Do algorithms provide consistency in clinical cue acquisition in telephone consultation at NHS direct?

Abstract

Background A number of studies have concluded that significant variation exists in nurse decision-making when using Clinical Decision Support Systems (CDSS) in the telephone consultation setting. This is surprising, since one of the major purported benefits of using CDSS is enhanced consistency and safety. It is thought that algorithm based CDSS reduces decision-making variance by providing a template of specimen questions to capture all relevant cues during a consultation. This study was undertaken to determine if a) the use of specimen questions and clinical cue acquisition does not differ significantly between two key independent variables when using an algorithm based CDSS (call centre and algorithm) and b) to determine whether there was any correlation between clinical cue acquisition and the consultation outcome (disposition). Two of the most frequently used algorithms at NHS Direct were chosen for the study; an adult specific presentation (Abdominal Pain) and a child specific presentation (Fever Toddler). The settings were two call centres at NHS Direct, the health advice and health information service delivering telephone consultation to the general public throughout England and Wales. Methods The audio recordings of 250 NHS Direct nurse consultations were examined and coded. The two algorithms chosen for the study were frequently used at NHS Direct and also provided the opportunity to compare two different consultation types; a direct interlocution with the patient (Abdominal Pain) and a third party interlocution with the patient’s parent or carer (Fever Toddler). Difference tests were conducted to determine variance in clinical cue acquisition and a correlation test was carried out to determine the relationship of clinical cue acquisition with the disposition. Results The frequency of clinical cue negative specimen questions (CCA-) differed by algorithm, (U=5314, Z=-4.457, p=<0.001). A significant correlation between CCA- specimen questions and the disposition (r= 0.230, p=<0.001). Between 1 and 11 specimen questions were CCA- in 70% (175/250) consultations. Across the 250 consultations, 6,501 specimen questions were available to Nurse Advisors using the NHS Clinical Assessment System, 91.66% (5,559/6,501) of specimen questions were CCA+, 8.34% (542/6,501) were CCA-. Conclusion Significant variance exists in the use of specimen questions in consultations at NHS Direct. This variance is apparent despite the use of an algorithm based CDSS which is designed to reduce variance in assessments and decision-making. Furthermore, clinical cue acquisition is related to the consultation outcome raising questions concerning the clinical safety in consultations that have a high frequency CCA- specimen questions. However, the within methods quantitative methodology of the study has limitations and further across methods research is required to fully explain the variance in the acquisition of clinical cues supported by an algorithm based CDSS.

Date of Award	2012
Original language	English
Awarding Institution	Coventry University University of Worcester