High Risk of Readmission After THA Regardless of Functional Status in Patients Discharged to Skilled Nursing Facility

Abstract

Background

The postoperative period and subsequent discharge planning are critical in our continued efforts to decrease the risk of complications after THA. Patients discharged to skilled nursing facilities (SNFs) have consistently exhibited higher readmission rates compared with those discharged to home healthcare. This elevated risk has been attributed to several factors but whether readmission is associated with patient functional status is not known.

Questions/purposes

After controlling for relevant confounding variables (functional status, age, gender, caregiver support available at home, diagnosis [osteoarthritis (OA) versus non-OA], Charlson comorbidity index [CCI], the Area Deprivation Index [ADI], and insurance), are the odds of 30- and 90-day hospital readmission greater among patients initially discharged to SNFs than among those treated with home healthcare after THA?

Methods

This was a retrospective, comparative study of patients undergoing THA at any of 11 hospitals in a single, large, academic healthcare system between 2017 and 2022 who were discharged to an SNF or home healthcare. During this period, 13,262 patients were included. Patients discharged to SNFs were older (73 ± 11 years versus 65 ± 11 years; p < 0.001), less independent at hospital discharge (6-click score: 16 ± 3.2 versus 22 ± 2.3; p < 0.001), more were women (71% [1279 of 1796] versus 56% [6447 of 11,466]; p < 0.001), insured by Medicare (83% [1497 of 1796] versus 52% [5974 of 11,466]; p < 0.001), living in areas with greater deprivation (30% [533 of 1796] versus 19% [2229 of 11,466]; p < 0.001), and had less assistance available from at-home caregivers (29% [527 of 1796] versus 57% [6484 of 11,466]; p < 0.001). The primary outcomes assessed in this study were 30- and 90-day hospital readmissions. Although the system automatically flags readmissions occurring within 90 days at the various facilities in the overall healthcare system, readmissions occurring outside the system would not be captured. Therefore, we were not able to account for potential differential rates of readmission to external healthcare systems between the groups. However, given the large size and broad geographic coverage of the healthcare system analyzed, we expect the readmissions data captured to be representative of the study population. The focus on a single healthcare system also ensures consistency in readmission identification and reporting across subjects. We evaluated the association between discharge disposition (home healthcare versus SNF) and readmission. Covariates evaluated included age, gender, primary payer, primary diagnosis, CCI, ADI, the availability of at-home caregivers for the patient, and the Activity Measure for Post-Acute Care (AM-PAC) 6-clicks basic mobility score in the hospital. The adjusted relative risk (ARR) of readmission within 30 and 90 days of discharge to SNF (versus home healthcare) was estimated using modified Poisson regression models.

Results

After adjusting for the 6-clicks mobility score, age, gender, ADI, OA versus non-OA, living environment, CCI, and insurance, patients discharged to an SNF were more likely to be readmitted within 30 and 90 days compared with home healthcare after THA (ARR 1.46 [95% CI 1.01 to 2.13]; p= 0.046 and ARR 1.57 [95% CI 1.23 to 2.01]; p < 0.001, respectively).

Conclusion

Patients discharged to SNFs after THA had a slightly higher likelihood of hospital readmission within 30 and 90 days compared with those discharged with home healthcare. This difference persisted even after adjusting for relevant factors like functional status, home support, and social determinants of health. These results indicate that for suitable patients, direct home discharge may be a safer and more cost-effective option than SNFs. Clinicians should carefully consider these risks and benefits when making postoperative discharge plans. Policymakers could consider incentives and reforms to improve care transitions and coordination across settings. Further research using robust methods is needed to clarify the reasons for higher SNF readmission rates. Detailed analysis of patient complexity, care processes, and causes of readmission in SNFs versus home health could identify areas for quality improvement. Prospective cohorts or randomized trials would allow stronger conclusions about cause-and-effect. Importantly, no patients should be unfairly “cherry-picked” or “lemon-dropped” based only on readmission risk scores. With proper support and care coordination, even complex patients can have good outcomes. The goal should be providing excellent rehabilitation for all, while continuously improving quality, safety, and value across settings.

Level of Evidence

Level III, therapeutic study.

Introduction

THA is a common surgical procedure, with more than 370,000 THAs performed in the United States in 2014; volumes are expected to increase substantially [30, 32, 33]. Proper discharge planning and selection of postacute care (PAC) facilities is critical to improving outcomes and reducing risks of complications and readmission after THA [7, 23, 31, 34]. Patients discharged to skilled nursing facilities (SNFs) and inpatient rehabilitation facilities have consistently exhibited higher complication rates compared with those discharged home with home healthcare [6, 19, 26, 27, 28]. This elevated risk has been attributed to factors such as increased exposure to healthcare-associated infections, variability in PAC facility care quality, and a higher comorbidity burden in patients requiring institutional PAC [11, 13, 18, 22, 24, 26, 36].

Patients discharged to SNFs often have medical complexity and functional impairment, needing close monitoring, skilled nursing, physician contact, and rehabilitation to promote recovery after THA [2, 8, 9, 25, 26]. However, despite these efforts, discharge to SNFs has been associated with increased readmission risk after THA [3, 18, 26]. Although prior studies have accounted for factors like age, gender, race, comorbidities, and surgical complications [3, 18, 26], to our knowledge, none have adequately controlled for functional status, availability of caregiver support at home, and social determinants of health (SDOH) when examining this relationship. Assessing functional status has major implications for outcomes and costs [23, 31], and the mandates of the Improving Medicare Post-Acute Care Transformation Act of 2014 (IMPACT Act) underscore the need to understand this association. Clarifying the relationship between discharge disposition and readmission risk, while accounting for functional status and other key factors, can inform discharge planning and interventions to improve recovery and reduce readmissions after THA.

We therefore asked: After controlling for relevant confounding variables (functional status, age, gender, caregiver support available at home, diagnosis [osteoarthritis (OA) versus non-OA], Charlson comorbidity index [CCI], the Area Deprivation Index [ADI], and insurance), are the odds of 30- and 90-day hospital readmission greater among patients initially discharged to SNFs than among those treated with home healthcare after THA?

Patients and Methods

Study Design and Setting

The present study was a retrospective cohort analysis of patients who underwent THA at any of the 11 hospitals within a large, single academic healthcare system. Demographic and clinical variables were extracted from each patient episode using the electronic health record (EHR) system.

Patient Population

Patients who underwent primary elective THA (CPT procedure code 27130) between January 2, 2017, and August 31, 2022, were identified for inclusion in the study. Exclusions were made for individuals who underwent bilateral or revision procedures, those with a discharge diagnosis for a nonorthopaedic condition, and those discharged from the hospital to an inpatient rehabilitation facility, other PAC facilities, or home without additional healthcare services. During this period, 16,063 patients underwent THA. After applying exclusion criteria, the final cohort for analysis was 13,262 patients (Fig. 1). Overall, 86% (11,466) of patients were discharged to home healthcare, and 14% (1796) of patients were discharged to SNF.

Fig. 1.

This flow chart shows the patient selection for this study. SNFs = skilled nursing facilities.

Patient Baseline Demographics

Patients discharged to SNFs were older (73 ± 11 years versus 65 ± 11 years; p < 0.001), less independent at hospital discharge (6-click score: 16 ± 3.2 versus 22 ± 2.3; p < 0.001), more were women (71% [1279 of 1796] versus 56% [6447 of 11,466]; p < 0.001), insured by Medicare (83% [1497 of 1796] versus 52% [5974 of 11,466]; p < 0.001), living in areas with greater deprivation (30% [533 of 1796] versus 19% [2229 of 11,466]; p < 0.001), and had less assistance available from at-home caregivers (29% [527 of 1796] versus 57% [6484 of 11,466]; p < 0.001) (Table 1).

Table 1.

Study patients baseline characteristics

	Home healthcare (n = 11,466)	SNF (n = 1796)	p value
Age in years	65 ± 11	73 ± 11	< 0.001
Women	56 (6447)	71 (1279)	< 0.001
Primary payer			< 0.001
Medicare	52 (5974)	83 (1497)
Commercial	39 (4427)	9 (155)
Medicaid	7 (854)	8 (136)
Other/self-pay	2 (211)	0.4 (8)
Primary diagnosis			< 0.001
Osteoarthritis	93 (10,612)	82 (1466)
Non-OA: necrosis	5 (529)	5 (90)
Non-OA: fracture or other trauma	2 (253)	13 (225)
Non-OA: other bone disease	1 (72)	1 (15)
Charlson comorbidity index	3.0 (1.0-6.0)	3.5 (1.0-6.0)	< 0.001
ADI national rank: quintile			< 0.001
1st (1-20): least disadvantaged	8 (962)	3 (52)
2nd (21-40)	22 (2518)	16 (289)
3rd (41-60)	25 (2911)	24 (423)
4th (61-80)	22 (2536)	25 (441)
5th (81-100)	19 (2229)	30 (533)
Missing	3 (310)	3 (58)
Assistance available at home			< 0.001
24 hours	57 (6484)	29 (527)
Occasionally	41 (4696)	54 (972)
Never	1 (148)	13 (232)
Not documented	1 (138)	4 (65)
Last 6-clicks mobility score	22 ± 2.3	16 ± 3.2	< 0.001

Data presented as mean ± SD or % (n).

Primary Study Outcomes

The primary outcomes assessed in this study were 30- and 90-day hospital readmissions within the 11 hospitals comprising our health system. Our primary study goal was to evaluate the association between discharge disposition (home healthcare versus SNF) and readmission after controlling for relevant confounding variables. Although the system automatically flags readmissions occurring within 90 days at the various facilities in the overall healthcare system, readmissions occurring outside the system would not be captured. Therefore, we were not able to account for potential differential rates of readmission to external healthcare systems between the groups. However, given the large size and broad geographic coverage of the healthcare system analyzed, we expect the readmissions data captured to be representative for the study population. The focus on a single healthcare system also ensures consistency in readmission identification and reporting across subjects. Covariates evaluated included age, gender, primary payer (Medicare, Medicaid, commercial, or other/self-pay), primary diagnosis (osteoarthritis [OA] versus non-OA), CCI [29], and ADI [11, 13, 20]. The ADI aggregates 17 specific indicators (such as, poverty, housing, education, employment) into an index of disadvantage, accessible through the Neighborhood Atlas. The ADI is scored from 1 to 100, with higher scores suggesting increased socioeconomic disadvantage. Following the methodology employed by Hu et al. [16], the national percentile rankings of the ADI were categorized into quintiles. Additional covariates documented in the EHR by in-hospital physical therapists were incorporated, such as the availability of at-home caregivers for the patient (24 hours, occasional, or never) and the last-recorded score on the Activity Measure for Post-Acute Care (AM-PAC) 6-clicks basic mobility short form, which has been validated for use with patients in the acute hospital setting [1, 14, 17]. This measure assesses six fundamental mobility tasks, including bed mobility, transfers to a chair, standing, walking, and negotiating stairs, with each task scored on a 4-point Likert scale based on the level of assistance required for the patient to complete the task. Total raw scores range from 6 to 24, with higher scores indicating greater independence.

Ethical Approval

Ethical approval for this study was obtained from the Cleveland Clinic Institutional Review Board (#19-1656).

Statistical Analyses

For the exploratory analysis, we first compared all baseline characteristics between the SNF and home healthcare groups using t-tests, Mann-Whitney tests, chi-square tests, and Fisher exact tests as appropriate based on variable type and distribution. Variables with statistically significant differences between groups (p < 0.05) were considered as potential confounders. The unadjusted analysis found higher 30- and 90-day readmission rates in the SNF group (Supplemental Table 1; http://links.lww.com/CORR/B269), but the likelihood of confounding called for an adjusted analysis. Based on clinical judgement, previous evidence, and availability in our dataset, we selected age, gender, diagnosis, insurance status, 6-clicks mobility score, ADI, CCI, and living environment for inclusion in the multivariable regression models. We believed these variables addressed key differences between the groups that could influence readmission risk. The adjusted relative risk (ARR) of readmission within 30 and 90 days of hospital discharge was estimated separately for the SNF and home healthcare groups using modified Poisson regression with robust variance estimation [5, 10, 35]. We nested episodes within hospitals to account for potential clustering effects. The home healthcare group served as the reference category. All analyses were performed using Stata version 17 (Stata Corp). P values less than 0.05 were considered statistically significant. We focused on the adjusted models rather than unadjusted comparisons to minimize issues of confounding between the groups.

Results

After adjusting for 6-clicks mobility score, age, gender, OA vs non-OA, living environment, CCI, ADI, and insurance, patients discharged to an SNF had a higher odds of readmission within 30 days after THA compared with those discharging to home healthcare (ARR 1.46 [95% CI 1.01 to 2.13 ]; p = 0.046) (Table 2). Patients discharged to an SNF also had slightly higher odds of 90-day readmission than did patients discharged to home healthcare (ARR 1.57 [95% CI 1.23 to 2.01]; p < 0.001) (Table 2).

Table 2.

Association between discharge disposition and 30- and 90-day readmission rates

Variable	30-day readmission after discharge to SNF vs home healthcare, ARR (95% CI)	90-day readmission after discharge to SNF vs home healthcare, ARR (95% CI)
SNF vs home healthcare	1.46 (1.01-2.13)	1.57 (1.23-2.01)
Last 6-clicks mobility score	0.93 (0.91-0.96)	0.93 (0.90-0.96)
Age	1.02 (1.01-1.03)	1.00 (0.99-1.01)
Female	1.12 (0.92-1.36)	0.92 (0.83-1.02)
CCI	1.02 (0.98-1.06)	1.00 (0.97-1.01)
Primary payer
Commercial	0.79 (0.62-1.02)	0.81 (0.70-0.94)
Medicaid	1.2 (1.89-1.88)	1.3 (1.07-1.81)
Other/self-pay	0.20 (0.22-1.74)	0.74 (0.34-1.58)
Diagnosis OA	0.66 (0.50-0.88)	0.62 (0.51-0.76)
Assist available at home
Occasionally	1.04 (0.83-1.30)	1.08 (0.95-1.21)
Never	1.10 (0.67-1.79)	0.95 (0.61-1.54)
Not documented	1.05 (0.66-1.68)	1.19 (0.91-1.55)

CCI = Charlson comorbidity index; OA = osteoarthritis.

Discussion

Proper discharge planning and selection of PAC facilities are critical steps for improving outcomes and reducing complications after THA. Patients discharged to SNFs consistently exhibit higher readmission rates compared with home discharge. This study identified that after controlling for functional status, age, gender, caregiver support, and SDOH, patients discharged to SNFs still had a higher likelihood of readmission within 30 and 90 days compared with home healthcare discharges. These findings highlight that discharge to SNFs is associated with increased readmission risk irrespective of patient functional status and other key factors.

Limitations

The retrospective observational design restricts the ability to make causal inferences about the relationship between discharge disposition and readmission risk. Our retrospective design also was unable to account for potential selection bias and inconsistencies in the discharge planning process across hospitals and providers. Details were not available on the decision-making protocols, methods, or responsible clinicians determining discharge disposition for each patient. Variability in discharge planning approaches could influence which patients get sent to SNFs versus home health. We did, however, nest episodes within hospitals in the regression models to attempt to control for some level of hospital-specific practices that could vary from one hospital to the next. Although patients who were discharged to SNFs overall had greater medical complexity and social needs, we aimed to account for key variables by adjusting for age, gender, insurance status, mobility score, ADI, and caregiver support in our regression models. Despite efforts to adjust for confounding, residual confounding is likely present, as evidenced by the attenuation of the effect size with covariate adjustment. This attenuation with more complete confounder adjustment is a reason to interpret the results cautiously. There are likely additional unmeasured confounders that could further weaken the association. For instance, details on medical and surgical complications, standardized facility-level quality metrics, and patient adherence/engagement were not available in the dataset. However, we believe the factors adjusted for capture some of the most salient differences between the SNF and home healthcare groups. To avoid issues of confounding, we focused our analyses solely on adjusted models rather than making unadjusted comparisons. Although residual confounding is still possible, we believe these adjusted analyses remain useful for understanding the relationship between discharge setting and readmission risk. Further research with more comprehensive patient data could clarify this association and its underlying drivers more definitively. Also, our retrospective design was unable to account for potential changes in care processes or policies over the 5-year study period that could have influenced discharge disposition practices. For instance, factors like adoption of alternative payment models, COVID-19, increases in outpatient joint arthroplasty, and bundled payment initiatives may have affected discharge patterns over time. Not accounting for the effects of these temporal changes in policy and practice represents an additional study limitation. The 6-clicks basic mobility short form used in this study is a validated measure of functional status. However, our observation of similar readmission rates between patients discharged to SNFs or home healthcare despite differences in 6-clicks scores warrants further investigation into the sensitivity of this tool and potential value of using complementary functional assessments in this population. Moreover, another limitation of this study was that only the 6-clicks basic mobility short form was available, rather than the full assessment including the daily activity domain. Prior research found the daily activity domain may have stronger associations with readmission risk. Not having access to those items may have restricted our ability to discern the relationship between functional status and readmissions. The exclusive reliance on data extracted from our health system’s electronic health record confines our ability to capture readmissions occurring at external hospitals, potentially leading to an underestimation of readmission rates. However, our health system has a very large coverage area and market share in the regions where these patients live, meaning most readmissions would be captured. Moreover, details on medical and surgical complications leading to readmission were not available in the dataset. Understanding the specific reasons and diagnoses for readmission in between patients discharged to SNFs versus home health could help identify potential areas for quality improvement initiatives to reduce readmissions across settings. Lastly, the study’s sample is constrained to patients undergoing THA within a single health system, which may limit the generalizability of our findings to broader populations or healthcare settings. With regard to these limitations, future research should consider employing prospective or experimental designs, incorporating a more comprehensive range of covariates, and utilizing multicenter data to enhance the generalizability of findings. Moreover, further investigations should strive to elucidate the underlying mechanisms linking discharge disposition to readmission risk, facilitating more informed decision-making in postoperative care and discharge planning.

Discussion of Key Findings

After controlling for a large number of relevant confounders, patients treated in SNFs after THA had a slightly higher odds of hospital readmission than did patients treated with home healthcare. This is an important finding because it suggests that discharge to SNFs may not always be the optimal option for post-THA recovery and could lead to worse outcomes. Although SNFs aim to provide rehabilitation and skilled nursing to medically complex, functionally impaired patients, our adjusted analyses indicate they still had higher readmission rates despite controlling for these factors. This is aligned with previous studies showing that patients discharged to SNFs had higher readmission risk after THA. For instance, Bini et al. [3] conducted a study in which they examined a sample of 9150 patients who had undergone total joint arthroplasty to ascertain whether discharge disposition had an effect on readmission rates. The authors found higher readmission rates in those discharged to SNFs compared with home (5.2% versus 2.4%, respectively; p < 0.001). Moreover, after controlling for variables such as gender, age, and American Society of Anesthesiologist scores, patients discharged to SNFs after THA experienced a twofold increase in the likelihood of readmission within 90 days postoperatively compared with those patients who were discharged to their homes. Interestingly, other research demonstrated that patients discharged to SNFs after THA experience a greater risk of readmission, even after accounting for more comorbidity variables, such as chronic obstructive pulmonary disease, coronary artery disease, dialysis, corticosteroids, bleeding disorders, and preoperative transfusion [26]. However, no study to date has included functional status at discharge, level of caregiver assistance at home, or SDOH in their models. In our study, after adjusting for mobility status at discharge, home environment, ADI, and comorbidities, patients discharged to SNFs continued to exhibit an elevated likelihood of readmission within 30 and 90 days after undergoing THA. Incorporating functional assessments is of paramount importance, as these evaluations have been identified as predictors of readmission rates [12, 21]. Moreover, this tool has demonstrated the ability to identify higher readmission rates when predicting admission to a facility rather than at home [12, 21]. This finding also has important implications within the context of the contemporary value-oriented healthcare paradigm, which emphasizes the importance of delivering cost-effective, high-quality, and patient-centered care. The mandates established by the IMPACT Act of 2014 require the Centers for Medicare and Medicaid Services to include measures of function in the set of quality measures for PAC settings. As a result, the ability to assess these variables has important implications for postoperative outcomes, encompassing both patient well-being and healthcare costs [23, 31]. Therefore, it is crucial for healthcare providers to consider discharging appropriate patients to their homes as a potentially more secure, economically efficient, and superior-quality alternative to institutional care in SNFs [7, 27, 31, 34]. This approach could not only enhance patient satisfaction and recovery but also may contribute to reducing the overall burden on healthcare systems. Notwithstanding the insights provided by the present study, further research endeavors are warranted to thoroughly investigate and identify the underlying determinants of readmission for this high-risk population of patients discharged to SNFs after THA. A comprehensive understanding of these factors is paramount to devising targeted interventions and refining care transition processes, ultimately contributing to the improvement of patient outcomes and the reduction of associated healthcare expenditures. By addressing the multifaceted aspects of readmission risk, future studies may shed light on critical areas for improvement within the continuum of care and help establish evidence-based best practices for discharge planning and postoperative management. The inclusion of functional status assessment during the discharge process underscores the critical role of discharge disposition in readmission risk. Consequently, it is essential to persistently strive for appropriate patient discharge to PAC facilities, as this plays a vital role in augmenting the efficacy of the prevailing value-based healthcare system. Implementing well-informed and evidence-based discharge planning strategies can aid healthcare providers in determining the most suitable PAC settings for individual patients, ultimately promoting better patient outcomes and satisfaction while minimizing unnecessary healthcare costs [8, 27]

Our findings suggest that the care delivered in SNFs may be an important contributor to the increased readmission risk observed among patients discharged to SNFs after THA, even after adjusting for patient factors. This highlights the need to critically evaluate care transitions and coordination between settings. As patients move from the hospital to SNFs, quality of handoffs and communication between providers has major implications for outcomes. For instance, research by Boockvar et al. [4] showed that ineffective information exchange during care transitions from hospitals to nursing homes was associated with worse outcomes and higher rates of adverse events. They found that key data regarding medications, diagnoses, functional status, and care plans was often missing or ambiguous during these transitions [4]. Improving handoffs through interventions like standardized communication tools, discharge summaries, and follow-up with outpatient providers could enhance continuity of care and potentially reduce complications [15]. As we work to understand drivers of readmissions, closer analysis of hospital-to-SNF transitions and implementing robust quality improvement initiatives around handoffs may represent fruitful areas for further investigation. Stronger coordination and team-based care across settings is needed to ensure patient needs are met during this vulnerable period. Further research is warranted on how specific interventions to improve hospital-to-SNF handoffs and coordination impact downstream outcomes like readmissions after THA. Such efforts will require engagement from leaders, providers, and staff in both hospitals and SNFs to meaningfully transform practice.

Conclusion

Patients discharged to SNFs after THA had a slightly higher likelihood of hospital readmission within 30 and 90 days compared with those discharged to home healthcare. This difference persisted even after adjusting for relevant factors like functional status, home support, and social determinants of health. These results indicate that for suitable patients, direct home discharge may be a safer and more cost-effective option than SNFs. Clinicians should carefully consider these risks and benefits when making postoperative discharge plans. Policymakers could consider incentives and reforms to improve care transitions and coordination across settings. Further research using robust methods is needed to clarify the reasons for higher SNF readmission rates. Detailed analysis of patient complexity, care processes, and causes of readmission in SNFs versus home health could identify areas for quality improvement. Prospective cohorts or randomized trials would allow stronger conclusions about cause-and-effect. Importantly, no patients should be unfairly “cherry-picked” or “lemon-dropped” based only on readmission risk scores. With proper support and care coordination, even complex patients can have good outcomes. The goal should be providing excellent rehabilitation for all, while continuously improving quality, safety and value across settings.

Supplementary Material

abjs-482-1185-s001.docx ^{(20.9KB, docx)}