Which types of emergency medical admissions may be preventable? Analyses of 132 randomly selected cases

Problem

In 2012-13, there were 5.3 million emergency admissions to hospitals, an increase of 47% over the last 15 years, costing approximately £12.5 billion and occupying 67% of hospital beds. In an attempt to reduce these admissions, lists of potentially avoidable admissions (termed ACSCs) have been developed based on consensus opinions of expert physician panels drawn from secondary and primary care, examining diagnostic disease codes. We aimed to compare this list of ACSCs used in the NHS with the conclusions of GP practice teams using significant event analysis (SEA) on emergency admissions.

Approach

We performed a study in 20 practices in East Anglia. We trained each practice to carry out SEAs in a systematic way. Hospitals randomly selected emergency admissions (one per week per practice) and practices approached patients for consent. The practice used SEA to examine (1) details of what happened before the admission (2) the people and systems that were involved (3) what might have been done differently to avoid the admission and (4) develop specific action points to promote change in practice that might help reduce future unplanned admissions. The practice reached a consensus as to whether the admission contained avoidable factors. The ICD 10 disease code for the primary cause of admission was used to identify ACSCs. An inter-rater reliability analysis using the Cohen Kappa statistic was performed to determine consistency between the lists of potentially avoidable admissions determined using SEA with those identified by the ACSC list.

Findings

SEAs were carried out on 26% of selected admissions (132 cases). Practice teams identified 13% (17) of cases as potentially avoidable, and the ACSC list categorised 28 % (36) as potentially avoidable. The inter-rater reliability was found to be Kappa = 0.253 (p=0.001), which represents only fair agreement. The ACSC list mainly comprised of respiratory admissions, Pneumonia 14 cases (39%), COPD 7 (19%), Asthma 3 (8%) and Bronchiectasis 1 (3%). Practice teams identified fewer respiratory cases as potentially avoidable, Pneumonia 4 (24%), Asthma 1 (6%), COPD 1 (6%), and Bronchiectasis 1 (6%). Practices identified a wider range of cases including admission with constipation, anxiety, cancer and intracranial haemorrhage.

Consequences

There was only a fair level of agreement on which admissions might be avoidable when comparing the ACSC list used by the NHS with the list generated by the practice team using SEA. It is plausible that case-based analysis of actual admissions may be more sensitive in detecting factors that might be associated with an avoidable admission than simple diagnostic labels. We propose further work is done using a case-based approach to review the ACSC list and to produce an updated list of conditions and factors (practitioner, patient and systems) which might identify potentially avoidable admissions.