AbstractIn shortening length of stay (LOS) in the public hospitals, there is a need to examine the trends of its changes and its influencing factors. This thesis explores most suitable modelling approaches for linking LOS and its influencing factors. It also discusses the relationship between LOS and Casemix funding.
In order to understand the relative importance of factors influencing LOS, a Delphi evaluation was conducted through interviewing health care practitioners. Six important, 48 significant and four not important determinants of LOS were identified. The relationship among these factors and the implications of this evaluation were explored.
Two modelling methods, namely a holistic approach and mixture distribution analysis, and their applications were discussed. In the holistic approach proposed, firstly, the underlying distribution of LOS should be determined. Secondly, based on the determined distribution, a discordancy test trimming method could be applied to determine the trim points. This procedure would assist in distinguishing the inliers from the outliers. Thirdly, since normality of the distribution might not be attained simply by transforming LOS, appropriate regression models such as a gamma regression model should be required to reflect the relationship between LOS and its influencing factors. This would provide insights into the different effects of influencing factors between inliers and outliers. It would also help prescribe appropriate policy measures to reduce LOS. The holistic approach was useful in simultaneously identifying the appropriate underlying influencing factors of LOS and determining trim points for funding purposes.
As an alternative way of modelling LOS and its influencing factors, mixture distribution analysis was found to he able to confirm the homogeneity of certain Diagnosis Related Groups (DRGs). It could also reveal the heterogeneous patterns of other DRGs. For those DRGs exhibiting heterogeneity in LOS, related socio-economic factors influencing LOS were compared and contrasted between components by Poisson mixture regressions. Such an analysis provided an integrated framework to link funding with relevant influencing factors of LOS. A Poisson mixture regression model could give useful insights for state health institutions to initiate efficient Casemix payments. It could also benefit hospital managers and clinicians to manage LOS effectively.
An evaluation of the effects of Casemix funding on LOS and other relevant hospital performance measures was presented. Some evidence of the influence of Casemix funding on LOS and other hospital performance indicators since the implementation was been identified. The implications of the study in shortening LOS and promoting the efficiency of hospital management were discussed.
A summary of the conclusions in terms of the shortening LOS in the public hospitals of the Northern Territory was presented, and further research topics were discussed.
|Date of Award||Jun 1998|
|Supervisor||Ram Vemuri (Supervisor)|