The Covid-19 pandemic has significantly impacted the operation of NHS trusts in the last six months and will continue to do so for the foreseeable future. Given the nature of the virus, there has been notable variation in the impact faced by trusts due to differences in patient casemix, hospital capacity and regional patterns in both Covid-19 spread and population vulnerabilities.
The variation in hospital’s activity relating to Covid-19 has in turn had an impact on mortality indicators and how these can be interpreted. As a result, trusts have found it challenging to identify appropriate trusts to benchmark themselves against. Benchmarking is vital to enable trusts to compare their performance with that of similar organisations, to see what they are doing well and to highlight areas for improvement.
Dr Foster has devised a new methodology to identify trusts that have been similarly impacted due to Covid-19. This will enable trusts to compare themselves to others who were faced by similar pressures during the pandemic.
New methodology allowing bespoke comparison under new circumstances
It is known that there has been regional variation in the prevalence and transmission of the virus. This makes it important to consider additional factors for benchmarking, such as patient casemix, which are especially informative of the impact trusts were likely to face and their capacity to respond to the arising challenges. Rather than focusing only on geographic location, hospital size and both Covid-19 related and overall activity are other examples of important factors to consider.
To identify comparable trusts, this analysis took into account casemix factors such as age, sex, ethnicity and a number of baseline comorbidities that meant patients were at a higher risk of admission, the number of Covid-19 spells at a trust and hospital baseline capacity indicators. Baseline capacity was proxied using factors including trust volume, general and critical care bed capacity and patient admission methods. Capacity indicators were measured before the pandemic to ensure that trusts are compared based on their performance rather than on their subsequent response to the crisis. For example, turning general wards into critical care facilities which would allow them to provide high quality patient care.
The appropriate data suppression rules have been applied to the above figure. Trusts which would require the proportion metrics to be calculated from small numbers have been excluded.
The similarity between pairs of trusts was measured by using these variables to calculate the Euclidean distance between them. This allowed the identification of trusts which were more similar in their experience over the pandemic and therefore more appropriate for peer group comparison. The closer the Euclidean distance, the greater the ability for comparison. Rather than using the simple nearest neighbours method to allocate peers, a cut off Euclidean distance between trusts was used to identify which were similar enough to be classed as ‘true peers’.
Creating bespoke peer groups enables us to see a clearer and truer picture of how trusts have performed compared to others in the same situation. Identifying performance issues over the past few months and learning from best practice may prove extremely important in the upcoming months and beyond.