Algorithm-based approach to CKD identification and its association with health inequalities in coding practices
Chronic Kidney Disease
This project aims to transform CKD detection through a data-driven tool that addresses coding inaccuracies, health inequities, and economic burdens. The early stage of the study consists of three parts: 1. Development and pilot testing of a novel CKD detection algorithm designed to identify undiagnosed CKD cases and incorrect CKD coding. 2. Optimisation and validation of the algorithm to improve CKD classification accuracy by enhancing its sensitivity and specificity. 3. Analysis of kidney care inequalities by comparing coded and uncoded CKD groups, stratified by socioeconomic status, age, sex, ethnicity, other health conditions (comorbidities) and major complications related to CKD (sequelae), including cerebrovascular disease. The study does not involve any patient contact and uses existing health records only. There is no additional sample or collection of data beyond what is routinely generated during standard care. How the participants' data is handled during the study (full details can also be found in the Protocol 6.4 Overview of data): GP practices securely send NHS numbers and CKD registers for all registered patients to the research team under Section 251 approval. No other identifiers are shared. Once received, NHS numbers are immediately converted into unique pseudo IDs within secure LTHTR systems. After conversion, all processing and analysis use only pseudo IDs. The original NHS number file is stored separately in an encrypted, access-restricted area. Laboratory results from the OMOP database are linked using pseudo IDs to combine serial tests for each patient. All data processing takes place within the Lancashire and South Cumbria Secure Data Environment. A pseudonymised analysis dataset is then created containing pseudo IDs, lab results, demographics, and comorbidity information. This dataset supports CKD algorithm development, performance testing, and inequalities analysis. All analyses use pseudonymised data only. Algorithm outputs remain ps
1. Adults aged 18 or over (and no upper age limit) 2. Patients registered with GP practices located within Lancashire & South Cumbria (L&SC) participating in the study 3. No other specific inclusion criteria apply
1. Patients under the age of 18 2. Patients registered with GP practices outside Lancashire & South Cumbria (L&SC) 3. No other exclusion criteria apply