Background Ambulance clinicians use pre-alert calls to advise Emergency Departments (EDs) about patients who may require immediate assessment on arrival. Despite pre-alerts playing a key role in the transfer of care between ambulance and ED clinicians, there is a lack of understanding about which patients should receive a pre-alert and potential variation in practice.

Methods We created a linked dataset from 3 ambulance services comprising of 999 call, electronic patient record and ambulance staff demographic and shift pattern data for all ED ambulance conveyances between July 2020–June 2021. We also compared pre-alert activity with the AACE/RCEM guidance on pre-alerts. We undertook descriptive analysis using the statistics package R.

Results Pre-alerts were recorded in 10.5% of conveyances (142,795/1,363,274) with significant variation in pre-alert rates between ambulance services (8.2%–15.0%) and between receiving hospitals. Paramedics pre-alerted 10.7% of their conveyances (107,309/1,002,733) with non-registered clinician staff pre-alerting 9.8% of their conveyances (35,486/360,541).

Patients who met the AACE/RCEM pre-alert criteria were more likely to be actually pre-alerted when they met non-physiological criteria (e.g. stroke), compared to physiological criteria (e.g. respiratory rate) (51.9% vs 28.5%). Only a third of cases that met any AACE/RCEM physiological criteria for pre-alert were actually pre-alerted (103,066/323,971). Around 15% (20,522/142795) of pre-alerts were for patients in low priority triage categories (3-5).

Conditions with the highest numbers of pre-alerts included suspected sepsis (21,479/142,795, 15.0%), unspecified medical conditions (16,884, 11.8%) and acute stroke (14,869, 10.4%). Covid-19, respiratory problems and lower respiratory tract infections made up a further 15.4% of pre-alerts. Highest pre-alert rates were found with suspected sepsis (21,749/34,821,62%), acute stroke (14,869/27,783, 54%) and STEMI (3,687/6,207, 59%).

Conclusion There is significant variation in patients who are pre-alerted. Improved clarity of criteria for pre-alert may help. Variation may be partly due to under-documentation of pre-alerts.