A matched-control analysis on the effects of alcohol use disorder following primary reverse shoulder arthroplasty

Abstract

Introduction

The purpose of this study was to determine whether alcohol use disorder (AUD) patients undergoing reverse shoulder arthroplasty (RSA) have increased: 1) lengths of stay (LOS); 2) complications; and 3) costs.

Methods

The study identified 19,168 patients in the study (n = 3198) and control (n = 15,970) cohort. In-hospital LOS, 90-day complications, and costs were assessed.

Results

AUD patients had significantly longer LOS (3- vs. 2-days, p < 0.0001), higher9 0-day medical complications (49.59 vs. 14.81%; p < 0.0001), and 90-day costs of care ($18,763.25 vs. $16,035.49, p < 0.0001).

Conclusions

The study is useful as it can allow healthcare professionals to adequately counsel these patients.

1. Introduction

Alcohol use disorder (AUD) is one of the leading mental health disorders in the United States.¹ Under the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), AUD is defined as a chronic relapsing disease characterized by an impaired ability to stop or control alcohol use despite adverse social, occupational, or health consequences.¹ Risk factors for AUD within the general population include male gender, genetics, family history, culture, and ethnicity. Patients with two of the eleven DSM-5 AUD symptoms within a 12-months period meets the criteria for a diagnosis of AUD.¹ In the United States, approximately one-third of all adults 18 years and older will meet criteria for AUD.² Moreover, it is currently estimated that approximately 14 million adults meet criteria for AUD in the previous 12 months.³ Among all hospitalized patients, it is estimated that 21%–42% have AUD – a cohort that is increasingly susceptible to sepsis and pneumonia and has a higher rate of morality.⁴

The number of shoulder arthroplasties in the United States increased by 350% between 2000 and 2008, which can be largely attributed to the introduction of primary reverse shoulder arthroplasty (RSA).⁵ Primary RSA was initially used to address rotator cuff tears in the 1980s and is now a commonly performed procedure that treats multiple pathologies of the shoulder.⁶ Since RSA is commonly used to salvage complex conditions, it is not surprising that the reported complication rate is relatively high. The reported overall complication rate of primary RSA is approximately 15%, but data on risk factors for postoperative outcomes remains limited.⁷ In a study of a level III prognostic database, Somerson et al. found that alcohol use is significantly associated with a higher “tipping point” for a patient to undergo elective shoulder arthroplasty, which represented 60% of their 931-patient study.⁸ Given the increasing popularity of primary RSA as well as the high prevalence of AUD (along with its implications in early postoperative morbidity and mortality), it is of importance to orthopedic surgeons to explore the effect of AUD in surgical candidates. This knowledge can help orthopedic surgeons minimize complications and achieve favorable outcomes following primary RSA in patients with AUD. The impact of AUD on outcomes following primary RSA, however, has not yet been investigated. In a more general population of orthopedic patients, alcohol use has been demonstrated to impact postsurgical complications and increase adverse outcomes, including increased surgical bleeding time, length of stay (LOS), and postoperative infections.9, 10, 11, 12, 13, 14, 15, 16 In an observational retrospective cross-sectional study from 2005 to 2016 using a questionnaire to target preoperative comorbidity, Esteras-Seranno et al. showed that adult patients undergoing primary RSA who had a history of alcohol consumption (p = 0.028) are at increased risk for worse postsurgical outcomes.¹⁷ Similarly, Poncet et al. analyzed the Nationwide Inpatient Sample in a retrospective cohort study from 2002 through 2011 and found that patients with preexisting AUD are 2.7 times more likely to experience perioperative complications after shoulder arthroplasty (SA).⁴ Furthermore, in other studies, Humphries et al. Chalmers et al., and Singh et al. all found that, in general, there is a statistical relationship between preoperative comorbidity, (e.g. AUD), and an increase in postoperative complications, hospital mortality, hospital stay, and cost following SA and total shoulder arthroplasty (TSA).18, 19, 20 SA and TSA, however, include different surgical techniques than primary RSA, so knowledge on how AUD affects this particular procedure is critical.

The purpose of this study was to utilize a nationwide administrative claims registry and determine whether AUD patients undergoing primary RSA have increased: 1) in-hospital lengths of stay (LOS); 2) medical complications; 3) costs of care.

2. Methods

A retrospective level III query from January 1st, 2005 to March 31st, 2014 was performed using the PearlDiver (PearlDiver Technologies, Fort Wayne, Indiana) claims database. The supercomputer contains the records of over 150 million patients within the 100% Medicare Standard Analytical Files (SAF) and a private payor database. Due to the number of patients within the database, PearlDiver has been used for extensively for shoulder-related research. As with previously published studies using the database, this study consisted of all patients from the Medicare SAF portion of the database. Patients and complications were identified using International Classification of Disease, Ninth Revision (ICD-9) codes. Since the information does not contain patient information, the current study was deemed exempt from our institution's Institutional Review Board (IRB) approvals.

The database was initially queried for all patients who underwent primary RSA using ICD-9 procedural code 81.80. Patients with AUD were identified using ICD-9 diagnostic codes: 303.90 to 303.92 and 305.00 to 305.02. Using Boolean command operators, the inclusion criteria for the study group consisted of all patients undergoing primary RSA with a concomitant diagnosis of AUD versus a control cohort of patients without AUD. The study group patients were randomly matched to controls in a 1:5 ratio by age, sex, chronic obstructive pulmonary disease (COPD), diabetes mellitus, hyperlipidemia, hypertension, obesity (defined as a body mass index (BMI) greater than 30 kg/m²⁾, and tobacco use. The query yielded a total of 19,168 patients with study (n = 3198) and control (n = 15,970) cohorts. Matching was successful as there was no statistical difference between the matched cohorts.

Primary endpoints of the study were to compare in-hospital LOS, 90-day medical complications, 2-year implant-related complications; in addition to day of surgery and 90-day episode of care (EOC) costs. 90-day complications analyzed included: myocardial infarction, cerebrovascular accidents, pneumonia, respiratory failure, acute renal failure, ileus, urinary tract infection, venous thromboemboli, deep vein thromboses, and pulmonary emboli. The 90-day timepoint was chosen for the endpoints such that the dependent variables were compliant with the bundled payment care initiative (BPCI) set in place by the Centers for Medicare and Medicaid Services (CMS).²¹

2.1. Data analyses

Welch's t-tests were used to test for significance for in-hospital LOS and costs of care between the cohorts. Multivariate logistic regression analysis was used to calculate odds-ratios (OR), 95% confidence intervals (95%CI), and p-values on the effects of AUD on medical complications adjusting for age, sex, region, and Elixhauser-Comorbidity Index (ECI). Bonferroni-correction was performed as a result of the ability to use the large administrative database to establish statistical significance. A p-value <0.002 was considered statistically significant for this study. Statistical analyses were performed using the programming language R (R, Foundation for Computational Statistics, Vienna, Austria).

3. Results

The study identified 19,168 patients who had primary RSA between January 1st, 2005 to March 31st, 2014 from the PearlDiver database. The study cohort (n = 3198) included patients diagnosed with AUD who subsequently underwent primary RSA. The study cohort was then compared to the control group (n = 15,970) which included patients with no known diagnosis of AUD before undergoing primary RSA. Demographics of patients and matched controls undergoing primary RSA are shown in Table 1.

Table 1.

Demographics of patients with alcohol use disorder and matched-controls undergoing primary reverse shoulder arthroplasty. COPD = chronic obstructive pulmonary disease.

Demographics	Alcohol Use Disorder		Controls		p-value¥
	n	%	n	%
Age (Years)					0.99
<64	535	16.73	2671	16.73
65 - 69	958	29.96	4790	29.99
70 - 74	766	23.95	3822	23.93
75 - 79	588	18.39	2934	18.37
80 - 84	275	8.60	1373	8.60
85<	76	2.38	380	2.38
Sex					0.99
Female	1548	48.41	7726	48.38
Male	1650	51.59	8244	51.62
Comorbidities
COPD	124	3.88	603	3.78	0.82
Depression	1676	52.41	8381	52.48	0.99
Diabetes Mellitus	1131	35.37	5656	35.42	0.99
Hyperlipidemia	2428	75.92	12,136	75.99	0.95
Hypertension	2953	92.34	14,754	92.39	0.95
Obesity	434	13.57	2158	13.51	0.95
Tobacco	1163	36.37	5803	36.34	0.99

^¥ = Assessed by Pearson's Chi-Square Analyses.

The study demonstrated AUD patients undergoing primary RSA had significantly longer in-hospital LOS (3- vs. 2-days, p < 0.0001). Additionally, study group patients were found to have significantly higher incidence and odds (49.59% vs. 14.81%; OR: 2.71, p < 0.0001) of 90-day medical complications compared to matched controls. Specifically, AUD patients had higher rates of cerebrovascular accidents (5.38% vs. 1.47%; OR: 2.87, p < 0.0001), acute renal failure (8.35 vs. 1.98%; OR: 2.75, p < 0.0001), urinary tract infection (13.29% vs. 4.26%; OR: 2.54, p < 0.0001), respiratory failure (6.38% vs. 1.69%; OR: 2.53, p < 0.0001), pneumonia (8.10 vs. 2.28%; OR: 2.48, p < 0.0001), myocardial infarction (1.56 vs. 0.51%; OR: 2.03, p < 0.0001) in addition to other medical complications (Table 2). These findings were adjusted for age, sex, geographic region, and ECI.

Table 2.

Comparison of ninety-day medical complications amongst alcohol use disorder patients and matched-controls undergoing primary reverse shoulder arthroplasty. OR = odds-ratio, 95%CI = 95% confidence interval.

Medical Complication Assessed	AUD (%)	Controls (%)	OR	95%CI	p-valuea
Cerebrovascular Accident	5.38	1.47	2.87	2.32–3.54	<0.0001
Acute Renal Failure	8.35	1.98	2.75	2.29–3.28	<0.0001
Urinary Tract Infection	13.29	4.26	2.54	2.22–2.91	<0.0001
Respiratory Failure	6.38	1.69	2.53	2.08–3.08	<0.0001
Pneumonia	8.10	2.28	2.48	2.09–2.95	<0.0001
Ileus	0.84	0.26	2.04	1.21–3.37	0.005
Myocardial Infarction	1.56	0.51	2.03	1.39–2.93	0.0001
Venous Thromboemboli	2.60	1.08	1.96	1.48–2.47	<0.0001
Deep Vein Thromboses	2.00	0.85	1.95	1.42–2.65	<0.0001
Pulmonary Emboli	1.09	0.44	1.65	1.07–2.51	0.019
Total Medical Complications	49.59	14.81	2.71	2.46–2.99	<0.0001

^a= Adjusted for Age, Sex, Region, and Elixhauser-Comorbidity Index.

Additionally, when analyzing economic data AUD undergoing primary RSA incurred significantly higher surgical ($14,539.47 vs. $13,324.77, p < 0.0001) as well as total global 90-day EOC costs ($18,763.25 vs. $16,035.49, p < 0.0001). The comparison of day of surgery and total global 90-day episode of care costs amongst AUD patients and matched controls is shown in Fig. 1.

Fig. 1.

Comparison of day of surgery and ninety-day episode of care costs between alcohol use disorder patients and matched controls following primary reverse shoulder arthroplasty. Aud = alcohol use disorder; EOC = episode of care.

4. Discussion

Current literature on postoperative complications from RSA in patients with AUD is limited and to date, there is no specific literature pertaining to patients undergoing primary RSA. While several studies list alcohol use as a preoperative comorbidity that leads to worse outcomes after orthopedic surgery, the effects of AUD are still largely unknown especially with respect to outcomes following primary RSA. Given that approximately one-third of U.S. adults will meet AUD criteria in their lifetime and that an estimated 60% of patients electing for RSA consume alcohol, investigation of the impact of AUD on primary RSA was warranted.^2,8 After adjusting for age, sex, and medical comorbidities, this study found patients with AUD undergoing primary RSA have significantly longer in-hospital LOS, in addition to higher rates of medical complications and costs of care.

While the current study has a large population of patients to ensure adequate statistical power, use of a large administrative claims database has inherent limitations. In addition, this study was reliant on accurate ICD-9 diagnostic and procedural coding of patients within the database. Several error sources have been identified that impede accurate ICD-9 coding, and the utility of ICS-9 coding for large database studies remains unestablished.²² It is currently estimated that there are 1.3% of coding errors within the Medicare database.²³ Additionally, it is plausible that patients in the control cohorts had not yet been diagnosed with AUD or did not meet AUD criteria at the time of evaluation, potentially underestimating the true association of AUD within the outcomes measured within the study. Although prevalent in the US population, AUD is not routinely screened for by healthcare professionals.²⁴ Furthermore, we are unable to stratify patients based on severity and duration of alcohol consumption to assess the relationship of these variables with rate of complications and cost of care. This, however, could serve as the basis for future prospective studies. Despite these limitations, this study provides valuable information to clinicians on the impact of AUD on primary RSA and can facilitate future research on the relationship between AUD and primary RSA.

The results of our study are consistent with findings in current literature, but also challenge others, warranting further studies in this topic.^4,17, 18, 19, 20 Leschinger et al. investigated alcohol use as a risk factor for postoperative complications in TSA and found no significant association (p > 0.05), however this could be attributed to the study design comparing outcomes in patients undergoing primary TSA rather than primary RSA.²⁵ Many of the orthopedic perioperative complications associated with AUD are well established (e.g., pneumonia, cardiopulmonary complication, blood transfusion, postoperative infection, poor wound and bony healing, and mortality) but studies have not examined the direct relationship between AUD and medical complications, LOS and cost of care following primary RSA.26, 27, 28 This sets up the need for more studies to further understand the relationship between AUD and primary RSA. Patients with AUD undergoing primary RSA in our study were found to have significantly higher OR for medical complications. While Humphries et al. used a preoperative comorbidity index to study the effect of AUD on postoperative complications, length of hospital stays, and hospital costs for RSA patients, demonstrating an OR of 1.1 (p = 0.098) for postoperative medical complications, our OR for medical comorbidities yielded 2.71 (p < 0.0001). Although both sets of data suggest an increase in odds for medical complication, our findings show a much higher value, potentially due to our focus on AUD. Additionally, their study was conducted on TSA, while primary RSA includes a different procedural technique. While it is helpful for orthopedic surgeons to look to similar studies to gain insight on outcomes for similar procedures, further study on the effects on AUD on primary RSA specifically can enhance care. AUD patients undergoing primary RSA in our study were found to have significantly longer in-hospital LOS, which is supported by current literature but only to some extent. Humphries et al. found the LOS mean to be 4.1 days, while our results yield LOS of 3-days (p < 0.0001). Although both our results suggest an increase in LOS, ours is lower and potentially due to our focus on AUD.¹⁹ Further, AUD patients undergoing primary RSA in our study were found to have higher day of surgery and total global 90-day EOC costs, which is supported by current literature. Humphries et al. found the cost mean to be $18,549, similar to the results of the current investigation $18,763.¹⁹ Although both results suggest an increase in cost, the findings presented within this study are specific to AUD patients undergoing primary RSA. Notably, their study was conducted on TSA, while primary RSA includes a different procedural technique. It is helpful for orthopedic surgeons to look to similar studies to gain insight on outcomes for similar procedures, however it is important for there to be further study of the effects on AUD on primary RSA specifically to enhance care.

5. Conclusion

The study found that patients with a history of alcohol abuse had a significantly increased risk and odds of medical complications, in-hospital lengths of stay and increased costs of care. These results show the importance of future studies to increase our understanding of the relationship between AUD and orthopedic surgical complications. This investigation's findings provided a well-developed and statistically significant demonstration of the effect of AUD and post-surgical complications in orthopedics. This study will provide orthopedists and other healthcare professionals increased awareness of possible complications and greater understanding of post-surgical management of patients suffering from AUD.

Declaration of competing interest

None.