# Case-mix Adjustment of Patient Reported Experience Measures (PREMs): a rapid review to inform benchmarking practices across inpatient health centers in Switzerland

## TL;DR

This study reviews how adjusting for patient differences can improve fair comparisons of patient experience in Swiss hospitals.

## Contribution

The paper evaluates case-mix adjustment practices for benchmarking Patient Reported Experience Measures (PREMs) in multilingual Switzerland.

## Key findings

- Seven studies support case-mix adjustment for PREMs, though variables used vary.
- Language spoken is a key case-mix variable in Switzerland but was only studied once.
- Reporting both adjusted and unadjusted scores is recommended to avoid masking quality differences.

## Abstract

The rapid review aimed to analyse current practices and recommendations regarding case-mix adjustment for benchmarking Patient Reported Experience Measures (PREMs) across inpatient health centres. Findings will inform the applicability of case-mix adjustment to PREMs in the Swiss context.

We searched PubMed, Embase, and Web of Science for studies which met the following criteria: PREMs is a main outcome, study from a European country with a national inpatient PREMs survey, study with adult patients in acute care setting, and evaluates the effect of case-mix adjustment on PREMs. Screening and appraisal were performed by an experienced epidemiologist. A narrative evidence synthesis was undertaken to address the review question, with support of tables to summarize evidence on case-mix variables and statistical methods.

Seven studies (n = 301,833) were included. All supported case-mix adjustment to some extent, though variables used for case mix varied, complicating standardization. Concerns included the risk of masking quality differences. To address this, several authors advocated reporting both adjusted and unadjusted scores. Only one study included language spoken as a case mix variable—a key factor in Switzerland.

Case-mix adjustment can enhance fairness in PREM-based benchmarking but must be applied cautiously. For multilingual contexts like Switzerland, local relevance of adjustment variables should be evaluated. A stepwise, transparent approach is recommended to avoid obscuring true performance differences.