Validating Studies to Assess the Diagnostic Accuracy of a Software Application for Detection and Monitoring of Attentional Deficits in Delirium

The investigators previously developed a new neuropsychological test for the objective measurement of inattention in delirium, which was implemented on a purpose-built computerised device (Delbox). A prototype software application (DelApp) for smartphones has now been developed based on the same test.

The DelApp has many advantages over existing tests, including objectivity, automated scoring and ease-of-use. Studies in 30 hospital patients showed highly comparable performance between the Delbox and DelApp. Further proof-of-principle single-rater studies using the DelApp in more than 200 hospital patients have provided additional evidence of good performance of the DelApp.

Though there are strong positive findings from investigator pilot studies, larger-scale formal studies with blinded raters and more representative patient samples are now required.

The proposed study is part of a larger programme of work funded by the Medical Research Council (grant value £1.01M) on the optimisation and clinical validation of the DelApp, which consists of three stages: Phase A - optimisation of the DelApp and feasibility testing (REC no 15/SC00/19); Phase B - case-control studies in selected patients; and Phase C - validation studies in unselected patients. Ultimately investigators aim to generate a software product which will be fully licensed and commercially available to healthcare organisations.

The proposed Phase B studies will be carried out in hospitalised patients in general and acute medical wards and ICU. The objectives of these studies are (a) to assess diagnostic performance of the DelApp in discrete patient groups and to inform optimal cut-points (Phase B); and (b) to evaluate the diagnostic performance of the DelApp in unselected patients (Phase C). The studies will also contribute to a better understanding of the nature of neuropsychological deficits underlying delirium. At the end of these studies, the DelApp may be modified depending on the study results. Investigators will select the stimuli and subtests that provide the best balance of sensitivity and specificity.

Two case-control studies will be conducted involving patients recruited from general and acute wards (Study B.1) and ICU (Study B.2). Each study will run in two sites, i.e. the Royal Infirmary of Edinburgh (RIE) and the Glasgow Royal Infirmary (GRI). The performance of the DelApp (index test) against reference standard diagnosis will be independently assessed by two trained raters, with one pair of raters on each site. As in our prior case-control studies, convenience samples will be used, where clinicians are asked to identify patients in the target groups.

Hospitalised patients will be asked to undertake a number of tests at their bedside. These tests will be administered to patients once only. Patients will be visited by two different researchers on the same day, with a gap of approximately 15-60 min between assessment visits assuming that consent is provided in the first visit. The first visit will involve assessment of capacity and (in those thought likely to be capable of giving consent) obtaining informed consent, as well as assessment of delirium (reference tests) and cognition. The second visit will involve the DelApp assessment (index test). Thus, the index test and reference tests are administered by independent raters. Where the patient is unable to provide consent for themselves, proxy consent will be sought, and if granted, the same visits as above will occur.

The Phase C formal validation studies have a similar design to the case-control studies. The main difference is that unselected samples of patients will be recruited from general and acute medical wards (Study C.1) and ICU (Study C.2) in the GRI and RIE.

A subset of patients in studies C.1 and C.2 (N=60 combined) will undergo up to 4 repeat assessments (at least one day apart), along with the reference standard, to explore the performance of the DelApp in tracking change in cognitive function.

In the longitudinal assessments, researchers will ask patients able to provide consent at the time of recruitment if they consent to subsequent assessments even if at the time of these subsequent assessments the patient lacks capacity. However, during the subsequent assessments, it will be carefully checked if the patient assents to continue their participation in the study. In patients who initially lack capacity and have been included through consent from a legal proxy, and who are undergoing serial assessments, researchers will seek informed consent from patients at the beginning of each new testing session. If the patient regains capacity after the final assessment, the investigators will seek retrospective consent from the patient where possible.