Population and patient factors affecting AED attendance in London

Problem

The prevalence of multimorbidity rises with an aging population, and is independently associated with social deprivation. In turn deprivation is linked with increased use of Accident and Emergency Department (AED) services, with expectations that improved access to primary care can moderate activity. Primary care remediable aspects include ability to get an appointment, and knowledge of out-of-hours GP services, but recent studies suggest these factors contribute only marginally to predictive explanatory models. We use linked patient level primary and secondary care data from 136 practices in a multi-ethnic inner east London population of 819,590 to identify demographic and clinical factors which predict AED use.

Approach

Primary and secondary demographic, attendance, diagnostic (QOF conditions) and clinical data were extracted for 819,590 cases for two years prior to December 2015 and linked with AED attendance data using hospital Secondary Use Service (SUS) data. Distance to the nearest AED for each patient was calculated from the centre of the LSOA of residence. GP Patient Survey access questions were linked to practices.

Findings

Crude adult AED attendance rate increased with deprivation. Comparing the most deprived to the least deprived population quintile, OR 1.42 (95% CI 1.39-1.45).Adjusted multilevel analysis for adults showed a progressive rise in AED attendance with increasing number of long term conditions (LTCs). Comparing two LTCs with no conditions OR 1.28 (CI 1.25-1.31); four or more conditions in the previous year compared to no conditions OR 2.66 (CI 2.55-2.78). Increasing annual GP surgery consultations predicted AED attendance, comparing zero with more than two consultations OR 8.48 (CI 8.32-8.62). Smoking (OR 1.33) and age were also important predictors of adult attendance. In the adjusted analysis deprivation remained significant, comparing the most deprived to the least deprived quintile OR 1.15 (CI 1.13-1.18). Distance from the AED unit and ethnicity remained significant, but less important, drivers of attendance. Patient reported ‘access’ scores based on GPPS data were not a significant predictor. For children younger age, male gender, White ethnicity and higher annual GP consultation rates were key predictors of high attendance rates

Consequences

Using patient level data, rather than practice data, we demonstrate that the population burden of multimorbidity is the strongest clinical predictor of AED attendance, which in turn is patterned by social deprivation. Low use of the GP surgery is associated with low attendance at AED. Unlike other studies we found that patient experience of GP access, as measured by the GPPS, was not a significant predictor of use. The implications for provision of AED units, and potential primary care factors in reducing AED attendance will be discussed