The A-STOP Study
Research type
Research Study
Full title
Antifungal stewardship opportunities with rapid tests for fungal infection in critically ill patients
IRAS ID
234779
Contact name
Ronan McMullan
Contact email
Sponsor organisation
The Queen's University of Belfast
ISRCTN Number
ISRCTN43895480
Duration of Study in the UK
3 years, 11 months, 31 days
Research summary
Research Summary:
The clinical problem we address is over-prescription of drugs used to treat Candida fungal infection in adults and children in intensive care units (ICU). Treatment with ‘antifungal’ drugs is started when patients are thought to be at risk of fungal infection, even though the large majority turn out not to have this infection.
The majority of ICU patients who are treated with an antifungal drug receive treatment on an empirical basis. Typically, 7% of patients in ICU receive treatment for fungal infection and the majority are started on a presumptive, basis. Of these, only 1 in 20 have fungal infection confirmed. Up to 11,000 patients receive potentially unnecessary antifungal treatment each year, at a cost of up to £12million to the NHS. Most patients treated fail to benefit and are disadvantaged by the risk of side effects. Over-treatment can also lead to resistance to these drugs in the wider population.
We propose to evaluate how accurately three new rapid tests can diagnose fungal infection in adults and children, started presumptively on antifungal treatment. We will need to study 1720 patients in UK ICUs to give us a statistically reliable result. After obtaining consent, we will test blood samples from patients who are being started on antifungal treatment. The results of the tests will not be made available to their doctors in this study and their treatment will not be affected by participating. We will determine the clinical and economic impact of implementing these rapid tests, based on how accurately they diagnose fungal infection.
The main aim of this study is to establish the ability of these tests, to rule out fungal infection in this patient group. We will use these results to develop a guideline that could be used by ICU staff to reduce unnecessary antifungal drug use.
Summary of Results:
Patients in intensive care units (ICU) are frequently prescribed treatment for fungal infection (called antifungal treatment) when doctors suspect that they may have a serious fungal infection. Currently used lab tests are slow, taking up to five days for the results. Delays to starting treatment increases the risk of patients dying, therefore, doctors tend to take a cautious approach by starting treatment before these results are available. This means that up to 95% of patients who receive antifungal treatment do not need it. This treatment can have side effects, is expensive, and its over-use leads to resistance to these medicines making them less effective when they are needed most. So, it would be best if we could avoid giving treatment to patients who do not need it.
The main purpose of this study was to assess whether a group of new rapid tests are reliable for detecting Candida infection using a blood sample. Candida is a type of fungal infection and concern about Candida infection is the main reason that patients in ICU are started on antifungal treatment. We studied how the tests performed in a group of ICU patients who had been started on antifungal treatment by their doctors. The new tests were compared to the results from standard care tests that are based on growing Candida in the lab (hence, are slower) since that is considered to be the gold standard for proving the presence of infection.
The new tests evaluated comprised three that detect DNA (T2 Candida, Fungiplex and CandID) and one that detects a component in the cell wall of fungi (Fungitell). The primary way that we evaluated the accuracy of these tests was in estimating their negative predictive value (NPV), which indicates the percentage of patients with a negative test result that truly do not have Candida infection. Since the main purpose of this test is to rule out infection, a very high NPV would allow doctors to be confident that a negative test was highly likely to mean that there is no infection, so that unnecessary antifungal treatment can be stopped.
We recruited 1251 patients to this study from 43 UK ICUs, of whom 6 subsequently withdrew their consent for continued participation. Of the remaining 1245 patients, 54 were proven to have Candida infection. The negative predictive value of each of the four tests evaluated was 97%, although other ways we evaluated the tests revealed that they did not all have similar accuracy. For example, the sensitivity of the tests was evaluated – this describes the proportion of patients who had been proven to have infection had a positive result with each test. Therefore, the sensitivity indicates how well the test is able to detect infection when it is present and this varied considerably between the tests, with Fungitell having the highest test sensitivity of 69%.
We also evaluated whether combining these tests produced better accuracy than using them individually. This resulted in a slight increase in both NPV (up to 98%) and sensitivity (up to 75%). Of note, by adding a pair of new tests into standard care tests (that are based on growing Candida in the lab) improved the test performance further, with NPV approaching 99% and sensitivity of 85%.
These results led us to the conclusion that no single test would be accurate enough to allow doctors to stop antifungal treatment. However, a combination of two new tests alongside standard care tests perform better and could be a useful way to rule out infection and avoid unnecessary antifungal treatment.
REC name
South Central - Hampshire A Research Ethics Committee
REC reference
17/SC/0613
Date of REC Opinion
4 Jan 2018
REC opinion
Further Information Favourable Opinion