Identifying triggers for optimal timing of advance care planning in electronic primary health care records: a nested case-control study

Objectives To explore whether routine electronic healthcare records can be used to identify triggers for initiating advance care planning (ACP) and the optimal time window to initiate ACP. We aimed to assess the prevalence of triggers for initiating ACP as defined for use in routine data, whether their presence is associated with death, and what their position is relative to a previously identified ‘optimal time window for ACP’. Design Nested case-control study within a large dynamic population cohort dataset. Setting Primary care population-based, anonymised data extracted from GP centres in the South Holland province, The Netherlands. Participants We selected records of individuals aged ≥65 registered with their general practice from 1 Jan 2014 to 1 Jan 2017. Cases were individuals who died between 1 Jan 2017 and 1 Jan 2020. Controls were individuals who remained alive. Cases were matched by age to controls in a 1:4 ratio. Main outcome measures Outcomes include prevalence of triggers for ACP in the records of deceased and living individuals; association of the triggers’ presence with death; timing of the identified triggers in deceased individuals relative to the ‘optimal time window for ACP’. Results We included 17098 records, 4139 from deceased individuals (mean age 81) and 12959 from living individuals (mean age 79). Triggers most strongly associated with death were consultations concerning malignancy (OR 8.35, 95%CI 7.42 to 9.41), hospital admissions (OR 7.32, 95%CI 6.75 to 7.94), emergency department referrals (OR 7.11, 95%CI 6.52 to 7.75), registered home visits (OR 5.97, 95%CI 5.51 to 6.47), consultations concerning heart failure (OR 5.25, 95%CI 4.59 to 5.99), dementia (OR 4.75, 95%CI 3.99 to 6.56), opioid prescriptions (OR 4.58 (4.25–4.93), consultations concerning general decline/feeling old (OR 4.15, 95%CI 3.72 to 4.64) and skin ulcers/pressure sores (OR 4.04, 95%CI 3.55 to 4.61). Those closest to the median of the optimal time window for ACP were consultations regarding dyspnoea, general decline/feeling old, heart failure, skin ulcers/pressure sores and fever, opioid prescriptions, emergency department referrals, registered home visits and hospital admissions. Conclusions Clinical triggers for initiating ACP in general practice can be recognised within the routine electronic health records and they align well with the ‘window of opportunity’ to initiate ACP.