Detailed analysis of knee replacement annual volume reveals its significant effect on readmission rates

Dr Foster has recently undertaken statistical analyses of abdominal aortic aneurisms and trans-catheter aortic valve implementations and found interesting correlations between provider annual volume and mortality. Following on from this, the Dr Foster team has carried out an analysis that examines how the number of annual knee replacement procedures performed within a trust influences the rate of readmission.

30 Jan 2020 | 3 min read

Image by skeeze from Pixabay

Knee replacement surgery is a common operation to replace a damaged knee with an artificial joint. The first knee replacement procedure a patient undergoes is known as the primary knee replacement, with subsequent operations on that knee being called ‘revision surgery’.

Dr Foster used data from Hospital Episode Statistics, analysing a total of 521,858 elective primary knee replacements in order to find the effect of provider annual volume on the outcome of 28-day readmissions post-operation. These surgeries all took place between April 2010 and March 2019.

Figure 1: Histogram of the average provider volumes for knee replacement surgery over the time period

The team found that annual volume was a significant variable, influencing readmissions at a 90 per cent confidence level. The results suggest that an increase in the number of knee replacements a provider carries out over a year would decrease the odds of readmissions following surgery. The crude rate of readmissions was 5.8 per cent. In the case of elective knee replacements, it adds to the evidence base that provider volume represents experience, and more experience means better outcomes.

Figure 2: Crude rate of readmissions for knee replacement surgery

Modelling the effect of surgeon and provider experience on patient outcomes following a procedure demanded specialist analytical skills. The data used in the analysis consisted of multiple procedures carried out by a single surgeon, with several surgeons operating at a hospital, and many hospitals across the country, which meant the data had a clustered structure with multiple levels. The complex nature of the data had to be taken into account, and analysts chose to use a linear mixed effect model, an advanced technique particularly useful for this type of data structure.

Figure 3: Adjusted rate of readmissions for knee replacement surgery

Clinical outcomes are influenced by a wide range of factors and this analysis was necessary to home in on the effect of provider volume alone without other variation between providers. Granular analyses of this kind can reveal relationships and patterns that would otherwise go undetected yet are capable of driving real improvement. Dr Foster analysts are skilled in the use of a wide range of statistical modelling techniques and always ensure the most appropriate, statistically robust model is used for the data at hand, meaning that conclusions are always as reliable as possible. Chosen models are also validated by the Dr Foster Unit at Imperial College London.

The new Analysis Managed Service offered by Dr Foster allows customers to utilise the breadth of knowledge and experience of the analytics team while maintaining one point of contact. This means that, regardless of the request, trusts will always have access to an analyst with the relevant skills. The service enables trusts and other NHS organisations to make use of the reams of data they collect, painting a full picture of clinical operations and transforming this into actionable insights.

You can find out more about our Analysis Managed Service here or contact us for more information.