A Pharmacy Liaison–Patient Navigation Intervention to Reduce Inpatient and Emergency Department Utilization Among Primary Care Patients in a Medicaid Accountable Care Organization

Key Points Question Is more frequent screening for health-related social needs and patient navigation embedded in pharmacy care associated with reduced hospital admissions and emergency department (ED) visits among primary care patients in a Medicaid accountable care organization when compared with usual pharmacy care? Findings In this nonrandomized controlled trial of 364 adults, patients in the enhanced pharmacy care group did not have a lower likelihood of any hospital admissions or ED visits vs the usual pharmacy care group over 12 months. Meaning These findings suggest that enhancing pharmacy services for patients with high levels of health care utilization does not lead to reduced health care utilization.


Introduction
Medicaid accountable care organizations (ACOs) assume responsibility for the health outcomes and costs of their members.Patients in low-resource communities served by these ACOs struggle with a number of health-related social needs, such as food insecurity, housing insecurity, and lack of transportation to medical appointments. 1,2Patients with unmet health-related social needs are at high risk for preventable health care utilization and high levels of medical expenditure 3 and may prioritize social needs over medication adherence or receipt of medical care, 4 which may lead to preventable utilization.
Prior studies 5 have tested screening-and-referral interventions for health-related social needs to reduce health care utilization, with mixed results.Patient navigation interventions also connect patients to social services and community-based organizations to address unmet social needs and have focused on patients with high levels of medical utilization and complex care needs. 6A recent trial based in an urban emergency department (ED) 7 found that patient navigation reduced ED visits and hospitalizations among Medicaid-insured patients.However, other studies 8,9 have had mixed results.We are unaware of prior trials that have embedded a health-related social needs screeningand-referral intervention or a patient navigator in a pharmacy care program to reduce health care utilization.By embedding these interventions in a pharmacy program, we sought to address the root causes of medication nonadherence that may drive preventable utilization.
In this nonrandomized controlled trial 10 involving patients who are members of a Medicaid ACO and receive primary care at a large urban safety-net hospital, our objective was to evaluate the effect of enhanced pharmacy care (a pharmacy technician with at least a high school degree serving as a patient navigator who screens for health-related social needs quarterly and provides navigation services to connect patients with community resources) in reducing inpatient hospital admissions and ED visits at 12 months.We hypothesized that the intervention would reduce health care utilization relative to usual pharmacy care.

Study Design
We conducted a parallel-group nonrandomized controlled trial with 1:1 allocation, whereby patients were assigned to 1 of 2 groups based on odd vs even medical record number.We did not randomize patients due to concerns it would disrupt patient flow.The study followed the Transparent Reporting of Evaluations With Nonrandomized Designs (TREND) reporting guideline.The study protocol found in Supplement 1 has been published elsewhere. 11The Boston Medical Center/Boston University Medical Campus Institutional Review Board approved the study protocol with a waiver of informed consent owing to the use of deidentified patient data.

Allocation Procedure
We assigned patients to study condition based on medical record number.The usual pharmacy care liaisons approached and recruited patients with even medical record numbers, while the enhanced pharmacy care liaisons approached and recruited patients with odd medical record numbers.
Intervention pharmacy liaisons did not provide details of their intervention during the recruiting process; patients were blinded to assigned study group.

Intervention and Control Conditions
The intervention has been described previously by Lasser et al, 11 and the trial protocol is found in Supplement 1.The usual pharmacy care group consisted of pharmacy liaisons focused on medication adherence.The enhanced pharmacy care group consisted of pharmacy liaisons dually trained as patient navigators who screened patients for health-related social needs quarterly, provided navigation services to connect patients with community resources, and carried a smaller caseload.
We anticipated the enhanced focus on social needs would help promote medication adherence that would in turn lead to reduced health care utilization.

Usual Pharmacy Care Group
The usual care group was standard of care in the practice, with some patients completing screening for health-related social needs during their primary care visit.While patients were initially screened for health-related social needs at every visit, due to patient screening fatigue, the clinic reduced the frequency of screening to every 6 months in June 2019. 12The screening tool included

Enhanced Pharmacy Care Group
Patients assigned to this study group received the aforementioned services.

Outcomes
The primary outcome was all-cause inpatient hospital admissions and all-cause ED visits (a combination of any visit [binary outcome] and the count of visits among those patients with any visit) in the 12 months following enrollment.Secondary outcomes included all-cause inpatient hospital admissions and all-cause ED visits as separate outcomes, all-cause 30-day inpatient hospital readmissions, 30-day ED revisits, and social needs outcomes (eg, connected to community-based organization, received services through referral to community-based organization, referred to hospital food pantry, or received utility shutoff protection letter).The data analyst (N.W.) was blinded to participant group assignment.

Covariates and Confounders
We collected demographic data and potential confounders such as medical comorbidity (Charlson Comorbidity Index score), 13 mental health and substance use diagnoses, and screening results.Race and ethnicity data were collected in the routine course of clinical care.While some participants may have identified their race or ethnicity, in some cases front desk staff may have provided the identification.Both methods used the same racial and ethnic categories for all patients treated at the hospital.Data are not available on the proportion of patients whose race and ethnicity were selfreported vs determined by front desk staff.Options were defined by the health system.We analyzed these data to capture unmeasured social factors (eg, structural racism, racial discrimination) and to

Statistical Analysis
Using a 2-sided type I error rate of 0.05, a sample size of 364 patients (182 in each group) would achieve 90% power to detect a 10-percentage point reduction in our primary outcome.We analyzed all outcomes according to the intention-to-treat principle. 14Due to a high number of patients with zero-count visits (136 of 364), we used a zero-inflated negative binomial regression, a 2-part model that simultaneously examined the odds of any visit and visit rates among patients who had any health care utilization.We controlled for potential confounders identified in bivariate analyses as well as variables of a priori clinical significance.We also analyzed the intervention effect according to whether participants received the minimum intervention dose and performed post hoc analyses examining the primary outcome among patients who had continuous ACO enrollment during the follow-up period.We compared secondary outcomes of social needs between the 2 groups and examined the adjusted intervention effect using log binomial regression with adjusted risk ratios (aRRs) and 95% CIs.All analyses were conducted using SAS, version 9.4 (SAS Institute Inc).

Results
Of Inpatient hospital admissions and ED visits decreased in both groups over the study period.In the usual care group, 129 (70.9%) had at least 1 inpatient hospital admission or ED visit compared with 118 (64.8%) in the 12-month follow up period.In the enhanced pharmacy care group, the corresponding numbers were 136 (74.7%) and 110 (60.4%).In analyses controlling for language, housing insecurity, referral to food pantry, visit to food pantry, posttraumatic stress disorder, and baseline hospital admissions and ED visits (in the year prior to enrollment), the enhanced pharmacy care group was not associated with the odds of having any hospital admissions or ED visits relative to the usual pharmacy care group (adjusted odds ratio, 0.62 [95% CI, 0.23-1.62];P = .32).Similarly, there was no between-group difference in the rates of visits among those with any visit in the follow-up period in the adjusted model (aRR, 0.93 [95% CI, 0.71-1.22];P = .62)(Table 2).Due to small numbers of patients with any hospital admission or ED visits, we did not run interaction terms between comorbid mental health and substance use diagnoses, language, and intervention group.
No significant associations were detected in analyses restricted to patients who had continuous ACO enrollment during the follow-up period (eTable 1 in Supplement 2).Similarly, outcomes did not differ according to receipt of the minimum intervention dose among members of the enhanced pharmacy care group (eTable 2 in Supplement 2).

Discussion
Medicaid ACO members receiving primary care at a large urban safety-net hospital assigned to receive care from pharmacy liaisons with training in motivational interviewing and patient navigation and who carried smaller caseloads had no difference in inpatient hospital admissions and ED visits at 12 months relative to patients who received usual care.Both groups experienced decreased health care utilization during the COVID-19 pandemic, which is consistent with prior studies. 15Our study adds to a growing literature finding that pharmacist interventions in primary care 16  The dually trained pharmacy liaisons in our study provided services similar to those used in other navigation programs, such as connecting patients to internal and community resources 8,18 to mitigate health-related social needs and coordinating medical care. 8,18,19As in prior studies, the navigators used motivational interviewing to promote behavior change. 17,20Aspects of the intervention differing from prior studies included enhanced training in screening for health-related social needs, partnership with a community-based organization, and incorporation of pharmacy liaison services.There are several reasons why we did not observe a difference in utilization outcomes at the end of the trial.First, patients in the usual care group also received a moderately intensive intervention.Like participants in the intervention group, they received monthly phone calls from a pharmacy liaison who could assist with barriers to medication adherence.Additionally, patients in the usual care study group were screened in a primary care practice for health-related social needs, and these patients were able to work with a clinic-based patient navigator to address such needs.Had our usual care study group been more reflective of usual care at other safety-net primary care practices, we might have observed an intervention effect.Second, only 57.7% of patients in the enhanced care intervention received the minimum intervention dose.With the exception of a face-to-face encounter for recruitment, the pharmacy liaison-patient navigators delivered the intervention entirely via telephone.We suspect that patients with unstable housing and limited telephone service were unable to engage with the intervention.
Interventions in health care settings are unlikely to improve health outcomes rooted in longstanding and intractable structural racism.A 2015 study 21 found the median net worth for White households in Greater Boston was a quarter million dollars, while for Black families, it was substantially lower.Given that the majority of study participants identified as members of a racial or ethnic group bearing a disproportionate burden of inequities in health care, it is unsurprising that a health system-based intervention did not improve outcomes.Implementation of policies to reduce income inequality would be more likely to improve health outcomes.

Limitations
This study had several limitations.First, it was not randomized.It is possible that there were unmeasured differences in the 2 study groups that we did not control for in our analysis.Considering that only 17.4% of eligible ACO patients participated in the trial, our findings might not be generalizable to all patients with high utilization and high cost from ACOs in safety-net institutions.
Our sensitivity analyses and post hoc subgroup analyses did not have enough sample size to achieve statistical power.Data on patient race and ethnicity were collected in the routine course of clinical care via a combination of self-report and determination by hospital staff and therefore may be inaccurate.Finally, usual care in our study included a pharmacy liaison and a clinic-based navigator.
Such staff may not be available in most primary care settings, thereby limiting generalizability of our findings.However, our intervention may be effective when implemented in clinics with fewer ancillary staff.

Conclusions
The findings of this nonrandomized controlled trial suggest that an intervention where Medicaid ACO members at a safety-net hospital received care from pharmacy liaison-patient navigators was not associated with a reduction in inpatient hospital admissions and ED visits relative to usual pharmacy care (pharmacy liaison and clinic-based navigator).Future studies might focus on preventing high levels of health care utilization and should identify which patients may benefit most from patient navigation to reduce health care utilization and in which settings such interventions should be delivered.
Enhanced pharmacy care liaisons had smaller caseloads (maximum of 91 patients), had no other responsibilities, and received training to serve as patient navigators (termed pharmacy liaison-patient navigators).In this role, they connected patients with community and hospital resources to address health-related social needs.A dedicated 0.50 full-time equivalent employee at a community-based organization helped connect the patients receiving enhanced pharmacy care to services.The pharmacy liaisonpatient navigators also received motivational interviewing training.A primary care physician with expertise in patient navigation (K.E.L.) and a social worker with expertise in motivational interviewing and in connecting patients with resources (J.B.-F.) met weekly with the liaisons to discuss challenging cases.The social worker also monitored pharmacy liaisons' motivational interviewing skills using role-play.The pharmacy liaison-patient navigators spent the first 30 to 45 days after patient enrollment in the program helping patients overcome health-related social needs beyond those identified as barriers to medication adherence.If the pharmacy liaison-patientnavigator documented during the intake assessment that the patient had received screening for health-related social needs in the past 3 months and rescreened the patient if they had not been screened in the past 3 months, we considered the patient to have received the minimum intervention dose.The pharmacy liaison-patient navigator used motivational interviewing skills to address domains in which a patient had positive screen results but did not indicate a desire to be connected to resources.The pharmacy liaison-patient navigator also provided patient education, assisted with scheduling appointments and reminders, and offered support to connect with internal and community resources to help patients mitigate health-related social needs.

JAMA Network Open | Diversity, Equity, and Inclusion Enhanced
Pharmacy Intervention Among Primary Care Patients in a Medicaid ACO Patients who qualified for the hospital's pharmacy program were eligible for the study.Qualification for the pharmacy program required that patients be aged 18 to 64 years, within the third to tenth percentile on an internal risk score (primarily comprised of inpatient and ED utilization over the last 12 months combined with indices of medical and social risk) among Medicaid ACO membership at the time of program enrollment, and attend a visit with a primary care clinician (nurse practitioner or physician).The pharmacy program constituted the usual pharmacy care group described below.

JAMA Network Open | Diversity, Equity, and Inclusion
Pharmacy liaisons could review the most recent health-related social needs screening on an ad hoc basis.If the patient requested resources, the pharmacy liaison could connect the patient to the clinic-based navigator.Pharmacy liaisons in the usual care group worked with a maximum caseload of 215 patients, as they managed additional patients who were not part of the trial.Pharmacy liaisons in the usual care group spent 24 hours per week on case management; the remaining time was spent on other responsibilities.
Downloaded From: https://jamanetwork.com/ on 09/25/2023 monthly over a 12-month period to confirm medication adherence and address new barriers to medication adherence and engagement in care.

Table 1 .
Baseline Characteristics of Participants by Treatment Group a was 50.1 (10.1) years.Two hundred sixteen participants were women (59.3%) and 148 were men (40.7%).Two hundred ninety-six participants (81.3%) were members of a racial or ethnic minority group (35 Hispanic [9.6%], 214 non-Hispanic Black [58.8%], and 47 other [Asian, American Indian or Alaska Native, Native Hawaiian or other Pacific Islander, other, or declined or not available; 12.9%]).More than one-half of patients had significant medical comorbidity, and nearly one-half carried a diagnosis of depression.The usual care and enhanced care groups differed in the 2097 potentially eligible patients, we assigned 364 (17.4%) to study condition.The Figure shows reasons for exclusion, assignment to study condition, and follow-up;Table 1 shows descriptive characteristics for the usual pharmacy care and enhanced pharmacy care groups.The mean (SD) age JAMA Network Open | Diversity, Equity, and Inclusion Enhanced Pharmacy Intervention Among Primary Care Patients in a Medicaid ACO JAMA Network Open.2023;6(1):e2250004.doi:10.1001/jamanetworkopen.2022.50004(Reprinted) January 9, 2023 5/11 Downloaded From: https://jamanetwork.com/ on 09/25/2023 b Unless otherwise indicated, data are expressed as No. (%) of participants.Percentages have been rounded and may not total 100.c Calculated via t test.d Collected in the routine course of clinical care through the patient portal, registration, and telephone encounters.e f Compared between English vs non-English language.g Indicates health-related social needs screening result on most recent screen prior to or on the date of enrollment.h Based on International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes on the electronic health record problem list and used for billing.JAMA Network Open | Diversity, Equity, and Inclusion Enhanced Pharmacy Intervention Among Primary Care Patients in a Medicaid ACO JAMA Network Open.2023;6(1):e2250004.doi:10.1001/jamanetworkopen.2022.50004(Reprinted) January 9, 2023 6/11 Downloaded From: https://jamanetwork.com/ on 09/25/2023 of all participants

JAMA Network Open | Diversity, Equity, and Inclusion
77d patient navigation-like interventions do not reduce utilization among safety-net patient populations.The most notable of the latter studies was the Health Care Hotspotting trial,6which found readmission rates were not lower among patients randomly assigned to the Camden Coalition's care-transition program than among those who received usual care.It is possible, however, that certain groups (eg, older patients9and those with limited English proficiency17) may benefit from patient navigation interventions.Patient navigation in the ED setting also shows promise.7

Table 2 .
Health Care Utilization at Baseline and 12 Months After Enrollment by Study Group the model fit is questionable in adjusted model due to a convergence problem.