A statistical analysis of spatially linked time series using data from the TOXBASE database to study Emergency Department and NHS phone line management of poisoned patients

Student thesis: Doctoral Thesis

Abstract

The aim of this thesis is to examine the overall trends in poisoning, which will lead to an assessment of the relationship between poisons information database TOXBASE use and the management of poisoned patients within UK based emergency departments. Previous studies on the demographics of poisoning will be presented. This will cover important concepts in the study of toxicology before examining how specific demographic variables are linked to incidences of poisoning. A discussion of the agents used in cases of poisoning in recent history will be followed by a summary of legislation pertaining to poisoning and dangerous substances. The various services examined throughout the thesis will also be introduced. The first of these services, NHS 24, is described as an out of hours service for use by members of the public. The NHS 24 operator can advise on appropriate action given a description of the symptoms. These descriptions can be categorised as poisoning based on the presence of specific key words or phrases. Using generalised additive models, a consistent seasonal trend in poisoning calls to NHS 24 was found. The second service examined, TOXBASE, is a database provided by the National Poisons Information Service which provides information to clinical professionals on how to treat poisoning by a variety of substances. As in the NHS 24 analysis, generalised additive models have been used in order to assess the trends present in accesses made to the TOXBASE database by clinicians. The results from this analysis found that there was a consistent seasonal trend in TOXBASE accesses which peaked over summer and was similar to that seen in the NHS 24 call data. A third temporal analysis was carried out on data obtained from NHS information services pertaining to admissions and attendances due to poisoning, again showing similar results to the previous two analyses.These analyses combined suggest an underlying trend in poisonings. Both the admissions data and TOXBASE access data were examined using funnel plots in order to determine whether there were any hospitals which were unusual in their admissions or their TOXBASE use. This analysis found some commonalities in those hospitals which are unusual in either their admission rates or TOXBASE use. The final step in this project was to link the TOXBASE access data with the attendances and admission data in order to examine whether there was any link between TOXBASE use and admission due to poisoning. The results of this indicate that there are associations between TOXBASE use and admissions, such that an increase in TOXBASE use indicates an increase in admissions due to drugs poisoning. However, it became clear that the data used were limited in their ability to show any direct impact of TOXBASE on admissions due to drug poisoning, and that more specific data, for example on toxicants involved and case severity, would potentially be useful in mitigating the obvious confounding present in these data. This thesis has provided new insight into patterns in cases of poisoning, as well as providing a strong basis for further analysis to establish whether there isa direct impact of TOXBASE use on patient management within UK emergency departments.
Date of Award19 Sep 2019
Original languageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsNational Poison Information Service (Edinburgh Royal Infirmary, NHS Lothian) & University of Strathclyde
SupervisorChris Robertson (Supervisor) & Alastair Rushworth (Supervisor)

Cite this