Article ID: | iaor20118386 |
Volume: | 3 |
Issue: | 1 |
Start Page Number: | 72 |
End Page Number: | 72 |
Publication Date: | Jan 2007 |
Journal: | Pakistan Journal of Statistics and Operation Research |
Authors: | Kamal Shahid, Lawrence Clive J, Trewin Vivian F, Sekandar Aisha |
Keywords: | statistics: inference, statistics: distributions |
The ability to make use of accurately documented, local and specific, reference ranges is an important part of clinical work. The most readily available, often the only, routinely collected data is that obtained from hospital admissions. A mixture distribution approach is employed to arrive at reference range appropriate to particular groups of patients. A classification rule is used to provide a spectrum of such ranges that allows for the lack of complete separation of the components of the mixture, a common feature when such models are fitted to data. The results of analyses are given for a range of biochemical parameters; in some examples, these are shown separately for males and females, and for those taking, and not taking, a diuretic drug – the most commonly occurring drug‐group in the elderly.