Incidence, Causes and Predictors of 30-Day Readmission After Shoulder Arthroplasty

Abstract

Background

The Center for Medicare and Medicaid Service has identified several quality metrics, including unplanned readmission within 30 days of surgery, to assess and compare surgeons and hospitals. The purpose of this study was to identify the incidence, causes and risk factors for unplanned 30-day readmission after total shoulder arthroplasty.

Methods

We identified patients undergoing primary elective shoulder arthroplasty performed at American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) participating hospitals in 2013. Cases were stratified by readmission status. Univariate and multivariate analyses were employed to assess patient demographics, comorbidities and operative variables predicting unplanned readmission.

Results

2779 patients undergoing shoulder arthroplasty were identified, with 74 (2.66%) requiring unplanned readmissions within 30 days of surgery. The most common surgical causes for unplanned readmission were surgical site infections (18.6%), dislocations (16.3%) and venous thromboembolism (14.0%). Medical causes for readmission were responsible for 51% of unplanned readmissions. Multivariate analysis identified patient age >75 (OR 2.62, 95% CI: 1.27 - 5.41), and ASA class of 3 (OR 1.79, 95% CI: 1.01 - 3.18) or 4 (OR 3.63, 95% CI: 1.31 - 10.08) as independent risk factors for unplanned readmission. Predictive modeling estimated that patients with ASA class of 4 and age >75 are 17.4 times more likely (95% CI 1.77-171.09) to be readmitted within 30 days of shoulder arthroplasty.

Conclusion

Unplanned readmission after shoulder arthroplasty is infrequent and medical complications account for more than 50% of occurrences. The risk of readmission exponentially increases when age and preoperative comorbidity burden are increased.

Introduction

Unplanned hospital readmissions are costly events¹ that have gained increased interest in recent orthopedic literature²,³ for good reason. The Readmissions Reduction Program, put forward in the Patient Protection and Affordable Care Act, would withhold payments from healthcare providers and hospitals with excessive unplanned readmission rates⁴. High volume orthopedic procedures in older patients are subject to unplanned readmission. The orthopedic community must perform comprehensive analyses of such procedures in order to understand factors associated with readmission.

Total shoulder arthroplasty (TSA) is a common treatment for glenohumeral arthritis⁵ with increasing surgical volumes in the United States⁶,⁷. Previous studies have estimated 14-day readmission rates of 5.6%⁸ and 90 day readmission rates of 5.9%-7.3%⁹,¹⁰ following TSA. Medical complications have previously been found to comprise the majority of readmissions in TSA. Surgical site infection (SSI) and management of shoulder dislocation/ instability are two of the more commonly cited operative complications necessitating readmission⁹. Increased age⁹, Medicaid insurance status⁹ and arthroplasty technique (hemi or reverse)¹⁰ have been associated with readmission following shoulder arthroplasty.

While descriptive studies⁹,¹⁰ exist, there is a paucity of studies in the current literature that comprehensively analyzes the incidence, causes, and risk factors for 30- day readmission following TSA in a large, multi-center cohort. Data on the frequency of and risk factors for readmission may aid healthcare systems in assessing and setting benchmarks for quality. Additionally, modifiable risk factors that are recognized prior to surgery may allow surgical teams to optimize patients prior to surgical intervention. Patients who are identified to be at-risk for readmission may also be managed differently with regard to hospital discharge criteria and clinic follow-up practices. The aims of the current study were to identify: (1) the incidence, (2) the most common causes and (3) patient risk factors for unplanned readmission in a group of patients undergoing elective TSA.

Methods

This study was deemed HIPAA-compliant and institutional review board-exempt. The American College of Surgeons National Surgical Quality Improvement Program (ACS - NSQIP) database is comprised of over 500 academic and private medical institutions from across the United States, which is well established in the orthopedic literature²,³,¹¹. The ACS- NSQIP collects prospective patient and operative data as well as 30-day outcomes. Patient morbidity and mortality data is collected for 30 days post-operatively. Trained surgical clinical reviewers (SCRs) hired by each institution review progress notes, operative notes and data from follow-up visits after surgery. SCRs may contact patients or surgeons in order to clarify any discrepancy in the medical record or find patients who have not presented for follow up within 30 days. SCRs abide by strict ACS NSQIP definitions to classify patient comorbidities and complications.

We surveyed the ACS NSQIP database using the Current Procedural Terminology (CPT) billing code 23472 to identify all cases of TSA performed in 2013. We excluded patients with preoperative wound infection, emergent surgery, preoperative sepsis, and patients with a contaminated surgical wound to create our elective cohort.

Patients who underwent unplanned readmission within 30 days after TSA were identified. Reasons for readmission were then assessed using NSQIP criteria and ICD-9 codes associated with subsequent hospital admissions. We conducted a univariate analysis to compare patients who were readmitted with those who were not. The analysis included assessment of patient demographics including age, gender, race, body mass index, current alcohol abuse, current smoking status, recent weight loss, dyspnea, chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), hypertension, diabetes, peripheral vascular disease, esophageal varices, disseminated cancer, steroid use, bleeding disorder, dialysis, chemotherapy in the previous 30 days, radiation therapy in the previous 90 days, operation in the previous 30 days, American Society of Anesthesiologists (ASA) class, operative time, resident involvement, and patient functional status. Multiple preoperative laboratory values were analyzed including sodium, blood urea nitrogen, albumin, white blood cell count, hematocrit, platelet count, and international normalized ratio. A two-tailed Student’s t-test was used for continuous variables and a chi-square test for categorical variables. The univariate analysis identified unadjusted differences between those with an unplanned readmission after TSA compared with those not readmitted. In order to build our multivariate logistic regression model, any univariate variable with a p-value < 0.1 and that had > 80% complete chart data was identified and included. A multivariate logistic regression analysis was conducted in an attempt to control for confounders utilizing SAS (Version 9.3; SAS Institute, Cary, NC, USA). The outcome variable was unplanned hospital readmission following TSA compared with no hospital readmission. Statistical significance was considered as p < 0.05. Model quality was evaluated for calibration using the Hosmer-Lemeshow test and for discrimination with C statistics. The calibration test yielded a modified Chi-Square statistic, and a p value > 0.05 indicated that the model was appropriate and fit the data well. Data from the multivariate analysis was used to construct a predictive model for readmission.

Results

2779 patients undergoing TSA in 2013 were identified at ACS NSQIP participating hospitals with 74 (2.66%) requiring unplanned readmission within 30 days of surgery. Of these, 28 patient required return to the operating room (1.01% overall). There were 5 deaths (0.18%) within 30 days of TSA.

Causes of Unplanned Readmission

The most common causes for unplanned readmission were surgical site infection (18.6% of readmissions n=8), management of dislocations (16.3% of readmissions, n=7), and venous thromboembolism (14.0% of readmissions, n=6). Medical causes for readmission (pneumonia/pulmonary, cardiac, renal, gastrointestinal, sepsis, altered mental status, n=22) were responsible for 51.16% of unplanned readmissions. Cause of unplanned readmission was not able to be identified in 31 patients using either ICD-9 or ACS NSQIP criteria.

Risk Factors for Unplanned Readmission

Patient variables were compared between patients requiring readmission (n = 74) with those not requiring readmission (n = 2705) using univariate analysis. Readmitted patients were older (70.3 vs 73.5 years p=0.010) and more frequently diagnosed with dyspnea (p=0.03) or hypertension (p=0.01) prior to surgery. Dependent functional status (p=0.03) and ASA class (p=0.0003) were also associated with readmission (Table I). Multivariate analysis identified patient age >75 (OR 2.62, 95% Confidence Interval (CI): 1.27 - 5.41), and elevated ASA class of 3 (OR 1.79, 95% CI: 1.01 - 3.18) or 4 (OR 3.63, 95% CI: 1.31 - 10.08) as independent risk factors for unplanned readmission following TSA.

Table I.

Patient demographics and specific CT image reconstruction parameters for each case.^†

Characteristic	No Readmission (n=2705)	Readmission (n=74)	Unadjusted P Value
Age, mean (standard deviation), yrs	70.32(9.57)	73.49(10.14)	0.0050*
Age (categorical), yrs			0.0114*
<60	12.94	12.16
60-69	32.31	17.57
70-75	23.77	22.97
>75	30.98	47.3
Gender			0.8009
Male	43.36	41.86
Female	56.64	58.11
Race			0.8330
Black	3.77	4.05
White	85.99	87.84
Other	10.24	8.11
BMI(kg/m2)			0.6319
≤ 35.0	77.37	79.73
> 35.0	22.63	20.27	181
Current Alcohol Abuse	2.57	4.55	0.4508
Current Smoker	9.57	9.46	0.9734
Dyspnea	6.06	12.16	0.0322*
COPD	6.40	9.46	0.2907
CHF	0.41	1.35	0.2771
Hypertension	67.54	82.43	0.0068*
Diabetes	17.38	17.57	0.9656
PVD	0.68	4.55	0.1693
Steroid Use	5.06	4.05	1.0000
Bleeding Disorder	3.40	4.05	0.7397
ASA Class			0.0003*
1 or 2 – No or Mild disturbance	48.93	29.73
3 - Severe Disturbance	48.48	62.16
4 - Life Threatening Disturbance	2.59	8.11
Operative Time, hrs			0.1283
≤2	60.13	51.35
Length of stay, days			0.0671
> 4	3.88	8.11
≤4 days	96.12	91.89
Functional Status			0.0263*
Independent	97.42	93.15
Totally or Partially Dependent	2.58	6.85

^†Listed as percentages (%) unless otherwise noted.

^*Denotes statistical significance (p < 0.05).

Predictive Modeling

A predictive model was built from the significant parameters identified by multivariate analysis. Patients with ASA class of 4 and age >75 are 17.40 times more likely (95% CI 1.77-171.09) to be readmitted within 30 days of TSA (Table II).

Table II.

Predictive Model of Combined Patient Age and ASA Class.

Variables	Odds Ratio	95% Confidence Interval
Age 60-69 and ASA 1 or 2	0.80	0.07 – 8.93
Age 60-69 and ASA 3	2.82	0.34 – 23.41
Age 60-69 and ASA 4	5.45	0.32 – 93.65
Age 70-75 and ASA 1 or 2	3.57	0.42 – 30.08
Age 70-75 and ASA 3	4.29	0.53 – 34.43
Age 70-75 and ASA 4	7.08	0.41 – 123.12
Age >75 and ASA 1 or 2	4.34	0.54 – 34.87
Age >75 and ASA 3	5.31	0.69 – 40.68
Age >75 and ASA 4	17.40	1.77 – 171.10

Discussion

Unplanned readmissions within 30 days of TSA are rare events. We determined the incidence to be less than 3%. Medical causes for readmission account for the majority of readmissions after TSA. The most common surgical causes of readmission are SSI and management of dislocation. We have determined that several patient factors are strongly associated with 30 day readmission including age greater than 75 years and increasing comorbidity burden (as assessed by ASA class). This is the first study to systematically evaluate the incidence, causes and risk factors for readmission after TSA.

Mahoney et al. evaluated readmissions following shoulder arthroplasty between 2005-2011 from a single hospital system¹⁰. They determined the 90-day readmission rate to be 5.9%. Schairer et al. analyzed readmissions using independent state databases in 7 states from 2005 to 2010⁹. They determined the 90-day readmission rate to be 7.3%. Fehringer et al. reported a 5.6% fourteenday readmission rate in their Veteran Health Administration (VHA) system study comparing complications between shoulder and lower extremity arthroplasty. ⁸. Anthony et al¹¹, when reporting complications from the ACS NSQIP, determined the 30-day readmission rate to be 4% in 2011. Our reported 2013 readmission rate of 2.66% is one year after the implementation of CMS’s Readmissions reductions program. As such, this may be a reflection of changes in practice of surgeons and hospitals aware of the consequences of unplanned readmission. Also, readmission rates at 90 days should intuitively be higher, especially since medical causes of readmission are common.

Causes for readmission in the present study support previous findings. Postoperative infections have been cited as a common cause of readmission following TSA⁹,¹⁰. Mahoney et al. reported that SSIs were responsible for 20% of readmissions¹⁰. This compares favorably with our data as we determined that 18.6% of readmissions in the ACS NSQIP database were for treatment of SSIs. Management of dislocations was the second most common cause of readmission related to surgery in our study. Mahoney et al¹⁰ reported that instability was responsible for 8 of 40 readmissions in their cohort and Schairer et al. determined that dislocation was the second most common cause of readmission related to surgery amongst their patients⁹. Medical causes of readmission were responsible for more than 50% of the readmissions in our study, and this compares favorably with the literature⁹,¹⁰.

No prior study has evaluated risk factors specifically for 30-day readmission following TSA. Two recent reports have evaluated 90-day readmission, but it is unclear if their data is generalizable to the RRP’s 30- day readmission window. Medicaid insurance status⁹,¹², increasing age¹² and low shoulder arthroplasty volume hospitals⁹ have been associated with increased rates of readmission. Schairer et al.⁹ determined that age was not a risk factor for readmission after multivariate analysis. We determined that age greater than 75 years and ASA class of 3 or 4 were independent risk factors for readmission following TSA. ASA 4 classification is the strongest predictor of readmission, as these patients are nearly 4 times more likely to be readmitted than their otherwise similar counterparts. When combining these factors, our predictive model would suggest that patients aged >75 years and ASA class of 4 are 17.3 times more likely to undergo unplanned readmissions after TSA.

The present study has several strengths; first, the NSQIP database is robust, multicenter, and representative of general practice with an approximate 50/50 mix of private and academic institutions. It employs strict definitions for each complication, and comprehensive 30-day follow up is standardized and large patient numbers are accumulated over a short period of time. Limitations of our study include follow-up limited to a 30-day window. However, because CMS’s Hospital Quality Initiative (HQI) has established 30-day readmission as an important quality metric, this limitation is relative. Cause for readmission was not available in all cases (41%). Preoperative laboratory results were not available in 80% of our patients, and could therefore not be built into our multivariate model. Finally, the CPT code used to identify cases in the ACS-NSQIP database does not differentiate between arthroplasty technique (reverse versus anatomic), so we are unable to comment on differences in readmission based on technique.

In conclusion, unplanned readmission after shoulder arthroplasty is infrequent and may be decreasing in response to increased focus resulting from the threat of a fiscal penalty. Medical complications are responsible for more than 50% of unplanned readmissions. Older patients (>75) with an increasing number of comorbid conditions (ASA 4) are 17 times more likely to undergo unplanned readmission after surgery. Since most patients are readmitted for medical reasons, and comorbidity burden is the strongest predictor of readmission, medical optimization prior to TSA is of upmost importance. Future studies aiming to decrease infection and dislocation rates will positively influence readmissions after TSA.