Inter-hospital variations in length of hospital stay following hip fracture.

Introduction

A hip fracture is the commonest reason for an acute admission to an orthopaedic ward, with a reported annual incidence of 58 970 in England for the year 1993/94 [1]. The financial burden to the health service has been estimated to be 288 million [pounds sterling] for hospital costs alone [2]. The length of hospital stay is an important and modifiable determinant of treatment costs. Previous reports have shown considerable variation (between 1 and 2 months) in length of total hospital stay [3-5]. We have investigated length of hospital stay and outcome for this group of patients in eight different hospitals in the East Anglian region.

Patients and methods

We audited hip fracture treatment in all orthopaedic departments in the East Anglian region to which hip fracture patients are admitted. Details for some 80 consecutive patients admitted with hip fracture to each hospital were recorded except for one hospital (number 2) where only 24 patients were admitted within the study period. The sample size of 80 per hospital was determined to allow both within- and between-hospital comparisons across important variables. Length of stay on the orthopaedic ward and any other wards before discharge was recorded for all patients. Additional details of the study methodology and results for mortality have been published previously [6,7].

At 90 days from injury, all surviving patients were contacted and their residential status determined. For this study a patient living `at home' included those living in a family member's house or in warden-aided accommodation. Receiving institutional care was defined as living in a residential or nursing home or already receiving long-term hospital care. Patients who sustained the hip fracture whilst a hospital inpatient had a length of stay recorded only for that time spent on the orthopaedic ward. Patients still in hospital at the end of the follow-up period had a length of stay adjudged as 90 days and were treated as censored observations in the statistical analysis.

Statistical evaluation was conducted using SPSS for Windows 6.1. Techniques included Kruskal-Wallis one-way analysis of variance, …