Corresponding Author
Dr Bob Phillips
Centre for Reviews and Dissemination, University of York, YO10 5DD
Email:
bob.phillips@york.ac.uk
Twitter: @drbobphillips
A. Centre for Reviews and Dissemination, University of York, YO10 5DD
B. Department of Paediatric Oncology, Leeds Teaching Hospitals NHS
Trust, LS1 3EX
The challenge of multiple paediatric febrile neutropenia clinical
decision rules has been well described, and the difficulties with
ever-expanding variants based on individual datasets discussed
extensively.1 The re-calibration of the SPOG rule
undertaken by Haeusler and colleagues2 (creating the
‘AUS’ rule) for a practical, bedside tool which would be able to risk
stratify episodes of FN. This letter reports a further validation in a
meta-analytic dataset from the ‘Predicting Infectious Complications in
Children with Cancer’ (PICNICC) collaboration.
Briefly, the PICNICC collaboration was formed to develop a robust
prediction rule for febrile neutropenia, with methods and materials
detailed in previous publications.1,3 Over 8000
episodes of 33 different candidate predictor variables and seven
clinically relevant outcomes have been collated from 26 study groups,
with varied completeness. The ‘AUS’ rule uses three equally weighted
factors to scale the risk of complications in an episode of FN, summed
between 0 and 3.These factors are: preceding chemotherapy more intensive
than acute lymphoblastic leukaemia maintenance, platelet count less than
50 g/L, total white cell count lower than 300
cells/mm3. The clinical response can be scaled
according to this score, with values of 0 and 1 considered as ‘lower
risk’.
We evaluated the association between the AUS rule and bacterial
infection using these data, reporting the discrimination (Area Under the
receiver-operator Curve, AUC) and proportion of episodes classed as
lower risk (scoring 0 or 1). Analyses were undertaken using R (v3.2.0).
1520 episodes contributed to the analysis, with 301 episodes of
documented bacterial infection. The discriminatory ability appeared very
similar to Haeusler’s values; AUC 0.64 (95% CI 0.61 to 0.68) compared
with the original AUC 0.67 (95% CI 0.63 to 0.71). Using the AUS rule on
the PICNICC dataset would have identified 44% (668/1520) of the
population as a lower risk group (score 1 or 0).
This reassuring data has led to the introduction of an AUS rule based
system to shorten the duration of antibiotic therapy, in concordance
with UK National and International guidelines[3,4]. This
now-validated rule will reduce hospitalisation for febrile neutropenic
episodes in the UK, which was of particular importance during the 2020
SARS-CoV-2 coronavirus pandemic.