Abstract
Objective
The purpose of this study was to determine the factors influencing the discharge to an inpatient rehabilitation facility (IRF) or a skilled nursing facility (SNF) of people poststroke with Medicare Advantage plans.
Methods
A retrospective cohort study was conducted with data from naviHealth, a company that manages postacute care discharge placement on behalf of Medicare Advantage organizations. The dependent variable was discharge destination (IRF or SNF). Variables included age, sex, prior living setting, functional status (Activity Measure for Post-Acute Care [AM-PAC]), acute hospital length of stay, comorbidities, and payers (health plans). Analysis estimated relative risk (RR) of discharge to SNF, while controlling for regional variation.
Results
Individuals discharged to an SNF were older (RR = 1.17), women (RR = 1.05), lived at home alone or in assisted living (RR = 1.13 and 1.39, respectively), had comorbidities impacting their function “some” or “severely” (RR = 1.43 and 1.81, respectively), and had a length of stay greater than 5 days (RR = 1.16). Individuals with better AM-PAC Basic Mobility (RR = 0.95) went to an IRF, and individuals with better Daily Activity (RR = 1.01) scores went to an SNF. There was a substantial, significant variation in discharge of individuals to SNF by payer group (RR range = 1.12–1.92).
Conclusions
The results of this study show that individuals poststroke are more likely to be discharged to an SNF than to an IRF. This study did not find a different discharge decision-making picture for those with Medicare Advantage plans than previously described for other insurance programs.
Impact
Medicare Advantage payers have varied patterns in discharge placement to an IRF or SNF for patients poststroke.
Keywords: Medicare, Rehabilitation, Stroke
Introduction
Stroke is a leading cause of long-term disability in the United States, and its prevalence is projected to increase by 20% over the next decade, along with a 150% increase in the cost of stroke from 72 billion to 184 billion dollars.1 Over the past 2 decades, there has been a focus on Medicare reform, including the Affordable Care Act, and the Bundled Care Payment Initiative, with a goal of reducing health care costs. Two targets of health care cost reduction were acute hospital patient length of stays and postacute care (PAC) services. PAC services now account for $60 billion per year for Medicare and demonstrate the largest source of regional variation in spending and use of facilities.2,3 Diagnosis of stroke accounts for the highest spending in PAC and the largest group to use inpatient rehabilitation facilities (IRFs) and skilled nursing facilities (SNFs).4
Among those referred for PAC in a facility, whether to select IRF or SNF represents a challenge for clinicians, patients, and families, because of the rapid decisions that must be made due to increasingly shorter hospital length of stay.5–7 Discharge planning is also complicated by the limited available evidence to aid prognosis of the long-term functional recovery patients will achieve.8,9 Policy guidelines for admission into IRF and SNF are related to expected outcomes.1,4,10 Specifically, individuals discharged to an IRF are expected to achieve significant functional improvement within a relatively short period (eg, 2 weeks) and return to a community setting.11 An SNF is recommended for clients who are expected to achieve only partial or slow recovery, or when maintenance of health status or prevention of deterioration are primary goals.11
Prior studies have suggested that individuals poststroke discharged to an IRF and SNF are characteristically different when assessing demographics, clinical characteristics, regional locations, and insurance/payer plans. Individuals that were older, Black, women, uninsured, and had lower income and lower premorbid and poststroke function were more likely to be discharged to an SNF than to an IRF.12–14 Local and regional market characteristics such as hospital competition, the availability of IRFs and SNFs in the area, insurance coverage, and limited networks have also been shown to affect use of IRFs or SNFs.15,16 Individuals covered by Medicaid are more likely to be placed in an SNF than an IRF.14,17 Thus, the variation in discharge placement to a PAC facility after stroke continues to demonstrate complex patterns influenced by demographic, clinical, or external factors.
Medicare Advantage (MA) is a type of health insurance plan offered by private companies that are approved by Medicare to manage the health care of beneficiaries enrolled in the plan. MA plans are comprised of different plan policies, such as health maintenance organization (HMO) or preferred provider organization (PPO). A primary goal for MA plans is to arrange for appropriate care to meet the member’s needs, while reducing variation of PAC cost.
Guidelines for discharge placement to a PAC facility under an MA plan follow the same criteria as stated above; however, it is not known which factors may be driving discharge placement under the MA policy model. The purpose of this research is to determine which factors influence discharge to IRF or SNF after the acute episode of care for patients poststroke who have an MA plan. We hypothesized that external factors such as insurance plan and policy would demonstrate greater influence on discharge placement to an IRF or SNF, than would demographic or clinical factors, for individuals with acute stroke who are enrolled in MA plans.
Methods
Dataset
A retrospective cohort study was conducted with data from naviHealth, Inc, a company that manages postacute care on behalf of MA organizations, to help guide the transition of care for Medicare patients. naviHealth provides clinical support services, proprietary decision-support tools, and advisory solutions, which the health plan partners use to make care and discharge decisions. naviHealth operates in 50 states and manages approximately 10 million MA members—approximately 36% of the MA population. naviHealth manages more SNF placements for MA Health Plans than IRF placements, thus more records in this dataset are from SNFs. The data from this study were extracted from the naviHealth dataset upon admission into the IRF or SNF (not from the acute care hospital stay) to ensure accuracy of data. Clinical documentation regarding relevant patients is securely transmitted to naviHealth. The documentation is extracted from the medical record by care coordinators at naviHealth and placed into discrete fields in the naviHealth database. Data were available from January 2012 to March 2020. PAC facilities can include home health and long-term care as well.
Cohort
Individual patients were included in the cohort if they had an MA plan, were 21 years or older, and had an IRF or SNF stay for cerebrovascular accident identified using ICD-9 codes 430–438 and ICD-10 codes I60-I69. Individuals enrolled in the Accountable Care Organization or contracts under the Bundled Payment Care Initiative or who were homeless were excluded. Individuals were removed if their age or Activity Measure for Post-Acute Care (AM-PAC) data (a score for each of the 3 domains) were missing, or if their length of stay in the acute care setting was greater than 70 days. The naviHealth dataset included deidentified data for MA members whom naviHealth manages for SNF and IRF placement, comprising 6 health plans (representing 6 states) (Figure). We excluded the areas where there was only an SNF contract because of the potential geographic bias that could not be accounted for. If the acute care provider Centers for Medicare & Medicaid Services (CMS) Certification Number was not identified, the individual was excluded from the dataset. This was to ensure the cohort demonstrated the care pattern of interest, where an individual had a stroke, was admitted to an acute care facility, and then discharged to an IRF or SNF. The study protocol was reviewed by the internal review board from the first author's (H.A.H.) institution (University of Utah, Salt Lake City, UT) and received an exempt determination (IRB_00124022). Data were deidentified in accordance with Safe Harbor guidelines (as defined by the Health Insurance Portability and Accountability Act).
Figure.
Consolidated Standards of Reporting Trials (CONSORT) diagram.
Conceptual Framework
Predictor variable selection was based on Andersen’s Behavioral Model of Health Services Use.18 There are 3 factors that make up the framework: (1) predisposing factors, defined here as demographic characteristics such as age and sex; (2) need factors, defined as clinical characteristics, which represent the immediate cause of health service use, the functional problem that generated the need for health care services, and the needs identified by the health care team; and (3) enabling factors, defined as external characteristics, which are the logistical aspects of obtaining care, and include health insurance policies and plans.18,19
Variables of Interest
Primary Outcome
The primary dependent variable was discharge destination after acute ischemic stroke: IRF or SNF.
Predictors of Discharge to IRF or SNF
Demographic characteristics include age, sex, and prior living setting. Age was calculated as a categorical variable based on a 10-year time period in the regression models for ease of interpretation. Prior living setting refers to the setting the individual lived in prior to the stroke episode of care. Prior living setting is comprised of 4 levels: living home alone, living at home with family or paid caregiver, required assistance (assisted living, board and care, long-term care, or hospice), or missing values.
Clinical characteristics include the functional status based on the AM-PAC scores, acute hospital length of stay (days), and additional clinical condition(s) that increase the complexity of health care management, including medical complexities and comorbidities.
AM-PAC is an item response theory-based measurement system assessing 3 functional domains: Basic Mobility, Daily Activities, and Applied Cognition. When the patient is admitted into the PAC setting, an initial evaluation is performed by the physical therapist, occupational therapists, or speech language pathologists. The naviHealth coordinator reviews the admission evaluations in the medical record within 24 hours. The naviHealth coordinator uses the evaluations to complete the AM-PAC Computer Adapted Test (CAT) v3.0.1. The AM-PAC demonstrates acceptable test–retest reliability for each scale with proxy assessment (0.68–0.90).20–22 The responses are transformed to an AM-PAC score on a 0–100 scale, with lower scores on this scale indicating worse functional ability. The 0–100 score is not the same as T-scale scores created by the AM-PAC CAT output; rather the score has been adjusted to allow ease of interpretation for clinicians and patients. The minimal detectable difference (MDD) values for the AM-PAC were established from the naviHealth database based on the adjusted 0–100 score and have the following values: Basic Mobility: 4.45 points; Daily Activity: 3.54 points; and Applied Cognition: 7.73 points (unpublished data).
Acute care hospital length of stay was coded as length of stay in days from admission and discharge dates.
Medical complexities are additional complications present above and beyond the primary admission diagnosis and comorbidities. These conditions may impact length of stay, tolerance to therapy hours, functional prognosis, and outcomes. A maximum of 3 complexities are recorded. The medical complexities variable was divided into 2 levels: (1) no medical complexities, and (2) presence of 1, 2, or 3 medical complexities.
Comorbidities are defined as comorbid conditions that impact function, other than the primary medical diagnosis. For this study, 3 levels were assessed and defined as comorbidities that had no impact on function, comorbidities that limit function, and comorbidities that severely limit function.
External characteristics included the payer product line (policy types) and payer groups (health/insurance plans). Payer product line refers to the type of managed care policy the individual has, either an HMO or PPO.
Payer groups are health plan partners for which naviHealth provides utilization management services and PAC management services, which contractually allow use of deidentified data for research purposes. In this dataset, there are 6 payer groups.
Statistical Analysis
Descriptive statistics were calculated for the entire cohort and by discharge destination (IRF, SNF), using mean, SD, median, 25th percentiles, and 75th percentiles for the continuous variables, and counts and percentages for the categorical variables. Descriptive comparisons between the groups were made for the continuous variables using an independent samples t test and for the categorical variables using a chi-square test. Univariable mixed effects regression models were constructed comparing each on-study variable with discharge destination, where hospital random intercepts were included to account for potential correlation of discharge destinations within hospitals. Poisson outcome models with robust error variance were used rather than conventional Bernoulli outcome models to facilitate interpretations as relative risks (RRs) rather than odds ratios (ORs).23 This approach is recommended in the medical literature when the outcome is nearly evenly split between groups.24 All on-study variables were included in the multivariable model. Multicollinearity was assessed using the variance of inflation factor (VIF), where we intended to drop variables with a VIF greater than 4 from the model. RRs and 95% CIs were reported. To assess the impact of AM-PAC scores on discharge, we used the MDD to operationalize units of AM-PAC. Analyses were performed using Stata 15/MP (StataCorp LLC, College Station, TX, USA).
Role of the Funding Source
The funder played no role in the design, conduct, or reporting of this study.
Results
Characteristics of the cohort are reported in Table 1. Among 14,281 individuals included in the final cohort, 6410 (44.8%) were discharged to an IRF and 7871 (55.2%) to an SNF. Length of stay was skewed, and this variable was dichotomized by the median of 5 days. All variables, except payer plan (HMO vs PPO) were statistically different from each other for the IRF and SNF (P < .05). Univariate analyses are reported in Table 2. No multicollinearity was observed.
Table 1.
Summary Descriptive Statistics of Predisposing, Need, and Enabling Factors for the Entire Cohort and for Each Setting: Inpatient Rehabilitation Facility (IRF) and Skilled Nursing Facility (SNF)
| Variables |
Entire Cohort N = 14,281 |
IRF n = 6410 (44.8%) |
SNF n = 7871 (55.2%) |
P | |
|---|---|---|---|---|---|
| Continuous variables: Mean (SD) Median (IQR) |
|||||
| Age, y | 79.4 (9.3) 80.0 (73.0–86.0) |
77.7 (9.0) 78.0 (72.0–84.0) |
80.8 (9.2) 82.0 (75.0–88.0) |
<.001a | |
| Categorical variables, n (%) | Variable coding | ||||
| Sex | Male | 6194 (43.4) | 3001 (46.8) | 3193 (40.6) | <.001b |
| Living setting prior to stroke | Home with family/other Home alone Assistance Missing |
5842 (40.9) 5838 (40.9) 2407 (16.9) 194 (1.4) |
3104 (48.4) 2521 (39.3) 606 (9.5) 179 (2.8) |
2738 (34.8) 3317 (42.1) 1801 (22.9) 15 (0.2) |
<.001b |
| Need—clinical characteristics | |||||
| Continuous variables: Mean (SD) Median (IQR) |
|||||
| Activity measure for postacute care | Basic mobility | 33.3 (10.3) 35.0 (30.0–42.0) |
35.6 (9.5) 36.0 (32.0–43.0) |
31.5 (10.6) 33.0 (22.0–40.0) |
<.001a |
| Daily activity | 27.0 (8.6) 29.0 (24.0–33.0) |
28.2 (7.1) 39.0 (25.0–33.0) |
26.1 (9.5) 28.0 (23.0–33.0) |
<.001a | |
| Applied cognition | 61.6 (11.1) 62.0 (56.0–69.0) |
62.8 (10.2) 63.0 (57.0–69.0) |
60.6 (11.6) 61.0 (55.0–67.0) |
<.001a | |
| Acute hospital length of stay, d | 6.6 (5.6) 5.0 (3.0–8.0) |
5.8 (4.7) 5.0 (3.0–7.0) |
7.3 (6.2) 6.0 (3.0–9.0) |
<.001a | |
| Categorical variables, n (%) | Variable coding | ||||
| Medical complexities | Count—none Count—1 or more (max. 3) |
12,306 (86.2) 1975 (13.8) |
5660 (88.3) 750 (11.7) |
6646 (84.4) 1225 (15.6) |
<.001b |
| Comorbidities | 0, 1, 2: no impact on function 3: limits function 4, 5: severely limits function |
3992 (28.0) 8318 (58.2) 1971 (13.8) |
2525 (39.4) 3418 (53.3) 467 (7.3) |
1467 (18.6) 4900 (62.3) 1504 (19.1) |
<.001b |
| Enabling—external characteristics | |||||
| Categorical variables, n (%) | Variable coding | ||||
| Payer/product line | HMO PPO |
7370 (51.6) 6910 (48.4) |
3367 (52.5) 3042 (47.5) |
4003 (50.9) 3868 (49.1) |
.046b |
| Contracts/payer groups/health plans | Pennsylvania 1 Michigan Pennsylvania 2 New Jersey South Carolina/Georgia New York |
8314 (58.2) 1951 (13.7) 1400 (9.8) 1077 (7.5) 970 (6.8) 569 (4.0) |
4238 (66.1) 748 (11.7) 145 (2.3) 501 (7.8) 563 (8.8) 215 (3.4) |
4076 (51.8) 1203 (15.3) 1255 (15.9) 576 (7.3) 407 (5.2) 354 (4.5) |
<.001b |
Independent samples t test. IQR = interquartile range.
Mann-Whitney U test.
Table 2.
Univariable and Multivariable Mixed Effects Generalized Linear Model Using the Poisson Regression Model to Estimate Relative Risks of Discharge to a Skilled Nursing Facility Versus an Inpatient Rehabilitation Facilitya
| Variable | Level |
Univariable Model RR (95% CI) |
P b |
Multivariable Model RR (95% CI) |
P b |
|---|---|---|---|---|---|
| Predisposing—demographic characteristics | |||||
| Age | For each 10-y increase | 1.20 (1.17–1.24) | <.001 | 1.17 (1.15–1.20) | <.001 |
| Sex | Female | 1.13 (1.07–1.18) | <.001 | 1.05 (1.01–1.09) | <.02 |
| Living setting prior to stroke | Home with family/other Home alone Assistance Missing |
Reference 1.13 (1.09–1.17) 1.61 (1.51–1.72) 0.15 (0.09–0.25) |
<.001 <.001 <.001 |
Reference 1.13 (1.09–1.18) 1.39 (1.33–1.46) 0.19 (0.12–0.31) |
<.001 <.001 <.001 |
| Need—clinical characteristics | |||||
| AM-PAC basic mobility | For each 4.45-point increase, MDD | 0.93 (0.92–0.94) | <.001 | 0.95 (0.94–0.96) | <.001 |
| AM-PAC daily activity | For each 3.54-point increase, MDD | 0.96 (0.94–0.97) | <.001 | 1.01 (1.00–1.02) | <.001 |
| AM-PAC applied cognition | For each 7.73-point increase, MDD | 0.93 (0.91–0.96) | <.001 | 1.00 (0.98–1.01) | .64 |
| Acute hospital length of stay | Median split ≤5 d >5 |
Reference 1.24 (1.19–1.20) |
<.001 | Reference 1.16 (1.12–1.20) |
<.001 |
| Medical complexities | Count—none Count—1 or more (max. of 3) |
Reference 1.20 (1.13–1.29) |
<.001 | 1.07 (1.02–1.12) | .03 |
| Comorbidities | 0, 1, 2: no impact on function 3: limits function 4, 5: severely limits function |
Reference 1.61 (1.52–1.70) 2.16 (1.96–2.38) |
<.001 <.001 |
Reference 1.43 (1.36–1.50) 1.81 (1.70–1.95) |
<.001 <.001 |
| Enabling—external characteristics | |||||
| Payer, product line | HMO PPO |
Reference 0.98 (0.94–1.01) |
.16 | Not included | NA |
| Contracts, payer groups, health plans | Pennsylvania 1 Michigan Pennsylvania 2 New Jersey South Carolina/Georgia New York |
Reference 1.20 (1.12–1.29) 0.87 (0.80–0.94) 1.05 (0.9–1.16) 1.74 (1.64–1.84) 1.12 (1.09–1.39) |
<.001 .001 .30 <.01 .001 |
Reference 1.28 (1.19–1.37) 1.00 (0.94–1.08) 1.12 (1.02–1.23) 1.92 (1.81–2.03) 1.18 (1.06–1.33) |
<.001 .90 .02 <.001 <.01 |
The setting acute hospital (using CMS certification number) was used as a random effect. AM-PAC = Activity Measure for Post-Acute Care; CMS = Centers for Medicare & Medicaid Services ; HMO = health maintenance organization; MDD = minimal detectable difference; NA = not applicable; PPO = preferred provider organization; RR = relative risk.
P values <.10 from the univariable model were included in the multivariable model.
The main findings of the multivariate analysis are presented in Table 2. All demographic characteristics were statistically significant predictors of discharge destination. Individuals who were older (RR = 1.17; 95% CI: 1.15–1.20; P < .001) were more likely to be discharged to an SNF (n = 184 of the entire sample were 29–54 years old). Individuals that lived home alone, or were in an assisted living, or Board and Care facility were more likely to be discharged to an SNF compared with individuals who had support at home (RR = 1.13; 95% CI: 1.09–1.18; RR = 1.39; 95% CI: 1.33–1.46, respectively; P < .001). Females were slightly more likely to go to an SNF (RR = 1.05; 95% CI: 1.01–1.09; P = .02).
All of the clinical characteristics were statistically significant predictors of discharge destination, except AM-PAC Applied Cognition score (RR = 1.00; 95% CI: 0.98–1.01; P = .64). The largest effects were noted in comorbidities. Individuals who had functional comorbidities prior to their stroke that limited their functional ability or severely limited their functional ability were 43% and 81%, respectively, more likely to be discharged to an SNF (RR = 1.43; 95% CI: 1.36–1.50; RR = 1.81; 95% CI: 1.70–1.95; P < .001). Individuals with a length of stay greater than 5 days were more likely to be discharged to an SNF (RR = 1.16; 95% CI: 1.12–1.20; P < .001). AM-PAC Basic Mobility and Daily Activities scores had a small effect on discharge placement. For every increase of 4.45 points on the AM-PAC Basic Mobility scale, individuals were 5% more likely to be discharged to an IRF (RR = 0.95; 95% CI: 0.94–0.96; P < .001), whereas for every increase of 3.54 points on the Daily Activities scale, individuals were 1% more likely to go to an SNF (RR = 1.01; 95% CI: 1.00–1.02; P < .001).
There was substantial variation in likelihood of discharge to an SNF (RR ranging from 1.12 to 1.92) by payer group, with all but one demonstrating significantly greater chance of discharge to SNF than the reference payer group (Pennsylvania 1).
Discussion
In this study of 14,281 individuals with postacute stroke who are enrolled in an MA plan, we identified factors that predicted discharge to an IRF versus an SNF setting. We sought to determine which factors (demographic, clinical, or external) had more influence on discharge placement and hypothesized that external characteristics such as insurance plan and policy would influence discharge placement, more than demographic or clinical characteristics. Our results suggest that our hypothesis was partly correct, because some variables in each category influenced discharge placement in this cohort. In summary, our results found that individuals that were older, lived at home alone, did not live at home prior to their stroke, had comorbidities that limited function prior to their stroke, had more medical complexities (adjustors), and all but one payer group was more likely to be discharged to a SNF compared with the reference payer group (Pennsylvania 1).
Our results related to demographic characteristics are consistent with published literature on discharge placement to an IRF or SNF for individuals poststroke.12–14 Prior studies have found that individuals discharged to an SNF are different from those discharged to an IRF; specifically, they are older and more likely female. It may be that older individuals are sicker, have more comorbidities, and that females may live at home alone, or the spouse is not able to assist. In sum, data from individuals covered by MA plans demonstrate similar discharge patterns related to demographic characteristics as other insurance programs.12–14
Our results related to clinical characteristics are similar to prior studies that suggest patients poststroke with more comorbidities and premorbid lower mobility are more likely to be discharged to an SNF.12,14,17,25 The AM-PAC functional assessment finds marginal difference between the groups. Clinically, the MDD score may not be meaningful. For example, the median score for the entire cohort was 33.3 (range 30.0–42.0), which places most individuals in the maximum to moderate assistance for Basic Mobility, suggesting similar clinical functional ability for these groups.
Prior studies have also found that external characteristics such as insurance plans influence discharge placement. Individuals with Medicaid, uninsured, and individuals with MA plans (compared with traditional Medicare) are more likely to be discharged to an SNF.14,17,26 Our results add to the understanding of discharge placement patterns for individuals with postacute stroke with MA health plans. There is variation in how these MA insurance plans manage PAC services for individuals poststroke. Some plans sent up to 86% of their patients to an SNF. Interestingly, the results of this study found that the insurance policy (HMO or PPO) did not influence discharge placement to an IRF or SNF. One explanation for this may be because MA plans seek to contract with SNFs who provide higher quality care or to larger facilities associated with chains.26–28
The implementation of MA was one of many solutions to attempt to reduce the geographic variation of Medicare spending and use of PAC services. These results suggest that geographic variation is still prevalent. The geographic locations in this study are also related to different payer groups, and large variation of discharge placement is observed by these payers. The variation in discharge placement has not been found to be related to differences in the health of the beneficiaries, which these results support as well.29
Limitations and Future Directions
This research has several limitations including using only 1 PAC contractor service, that is, naviHealth and their contracted MA payers. There is a limitation in using only an administrative dataset because there is a lack of detailed clinical/patient-level data, such as additional functional and severity assessments, such as the modified Rankin scale or National Institutes of Health scale, which are commonly used in clinical assessments. Moreover, there is a lack of environmental details, for example, home setup, and a lack of social demographics, such as zip code or race and ethnicity. A further limitation is a lack of understanding of hospital variability and plan and regional variability, which need further investigation. Also, there is no comparison with traditional Medicare data, which is recommended in future studies to aid in understanding the discharge decision-making. There is limited geographic representation, which has been shown to have a difference in discharge placement patterns. Future studies should also include additional PAC settings, such as home health or long-term care. Future research needs to include cost-effectiveness studies and understand long-term functional outcomes for patients poststroke in order to understand the relative costs and outcomes of different courses of action.
Conclusion
The results of this study suggest that in this cohort of individuals poststroke who have an MA plan the beneficiaries are more likely to be discharged to an SNF over an IRF for their PAC needs. Furthermore, variables that relate to demographic characteristics, such as increased age and female sex, are associated with discharge to an SNF over an IRF. Clinical characteristics that are associated with discharge to an SNF over an IRF are related to premorbid conditions, medical complexities, and comorbidities, all of which impact functional ability and potential functional recovery. In this study, insurance payer groups and regional location accounted for the largest variation in discharge to an SNF versus an IRF. Medical necessity is an additional influence on discharge decision that cannot be captured from this type of dataset. The discharge placement of individuals poststroke to an IRF or SNF continues to demonstrate influence by nonclinical factors, even for patients with an MA plan. In sum, the results from this analysis of MA plans do not demonstrate a different discharge picture than previously described when assessing other insurance programs.
Acknowledgments
The authors acknowledge Amy Leibensberger, Angela Gaetano, Alice Townsend, Jay LaBine, Gina Bruno, Gregory Gadbois, Jennifer Terrell, Tessa Balkon, Albert de Hombre, Katie Atikins, and Colleen O’Rourke for their work.
Contributor Information
Heather A Hayes, Department of Physical Therapy and Athletic Training, University of Utah, Salt Lake City, Utah, USA.
Vincent Mor, Department of Health Services, Policy & Practice, Brown University School of Public Health, Providence, Rhode Island, USA; Providence Veteran’s Administration Medical Center, Providence, Rhode Island, USA.
Guo Wei, Study Design and Biostatistics Center, University of Utah, Salt Lake City, Utah, USA.
Angela Presson, Department of Internal Medicine, University of Utah, Salt Lake City, Utah, USA.
Christine McDonough, Department of Physical Therapy, University of Pittsburgh, Pittsburgh, Pennsylvania, USA.
Author Contributions
Concept/idea/research design: H.A. Hayes, V. Mor, C. McDonough
Writing: H.A. Hayes, C. McDonough
Data analysis: H.A. Hayes, V. Mor, G. Wei, A. Presson
Project management: H.A. Hayes
Fund procurement: H.A. Hayes
Providing institutional liaisons: H.A. Hayes, V. Mor
Funding
This study was funded by a 2019 Center on Health Services Training and Research (CoHSTAR) Faculty Fellowship grant from the Foundation for Physical Therapy Research (to H.A. Hayes).
Ethics Approval
Each author certifies that their institution approved the human protocol for this investigation and confirmed that all investigations were conducted in conformity with ethical principles of research. This is a retrospective cohort data study (chart review) and did not require consent. Data were deidentified.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Disclosures
The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.
References
- 1. Alcusky M, Ulbricht CM, Lapane KL. Postacute care setting, facility characteristics, and poststroke outcomes: a systematic review. Arch Phys Med Rehabil. 2018;99:1124–1140 e1129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Mechanic R. Post-acute care—the next frontier for controlling Medicare spending. N Engl J Med. 2014;370:692–694. [DOI] [PubMed] [Google Scholar]
- 3. Medicare Payment Advisory Commission . A data book: health care spending and the Medicare program. MedPAC; 2020. Accessed April 13, 2021. http://www.medpac.gov/docs/default-source/data-book/july2020_databook_entirereport_sec.pdf?sfvrsn=0. [Google Scholar]
- 4. Regenbogen SE, Cain-Nielsen AH, Syrjamaki JD, Chen LM, Norton EC. Spending on postacute care after hospitalization in commercial insurance and Medicare around age sixty-five. Health Aff (Millwood). 2019;38:1505–1513. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Zarshenas S, Colantonio A, Alavinia SM, Jaglal S, Tam L, Cullen N. Predictors of discharge destination from acute care in patients with traumatic brain injury: a systematic review. J Head Trauma Rehabil. 2019;34:52–64. [DOI] [PubMed] [Google Scholar]
- 6. Greene NH, Kernic MA, Vavilala MS, Rivara FP. Variation in adult traumatic brain injury outcomes in the United States. J Head Trauma Rehabil. 2018;33:E1–E8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Wang H, Sandel ME, Terdiman J, et al. Postacute care and ischemic stroke mortality: findings from an integrated health care system in northern California. PM R. 2011;3:686–694. [DOI] [PubMed] [Google Scholar]
- 8. Stinear CM, Smith MC, Byblow WD. Prediction tools for stroke rehabilitation. Stroke. 2019;50:3314–3322. [DOI] [PubMed] [Google Scholar]
- 9. Cormier DJ, Frantz MA, Rand E, Stein J. Physiatrist referral preferences for postacute stroke rehabilitation. Medicine (Baltimore). 2016;95:e4356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Enderby P, Pandyan A, Bowen A, et al. Accessing rehabilitation after stroke—a guessing game? Disabil Rehabil. 2017;39:709–713. [DOI] [PubMed] [Google Scholar]
- 11. Winstein CJ, Stein J, Arena R, et al. Guidelines for adult stroke rehabilitation and recovery: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2016;47:e98–e169. [DOI] [PubMed] [Google Scholar]
- 12. Rakesh N, Boiarsky D, Athar A, Hinds S, Stein J. Post-stroke rehabilitation: factors predicting discharge to acute versus subacute rehabilitation facilities. Medicine (Baltimore). 2019;98:e15934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Stein J, Bettger JP, Sicklick A, Hedeman R, Magdon-Ismail Z, Schwamm LH. Use of a standardized assessment to predict rehabilitation care after acute stroke. Arch Phys Med Rehabil. 2015;96:210–217. [DOI] [PubMed] [Google Scholar]
- 14. Freburger JK, Holmes GM, Ku LJ, Cutchin MP, Heatwole-Shank K, Edwards LJ. Disparities in postacute rehabilitation care for stroke: an analysis of the state inpatient databases. Arch Phys Med Rehabil. 2011;92:1220–1229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Chen LM, Norton EC, Banerjee M, Regenbogen SE, Cain-Nielsen AH, Birkmeyer JD. Spending on care after surgery driven by choice of care settings instead of intensity of services. Health Aff (Millwood). 2017;36:83–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Stein J, Bettger JP, Sicklick A, Hedeman R, Magdon-Ismail Z, Schwamm LH. Use of a standardized assessment to predict rehabilitation care after acute stroke. Arch Phys Med Rehabil. 2015;96:210–217. [DOI] [PubMed] [Google Scholar]
- 17. Hong I, Goodwin JS, Reistetter TA, et al. Comparison of functional status improvements among patients with stroke receiving postacute care in inpatient rehabilitation vs skilled nursing facilities. JAMA Netw Open. 2019;2:e1916646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? J Health Soc Behav. 1995;36:1–10. [PubMed] [Google Scholar]
- 19. Travers JL, Hirschman KB, Naylor MD. Adapting Andersen's expanded behavioral model of health services use to include older adults receiving long-term services and supports. BMC Geriatr. 2020;20:58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Andres PL, Haley SM, Ni PS. Is patient-reported function reliable for monitoring postacute outcomes? Am J Phys Med Rehabil. 2003;82:614–621. [DOI] [PubMed] [Google Scholar]
- 21. Andres PL, Black-Schaffer RM, Ni P, Haley SM. Computer adaptive testing: a strategy for monitoring stroke rehabilitation across settings. Top Stroke Rehabil. 2004;11:33–39. [DOI] [PubMed] [Google Scholar]
- 22. Haley SM, Siebens H, Coster WJ, et al. Computerized adaptive testing for follow-up after discharge from inpatient rehabilitation: I. Activity outcomes. Arch Phys Med Rehabil. 2006;87:1033–1042. [DOI] [PubMed] [Google Scholar]
- 23. Zou G. A modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159:702–706. [DOI] [PubMed] [Google Scholar]
- 24. Zhang J, Yu KF. What's the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998;280:1690–1691. [DOI] [PubMed] [Google Scholar]
- 25. Hayes HA, Marcus R, Stoddard GJ, McFadden M, Magel J, Hess R. Is the activity measure for postacute care “6-clicks” tool associated with discharge destination postacute stroke? Arch Rehabil Res Clin Transl. 2022;4:100228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Huckfeldt PJ, Escarce JJ, Rabideau B, Karaca-Mandic P, Sood N. Less intense postacute care, better outcomes for enrollees in Medicare Advantage than those in fee-for-service. Health Aff (Millwood). 2017;36:91–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Rahman M, Meyers DJ, Mor V. The effects of Medicare Advantage contract concentration on patients' nursing home outcomes. Health Serv Res. 2018;53:4087–4105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Jung HY, Li Q, Rahman M, Mor V. Medicare Advantage enrollees' use of nursing homes: trends and nursing home characteristics. Am J Manag Care. 2018;24:e249–e256. [PMC free article] [PubMed] [Google Scholar]
- 29. Newhouse JP, Garber AM. Geographic variation in health care spending in the United States: insights from an Institute of Medicine report. JAMA. 2013;310:1227–1228. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.