Implant-related complications in patients with opioid use disorder undergoing primary shoulder arthroplasties: a matched-controlled analysis

Abstract

Background

The purpose of this study was to investigate whether patients undergoing primary shoulder arthroplasty with opioid use disorder have higher rates of (1) implant-related complications; (2) in-hospital lengths of stay; (3) readmission rates; and (4) costs of care.

Methods

Opioid use disorder patients undergoing primary shoulder arthroplasty were queried and matched in a 1:5 ratio to controls by age, sex, and medical comorbidities within the Medicare database. The query yielded 25,489 patients with (n = 4253) and without (n = 21,236) opioid use disorder. Primary outcomes analyzed included: 2-year implant related complications, in-hospital lengths of stay, 90-day readmission rates, and 90-day costs of care. A p value less than 0.01 was considered statistically significant.

Results

Opioid use disorder patients had significantly longer in-hospital lengths of stay (3 days vs. 2 days; p < 0.0001) compared to matched controls. Opioid use disorder patients were also found to have higher incidence and odds (OR) of readmission rates (12.84 vs. 7.45%; OR: 1.16, p < 0.0001) and implant-related complications (20.03 vs. 7.95%; OR: 1.82, p < 0.0001). Study group patients also incurred significantly higher 90-day costs of care ($16,918.85 vs. $15,195.37, p < 0.0001).

Discussion

This study can be used to help further augment efforts to reduce opioid prescriptions from healthcare providers in shoulder arthroplasty settings.

Introduction

Primary shoulder arthroplasty is one of the fastest growing orthopedic procedures performed annually in the United States, yet there is limited data in the literature about the effect of opioid usage on the perioperative outcomes of this surgery.^1,2 As the baby-boomer generation continues to age, there are increasing surgical indications for these procedures, with one study projecting a 10.6% annual increase in total shoulder arthroplasties (TSAs) from 1993 to 2007, and further increases in the coming years. ³ With the surge in the number of procedures, the volume of opioid medications distributed for post-operative pain control will continue to rise as well. Partly due to the nature of the surgeries performed, orthopedic surgeons are the third highest prescribers of opioid pain relievers in the United States.^4–6 Subsequently, the rise in number of opioids prescribed has led to an epidemic of patients with higher levels of opioid dependency and significantly deleterious outcomes secondary to these medications. ⁴

The Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-V) defines opioid use disorder (OUD) as a problematic pattern of opioid abuse and opioid dependency leading to problems or distress, with at least two of the following occurring within a 12-month period: spending a great deal of time obtaining or using the opioid or recovering from its effects, cravings, or a strong desire or urge to use opioids, problems filling obligations at work, school, or home, giving up or reducing activities because of opioid use, continued opioid use despite ongoing physical or psychological problem likely to have been caused or worsened by opioids in addition to other criteria.^7,8 Multiple studies have already shown that patients with opioid abuse and dependency are more likely to develop adverse events such as myocardial infarctions, infections, constipation episodes, and other metabolic derangements leading to osteoporosis and cognitive impairments.^8–12

With the increasing number of primary TSA and reverse shoulder arthroplasty (RSA) procedures being performed within the United States, the impact of OUD following these procedures has not been properly documented. Therefore, the purpose of this study was to investigate whether OUD patients undergoing shoulder arthroplasty have higher rates of: (1) in-hospital lengths of stay (LOS); (2) readmission rates; (3) implant-related complications; and (4) costs of care.

Methods

Database

A retrospective query from 1 January 2005 to 31 March 2014 was performed using the Medicare Standard Analytical Files from the PearlDiver (PearlDiver Technologies, Fort Wayne, Indiana) supercomputer. The database contains the records of over 100 million patients from Humana and Medicare and has been used extensively for orthopedic-related research. The database utilizes International Classification of Disease, ninth revision (ICD-9) and Current Procedural Terminology coding. The database provides information such as diagnoses, procedures, complications, and discharge disposition, in addition to other metrics. Since the database contains deidentified information, our study was exempt from our institution’s Institutional Review Board approvals.

Study population

The inclusion criteria consisted of all patients undergoing either TSA or RSA with OUD. Patients without OUD undergoing TSA or RSA served as controls. The database was first queried for all patients undergoing TSA or RSA using ICD-9 procedural codes 81.80 and 81.88. Patients with OUD were queried using ICD-9 diagnosis codes 304.00 to 304.02 and 305.50 to 305.52. Study group patients were randomly matched to controls in a 1:5 ratio by age, sex, and medical comorbidities: alcohol abuse, chronic obstructive pulmonary disease, diabetes mellitus, hyperlipidemia, hypertension, rheumatoid arthritis, and tobacco use. The query yielded 25,489 patients with (n = 4253) and without (n = 21,236) OUD undergoing primary shoulder arthroplasty. Matching was successful as there was no statistical difference between the matched cohorts (Table 1).

Table 1.

Demographics of patients with and without opioid use disorder undergoing shoulder arthroplasty.

	Opioid use disorder		Controls
Demographics	n	%	n	%	p value
Age (tears)		4253		21,236	0.99
<64	1662	39.08	8287	39.02
65–69	1069	25.14	5339	25.14
70–74	753	17.71	3766	17.73
75–79	473	11.12	2365	11.14
80–84	212	4.98	1059	4.99
85<	84	1.98	420	1.98
Sex					0.99
Female	2882	67.76	14,390	67.76
Male	1371	32.24	6846	32.24
Comorbidities
EtOH abuse	727	17.09	3614	17.02	0.99
COPD	302	7.10	1501	7.07	0.99
Diabetes mellitus	1925	45.26	9606	45.23	0.99
Hyperlipidemia	3160	74.30	15,779	74.30	0.99
Hypertension	3987	93.75	19,914	93.77	0.99
Obesity	772	18.15	3839	18.08	0.99
Tobacco	1673	39.34	8359	39.36	0.99

EtOH: alcohol; COPD: chronic obstructive pulmonary disease.

Outcomes analyzed

Primary outcomes analyzed included implant-related complications, in-hospital LOS, readmission rates, and costs of care. Two-year implant related complications analyzed and compared included: peri-prosthetic fractures, fractured prosthetic implants, per-prosthetic joint infections, mechanical loosening, prosthetic dislocations, and mechanical complications. Reimbursements were used as a substitute for cost as it is a more accurate representation of what providers are paid from the providers. Ninety days were chosen to be compliant with the bundled payment care initiative set by the Centers for Medicare and Medicaid Services.

Data analysis

Patient demographics of age, sex, and medical comorbidities were compared by Pearson’s X² analyses. Multivariate logistic regression analyses were used to calculate odds ratios (OR) and 95% confidence intervals (95%CIs) on the effects of OUD on implant-related complications and readmission rates, adjusting for age, sex, region, and Elixhauser-Comorbidity Index. Welch’s t-test was used to test for significance in LOS and costs of care between the OUD and non-OUD cohort. Statistical analyses were performed using the programming language R (R Foundation for Statistical Computing, Vienna, Austria). Due to the ease of finding statistical significance with large administrative databases, Bonferroni-corrections were performed to reduce the probability of a type I error. Thus, a p value less than 0.01 was considered statistically significant.

Results

Two-year implant-related complications

OUD patients undergoing shoulder arthroplasty were found to have significantly higher incidence and odds (20.03 vs. 7.95%; OR: 1.82, 95%CI: 1.64–2.03; p < 0.0001) of implant-related complications two-years following the index procedure. The study found OUD patients had significantly higher incidence and odds of mechanical complications of prosthetic joint peri-prosthetic fractures (2.00 vs. 0.58%; OR: 2.53, 95%CI: 1.87–3.41; p < 0.0001), broken prosthetic implants (1.72 vs. 0.46%; OR: 2.38, 95%CI: 1.71–3.31; p < 0.0001), peri-prosthetic joint infections (5.15 vs. 1.77%; OR: 2.10, 95%CI: 1.75–2.52; p < 0.0001), mechanical loosenings of prosthetic joints (3.27 vs. 1.49; OR: 1.71, 95%CI: 1.37–2.12; p < 0.0001), dislocations of prosthetic joints (5.83 vs. 2.90%; OR: 1.50, 95%CI: 1.27–1.76; p < 0.0001), and mechanical complications of prostheses (2.00 vs. 0.58%; OR: 2.53, 95%CI: 1.87–3.41; p < 0.0001; Table 2).

Table 2.

Comparison of implant-related complications among opioid use disorder patients and matched controls within 2-years following primary total shoulder arthroplasty.

Implant complication	OUD (%)	Controls (%)	OR	95%CI	p value*
MC of prosthetic joint	2.00	0.58	2.53	1.87–3.41	<0.0001
Broken prosthetic implant	1.72	0.46	2.38	1.71–3.31	<0.0001
Peri-prosthetic joint infection	5.15	1.77	2.10	1.75–2.52	<0.0001
Peri-prosthetic fractures	1.65	0.56	1.91	1.38–2.62	<0.0001
Articulating bearing wear	0.42	0.19	1.89	1.02–3.39	0.035
Mechanical loosening	3.27	1.49	1.71	1.37–2.12	<0.0001
Dislocation of prostheses	5.83	2.90	1.50	1.27–1.76	<0.0001
Total complications	20.03	7.95	1.82	1.64–2.03	<0.0001

OUD: opioid use disorder; OR: odds-ratio; 95%CI: 95% confidence interval; MC: mechanical complication.

^*Adjusted for age, sex, region, and Elixhauser-Comorbidity Index Score.

In-hospital lengths of stay and readmission rates

OUD patients had significantly longer in-hospital LOS (3 days vs. 2 days; p < 0.0001) compared to matched controls. Ninety-day readmission rates were significantly higher (12.84 vs. 7.45%; OR: 1.16, 95%CI: 1.04–1.30, p < 0.0001) in OUD patients following primary shoulder arthroplasties compared to matched controls.

Costs of care

OUD patients incurred significantly higher day of surgery costs ($13,073.09 vs. $12,047.03, p < 0.0001) compared to non-OUD patients following shoulder arthroplasties. Ninety-day costs of care were also significantly higher in study group patients ($16,918.85 vs. $15,195.37, p < 0.0001) compared to matched controls.

Discussion

While opioid prescription rates from the general orthopedic community are decreasing due to the introduction of strict protocols, the importance of recognizing the impact of OUD following shoulder arthroplasties continues to be of great importance. This continues to remain a societal problem as 99% of the hydrocodone supply is consumed in the United States, even though this country accounts for only 5% of the world population.^4,13 Given that this disease is still very prevalent in the population, it is important to identify its impact on shoulder arthroplasty, as previous studies have shown that 58% of all total joint arthroplasty complications were derived from opioid-related adverse drug events.^13,14 Considering that there is minimal literature focused specifically on shoulder arthroplasty, we were able to demonstrate that patients with OUD undergoing primary shoulder arthroplasty have higher rates of implant-related complications, longer in-hospital LOS, higher rates of readmissions and higher day of surgery, and 90-day costs of care compared to matched controls.

The current study is not without limitations, most of which are inherent to the use of an administrative database study. It is possible that patients in the control group have not been diagnosed with OUD, potentially underestimating the impact of OUD on the dependent variables tested within the study. Additionally, the authors of the study only analyzed a single insurance database, and the results of the study might not be a true cross-sectional representation on the impact of OUD following shoulder arthroplasty. ¹⁵ Furthermore, the results of the study are reliant on accurate diagnostic and procedural coding, and it is currently estimated that there are up to 1.3% of coding errors within the Medicare database. ¹⁶ Finally, the terminology in the database does not contain explicit definitions of what are inferred by terminology such as mechanical loosenings or mechanical complications of prostheses; we are unable to determine whether or not these are representing more specific complications such as aseptic necrosis or loosening of the humeral implant/glenoid base plate. Despite these limitations, the study is the first, to best of the authors’ knowledge, of investigating the impact of OUD on shoulder arthroplasty outcomes in a large well-powered population while controlling for multiple patient confounders.

There is a paucity of information describing the rate of implant complications in patients with and without OUD in the literature for shoulder arthroplasty. Vakharia et al. examined implant complications in the setting of total knee arthroplasty (TKA) for patients with and without OUD and discovered that patients with a diagnosis of OUD were more likely to require a revision procedure, sustain a periprosthetic fracture, have a prosthetic dislocation and loosening of the prosthesis compared to the normal controls. ¹⁷ The elevated risk for fractures has been shown in other studies, as Saunders et al. found that patients aged 60 and older who used opioids for chronic non-cancer pain were at 28% increase in fracture risk compared to those patients not taking opioids. ¹⁸ One meta-analysis of six studies that examined the effect on opioids and fracture risk demonstrated a pooled relative risk of 1.32 and 1.42 elevated fracture risk among those patients taking chronic opioids. ¹⁹ Other studies have placed the hazard ratio for increased fracture risk at close to 4.9 when comparing patients taking opioids to age-matched controls taking NSAIDS. ²⁰ In another study of patients undergoing a primary total hip arthroplasty (THA) or TKA, patients who were either short-acting opioid users (hydromorphone, morphine, hydrocodone, oxycodone) or long acting opioid users (i.e. MS-contin, oxycontin, methadone), short acting chronic opioid users had a 6.8% chance of sustaining a periprosthetic fracture compared to 1.7% for non-opioid users, and long-term opioid users were 6.15 more likely to develop a complication post-operatively than non-users. ²¹

The reason for increased fracture risk is likely multifactorial. Long-term opiate use has been shown to have deleterious effects on motor function, as Mintzer et al. demonstrated that daily dosing of morphine in opioid dependent volunteers showed a significant decrease in psychomotor function, coordination, and speed on a variety of tasks. ²² In addition to decreased cognitive function, opioids can contribute to increased fracture risk through central nervous system effects such as dizziness and reduced alertness, subsequently leading to increased falls. ⁹ To compound this issue, opioids have been theorized to impair osteoblast function and reduce osteocalcin synthesis leading to decreased bone formation.^23,24 In particular, osteoporosis classically disproportionally affects females compared to males, and recent literature suggests that through inhibition in the central nervous system, women who chronically consumed sustained-release opioids for non-cancer pain were found to have profound inhibition of ovarian sex hormones leading to decreased estrogen levels.^23,25 Estrogen affects bone development by reducing cytokine reduction that induces bone resorption, altering production of RANK ligand (to stimulate osteoclast development), as well as inhibit differentiated osteoclasts.^23,26 Without the osteoprotective effects of estrogen, increased levels of osteoporosis and decreased motor coordination all predispose chronic opioid users to sustaining elevated levels of periprosthetic fractures compared to a opioid naïve cohort after arthroplasty surgery.

The immunosuppressive aspect of chronic opioid consumption is particularly important to examine in the setting of arthroplasty surgery. Periprosthetic joint infections (PJIs) can have devastating consequences not only for the health of the patient, but also from the added cost burden on the healthcare system in an attempt to remove the offending agent. Bell et al. identified in their review of 23,754 patients who underwent a TKA or THA procedure at their institution over a 12-year period, patients with a history of opioid uses were 1.53 times more likely to develop a PJI than the opioid naïve group within 2 years after surgery. ²⁷ Opioids suppress both the adaptive and innate immune system responses via various pathways and can also increase the activation of the hypothalamic–pituitary–adrenal axis, leading to increased production of corticosteroids and inflammatory cytokines.^27,28 In rodent models, administration of morphine diminished the activity of natural killer cells, T cells, B cells, macrophages, and polymorphonuclear lymphocytes, as well as suppressed the formation of various cytokines that allow the body’s immune system to not only identify potential pathogens, but respond appropriately with the tools necessary to destroy disease causing organisms.^29–31 With a reduced ability to fight off infections due to a hindered immune system, patients are at an elevated risk of developing a PJI and possibly require a revision surgery to eliminate the pathogen.

A significant finding of this study that is concerning for any shoulder surgeon is the elevated risk of dislocations in the chronic opioid user population. While we are unable to draw a direct causative link for the reason why this might occur, it is possible to speculate on the potential predisposing factors. In a study of inpatient falls after shoulder arthroplasty, Menendez et al. found that patients with an OUD had an OR of 3.33 (95% CI, 1.98–5.59) of sustaining a fall after a shoulder arthroplasty, which was the second highest risk factor identified leading to a fall behind having a fluid or electrolyte disorder. ³² Even though this study was a review of only TSA procedures, it is reasonable to assume that this could be extrapolated to the similar RSA procedure as well. In a similar light, Morris et al. found in their analysis that patients with a history of pre-operative opioid use were associated with lower shoulder function scores at the conclusion of their study period and had lower pre-operative baseline scores than patients without a history of pre-operative opioid use. ⁵ In combination with the above described motor and cognitive deficits, it is possible that the average OUD patient is predisposed at putting themselves in compromising anatomic positions due to reduced shoulder function in the post-operative period that would lead to a higher incidence of a prosthetic dislocation.

While there are limited results in the literature regarding the effect of chronic opioid use on LOS in shoulder arthroplasty settings, our results are consistent with those findings in other orthopedic and hospital settings. In a retrospective institutional study of 176 patients undergoing either a TKA or THA, Sing et al. compared in-hospital LOS, readmissions, and discharge disposition between patients taking long-acting opioids, short-acting opioids, to non-users; they found patients consuming opioids preoperatively had on average 1.4 days longer in-hospital LOS (p = 0.005) compared to controls. ³³ In a review of the National Inpatient Sample dataset from 2002 to 2011, Menendez et al. found that 15,901 patients identified with OUD undergoing a major elective orthopedic surgery were 2.5 times more likely to have a prolonged LOS compared to the opioid naïve cohort. ³⁴ On the other hand, specifically in the setting of shoulder arthroplasty, a retrospective study of 262 patients undergoing shoulder arthroplasty (TSA = 170; RSA = 92), Cheah et al. found patients with chronic preoperative opioid use was not associated with longer in-hospital LOS, complications, or readmission rates. ³⁵ However, a large limitation to this study was due to the small sample size which could potentially be underpowered and making it difficult to find any possible association between opioid use and the dependent variables we measured in our study. ³⁵

Readmission rates and costs of care were elevated compared to opioid naïve cohorts as well in our study. In looking specifically at 30-day readmission rates after major operating room procedures, Gupta et al. determined that the most common cause for readmission in patients with a history of OUD was infection (namely cellulitis, other subcutaneous infections, bone and joint infections, endocarditis, central nervous system abscesses), nearly 9% higher than patients without a history of OUD. ³⁶ This evidence certainly supports the elevated risk of periprosthetic infections that was found in our study. In addition, the hospital stay was longer at both the initial admission (6 vs. 4 days, p < 0.0001) and at readmission (6 vs. 5 days, p < 0.0001), and OUD patients also had higher costs of care ($18,528 vs. $16,617, p < 0.0001) after adjusting for covariates. ³⁶ In a study examining patients undergoing primary 1-2 level posterior lumbar spinal fusion, Jain et al. found that patients with a pre-operative diagnosis of chronic opioid therapy were 1.15 times more likely to be readmitted for all complications within 90 days compared to matched controls. ³⁷ In a literature review of the clinical and economic burden of opioid abuse, Meyer et al. found that in one study population using Florida Medicaid administrative claims, opioid abuse patients incurred $24,724 in total direct health care costs per patient compared to $11,541 for matched controls.^38,39 In a separate cohort, privately insured opioid abuse patients incurred $24,193 in total direct health care costs while non-opioid controls incurred only $3467 in total direct health care costs.^38,39 Finally, in another study that examined opioid users in varying stratification by their use, patients with a diagnosis of OUD were estimated to have a 28% greater total care costs after matching and adjusting for confounders. ⁴⁰

Conclusion

In conclusion, this study demonstrates that patients with OUD undergoing primary shoulder arthroplasty are at an elevated risk for implant complications such as peri-prosthetic fractures, broken prosthetic implants, peri-prosthetic infections, and dislocations of their operative joint. They were also found to have greater LOS, 90-day readmission rates, and total episode of care costs. The results of this study emphasize a need to continue to identify and treat the underlying cause of OUD prior to surgical intervention in many cases of shoulder arthritis, as our study shows that these patients have worse outcomes compared to patients without a history of OUD.

ORCID iDs

Samuel J Swiggett https://orcid.org/0000-0001-8551-4884

Rushabh M Vakharia https://orcid.org/0000-0003-2008-0821