Introduction
Casemix is a system of classifying 'cases' (patient contacts, episodes or visits) according to characteristics such as diagnosis, treatment, severity of illness, potential for health care improvements or costs.¹ Healthcare Resource Groups (HRGs) have been developed in the UK as a suitable casemix measure to monitor efficiency within the NHS. The aim of the present study was to evaluate the usefulness of HRGs in determining reasons for a rise in non-elective readmission rates and length of acute hospital stay in the study site hospital, despite increased attention being given to discharge planning.

Methods
Medical, demographic and socio-economic data were gathered for three groups of elderly patients with unscheduled general medical admissions over the periods: June to December, 1995 (n=299), June to December 1996 (n=460) and June 1997 to January 1998 (n=487), together with data on their length of hospital stay, time to first unplanned readmission, primary admission diagnoses and unplanned readmission's to the study site hospital over a one year period post-discharge. Patients who died during the original admission, or follow-up period, were eliminated, leaving final groups of 298, 458 and 412 patients for the three respective periods. Eighty-two HRGs, which were most applicable to the study population, were selected from version 3.1 HRGs² and all patients were assigned to the most applicable HRG. The percentages of patients in each HRG were calculated together with the trimmed (excluding outliers) and untrimmed lengths of acute hospital stay and compared with National Casemix Statistics for England³.

Results
Twelve-month readmission rates of 19.2%, 26.5% and 36.2% for 1995, 1996 and 1997/98 were noted. Corresponding mean (+ SD) lengths of hospital stay were 8.6+9.2 days, 7.7+11.4 days and 10.5+13.3 days. Seventeen percent of patients with non elective admissions in 1995 were assigned to a cardiovascular HRG. Since this figure rose to 53% and 55% in 1998 and 1997/8 respectively, these HRGs were further investigated. HRGs E11 and E12, which account for all myocardial infarction patients, showed sharp rises in readmission rates e.g. a twelve-month rate of >45% being recorded in 1997/8 for E11 (acute MI with comorbidities and complications [wcc]). E33 (angina, aged >69 years or wcc) also demonstrated a high readmission rate (>45%). In terms of length of stay, the HRGs A20 (TIA, >69 years or wcc) and A22 (non-transient stroke, >69 years or wcc), were trimmed by as much as 50% in each of the three years, indicating that these patients were staying in hospital for longer than was necessary e.g. while secondary rehabilitation arrangements were being made. In contrast, HRG D20 (COPD/Bronchitis) did not require the removal of any outliers indicating that these patients were not exceeding their expected length of stay.

Discussion
The use of casemix analysis assisted in the identification of cardiovascular patient groups who may require more structured discharge planning to reduce excessive length of hospital stay or early readmission. Casemix analysis, however, does not account fully for changed severity of patient illness. Future research will therefore concentrate on developing severity of illness indices.

References

1. Hutchinson, A., Parkin, D., Philips, P. (1991) Casemix measures for ambulatory care. Journal of Public Health Medicine, 13: 189-197.
2. National Casemix Office (1997) Version 3.1 Healthcare Resource Groups. The National Casemix Office, NHS Executive Headquarters, Winchester
3. National Casemix Office (1998) National Casemix Statistics (England), 1996097. The National Casemix Office, NHS Executive Headquarters, Winchester