Table 2. Unadjusted and Adjusted Association of Limited English Proficiency With Postoperative Outcomes for Patients Undergoing Lower Extremity Total Joint Arthroplasty.
| Outcome variables | Univariate analysesa | Adjusted multivariable modelsb | ||||
|---|---|---|---|---|---|---|
| Total (N = 4721) | English proficient (n = 4343) | Limited English proficiency (n = 378) | P value | Adjusted IRR or OR (95% CI) | P value | |
| Length of stay, median (IQR), d | 2 (1-3) | 2 (1-3) | 3 (2-4) | <.001c | 1.15 (1.07-1.25)d | <.001 |
| Total costs, median (IQR), $ | 14 264 (12 184-19 388) | 14 179 (12 142-19 207) | 15 387 (13 016-21 566) | <.001c | 1.08 (1.04-1.12)d | <.001 |
| Discharge to skilled facility, patients, No. (%) | 1050 (22.3) | 889 (20.5) | 161 (42.6) | <.001e | 1.41 (1.03-1.93)f | .03 |
| 30-d Readmission, patients, No. (%) | 342 (7.2) | 312 (7.2) | 30 (7.9) | .59e | 0.80 (0.49-1.28)f | .35 |
Abbreviations: IRR, incidence rate ratio; OR, odds ratio.
Univariate data are presented as median (IQR) for continuous measures or as frequency (percentage) for binary measures.
Negative binomial regression models accommodated right skew exhibited by length of stay and total cost; the resulting IRRs are given. Logistic regression analyses determined ORs for dichotomous outcome variables. IRRs and ORs were adjusted for race and ethnicity, age, sex, primary insurance, American Society of Anesthesiologists status (proxy for medical comorbidities), body mass index, surgical case class, case length, and estimated blood loss. Length of stay and cost rate ratios additionally were adjusted for disposition location. Reference group is patients with English proficiency.
P values were calculated with Wilcoxon rank-sum test.
Data are IRRs.
P values were calculated with Pearson χ2 test.
Data are ORs.