Shoulder arthroplasty (SA), including total SA (TSA) and reverse TSA, is an effective surgical treatment for fracture and primary or secondary degenerative disease of the shoulder.1 Over the past few decades, use of SA has increased dramatically, from about 5000 cases in 1990 to 7000 in 2000 and more than 26,000 in 2008.1,2
Complications associated with SA generally are classified as perioperative (occurring during the operative index) or long-term (postdischarge).3 Long-term complications include implant loosening, instability, revision, infection, rotator cuff tear, neural injury, and deltoid detachment.1,4,5 Perioperative complications, which are less commonly reported, include intraoperative fracture, infection, neural injury, venous thromboembolic events (VTEs, including pulmonary embolism [PE] and deep vein thrombosis [DVT]), transfusion, and death.3,6-10
SA is an attractive treatment option for patients with rheumatoid arthritis (RA), as the effects of pain on these patients are greater in the shoulder joint than in any other joint.11 Patients with RA pose unique orthopedic surgical challenges, including any combination of decreased bone mineralization, poor capsular tissue integrity, and osteonecrosis.3,12 In addition, RA patients may be taking immunosuppressive medications that have severe side effects, and they may require multiple surgeries.12,13 These factors predispose patients with RA to complications that include infection and wound dehiscence.3,5,12-14
The complex nature of RA has prompted investigators to examine outcome measures in this patient group. Hambright and colleagues3 used the Nationwide Inpatient Sample (NIS) to examine perioperative outcomes in RA patients who underwent TSA between 1988 and 2005.3 They found that TSA patients with RA had shorter and less costly hospital stays and were more likely to have a routine discharge.3 Using the same patient population drawn from the period 2006–2011, we conducted a study to determine if this unexpected trend persists as the number of TSAs and quality of postoperative care continue to increase. Given the potential for anemia of chronic disease and the systemic inflammatory nature of RA, we hypothesized that the perioperative complication profile of RA patients would be worse than that of non-RA patients.
Materials and Methods
NIS data were acquired for the period 2006–2011. The NIS is the largest publicly available all-payer inpatient database, with a random 20% sample of about 1000 US hospitals accounting for 7 to 8 million inpatient stays. The database supplies weights used to estimate national totals, at about 35 million inpatient visits per year. NIS inpatient data are limited to the operative index. Postdischarge information is not available. The NIS is managed by the Healthcare Cost and Utilization Project, which is sponsored by the Agency for Healthcare Research and Quality. The quality of NIS data is assessed and validated by an independent contractor. NIS data have been widely used to examine perioperative outcomes.15-17
NIS data cover patient and hospital demographics, hospital length of stay (LOS), discharge status, payer information, charges, and perioperative outcomes and procedure/diagnosis codes (ICD-9; International Classification of Diseases, Ninth Revision18).
As our Institutional Review Board (IRB) reviewed the database and determined the project was not human subject research, IRB involvement was not required. This study paralleled successful efforts with similar RA and non-RA patients who had shoulder and elbow surgery.3,19 SA patients were identified by ICD-9 procedure code 81.80, but this code does not specify whether the prosthesis was unconstrained, semiconstrained, or constrained. ICD-9 coding also does not specify whether the TSA was traditional or reverse. Patients with RA were identified by ICD-9 diagnosis codes 714.0, 714.1, and 714.2. Patients without one of these codes were placed in the non-RA cohort. Patients with codes associated with pathologic fractures secondary to metastatic cancer or bone malignant neoplasm as a secondary or primary diagnosis and patients who had revision surgery indicated by code 81.83 were excluded, as they have a disproportionately higher comorbidity burden.
After each cohort was defined, demographic data (age, sex, race, income quartile based on ZIP postal code) were compared, as were data on primary payer, hospital demographics, LOS (≤5 days, defined as perioperative index), discharge type, inflation-adjusted charges in 2014 dollars based on the Consumer Price Indexes (http://www.bls.gov/cpi/), and mortality. Perioperative complications—respiratory, gastrointestinal, genitourinary, accidental puncture/laceration, central nervous system, wound dehiscence, device-related (including embolism, fibrosis, hemorrhage, pain, stenosis, or thrombus caused by any device, implant, or graft), cardiac, hematoma/seroma, acute respiratory distress syndrome, postoperative shock, VTE, postoperative infection complications, and intraoperative transfusions—were considered using ICD-9 codes (996.X-999.X and 99.X, respectively).20 Although commonly used to determine perioperative comorbidity burden using ICD-9 coding, the modified Charlson index was not considered because RA is a component of the index and would therefore bias the variable.3,21
Statistical analyses, including χ2 tests and 2-sample t tests, were performed for categorical and continuous variables, respectively. P < .05 was considered significant. Fisher exact test was used for cohorts with fewer than 5 occurrences. Multivariate logistic regression models were then calculated to determine the effect of RA on different outcomes and complications, with age, race, sex, hospital region, hospital type, number of hospital beds, primary payer, and hospital ownership as covariates. Statistical analyses were performed using the R statistical programming language.22