Readmission is not simply correlated to length of stay

Increasing levels of day-case surgery and zero day emergency admissions make analysing length of stay difficult. However, elective hip replacements provided an opportunity to examine changes in length of stay without complications from these factors.

The first graph shows a year-on-year reduction in the median length of stay for elective hip replacements. Patients who died have been excluded from the analysis, as have trusts that carried out fewer than 200 hip replacements in any year.

The graph shows the percentage of patients who have had an emergency readmission to any hospital in the country, within 28 days of being discharged after a hip replacement. Readmissions appear to rise as length of stay falls, but this hides considerable variation between trusts both for readmissions and length of stay.

The second graph shows the relative risk of readmission for 2006-07 after the variation that can be attributed to age, gender and deprivation has been removed. A relative risk of 100 suggests that a trust is average. The variation in readmissions shown is remarkable.

A patient from the worst-performing trust is four times as likely to be readmitted as a patient from the best-performing trust.

The bottom graph shows the variation in median length of stay for hip replacements and we can see that three trusts have a length stay of more than twice that of the trust with the shortest length of stay.

This is a much more complex picture than simply attributing readmissions to short length of stay.

While one of the trusts with shortest length of stay has a high risk of readmission, this is also true of one of the trusts with the longest length of stay.

For hip replacements, inter-patient variability in length of stay is a better indicator of readmission rates than length of stay itself, suggesting that care processes are more important than length of stay in preventing readmissions.