Antifungal stewardship opportunities with rapid tests for fungal infection in critically-ill patients (The A-Stop Study)

DESIGN: Multicentre, prospective, diagnostic test accuracy study SETTING: 35 UK adult and paediatric ICUs. TARGET POPULATION: Critically-ill adults and children INCLUSION CRITERIA: Patients started empirically on systemic antifungal therapy for suspected Candida infection; age over 4 weeks EXCLUSIONS: Treatment with systemic antifungal for more than 24hr in the past 7 days; antifungal treatment for suspected non-Candida infection; neutropenia; BMT/acute leukaemia; proven invasive candidiasis in past 28 days. HEALTH TECHNOLOGIES: Two PCR tests that detect Candida species (Renishaw Diagnostics and T2 biosystems assays) and a fungal antigen test (beta-D-glucan, BDG). MEASUREMENT OF COSTS AND OUTCOMES: At enrolment, patients will have a blood sample taken that will be tested by conventional culture (forming part of the reference standard) and each of the three index tests being assessed; all samples will undergo all tests. The study primary outcome of diagnostic accuracy will be derived for each index test alone and in combination: sensitivity; specificity; positive/negative predictive values; and positive/negative likelihood ratios. This analysis will be conducted using an expert consensus definition for proven infection as the reference standard and then repeated using a combination of both this and the new consensus definition for probable infection (to take account of no perfect reference standard being available). A comparison of the primary outcome in subgroups of interest will be included. Data will be collected to measure the cost of care, including antifungal drug use, with follow-up until hospital discharge enabled by data linkage to the ICNARC case-mix programme. The number of patients receiving unnecessary antifungal treatment will be estimated and cost-utility analyses will be conducted. Recommendations for future research will focus on trialling a test-based protocol in a subsequent RCT. This protocol (using a single test or combination) will be based on the diagnostic accuracy metrics and clinical/cost-effectiveness modelling data we produce; physician and patient risk preferences will also contribute. SAMPLE SIZE: The sample size of 1720 gives 90% power to exclude the NPV being below 99%. Furthermore, this sample gives the study 97% power for producing a point estimate for sensitivity above 90% - a decisive threshold for clinicians. With an expected point sensitivity of 95% for the best index test (or combination), the 95% CI around this would be 88%-98.7% which reflects a degree of precision that is likely to be acceptable to clinicians. PROJECT TIMETABLE AND RECRUITMENT: 48 months comprising: 6-month set-up; 12-month internal pilot to confirm feasibility; a 24-month main recruitment period; and 6-months trial close down/analysis. Recruitment will average 1-2 patients per ICU per month and is achievable based on FIRE Study data with, on average, 4 eligible patients per ICU/month. TEAM EXPERTISE: Expertise includes: microbiologists experienced in trialling novel fungal diagnostic tests, particularly PCR; intensivists with experience of large clinical trials in ICU, including previous NIHR-funded studies; and a cost-effectiveness modeller with experience of NICE/HTA assessments. There is also input from ICNARC, experienced trials methodologists/statisticians, a UKCRC-registered CTU, and NIHR DEC, Newcastle.