Hospital patients with severe wounds: early evidence on the impact of Medicare payment changes on treatment patterns and outcomes

Abstract

Aims: This study examines the effects of recent changes in Medicare long-term care hospital (LTCH) payments on treatment patterns and outcomes for severe wound patients discharged from short-term acute care hospitals (STACHs).

Materials and methods: The rolling implementation of a new Medicare payment policy was used to develop a difference-in-difference model. The study population consisted of Medicare beneficiaries subjected to the payment policy changes and hospitalized for stage 3, 4, or unstageable wounds; non-healing surgical wounds; and fistula. Using 2015-Q1-2017 Medicare claims data, changes in outcomes were examined for severe wound patients exposed to the new policy (treatment) and those that were not (comparison). All outcomes were modeled using linear regressions and adjusted for patient clinical characteristics. Analysis was conducted in a full sample and a sample with high-LTCH-use propensity.

Results: Severe wound patients exposed to the new policy experienced 4.1 and 7.5 percentage point (pp) reductions in LTCH use relative to the comparison group in the full sample and high-LTCH-propensity sample, respectively (p < .01 and p = .039). No statistically significant change was found in 60-day mortality or Medicare spending after the policy change in the treatment group as compared to the comparison group (p > .10). However, among severe wound patients who are exposed to the new policy in the high-LTCH-propensity sample, readmission and post-discharge sepsis rates increased after the policy change relative to the comparison group (readmission rate = 8.1 pp, p = .075; sepsis rate = 7.0 pp, p = .033).

Limitations: The findings are based on data from a limited timeframe around the policy change and, thus, provide only early evidence on the effects of the new policy.

Conclusion: The new LTCH payment policy is associated with no changes in Medicare spending and mortality, but higher readmissions and post-discharge sepsis rates among severe wound patients with a high likelihood to use an LTCH.

Keywords:

• Medicare
• long-term care hospital
• wounds
• mortality
• spending
• hospital readmission

JEL classification codes:

• I11
• I18

Introduction

Wound patients constitute an important patient population that requires prolonged and often expensive treatment. Population prevalence rates for chronic non-healing wounds are estimated to be ∼1–2%, with estimates of cost ranging from $25 billion to upwards of $50 billion annually^1,2. According to a recent study, within the Medicare population, ∼14.5% of beneficiaries were diagnosed with non-healing chronic wounds in 2014, with Medicare spending of ∼$31.7 billion³. Although treatment of individuals with chronic wounds account for significant healthcare resources, non-healing wounds have not received much attention in healthcare policy discussions, partially because they often appear as a comorbidity, whose underlying diseases may span multiple medical specialties. Sometimes referred to as the “silent epidemic”, the incidence of chronic wounds and the economic burden of treating them are expected to increase in the future with the aging of the population and rising rates of diabetes and obesity¹.

Patients with severe wounds, defined as those diagnosed with stage 3 or 4 wounds, non-healing surgical wounds, or fistulas, represent the most acute segment of the wound population⁴. An analysis of Medicare fee-for-service (FFS) patients in seven states in 2014 showed that patients hospitalized with severe wounds were discharged to an inpatient post-acute care (PAC) setting at almost twice the rate as all Medicare FFS hospitalized patients (54% vs 29%)⁵. Average length of stay for hospitalized severe wound patients in Medicare FFS was also more than twice the average length of stay of all Medicare FFS hospitalizations (10.9 days vs 5.2 days).

Long-term acute care hospitals (LTCHs), which treat chronically critically ill and medically complex patients who require hospital-level care for extended periods⁶, are a particularly important care setting for severe wound patients. Medicare FFS patients hospitalized with severe wounds in 2015 were 6-times more likely to be discharged to an LTCH compared to all Medicare FFS discharges (7.1% vs 1.2%)⁷. Wound patients treated in LTCHs often suffer from severe wounds that are difficult to heal, have complex underlying medical conditions, and require extensive treatment. A 2012 report found that LTCHs were 2–3-times more likely to care for cases requiring complex wound management than step-down units of short-term acute care hospitals (STACHs) and 5–10-times more likely than skilled nursing facilities (SNFs)⁸.

Recent changes to the LTCH payment system in Medicare, however, have reduced payments for patients who do not meet certain criteria. In this study, we investigate the effects of these changes in Medicare payments to LTCHs on discharge patterns and outcomes for severe wound patients.

Background: Medicare payment policy for severe wound cases cared for in LTCHs

Between Fiscal Years 2002 and 2016, payments for all Medicare fee-for-service cases treated in an LTCH were determined by the LTCH Prospective Payment System (PPS), which paid a fixed amount per discharge based largely on a patient’s diagnosis. In Fiscal Year 2016, Medicare started paying LTCHs based on a dual payment system. As established in the SGR Reform Act of 2013, LTCHs are paid based on the LTCH PPS for cases that spent at least 3 days in an intensive care unit in a STACH stay immediately preceding the LTCH stay and for cases that receive mechanical ventilation for at least 96 hours in the LTCH⁶. For the remaining cases, called “site-neutral” cases, Medicare will pay LTCHs the lower of cost or an amount that is comparable to what STACHs receive for similar cases, once the policy is fully implemented. During the initial 4-year phase-in period for site-neutral cases, LTCHs receive a blended payment, which is an average of the full site-neutral payment and the LTCH PPS payment. The dual payment system was introduced to each LTCH on a rolling basis, starting with discharges in an LTCH’s cost reporting period beginning on or after October 1, 2015. Based on our analysis of Medicare claims data, 54% of severe wound patients treated in LTCHs in Fiscal Year 2015 would not have met criteria for full LTCH PPS payment, and LTCHs treating these patients are at risk of receiving 20–40% less than the full LTCH PPS payment during the phase-in period. These site neutral severe wound cases made up 14.0% of total LTCH discharges and 16.1% of LTCH’s total covered days in Fiscal Year 2015. According to analyses conducted by the American Hospital Association, the average blended payment during the phase-in period covers only 79% of the cost of care in LTCHs⁹.

The change in LTCH payments has the potential to reduce access to LTCH care for site neutral severe wound patients. Lower payments for site neutral severe wound patients would reduce the financial incentive for LTCHs to accept these patients and incentivize them to increase their population of cases qualifying for the full LTCH payment. In that case, we would expect to observe a reduction in utilization of LTCH care among site neutral severe wound cases. If there are other care pathways for these site neutral severe wound patients that result in similar quality of care, we may not see a change in the outcomes of these patients as a result of the change in LTCH payment policy. Alternative post-acute care providers include skilled nursing facilities, inpatient rehabilitation facilities, and home health. Skilled nursing facilities are the most frequent discharge destination for Medicare FFS beneficiaries who are discharged to a post-acute care setting after an acute care hospital stay. They provide skilled nursing and rehabilitation care for patients who need inpatient post-acute care. Inpatient rehabilitation facilities provide intensive inpatient rehabilitation services, such as physical, occupational, or speech therapy patients who can tolerate at least from 3 hours of therapy per day. Home health agencies provide skilled care for patients who have difficulty leaving their homes. These other post-acute care settings, however, may not be adequately equipped to provide the level of care required by site neutral severe wound patients who would be candidates for LTCH care. If so, the change in LTCH payments could result in worsened outcomes for site neutral severe wound patients. In this study, we examine these hypotheses by asking the following two questions: (1) did the introduction of the site-neutral payments for LTCHs lead to a reduction in LTCH use among site neutral severe wound patients? (2) Did it lead to a change in patient outcomes and Medicare spending?

Methods

Overview of study design

LTCHs transitioned to the dual payment system based on the start of their cost reporting period, which can vary across providers. In order to use a consistent set of clinical coding based on ICD-10 to identify severe wound patients, we started our study period in October 2015, when ICD-10 codes went into effect. Reserving October–December 2015 as a period prior to transition, we focused on 162 LTCHs that transitioned to the new payment system between January 2016 and June 2016. These LTCHs constitute 38% of all LTCHs (Supplementary Appendix, Table A1). We used the staggered implementation of the new patient criteria across LTCHs to develop a difference-in-difference model. Specifically, we constructed a treatment group of site neutral severe wound cases at STACHs that were “exposed” (as described below) to the new LTCH patient criteria between January 2016 and June 2016 and a comparison group of site neutral severe wound cases at STACHs that had not yet been exposed to the new criteria during this period. We examined the risk-adjusted changes in LTCH use and outcomes in the treatment and comparison groups over the period in which the treatment group transitioned to the new LTCH criteria. We attributed the difference between the treatment and comparison groups in terms of the change they experienced in LTCH use and outcomes between the pre-criteria and post-criteria periods to the new patient criteria.

As specialty hospitals designed to treat chronically critically ill and medically complex patients, LTCHs treat only a small and select portion (7.1% in 2015) of all Medicare beneficiaries with severe wounds and discharged from STACHs. The low probability of LTCH transfer among severe wound patients introduces variability in our estimates. To focus on cases that are most likely to be appropriate for LTCH care and reduce noise in our estimates, we construct a sub-set of “high LTCH propensity” patients identified as those with relatively high predicted probability of being treated in an LTCH. In addition to estimating the effects of the new criteria on outcomes in the overall sample of site neutral severe wound patients, we also examined these effects for the high-LTCH-propensity sub-sample of severe wound patients.

Data and study population

We used 2015-Q1-2017 Medicare Inpatient, Skilled Nursing Facility, and Home Health Standard Analytic Files and associated Denominator Files to conduct our analysis. Using these files, we obtained 100% Medicare claims for services provided in STACHs, LTCHs, inpatient rehabilitation facilities (IRFs), skilled nursing facilities (SNFs), and home health agencies (HHAs). The Medicare claims data used in the analysis are obtained as limited data sets (LDS) under a data use agreement (DUA) and, therefore, do not require authorization from study subjects or documentation of a waiver according to the Privacy Rule of Health Insurance Portability and Accountability Act (HIPAA).

The study population consisted of Medicare beneficiaries who were discharged from a STACH between October 1, 2015 and July 31, 2016 and treated for severe wound in the STACH. Severe wound diagnoses were based on CMS’ clinical criteria, as used to implement a provision in the 21st Century Cures Act, and included stage 3 wounds, stage 4 wounds, unstageable wounds, non-healing surgical wounds, and fistula identified using principal and secondary diagnoses in a STACH⁴. To focus on severe wound cases that would be considered site-neutral were they to be treated in an LTCH, we excluded patients with 3 or more days in an ICU during their STACH stay. We calculated ICU days by summing Revenue Center Unit Count associated with Revenue Centers 0200–0219 for each hospital stay. In reporting days, hospitals do not count the day of discharge or death unless the admission and discharge day are the same day. We only use the ICU day criterion to determine site neutral status of severe wound cases - and not the criterion of being on a ventilator in an LTCH—because we do not know whether cases that are not discharged to an LTCH would have been on a ventilator had they been discharged to an LTCH. However, this is not a significant limitation for our study as ventilator use among severe wound cases with fewer than 3 ICU days is infrequent. For example, only 3.2% of severe wound cases with fewer than 3 ICU days in STACH and discharged to an LTCH in our full study sample are on a ventilator in an LTCH.

Our investigation centers on the direct effects of the LTCH payment policy change, i.e. the effect on the discharge behavior of STACHs that had prior experience sending severe wound patients to LTCHs. Therefore, we limited the study population to patients discharged from a STACH that transferred at least three severe wound cases to LTCHs in Fiscal Year 2015. CMS’ severe wound definition is only available based on ICD-10 diagnosis codes. In order to identify severe wound cases in STACHs in Fiscal Year 2015, which was the last year of ICD-9, we developed a list of ICD-9 codes for severe wounds by translating ICD-10 codes in CMS’ definition into ICD-9 codes using General Equivalence Mapping. We present the list of ICD-9 codes and the results of our validation exercise in the Supplementary Appendix.

In addition, we limited the study population to cases treated in STACHs that had a majority of their LTCH severe wound transfers going to a single LTCH (“primary LTCH”). We applied this restriction to identify severe wound cases in a STACH that were “exposed” to the new patient criteria. We associated a new-criteria transition date to each STACH based on the criteria transition date of its primary LTCH. Among the 1,415 STACHs that transferred at least three severe wound cases to LTCHs in FY 2015, 1,144 transferred a majority to a single LTCH. Finally, we excluded beneficiaries from the study population if they were not continuously enrolled in Medicare FFS between LTCH admission and 180 days after LTCH discharge, were discharged against medical advice, or admitted to a STACH for medical treatment of cancer, primary psychiatric disease, or rehabilitation care. In examining readmissions and transfers to PAC settings, only cases that were discharged alive were included in the analysis. We present our study cohort construction process in a figure included in the Supplementary Appendix.

In constructing the high-LTCH-propensity sub-population, we estimated a linear probability model of being transferred from STACH to LTCH using the number of days in an ICU at the STACH stay, indicators for wound type (stage 3 wound, stage 4 wound, unstageable wound, non-healing surgical wound, or fistula), and APR-DRG combined with Severity of Illness as control variables. We ran the model of being transferred from STACH to LTCH only on cases in our pre-period, but calculated a predicted likelihood of being treated in an LTCH for pre- and post-period cases, based on model estimates. Cases in the top 25th percentile in terms of predicted likelihood of being treated in an LTCH were included in the high-LTCH-propensity sub-population. This exclusion retained 24.8% of severe wound cases in the overall sample and 55.8% of severe wound cases discharged to LTCHs. We also conducted sensitivity analyses using sub-populations defined by top 20th and 30th percentiles in terms of predicted likelihood to be treated in an LTCH.

Treatment and comparison groups

We matched STACHs to a policy transition date based on the cost reporting period of the LTCH that received the majority of the STACH’s LTCH discharges among severe wound cases in Fiscal Year 2015. Therefore, implementation of the treatment is based on the date in which the STACH’s primary LTCH transitioned to the new patient criteria. We formed six cohorts based on policy transition month. The treatment group for cohorts 1–6 was formed from severe wound cases discharged from STACHs with policy transition month January 2016 to June 2016, respectively (Supplementary Appendix, Table A2). The comparison group for cohorts 1–6 was formed from severe wound cases discharged from STACHs with policy transition month April 2016 to September 2016, respectively. Thus, for each cohort, comparison group cases came from STACHs with policy transition 3 months after the treatment group STACHs’ policy transition. This study design is similar to the stepped-wedge design used in clinical trials where randomization at the individual level is not feasible.

For each cohort, we examined cases in a pre-criteria period and a post-criteria period. The pre-criteria period covered the second and third months prior to the cohort’s treatment group policy transition. The post-criteria period covered the month of the cohort’s treatment group policy transition and the following month. For example, for cohort 1, we looked at severe wound discharges from October 1, 2015 to November 30, 2015 (pre-period) and January 1, 2016 to February 29, 2016 (post-period), among STACHs with policy transition in January 2016 (treatment) and also among STACHs with policy transition in April 2016 (comparison). This excludes the month immediately prior to treatment hospitals’ implementation of the new criteria because admissions to LTCH in the prior month could be discharged in the first month of implementation and, thus, be subject to the new criteria. For each treatment and associated comparison cohort, we then examined changes in post-acute care (PAC) utilization patterns and outcomes between discharges that occurred in the pre-criteria period and discharges that occurred in the post-criteria period. The analysis was conducted at the patient discharge level.

Outcomes

To investigate changes in PAC utilization patterns, we examined discharge rates to each PAC setting as well as the percentage of All Patient Refined Diagnosis Related Group (APR-DRG) (3M) severity of illness 4 patients among those discharged to each PAC setting. APR-DRG severity of illness categories range from 1–4, with 4 indicating highest severity. We assessed mortality and Medicare spending over 60-day episodes starting with discharge from STACH. The readmission rate is constructed following CMS’ hospital-wide readmission measure and excludes planned readmissions. We also examined the sepsis rate during the initial STACH stay and the 60-day post-discharge period. Medicare spending analyzed in our study includes institutional payments from the Medicare trust to STACHs, LTCHs, IRFs, SNFs, other inpatient settings, and HHAs.

Statistical analysis

All outcomes are modeled using linear regression, which are estimated at the discharge level. The key explanatory variable is an indicator variable that takes on the value 1 if the STACH discharge occurred in the post-criteria period in a treatment group and 0 otherwise. We controlled for age, sex, race indicators, type of wound (i.e. stage 3, stage 4, unstageable wound, non-healing surgical wound, and fistula), number of days in the ICU (0, 1, or 2 days), interaction between APR-DRG and risk of mortality in estimating mortality, and interaction between APR-DRG and severity of illness in all other regressions. All clinical variables are measured based on STACH claim. The regressions also included indicator variables for month-cohort and STACH-cohort fixed effects. Standard errors are clustered at the STACH level. Our regression model is presented in Equation (1)(1) $$Y_{\mathit{\text{ijgt}}}={\gamma }_{jg}+{\theta }_{tg}+\delta T_{jgt}+{\beta X_{igt}+\epsilon }_{igt}$$(1) , where $Y_{\mathit{\text{ijgt}}}$ is outcome for discharge i, in STACH j, in cohort g, and time period t. ${\gamma }_{jg}$ is the STACH-cohort fixed effects, ${\theta }_{tg}$ is the time-cohort fixed effect, $T_{jgt}$ is an indicator variable that takes the value 1 if new policy is in effect for STACH j, in cohort g, at time t, and $X_{\mathit{\text{ijgt}}}$ includes discharge-level control variables. Cohorts are defined by when the new criteria started and include treatment and comparison observations. The time variables are the calendar month-year indicators: (1) $$Y_{\mathit{\text{ijgt}}}={\gamma }_{jg}+{\theta }_{tg}+\delta T_{jgt}+{\beta X_{igt}+\epsilon }_{igt}$$(1)

Results

The full study population consisted of 21,929 cases, of which 7,595 were in the treatment group. The high-LTCH-propensity sample consisted of 5,449 cases, including 1,923 in the treatment group. During the pre-criteria period, the discharge rate of severe wound cases to an LTCH was 7.5% in the full sample and 18.2% in the high-LTCH-propensity sample (Tables 1 and 2). In both populations, SNFs and HHAs were the most commonly used PAC settings for severe wound patients, with 57.7% of the overall sample and 56.3% of the high-LTCH-propensity sample discharged to a SNF or HHA. As expected, severe wound patients with the highest probability of being transferred to an LTCH are more acute than the full severe wound sample. Specifically, cases in the high-LTCH-propensity sample have a higher average length of stay in the STACH (9.1 vs 7.6 days) and higher likelihood of having severity of illness 4 (37.0% vs 23.3%), as compared to the full sample of severe wound patients (Tables 1 and 2).

Table 1. Estimation results for full study population (n = 21,929).

Outcomes	Pre-criteria level	Differential change after new criteria	p-value
Discharge destination after STACH stay
LTCH	0.0746	−0.0414	.000
Other STACH	0.0211	−0.0018	.742
IRF	0.0271	0.0081	.347
SNF	0.3510	−0.0003	.988
HHA	0.2257	0.0556	.004
Hospice	0.0590	0.0010	.915
Home	0.2018	−0.0254	.173
Rate of APR-DRG Severity of Illness 4
At LTCH	0.3173	0.0075	.951
At IRF	0.1434	−0.0915	.586
At SNF	0.2429	0.0828	.010
At HHA	0.1446	−0.0598	.154
At Hospice	0.4918	0.0013	.995
At Home	0.0823	−0.0274	.349
Length of Stay STACH	7.5605	−0.1505	.451
60-day mortality	0.2218	−0.0178	.303
60-day readmission rate	0.2982	0.0143	.469
Sepsis rate in STACH	0.2104	0.0080	.365
Sepsis rate after discharge from STACH	0.1177	−0.0147	.319
Medicare spending	30,619	−1,418	.186

Authors’ analysis of 2015-Q1-2017 Medicare claims data.

Table 2. Estimation results for High LTCH propensity population (m = 5,449).

Outcomes	Pre-criteria level	Differential change after new criteria	p-value
Discharge destination after STACH stay
LTCH	0.1821	−0.0748	.039
Other STACH	0.0162	0.0016	.872
IRF	0.0291	0.0057	.681
SNF	0.3796	0.0167	.744
HHA	0.1833	0.0459	.263
Hospice	0.0708	−0.0012	.963
Home	0.0996	−0.0050	.881
Rate of APR-DRG severity of illness 4
At LTCH	0.3644	−0.1367	.477
At IRF	0.2500	−1.0463	.746
At SNF	0.3294	0.1061	.185
At HHA	0.3002	0.0792	.586
At Hospice	0.5393	0.2023	.743
At Home	0.2236	−0.2682	.564
Length of Stay STACH	9.0814	−0.8527	.193
60-day mortality	0.2878	−0.0028	.945
60-day readmission rate	0.3187	0.0806	.075
Sepsis rate in STACH	0.3559	−0.0260	.295
Sepsis rate after discharge from STACH	0.1751	0.0700	.033
Medicare spending	40,025	−1,302	.644

Authors’ analysis of 2015-Q1-2017 Medicare claims data.

Our examination of discharge patterns revealed that LTCH transfer rate decreased by 4.1 percentage points among severe wound patients exposed to the new LTCH payment policy relative to the comparison group (Table 1). This change constituted a 55% reduction (4.1/7.5 = 55%) in LTCH transfer rate in the full sample. Conversely, discharge rate to HHA increased by 5.6 percentage points relative to the comparison group. Severe wound patients who are most likely to be treated in an LTCH and exposed to the new LTCH payment policy experienced a 7.5 percentage point reduction in LTCH use, relative to the comparison group (Table 2). This change was equivalent to a 41% reduction (7.5/18.2 = 41%) in LTCH transfer rate in the high-LTCH-propensity sample. Changes in discharge rates to other PAC settings were not statistically significant in both study populations.

The change in LTCH use was accompanied by a change in the severity of patients discharged to SNFs in the full population. Specifically, we found that the rate of severity of illness 4 increased among severe wound patients discharged to SNFs in the post-criteria period compared to severe wound patients who were not exposed to the new LTCH policy (Table 1). We did not find a statistically significant change in patient severity among high-LTCH-propensity severe wound patients discharged to PAC settings (Table 2). We also examined changes in average length of stay in the STACH during the post-criteria period and found no statistically significant change in average length of stay in STACH in either the overall or the high-LTCH-propensity samples as compared to the comparison group. Similarly, we did not find a statistically significant change in average length of stay in LTCH in either of the samples after the implementation of the new criteria (not shown).

Our analysis of patient outcomes revealed varying results. We found no statistically significant effect of the new LTCH payment policy on 60-day mortality of severe wound patients in either the full or high-LTCH-propensity samples. Similarly, there was no statistically significant change in 60-day readmission rate among severe wound patients exposed to the new LTCH criteria relative to the comparison group in the full sample. However, among severe wound patients who are most likely to be treated in an LTCH, the 60-day readmission rate increased by 8.1 percentage points in the post-criteria period in the treatment group relative to the comparison group. We also found that severe wound patients exposed to the new LTCH criteria in the high-LTCH-propensity population experienced a 7 percentage point increase in post-discharge sepsis rate as compared to the comparison group. Differential changes (treatment vs comparison) in post-discharge sepsis rate for the full sample and sepsis rate in the initial STACH stay in either sample were statistically insignificant. Finally, there was no statistically significant change in 60-day Medicare spending for either treatment group in the post-criteria period relative to the comparison groups.

Some of our findings were robust to changing the definition of the high-LTCH-propensity sample, while others showed some variability (Supplementary Appendix Table A3). The estimated effect of the policy change on Medicare spending and mortality continued to be statistically insignificant in alternative high-LTCH-propensity samples based on top 20th and 30th percentiles of predicted LTCH transfer propensity. The effects on 60-day readmission rate and post-discharge sepsis rate were positive and statistically significant in the top 20th percentile sample, while they were statistically insignificant in the top 30th percentile sample.

Discussion

In this study, we used the rolling implementation of new Medicare patient criteria for LTCHs to estimate the effects of the policy on severe wound patients’ access to LTCHs and outcomes. We found that the new policy, which significantly reduced payments to LTCHs for severe wound patients that did not meet criteria, resulted in a significant reduction in admissions to LTCHs among severe wound cases. This reduced utilization was observed in the full sample and the high-LTCH-propensity sample. Surprisingly, we did not find evidence that the reduced utilization of LTCHs resulted in longer stays at STACHs. Such a finding is inconsistent with other studies that have found that LTCH use was associated with reduced STACH stays and may suggest some unique aspects of the severe wound population or initial response to the policy change¹⁰. In addition, we did not find a statistically significant change in average length of stay at the LTCHs after the implementation of the new criteria. Interestingly, we found a significant increase in the use of home health in the overall sample, but no significant change in SNF use. However, we did find a significant increase in the percentage of the highest severity of illness cases going to SNFs. While additional analyses are needed to further elucidate the impact of the policy on patterns of care, our results are consistent with the hypothesis that SNF care substituted for LTCH care among the highest severity cases (with resulting higher readmission and increased post-discharge sepsis), while HHA overall volume increased, to some extent, as a result of receiving relatively lower severity cases who would have otherwise been discharged to SNFs.

We did not find evidence that the policy resulted in savings to the Medicare program in the full sample or the high-LTCH-propensity sample. We found a potential explanation for the lack of evidence on savings in our analysis of the high-LTCH-propensity sample. Specifically, we found an increase in readmissions and rates of post-discharge sepsis. Our results suggest that some of the potential savings from the policy may have been offset by high spending on readmissions and care for costly conditions, like sepsis.

Congress passed new patient criteria for LTCHs in Medicare to increase the efficiency of the healthcare delivery system by focusing LTCH care on high severity patients, which it defined as those patients who spent 3 or more days in an ICU or who need prolonged mechanical ventilation. The current criteria has some support in the literature; Koenig et al.¹¹ found LTCH care to be cost-effective for patients with 3 or more STACH ICU days in three of the five conditions studied and to be associated with lower mortality but higher spending in the remaining two conditions. However, industry representatives have raised concern that certain high-severity patient populations, such as severe wounds, may be negatively impacted by the new criteria. In the 21st Century Cures Act, Congress has mandated a study by the Government Accountability Office, due by October 2020, to examine treatment needs of severe wound patients and the impact of new patient criteria on access to care for this population.

The patient criteria and associated payment policies were originally scheduled to be fully implemented during FY 2018 cost report periods. However, Congress extended the phase-in until FY 2020 cost reporting periods in the 21st Century Cures Act. Once fully implemented, LTCHs will receive the lower of cost or an amount comparable to what a STACH would receive for a similar patient. As compared to current payment levels, full implementation is expected to result in an additional 20–40% reduction in payments for site neutral cases. Our results suggest such additional payment reductions will result in further reduced access to LTCHs among severe wound cases. The impact on patients and the Medicare trust fund is uncertain and will depend on the extent to which other providers are able to adapt to the shifts created by the policy. Our results indicate that the full implementation of criteria for severe wound cases may produce unintended consequences and not yield the anticipated level of savings.

Our study has several limitations. Our estimates on the effect of the new LTCH payment policy are based on severe wound patients who are hospitalized in STACHs with at least three severe wound patients and limited to hospitals that send the majority of their severe wound LTCH patients to a single LTCH. As a result, our results may not be generalizable to severe wound patients hospitalized in other STACHs. Our baseline estimates of utilization and outcomes come from the second and third month before transition to the new payment rule. To the extent that LTCHs started responding to the new policy during these months, our methodology would under-estimate the effect of the new LTCH payment policy on LTCH use. Similarly, comparison group hospitals do ultimately experience the new LTCH payment policy. Anticipating this future payment policy change, comparison group hospitals may have begun to transition to the new policy during the study period, leading us to under-estimate the effect of the new LTCH payment policy. We associate each STACH with a single LTCH, but there may be other LTCHs in a STACH’s market with different policy implementation dates, increasing variability in our estimates. Our measure of Medicare spending (institutional reimbursements from the Medicare trust to STACHs, LTCHs, IRFs, other inpatient settings, SNFs, and HHAs) does not include outpatient services or other Medicare Part B covered services. As a result, the effects of the policy on spending may be under-estimated since certain services (such as wound debridement) are bundled in the LTCH payment while they are not bundled for SNFs and HHAs. Moreover, we used CMS’ definition of severe wound cases to identify our study population. Other definitions may be used to identify severe wound cases that are appropriate for LTCH care.

Finally, our study examines only partial implementation of the policy and, thus, provides only early evidence on the effects of the new policy. How these early effects will change over time is ambiguous. On one hand, these initial effects may increase in magnitude after the phase-in period and blended payments end and the full implementation of the criteria begin. On the other hand, the effects may lessen over time if SNFs become better equipped to substitute for LTCH care in the future.

Conclusions

The Centers for Medicare & Medicaid Services recently implemented LTCH payment system changes, leading to a reduction in payments for cases that do not meet certain criteria. We examined the effects of this change on access to care and outcomes for site-neutral severe wound patients in the first year of policy implementation. Our findings showed that the policy change has resulted in a reduction in LTCH use among severe wound patients, but no statistically significant change in Medicare spending over a 60-day episode. We also found evidence for higher readmissions and sepsis rate among severe wound patients who are most likely to receive LTCH care. Policymakers should consider actions to minimize potential unintended effects of the policy prior to its full implementation. While our study focused on the effects of the new LTCH payment policy on site neutral severe wound patients, other patient populations are expected to be influenced by this policy change as patient populations in LTCHs and other post-acute care providers change as a result of the policy. Future research should investigate the LTCH payment policy change on broader patient population.

Transparency

Declaration of funding

This study was funded by the National Association of Long-Term Hospitals (NALTH). NALTH approved the manuscript before submission.

Declaration of financial/other relationships

B.D., S.S., and L.K. are employees of KNG Health Consulting, LLC, which received consulting fees from NALTH. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors would like to thank Elizabeth Hamlett and NALTH’s Research and Quality Committee.