What Are the Migration Patterns for U.S. Primary Total Joint Arthroplasty Patients?

Background: Evaluation of total joint arthroplasty (TJA) patient-reported outcomes and survivorship requires that records of the index and potential revision arthroplasty procedure are reliably captured. Until the goal of the American Joint Replacement Registry (AJRR) of more-complete nationwide capture is reached, one must assume that patient migration from hospitals enrolled in the AJRR to nonAJRR hospitals occurs. Since such migration might result in loss to followup and erroneous conclusions on survivorship and other outcomes of interest, we sought to quantify the level of migration and identify factors that might be associated with migration in a specific AJRR population.

Questions/purposes: (1) What are the out-of-state and within-state migration patterns of U.S. Medicare TJA patients over time? (2) What patient demographic and institutional factors are associated with these patterns?

Methods: Hospital records of Medicare fee-for-service beneficiaries enrolled from January 1, 2004 to December 31, 2015, were queried to identify primary TJA procedures. Because of the nationwide nature of the Medicare program, low rates of loss to followup among Medicare beneficiaries, as well as long-established enrollment and claims processing procedures, this database is ideal for examining patient migration after TJA. We identified an initial cohort of 5.33 million TJA records from 2004 to 2016; after excluding patients younger than 65 years of age, those enrolled solely due to disability, those enrolled in a Medicare HMO, or residing outside the United States, the final analytical dataset consisted of 1.38 million THAs and 3.03 million TKAs. The rate of change in state or county of residence, based on Medicare annual enrollment data, was calculated as a function of patient demographic and institutional factors. A multivariate Cox model with competing risk adjustment was used to evaluate the association of patient demographic and institutional factors with risk of out-of-state or out-of-county (within-state) migration.

Results: One year after the primary arthroplasty, 0.61% (95% confidence interval [CI], 0.60-0.61; p < 0.001 for this and all comparisons in this Results section) of Medicare patients moved out of state and another 0.62% (95% CI, 0.60-0.63) moved to a different county within the same state. Five years after the primary arthroplasty, approximately 5.41% (95% CI, 5.39-5.44) of patients moved out of state and another 5.50% (95% CI, 5.46-5.54) Medicare patients moved to a different county within the same state. Among numerous factors of interest, women were more likely to migrate out of state compared with men (hazard ratios [HR], 1.06), whereas black patients were less likely (HR, 0.82). Patients in the Midwest were less likely to migrate compared with patients in the South (HR, 0.74). Patients aged 80 and older were more likely to migrate compared with 65- to 69-year-old patients (HR, 1.19). Patients with higher Charlson Comorbidity Index scores compared with 0 were more likely to migrate (index of 5+; HR, 1.19).

Conclusions: Capturing detailed information on patients who migrate out of county or state, with associated changes in medical facility, requires a nationwide network of participating registry hospitals. At 5 years from primary arthroplasty, more than 10% of Medicare patients were found to migrate out of county or out of state, and the rate increases to 18% after 10 years. Since it must be assumed that younger patients might exhibit even higher migration levels, these findings may help inform public policy as a "best-case" estimate of loss to followup under the current AJRR capture area. Our study reinforces the need to continue aggressive hospital recruitment to the AJRR, while future research using an increasingly robust AJRR database may help establish the migration patterns of nonMedicare patients.

Level Of Evidence: Level III, therapeutic study.