Insomnia symptom prevalence in England: a comparison of self-reported data and primary care records in the UK Biobank (Poster 1352)

Background: Estimates of insomnia prevalence have relied on participants self-reporting symptoms via surveys, and may be subject to recall and selection bias. Electronic health records (EHRs) offer more objective, large-scale data, but only capture events when patients visit a health care professional. It is therefore not clear how useful EHRs are in measuring insomnia prevalence.

Objectives: 1) Compare self-reported insomnia symptom prevalence to insomnia symptom prevalence recorded in the same individuals’ linked primary care records. 2) Establish whether the two datasets identify participants with similar characteristics.

Methods: We included 163,748 participants enrolled in the UK Biobank in England (aged 38–71) between 2006 and 2010 who had linked primary care records. We compared the percentage self-reporting ‘usually’ having insomnia symptoms at UK Biobank enrolment to the proportion with a Read code for insomnia symptoms in their entire primary care records prior to this baseline. We examined the overlap between these groups. We stratified prevalence by self-reported sociodemographic and lifestyle characteristics. Sensitivity analyses examined different definitions of primary care insomnia symptoms, including having a hypnotic prescription alone or having a Read code for insomnia symptoms and a concomitant hypnotic prescription within 90 days of the Read code.

Results: In the total sample, 29% self-reported having insomnia symptoms, whilst only 6% had an insomnia symptom Read code in their primary care records. Examining the overlap, only 10% of self-reported cases had an insomnia symptom Read code, whilst 49% of primary care cases self-reported having insomnia symptoms. Both self-reported and primary care-identified insomnia symptom cases were highest in females, older participants and those with the lowest self-reported household incomes. Sensitivity analyses showed 11% of the total sample had a prescription for a hypnotic in their primary care records, but only 2% had both an insomnia symptom Read code and a hypnotic prescription within 90 days of the Read code.

Conclusions: Primary care records only capture a small proportion of individuals self-reporting insomnia symptoms. As these are likely to be the most extreme cases, associations between insomnia and other health outcomes may be amplified in primary care data. However, the sociodemographic characteristics of primary-care identified and self-reported insomnia symptom cases were consistent, thus primary care records may be a valuable data source for studying population-level risk factors for insomnia. Further studies should replicate our findings in other populations and examine ways to increase discussions about sleep health in primary care.