# Development of a simple clinical score to estimate in-hospital adverse event risk after elective hip arthroplasty: a retrospective cohort study in a high-risk population

## TL;DR

This study created a simple clinical score to predict in-hospital complications after hip surgery in high-risk patients, helping identify those needing closer monitoring or suitable for fast-track recovery.

## Contribution

A validated, pragmatic clinical score for risk stratification in high-risk hip arthroplasty patients using LASSO regression.

## Key findings

- High-risk patients had more comorbidities like cardiac and diabetes compared to national data.
- A clinical score with a cutoff ≥2 identified patients with >7% AE risk, while scores <2 had 99% negative predictive value.
- The model showed strong discrimination (AUC 0.80) and calibration (Brier score 0.07).

## Abstract

Patients undergoing total hip arthroplasty (THA) in tertiary centres often present with complex comorbidities that increase the risk of perioperative adverse events (AE). While fast-track and outpatient protocols are expanding, reliable risk stratification tools tailored to high-risk European populations remain limited. This study aimed to (1) compare comorbidity burden in a high-risk population to national data, (2) determine incidence and risk factors for in-hospital AE and (3) develop a simple, pragmatic score to identify patients at elevated AE risk.

We retrospectively analyzed 4,101 elective primary THA cases from a German tertiary care centre (2010–2019). Comorbidity burden was quantified using the Elixhauser Comorbidities (EC) and benchmarked against national registry data (EPRD). Independent predictors of in-hospital AE were identified using multivariate logistic regression. These variables were then used to develop a pragmatic preoperative clinical risk score via LASSO regression, internally validated with 10-fold cross-validation and bootstrapping.

Compared to the national registry, our cohort showed significantly higher rates of major comorbidities, including cardiac valvular disease, diabetes, and fluid/electrolyte disorders. The overall in-hospital AE rate was 2.6%. Six comorbidities—including pulmonary circulation disorders (OR 10.7, 95% CI: 3.6–31.8)—were independently associated with AE. The derived LASSO model demonstrated strong discrimination (AUC 0.80; 95% CI: 0.75–0.84) and calibration (Brier score 0.07). A cutoff score ≥ 2 identified patients with an AE rate of > 7%, while scores < 2 corresponded to an NPV of 0.99, supporting its utility in identifying low-risk patients for fast-track pathways.

Patients treated in tertiary centres exhibit elevated comorbidity burden but maintain acceptable perioperative AE rate. A simple, validated clinical score can flag patients at substantially increased risk of in-hospital AE who may benefit from closer in-hospital surveillance and effectively identify low-risk candidates for fast-track THA pathways. Further external validation is warranted.