Development of a data-driven case-mix adjustment model for comparison of hospital performance in hip fracture care

Franka S. Würdemann*, Arthur K.E. Elfrink, Janneke A. Wilschut, Crispijn L. van den Brand, Inger B. Schipper, Johannes H. Hegeman

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Summary: To compare hospitals’ hip fracture patient mortality in a quality of care registry, correction for patient characteristics is needed. This study evaluates in 39,374 patients which characteristics are associated with 30 and 90-day mortality, and showed how using these characteristics in a case mix-model changes hospital comparisons within the Netherlands. Purpose: Mortality rates after hip fracture surgery are considerable and may be influenced by patient characteristics. This study aims to evaluate hospital variation regarding patient demographics and disease burden, to develop a case-mix adjustment model to analyse differences in hip fracture patients’ mortality to calculate case-mix adjusted hospital-specific mortality rates. Methods: Data were derived from 64 hospitals participating in the Dutch Hip Fracture Audit (DHFA). Adult hip fracture patients registered in 2017–2019 were included. Variation of case-mix factors between hospitals was analysed, and the association between case-mix factors and mortality at 30 and 90 days was determined through regression models. Results: There were 39,374 patients included. Significant variation in case-mix factors amongst hospitals was found for age ≥ 80 (range 25.8–72.1% p < 0.001), male gender (12.0–52.9% p < 0.001), nursing home residents (42.0–57.9% p < 0.001), pre-fracture mobility aid use (9.9–86.7% p < 0,001), daily living dependency (27.5–96.5% p < 0,001), ASA-class ≥ 3 (25.8–83.3% p < 0.001), dementia (3.6–28.6% p < 0.001), osteoporosis (0.0–57.1% p < 0.001), risk of malnutrition (0.0–29.2% p < 0.001) and fracture types (all p < 0.001). All factors were associated with 30- and 90-day mortality. Eight hospitals showed higher and six showed lower 30-day mortality than expected based on their case-mix. Six hospitals showed higher and seven lower 90-day mortality than expected. The specific outlier hospitals changed when correcting for case-mix factors. Conclusions: Dutch hospitals show significant case-mix variation regarding hip fracture patients. Case-mix adjustment is a prerequisite when comparing hospitals’ 30-day and 90-day hip fracture patients’ mortality. Adjusted mortality may serve as a starting point for improving hip fracture care.

Original languageEnglish
Article number73
JournalArchives of Osteoporosis
Issue number1
Early online date27 Apr 2022
Publication statusPublished - Dec 2022
Externally publishedYes


  • Case-mix correction
  • Case-mix factors
  • Confounders
  • Database
  • Hip fractures
  • Hospital comparison
  • Mortality
  • Outcomes
  • Quality of care
  • Registry

Cite this