Abstract
Objectives
Physical and cognitive conditions of patients discharged to skilled nursing facilities (SNF), inpatient rehabilitation facilities (IRF), and home with home health agencies (HHA) following total joint arthroplasty (TJA) have not been evaluated. The purpose of this study is to examine physical and cognitive function trends of Medicare beneficiaries discharged to SNFs, HHAs, and IRFs following TJA from 2013 to 2018.
Design
Observational study using Medicare enrollment, claims, and assessment data from 2013–2018.
Setting and Participants
1,278,939 Medicare beneficiaries discharged to SNF, HHA, or IRF for post-acute care following TJA from 2013–2018.
Methods
Medicare data was used to examine the association between the endpoints of interest (discharge destination [SNF, HHA, or IRF], and the physical [measured using activities of daily living] and cognitive [measured using a range of setting-specific metrics] status of patients in each setting) and the year of TJA (2013–2018) by estimating multivariable models that controlled for patient- and hospital-level covariates.
Results
On multivariable analysis of 1,278,939 TJAs, SNF discharge decreased (44.15%[2013] to 21.57%[2018], p<0.001), HHA increased (46.72% to 72.47%, p<0.001), and IRF decreased (9.13% to 5.69%, p<0.001). For SNF, mean physical function scores (14.61[2013] to 14.23[2018], p<0.001) and cognitive impairment (13.25% to 12.33%, p=0.01) decreased, indicating less dependence. Physical function scores (3.09 to 3.94, p<0.001) and cognitive impairment (13.95% to 16.52%, p<0.001) increased for HHA patients, indicating greater dependence. For IRF, motor functional independence measure decreased (38.81 to 37.78, p<0.001) and cognitive dependence increased (39.08% to 46.36%, p<0.001), indicating greater dependence.
Conclusions and Implications
From 2013–2018, patients were increasingly discharged to HHA. While SNF patients were less dependent over time, HHA and IRF patients were physically and cognitively more dependent. Each setting is likely to benefit from policy and fiscal supports that help them manage changes in the volume and clinical intensity of patients requiring their services.
Keywords: Total joint arthroplasty, Medicare, post-acute care
Brief Summary
From 2013–2018, patients were increasingly discharged to HHA after total joint arthroplasty. While SNF patients were less dependent over time, HHA and IRF patients were physically and cognitively more dependent.
Introduction
Lower extremity total joint arthroplasty (TJA) (specifically total hip and knee arthroplasty) is one of the most common and successful procedures for Medicare beneficiaries to improve their physical function, pain, and overall quality of life.1,2 Therefore, TJA remains the largest surgical expense for Centers for Medicare and Medicaid Services and a target for Medicare payment and delivery models to reduce cost while improving quality, safety and efficiency.1,2 Due to the expansion of bundled payment programs, such as the Comprehensive Care for Joint Replacement (CJR)3, and evidence of substantial variation in the use, outcomes, and spending for post-acute care4–6, a key target of cost reduction is post-acute care following TJA.2,7
Since introduction of Medicare bundled payments, several studies demonstrate decreasing proportion of patients being discharged to post-acute care facilities such as skilled nursing facilities (SNF) and inpatient rehabilitation facilities (IRF), favoring post-operative discharge to home with the care of home health care agencies (HHA).1,8 From 2013 to 2016, institutional post-acute care use was found to decrease by 11.2% for Medicare-only patients, 7.8% for dual-eligible patients with full benefits, and by 8.8% for dual-eligible patients with partial benefits.1 Furthermore, during the first 18 months of the mandatory Medicare bundled payment program, there was a 6.1% relative reduction in discharge to institutional post-acute care amongst traditional Medicare patients. 8 However, despite these trends, over 30% of patients continue to be discharged to post-acute care facilities following TJA.1,8 Although existing literature demonstrates the general post-acute care trends above, it does not evaluate the physical and cognitive conditions of patients in various post-acute care settings (SNF, HHA, IRF), which is critical to understand for several reasons. First, the physical and cognitive condition of patients may explain the increased risk of post-discharge adverse events for patients discharged to IRFs or SNFs.9 Second, changing trends in the volume and clinical complexity of patients discharged to post-acute care settings may justify the need for policy and fiscal support.
Hence, the purpose of this study is to examine trends in the physical and cognitive function of Medicare beneficiaries from 2013 to 2018 at the start of their post-acute care following TJA.
Methods
Data sources and study cohort
We used the 2013–2018 100% Medicare Master Beneficiary Summary File (MBSF) to identify fee-for-service Medicare beneficiaries, 65 years and older, and eligible for Medicare due to age (Appendix Exhibit 1A).10 We used the 2013–2018 100% Medicare Provider Analysis and Review (MedPAR) files to identify short, inpatient stays for non-Hispanic White, non-Hispanic Black, and Hispanic beneficiaries who underwent total hip or knee replacements, were discharged in the respective calendar year, and met the MBSF inclusion criteria listed above (Appendix Exhibit 1B).11 Specifically, we used Medicare Severity-Diagnosis Related Groups 469 and 470, and International Classification of Diseases, 9th and 10 Revision, Procedure Coding System (Appendix Exhibit 2) to identify the cohort.12
We applied additional exclusion criteria (Appendix Exhibit 1B) and linked these inpatient stays to the 100% Minimum Data Set (3.0) for SNF,13 cohort-specific Home Health Outcome and Assessment Information Set (OASIS) for HHA,14 and cohort-specific IRF – Patient Assessment Instrument (IRF-PAI)15 to identify post-acute care encounters that appeared in both the MedPAR files and assessment files, and were initiated within three days of discharge from the inpatient setting. Because the discharge destination in MedPAR files may be misclassified, we verified the post-acute care discharge destinations by determining whether the start of care in the three settings were within 3 days of the inpatient discharge date in the MedPAR data.16,17 This determination required us to use the following years data for patient’s discharged in the last three days of a year.
Our final cohort was 1,278,939 TJAs, where 483,990 were discharged to SNF, 708,885 to HHA, and 86,064 to IRF following TJA (Appendix Exhibits 1A and 1B). We used Medicare’s Impact Files18 and Hospital Compare19 data to obtain hospital characteristics.
Outcomes
We examined trends in the discharge destination (SNF, HHA, or IRF) by modeling these destinations in a categorical variable.
For patients discharged to SNFs, outcomes were (i) physical function - sum of scores for assistance needed while performing activities of daily living (ADLs) (bed mobility, transfer, locomotion, dressing, eating, toilet use, and personal hygiene) with higher scores representing higher degrees of dependence,20 (ii) Cognitive Function Scale21 which includes scores for Brief Interview of Mental Status22 and Cognitive Performance Scale, and was specified as a binary variable (Appendix Exhibit 3).23
For patients discharged to HHAs, outcomes were (i) physical function - a sum of scores for assistance needed for ADLs (grooming, dressing upper and lower body, toileting, bathing, transferring, ambulation, and eating) with higher scores representing higher degrees of dependence,20,24 (ii) binary indicator of cognitive function ranging from alert and oriented to mild to severe decline in cognitive function.25
For patients discharged to IRFs, outcomes were (i) functional independence measure (motor) - sum of four domains of scores, including assistance for self-care (eating, grooming, bathing, upper and lower body dressing, toileting), bowel and bladder care, transferring (bed/chair/wheelchair, toilet), and locomotion (walk/wheelchair). Note that for IRF physical function measures, lower scores represent dependence and higher scores represent independence, (ii) functional independence measure (cognition) as a sum of two domains of scores, including communication (comprehension and expression), and social cognition (social interaction, problem solving, memory)26,27 and specified as a binary variable.
Each of these endpoints were measured at the start of the post-acute care stay.
Key independent variable
The key independent variable was a categorical indicator for the year (2013 to 2018), which is the year of discharge for an inpatient TJA encounter
Covariates
We controlled for patient- and hospital-level covariates that were likely to confound the association of interest. Patient covariates included continuous specification of age, and categorical indicators for sex (male, female), race (non-Hispanic White, non-Hispanic Black, Hispanic), dual-eligibility for both Medicare and Medicaid (defined using the state-reported dual-eligibility status code; a marker for socioeconomic risk),28 admission type (emergent, urgent, or other), source of admission (from home or not), urgency of surgery identified using Medicare’s algorithm for identification of elective primary TJA (elective or not), hip or knee arthroplasty, and comorbidities identified using Elixhauser’s algorithm.29
Hospital-level covariates included continuous indicators for transfer-adjusted case mix index to account for medical complexity of patients, Disproportionate Share Hospital Patient Percentage to account for hospital volume of socially vulnerable patients, hospital TJA volume, and market competition defined using the Herfindahl-Hirschman Index. Hospital-level categorical indicators included hospital-ownership (government, private for-profit, private not-for-profit), number of beds (small [<200 beds], medium [>=200 beds and <400 beds], large [>=400 beds]), medical school affiliation, and urban/rural location.
Statistical analysis
For descriptive analysis, we report means (and standard deviations) for normally distributed continuous variables, medians (and interquartile ranges) for skewed continuous variables, and numbers (and percentages) for categorical variables in the baseline year (2013). We used Kruskal-Wallis and chi-square tests to determine whether distribution of key variables varied across post-acute care settings.
For multivariable analysis, we estimated multinomial logistic regression models with state-level clustering to determine adjusted trends in discharge destinations. We estimated hierarchical linear regression models (for continuous outcomes) and linear probability models (for binary outcomes) with state-level random effects to determine adjusted trends in physical and cognitive function of patients following TJA. We report adjusted means and p-values for tests of linear combination of estimates.
In additional analysis, we included all patients who underwent TJAs (and not only those that were discharged to SNF/HHA/IRF). This led to the inclusion of patients who were (i) discharged to home without home healthcare or (ii) did not meet the post-acute care inclusion criteria (i.e., admission to post-acute care setting could not be verified in both the MedPAR and post-acute care assessment files, and those that may have been admitted to a post-acute care setting after more than 3 days of inpatient discharge). We used this updated cohort to compute the percentage of patients discharged to various post-acute care settings.
Furthermore, we also determined the race/ethnicity-based trends in the physical and cognitive status of patients starting post-acute care by estimating multivariable multinomial logistic regression models with state-level clustering (similar to those as described above) and included an interaction between the discharge year and race/ethnicity of the patient (non-Hispanic White, non-Hispanic Black, and Hispanic) as the key estimate of interest.
Results
Descriptive Results from 2013
Of 233,091 TJAs, 102,505 (43.98%) were discharged to SNFs, 109,569 (47.01%) to HHA, and 21,017 (9.02) to IRFs (Table 1 and Table 2). Mean age was 74.72 years (standard deviation [SD]: 6.46 years), 65.82% were female, 93.57% were White, 7% were dually eligible for Medicare and Medicaid, and 93.22% were elective admissions. Of all encounters, 78,477 (33.67%) were for hip arthroplasty and 154,614 (66.33%) for knee arthroplasty.
Table 1:
Patient characteristics of patients in the baseline year (2013)
| SNF | HHA | IRF | Total | p-value | |
|---|---|---|---|---|---|
| N | |||||
| Encounters | 102,505 | 109,569 | 21,017 | 233,091 | |
| Hospitals | 2,686 | 2,585 | 1,719 | 2,907 | |
| Patient characteristics | |||||
| Age in years: Mean (SD) | 76.27 (6.67) | 72.88 (5.57) | 76.78 (7.10) | 74.72 (6.46) | <0.001 |
| Female: N (%) | 74,771 (72.94) | 64,086 (58.49) | 14,571 (69.33) | 153,428 (65.82) | <0.001 |
| Race: N (%) | <0.001 | ||||
| White | 94,910 (92.59) | 103,996 (94.91) | 19,194 (91.33) | 218,100 (93.57) | |
| Black | 6,491 (6.33) | 4,615 (4.21) | 1,493 (7.10) | 12,599 (5.41) | |
| Hispanic | 1,104 (1.08) | 958 (0.87) | 330 (1.57) | 2,392 (1.03) | |
| Dual-eligibility: N (%) | 9,361 (9.13) | 4,999 (4.56) | 1,953 (9.29) | 16,313 (7.00) | <0.001 |
| Admission type: N (%) | <0.001 | ||||
| Elective | 94,633 (92.32) | 103,906 (94.83) | 18,747 (89.20) | 217,286 (93.22) | |
| Emergent | 2,889 (2.82) | 718 (0.66) | 1,256 (5.98) | 4,863 (2.09) | |
| Urgent | 4,813 (4.70) | 4,851 (4.43) | 920 (4.38) | 10,584 (4.54) | |
| Other | 170 (0.17) | 94 (0.09) | 94 (0.45) | 358 (0.15) | |
| Home or non-healthcare source of admission: N (%) | 101,762 (99.28) | 109,324 (99.78) | 20,950 (99.68) | 232,036 (99.55) | <0.001 |
| Elective surgery: N (%) | 97,370 (94.99) | 107,846 (98.43) | 18,847 (89.68) | 224,063 (96.13) | <0.001 |
| Type of surgery: N (%) | <0.001 | ||||
| Hip arthroplasty | 35,086 (34.23) | 34,934 (31.88) | 8,457 (40.24) | 78,477 (33.67) | |
| Knee arthroplasty | 67,419 (65.77) | 74,635 (68.12) | 12,560 (59.76) | 154,614 (66.33) | |
| Elixhauser’s comorbidities: N (%) | |||||
| AIDS/HIV | § | § | § | 35 (0.02) | 0.48 |
| Alcohol Abuse | 941 (0.92) | 908 (0.83) | 196 (0.93) | 2,045 (0.88) | 0.06 |
| Blood Loss Anemia | 1,593 (1.55) | 1,176 (1.07) | 347 (1.65) | 3,116 (1.34) | <0.001 |
| Cardiac Arrhythmias | 19,483 (19.01) | 14,999 (13.69) | 4,506 (21.44) | 38,988 (16.73) | <0.001 |
| Chronic Pulmonary Disease | 18,792 (18.33) | 14,775 (13.48) | 3,978 (18.93) | 37,545 (16.11) | <0.001 |
| Coagulopathy | 3,295 (3.21) | 2,480 (2.26) | 887 (4.22) | 6,662 (2.86) | <0.001 |
| Congestive Heart Failure | 5,873 (5.73) | 2,862 (2.61) | 1,443 (6.87) | 10,178 (4.37) | <0.001 |
| Deficiency Anemia | 1,707 (1.67) | 1,216 (1.11) | 397 (1.89) | 3,320 (1.42) | <0.001 |
| Depression | 15,516 (15.14) | 11,656 (10.64) | 2,996 (14.26) | 30,168 (12.94) | <0.001 |
| Diabetes, Complicated | 2,234 (2.18) | 1,229 (1.12) | 601 (2.86) | 4,064 (1.74) | <0.001 |
| Diabetes, Uncomplicated | 22,279 (21.73) | 19,109 (17.44) | 4,924 (23.43) | 46,312 (19.87) | <0.001 |
| Drug Abuse | 1,168 (1.14) | 504 (0.46) | 226 (1.08) | 1,898 (0.81) | <0.001 |
| Fluid and Electrolyte Disorders | 13,486 (13.16) | 9,080 (8.29) | 3,029 (14.41) | 25,595 (10.98) | <0.001 |
| Hypertension, Complicated | 9,413 (9.18) | 5,995 (5.47) | 2,204 (10.49) | 17,612 (7.56) | <0.001 |
| Hypertension, Uncomplicated | 69,361 (67.67) | 71,977 (65.69) | 14,183 (67.48) | 155,521 (66.72) | <0.001 |
| Hypothyroidism | 22,728 (22.17) | 19,564 (17.86) | 4,769 (22.69) | 47,061 (20.19) | <0.001 |
| Liver Disease | 940 (0.92) | 848 (0.77) | 194 (0.92) | 1,982 (0.85) | 0.001 |
| Lymphoma | 445 (0.43) | 380 (0.35) | 108 (0.51) | 933 (0.40) | <0.001 |
| Metastatic Cancer | 174 (0.17) | 83 (0.08) | 81 (0.39) | 338 (0.15) | <0.001 |
| Obesity | 19,095 (18.63) | 16,496 (15.06) | 4,092 (19.47) | 39,683 (17.02) | <0.001 |
| Other Neurological Disorders | 3,291 (3.21) | 1,554 (1.42) | 1,062 (5.05) | 5,907 (2.53) | <0.001 |
| Paralysis | 127 (0.12) | 45 (0.04) | 91 (0.43) | 263 (0.11) | <0.001 |
| Peptic Ulcer Disease Excluding Bleeding | 630 (0.61) | 550 (0.50) | 140 (0.67) | 1,320 (0.57) | <0.001 |
| Peripheral Vascular Disorders | 3,755 (3.66) | 2,666 (2.43) | 962 (4.58) | 7,383 (3.17) | <0.001 |
| Psychoses | 1,104 (1.08) | 355 (0.32) | 205 (0.98) | 1,664 (0.71) | <0.001 |
| Pulmonary Circulation Disorders | 1,938 (1.89) | 975 (0.89) | 492 (2.34) | 3,405 (1.46) | <0.001 |
| Renal Failure | 9,375 (9.15) | 5,885 (5.37) | 2,185 (10.40) | 17,445 (7.48) | <0.001 |
| Rheumatoid Arthritis/Collagen Vascular | 5,521 (5.39) | 4,677 (4.27) | 1,328 (6.32) | 11,526 (4.94) | <0.001 |
| Solid Tumor Without Metastasis | 716 (0.70) | 644 (0.59) | 193 (0.92) | 1,553 (0.67) | <0.001 |
| Valvular Disease | 6,412 (6.26) | 4,740 (4.33) | 1,454 (6.92) | 12,606 (5.41) | <0.001 |
| Weight Loss | 709 (0.69) | 272 (0.25) | 209 (0.99) | 1,190 (0.51) | <0.001 |
Abbreviations:
SNF: Skilled nursing facility; HHA: Home health agency; IRF: Inpatient rehabilitation facility; N: Number; SD: Standard deviation; %: column percentage
Notes:
: Values masked in compliance with Medicare’s cell size policy.
Table 2:
Hospital characteristics of patients in the baseline year (2013)
| SNF | HHA | IRF | Total | p-value | |
|---|---|---|---|---|---|
| N | |||||
| Encounters | 102,505 | 109,569 | 21,017 | 233,091 | |
| Hospitals | 2,686 | 2,585 | 1,719 | 2,907 | |
| Hospital characteristics | |||||
| Ownership: N (%) | <0.001 | ||||
| Government | 11,221 (10.95) | 14,261 (13.02) | 2,209 (10.51) | 27,691 (11.88) | |
| Private for-profit | 15,266 (14.89) | 20,447 (18.66) | 4,398 (20.93) | 40,111 (17.21) | |
| Private not-for-profit | 76,018 (74.16) | 74,861 (68.32) | 14,410 (68.56) | 165,289 (70.91) | |
| Number of beds: N (%) | <0.001 | ||||
| Small (<200 beds) | 37,865 (36.94) | 42,336 (38.64) | 8,308 (39.53) | 88,509 (37.97) | |
| Medium (>=200 and <400 beds) | 35,279 (34.42) | 35,558 (32.45) | 6,848 (32.58) | 77,685 (33.33) | |
| Large (>=400 beds) | 29,361 (28.64) | 31,675 (28.91) | 5,861 (27.89) | 66,897 (28.70) | |
| Medical school affiliation: N (%) | 50,491 (49.26) | 49,392 (45.08) | 10,345 (49.22) | 110,228 (47.29) | <0.001 |
| Urban location: N (%) | 92,105 (89.85) | 97,726 (89.19) | 19,113 (90.94) | 208,944 (89.64) | <0.001 |
| Disproportionate patient percentage: Mean (SD) | 0.23 (0.12) | 0.23 (0.13) | 0.24 (0.15) | 0.23 (0.13) | <0.001 |
| Transfer-adjusted casemix index: Mean (SD) | 1.64 (0.26) | 1.68 (0.27) | 1.67 (0.28) | 1.66 (0.27) | <0.001 |
| Volume of hip/knee arthroplasty: Mean (SD) | 197 (108, 350) | 226 (124, 401) | 159 (87, 307) | 207 (113, 366) | <0.001 |
| Market competition: Mean (SD) | 0.51 (0.32) | 0.47 (0.33) | 0.48 (0.32) | 0.49 (0.32) | <0.001 |
Abbreviations:
SNF: Skilled nursing facility; HHA: Home health agency; IRF: Inpatient rehabilitation facility; N: Number; SD: Standard deviation; %: column percentage
Overall Discharge Destination Results
The unadjusted results are in Table 3.
Table 3:
Unadjusted outcomes from 2013 to 2018
| 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | Total | p-value | |
|---|---|---|---|---|---|---|---|---|
| Discharge destination | <0.001 | |||||||
| SNF | 102,505 (43.98) | 97,474 (43.26) | 92,191 (41.02) | 86,023 (37.85) | 72,819 (33.76) | 32,978 (21.58) | 483,990 (37.84) | |
| HHA | 109,569 (47.01) | 109,854 (48.75) | 116,842 (51.99) | 128,718 (56.63) | 133,055 (61.68) | 110,847 (72.55) | 708,885 (55.43) | |
| IRF | 21,017 (9.02) | 18,001 (7.99) | 15,688 (6.98) | 12,551 (5.52) | 9,845 (4.56) | 8,962 (5.87) | 86,064 (6.73) | |
| SNF | ||||||||
| ADL: Mean (SD) | 14.88 (4.15) | 14.92 (4.10) | 14.84 (4.11) | 14.79 (4.10) | 14.73 (4.08) | 14.87 (4.00) | 14.84 (4.10) | <0.001 |
| Cognitive function: N (%) | <0.001 | |||||||
| Cognitively intact | 89,018 (88.09) | 84,937 (88.54) | 80,307 (88.74) | 74,734 (88.65) | 63,198 (88.41) | 26,992 (87.66) | 419,186 (88.43) | |
| Mild to severe impairment | 12,030 (11.91) | 10,998 (11.46) | 10,194 (11.26) | 9,564 (11.35) | 8,286 (11.59) | 3,799 (12.34) | 54,871 (11.57) | |
| HHA | ||||||||
| ADL: Mean (SD) | 3.16 (1.01) | 3.29 (0.99) | 3.46 (0.99) | 3.70 (0.98) | 3.90 (0.94) | 4.05 (0.92) | 3.61 (1.02) | <0.001 |
| Cognitive function: N (%) | <0.001 | |||||||
| Cognitively intact | 96,099 (87.71) | 95,362 (86.81) | 100,861 (86.32) | 108,955 (84.65) | 112,465 (84.53) | 93,287 (84.16) | 607,029 (85.63) | |
| Mild to severe impairment | 13,470 (12.29) | 14,492 (13.19) | 15,981 (13.68) | 19,763 (15.35) | 20,590 (15.47) | 17,560 (15.84) | 101,856 (14.37) | |
| IRF | ||||||||
| ADL: Mean (SD) | 38.47 (8.54) | 38.04 (8.60) | 37.61 (8.65) | 37.18 (8.73) | 36.79 (8.71) | 36.56 (8.74) | 37.64 (8.67) | <0.001 |
| Cognitive dependence: N (%) | 7,837 (37.29) | 7,165 (39.80) | 6,300 (40.16) | 5,355 (42.67) | 4,366 (44.35) | 4,325 (48.26) | 35,348 (41.07) | <0.001 |
Abbreviations:
SNF: Skilled nursing facility; HHA: Home health agency; IRF: Inpatient rehabilitation facility; N: Number; SD: Standard deviation; %: Column percentage; ADL: Activities of daily living
Notes:
Definitions of the physical and cognitive function measures are in Appendix Exhibit 3.
On multivariable analysis, proportion of patients discharged to SNF more than halved from 44.15% (95% Confidence Interval [CI]: 40.85, 47.45) in 2013 to 21.57% (95% CI: 19.65, 23.49) in 2018 (p-value for comparison: p<0.001) and the proportion of patients discharged to IRFs also decreased from 9.13% (95% CI: 6.65, 11.61) in 2013 to 5.69% (95% CI: 3.83, 7.55) in 2018 (p<0.001) (Table 4, Full model estimates in Appendix Exhibit 4). In contrast, the proportion of patients discharged to HHA increased from 46.7% (95% CI: 43.91, 49.53) in 2013 to 72.74% (95% CI: 71.02, 74.46) in 2018 (p<0.001).
Table 4:
Adjusted percentages for trends in discharge destination from 2013–2018
| Year | SNF | HHA | IRF |
|---|---|---|---|
| N=1,277,175 | |||
| 2013 | 44.15 (40.85, 47.45) | 46.72 (43.91, 49.53) | 9.13 (6.65, 11.61) |
| 2014 | 43.37* (39.98, 46.76) | 48.57*** (45.94, 51.20) | 8.06*** (5.80, 10.32) |
| 2015 | 41.02*** (37.94, 44.10) | 51.95*** (49.60, 54.30) | 7.03*** (4.95, 9.11) |
| 2016 | 37.76*** (34.94, 40.58) | 56.79*** (54.73, 58.85) | 5.45*** (3.54, 7.36) |
| 2017 | 33.57*** (30.94, 36.20) | 61.90*** (59.93, 63.87) | 4.53*** (2.82, 6.23) |
| 2018 | 21.57*** (19.65, 23.49) | 72.74*** (71.02, 74.46) | 5.69*** (3.83, 7.55) |
p<0.05
p<0.001
Abbreviations:
SNF: Skilled nursing facility; HHA: Home health agency; IRF: Inpatient rehabilitation facility; N: Number
Notes:
Adjusted estimates from multivariable multinomial logistic regression models. Full model estimates are in Appendix Exhibit 4.
Patients Discharged to SNFs
The unadjusted physical and cognitive function scores for all three post-acute care settings are in Table 3.
On multivariable analysis, mean physical function score decreased from 14.61 (95% CI: 14.34, 14.89) in 2013 to 14.23 (95% CI: 13.94, 14.51) in 2018 (p<0.001), indicating patients discharged to SNFs in 2018 were significantly less physically dependent compared to SNF patients in 2013 (Table 5, Appendix Exhibit 5). The proportion of patients with mild to severe cognitive impairment discharged to SNFs decreased from 13.25% (95% CI: 12.56, 13.95) in 2013 to 12.33% (95% CI: 11.47, 13.20) in 2018 (p<0.01), indicating patients discharged to SNFs were also less cognitively impaired over time.
Table 5:
Adjusted means (for physical function) and percentages (for cognitive function) for patients discharged to skilled nursing facilities, home health agencies, and inpatient rehabilitation facilities following total joint arthroplasty from 2013–2018
| Year | SNF | HHA | IRF |
|---|---|---|---|
| Physical function | |||
| N | 468,709 | 707,934 | 79,233 |
| 2013 | 14.61 (14.34, 14.89) | 3.09 (3.01, 3.17) | 38.81 (38.09, 39.52) |
| 2014 | 14.61 (14.31, 14.90) | 3.21*** (3.13, 3.29) | 38.65 (37.99, 39.32) |
| 2015 | 14.46*** (14.17, 14.75) | 3.38*** (3.30, 3.45) | 38.35* (37.66, 39.05) |
| 2016 | 14.32*** (14.03, 14.61) | 3.61*** (3.53, 3.69) | 38.04*** (37.40, 38.68) |
| 2017 | 14.19*** (13.91, 14.48) | 3.81*** (3.73, 3.88) | 37.74*** (37.06, 38.43) |
| 2018 | 14.23*** (13.94, 14.51) | 3.94*** (3.86, 4.01) | 37.78*** (37.05, 38.51) |
| Cognitive function | |||
| N | 473,422 | 707,934 | 85,895 |
| 2013 | 13.25 (12.56, 13.95) | 13.95 (12.40, 15.50) | 39.08 (36.00, 42.15) |
| 2014 | 12.72** (11.97, 13.46) | 14.80*** (13.18, 16.41) | 40.73* (37.80, 43.65) |
| 2015 | 12.42*** (11.73, 13.11) | 15.15*** (13.65, 16.65) | 40.51 (37.51, 43.50) |
| 2016 | 12.18*** (11.43, 12.92) | 16.66*** (15.07, 18.25) | 42.20* (38.93, 45.47) |
| 2017 | 12.15*** (11.40, 12.90) | 16.58*** (14.83, 18.32) | 43.32*** (40.40, 46.24) |
| 2018 | 12.33** (11.47, 13.20) | 16.52*** (14.79, 18.24) | 46.36*** (42.91, 49.82) |
p<0.05
p<0.01
p<0.001
Abbreviations:
SNF: Skilled nursing facility; HHA: Home health agency; IRF: Inpatient rehabilitation facility; N: Number
Notes:
Adjusted estimates from multivariable linear (for physical function) and linear probability (for cognitive function) regression models. Full model estimates are in Appendix Exhibit 5 (home health), Appendix Exhibit 6 (skilled nursing facility), and Appendix Exhibit 7 (inpatient rehabilitation facility).
Patients Discharged to HHAs
On multivariable analysis, mean physical function score increased from 3.09 (95% CI: 3.01, 3.17) in 2013 to 3.94 (95% CI: 3.86, 4.01) in 2018 (p<0.001), indicating patients discharged to HHA were more physically dependent over time (Table 5, Appendix Exhibit 6). The proportion of patients with mild to severe cognitive impairment also increased from 13.95% (95% CI [12.40, 15.50]) in 2013 to 16.52% (95% CI [14.79, 18.24]) by 2018 (p<0.001), indicating more patients with cognitive impairment being discharged to HHA over time.
Patients Discharged to IRFs
On multivariable analysis, mean motor functional independence measure decreased from 38.81 (95% CI: 38.09, 39.52]) in 2013 to 37.78 (95% CI: 37.05, 38.51) in 2018 (p<0.001), indicating greater physical dependence in 2018 compared to 2013 (Table 5, Appendix Exhibit 7). The proportion of patients with cognitive dependence discharged to IRFs also increased from 39.08% (95% CI: 36.00, 42.15) in 2013 to 46.36% (95% CI: 42.91, 49.82) in 2018 (p< 0.001), indicating more patients with cognitive impairment were discharged to IRF over time.
Additional analysis
On inclusion of all TJA patients (and not limiting to those discharged to SNF/HHA/IRF), the percentage of patients discharged to SNFs and IRFs decreased by a third, the percentage of patients discharged to HHA remained stable, and the percentage of patients that underwent TJRs but did not meet the inclusion criteria for discharge to SNF/HHA/IRF doubled over the study period (Appendix Exhibit 8). In the race/ethnicity-based analysis, the mean physical function for Hispanic patients discharged to HHA decreased by 0.21 points (95% CI: −0.28 to 0.13; p<0.001) compared to White patients indicating that Hispanic patients discharged to HHA were less dependent than White patients in 2018 (Appendix Exhibit 9). The percentage of Black patients with cognitive impairment discharged to SNFs increased by 2.10 percentage-points (95% CI: 0.87 to 3.33, p<0.001) and to IRFs increased by 4.62 percentage-points (95% CI: 0.35 to 8.89, p=0.03) compared to White patients indicating that more cognitively dependent Black patients were discharged to SNFs and IRFs compared to White patients in 2018.
Discussion
TJA is one of the largest expenses for the Centers for Medicare and Medicaid Services. Therefore, there is a push to lower costs by targeting post-acute care following TJA, which has led to a larger proportion of patients being discharged home rather than post-acute care facilities. Although this trend is well established, existing literature does not evaluate the physical and/or cognitive conditions of patients being discharged to various post-acute care settings (HHA, IRFs, SNFs) and how those conditions have changed over time. Therefore, we performed this study to examine trends in the physical and cognitive function of Medicare beneficiaries at the start of their post-acute care following TJA. Our results confirm that from 2013 to 2018, the majority of patients were discharged to HHA after TJA, while discharges to SNF and IRF drastically decreased. Notably, patients discharged to HHA were more physically dependent and cognitively impaired in 2018 compared to 2013 counterparts. On the contrary, patients discharged to SNF were less physically dependent and less cognitively impaired as a whole. Patients discharged to IRF continued to be more physically and cognitively dependent in 2018 compared to 2013.
Our findings of increase in HHA discharges and decrease is SNF/IRF discharges are consistent with existing literature. Li et al.1 demonstrated that among Medicare patients, institutional post-acute care use decreased from 43.7% in 2013 to 32.5% by 2016. Burnett et al.30 also analyzed over 79,000 TKAs from 2007–2016 in the Humana administrative claims database and determined that patients in 2016 had 26% greater odds of being discharged home compared to 2007 and were also less likely to be discharged to SNF or IRF. A driving force for this shift in post-acute care trends may be cost-saving initiatives such as the CJR and the Bundled Payments for Care Improvement (BPCI), where one of the avenues to increase margin and overall profit is to reduce the cost of post-operative care since the cost of discharge to SNF or IRF is substantially greater than HHA.2,7,8,30–32 Notably, Serino et al.2 determined that compared to 2010, before the initiation of bundled payment programs, costs of both the post-acute period and the overall episode for TJA were significantly lower in 2018.
For patients discharged to SNF following TJA, patients in 2018 were significantly less physically dependent (i.e., more independent). As for cognitive function, patients discharged to SNF were also significantly less cognitively impaired in 2018 compared to 2013 counterparts. These results were surprising because patients who require additional physical assistance or are cognitively impaired and cannot safely function independently are generally appropriate for SNF discharge. Our results instead indicate that in a climate where SNF discharge is not preferred unless absolutely required, patients discharged to SNF in 2018 were both physically and cognitively more independent than their 2013 counterparts. These results suggest there may be additional clinical or socioeconomic determinants confounding this finding. Since we demonstrated that SNF discharge patients were significantly more cognitively impaired before adjustment for covariates, patient and/or hospital-based covariates may explain the variation in the association between cognitive impairment and SNF discharge. Although not a focus of this study, further multivariable analysis is necessary to determine if there are any patient factors such as race and/or Medicare dual-eligibility status or hospital-based factors such as TJA volume that contribute to decisions to discharge patients to SNF after TJA.
For patients discharged home with HHA following TJA, patients in 2018 were both significantly more physically dependent (i.e. less independent) and more cognitively impaired compared to 2013. A potential explanation for this phenomenon is due to the majority of patients being discharged home with HHA, a large subset of patients that previously were being discharged to SNF or IRF due to requiring mild to moderate physical assistance or had mild to moderate cognitive impairment are likely now discharged home. As a result of these patients now being discharged to HHAs, patients in 2018 were collectively more physically dependent and more cognitively impaired than their 2013 counterparts, where greater than 50% of patients were discharged to SNF or IRF. This finding has important implications. Due to the fact that over 70% of TJA patients are being discharged to HHA as of 2018 and the medical complexity within this cohort has substantially increased as a whole, it is imperative to determine if the current availability of home health care is able to meet the demands of the growing volume of complex patients. A study previously did determine that between 2002–2003 and 2014–2015, there was a relative increase of 2.84 in the number of Medicare certified home health care agencies per 100,000 people aged at least 18 years33, but the existing body of literature does not evaluate if there have been any significant recent changes in post-discharge adverse events or readmissions for TJA patients discharged to HHA. Previous literature supports increased risk of post-discharge adverse events for patients discharged to SNF or IRF compared to home, but this research involves patients from 2011–2013, which does not reflect the current conditions of patients being discharged to HHA.9 In fact, when Keswani et al.9 analyzed post-discharge outcomes after TJA, poor patient functional status was a key risk factor contributing to severe adverse events after discharge and for 30-day readmission. Therefore, due to the findings above, efforts to continue to increase resources and funding for HHAs are paramount to ensure discharge home remains both cost-effective and safe for patients.
Lastly, for patients discharged to IRF following TJA, patients in 2018 were significantly more physically and cognitively dependent or impaired compared to 2013. Since IRF is the most intensive and costly setting for post-acute care discharge following TJA, approximately 81% more costly compared to SNF and 230% more costly compared to home health per patient based on figures from 20182, it is appropriate that the most dependent and cognitively impaired patients, who require the most assistance and overall resource utilization, are being selectively discharged to IRF. This finding has important implications because it may further explain why patients discharged to IRF have an increased incidence of adverse events. According to Keswani et al.9, patients discharged to IRF are most dependent, have multiple significant comorbidities, and exhibit the greatest proportion of severe and minor adverse events both before and after discharge.
There are several limitations to this study. First, although we use granular claims and assessment data that are national in scope, it is likely we have not accounted for all unmeasured confounding. For example, clinical measures such as severity of osteoarthritis or social measures such as availability of caregivers may determine a patient’s discharge to post-acute care settings. Nevertheless, our national analysis of Medicare beneficiaries undergoing TJA fills an important gap in our understanding of the physical and cognitive status of patients during post-acute care. Second, the physical and cognitive measures are not directly comparable between settings. The ongoing work on standardization of these measures34 will likely address these concerns in the future. Third, our study focused on fee-for-service Medicare beneficiaries. Hence, our findings may have limited generalizability to Medicare Advantage beneficiaries and younger patients undergoing TJAs. Finally, we limited our study to non-Hispanic White, non-Hispanic Black, and Hispanic beneficiaries to limit the heterogeneity in key factors such as race/ethnicity in the cohort. Because this criterion excluded less than 3% of the study cohort, we expect our findings to be generalizable to most Medicare beneficiaries. However, a detailed analysis of social risk factors such as race, ethnicity, and dual-eligibility is beyond the scope of our study. We have shown that Black, Hispanic, and dual-eligible patients were over-represented in SNFs and IRFs. Therefore, future studies are required to evaluate how the trends vary by these key social factors and how those factors may influence post-acute care following TJA.
Conclusions and Implications
Compared to 2013, an overwhelming majority of patients in 2018 are discharged to HHA rather than SNF or IRF likely due to the implementation of several cost-saving Medicare initiatives. As a consequence, patients discharged to HHA in 2018 were overall more physically and cognitively impaired compared to their 2013 counterparts, while patients discharged to SNFs are less physically and cognitively dependent in 2018. Patients discharged to IRFs continue to increase in their physical and cognitive dependence. These findings highlight important changes in the use and selection of patients for each of the post-acute care settings. Further research is required to understand the implications of these changes, and additional policy and fiscal supports may be needed for these settings to provide high-quality and efficient care for the changing clinical profile of their patients.
Supplementary Material
Acknowledgements
The funding agencies (National Institute on Minority Health and Health Disparities (R01MD018573) and National Institute on Aging (R21AG07855)) had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Sponsor Role
The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest
NA does not report any conflicts of interest. MY reports receiving a research grant from the Department of Defense. CPT reports receiving speaker honoraria from the National Institutes of Health, Brown University and Boston University; reviewer honoraria from the National Institutes of Health and the Veterans Affairs; and serves as the Disparities Section Editor for Current Osteoporosis Reports, and Anesthesia and Analgesia. BFR receives research funding outside this work from Johnson and Johnson and the National Institutes of Health; serves as an editorial board member of Clinical Orthopaedics and Related Research, HSS Journal, The Knee, and Arthroplasty today; serves as a committee member for the American Academy of Orthopaedic Surgeons and the American Association of Hip and Knee Surgeons.
References
- [1].Li Y, Ying M, Cai X, Kim Y, Thirukumaran CP. Trends in postacute care use and outcomes after hip and knee replacements in dual-eligible Medicare and Medicaid beneficiaries, 2013–2016. JAMA Netw Open 2020;3(3):e200368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Serino J, Burnett RA, Della Valle CJ, Courtney PM. National trends in post-acute care costs following total hip arthroplasty from 2010 through 2018. J Bone Joint Surg Am 2022;104(3):255–264. [DOI] [PubMed] [Google Scholar]
- [3].Comprehensive Care for Joint Replacement Model. Centers for Medicare & Medicaid Services. Updated May 31, 2023. Accessed August 25, 2023. https://innovation.cms.gov/innovation-models/cjr
- [4].Mechanic R Post-acute care--the next frontier for controlling Medicare spending. N Engl J Med 2014;370(8):692–694. [DOI] [PubMed] [Google Scholar]
- [5].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(9):1505–1513. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].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(1):91–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Meath THA, Juarez C, McConnell KJ, Kim H. Hospital characteristics associated with heterogeneity in institutional postacute care spending reductions under the comprehensive care for joint replacement model. JAMA Health Forum 2022;3(6):e221657. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Meyers DJ, Kosar CM, Rahman M, Mor V, Trivedi AN. Association of mandatory bundled payments for joint replacement with use of postacute care among medicare advantage enrollees. JAMA Netw Open 2019;2(12):e1918535. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Keswani A, Tasi MC, Fields A, Lovy AJ, Moucha CS, Bozic KJ. Discharge destination after total joint arthroplasty: an analysis of postdischarge outcomes, placement risk factors, and recent trends. J Arthroplasty 2016;31(6):1155–1162. [DOI] [PubMed] [Google Scholar]
- [10].Research Data Assistance Center (ResDAC). Master Beneficiary Summary Files - Base Segment (online). Available at: https://www.resdac.org/cms-data/files/mbsf-base. Accessed May 1, 2022.
- [11].Research Data Assistance Center (ResDAC). MedPAR RIF (online). Available at: https://www.resdac.org/cms-data/files/medpar. Accessed May 1, 2022.
- [12].Centers for Medicare and Medicaid Services. 2019. Procedure-Specific Complication Measure Updates and Specifications: THA/TKA (online). Available at: https://www.qualitynet.org/dcs/ContentServer?c=Page&pagename=QnetPublic%2FPage%2FQnetTier4&cid=1228772782693. Accessed May 1, 2022.
- [13].Research Data Assistance Center (ResDAC). Long Term Care Minimum Data Set 3.0 (online). Available at: https://resdac.org/cms-data/files/mds-30. Accessed May 1, 2022.
- [14].Research Data Assistance Center (ResDAC). Home Health Outcome and Assessment Information Set (OASIS) (online). Available at: https://resdac.org/cms-data/files/oasis. Accessed May 1, 2022.
- [15].Research Data Assistance Center (ResDAC). Inpatient Rehab Facility - Patient Assessment Instrument (IRF-PAI) (online). Available at: https://resdac.org/cms-data/files/irf-pai. Accessed May 1, 2022.
- [16].Kim H, Grunditz JI, Meath THA, Quiñones AR, Ibrahim SA, McConnell KJ. Accuracy of hospital discharge codes in medicare claims for knee and hip replacement patients. Med Care 2020;58(5):491–495. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Kahn JM, Iwashyna TJ. Accuracy of the discharge destination field in administrative data for identifying transfer to a long-term acute care hospital. BMC Res Notes 2010;3:205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Centers for Medicare and Medicaid Services. Acute Inpatient Prospective Payment System Impact Files (online). Available at: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Historical-Impact-Files-for-FY-1994-through-Present. Accessed May 1, 2022.
- [19].Centers for Medicare and Medicaid Services. Hospital Compare (online). Available at: https://www.medicare.gov/hospitalcompare/search.html. Accessed May 1, 2022.
- [20].Osakwe ZT, Larson E, Agrawal M, Shang J. Assessment of activity of daily living among older adult patients in home healthcare and skilled nursing facilities: an integrative review. Home Healthc Now 2017;35(5):258–267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Thomas KS, Dosa D, Wysocki A, Mor V. The minimum data set 3.0 cognitive function scale. Med Care 2017;55(9):e68–e72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Chodosh J, Edelen MO, Buchanan JL et al. Nursing home assessment of cognitive impairment: development and testing of a brief instrument of mental status. J Am Geriatr Soc 2008;56(11):2069–2075. [DOI] [PubMed] [Google Scholar]
- [23].Morris JN, Fries BE, Mehr DR et al. MDS cognitive performance scale. J Gerontol. 1994;49(4):M174–82. [DOI] [PubMed] [Google Scholar]
- [24].Scharpf TP, Madigan EA. Functional status outcome measures in home health care patients with heart failure. Home Health Care Serv Q 2010;29(4):155–170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Tullai-McGuinness S, Madigan EA, Fortinsky RH. Validity testing the outcomes and assessment information set (OASIS). Home Health Care Serv Q 2009;28(1):45–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Cao Y, Nie J, Sisto SA, Niewczyk P, Noyes K. Assessment of differences in inpatient rehabilitation services for length of stay and health outcomes between US Medicare advantage and traditional Medicare beneficiaries. JAMA network open 2020;3(3):e201204–e201204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Ottenbacher KJ, Smith PM, Illig SB, Linn RT, Ostir GV, Granger CV. Trends in length of stay, living setting, functional outcome, and mortality following medical rehabilitation. JAMA 2004;292(14):1687–1695. [DOI] [PubMed] [Google Scholar]
- [28].Centers for Medicare and Medicaid Services. Beneficiaries dually eligible for Medicare and Medicaid (online). Available at: https://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNProducts/downloads/Medicare_Beneficiaries_Dual_Eligibles_At_a_Glance.pdf. Accessed May 1, 2022.
- [29].Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36(1):8. [DOI] [PubMed] [Google Scholar]
- [30].Burnett RA, Serino J, Yang J, Della Valle CJ, Courtney PM. National trends in post-acute care costs following total knee arthroplasty from 2007 to 2016. J Arthroplasty 2021;36(7):2268–2275. [DOI] [PubMed] [Google Scholar]
- [31].Fisher ES. Medicare’s bundled payment program for joint replacement: promise and peril? JAMA 2016;316:1262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].McWilliams JM, Hatfield LA, Landon BE, Hamed P, Chernew ME. Medicare spending after 3 years of the Medicare shared savings program. N Engl J Med 2018;379(12):1139–1149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Wang Y, Leifheit-Limson EC, Fine J, Pandolfi MM, et al. National trends and geographic variation in availability of home health care: 2002–2015. J Am Geriatr Soc 2017;65(7):1434–1440. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Chen EK, Edelen MO, McMullen T, et al. Developing standardized patient assessment data elements for Medicare post-acute care assessments. J Am Geriatr Soc 2022;70(4):981–9. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.