Cost and Cost-Savings of Navigation Services to Avoid Re-hospitalization (NavSTAR) for a Comorbid Substance Use Disorder Population

Abstract

Background:

A randomized clinical trial found that patient navigation for hospital patients with comorbid substance use disorders (SUDs) reduced emergency department (ED) and inpatient hospital utilization compared to Treatment-As-Usual (TAU).

Objective:

To compare the cost and calculate any cost savings from the Navigation Services to Avoid Re-hospitalization (NavSTAR) intervention over TAU.

Research Design:

This study calculates activity-based costs from the healthcare providers and uses a net benefits approach to calculate the cost savings generated from NavSTAR. NavSTAR provided patient navigation focused on engagement in SUD treatment, starting prior to hospital discharge and continuing for up to three months post-discharge.

Subjects:

Adult hospitalized medical/surgical patients with comorbid SUD for opioids, cocaine, and/or alcohol

Cost Measures:

Cost of the three-month NavSTAR patient navigation intervention and the cost of all inpatient days and ED visits over a 12-month period.

Results of Base Case Analysis:

NavSTAR generated $17,780 per participant in cost-savings. Ninety-seven percent of bootstrapped samples generated positive cost-savings and our sensitivity analyses did not change our results.

Limitations:

Participants were recruited at one hospital in Baltimore, MD through the hospital’s addiction consultation service. Findings may not generalize to the broader population. Outpatient healthcare cost data was not available through administrative records.

Conclusion:

Our findings show that patient navigation interventions should be considered by payors and policy makers to reduce the high hospital costs associated with comorbid SUD patients.

INTRODUCTION

Substance use disorders (SUDs) are associated with substantial hospitalization costs. An estimated $15 billion was spent on hospitalizations related to opioid use or dependence in 2012 (1). Excessive drinking cost an estimated $5.9 billion in-hospital care in 2014 (2). Individuals with comorbid SUDs need both medical and behavioral interventions which are challenging to treat in a hospital setting (1) and they are more likely to experience inpatient readmissions (3, 4).

Reducing avoidable inpatient readmissions and associated costs is a critical need. One potential solution for patients with SUDs is patient navigation, an intervention that provides case management and encouragement to engage in recommended SUD treatment and medical care in the community following hospital discharge. Currently, there have been no randomized controlled trials to study the costs of patient navigation in reducing rehospitalizations in a population with comorbid SUD. Instead, studies have been conducted with general patient populations and patients with high ED utilization. Observational studies in general patient populations indicate patient navigation and similar case management interventions can reduce avoidable emergency department (ED) visits, hospitalizations, and costs among frequent users of ED services (5–9). Two randomized controlled studies have estimated the cost of patient navigation interventions compared to treatment as usual in reducing ED use and costs for general population patients with high ED utilization in the past year (10, 11). These two studies have mixed findings on reducing costs: Shumway et al. showed no difference in hospital and other healthcare costs for patients receiving patient navigation, while Seaberg et al. found that patient navigation reduces hospital inpatient and ED costs.

Recent years have seen a growth in hospital SUD consultation services, which often include initiating SUD treatment and providing referrals to continue care in the community (12, 13). Building on these efforts and to address re-hospitalizations for patients with SUDs, the present study team developed and implemented a patient navigation service, referred to as Navigation Services to Avoid Re-hospitalization (NavSTAR). NavSTAR is a patient navigation intervention designed to facilitate entry into SUD treatment and recommended medical care upon hospital discharge, support and motivate patients to engage in recommended services, and address various barriers to treatment access (14). Gryczynski et al., (15) found that the NavSTAR intervention reduced inpatient admissions and ED visits 12 month post-discharge. Participants in NavSTAR were more likely to enter SUD treatment within 3 months of discharge and reported similar quality of life 12 months post-discharge. This current study estimates the costs and cost savings of the NavSTAR intervention.

METHODS

Study Design

The NavSTAR trial included 400 adult hospital patients randomized on a 1:1 basis to either NavSTAR or Treatment-as-Usual (TAU). Participants were recruited during an index hospitalization at the University of Maryland Medical Center (UMMC) in Baltimore and were required to be over 18 years old and have a current DSM-5 SUD diagnosis for opioids, cocaine, or alcohol. Participants were excluded if they were currently enrolled in SUD treatment, lived outside of Baltimore City, were pregnant, had a terminal medical condition, or were hospitalized for a suicide attempt (14).

Participants in both arms received usual care from UMMC and the long-standing addiction consultation service (13), a multidisciplinary team that includes psychiatrists, nurses, addiction counselors, and social workers (15–17). During the study, this team conducted bedside assessments, counselling, motivational interviewing, and referral to SUD treatment (including medications if indicated) for all patients. Participants randomized to the NavSTAR arm were also connected to a masters-level social work patient navigator prior to discharge from the hospital and provided 3 months of patient navigation post-discharge. Nordeck et al. (14) provides a full description of the two study arms.

DATA AND ANALYSIS

We adapted the approach used by the Substance Abuse Services Cost Analysis Program (SASCAP) questionnaire to collect activity-level resource use and cost data on patient navigation (18). This approach calculates costs from the healthcare provider perspective, which for this study includes patient navigation services, hospital inpatient days, and ED visits; we included no other societal costs (19). In our study, providers are considered the study-funded patient navigation services and hospitals. Focusing on the healthcare provider perspective makes our cost estimates most relevant to possible providers for patient navigation, such as hospital administrators.

Patient Navigation Data and Costs

NavSTAR was designed to augment UMMC’s addiction services, and participants in both arms received the standard UMMC substance use consults. As our focus was on NavSTAR, we did not cost the UMMC services and focused instead on the incremental costs of the NavSTAR intervention. We collected data on patient navigator time through daily time logs, which were completed for 34 working days spread over the study intervention period. The daily time logs recorded time spent directly working with participants, as well as time spent traveling and doing administrative work. We then interviewed study staff to collect patient navigator wages, the amount of building space used in the intervention, and the cost of other support services provided. Separately, the study tracked the total number of patient navigation hours each patient received, and any participant support funds used to pay for cellphones and other minor expenses (e.g., transportation to medical appointments).

To calculate the total hours of patient navigation per participant, we combined the administrative time reported on the daily time logs with the number of patient navigation hours provided for each participant. We then multiplied by the patient navigator’s average hourly wage plus fringe benefits and taxes to calculate the labor costs of patient navigation. For non-labor costs, we estimated building costs and patient navigator travel costs. These costs were allocated proportionally to participants based on the number of hours of patient navigation received. Lastly, patient support funds were added to the total costs for each participant.

Healthcare Data and Costs

Hospital utilization data were obtained from medical records abstracted from the Chesapeake Regional Information System for our Patients (CRISP), Maryland’s statewide health information exchange (14). These data indicated the number of inpatient days, ED visits, and deaths for 12 months post-discharge from the index hospitalization. Medical records were obtained for all 400 participants randomized over the 365-day study period, and there are no missing inpatient and ED utilization data in our sample. We summed all inpatient days per participant that included admissions for medical, psychiatric, and surgical reasons and planned hospital stays. For ED utilization, we summed psychiatric and medical ED visits per participant. To calculate costs, we multiplied a unit cost by the number of inpatient days and ED visits for each participant. Unit costs for healthcare were found in the literature (20) and inflated to 2019 dollars using the general consumer inflation calculator (https://www.bls.gov/data/inflation_calculator.htm). To test for statistically significant differences, we used a two-sided t-test of means using a Type I error rate of .05.

Cost Savings Analysis

The net benefits of NavSTAR are the cost-savings associated with NavSTAR relative to TAU. To calculate cost-savings, we summed the cost of inpatient days and ED visit costs for participants in each arm, averaged across participants by arm, and then calculated the difference in average costs across the arms (21, 22). To model uncertainty around the cost savings, we used a bootstrap analysis to draw 10,000 random sample datasets with replacement (23). We then re-calculated cost savings for each bootstrapped sample. From these data, we calculated the percentage of draws with cost savings.

Sensitivity and Supplemental Analysis

We conducted sensitivity analyses by re-running our analysis with model-adjusted costs controlling for participant exposure, measured in days in the community post discharge from the index hospital and arm. These are the same independent variables used in the analyses described in Gryczynski et al. (15). To account for skewed cost data, we also used ordinary least squares (OLS) regression with logged-transformed costs to confirm results from the two-sided t-test (24). We also conducted a supplemental analysis that included participant self-report data on outpatient healthcare and SUD treatment utilization and costs 12-month post-discharge from the index hospitalization.

RESULTS

Table 1 lists intervention services and costs over the 3-month intervention period. For NavSTAR participants, patient navigators spent an average of 8.7 hours at an hourly labor cost of about $33. On average, labor costs were $286, and non-labor costs were $57 per participant. Next, Table 1 shows average inpatient days and ED visits reported in CRISP. On average, TAU participants had 15.2 inpatient days compared to 12.0 days for NavSTAR participants, which is approximately 80% of the time spent by TAU participants. For ED visits, TAU participants had 9.4 visits compared to 6.1 visits for NavSTAR participants, which is approximately 64% of the number of ED visits for TAU participants. Gryczynski et al. (15) found significant differences between NavSTAR and TAU on inpatient readmissions and ED visits using Cox regression and count model analyses.

Table 1.

Average intervention service use and costs and inpatient and ED use (USD 2019)

Group	n	Total hours of patient navigation per participant a	Labor cost per participant	Non-labor costs per participant c	Inpatient days per participant d, e	Emergency department (ED) visits per participant d, e
TAU	200	--	--	--	15.2	9.4
NavSTAR	200	8.7	$286	$57	12.0	6.1
Unit costs	—	$33b	—	—	$2,346	$3,189

^aIncludes time for direct patient navigation services, administrative work, and travel time.

^bHourly wage plus fringe benefits, tax, and overhead rates for patient navigators.

^cIncludes funds used for participant IDs and other fees, building space costs, and estimated patient navigator travel costs.

^dBased on hospital utilization data through 12-month follow-up or death. Source: Health Information Exchange, the Chesapeake Regional Information System for Our Patients.

^eInpatient days includes inpatient hospital stays resulting from medical, psychiatric, surgery, and planned inpatient admissions.

Total average costs are presented in Table 2. The average patient navigation cost for NavSTAR participants was $343. The average inpatient hospitalization cost for TAU participants was $35,597, while the average cost for NavSTAR participants was $28,174, with average difference of $7,423. The average ED visit costs for TAU participants were $30,041 compared to $19,343 for NavSTAR participants with average difference of $10,698. The average total cost for TAU participants was $65,639 compared to $47,829 for NavSTAR participants over the 12–month post-discharge period; this difference is statistically significant (p = .048). Including the cost of patient navigation, the costs for NavSTAR participants were still 27% lower than for TAU participants, resulting in cost savings of $17,780 per participant from reduced inpatient days and ED visits over 12 months. Figure 1 shows the plot of cost savings from 10,000 bootstrapped samples. Ninety seven percent of the bootstrapped samples produced cost savings for NavSTAR participants. Table 3 presents our sensitivity analyses using predicted cost data from both a generalized linear model (GLM) and OLS regression. In both cases, the NavSTAR arm generated cost savings per participant. Model adjusted costs from the GLM model slightly changed average total costs per patient while the OLS model’s average total costs per patient match the observed data when rounded to the nearest dollar. For both models and NavSTAR generates approximately the same cost savings. Finally, the OLS regression of the log-transformed costs showed a statistically significant difference (p < .001) in total costs between TAU and NavSTAR.

Table 2.

Average Costs and Cost Savings (USD 2019)

Group	n	Patient navigation cost per participant	Inpatient costs per participant	ED visit cost per participant	Total intervention, inpatient, and ED visit cost per participant	Cost Savings per participanta
TAU	200	$0	$ 35,597	$30,041	$65,639	--
NavSTAR	200	$343	$ 28,174	$19,343	$47,859*	$17,780

^*Statistically significant difference with p = .048 based on two-tailed t-test results.

^aCost savings are calculated as follows: $65,639-$47,859 = $17,780

Figure 1.

Distribution of Bootstrapped Cost Savings

Table 3:

Sensitivity Analysis: Cost Savings

Group	n	Observed intervention, inpatient, and ED visit costs per participant	Cost savings per participant^^
NavSTAR	200	$47,792	--
TAU	200	$65,639	$17,780
		GLM predicted intervention, inpatient, and ED visit costs participant	Cost savings per participant^^
NavSTAR	200	$48,099	--
TAU	200	$65,377	$17,278
		OLS intervention, inpatient, and ED visit costs per participant	Cost savings per participant^^
NavSTAR	200	$47,859	--
TAU	200	$65,639	$17,780
		Average intervention, outpatient, inpatient hospital costs per participant	Cost savings per participant^
NavSTAR	111	$56,528	--
TAU	107	$73,939	$18,578

^{^^}Bootstrapped analysis finds that 97% of draws show NavSTAR costs less per participant than TAU and creates cost savings.

^{^}Bootstrapped analysis finds that 93% of draws show NavSTAR costs less per participant than TAU and creates cost savings.

Table 3 also shows summary results from a supplemental analysis (see Supplement Digital Content) on a subset of patients, which included 107 participants in TAU and 110 participants in NavSTAR with self-report data on outpatient healthcare and SUD treatment costs. On this sub-sample, total costs, including hospitalization costs, were $73,939 per participant in TAU and $56,528 in NavSTAR. As above, NavSTAR generated substantial cost savings of $17,411 per participant and the inclusion of outpatient healthcare and SUD treatment costs did not change our conclusions.

DISCUSSION

Participants with comorbid SUDs incur substantial hospitalization costs and have high rates of inpatient readmission (1–4). No cost analysis of patient navigation to reduce hospital readmission with comorbid SUD has been previously conducted. Our study shows that NavSTAR has lower total costs per participant without reducing patients’ quality of life (15). NavSTAR generates a statistically significant average cost savings of over $17,000 per participant and generates cost savings in 97% of our bootstrapped samples.

Our analysis estimates average NavSTAR costs of $343 per participant for 8.7 hours of patient navigation. In the context of hospital costs, this additional expenditure is minimal and is an important finding. To better treat patients with SUD, hospitals have begun to incorporate inpatient addiction consultation services to manage co-morbid SUDs during hospitalization and to generate referrals to community-based SUD treatment following discharge (15–17). Our findings suggest that SUD treatment referrals and in-hospital addiction consultation services could be expanded to include patient navigation and create cost savings even if NavSTAR was more costly or provided by different staff than in our analysis. If NavSTAR costs were doubled or tripled, per patient costs would still be substantially less than the cost of a single inpatient day or ED visit. Alternatively, patient navigation services could be provided by lower cost staff, such as peer specialists; while this would reduce per participant costs, effectiveness of the intervention may be lower resulting in lower cost savings. But even a 50 percent drop in cost savings (e.g., $17,000 to $8,500) would still provide substantial savings.

These results have the potential for meaningful, real-world impacts. For example, under the Affordable Care Act, hospitals with higher-than-expected risk-adjusted readmissions are financially penalized (25). Specific to Maryland, higher readmissions and costs could push hospitals over their readmission targets and per capita budgets set by Maryland’s All-Payer Model (26). Instead, if hospitals provided funding for NavSTAR patient navigation services, our findings show the average cost savings for each dollar invested in NavSTAR intervention is $51 (average cost savings divided by average NavSTAR cost = $17,780/$343). Hospitals that adopted NavSTAR could reduce their readmission and the associated punitive penalty, save money, potentially bill for patient navigation services, and improve care for patients with comorbid SUD.

Our findings are also relevant to the ongoing opioid epidemic. Of study participants, 78% used opioids in the past 30 days at baseline (15). With the high percentage of OUD participants, the NavSTAR patient navigation intervention may be able to substantially reduce hospitalization costs specifically for OUD patients. Ronan and Herzig (1) reported that 520,275 patients with OUD incurred $15 billion in hospital costs in 2012. Applying the average cost per patient of NavSTAR patient navigation, it would cost over $178 million with an expected cost savings of approximately $4 billion based on our calculated 27% decrease in hospital costs between TAU and NavSTAR. In addition to decreasing hospital costs, Gryczynski et al., (15) also shows that NavSTAR was effective in increasing the likelihood of entering community SUD treatment after discharge (treatment entry within 3 months, 50.3% for NavSTAR compared to 35.3% for TAU).

Limitations

This study has several limitations. First, our results may not be generalizable to other patient populations as all patients were recruited from one site, a university hospital in Baltimore, MD, a city with a high opioid prevalence. The hospital has a well-established and staffed addictions consultation service that provides in-hospital treatment and community referrals to patients with comorbid SUDs, which may not be common in other hospitals. As noted in Gryczynski et al. (15), the fact that UMMC has a well-established addiction consultation service could have reduced the difference between the TAU and NavSTAR arms, implying that hospitals without an addiction consultation service could see larger benefits from NavSTAR. As for costs, NavSTAR patient navigators were able to operate in parallel with the existing addiction consultation service and adding stand-alone patient navigators in other hospitals may require more staffing and implementation costs than we captured here. Future studies should aim to replicate and extend our results across a diverse set of hospitals. Second, we collected self-report data on SUD treatment and outpatient care; however, our response rate at 12 months was only 55%, limiting our ability to fully include SUD treatment and outpatient care costs. Instead, we analyzed the sub-sample of patients with complete data and found that NavSTAR still generated cost savings. Finally, we included intervention and hospital costs only from the healthcare perspective and did not have data to calculate quality adjusted life years (QALYs). Future studies should look to include societal perspectives such as the impact on patient costs, employment, other social services costs and calculate QALYs (19).

Although additional study is needed to address these limitations and confirm our findings, we find that NavSTAR incurred low total costs compared TAU and that the amount of cost savings is substantial. Our findings show that patient navigation interventions should be considered by hospitals to reduce the inpatient and ED visits costs associated with comorbid SUD patients.