1 Introduction

Discharge disposition is of increasing interest to health care providers and insurance carriers as leaner models of health care management including bundled care have become prevalent. Discharge directly to home avoids the high costs of discharge to skilled nursing facilities (SNF), rehabilitation centers (RC), and nursing homes (NH). Discharge to extended care facilities (ECF) correlates with worse outcomes in terms of infections, 30-day readmission rates, and overall complication rates.¹ Hence, prediction of discharge destination is important. Elective total hip (THA) and knee (TKA) arthroplasties are two of the most common hospital orthopedic procedures with over 1 million performed annually in the United States.² The utilization of TKA alone doubled in the decade starting in 2003.³ Though there is an increasing number of arthroplasties performed as outpatients, inpatient procedures still dominate.⁴ Education and procedure timing can be optimized by accurate prediction of discharge disposition.

No National Surgical Quality Improvement Program (NSQIP) studies regarding discharge after knee or hip arthroplasties have examined data beyond 2016.^5–7 Prior NSQIP and single institution studies in patients undergoing knee and hip arthroplasty procedures have identified variables that help predict discharge destination, including age, race, American Society of Anesthesiologists (ASA) class, obesity, insurance, female sex, preoperative functional status, comorbidities, and arthroplasty procedures done after fracture.^5–14 The only information on race suggests that Blacks/African Americans are more likely than Whites to be discharged to an extended care facility.^5,11,12 The importance of smoking, which would intuitively seem to adversely affect the rate of discharge to home, has yielded mixed results.^7,10,13 The only NSQIP study in this group found that smoking negatively affected discharge to home.⁷

Three published nomograms predict a patient's post-operative course, complication risk, and discharge disposition. The NSQIP based calculator was updated in 2016 and includes predictive statistics based on a rigorous statistical methodology that has been thoroughly evaluated in terms of discrimination and calibration.¹⁵ The RAPT (Risk Assessment and Prediction Tool) was developed in Australia in 2001 to predict discharge destination of patients who had elective hip and knee replacements.¹⁶ The Predicting Location after Arthroplasty Nomogram (PLAN) was published in 2010 by Barsoum et al.¹⁷ Each of these systems is based upon statistically robust variables.

In this paper, the NSQIP database from 2015 through 2018 is examined to evaluate the recent effects of patient socio-demographics, risk assessment measures, hospitalization factors, comorbidities, and chronology on discharge disposition for hip and knee arthroplasty patients. Our primary hypothesis was that female sex, as reported previously, would affect discharge disposition. Our secondary hypotheses were that (a) increased risk assessment measures such as ASA class, (b) adverse hospitalization factors such as operative time, and (c) presence of any comorbidity, including smoking, would predict discharge to an ECF, and, finally, that the percentage of patients discharged to home would increase over the four-year study period.

2 Material and methods

The American College of Surgeons (ACS) NSQIP is a risk-adjusted, validated, outcome based, hospital surgical quality improvement approach. Data are collected prospectively and peer controlled. The primary purpose is to provide a validated database with which to quantify and compare 30-day risk-adjusted surgical outcomes among all hospitals in the program. Discharge disposition information is also recorded. Approximately 700 American hospitals use the database. The NSQIP removes all identifiers from the Participant Use Data File (PUF), and in accordance with the data use agreement entered upon requesting the specific data set for this project, this information remains deidentified. Given the deidentified nature of the NSQIP database, this study was deemed exempt from formal review by the Institutional Review Board for Protection of Human Subjects (IRBPHS). The ACS NSQIP and the hospitals participating in the ACS NSQIP are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions arrived by the authors.

Cases listed in the NSQIP as receiving a surgical intervention for a joint were the primary population of interest (All Joints, n = 413,267), using Current Procedural Terminology (CPT) codes. Subcategories for primary vs. revision procedures, and for knee (All Knee) vs. hip (All Hip) procedures, were also created. Creation of these variables via CPT code is further outlined in Fig. 1.

Close

Fig. 1 All CPT Codes- All joints-primary/revision, CPT, Current Procedural Terminology. All joints procedures CPT codes - 27130, 27132, 27447, 27446, 27445, 27443, 27442, 27441, 27440, 27134, 27137, 27138, 27486, 27487, 27488, 27091, 27090 Primary total joints procedures CPT codes - 27130, 27132, 27447, 27446, 27445, 27443, 27442, 27441, 27440 Revision total joints procedures CPT codes - 27134, 27137, 27138, 27486, 27487, 27488, 27091, 27090 ALL KNEE CPT codes - 27440, 27441, 27442, 27443, 27445, 27446, 27447, 27486, 27487, 27488 ALL HIP CPT codes 27130, 27132, 27134, 27137, 27138, 27091, 27090.

Export to PPT

The primary outcome of interest, discharge destination, was condensed into two categories to create a binary variable, indicated “discharge home” vs. “discharge other”. Individuals who were discharged “Against Medical Advice (AMA)” were removed from analyses (n = 340 out of 883,309). Variables that theoretically or empirically affect discharge destination in patients receiving a surgical intervention in a joint (“All Joint”) were identified within the dataset (total of 24), and a priori testing was conducted via simple crosstabulation against discharge destination, with chi square calculation or analysis of variance (ANOVA) employed as appropriate to determine statistically significant effects. A list of all 24 variables included in this analysis can be found in Table 1. All 24 variables were significantly associated with discharge; race, ethnicity, gender, smoking, and year were each identified as having counterintuitive statistical relationships with discharge destination and proceeded to the main analytic phase. Race and ethnicity were combined into a single “Race/Ethnicity” variable to simplify the analyses. Subjects identifying as either Native American/Alaskan Native or as Native Hawaiian/Pacific Islander appeared in very small numbers in the data set, and the two groups exhibited very similar and statistically indistinguishable outcomes. To reduce the complexity of the data set and increase the utility of the categories, we merged the two into a single “Indigenous” category. Unadjusted and adjusted odds ratios for the four resulting independent variables (race/ethnicity, gender, smoking and year) were then created via binary logistic regression. All analyses presented were conducted in Statistical Package for the Social Sciences (SPSS) v.27.

3 Results

The dataset contained 883,309 cases of orthopedic procedures, with an increasing number of cases annually. Of these, 413,627 cases represent “All Joint” procedures, which are further divided into primary and revision total joint procedures. A breakdown of the dataset can be found in Table 2. Within the “All Joints” cases, the dataset had more females than males, and the majority were non-Hispanic and White. Non-smokers were more prevalent than current smokers. Of those individuals with a recorded discharge destination, the majority were discharged home as shown in Table 3.

Male gender and current smoker status were each associated with a significant increase in the odds of a patient being discharged home (Adjusted Odds Ratio [AOR] = 1.76, 1.18, respectively, p < 0.001). Year also had a direct association with being discharged home, with more individuals going home over time. From 2015 to 2018, the odds of being discharged home more than doubled (OR = 2.04, p < 0.001). Individuals who received an “all knee” or primary total joint procedure were more likely to be discharged home compared to those who received an “all hip” or revision total joint procedure (AOR = 1.05, 1.67, respectively, p < 0.001).

Race and Ethnicity varied the likelihood of a patient going home. Regardless of Ethnicity, Black or African American individuals were the least likely to be discharged home. Of patients with a reported Ethnicity, Indigenous Non-Hispanics and Indigenous Hispanics had the highest odds of being discharged home (OR = 1.94; 1.77, p < 0.001; 0.004, respectively). Individuals with Unknown Race or Ethnicity were more likely to be discharged home. Table 4 contains more detail regarding results of logistic regression analyses.

4 Discussion

Social factors and context have an impact on a patient's ability to go directly home after a hip or knee arthroplasty. No prior NSQIP study regarding the effect of demographic variables and comorbidities on arthroplasty patient discharge has evaluated data through 2018. None of the three published nomograms utilizing NSQIP data predicting a patient's post-operative course, complication risk, and discharge disposition have incorporated the unique findings from this study.^15–17

The latest NSQIP calculator was updated in 2016 and includes predictive statistics based on rigorous methodology including statistical discrimination and calibration.¹⁵ Variables used in this calculator include for example, sex and smoking, but race was not included. The RAPT was developed in Australia in 2001 to predict patient discharge destination following elective hip or knee replacements.¹⁶ The RAPT system includes age, sex, distance able to be walked, use of gait aids, community support, and care at home as the six variables. Smoking and race variables are not included. The PLAN includes 17 variables with 7 having independent significance. They include procedure type, age, sex, heart disease, diabetes, chronic obstructive pulmonary disease (COPD), and caregiver. Each of these systems is based upon statistically robust variables. However, neither system individually includes all the significant variables found in this study. This study confirms the importance of sex as a predicter in all three prior nomograms, smoking and race as important variables. However, the unique findings of this paper are that smokers are more, not less, likely to be discharged home and that Indigenous peoples as a subset of race are also more likely to be discharged home. Not surprisingly, our data confirms that a higher percentage of patients are being discharged home every year.

Female sex has been shown to be predictive of discharge to an ECF.^{5,7,8,12–14} In a study of elective primary THA and TKA from the NSQIP database 2011–2013, Keswani et al. found that patients going home were younger, male, functionally independent, and not obese.⁷ Our data confirm this finding and supports our hypothesis regarding the effect of sex on discharge disposition. There are several hypotheses related to gender roles and norms that could follow from this result but would require different or new data collection. For example, male-identifying patients may be discharged home more often if they have partners to provide the support system or fill the caregiver role at home. Conversely, those who identify as female may be less likely to have a partner that can function as a caregiver to allow discharge home or may wait until their disease/disability is worse before undergoing the procedure. Males historically also die at younger ages (18), which could lead to trends where female partners are more frequently left without a support system at home.¹⁸

In this study, smokers were more likely to be discharged from home, disproving our hypothesis, and contradicting the previous literature. This novel finding is contrasted with that of the 2016 NSQIP calculator, which found smokers less likely to go home.⁷ Two other studies found no significant relationship between smoking and discharge destination.^6,10 Smokers may be more motivated to go directly home as opposed to transferring to a facility that does not allow smoking. We are not aware of a societal change between the 2016 study and this study including data through 2018 that would have reversed these findings.

Previous studies in joint arthroplasty patients have identified significant associations between discharge and race. White patients are consistently more likely to be discharged to home than Black patients (5, 6,11,12). In a study of specifically total knee arthroplasty patients from NSQIP 2011–2016, Gwam et al. found that in addition to other variables, Black race significantly predicted discharge to a skilled nursing or rehabilitation center.⁶ Our findings supported the relationship between race and discharge and confirmed the prior literature showing Black people are less likely to be discharged home. However, in this paper, a novel finding was that people whose race was identified as Indigenous were significantly more likely to go home. Although the reasons for this are unclear, it may be due to better social support systems or cultural emphasis on family support among these patients.

Our hypothesis that the rate of home discharge has increased over the time of the study was also confirmed. This trend has been shown in years prior to 2016.¹⁹ Our study affirms that this trend continues through the year 2018. As time has passed, more patients are going home directly from the hospital. The percentage of patients being discharged home increased each year studied. This is almost certainly due to the emphasis on lowering costs, especially within bundled payment models.

Limitations of this study include the inherent weakness of any national database, including potentially important variables such as insurance status not included in the database. The 2019 data were not available at the time of statistical analysis so the data through 2019 may show unique findings. As with any large population-based study, the results are overpowered. Hence, even small absolute differences can be statistically significant while not necessarily clinically important. Despite those weaknesses, this study examined the broadest period to date, including data through 2018. The NSQIP is a large, validated data set, and the wide sampling allows the results to be generalizable to many patient populations without fear of an inherent selection bias.

5 Conclusions

Discharge destination has many determining factors in hip or knee arthroplasty patients. Male sex, smokers, and Indigenous patients are more likely to go directly home, and more patients are going home as time goes on. While sex and race are not modifiable patient factors, they are important in assessing a patient's risk and outcome for elective hip and knee arthroplasties.

Funding/sponsorship

This work was funded by SUNY Upstate Medical University Department of Orthopedics. This research did not receive any specific grant from other funding agencies in the public, commercial or not-for-profit sectors.

Informed consent

Not applicable (database study).

Institutional ethical committee approval

Not applicable (database study).

Guardian/patient's consent

This is a national database study with deidentified information and has been deemed IRB-exempt.

Ethical statement

The corresponding author, on behalf of all authors, jointly and severally, certifies that their institution has approved the protocol for any investigation involving humans or animals and that all experimentation was conducted in conformity with ethical and humane principles of research. Authors Contribution (CRediT author statement).

CRediT authorship contribution statement

Matthew R. Bauer: Methodology, Investigation, Writing – original draft, Writing – review & editing. Sean Karott: Methodology, Investigation, Writing – review & editing. Laura A. Schad: Methodology, Software, Validation, Formal analysis, Data curation, Writing – review & editing. Christopher P. Morley: Methodology, Software, Validation, Formal analysis, Data curation, Writing – review & editing. Timothy A. Damron: Conceptualization, Methodology, Investigation, Writing – original draft, Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Timothy Damron received grants or contracts from Jim and Juli Boeheim Foundation and Carol Baldwin Breast Cancer Research Foundation which ended in 2021. Timothy Damron reports a relationship with BoneSource (Cerament) Physician Advisory Board that includes: consulting or advisory. Timothy Damron reports a relationship with Jim and Juli Boeheim Foundation and Carol Baldwin Breast Cancer Research Foundation that includes: funding grants. Timothy Damron reports a relationship with JAAOS Deputy Editor for Research, AAOS Editorial Board Member for the following journals: Clinical Orthopedics and Related Research, Journal of Orthopedic Research, Journal of Surgical Oncology, JAAOS Global Reviews and Reports, BMC Musculoskeletal Disorders, The Online Orthopedic Journal, Medicina, PLOS One that includes: board membership. Timothy Damron reports a relationship with David G. Murray Research Endowment, Upstate Research Foundation, David Smyth Cancer Research Fund and Marvin Damron Cancer Research Fund that includes: employment and non-financial support. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Laura A. Schad reports a relationship with NYS Health Foundation and NYS Department of Health/Health Research Inc that includes: funding grants.

Christopher P. Morley reports a relationship with New York City MTA, NYS Health Foundation and NYS Department of Health/Health Research Inc that includes: funding grants. Christopher P. Morley reports a relationship with HRSA Grant Review, SUNY Downstate Research Office – Grant Review and Michigan State University that includes: speaking and lecture fees. Christopher P. Morley reports a relationship with Journal editor, Society for Teachers of Family Medicine that includes: board membership. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The rest of the authors have no interests to declare.