Article ID: | iaor201523852 |
Volume: | 30 |
Issue: | 6 |
Start Page Number: | 843 |
End Page Number: | 855 |
Publication Date: | Oct 2014 |
Journal: | Quality and Reliability Engineering International |
Authors: | Zeng Li, Peterson Ronnie |
Keywords: | health services |
Telephone nurse triage (TNT) services are an important component of the health care delivery system. Thanks to the advancement of telephone/information technologies, abundant data on the daily operations of telephone triage centers are now readily available that can be used to assess their performance. However, the current way to use such data is very primitive, which mainly focuses on simple descriptive statistics of the data. This study provides an approach to mine information from count data from the TNT service, including the daily inbound and outbound call volumes. A class of overdispersed generalized linear models is proposed to model the outbound–inbound relationship, and Bayesian algorithms for model estimation and comparison are provided. The established model reveals the variation in the accessibility of the triage service and the effect of weekday/weekend, which will provide significant information for performance assessment and improvement in such services. In the case study, the proposed approach is applied to a dataset from a triage call center in Wisconsin to demonstrate its use in practice. A numerical study is also done to examine the properties of the proposed model and accuracy of parameter estimation.