Identifying Risk Factors for Complication and Readmission with Same-Day Discharge Arthroplasty

Abstract

Background

The COVID-19 pandemic caused a surge of same-day discharge (SDD) for total joint arthroplasty. However, SDD may not be beneficial for all patients. Therefore, continued investigation into the safety of SDD is necessary as well as risk stratification for improved patient outcomes.

Methods

This retrospective cohort study examined 31,851 elective SDD hip and knee arthroplasties from 2016 to 2020 in a large national database. Logistic regression models were used to identify patient variables and preoperative comorbidities that contribute to postoperative complication or readmission with SDD. Adjusted odds ratios (AOR) and 95% confidence intervals (CI) were calculated.

Results

SDD increased from 1.4% in 2016 to 14.6% in 2020. SDD is associated with lower odds of readmission (AOR: 0.994, CI: 0.992-0.996) and postoperative complications (AOR: 0.998, CI: 0.997-1.000). Patients who have preoperative dyspnea (AOR: 1.03, CI: 1.02-1.04, P < .001), chronic obstructive pulmonary disease (AOR: 1.02, CI: 1.01-1.03, P = .002), and hypoalbuminemia (AOR: 1.02, CI: 1.00-1.03, P < .001), had higher odds of postoperative complications. Patients who had preoperative dyspnea (AOR: 1.02, CI: 1.01-1.03), hypertension (AOR: 1.01, CI: 1.01-1.03, P = .003), chronic corticosteroid use (AOR: 1.02, CI: 1.01-1.03, P < .001), bleeding disorder (AOR: 1.02; CI: 1.01-1.03, P < .001), and hypoalbuminemia (AOR: 1.01, CI: 1.00-1.02, P = .038), had higher odds of readmission.

Conclusion

SDD is safe with certain comorbidities. Preoperative screening for cardiopulmonary comorbidities (eg, dyspnea, hypertension, and chronic obstructive pulmonary disease), chronic corticosteroid use, bleeding disorder, and hypoalbuminemia may improve SDD outcomes.

Total joint arthroplasties (TJAs) have increased over the past 20 years and will continue to rise [1]. As osteoarthritis prevalence and TJA demand increases, providers are increasingly focused on maximizing quality while minimizing cost. Same-day discharge (SDD) uses fewer hospital resources than traditional inpatient stays [[2], [3], [4]]. The COVID-19 pandemic increased health care costs and limited capacity for inpatient procedures [5]. Although TJA SDD has increased over the past decade, the COVID-19 pandemic created a surge in this demand [6].

The definition of SDD for TJA may vary from one author to another or from one database to another. In National Surgical Quality Improvement Project (NSQIP), comparing inpatient/outpatient status versus actual length of stay leads to different outcomes [7]. For the purposes of this paper, SDD is defined as a patient discharged within 24 hours of having an arthroplasty [7]. While some studies found similar or improved outcomes following SDD [[8], [9], [10], [11], [12]], others found SDD associated with postoperative bleeding requiring transfusion [13], and deep vein thrombosis, renal failure, component failure [14], and major and minor complications in total knee arthroplasty (TKA) [15]. More recent studies found SDD unsafe in certain populations [[16], [17], [18]]. Sher et al, analyzing the NSQIP database from 2011 to 2014, identified age older than 80 years, smoking, bleeding-causing disorders, and American Society of Anesthesiologist (ASA) ≥3 as independent predictors for severe adverse events or readmissions [16]. Courtney et al, reviewing 1,012 patients in a single institution, identified chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), coronary artery disease, and cirrhosis as independent predictors for major medical complications requiring physician intervention within 24 hours [17].

It is important to update our understanding of SDD with its increase during the COVID-19 pandemic, which may have led to unintended consequences on surgical outcomes on patients following TJA. The pandemic delayed elective procedures, and longer wait times were associated with decreasing quality of life, nearly doubling the number of patients in a health state defined as “worse than death” [19]. This may have caused surgeons to be more lenient on their criteria for SDD [[6], [20]] and may have motivated patients to avoid hospitalization altogether [21]. With the growing pressure to lower TJA procedure costs and improve patient quality care, the challenge will be for orthopaedic surgeons and hospitals to identify patients appropriate for SDD by examining certain necessary risk factors and establishing preoperative guidelines for SDD consideration [16]. Due to conflicting results in the literature, analysis of outcomes following SDD in TJA is warranted. Additionally, further risk stratification is necessary to improve patient outcomes following SDD.

Methods

Setting and Patient Sample Size

We included primary total hip arthroplasty (THA, Current Procedural Terminology [CPT] 27130) and primary total knee arthroplasty (TKA, CPT 27447) patients from 706 hospitals participating in the American College of Surgeons National Surgical Quality Improvement Project (NSQIP) [22]. Our initial analysis between 2016 and 2020, analyzed patients whose index procedure was an arthroplasty procedure and excluded emergency cases (1,746, 0.35%), assessing 495,727 TJAs. In the secondary analysis, the sample was limited to 31,851 elective hip and knee arthroplasties that had a total hospital length of stay <24 hours, which we consider SDD [7]. Data from each site are extracted by surgical certified reviewers that are intensively trained with continuing education courses to standardize data collection. Data consistency and reliability at each NSQIP hospital are evaluated by an inter-rater reliability audit program (1.6% was the disagreement rate in 2008) [23]. NSQIP contains deidentified data, so no institutional review board approval was needed.

Variables

We included patient characteristics including age, sex, race, ethnicity, smoking status, body mass index (BMI), operative time, ASA physical status classification score ≥3, and preoperative medical comorbidities. Operative time was converted into 15-minute increments so that OR in the multivariables were more clinically interpretable.

Outcomes

Adverse outcomes within 30 days of arthroplasty were of interest: readmission for any reason, postoperative medical complication, and reoperation for any reason.

Data Analyses

Missing or null data (5,879, 1.17%) were removed from analyses. Univariate logistic regression models assessed the association between year, as a continuous variable, and rates of SDD and outcomes for all TJA procedures from 2016 to 2020 (See Supplementary Table 1). Pairwise t-tests compared continuous patient characteristics (eg, age, BMI, and operative time), and chi-squared tests compared categorical patient characteristics (eg, sex, race, ethnicity, smoking status, ASA, and one or more preoperative comorbidities) and surgical outcomes between those who underwent SDD and hospital length of stay ≥1 day (See Table 1 ). Patients who underwent an SDD were more likely to be younger, male, White, non-Hispanic, and non-smokers with a BMI <30.6 ± 5.7 kg/m², have operative times <82.9 ± 27.5 minutes and be less likely to have preoperative medical comorbidities when compared to patients that did not undergo an SDD (P < .001, See Table 1). SDD patients were also associated with a 1.4% lower incidence of readmission, a 2.9% lower incidence of postoperative medical complications, and a 0.5% lower incidence of reoperation when compared to patients that did not undergo SDD (P < .001, See Table 1).

Table 1.

Patient Characteristics and Postoperative Complications of Same-Day Discharge (SDD) and Hospital Length of Stay ≥ 1 d After Arthroplasty.

Patient Characteristics	SDD, % (n)	Hospital Length of Stay ≥1 d, % (n)	P-Value
Mean age in years (range)	64 (56 to 74)	67 (57 to 77)	<.001
Sex (Women)	51.3 (16,353)	59.1 (274,062)	<.001
Race (White)	77.2 (24,590)	69.7 (323,235)	<.001
Ethnicity (Hispanic)	10.1 (2,357)	19.4 (20,951)	<.001
Mean BMI in kg/m2 (Range)	31 (25 to 37)	32 (25 to 39)	<.001
Mean operative time in minutes (Range)	83 (55 to 111)	91 (54 to 128)	<.001
Active smoking status	7.2 (2,305)	9.6 (44,364)	<.001
ASA ≥3	33.1 (10,554)	49.9 (231,373)	<.001
One preoperative comorbidity	13.9 (4,441)	22.8 (105,619)	<.001
≥2 preoperative comorbidities	41.6 (13,242)	44.5 (206,637)	<.001
Postoperative complications	SDD, % (n)	Hospital Length of Stay ≥ 1 d, % (n)	P-Value
Readmission	1.8 (586)	3.2 (14,616)	<.001
Postoperative medical complications	2.5 (783)	5.4 (25,004)	<.001
Reoperation	0.9 (286)	1.4 (6,416)	<.001

BMI Body mass index, ASA American Society of Anesthesiologist, SDD is defined as total hospital length of stay of 0 in the NSQIP database. Preoperative comorbidities included diabetes mellitus requiring therapy with noninsulin agents or insulin, partially or fully dependent functional health status, disseminated cancer, open wound with or without infection prior to surgery, corticosteroid/immunosuppressant use for chronic condition, more than 10% loss of body weight in the 6 mo prior to surgery, bleeding disorders, preoperative transfusion, and sepsis within 48 h prior to surgery, hypertension requiring medication, ascites, renal failure, dialysis, congestive heart failure (CHF) , history of ventilator dependence within 48 h prior to surgery, chronic obstructive pulmonary disease (COPD), and dyspnea. Preoperative assessment of labs for hypoalbuminemia was included (albumin < 3.5 g/dL).

Univariate logistic regression models examined the association of each surgical outcome with patient characteristics and SDD status. Covariates with a P-value < .25 in univariate analyses were included as variables in the multivariable models [24]. Multivariable analyses assessed the association between SDD (hospital length of stay ≥1 day as the reference) and readmission, medical complication, and reoperation, while controlling for patient characteristics (ie, age, women, BMI operative time, ASA ≥3, active smoking status, race, and one or multiple preoperative medical comorbidities, See Supplementary Table 2). There were no highly correlated baseline factors (below −0.7 or above 0.7) and variance inflation factor for each was less than 5. Next, the number needed to expose (NNE) was calculated from AORs and the unexposed event rate estimated by multiple logistic regression models [25]. The NNE is analogous to the number needed to treat with adjustment for confounding variables, making it a more appropriate measure in observational studies [25]. The NNE was calculated to understand the AORs in absolute terms for more relevant comparison between the SDD group and longer hospital stay group. The NNE for one additional person to be harmed was calculated when the AOR was > 1 (harmful exposure), and the NNE for one additional person to benefit was calculated with the AOR < 1 (beneficial exposure).

Next, the analyses were limited to same-day procedures and the models were repeated. Initially, univariate analyses assessed all available patient covariates on readmission and postoperative medical complications. Significant variables included in the multivariable included age, sex, BMI operative time, ASA ≥3, diabetes, dyspnea, COPD, hypertension, steroid, bleeding disorder, and hypoalbuminemia contributing to readmission and reoperation after SDD. Then we analyzed the incidence of these comorbidities in patients selected for SDD between 2016 and 2020 (See Supplemental Table 3). Univariable analyses were done to assess the relationship between year and each of these significant preoperative comorbidities to determine the incidence of comorbidities throughout the years. There were fewer patients with hypoalbuminemia (B = −0.128, P < .001); no change in the patients with COPD, hypertension, chronic steroid use, or bleeding disorders; and more patients with dyspnea (B = 0.088, P < .021, Supplemental Table 3) Multivariable analyses assessed the association of year with dyspnea and hypoalbuminemia, while adjusting for patient characteristics (ie, age, women, BMI operative time, ASA ≥3, active smoking status, and race, Supplementary Table 4).

AORs and 95% confidence intervals (CI)s were calculated for all logistic regression models. Significance was set at a P-value < .05. Statistical analyses were performed using R software (version 4.1.0, Vienna, Austria) [26] RStudio (version 1.4.1717, Boston, MA) [27] and the “tidyverse” package [28].

Results

From 2016 to 2020, the rate of SDD increased 10-fold from 1.3 to 14.6%, while the rates of readmission decreased from 3.2 to 2.8% (P < .001), postoperative complications decreased from 6.1 to 5.0% (P < .001), and reoperation decreased from 1.5 to 1.3% (P < .001, See Fig. 1 and Supplementary Table 1).

Fig. 1.

Frequency of same-day discharge (SDD) over the years. The frequency of patients undergoing SSD for total joint arthroplasty (TJA) increased from 2016 to 2020 (pairwise t-test between each year: P < .001).

In our multivariable analyses, SDD was associated with 0.6% lower odds of a readmission (AOR: 0.994, CI: 0.992-0.996, P < .001) and 1.4% lower odds of a postoperative medical complication (AOR: 0.986, CI: 0.983-0.988, P < .001, See Table 2). To place these readmission statistics in absolute terms, preventing one readmission would require exposure of SDD to 9,428 patients, preventing 1 postoperative medical complication would require exposure of SDD to 2,929 patients. Reoperation odds did not differ significantly between those who did or did not undergo SDD (P = .110, See Table 2); therefore, reoperation was excluded from further analyses.

Table 2.

Multivariable Models for Surgical Outcomes for Same-Day Discharge (SDD).

Surgical Outcomes	Same Day Discharge (ref = Hospital Length of Stay ≥1 d), Adjusted Odds Ratio (95% CI)	P-Value	NNE
Readmission	0.994 (0.992-0.996)	<.001	−9,428
Postoperative medical complications	0.986 (0.983-0.988)	<.001	−2,929
Reoperation	0.998 (0.997-1.000)	.110	−56,059

NNE Number Needed to Expose, CI Confidence Interval, Same Day Discharge is defined as total hospital length of stay of 0 in the NSQIP database. All patient characteristics were used to predict each surgical outcome (See Supplementary Table 2). A postoperative medical complication was defined as any one of these predetermined events–wound complication (ie, superficial incisional surgical site infection (SSI), deep incisional SSI, organ space SSI, and wound disruption), pulmonary complications (ie, pneumonia, unplanned reintubation, use of ventilator for more than 48 h), renal complications (ie, renal insufficiency, renal failure, urinary infection), central nervous system complication (ie, stroke), cardiovascular complications (ie, cardiac arrest, myocardial infarction), venous thromboembolic event (ie, deep vein thrombosis and pulmonary embolism), and systemic complications (ie, systemic sepsis, septic shock, and blood transfusion).

Of those who had SDD between 2016 and 2020, subgroup analyses were conducted to identify the preoperative medical comorbidities of patients associated with readmission or a postoperative medical complication within 30 days of SDD. The preoperative medical factors associated with readmission at 30 days included dyspnea (AOR: 1.02, CI: 1.01-1.03, P < .001), hypertension (AOR: 1.01, CI: 1.01-1.03, P = .003), chronic corticosteroid use (AOR: 1.02, CI: 1.01-1.03, P < .001), bleeding disorders (AOR: 1.02; CI: 1.01-1.03, P < .001), and hypoalbuminemia (AOR: 1.01, CI: 1.00-1.02, P = .038, See Table 3). To place these readmission statistics in absolute terms, the number needed to undergo SDD for one additional to have a readmission is 899 patients with dyspnea, 879 patients with hypertension, 1,341 patients with chronic steroid use, 2,004 with bleeding disorder, and 2,313 patients with hypoalbuminemia (See Table 3). The preoperative medical factors associated with postoperative medical complication at 30 days included dyspnea (AOR: 1.03, CI: 1.02-1.04, P < .001), COPD(AOR: 1.02, CI: 1.01-1.03, P = .002), and hypoalbuminemia (AOR: 1.02, CI: 1.01-1.03, P < .001). To place these readmission statistics in absolute terms, the number needed to undergo SDD for one additional to have a postoperative medical complication is 719 patients with dyspnea, 1,174 patients with COPD, or 1,003 patients with hypoalbuminemia (See Table 3). The number of patients with hypoalbuminemia with SDD decreased significantly from 2016 to 2020 (AOR: 0.860, CI: 0.842-0.878, P < .001), while the number of patients with SDD was unchanged from 2016 to 2020 (AOR: 1.074, CI: 0.998-1.157, P = .059, Table 4).

Table 3.

Incidence and Multivariable Models for Readmission and Postoperative Medical Complications for Same-Day Discharge (SDD) Arthroplasty.

Patient Factors	Incidence for Readmission, % (n)	Readmission, AOR (95% CI)	P-Value	NNE	Incidence for Postoperative Complication, % (n)	Postoperative Complications, AOR (95% CI)	P-Value	NNE
Mean age in years (range)	69 (58 to 80)	1.001 (1.001-1.001)	<.001	NA	69 (58 to 80)	1.001 (1.000-1.001)	<.001	NA
Women	56.4 (8,567)	1.002 (0.999-1.005)	.111	2036	62.4 (16,092)	NA	NA	NA
Mean BMI in kg/m2 (range)	32 (25 to 39)	1.000 (1.000-1.001)	.051	NA	32 (25 to 39)	1.001 (1.000-1.001)	.002	NA
Mean operative time in minutes (range)	95 (55 to 135)	1.001 (1.001-1.002)	<.001	NA	105 (52 to 158)	1.003 (1.002-1.004)	<.001	NA
ASA ≥ 3	46.1 (270)	1.004 (1.001-1.008)	.020	1,008	46 (360)	1.008 (1.004-1.012)	<.001	269
Diabetes	15.9 (93)	1.002 (0.997-1.006)	.497	3,740	15.3 (120)	1.002 (0.997-1.008)	.396	3,233
Dyspnea	5.1 (30)	1.023 (1.012-1.034)	<.001	899	4.9 (38)	1.028 (1.015-1.041)	<.001	719
COPD	3.6 (21)	1.010 (0.998-1.021)	.104	2,883	4.0 (31)	1.021 (1.007-1.034)	.002	1,174
Hypertension	64.7 (379)	1.005 (1.001-1.008)	.004	879	57.5 (450)	1.000 (0.996-1.003)	.885	NA
Corticosteroid	4.6 (27)	1.017 (1.007-1.027)	<.001	1,341	3.3 (26)	1.009 (0.997-1.020)	.144	3,298
Bleeding disorder	2.7 (16)	1.019 (1.005-1.034)	.008	2004	1.9 (15)	1.009 (0.993-1.026)	.270	5,665
Hypoalbuminemia	4.1 (24)	1.011 (1.001-1.022)	.027	2,313	4.5 (35)	1.022 (1.010-1.034)	<.001	1,003

All patient characteristics and medical comorbidities listed above, except women in the postoperative medical complication model, were included in both multivariable models.

BMI, body mass index; ASA, American Society of Anesthesiologist; COPD, Chronic Obstructive Pulmonary Disease; Corticosteroid Steroid/immunosuppressant use for chronic condition; NNE, number needed to expose.

Table 4.

Multivariable Models for Dyspnea and Hypoalbuminemia.

Medical Comorbidity	Year, AOR (95% CI)	P-Value
Dyspnea	1.074 (0.998-1.157)	.059
Hypoalbuminemia	0.860 (0.842-0.878)	<.001

The multivariable analyses examined the association between year and dyspnea and hypoalbuminemia, while adjusting for patient characteristics (ie, age, women, BMI operative time, ASA ≥ 3, active smoking status, and race, See Supplementary Table 4).

Discussion

The rate of SDD increased 10-fold from 2016 to 2020. The literature regarding outcomes following SDD compared to inpatient hospital stays has shown either clinical indifference or improvement [[8], [9], [10], [11], [12]] or worse outcomes [[13], [14], [15]]. This trending demand for SDD stems from an effort to circumvent limited health care resources including limited hospital capacity for inpatient admissions faced during the COVID-19 pandemic [5,6,29]. With the rise of SDD in the setting of COVID-19, it was important to reassess SDD outcomes. Our study from 2016 to 2020 identified a decrease in observed readmission, postoperative complication, and reoperation rates. Since 2016, we noticed that SDD patients were more likely younger, male, non-Hispanic White, lower average BMI, shorter operative time, less likely to smoke, lower ASA score, and less preoperative comorbidities compared to patients with longer inpatient hospital stays. We suggest this may have influenced the improved results seen in this study regarding SDD in TJA. To prevent one medical complication after TJA, an institution would need to keep nearly 3,000 patients in the hospital past 24 hours. So even a high-volume institution with 14 board-certified orthopaedic surgeons performing 1,743 primary unilateral, SDD arthroplasties would need to deny SDD for almost 2 years to prevent one complication [30]. Denying SDD at that same institution for nearly 5 years would prevent one readmission (See Table 2). Hence, keeping patients eligible for SDD in the hospital longer than 24 hours after arthroplasty to prevent that single complication or readmission is a poor use of health care resources. We showed a 0.6% decreased risk for readmission and 1.4% decreased risk for postoperative complication following SDD compared to inpatient stays. These values, while significant, reflect limited clinical improvement following SDD. Therefore, it is still important to identify risk factors that contribute to readmission and postoperative complication following SDD. With appropriate risk stratification, orthopaedic surgeons can improve patient care and minimize postoperative episodes requiring additional hospital care.

Our study adds on the previous findings similar to the findings of Sher et al, which identified ASA ≥3 and bleeding disorders as contributors to readmission and serious adverse events [16] and Courtney et al [17], which identified COPD as a contributor to postoperative medical complications. For further understanding of adverse SDD outcomes, our study identified risk factors that contribute distinctly to readmission and postoperative complications. Hypertension, chronic corticosteroid use, and bleeding disorder uniquely increased risk for readmission, and COPD increased risk for postoperative complications. We suggest that bleeding disorder likely leads to risk for postoperative transfusion requiring readmission into the hospital [13]. Chronic corticosteroid use is associated with poor wound healing [31] and has been shown to increase risk for surgical site infection, wound dehiscence, pneumonia, and urinary tract infections, which could lead to sepsis, and thus readmission [32,33].

Interestingly, from 2016 to 2020, less number of patients with hypoalbuminemia and more patients with preoperative dyspnea underwent SDD. Due to the large amount of evidence suggesting hypoalbuminemia as a risk factor for postoperative complication [16,18], perhaps more orthopaedists are refusing SDD to patients with hypoalbuminemia. While we do not agree with strict refusal for patients with hypoalbuminemia, we believe that patients may benefit from preoperative strategies to modify their albumin levels prior to SDD. Treatment approaches include direct intravenous human albumin administration or indirect parenteral nutrition or enteral protein supplementation [34]. Dyspnea is not typically used for selection criteria for SDD [18], which is likely why more patients with dyspnea are being found to undergo SDD. It is also possible that dyspnea is often overlooked by orthopaedic physicians as a possible risk factor for adverse outcomes. Other manifestations of dyspnea, such as COPD and CHF, are risk factors for a major medical complication following short-stay arthroplasties [17,35]. However, Courtney et al and Lovald et al did not identify dyspnea as an independent risk factor [17,35]. Because CHF was not a significant contributor to adverse outcomes in our study, our data suggest that surgeons are already appropriately restricting patients with diagnosed CHF from undergoing SDD. Dyspnea is suggestive of undiagnosed CHF or other cardiac or pulmonary conditions not necessarily offered in NSQIP [36]. While we agree that preoperative dyspnea should not be used to exclude patients from SDD, it is likely that patients will benefit from preoperative strategies to identify and address the underlying cause of their dyspnea prior to SDD. To do this, the preoperative exam could also include a more thorough physical examination, noninvasive tests like pulse oximetry, and pulmonary function tests to identify patients with dyspnea, with potential consideration for further confirmatory testing with electrocardiography, chest radiography, cardiac stress test, and a ventilation-perfusion scan [36].

There are several limitations to this study. NSQIP permits only 30-day postoperative outcomes without providing intermediate and long-term complication and readmission rates, thus limiting the full scope of complications that can occur after discharge. Also, we assessed patients discharged from hospitals that contribute to NSQIP within 24 hours and not necessarily those from an outpatient facility such as an ambulatory surgical center. However, our study does provide insights into risks for complications from discharge within 24 hours from a hospital setting. In addition, since this is a retrospective cohort study, it is not methodologically free of significant biases. However, it is important to point out that NSQIP collects its data in a prospective manner. It contains large data from 706 participating hospitals, and as such, it is a good resourceful tool for preoperative information for surgeons.

Future quality improvement efforts should be focused on coordinated postoperative care, preoperative optimization, and preoperative risk assessment to target improved health outcomes for arthroplasty patients. Additional research and resources to track long term outcomes for SDD patients are recommended to improve evidence-based knowledge and clinical practice on this priority public health issue. As the demand for SDD continues to rise, appropriate risk stratification is critical for patient safety. To our knowledge, there is no national guideline for appropriate patient selection. Our study points out dyspnea and hypoalbuminemia as the most significant preoperative key risk factors that result in readmission and postoperative medical complications following SDD. While these should not solely inhibit a patient from SDD, we suggest that appropriate measures to address the dyspnea and hypoalbuminemia prior to surgery will be beneficial.

Appendix

Supplementary Table 1.

Rate of Same-Day Discharge (SDD) and Surgical Complication Following Arthroplasty From 2016 to 2020.

	2016	2017	2018	2019	2020	β	Standard Error	Z-value	P-Value
SDD, % (No.)	1.3% (1,182)	2.8% (2,758)	5.7% (5,933)	8.3% (9,696)	14.6% (12,282)	0.0316	0.0003	125.9	<.001
Readmission, % (No.)	3.2% (3,023)	3.2% (3,147)	3.1% (3,221)	3.0% (3,471)	2.8% (2,340)	−0.0011	0.0002	−6.11	<.001
Postoperative Medical Complications, % (No.)	6.1% (5,667)	5.3% (5,192)	4.8% (4,952)	5.0% (5,808)	5.0% (4,168)	−0.0025	0.0002	−10.92	<.001
Reoperation, % (No.)	1.5% (1,383)	1.4% (1,347)	1.4% (1,431)	1.3% (1,483)	1.3% (1,058)	−0.0406	0.0090	−4.51	<.001

β Beta coefficient. Preliminary analysis was done to detect differences in continuous patient factors or outcomes by year. First, an overall F-test of one-way analysis of variance (ANOVA) test was utilized, followed by the post hoc Benjamini-Hochberg (BH) procedure to assess differences between specific year (ie, 2016 versus 2020), which showed significant of P < .001 for comparisons by each year. Then univariate logistic regression models examined the association between year, as a continuous variable, and SDD, readmission, postoperative medical complications, and reoperation.

Supplementary Table 2.

Multivariable Analysis of Patient Factors Contributing to Readmission, Postoperative Complications, and Reoperation.

Patient Factors	Readmission	P-Value	Postoperative Medical Complications	P-Value	Reoperation	P-Value
Age	1.001 (1.001-1.001)	<.001	1.001 (1.001-1.001)	<.001	1.000 (1.000-1.000)	<.001
Women	0.998 (0.997-0.999)	<.001	1.012 (1.011-1.013)	<.001	0.999 (0.999-1.000)	.095
BMI	1.000 (1.000-1.000)	.401	0.999 (0.999-0.999)	<.001	1.000 (1.000-1.000)	<.001
Operative time	1.002 (1.002-1.002)	<.001	1.009 (1.009-1.010)	<.001	1.002 (1.001-1.002)	<.001
ASA ≥ 3	1.014 (1.012-1.015)	<.001	1.026 (1.025-1.027)	<.001	1.005 (1.004-1.006)	<.001
Active smoking status	1.012 (1.010-1.014)	<.001	1.008 (1.006-1.010)	<.001	1.008 (1.007-1.010)	<.001
Black or African American	1.003 (1.001-1.005)	.002	1.002 (1.000-1.005)	.056	0.999 (0.998-1.000)	.066
Asian	0.992 (0.988-0.995)	<.001	0.992 (0.988-0.997)	.001	0.994 (0.992-0.997)	<.001
American Indian or Alaskan Native	1.000 (0.993-1.007)	.968	1.001 (0.992-1.010)	.850	1.001 (0.997-1.006)	.581
Native Hawaiian or Pacific Islander	0.994 (0.986-1.003)	.211	0.990 (0.979-1.001)	.075	0.995 (0.989-1.000)	.066
One preoperative comorbidity	1.004 (1.003-1.005)	<.001	1.003 (1.002-1.005)	<.001	1.002 (1.001-1.002)	<.001
≥2 Preoperative comorbidities	1.019 (1.017-1.020)	<.001	1.030 (1.028-1.032)	<.001	1.006 (1.005-1.007)	<.001

BMI, body mass index; ASA, American Society of Anesthesiology physical status classification system. Patient factors and preoperative medical comorbidities were controlled for in the multivariable analysis assessing SDD and 30-d surgical outcomes: readmission, postoperative medical complications, and reoperation from Table 2.

Supplementary Table 3.

Incidence and Univariate Models for Preoperative Comorbidities by Year.

Medical Comorbidity	2016	2017	2018	2019	2020	Β	Standard Error	Z-value	P-Value
Dyspnea, % (No.)	2.7% (32)	1.3% (35)	1.3% (79)	2.2% (218)	2.1% (256)	0.0878	0.0379	2.318	.021
COPD, % (No.)	2.0% (24)	2.2% (62)	1.6% (94)	1.7% (166)	1.7% (212)	−0.0438	0.0375	−1.166	.244
Hypertension, % (No.)	52.3% (618)	50.4% (1,389)	49.4% (2,930)	50.0% (4,843)	51.1% (6,277)	0.0083	0.0100	0.831	.406
Steroid, % (No.)	2.9% (34)	1.9% (53)	2.0% (119)	2.4% (228)	2.4% (298)	0.0381	0.0341	1.117	.264
Bleeding Disorder, % (No.)	1.0% (12)	1.2% (33)	1.0% (59)	1.1% (110)	1.2% (146)	0.0360	0.0483	0.744	.457
Hypoalbuminemia, % (No.)	2.0% (24)	0.9% (25)	2.2% (133)	3.3% (316)	1.8% (220)	−0.1280	0.0102	−12.59	<.001

β Beta coefficient; COPD, Chronic Obstructive Pulmonary Disease, Steroid Steroid/immunosuppressant use for chronic condition. Univariable analysis was done to assess the relationship between year and each of the significant preoperative comorbidities.

Supplementary Table 4.

Multivariable Models of Dyspnea and Hypoalbuminemia.

Patient Characteristics	Dyspnea, AOR (95% CI)	P-Value	Hypoalbuminemia, AOR (95% CI)	P-Value
Year	1.074 (0.998-1.157)	.059	0.860 (0.842-0.878)	<.001
Age	1.032 (1.022-1.041)	<.001	1.008 (1.006-1.011)	<.001
Women	1.323 (1.124-1.560)	<.001	1.025 (0.979-1.073)	.286
BMI	1.062 (1.048-1.076)	<.001	0.999 (0.995-1.003)	.624
Operative time	NA	NA	0.915 (0.903-0.927)	<.001
ASA ≥ 3	3.240 (2.702-3.897)	<.001	0.977 (0.928-1.028)	.372
Active smoking status	2.368 (1.845-3.006)	<.001	0.930 (0.851-1.017)	.112
Black	0.966 (0.693-1.313)	.833	0.977 (0.892-1.069)	.613
Asian	0.803 (0.456-1.306)	.411	1.560 (1.392-1.750)	<.001
American Indian or Alaskan Native	1.206 (0.197-3.884)	.795	1.691 (1.128-2.564)	.012
Native Hawaiian or Pacific Islander	0.581 (0.095-1.864)	.452	2.379 (1.699-3.376)	<.001
Hispanic ethnicity	0.601 (0.395-0.876)	.012	1.251 (1.146-1.364)	<.001
Unknown ethnicity	1.659 (1.154-2.365)	.006	2.540 (2.246-2.874)	<.001

The multivariable analyses adjust for patient characteristics (ie, age, women, BMI operative time, ASA ≥ 3, active smoking status, and race).