## Introduction

Effective pain management among hospitalized patients is associated with better health outcomes1 and increased patient satisfaction.2 Traditionally, pharmacologic therapies such as opioids form the cornerstone of pain management in the inpatient setting. However, while opioids are effective in reducing pain, they are also associated with side effects including sedation, dizziness, nausea, and constipation, among others. These adverse effects can prolong length of stay (LOS) in the hospital, increase healthcare costs, and decrease patient satisfaction.3,4

Non-pharmacologic therapies may contribute to the efficacy of an overall pain management strategy and provide alternatives to traditional opioid treatments.5,6,7 Recently, virtual reality (VR), a computer-generated simulation of a three-dimensional environment which can be explored and interacted with by the user, has emerged as a novel non-pharmacologic therapy for pain. There is an increasing body of evidence that demonstrates the effectiveness of VR on pain reduction in the outpatient setting.8,9,10,11,12 Aside from pain management, VR has also been tested in a variety of other disease states such as anxiety,13,14,15 obesity,16,17,18 oncology,8 and neurorehabilitation.19,20

Investigators, including those from our own group, have also examined the impact of VR on patients in the inpatient setting. In a feasibility study, we found that while few inpatients were both eligible and willing to use VR, those that used VR reported that it was a positive experience and that it improved their pain and anxiety.21 In a separate study, we found that 65% of hospitalized patients receiving a VR experience achieved a clinically significant pain response vs. 40% of controls watching a relaxation video (p = 0.01; number needed to treat = 4) without any adverse events reported.22 Given the effectiveness of VR therapy for pain management, VR as an adjunctive non-pharmacologic pain therapy program has potential to reduce opioid utilization.23 Other possible benefits of inpatient VR therapy include reduction in hospital LOS and increased patient satisfaction.24,25 While the use of VR in the hospital is promising, no study to our knowledge has yet examined the cost and effectiveness thresholds required for an inpatient VR program to be cost-saving.

To address this gap in knowledge, we sought to estimate the projected cost savings and budget impact of implementing a VR pain management program for hospitalized patients. We performed health economic decision analyses incorporating costs related to VR implementation, potential savings from decreases in opioid utilization and hospital LOS, and effects on Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) scores and resulting adjustments in Centers for Medicare & Medicaid Services (CMS) Hospital Value-Based Purchasing (VBP) payments. We then performed sensitivity analyses to create a return on investment (ROI) lookup table for hospitals of varying size and staffing costs that are considering implementation of an inpatient VR program. Our objective with this hypothesis-generating analysis was to create a framework for future health economic analyses of inpatient VR therapy and to determine the cost and effectiveness thresholds at which point inpatient VR therapy for pain management becomes cost-saving.

## Results

### Base-case results

#### Reduced length of stay

We sought to estimate the potential savings from a decreased hospital LOS as a result of VR therapy. Given prior research demonstrating that the majority of costs associated with a hospital admission are fixed and incurred early in the admission, we focused on end-of-stay variable costs to estimate cost savings associated with a reduction in LOS.51,52,53,54 Taheri et al. in 2000 estimated the variable costs of the final day of a general hospitalization to be $42051 (adjusted to 2016 dollars,$584). Any potential cost-savings from a VR-related LOS decrease were based on this figure. For the base-case, patients who completed VR and did not experience side effects were assumed to, on average, have a 20% reduction in costs for the final day of hospitalization; we varied this estimate over a wide range in sensitivity analysis. Further, we assumed that patients who completed VR therapy but had a minor side effect achieved no reduction in costs for the final day of hospitalization; this estimate was also varied in sensitivity analysis. We assumed that patients who did not experience VR had no change in the marginal costs associated with the last day of hospitalization.

#### HCAHPS and hospital VBP reimbursements

Patient satisfaction as measured by the HCAHPS surveys impacts CMS Hospital VBP reimbursements. We therefore analyzed the possible effects of offering VR therapy for inpatients on overall HCAHPS scores and reimbursement.

We utilized the CMS VBP calculation to estimate the effects of varying the pain control and overall hospital rating dimension scores from 0 to 10 on Medicare reimbursements for FY 2016.32,33,34,36,37 We used a CMS estimate of the average hospital base operating diagnosis-related group (DRG) payment for FY 2016 of $28,162,06634 and held all other variables in the VBP calculation constant at the national averages. Varying the pain and overall dimension scores from 0 to 10 correlated to a minimum CMS incentive payment of$4.39 per patient and a maximum of $11.70 per patient (maximum difference of$7.31 per patient). The maximum difference from the average national status quo scores (i.e., scoring a 10 for both pain and overall dimensions versus the current national averages of 2.35 and 2.62, respectively) was \$5.50 per patient.

We included the HCAHPS calculation in the decision tree by incorporating each patient’s probability of selecting the best possible answer for the questions related to the pain and overall dimensions. Each tree branch was associated with an average probability of selecting the best possible answer, which translated to a dimension score and resultant changes in hospital reimbursement. Supplementary Table 6 and Supplementary Table 7 demonstrate the lookup table used to determine the change in reimbursement per patient resulting from the various probabilities of selecting the best answer for the questions on the HCAHPS survey.

### Sensitivity analyses

Because some of the probabilities included in our analysis are supported by only limited data and/or may not apply to all hospitals, we performed extensive sensitivity analyses for all cost and probability estimates. We first tested the influence of all variables on the model by performing multiple univariate sensitivity analysis (i.e., tornado analysis). Based on the results of the tornado analysis, we subsequently completed one-way sensitivity analyses on the most influential variables to identify thresholds where VR became cost-saving. For all sensitivity analyses, the model accounted for additional costs associated with purchasing more VR licenses and hiring additional staff as the number of patients using VR therapy increased (Supplementary Table 8).

We also conducted Monte Carlo probabilistic sensitivity analysis using 1000 simulations and assumed that all variables followed a triangular distribution with base-case, minimum, and maximum values listed in Table 3.

### Data availability

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.