Development and Validation of a Primary Care Electronic Health Record Phenotype to Study Migration and Health in the UK

Abstract

International migrants comprised 14% of the UK population in 2020, but migrant health in the UK has rarely been studied at a population level using primary care electronic health records (EHRs). Given the difficulty of determining migration status using EHRs, this study developed a migration phenotype and assessed its validity. We developed a phenotyping algorithm using codes for country of birth, visa status, non-English main/first language and non-UK origin. It was applied to a Clinical Practice Research Datalink (CPRD) GOLD database of 16,071,111 primary care patients between 1997 and 2018. We compared the completeness and representativeness of the identified migrant population to Office for National Statistics (ONS) country of birth and 2011 census data by year, age, sex, geographic region of birth and ethnicity. Between 1997-2018, 403,768 migrants (2.51% of the CPRD GOLD population) were identified using the phenotype. 178,749 (1.11%) of these migrants were identified by codes indicating foreign country of birth or visa status, 216,731 (1.35%) a non-English main/first language, and 8,288 (0.05%) non-UK origin. The cohort was similarly distributed compared to ONS migration statistics in terms of sex and region of birth. Recording of migration improved from identifying approximately one-tenth of the expected proportion of migrants according to the ONS in 2004 to a quarter in 2018. Younger migrants were better represented than those aged 50 and over. The migration phenotype identified a large number of migrants and can be used to undertake large-scale migration health research in CPRD GOLD to inform healthcare policy, practice and action. While the cohort was representative of the UK migrant population in terms of sex and region of birth, migration status was under-recorded in earlier years and older ages, and future studies for these groups should therefore be interpreted with caution.

Migrants’ primary care utilisation before and during the COVID-19 pandemic in England: An interrupted time series

“…These biases would likely result in an underestimation of differences between groups. As we reported previously, the migration code list used in this study is less representative of migrants aged over 50 (20). As a result, findings concerning older migrants should be interpreted with caution.…”

“…The exposure of interest was international migration to the UK, determined by a validated migration code list comprising migration-related 'Read Version 2' codes (20). The code list produces a binary indicator (migrants, non-migrants).…”

“…In the pre-pandemic analysis, we examined effect modification by ethnicity using a two-way interaction term between migration and ethnicity, based on the 2011 Census' 18 categories grouped into 6 broader categories and derived using an ethnicity code list adapted from Pathak et al (20), with an 'Unknown' category for individuals with no ethnicity data. We calculated the relative excess risk due to interaction (RERI, additive) (24) and multiplicative effects.…”

“…We, therefore, ran multivariable models with and without adjusting for index of multiple deprivation (IMD; patient-level IMD for individuals with linked IMD data or else practice-level IMD). We stratified models by larger age groups (0-15, 16-24, 25-34, 35-49, 50-64, and 65 and over) to account for differences in representativeness of the migration code list across different age groups (20). Due to local policy interest, we conducted a secondary analysis of individuals in London only.…”

“…Recent validation of an electronic health record (EHR) code list to identify a cohort of migrants largely representative of the broader migrant population (20) presents a unique opportunity to quantify differences in primary care usage between migrants and nonmigrants in England before and during the COVID-19 pandemic.…”