Hospital services and casemix in Western Australia

Predictors, consequences and costs of diabetes-related lower extremity amputation complicating type 2 diabetes: the Fremantle Diabetes Study

AIMS/HYPOTHESIS

The aims of this study were to assess the incidence, predictors, consequences, and inpatient cost of lower extremity amputation (LEA) in a community-based cohort of type 2 diabetic patients.

METHODS

Between 1993 and 1996, 1,294 patients with type 2 diabetes were recruited to the longitudinal, observational Fremantle Diabetes Study. LEAs and mortality from cardiac causes were monitored until 30 June 2005. Inpatient costs (in Australian dollars in year 2000), derived using a case-mix approach, were available for the period from 1 July 1993 to 30 June 2000.

RESULTS

During follow-up 44 patients without LEA at baseline had a first-ever diabetes-related LEA, an incidence of 3.8 per 1,000 patient-years. Independent predictors of first-ever LEA included foot ulceration (hazard ratio [95% CI]: 5.56 [1.24-25.01]), an ankle brachial index < or =0.90 (2.21 [1.11-4.42]), HbA(1c) (increase of 1%: 1.30 [1.10-1.54]) and neuropathy (2.65 [1.30-5.44]). The risk of cardiac death was significantly increased in patients with LEA at baseline, although this was not an independent risk factor. The median (interquartile range) inpatient cost per LEA admission was 12,485 Australian dollars (6,037 Australian dollars-24,415 Australian dollars), with a median length of stay of 24 (10-43) days.

CONCLUSIONS/INTERPRETATION

First-ever LEAs in type 2 patients were associated with poor glycaemic control, foot ulceration and evidence of microvascular and macrovascular disease. Patients with LEA were at increased risk of cardiac death. LEAs contribute disproportionately to diabetes-related inpatient costs.

Interstate comparisons of public hospital outputs using DRGs: are they fair?

OBJECTIVE

To assess whether there is variation among Australian States in the reporting and coding of important and relevant secondary diagnoses in public hospital data. Such variation is a potentially important problem because it may invalidate interstate (and other) comparisons of hospital outputs based on Diagnosis Related Groups (DRGs).

METHODS

Our outcome measure was the percentage of separations in the lowest-resource split for adjacent DRGs. To reduce potential bias due to interstate differences in the complexity of cases treated in public hospitals, we directly standardised by adjacent-DRG and analysed two subgroups of adjacent-DRGs where there is less discretion about the threshold for admission: obstetrics and major medical conditions.

RESULTS

There was important interstate variation in the percentage of separations in the lowest-resource split. The statistically significant differences represent a large number of medical records that might have been documented or coded differently if the separation had occurred in another State. For example, if Queensland had the same standardised percentage as South Australia, then an extra 10,100 separations in Queensland would have been allocated to a DRG with a higher cost weight.

CONCLUSIONS

The task of perfecting a national database is never complete and it makes sense to superimpose a permanent cycle of monitoring, followed by more detailed audits. The methods used in this paper could be routinely used to screen administrative hospital data to identify where detailed audits with feedback might be implemented with best effect. Unless interstate variation in the reporting and coding of important additional diagnoses is reduced, measuring public hospital outputs using DRGs will be of limited value at a national level.