Case study of the prediction of elective surgery durations in a New Zealand teaching hospital

Summary We present an elective surgery redesign project involving several New Zealand hospitals that is primarily data‐driven. One of the project objectives is to improve the predictions of surgery durations. We address this task by considering two approaches: (a) linear regression modelling, and (b...

Full description

Saved in:
Bibliographic Details
Published in:The International journal of health planning and management 2020-11, Vol.35 (6), p.1593-1605
Main Authors: Soh, Kian Wee, Walker, Cameron, O'Sullivan, Michael, Wallace, Jonathan, Grayson, David
Format: Article
Language:English
Subjects:
Classification
Data quality
Elective surgery
electronic booking form
Health administration
Hospitals
linear regression
prediction
Predictions
Redesign
scheduling
Specification
Surgeons
Surgery
Teaching
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Summary We present an elective surgery redesign project involving several New Zealand hospitals that is primarily data‐driven. One of the project objectives is to improve the predictions of surgery durations. We address this task by considering two approaches: (a) linear regression modelling, and (b) improvement of the data quality. For (a) we evaluate the accuracy of predictions using two performance measures. These predictions are compared to the surgeons' estimates that may subsequently be adjusted. We demonstrate using the historical surgical lists that the estimates from our prediction techniques improve the scheduling of elective surgeries by minimising the occurrences of list under‐ and over‐runs. For (b), we discuss how the surgical data motivates a review of the surgery procedure classification which takes into account the design of the electronic booking form. The proposed hierarchical classification streamlines the specification of surgery types and therefore retains the potential for improved predictions.
ISSN:0749-6753
1099-1751
DOI:10.1002/hpm.3046