Acute undifferentiated febrile illness in the UK and investigating the diagnostic utility of metagenomic next generation sequencing

Abstract

Aims Acute undifferentiated febrile illness (AUFI) has not been well studied in the UK. The global literature suggests that the majority of AUFI is likely to be attributable to infection, however a high proportion of individuals remain without a microbiologically confirmed diagnosis. Indiscriminate broad-spectrum antimicrobial use is common and mortality can be high. This proof of concept study aimed to characterise the clinical features, aetiology, antimicrobial use and clinical outcomes of adults hospitalised with AUFI at a large teaching hospital in the UK. In addition, metagenomic next generation sequencing (mNGS) was retrospectively used as an experimental diagnostic tool in an attempt to further elucidate the microbiological causes of AUFI. Methods One hundred adults with AUFI and fifty healthy volunteers were recruited in this prospective controlled cohort study. AUFI participants were recruited in hospital and followed up at 4 to 6 weeks following onset of fever. A standardised set of diagnostic tests were performed in addition to those requested by the treating physician. EDTA whole blood, serum, nasopharyngeal swabs and urine were taken for research purposes. Final diagnoses were recorded once follow up was complete. Where there was uncertainty in the final diagnosis this was resolved by independent adjudication between the principal investigator and the lead researcher; clinicians both trained in infectious diseases. The same research samples were taken from healthy volunteers and telephone follow up occurred one week after recruitment. Those who developed infective symptoms at follow up were excluded. Retrospective mNGS of EDTA whole blood and serum samples from twenty-five AUFI participants and five healthy volunteers was performed. Results Over half of participants with AUFI did not receive a clinically credible diagnosis, 52% (52/100). Of those with a confirmed diagnosis, infectious causes predominated (viral 27% (27/100), bacterial 18%; (18/100)), with few non-infectious diagnoses made (3%; 3/100). Extensive microbiological and radiological investigations were performed, but only a small proportion (5.0%; 70/1401) contributed to the final diagnosis. Empirical antimicrobial use was common (81%; 81/100). Antimicrobials were prescribed in 84.6% (44/52) of those without a diagnosis supported by standard diagnostic testing and 77.1% (37/48) of those with a diagnosis, (OR 0.61; 95% CI 0.211.77; p=0.45). The median [IQR] antimicrobial duration was 6.0 [1.0-12.0] days across the entire AUFI cohort, 5.6 [1.6 to 9.9] days in those without a diagnosis and 5.0 [0.1 to 12.1] days in those with a diagnosis (difference 0.63, 95% CI -2.46-1.96; p=0.99). The median [IQR] length of stay for the whole cohort was 2.0 [1.0-4.1] days. Undiagnosed participants had significantly longer hospital stays than those with a diagnosis (median 2.9 IQR [1.6-4.9] days versus 1.7 [0.8-1.6] days; difference of 1.2 days (95%CI 0.04 to 1.66); p=0.036). There was a trend for undiagnosed participants having on-going symptoms at follow up (undiagnosed 59.6% (31/52) versus diagnosed 39.6% (19/48)(OR 0.44, 95% CI 0.2-4.9; p=0.07) although this was not statistically significant. In this study, mNGS was not able to confirm microbiological diagnoses made by standard of care diagnostics or provide additional diagnostic information for undiagnosed participants. Conclusions This is the first proof of concept prospective study of adults presenting to a UK hospital with AUFI. Accepting of its small size, this study highlights the high proportion or participants who remain undiagnosed despite extensive investigation, that the majority of participants received antimicrobials and the use of broad-spectrum agents was common. Infection was the predominant cause of AUFI with a wide range of pathogens detected. Despite a growing body of evidence demonstrating mNGS can contribute to the diagnosis of infection, mNGS was not proven to be a useful diagnostic tool in this study, however a number of study limitation may have contributed to this finding. The wide range of pathogens identified and limited contribution of standard diagnostic testing suggests a broad, untargeted approach to infection diagnostics should be sought. Large, multi-centre studies should be considered to further characterise the burden of AUFI in the UK and to further explore the use of untargeted diagnostic technologies such as mNGS to improve the diagnosis.

More information

Identifiers

ORCID for Tristan Clark: orcid.org/0000-0001-6026-5295

Catalogue record

Export record