Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Oct 1.
Published in final edited form as: Med Care. 2019 Oct;57(10):757–765. doi: 10.1097/MLR.0000000000001184

Trends in the Use of Skilled Nursing Facility and Home Health Care under the Hospital Readmissions Reduction Program: An Interrupted Time-series Analysis

Ioana Popescu 1, Neeraj Sood 2, Sushant Joshi 2, Peter Huckfeldt 3, José Escarce 1, Teryl Nuckols 4
PMCID: PMC6742564  NIHMSID: NIHMS1535628  PMID: 31453891

Abstract

Background:

Medicare’s Hospital Readmission Reduction Program (HRRP) penalizes hospitals with elevated 30-day readmission rates for acute myocardial infarction (AMI), heart failure (HF), or pneumonia. To reduce readmissions, hospitals may have increased referrals to skilled nursing facilities and home health care.

Research Design:

Outcomes included 30-day post-discharge utilization of skilled nursing facility and home health care, including any use as well as days of use. Subjects included Medicare fee-for-service beneficiaries aged ≥65 who were admitted with AMI, HF, or pneumonia to hospitals subject to the HRRP. Using an interrupted time-series analysis, we compared utilization rates observed after the announcement of the HRRP (April 2010-September 2012) and after the imposition of penalties (October 2012-September 2014) with projected utilization rates that accounted for pre-HRRP trends (January 2008-March 2010). Models included patient characteristics and hospital fixed effects.

Results:

For AMI and HF, utilization of skilled nursing facility and home health care remained stable overall. For pneumonia, observed utilization of any skilled nursing facility care increased modestly (1.0%, p<0.001 during anticipation; 2.4%, p<0.001 after penalties) and observed utilization of any home health care services declined modestly (−0.5%, p=0.008 after announcement; −0.7%, p=0.045 after penalties) relative to projections. Beneficiaries with AMI and pneumonia treated at penalized hospitals had higher rates of being in the community 30 days post-discharge.

Conclusions:

Hospitals might be shifting to more intensive post-acute care to avoid readmissions among seniors with pneumonia. At the same time, penalized hospitals’ efforts to prevent readmissions may be keeping higher proportions of their patients in the community.

Keywords: (3 – 5) Readmission, Medicare, home health care, skilled nursing facility, healthcare policy

BACKGROUND

The Medicare Hospital Readmissions Reduction Program (HRRP) financially penalizes hospitals with higher than expected 30-day readmission rates for Medicare fee-for-service beneficiaries age 65 years and older. Initially, the HRRP focused on three target conditions, acute myocardial infarction (AMI), heart failure (HF), and pneumonia, adding additional conditions later.1,2 In response to the policy, hospitals have several strategies available to decrease readmissions and reduce the risk of penalty.3,4 First, hospitals may invest in transition-related interventions that promote patient self-management activities, update primary care physicians, monitor patients more closely, and ensure timely follow-up visits—ideally, keeping patients at home during those first 30 days.5 These strategies, albeit costly and complex,6,7, are within hospitals’ reach and, in fact, may have contributed to the decrease in readmissions observed during the first years after the HRRP was implemented.3 However, transition-related care alone may be insufficient to reduce readmissions among seniors with complex treatment needs, functional impairments, or challenges traveling to post-hospitalization ambulatory visits. Among such vulnerable populations, referrals to post-acute care is another strategy for reducing readmissions.

Post-acute care offers important rehabilitative, therapeutic, and nursing services to Medicare beneficiaries after hospitalization. These services encompass an array of providers, including skilled nursing facilities, home health agencies, inpatient rehabilitation facilities, and long-term acute care hospitals. Skilled nursing facility (SNF) care and home health care are, by far, the most common types of post-acute care services used by seniors, whereas inpatient rehabilitation facility (IRF) and long-term acute care hospital (LTACH) use typically account for less than 5% post-acute care services used.8 Because the vast majority (80%) of hospital discharges to post-acute care for seniors with AMI, HF and pneumonia are for SNF and home health care,8–10 we focused on these two types of services.

Patients are eligible to receive home health and skilled nursing services if they have a specific need for skilled clinical care, such as nursing (e.g., for intravenous antibiotic administration or wound care), physical therapy, respiratory therapy, speech therapy, or other rehabilitation services. A need for assistance with basic life functions like eating, toileting, or bathing cannot be used as the sole basis for obtaining home health and skilled nursing services.11

Nonetheless, functional impairments are a common cause of readmission.12 Post-acute care services have been shown to improve functional outcomes, keep frail seniors in the community after hospitalization, and reduce preventable hospitalizations.13–15 When patients could have difficulty getting functional needs met at home, hospital physicians may be more likely to refer patients to home health and skilled nursing care as a strategy to prevent 30-day readmissions, if the requisite skilled needs are also present.

On the other hand, post-acute care utilization rates have risen steadily over the past decades as hospitals began reducing length of stay.8–10,16 Currently, post-acute care is considered a significant source of variation in Medicare spending and, as such, it represents a major target of payment reforms and programs aimed at controlling health care costs.17,18 Thus, pressures to increase in post-acute care use in response to the HRRP may be counterbalanced by efforts to rein in post-acute care costs.

The current study sought to examine whether trends in the utilization of post-acute care during the first 30 days after hospitalization changed following the announcement and implementation of the HRRP.

METHODS

Study Population

Subjects included Medicare fee-for-service beneficiaries aged 65 and older admitted between January 1, 2008 and September 30, 2014, with a principal discharge diagnosis of AMI (International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes 410 except 410.x2), HF (ICD-9-CM codes 402, 404 and 428), or pneumonia (ICD-9-CM codes 480–488), following previously established Medicare methodology.19

Episodes of Care

For each index hospitalization, we created a longitudinal episode of care consisting of the index hospitalization and any home health and skilled nursing care use during the first 30 days after discharge or until the beneficiary’s death (including post-acute care occurring after any readmissions). Following the HRRP penalty calculation methodology, we excluded episodes of care for beneficiaries who were released against medical advice, were hospitalized over 365 days, were discharged from an acute care hospital for the same condition in the preceding 30-days, or were not continuously enrolled in fee-for-service Medicare during the previous year or the month after the index hospitalization. We also excluded beneficiaries who died during the index hospitalization, or were admitted to hospitals ineligible for the HRRP. If an inter-hospital transfer occurred during the index hospitalization, episode outcomes were attributed to the hospital that discharged the patient.

Data Sources

We used Medicare Provider and Analysis Review (MedPAR) files to identify eligible index hospitalizations and post-acute SNF stays during the episode of care, and Home Health Agency (HHA) files to identify home health visits during that period. We also used the MedPAR files to obtain information on beneficiary clinical characteristics including target condition and comorbidities, identified from secondary diagnoses (from ICD-9-CM diagnosis and procedure codes) during the index hospitalization, adapting previously validated Medicare methods.19 The Medicare Beneficiary Summary file provided sociodemographic information including age, sex, race, and Medicaid dual eligibility status. We used the Medicare Provider of Services (POS) to obtain information on admitting hospital characteristics. Hospital eligibility for the HRRP, penalty status, and size were ascertained from publicly reported data on Medicare Hospital Compare.20 We employed the penalty factor released by Medicare in September 2012 for Fiscal Year 2013 (when HRRP penalties were implemented).21

Measures

The study outcomes were the condition-specific rates of home health and skilled nursing service utilization during the episode of care, and the rate of being continuously at home during the 30-day postdischarge period without the use of any type of post-acute care. We measured utilization outcomes both as binary variables (any use during the episode) and continuous variables (number of days these services were used during the episode) for all episodes of care. The total number of days was calculated for the entire beneficiary sample (i.e., users and non-users), and for users only.

We divided the study timeline into a baseline period, prior to the HRRP announcement (January 2008 through March 2010, 9 quarters); an anticipation period, occurring between the announcement of the policy and the imposition of penalties (April 2010 through September 2012, 10 quarters); and a penalty period, after the imposition of HRRP penalties (October 2012 through September 2014, 8 quarters). Our main independent variables were indicators for whether an episode of care occurred during the baseline, anticipation, or penalty period.

Analyses

First, we described patient and hospital characteristics for the study population, and average rates of readmission and post-acute care service use over the study period.

Next, we used linear probability models to estimate observed and expected utilization rates during each study period. Models controlled for patient characteristics (age, sex, race, and clinical characteristics), season, and included hospital fixed effects to account for between-hospital variations in post-acute care use patterns. Clinical characteristics consisted of comorbidities (e.g., diabetes, COPD, cancer) and severity measures (e.g., history of open-heart surgery, or anterior location of AMI), identified from secondary discharge diagnosis and procedure ICD-9-CM codes. We selected these variables following Medicare methodology.19 Model standard errors were corrected for clustering within hospital markets (hospital referral regions).22 Separate analyses were conducted for each primary outcome (any and total use of post-acute care services and being continuously in the community without post-acute care services during the 30-day postdischarge period), and by target condition. Expected outcomes reflect what conditions might have been in the absence of the HRRP.

Observed and expected outcomes were aggregated by quarter-year. We then graphically displayed the observed and expected risk-adjusted quarterly outcome rates in each study period, and we tested whether observed and expected outcome rates differed significantly during the anticipation or penalty periods using interrupted time-series analyses. A difference between observed and expected outcome rates during the anticipation and penalty periods suggests a change potentially associated with the HRRP.

To evaluate whether initial HRRP penalties influenced rates of post-acute care use we conducted sub-analyses stratified by penalty status. To assess whether outcomes of minority and low-income populations were affected differently during the HRRP implementation, we also conducted sub-analyses that compared changes between black patients and white patients, and between patients who were dually eligible for Medicaid and patients who were not dually eligible.

Analyses were performed in Stata version 14 (StataCorp LLC, College Station, TX).

RESULTS

Characteristics of the study population including average rates of home health and skilled nursing care utilization, and rates of being in the community without post-acute care service use services during the 30-days post-discharge period are presented in Table 1, for each condition separately and overall, for the combined cohorts. Overall, the mean beneficiary age was 80 years. Most beneficiaries were white (86%), followed by black (9%); only 5% were other races/ethnicities. Over a fifth of beneficiaries had Medicaid dual eligibility. Comorbidities were common; for example, 29% of beneficiaries had COPD, 28% had diabetes, and more than 10% had dementia. Demographic and clinical characteristics of the three cohorts were similar.

Table 1.

Characteristics of Medicare Beneficiaries with Index Hospitalizations for Acute Myocardial Infarction, Heart Failure, and Pneumonia from January 2008 through September 2014

Acute Myocardial Infarction Heart Failure Pneumonia Overall
N=1,104,349 N=2,538,244 N=2,076,809 N=5,719,402
Patient Sociodemographic Characteristics
Male, % 50.7 44.3 44.7 45.7
Age, mean 78.7 80.9 80.2 80.2
Race/ethnicity
 White, % 87.9 83.6 88.0 86.0
 Black, % 7.5 11.8 7.1 9.3
 Asian, % 1.3 1.2 1.4 1.3
 Hispanic, % 1.6 2.0 1.9 1.9
 Other, % 1.7 1.4 1.7 1.6
Dual Medicaid eligibility, % 17.7 22.3 23.9 22.0
Patient Comorbidities
COPD, % 16.0 27.5 38.9 29.4
Coronary heart disease, %* . 42.9 24.4 27.9
Diabetes mellitus, % 27.6 31.3 24.8 28.2
Anemia, % 16.9 18.8 21.0 19.3
Gastrointestinal disorders, %* . 11.9 20.9 12.9
Valvular heart disease, % 11.3 18.1 4.6 11.9
Dementia and other brain disorders, % 8.4 8.7 12.4 10.0
Pneumonia, % 9.3 15.0 . 8.5
Protein-calorie malnutrition, % 3.1 4.1 7.1 5.0
Decubitus ulcer or chronic skin ulcer, % 0.8 2.1 0.9 1.4
Urinary tract disorders, % 3.3 3.8 3.6 3.6
Disorders of fluid/electrolyte/acid-base, % 2.1 3.0 3.1 2.9
Cancer, % 4.4 4.7 9.9 6.5
Psychiatric disorders, %* . 6.4 6.2 5.1
Asthma, % 1.5 1.6 3.3 2.2
Disease Severity
History of prior angioplasty, %* 41.9 . . 8.1
History of prior open heart surgery, % 13.2 7.4 3.6 7.1
Anterior myocardial infarction, %* 7.8 . . 1.5
Other location of myocardial infarction, %* 11.6 . . 2.3
Use of Post-Acute Care during the 30 Days after Hospital Discharge
Any use of home health, % 23.9 28.9 23.7 26.0
Any use of skilled nursing, % 20.7 24.0 27.3 24.6
Total days of use of home health, mean 4.6 5.7 4.5 5.1
Total days of use of home health conditional on any use, mean 19.3 19.6 19.2 19.4
Total days of use of skilled nursing, mean (standard deviation) 4.0 4.7 5.5 4.8
Total days of use of skilled nursing conditional on any use, mean 19.2 19.5 20.1 19.7
At home continuously for 30-days, without post-acute care, % 46.4 36.9 39.8 39.8
Open in a new tab
*

According to Medicare risk-adjustment methodology, these comorbidities were only included for some conditions

Overall, one in four beneficiaries used home health and one in five beneficiaries used skilled nursing services during the 30 days after hospital discharge. On average, beneficiaries used these services for 5.1 (home health) and 4.8 (skilled nursing) of the 30 days after discharge (or 19.4 days for home health and 19.7 days for skilled nursing among those with any use). Characteristics of study hospitals are presented in the Supplemental Digital Content 1 (Appendix Table 1).

Figures 1 to 3 depict trends in observed and expected rates of any use of home health and skilled nursing services by quarter year, and the rate of being continuously in the community without post-acute care during the 30-day post-discharge period, for each target condition. Trends for total days of home health and skilled nursing services utilization are presented in Supplemental Digital Content 1 (Appendix Figures 1–2).

Figure 1.

Figure 1.

Open in a new tab

Home health care utilization trends before and after HRRP implementation for AMI, HF and pneumonia

Figure 3.

Figure 3.

Open in a new tab

Rates of being in the community for 30 days post-discharge, without the use of post-acute care, before and after HRRP implementation for AMI, HF and pneumonia

Observed and expected post-acute care utilization trends differed among the target conditions. Any home health utilization throughout the study period was stable for AMI, and slightly increased for HF and pneumonia (Figure 1). Total home health use followed similar trends (Appendix figure 1). Any skilled nursing service use was relatively stable for all three conditions throughout the study period, and so was total skilled nursing service use (Figure 2 and Appendix Figure 2). For both AMI and HF, observed and expected trends during the anticipation and penalty periods tracked each other closely. However, in the case of pneumonia, observed and expected utilization trends diverged. Home health use was lower during the anticipation and penalty periods than expected based on pre-HRRP trends, while SNF use was higher than expected. The rate of being continuously in the community for 30 days post-discharge increased modestly for AMI and decreased for HF and pneumonia; observed and expected trends followed each other closely for all conditions (Figure 3).

Figure 2.

Figure 2.

Open in a new tab

Skilled nursing care utilization trends before and after HRRP implementation for AMI, HF and pneumonia

Table 2 summarizes the mean differences between observed and expected home health and skilled nursing utilization rates, and rates of continuously being in the community for 30 days post-discharge, during the anticipation and penalty periods. Significant differences represent changes potentially associated with the HRRP. As seen in the Figures, for AMI and HF, there were no significant differences in either whether post-acute care services were used or in the total days of use. In the case of pneumonia there were small but significant changes in the use of home health and skilled nursing service use during the HRRP anticipation and penalty periods. Specifically, the percentage of episodes during which skilled nursing care was used increased (1 percentage point, p<0.001 during anticipation period; 2.4 percentage points, p<0.001 during penalty period) as did the total days of skilled nursing use during the penalty period (0.3 days, p<0.001 during penalty period). This coincided with decreases in any use of home health (−0.5 percentage points, p=0.008 during anticipation period). Similarly, small but significant decreases were observed for total days of home health use (−0.16 days, p<0.001 during anticipation period; −0.22 days, p=0.007 during penalty period). Observed and expected trends in being in the community continuously during the 30-day post-discharge period did not differ significantly for any of the target conditions.

Table 2.

Use of post-acute care and being continuously at home for 30 days after hospital discharge during the anticipation* and penalty† periods of the Hospital Readmissions Reduction Program: differences between risk-adjusted observed rate and risk-adjusted expected rates based on pre-policy trends

Any use of skilled nursing facility care Any use of home health care Continuously at home for 30 days post-discharge
Percentage points p-value Percentage points p-value Percentage points p-value
Acute myocardial infarction
Anticipation Period, Mean Difference between Observed and Expected Rates −0.1 0.828 −0.40 0.216 0.55 0.060
Penalty Period, Mean Difference between Observed and Expected Rates −0.1 0.896 −0.30 0.605 1.05 0.062
Difference between Anticipation and Penalty Periods, Mean 0.979 0.768 0.100
Heart Failure
Anticipation Period, Mean Difference between Observed and Expected Rates 0.4 0.024 −0.20 0.316 −0.16 0.468
Penalty Period, Mean Difference between Observed and Expected Rates 0.5 0.146 −0.5 0.252 0.06 0.468
Difference between Anticipation and Penalty Periods, Mean 0.689 0.266 0.315
Pneumonia
Anticipation Period, Mean Difference between Observed and Expected Rates 1.0 <0.001 −0.5 0.008 −0.06 0.790
Penalty Period, Mean Difference between Observed and Expected Rates 2.4 <0.001 −0.7 0.045 −0.84 0.049
Difference between Anticipation and Penalty Periods, Mean <0.001 0.289 <0.001
Open in a new tab
*

Anticipation period starts in April 2010 with the announcement of the policy and ends in September 2012 immediately before the imposition of penalties.

†

Penalty period starts in October 2012 with the imposition of financial penalties for hospitals and ends in September 2014.

Analyses of home health and skilled nursing use stratified by hospital penalty status (Digital Supplemental Content 2) were largely negative, except for a small but significant increase in any SNF use above expected trends for beneficiaries with pneumonia treated at non-penalized hospitals (2.7%, p<0.001 during the penalty period). Beneficiaries with AMI and pneumonia treated at penalized hospitals had higher than expected rates of being continuously at home for 30 days postdischarge (1.9%, p=0.0024 during anticipation and 3.8%, p=0.0015 during the penalty period for AMI, and 2.5%, p=0.0058 for pneumonia during the penalty period). On the other hand, beneficiaries admitted with pneumonia at non-penalized hospitals had lower than expected rates of being continuously at home for 30 days postdischarge (2.6%, p=0.0014 during the penalty period).

Analyses stratified by race showed that there were no significant differences in trends in the utilization of home health and skilled nursing facility care between black and white patients, or between dually eligible and non-dually eligible patients, for any of the target conditions. Similarly, there were no differences by race or dual eligibility in trends in the likelihood of being continuously at home.

DISCUSSION

Given the pressures to decrease readmission rates imposed by the HRRP and prior evidence that post-acute care services may help reduce preventable hospitalization, we had expected use of post-acute care within 30 days after hospitalization to rise following the announcement and penalty of the HRRP. Instead, we found nuanced, condition-specific changes. Among patients hospitalized for AMI and HF, trends in the use of post-acute care during the HRRP anticipation and penalty periods were generally consistent with pre-HRRP trends. Among patients with pneumonia, a small increase in the use of skilled nursing care coincided with a modest decrease in use of home health services during the HRRP anticipation and penalty periods. In addition, rates of being in the community without post-acute care service use for 30 days post-discharge increased for beneficiaries with AMI and pneumonia admitted to penalized hospitals and decreased for beneficiaries with pneumonia admitted to non-penalized hospitals, in association with the HRRP implementation. Findings did not differ by race or by socioeconomic position. Several of these findings merit further discussion.

The fact that the HRRP was not associated with changes in the use of SNF and home health care after hospitalization for AMI and HF could have several potential explanations. First, there may be a ceiling effect on the degree to which the HRRP could increase the use of post-acute care because hospital physicians can only increase referrals for patients with needs for skilled care; patients with basic functional limitations alone do not qualify. Second, it is possible that most patients with AMI or HF who needed skilled care were already being referred to SNF and home health care services prior to the HRRP. Third, due to the preceding factors, hospitals may have prioritized other strategies for reducing readmissions that were more under their control, such as improving discharge processes and post-hospitalization coordination of care.23 This hypothesis is strengthened by the finding that rates of being continuously in the community for 30 days post-discharge increased at hospitals that incurred penalties, despite stable use of post-acute care. Finally, instead of SNF and home health care, some patients with AMI and HF may have accessed condition-specific rehabilitation services that are available as stand-alone outpatient services or part of post-acute care. Cardiac rehabilitation programs, in particular, reduce readmissions and improve outcomes for patients with heart disease and are offered as both outpatient and post-acute services.16 As of 2015, 35.5%, of AMI survivors received outpatient rehabilitation.24 Earlier, in 2008, 19% of fee for service Medicare beneficiaries with AMI attended a cardiac rehabilitation program in the outpatient setting, whereas only 13% and 5% attended a program through home health and skilled nursing units, respectively.25

In contrast to AMI and HF, use of home health care for pneumonia declined during the HRRP anticipation and penalty periods relative to what was expected based on historical trends, while use of SNF services increased, suggesting a possible shift from home health care, which offers less intensive care, to SNF services, which include more intensive and diverse services. Although the HRRP could have prompted a shift from home health care to SNF care among patients with the requisite needs for skilled care, other secular trends during this period might have played a role as well. Community-acquired pneumonia is increasingly treated in the outpatient setting, resulting in higher severity of illness for patients hospitalized for pneumonia.26,27 Additionally, hospitalizations for sepsis, including sepsis from pneumonia, have been on the rise in recent years,28 and are associated with higher likelihood of SNF admission.28,29 At the same time, hospitals have experienced pressures to reduce the number of hospitalizations, and the average length of stay, which resulted in decreased hospitalization volumes and length of stay for pneumonia over the time period overlapping our study.30 Patients who are clinically stable but who have ongoing needs for intravenous antibiotics, respiratory therapy, or rehabilitation, particularly elderly patients with functional or cognitive impairments, are increasingly referred to SNF services.16 SNFs are better equipped to treat complex cases than HHA are. Indeed, evidence suggests that post-acute care may substitute for prolonged hospitalizations for seniors with pneumonia. Data from the annual National Hospital Discharge Survey have revealed that pneumonia and sepsis were the two diagnoses most commonly associated with decreasing length of stay and increasing discharges to post-acute care facilities (e.g., SNFs) between 1996 and 2010.27

The relationship (or lack thereof) between the HRRP and trends in post-acute care use for the study conditions should not be interpreted in isolation. Other policies announced or implemented during the study period may have exerted various pressures on utilization of post-acute care. For example, Medicare has implemented the Hospital Value-Based Purchasing Program, which penalizes hospitals on a variety of quality measures including worse-than-expected 30-day mortality and may incentivize hospitals to increase use of post-acute care services known to improve outcomes. On the other hand, Medicare oversaw a national demonstration project aimed at decreasing variation in post-acute care referral patterns concurrently with the HRRP anticipation period (2008–2011)31, which may have exerted opposite effects. Medicare Accountable Care Organizations receive financial incentives for improving the quality and lowering the cost of care for populations of patients, creating incentives for affiliated hospitals to rein in major sources of spending variation like post-acute care.32 In addition, the mandatory Medicare Bundled Payments for Care Improvement Initiative started test payment models for related episodes of care, such as a single payment for the hospitalization and post‐discharge period, in selected geographic areas in 2013.33 Although this program was implemented after the HRRP, hospitals may have taken anticipatory measures to use post-acute care referrals more judiciously.

As hospitals seek to reduce readmissions and avoid HRRP penalties, they will continue to have incentives to refer patients for post-acute care and to shift the most vulnerable patients from home health care to SNF-based care. However, other Medicare policies designed to contain costs provide countervailing pressures to limit how often hospitals use these strategies. Importantly, Medicare has announced the introduction of penalties for long-term care facilities, including SNFs, with higher than expected readmission rates in 2018,34 which may result in SNF facilities being more selective about which patients they admit. Strong linkages between hospitals and carefully selected SNFs, which have been shown to prevent re-hospitalizations for a wide range of medical and surgical conditions24,35,36,21 may be more useful to future efforts to ensure judicious use of services, rather than simply increasing referrals to post-acute care.

Supplementary Material

Supplemental Data File (.doc, .tif, pdf, etc.)_1
Supplemental Data File (.doc, .tif, pdf, etc.)_2

Acknowledgements:

Margaret Kelley, MSEd, Cedars-Sinai Medical Center; Jillian Wallis, PhD, University of Southern California; Brendan Rabideau, BS, University of Southern California contributed to the research effort.

Funding Sources: Dr. Nuckols was funded by a grant (R01 HS024284) from The Agency for Healthcare Research and Quality, Dr. Escarce and Dr. Nuckols were funded by a grant (R01 HS025394) from The Agency for Healthcare Research and Quality, and Dr. Sood was funded by a grant (R01AG046838) from The National Institute on Aging.

The funding organization had no involvement in any aspect of the study, including design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Footnotes

All other authors have no relevant conflicts of interest with the work.

REFERENCES

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Data File (.doc, .tif, pdf, etc.)_1
NIHMS1535628-supplement-Supplemental_Data_File___doc___tif__pdf__etc___1.docx (357.4KB, docx)
Supplemental Data File (.doc, .tif, pdf, etc.)_2
NIHMS1535628-supplement-Supplemental_Data_File___doc___tif__pdf__etc___2.docx (718.8KB, docx)

RESOURCES