Use of point-of-care testing for respiratory viruses in hospital

In The Lancet Respiratory Medicine, Tristan Clark and colleagues report the results of a randomised controlled trial of point-of-care testing for influenza in patients in two UK hospitals in Hampshire, UK.¹ Before the emergence of COVID-19, influenza was the most important cause of admission to hospital with a respiratory virus. In England, influenza is estimated to have caused an average of 11 300 deaths per year between the 2015–16 and 2019–20 influenza seasons.² Influenza often goes undiagnosed in hospital, in part because of an absence of routine testing of patients with respiratory illness in this setting. For example, a study from Canada found that only around a quarter of patients with hospital admissions due to influenza had a laboratory-confirmed diagnosis.³ Standard PCR-based laboratory tests in hospital typically have a turnaround time of at least 24 h and sometimes longer, so cannot affect initial treatment decisions or early decisions about isolation for patients admitted as an emergency.⁴ This long turnaround time might explain, in part, why widespread ordering of influenza tests does not occur in hospital. Given that early antiviral treatment in patients with influenza is more likely to reduce mortality than is later treatment,⁵ these missed or delayed diagnoses could contribute to mortality. Data suggest that about 15% of influenza cases in hospital are due to nosocomial transmission, and around 9% of new influenza cases seed ongoing transmission in hospital; influenza outbreaks in hospital are thought to be common but under-recognised.4, 6 Hence, routine use of rapid diagnostics could contribute substantially to the prevention of influenza and influenza-related mortality among susceptible patients in hospital, but only if the diagnostics change clinical practice through earlier initiation of antivirals and appropriate isolation.

It is unusual to see a well conducted randomised controlled trial of the effect on practice of a new diagnostic tool. Clark and colleagues have run exactly such a trial in their assessment of a molecular point-of-care test (mPOCT) for influenza in adults admitted to hospital.¹ The investigators randomly assigned 623 patients with acute respiratory illness to mPOCT, with results back within 1 h (n=307), or to usual practice (n=306). If influenza was detected by mPOCT, clinical teams followed national guidelines for the treatment of influenza. Median age of participants was 62 years (IQR 45–75) and 332 (54%) of 612 participants with data were female. Although 202 (33%) of 613 participants had influenza infection, only 63 (62%) of 102 with influenza in the control group were treated with antivirals within 5 days of admission to hospital compared with 99 (99%) of 100 diagnosed by mPOCT. Furthermore, the median time to initiation of antivirals was 1 h (IQR 0·0–6·0) in the intervention group compared with 6 h (0·0–12·0) in the control group. The authors found that, compared with standard of care, the use of mPOCT strongly affected the likelihood that patients with influenza were appropriately isolated in single-room accommodation, minimising the risk of onward transmission. 70 (70%) of 100 patients with influenza in the intervention group were accommodated in this way compared with only 39 (38%) of 102 in the control group. This observation suggests that routine use of rapid diagnostics could help to prevent influenza in vulnerable patients by supporting early isolation that prevents nosocomial transmission.

The trial is pragmatic in the sense that the outcome measures are changes in clinical practice rather than hard measures of reductions in mortality, morbidity, or transmission. Nevertheless, unless diagnostics alter clinicians' behaviour, they cannot change clinical outcomes or affect onward transmission. We often rely on assessment of test characteristics such as accuracy, speed, and point-of-care delivery and assume that improving these characteristics will lead to meaningful changes in management.⁷ Therefore, standards for assessing the clinical utility of diagnostics are considerably lower than those for new therapeutics. There is a need for improved evidence on the impact of novel diagnostics because the introduction of new technologies always leads to some disruption, additional costs, possibilities of unwanted consequences, and uncertain effects on practice.

At the height of the first wave of the COVID-19 pandemic, about 15% of COVID-19 cases in a hospital in London, UK, were due to nosocomial transmission; 36% of patients who became infected in hospital subsequently died.⁸ However, despite the recommendation from April, 2020, that all patients admitted to hospital across the UK National Health Service be tested routinely using PCR for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2),⁹ data indicate that an increasing proportion of patients with COVID-19 in hospital developed symptoms more than 7 days after admission, suggesting that they probably acquired infection in hospital.¹⁰ These findings point to the need for increased use of rapid turnaround tests—an approach that was shown to be feasible for SARS-CoV-2 in a non-randomised trial of point-of-care testing by the same study group¹¹—to enable appropriate isolation or cohorting of symptomatic patients while awaiting laboratory results.

When considering the use of rapid diagnostic tests for control of respiratory infections in any setting, there are important trade-offs between accuracy, speed, ease of use, coverage, the usefulness of true-positive and true-negative tests, harms associated with false-positive and false-negative results, the likelihood that individuals or clinicians will act on the test results, availability, and cost. Although trials of the use of rapid diagnostics are ideal to assess the potential net effect of all of these variables on important outcomes, doing such trials across the wide range of potential uses of rapid diagnostics, especially in the context of a global health emergency, is challenging. Modelling studies that incorporate these parameters and are informed by real-world data from field studies can and should inform the investments being made in the development and use of rapid diagnostics. In future years, we will probably continue to need to respond to annual cycles of respiratory infection, including variants of SARS-CoV-2 and seasonal strains of influenza. Routine use of rapid diagnostic tests in emergency rooms and inpatient areas could play an important part in reducing mortality associated with these infections.

© 2021 Dr P Marazzi/SPL