Does access to free medication reduce health system costs? An evaluation of the Dispensary of Hope program

Abstract

BACKGROUND:

Cost-related medication nonadherence—when patients fail to take medication as prescribed because of the cost of the medication—has numerous consequences: more hospitalizations, avoidable deaths, and greater health care expenditures. Dispensary of Hope is a charitable medication access program that collects and distributes pharmaceuticals to pharmacies to dispense free of charge to patients with no insurance, low incomes, and chronic conditions.

OBJECTIVE:

To estimate the differences in medical costs and utilization of hospital patients enrolled in the Dispensary of Hope program relative to those who were not enrolled.

METHODS:

We used administrative claims data from 2 health systems participating in Dispensary of Hope to identify those in the program and a comparison group, respectively. Claims data included emergency department (ED) encounters, inpatient encounters, costs, and prescriptions. Health system 1 (HS1) data began July 1, 2016, and ended December 31, 2019; health system 2 (HS2) data ran from March 10, 2014, to December 31, 2019. Program enrollment dates (index dates) were identified via program registration or prescription fills. We propensity score weighted a comparison population from HS1 and HS2, respectively, to match program patient demographic and comorbidity characteristics. We estimated changes in costs, ED visits, inpatient stays, and primary care sensitive ED visits over time between the 2 groups (difference-indifference) over 18 months preenrollment and postenrollment.

RESULTS:

HS1 comparison (n = 6,714) and Dispensary of Hope (n = 880) groups were balanced on age, sex, race and ethnicity, and comorbidities; both populations were approximately 46 years old, 43% female, 64% White, with an average of 3.0 comorbidities. The HS2 samples were almost 50 years old and a majority female (56%) and Black (55%). Per-person annual costs at HS1 decreased by $3,161 (P < 0.05) more in the Dispensary of Hope group than in the comparison group from the preenrollment to the postenrollment period. Inpatient stays decreased by 200 stays per 1,000 patients per year (P = 0.02) and ED visits increased by 0.32 (P < 0.01) on a yearly basis relative to the comparison group. Primary care sensitive ED visits increased over the period. No results were statistically significant in HS2.

CONCLUSIONS:

We found substantial reductions in costs and inpatient stays for Dispensary of Hope HS1 participants, and we did not find significant results at HS2. Differences between the health systems or patient populations could explain these varying results. Our study represents a rigorous, multistate evaluation that highlights the impact of a charitable medication access program on hospital utilization for the medically underserved population.

Plain language summary

The Dispensary of Hope provides medications free of charge to patients without insurance. We investigated the impact Dispensary of Hope’s services had on costs, stays in the hospital, and visits to the emergency department in 2 health systems. One health system showed drops in costs and stays in the hospital, as well as increases in visits to the emergency department; we did not see any changes in the other system.

Implications for managed care pharmacy

Patients across the United States who are uninsured or underinsured face significant barriers to medication access, including the financial burden from medication costs. Providing access to medications free of charge can improve patient outcomes and lead to healthier lives. The Dispensary of Hope program demonstrates the potential for health systems and affiliated pharmacies to improve access to medications. Improving access to medications is a key step to reducing health disparities and social inequities.

Medication nonadherence—when patients fail to take their medication as prescribed—has numerous consequences: increased rates of comorbid diseases, more hospitalizations, avoidable deaths, and greater health care expenditures.¹ In the United States alone, medication nonadherence is responsible for an estimated 10% of hospitalizations and 125,000 avoidable deaths each year.^2,3 Nonadherence is associated with significantly more annual inpatient health system days, ranging from 1.2 to 5.7 more days per year, depending on the condition.⁴ Nonadherent patients experience 27% more hospital visits than adherent patients.⁵

Medication nonadherence also carries a significant financial burden on the US health care system, which spends up to $300 billion per year in additional hospitalizations, outpatient visits, and other additional medical costs.6 The typical nonadherent patient requires 3 extra medical visits per year, increasing per-patient treatment costs by $2,000 per year.⁷ Medical complications arising from nonadherence result in further annual expenditures per patient: $9,204 for cardiovascular disease, $11,052 for mental health, and $6,310 for diabetes.⁷

Although the reasons for medication nonadherence are complex, one major barrier to adherence is the cost of medicine prescribed; patients consistently report out-of-pocket costs as a top reason for medication nonadherence.^8-11 This concern is voiced by patients experiencing a spectrum of chronic conditions, for which reliably taking prescribed medications can aid in symptom control, prevent disease progression, and improve quality of life. For example, among the US population with atherosclerotic cardiovascular disease, 13% experienced cost-related nonadherence, including 9% each who missed doses and took lower-than-prescribed doses and 11% who intentionally delayed filling a prescription to save money.¹² Of adults with type 1 or type 2 diabetes, 25% reported rationing insulin in the previous year to manage costs.¹³ Patients report even higher nonadherence rates for psychiatric medications because of financial strain (a key aspect of medication access addressed by the Dispensary of Hope).¹⁴

An underlying cause of nonadherence is a lack of insurance or the presence of underinsurance, whereby a patient may have high out-of-pocket costs relative to their income. One example is patients who are ineligible for Medicaid and with incomes below the Federal Poverty Level (FPL) in the so-called Medicaid coverage gap. Patients in this situation may have difficulty locating a dependable source of medication. It can also carry substantial financial risk from out-of-pocket medication costs. That risk can often translate into nonadherence, especially for those with chronic conditions. According to one study, only 47% of underinsured patients with diabetes in the Medicaid gap were adherent to their medications, and 41% of those with major depressive symptoms were adherent and less than 40% with cardiovascular risk factors were adherent.¹⁵

One pathway to improving medication adherence, and therefore potentially reducing utilization, is via charitable medication access programs.^16,17,18 Charitable medication access programs allow patients access to medications that have been made available at the program’s dispensing site for little to no cost. No-cost or low-cost access to medications for eligible patients ensures the patient receives the appropriate medication and counseling without having to consider coverage or copayments, working through a third-party insurer, or requiring adjudication.

In this study, we evaluated Dispensary of Hope, a charitable medication access program headquartered in Nashville, Tennessee. Dispensary of Hope collects and distributes millions of dollars of donated pharmaceuticals. In fiscal year 2022, there were 1,017,732 prescriptions filled by more than 220 pharmacies and clinics across the country for nearly 150,000 patients (see Chima et al 2020¹⁶ for a discussion of Dispensary of Hope distribution strategies and Mospan and Alexander 2015¹⁹ for a description of Dispensary of Hope’s establishment in Tennessee). Dispensary of Hope’s formulary of direct manufacturer/ distributor-donated medication mostly consists of primary care medications to treat chronic diseases, including diabetes, cardiovascular disease, hypertension, dyslipidemia, mental health, and infections. We partnered with 2 health systems to obtain data on pharmacy fills and health system encounters spanning 4 states. Our main objective was to estimate the impact of Dispensary of Hope enrollment on medical costs and utilization.²⁰

METHODS

STUDY DESIGN

This study is a retrospective observational study of pharmacy fills and health system encounters. We used a difference-in-difference estimation approach.

SITE SELECTION

Several health systems partnering with Dispensary of Hope were reviewed for potential partnership with the study. The 2 we included were selected because they had been in the program long enough to permit evaluation (>4 years) and the number of patients served by the program (>300 patients) in each system. We collaborated with health system and pharmacy analysts from each system to assemble data on pharmacy fills, health system stays, and emergency department (ED) visits. This included data for patients enrolled in Dispensary of Hope and a comparison group of similar patients not participating in the program. For health system 1 (HS1), data were provided from hospital facilities and 8 pharmacies affiliated with the health system across 4 states in the Midwest and Southeast. The Dispensary of Hope start date for the 8 pharmacies in HS1 ranged from March of 2008 to May of 2017. For health system 2 (HS2), there was a single hospital and pharmacy location. The pharmacy in HS2 started Dispensary of Hope in March of 2016. Four states were represented in the 2 health systems: Tennessee, Kansas, Wisconsin, and North Carolina.

DATA

We obtained (1) patients’ Dispensary of Hope enrollment data or their prescriptions filled with Dispensary of Hope medication, (2) facility data from the health systems’ administrative claims, and (3) retail pharmacy data for each eligible patient who met the inclusion criteria (details in the Population Descriptions section). These data included ED encounters, inpatient encounters, costs per encounter (which included direct, indirect, fixed, and variable costs for labor, equipment, medical supplies, and drugs), and the prescription history of the Dispensary of Hope patients.

Data from HS1 covered the period from July 1, 2016, through December 31, 2019. The data coverage period began after HS1 completed its transition to a new national electronic health record system in July 2016. Data after December 2019 were not used in the analysis to avoid confounding factors that may have appeared during the COVID-19 pandemic in 2020. Data from HS2 covered the period from March 10, 2014, through December 31, 2019.

SAMPLE SELECTION

The index date for the Dispensary of Hope patients was the first of either the enrollment date for the program or the first fill paid for by the program. Dispensary of Hope patient index dates were limited to the period between January 1, 2018, and June 30, 2018, for HS1 patients and March 4, 2016, through June 30, 2018, for HS2 patients to ensure symmetric 18-month data preenrollment and postenrollment periods. Figures 1 and 2 provide an illustration of our analysis time frame. Almost all (98%) Dispensary of Hope patients included in the study had a least 1 encounter (inpatient or ED visit) in the preenrollment period and 80% of Dispensary of Hope patients had an encounter within 2 weeks of enrollment. The comparison group was also required to have an encounter and a prescription fill. The comparison group’s index date was set 2 weeks after their first inpatient encounter to mimic the timing of enrollment by the Dispensary of Hope patients.

FIGURE 1.

Time Frame of Health System 1 Data

FIGURE 2.

Time Frame of Health System 2 Data

HS1 and HS2 both provided data for 2 groups of patients: the Dispensary of Hope group and a comparison group of patients who did not participate in the Dispensary of Hope program at any point in the study period. The Dispensary of Hope group consisted of patients who were uninsured and at or below 200% of the FPL. They may be enrolled on-site at the pharmacy with a 1-page attestation form or through other processes that the hospital has (ie, case management identifies eligibility during discharge planning or financial services determines patient to be in need of charity care). Pharmacies requalify patients every 6 to 12 months. Eligible patients may not receive all of their prescribed medications from Dispensary of Hope.

The comparison patients illustrate what health system service utilization by the Dispensary of Hope patients would have been during the same time had Dispensary of Hope not subsidized prescription medications. We selected a group of comparison patients in the same locations (ie, HS1 or HS2) as the Dispensary of Hope population included in the study. The comparison group for this analysis included a group of patients who had filled prescriptions at the retail pharmacy as either Medicaid, charity care, or self-pay.

The RTI Institutional Review Board determined this study to be exempt from review (IRB ID STUDY00021180).

COMORBIDITY PROFILES

To adjust for differences in underlying health issues, we created a comorbidity count for the Dispensary of Hope patients and the comparison group. We used the International Classification of Diseases, Ninth Revision (and Tenth Revision), Clinical Modification (ICD-9-CM; ICD-10-CM) diagnosis codes found in administrative data found in the Elixhauser Comorbidity Software from the Administration for Healthcare Quality to identify patients with comorbidities.^21,22 We removed 4 comorbidities (paralysis, peripheral vascular disease, hypothyroidism, and weight loss) based on their costs, prevalence, and whether the medications provided by the Dispensary of Hope would be expected to treat or lessen the cost burden of these diseases. For ease of exposition, we also combined some of the comorbidities into the same category (eg, “diabetes with complications” and “diabetes without complications” into “diabetes”). The final list consisted of 17 comorbidities.

A patient was identified as having a comorbidity if they had at least 1 health system stay or ED visit during the study period with a corresponding primary or secondary diagnosis code for 1 of the 17 comorbidities from our list. To ensure that the impact of the program across patients with different comorbidities equally, we then summed the number of comorbidities to calculate the total count of comorbidities identified during the study period. HS2 data included claims before and after October 1, 2015; thus, we used both ICD-9-CM and ICD-10-CM codes to identify comorbidities (hospitals switched ICD-9-CM to ICD-10-CM codes on October 1, 2015). HS1 data started after October 1, 2015, and thus only had ICD-10-CM codes. We used the total number of comorbidities in our propensity score–weighting approach, as well as in the regression analysis, to further control for any remaining differences between the treatment and control groups.

OUTCOME MEASURES

Our outcomes of interest were health system costs and health care utilization, including ED visits, inpatient stays, and primary care sensitive (PCS) ED visits. Health system costs included the total expenditures for a patient’s health system encounters using contracted rates in each period, for all visits in the period. We inflated health system costs to 2020 using the Personal Consumer Expenditures Health Index.²³ ED visits were identified using an encounter description variable on administrative claims describing whether a health system encounter was to the ED. PCS visits are ED visits that could be reduced in the presence of high-quality care. Using ICD-9-CM and ICD-10-CM codes and the New York University ED Visit Algorithm, we retrospectively assessed the probability that ED visits were PCS visits to the ED.²⁴ Although our study period included 18 months preenrollment and 18 months postenrollment, we annualized our outcome measures to represent costs and utilization over a 12-month period.

ANALYSIS

We applied propensity score weights to the comparison group so that patients with similar characteristics to patients in the Dispensary of Hope group were weighted more heavily in the analysis. This helped ensure a good balance between groups, resulting in an appropriate comparison group for the Dispensary of Hope group. We estimated separate propensity scores for HS1 and HS2. Logistic regression models produced inverse probability weights based on patient age, sex, location, payer, and number of comorbidities.²⁵ We calculated standardized differences between the means of patient characteristics in the weighted comparison group and the Dispensary of Hope group. Covariates with standardized differences below 0.1 were considered well balanced.²⁵ Unbalanced covariates were included in the model executed on the propensity score–weighted sample.

We estimated difference-in-difference models weighted by propensity scores to assess Dispensary of Hope’s impact on outcomes. We selected a generalized linear model with a log link and gamma distribution to estimate costs because costs are nonnegative and have a large portion of observations at zero and a set of observations with large values.²⁶ For inpatient stays, ED visits, and PCS ED visits, we chose a negative binomial model. For combined analyses, we propensity score weighted by HS1 then combined the 2 regressions and adjusted weights. Each model included indicators for the Dispensary of Hope group (vs comparison group) and postenrollment period (vs preenrollment) and interactions between the 2 indicators. The models also accounted for the influence of several other factors that could affect costs and utilization, including age, sex, race and ethnicity, payer type, pharmacy-location fixed effects, and the number of comorbidities. SEs were clustered at the pharmacy-location level to account for correlation within each pharmacy.

Additionally, we ran additional regressions for patients considered to be high users and consistent users. The high users group included patients with 2 or more ED visits or inpatient stays in the preenrollment period. The consistent users included patients who had at least 12 Dispensary of Hope fills in the 18-month postenrollment period. These results are available in the Supplementary Table 1 ^{(136.5KB, pdf)} (available in online article).

Results

PATIENT DEMOGRAPHICS AND COMORBIDITY PROFILES

Table 1 shows the patient demographics and comorbidity profile for the Dispensary of Hope groups and comparison groups at each health system. Comparing the health systems, HS1 and HS2 patients were similar in average age (46-49 years) but differed across sex, race and ethnicity, and comorbidity profiles. The patient population from HS1 had a lower percentage of female patients than HS2 (43% vs 56%) and a higher percentage of White patients (64% vs 30%). At HS2, more than half of the patients reported identifying as Black or African American (55%) compared with about a quarter of patients at HS1. About three-quarters of comparison group patients listed Medicaid or charity care as their insurance type, and the remainder reported a mix of other insurance types.

TABLE 1.

Patient Demographics and Comorbidity Profiles Before and After Propensity Score Weighting

Demographic or comorbidity	HS1			HS2
Before propensity score weighting	Dispensary of Hope (N = 880)	Comparison group (N = 6,714)	Standardized difference	Dispensary of Hope (N = 3,059)	Comparison group (N = 579)	Standardized difference
Age at preenrollment, years	46.2	42.8	0.22	49.0	46.2	0.22
Sex, %
Female	43.1	56.6	0.27	56.1	54.4	0.03
Race and ethnicity, %
Black or African American	23.9	59.5	0.78	55.2	62.2	0.14
Hispanic	3.8	2.1	0.10	–	–	0.00
White	64.4	32.8	0.67	29.6	22.8	0.15
Other race, declined, or missing	8.0	5.5	0.10	15.2	15.0	0.00
Total number of comorbidities	3.0	3.1	0.01	0.9	2.8	0.87
Demographic or comorbidity	HS1			HS2
Before propensity score weighting	Dispensary of Hope (N = 880)	Comparison groupa (N = 6,714)	Standardized difference	Dispensary of Hope (N = 3,059)	Comparison groupa (N = 579)	Standardized difference
Age at preenrollment, years	46.2	46.7	0.04	49.0	48.2	0.06
Sex, %
Female	43.1	43.2	0.00	56.1	55.1	0.06
Race and ethnicity, %
Black or African American	23.9	25.3	0.03	55.2	54.1	0.02
Hispanic	3.8	3.2	0.03	–	–	–
White	64.4	63.6	0.02	29.6	30.0	0.01
Other race, declined, or missing	8.0	7.9	0.00	15.2	15.9	0.02
Total number of comorbidities	3.0	3.0	0.01	0.9	1.5	0.33

^a Comparison group Ns reflect the weighted sample size.

HS = health system.

Comparing the 2 groups at HS1, the average ages of the Dispensary of Hope group and the comparison group were 46.2 and 46.7, respectively. The HS1 groups were balanced on sex and race and ethnicity; both populations were 43% female and 64% White. Each group had an average of 3.0 comorbidities in total. In the HS2 sample, Dispensary of Hope patients averaged about 0.6 fewer comorbidities than the comparison group patients. As a result of these differences, we included the number of comorbidities as a control variable in our outcome regressions.

MONTHLY COSTS

We plotted monthly mean health system costs relative to the enrollment for HS1 and HS2 and their respective comparison groups (Figure 3). Both HS1 and HS2 patients experienced a peak during their enrollment month, indicating that a major event (hospitalization or ED visit) preceded enrollment. By construction, the comparison groups for both locations also contained a significant event during the quarter of pseudo-enrollment. Health system costs for HS1 Dispensary of Hope patients in the postenrollment period is consistently lower than comparison group costs, whereas costs for both groups at HS2 follow a similar trend during the study period.

FIGURE 3.

Monthly Costs by Hospital System and Treatment Group

UNADJUSTED AND REGRESSION-ADJUSTED RESULTS

Table 2 presents results for the 4 outcomes and for 3 different perspectives: HS1 and HS2 together and HS1 and HS2 separately. Unadjusted results are average outcomes for the preenrollment and postenrollment periods for the Dispensary of Hope group and the comparison group, respectively.

TABLE 2.

Changes in Outcomes per Patient per Year, by Dispensary of Hope Status

Outcome by HS	Unadjusted results (weighted)							Regression-adjusted results
	Dispensary of Hope			Comparison group			D-in-D (SE)
	Preenrollment mean (SD)	Postenrollment mean (SD)	Change (SE)	Preenrollment mean (SD)	Postenrollment mean (SD)	Change (SE)		D-in-D (SE)
HS costs
HS1 and HS2	$5,681 ($10,895)	$4,981 ($14,293)	−$700 ($249)	$4,473 ($11,845)	$4,282 ($15,569)	−$191 ($305)	−$509 ($494)	−$35 ($465)
HS1	$7,709 ($12,255)	$6,997 ($23,139)	−$712 ($790)	$8,854 ($19,826)	$9,538 ($28,633)	$684 ($978)	−$1,396 ($1,494)	−$3,161 ($1,251)b
HS2	$5,098 ($10,339)	$4,401 ($10,280)	−$697 ($226)	$3,007 ($6,868)	$2,522 ($6,068)	−$485 ($242)	−$212 (406)	$397 ($632)
Inpatient stays
HS1 and HS2	0.42 (0.70)	0.29 (0.74)	−0.13 (0.01)	0.26 (0.74)	0.21 (0.66)	−0.05 (0.02)	−0.08 (0.03)a	−0.06 (0.05)
HS1	0.57 (0.57)	0.40 (0.83)	−0.17 (0.03)	0.47 (0.97)	0.47 (1.01)	0.01 (0.04)	−0.18 (0.06)b	−0.20 (0.04)b
HS2	0.37 (0.73)	0.26 (0.71)	−0.11	0.19 (0.63)	0.13 (0.45)	−0.07	−0.05 (0.03)	0.02 (0.04)
ED visits
HS1 and HS2	1.28 (2.35)	1.20 (2.66)	−0.08 (0.04)	1.72 (4.19)	1.54 (4.99)	−0.18 (0.10)	−0.10 (0.20)	0.06 (0.06)
HS1	0.75 (1.34)	1.04 (2.56)	0.29 (0.08)	2.04 (5.85)	1.88 (5.52)	−0.16 (0.09)	0.45 (0.32)	0.32 (0.03)b
HS2	1.43 (2.55)	1.24 (2.69)	−0.19 (0.04)	1.61 (3.47)	1.43 (4.80)	−0.19 (0.13)	0.00 (0.25)	0.01 (0.20)
PCS ED visits
HS1 and HS2	0.09 (0.39)	0.11 (0.53)	0.01 (0.01)	0.29 (1.38)	0.25 (1.07)	−0.05 (0.03)	0.06 (0.06)	0.03 (0.01)b
HS1	0.11 (0.35)	0.17 (0.77)	0.06 (0.02)	0.31 (1.00)	0.27 (1.04)	−0.04 (0.03)	0.10 (0.06)a	0.06 (0.03)b
HS2	0.09 (0.40)	0.09 (0.43)	0.00 (0.01)	0.29 (1.49)	0.24 (1.08)	−0.05 (0.04)	0.05 (0.08)	0.02 (0.03)

Regression-adjusted results controlled for age, age squared, sex, race and ethnicity, pharmacy-location fixed effects, and the number of comorbidities. These

results are also weighted by propensity score weights as described in the methods.

^a Statistically significant at P < 0.10.

^b Statistically significant at P < 0.05.

D-in-D = difference-in-difference; ED = emergency department; HS = health system; PCS = primary care sensitive.

The unadjusted results reveal that, on average, health system costs decreased in the postenrollment period by about $700 for Dispensary of Hope patients at HS1 and HS2. More variation was present in the comparison group, in which the HS1 comparison group increased by $684 and the HS2 comparison group decreased by $485. Adjusted for patient characteristics, per-person annual health system costs at HS1 decreased by $3,161 (P < 0.05) more in the Dispensary of Hope group than the comparison group. This change represents a 27% reduction in health system costs for program participants at HS1 (adjusted preenrollment spending was $11,536 at HS1). Adjusted results for HS2 and the combined sample were not statistically significant.

Reductions in inpatient stays were a primary driver of health system costs reductions. On average, Dispensary of Hope patients in the combined sample decreased inpatient stays by 0.13 stays per year (or 130 inpatient stays per 1,000 program participants in the unadjusted results). Comparison group changes in the combined group were more modest: there was a reduction of 0.05 inpatient stays per year in the postenrollment period. Adjusted for patient characteristics and annualized, the number of inpatient stays per person at HS1 decreased by 0.20 (P < 0.05) fewer stays per year in the Dispensary of Hope group than in the comparison group. This change represents a reduction of 200 inpatient stays per 1,000 program participants annually at HS1. Adjusted results for HS2 and the combined sample were not statistically significant.

Although inpatient stays generally decreased during the study period, ED visits did not change significantly for Dispensary of Hope patients across the 2 health systems. However, when looking at HS1 alone, there was a notable increase in ED visits. Adjusted for patient characteristics, per-person number of annual ED visits at HS1 increased by 0.32 (P < 0.05) more visits in the Dispensary of Hope group than in the comparison group. PCS ED visits exhibited a small but statistically significant increase in the regression-adjusted results for the Dispensary of Hope patients relative to the comparison group (0.03 visits per year). Similar to ED visits overall, this increase in PCS ED visits was primarily driven by results from HS1 (an increase of 0.06 more visits per year). Supplementary analyses restricting the study sample to high users and consistent users were consistent with the full sample in both magnitude and statistical significance (results available in the Supplementary Table 1 ^{(136.5KB, pdf)} ). In HS1, we found statistically significant reductions in inpatient stays and increases in ED visits among high users and statistically significant reductions in health system costs among consistent users.

Discussion

Dispensary of Hope distributes donated medications to clinics and pharmacies around the country, which are used to fill prescriptions free of charge to people with lower incomes and no health insurance. Results of our analyses revealed substantial reductions in costs and inpatient stays for HS1 participants of the Dispensary of Hope program. Specifically, in HS1, hospital costs decreased by $3,161 per patient over a 12-month follow-up period, representing a 27% reduction in costs. Inpatient stays decreased by 200 stays per 1,000 patients per year. Although ED visits increased during the study period, this increase was offset by a reduction in inpatient stays, which resulted in a net reduction in costs. In HS2, hospital costs, inpatient stays, and ED visits among Dispensary of Hope participants showed changes, but they were smaller in magnitude and were not statistically significant.

A variety of factors could be contributing to the results differences we found between HS1 and HS2 participants of Dispensary of Hope, including specific features of the program unique to each health system and pharmacy. One of the differences that we observed between the 2 hospital systems was the patient profile. Specifically, HS1 patients who participated in Dispensary of Hope appeared to be less healthy than the HS2 patients who participated in the program. The average number of comorbidities among Dispensary of Hope participants was 3.0 at HS1 and 0.9 at HS2. These differences in the health profiles of patients in the 2 health systems could be the major driver of the varying results showing program impacts. The changes observed among HS1 patients were likely larger because those patients were sicker and thus had more room for improvement than the HS2 patients.

Our findings were robust to subgroup analyses including the consistent users and high users, which were consistent with the full sample in both magnitude and statistical significance.

The results we present in this study are similar to other literature on charitable medication access programs. Stickel and Kim (2021)¹⁸ conducted a prestudy/poststudy of patients enrolled in several charitable medication access programs similar to Dispensary of Hope. They found that acute care visits (hospitalizations and ED visits) decreased by 0.23 (P = 0.02) and inpatient care visits fell by 0.17 (P = 0.01) in the 90 days after enrollment and the change in ED visits (0.06, P = 0.34) was not significant. Our results are also consistent with the reductions in health system utilization and savings previously reported for the Dispensary of Hope program by the Advisory Board Company.²⁷ The authors of that analysis calculated reductions of about $2,300 (inflated to 2020 medical prices) or 20% in average costs and a 37% reduction in inpatient utilization for a sample of approximately 400 patients at a single pharmacy.

Further expansion of the Dispensary of Hope program may require support from other members of the pharmacy supply chain, including pharmacy benefit managers, distributors, and manufacturers. US hospitals and health systems already provide care for uninsured patients in their ED or hospital. By participating in the Dispensary of Hope program, they provide uninsured patients with consistent access to essential medications.

LIMITATIONS

Our analyses are subject to several limitations. First, we did not account for the frequency of use of the Dispensary of Hope program, only that a patient had started in the program; although, for HS1, we conducted sensitivity tests looking at Dispensary of Hope participants who had more than 12 fills in an 18-month period and found that they were similar in magnitude to the overall sample. Second, we cannot account for unobserved programs cooccurring with the Dispensary of Hope program that could contribute to changes in costs and utilization. Third, our analyses were limited by the lengths of the available preenrollment and postenrollment periods. It may take additional time to see the impact of free access to medication on health care costs and utilization, and we were unable to capture those here. Fourth, the available data covered only those encounters that occurred within the health systems; we did not have access to other encounters or care that these patients may have received in other settings or facilities during the study period. Fifth, uninsured patients may be less likely to have an inpatient visit. Our analyses may overstate the cost avoidance due to Dispensary of Hope because they are able to enroll the sickest uninsured patients unable to avoid a hospital stay. Finally, we did not have the data necessary to assess the impact of Dispensary of Hope on either adherence or clinical outcomes of patients, but rather we examined the impact on health care utilization and costs.

Conclusions

The results of this study indicate that patients who received medications through the Dispensary of Hope program experienced significant reductions in total health system costs and inpatient utilization compared with patients who did not participate in the program in HS1. Future research could consider Dispensary of Hope’s impact on clinical outcomes of patients. These findings can play an important role in providing a consistent source of medication for those without insurance.