Validity of smoking prevalence estimates from primary care electronic health records compared with national population survey data for England, 2007 to 2011.

PURPOSE: Primary care electronic health records (EHRs) are increasingly used as a resource for epidemiological research. Cigarette smoking is an important variable in many epidemiological studies. We evaluated the validity of smoking records in primary care EHRs by comparing estimates for smoking prevalence from primary care EHRs with national health survey data.
METHODS: Data were analysed for adults over 30 years of age from the Clinical Practice Research Datalink (CPRD) in comparison with data from the Health Survey for England between 2007 and 2011. Electronic health records in the CPRD were searched for records of smoking status and smoking cessation treatment. Annual age- and sex-standardised prevalence of current-, non- and former smoking were calculated, and compared with equivalent data from the Health Survey for England (HSE).
RESULTS: The difference between estimates of current smoking in CPRD and HSE was generally <1% from 2007 to 2011. In 2011, the prevalence of current smoking in men was: CPRD 24.3%, HSE 24.2%. The mean difference was 0.1% (95% confidence interval −1.5 to 1.7%). In women, current smoking prevalence was CPRD 20.3%, HSE 19.0%; mean difference 1.3% (0.0 to 2.6%). Estimates for former smoking were lower in CPRD than HSE for men (CPRD 26.7%, HSE 31.3%) and women (CPRD 22.9%, HSE 25.0%).
CONCLUSIONS: Prevalence estimates for current smoking, and non-smoking, from primary care EHRs are similar to those from nationally representative surveys, but former smoking may be under-recorded.