Most patients with dementia in the UK use their local hospitals and general (family) practices throughout their illness. Linked electronic health records from primary care, hospital and death certificates records therefore provide useful information about the diagnosis and prognosis of patients who develop dementia.

In this study we will assess the validity of dementia diagnoses in linked primary care, hospital and death records, by examining the timing of important health transitions in patients with recorded dementia, and we will estimate the lifetime risk of recorded dementia in different age and sex groups

Dementia is a clinical syndrome with insidious onset that is difficult to diagnose in its earliest stages. Presentation to healthcare depends not only upon the rates of disease progression, but also on social support, recognition by clinicians, and patients' and carers' fear of diagnosis. Maintaining complete follow up in cohorts of patient with dementia is difficult, because patients with dementia are frequently lost to follow up.

Most patients with dementia in the UK use their local hospitals and general (family) practices throughout their illness. Linked electronic health records from primary care, hospital and death certificates records should therefore provide useful information about the diagnosis and prognosis of patients who develop dementia with minimal loss to follow-up rates and improved completeness of diagnosis.

Demonstrating that patients with recorded dementia have an earlier onset of typical symptoms, functional impairment and death than patients in the general population would support the veracity of diagnosed dementia recorded in electronic health records and its use as an outcome or exposure in cohort studies and for evaluating policy. Previous studies have found that dementia is poorly recorded in routine clinical practice in comparison to face-to-face studies, although this varies by setting and region. However, ascertainment may be improved by examining linked, longitudinal resources. Comparing the lifetime risk of dementia calculated from linked electronic health records with lifetime risks from other sources will also be a useful information to support the use of linked electronic health records in dementia research.

Electronic health records contain information on important health transitions in patients with dementia: from the earliest stage of the illness (depression, anxiety, memory complaints); the development of cognitive impairment that manifest as loss of capacity or missed appointments; and significant functional impairment, with admission to nursing homes or hospital admission. In this study, we will assess the validity of dementia diagnoses in linked primary care, hospital and death records, by examining the timing of important health transitions in patients with recorded dementia, and estimate the lifetime risk of recorded dementia in different age and sex groups.

Inclusion Criteria:

• Patients aged 18 years and over
• Registered with a participating general practice during the study period
• Minimum one year of records prior to study entry meeting CPRD data quality criteria
• Followed on or after 1 January 1997

Exclusion Criteria:

• Patients without recorded gender
• Less than 1 year of follow-up between study entry and date of administrative censoring

Patients registered in English general practices contributing with data to the CALIBER research platform from 1997 onward.