NICE CCAA Collaboration: Federated analysis scripts for ICU benchmarking using OMOP CDM

These analysis scripts are supplementary to the manuscript titled "Using the OMOP Common Data Model for a multi-registry ICU benchmarking federated analysis: lessons learned"

Description

These scripts were developed by the NICE¹ and CCAA² CC-CQRs³ to test the OMOP⁴ CDM's⁵ capability for inter-registry APACHE⁶ II benchmarking through federated analysis. The main script is benchmarking-OMOP. It aims to create demographics tables and an SMR⁷ funnel plot. It does so in the following steps:

1. Create a configuration file to specify connection parameters to the local OMOP CDM database.
2. Create a configuration file to specify both the OMOP table and unique OMOP Standardized Vocabularies ID for comorbidities relevant to the APACHE II severity of illness calculation.
3. Extract physiological data, comorbidity data, and laboratory measurement data from the local OMOP CDM database using scripts from the SeverityScoresOMOP R-package.
4. Extract outcome and diagnosis data from the local OMOP CDM database using SQL queries.
5. Perform CC-CQR specific exclusions.
6. Perform APACHE II exclusions.
7. Perform multiple imputation if applicable.
8. Calculate the SMRs for each CC-CQR in the dataset. SMRs are calculated for an imputed and not imputed dataset.
9. Create tables for demographics, data availability, and data distribution.
10. Create SMR funnel plots.

Abbreviations

1. NICE: National Intensive Care Evaluation
2. CCAA: Critical Care Asia Africa
3. CC-CQR: Critical Care Clinical Quality Registry
4. OMOP: Observational Medical Outcomes Partnership
5. CDM: Common Data Model
6. APACHE: Acute Physiology and Chronic Health Evaluation
7. SMR: Standardized Mortality Ratio