicator in people aging with HIV. Eur Geriatr Med, 2019; 10:219–226.