LEP status independently predicted decreased 1-year revision surgery (aOR 0.45) after arthroplasty

Description #

On multivariable logistic regression adjusting for age, sex, BMI, ASA physical status, smoking status, case urgency, and surgery type, limited English proficiency (LEP) was independently associated with lower odds of undergoing revision surgery at the same institution within 1 year of primary hip or knee arthroplasty, with an adjusted odds ratio of 0.45 (95% CI 0.22–0.92, P = .03) relative to English-proficient patients (Table 2) — roughly a 55% reduction in the odds of receiving a 1-year revision.

"In multivariable analyses adjusted for age, sex, BMI, ASA physical status, smoking status, case urgency, and surgery type, patients who have LEP compared to EP patients had an adjusted odds ratios of 0.45 (Confidence Interval [CI]: 0.22-0.92, P = .03) and 0.44 (CI: 0.23-0.85, P = .01) of undergoing revision surgery at the same institution within 1 year and 2 years of the initial arthroplasty procedure, respectively (Table 2)." (Kevin, 2023, p. 1431)

Methods Context #

What? #ⓘ

The observable: undergoing a revision surgery at the same institution within 1 year, modeled as a binary outcome in the multivariable regression.

"Primary outcome variables were proportions of patients who underwent revision surgery at the same institution within 1 and 2 years after the initial hip or knee arthroplasty procedure." (Kevin, 2023, p. 1430)

How? #ⓘ

Multivariable logistic regression producing adjusted odds ratios, with covariates chosen a priori; race/ethnicity was deliberately excluded from the model for collinearity with English proficiency.

"Multivariable regressions controlling for potential demographic and clinical confounders were used to calculate adjusted odds ratios of undergoing revision surgery within 1 and 2 years after primary arthroplasty for patients who have LEP, compared to English proficient patients." (Kevin, 2023, p. 1429)

Who? #ⓘ

7,985 hip and knee arthroplasty procedures at a single academic medical center (2013–2021), 577 (7.2%) classified as LEP by self-reported non-English primary language.

"A total of 7,985 hip and knee arthroplasty surgeries were included in the analysis. There were 577 (7.2%) patients who were classified as having LEP." (Kevin, 2023, p. 1429)

Other Notes #

In the same models, higher BMI, ASA physical status ≥3, current smoking, and urgent/emergent cases were associated with increased odds of revision, while older age was associated with decreased odds (Table 2). Race/ethnicity was excluded from the model for collinearity with LEP (see qualifying caveat on residual confounding).

Caveats #

• Single-institution revision capture could not detect revisions performed elsewhere (Kevin 2023 arthroplasty cohort) Revision surgeries were counted only if performed at the study's own institution; the design could not detect revisions that LEP or EP patients received at other hospitals. Because differential out-of-institution revision could plausibly differ by language group, the observed lower same-institution revision utilization among LEP patients may partly reflect care being obtained elsewhere rather than a true decrease in revision. The authors note the study site is one of the main revision centers in the area, which mitigates but does not eliminate this concern.
• No revision indications or follow-up data; differential loss to follow-up could confound revision-utilization findings (Kevin 2023) The dataset did not include indications for revision surgery, so the reasons LEP versus EP patients did or did not undergo revision cannot be characterized, and the database could not distinguish whether the lower revision rate among LEP patients reflected reduced access versus differential loss to follow-up. The authors state a more granular assessment of frequency and duration of follow-up care is needed to explain the findings. This leaves open a non-access explanation for the observed decreased-utilization signal.
• Race and ethnicity excluded from adjustment for collinearity with LEP, leaving residual confounding (Kevin 2023) Race/ethnicity was dropped from the multivariable models because of its high collinearity with English proficiency status. [Inferred: Because race/ethnicity is strongly correlated with LEP in this cohort and independently tracks documented disparities in arthroplasty revision, omitting it from the adjustment means the adjusted odds ratios cannot separate a language-proficiency effect from confounding by race/ethnicity and its correlated socioeconomic factors — residual confounding that could inflate or otherwise bias the apparent LEP–revision association.]
• LEP dichotomized by self-reported primary language, risking proficiency misclassification (Kevin 2023 arthroplasty cohort) English proficiency was operationalized as a binary variable derived solely from self-reported primary language. This risks misclassification in both directions: patients coded English-proficient may have had inadequate health or English literacy to navigate postoperative care, and some patients with limited English proficiency may have self-reported English as their primary language. Such misclassification would tend to attenuate the true association between language proficiency and revision-surgery utilization. The authors note that future studies will aim to use more objective measures of language proficiency.