SCH Journal ClubUse of time from fever onset improves the diagnostic accuracy of C-reactive protein in identifying bacterial infections
Wednesday 13th May 2015
Nic Seneviratne
Case Presentation• A 6 month old presents to the hospital with evidence of a
LRTI and fever of 38.5oC • They are not clinically septic• How will you decide whether this needs treatment with
antibiotics or is likely to be viral?• Will initial inflammatory markers help?• Does consideration of the duration of their fever prior to
presentation make any difference?
The Clinical Question• In a feverish child, does the use of time of onset of fever
improve the ability of CRP to differentiate between bacterial & viral infections?
• P- feverish children• I – time from onset of fever & CRP• C – CRP alone• O – differentiation between bacterial & viral infections
Paper• Use of time from fever
onset improves the diagnostic accuracy of C-reactive protein in identifying bacterial infections.
• Idan Segal, Matityahu Ehrlichman, Joseph Urbach
• Arch Dis Child 2014 99: 974-978
Current Practice / Guidelines• NICE Guidelines for Early Onset Neonatal Sepsis
recommends a repeat CRP at 18-24h as the initial result may be falsely reassuring
• Can the same presumption be made about older patients?
• NICE guidelines for feverish children cautions that a low CRP at presentation does not rule out serious infection
Methods• A prospective observational study• Single hospital• Children with fever where it was felt blood tests were
clinically appropriate• Only enrolled when study investigators were available• WBC and differential, CRP, blood cultures, and time from
fever onset in all patients• CXR, urine & CSF culture semi-dependent on clinical
picture, although some guidance
Study Flow Chart1158 children with fever
229 met exclusion criteria 929 eligible patients
373 enrolled
0-28 days
Full septic screen (blood, urine & CSF
cultures)
28 days to 8 weeks
Focus of infection or WBC >15,000 or
<5,000 or white cells in urine
Full septic screen
>8 weeks
Abnormal urinalysis or symptoms of urine
infection
Urine culture
Signs of pneumonia
Chest X-ray
No focus of infection
Chest X-ray
Appeared unwell
Full septic screen
All enrolled patients had blood culture
and WBC with differential sent
CRP was sent and parents determined
time from fever onset
Outcome Measures• Patients classified as
– Bacterial infection– Presumed viral infection– Acute otitis media– Inconclusive
• CRP value at different time points compared between bacteria & viral groups
• Sensitivity, specificity & likelihood ratios calculated for diagnosis of bacterial infection.
C-reactive protein (CRP) values by time from fever onset for bacterial (circles) and viral (crosses) infections.
Idan Segal et al. Arch Dis Child 2014;99:974-978
Copyright © BMJ Publishing Group Ltd & Royal College of Paediatrics and Child Health. All rights reserved.
Receiver operating characteristic curves for C-reactive protein at different time points from fever onset.
Idan Segal et al. Arch Dis Child 2014;99:974-978
Copyright © BMJ Publishing Group Ltd & Royal College of Paediatrics and Child Health. All rights reserved.
Positive & Negative Post Test Probability for Bacterial Infection by CRP value and time from fever onset.
C – rule in, D - rule out.
Idan Segal et al. Arch Dis Child 2014;99:974-978
Copyright © BMJ Publishing Group Ltd & Royal College of Paediatrics and Child Health. All rights reserved.
CASP• Was there a clear question for the study to address?
» Clear regarding the setting and the test» Less so about the population or the outcomes and
how these were defined
• Was there a comparison with an appropriate reference standard?
» Unclear» The definition of viral infection is the absence of
evidence of bacterial infection
CASP 2• Did all the patients get the diagnostic test and reference
standard? Yes• Could the results of the tests have been influenced by
the results of the reference test? No• Is the disease status of the tested population clearly
defined? No• Were the methods for performing the test described in
sufficient details? Not really
What Are the Results?CRP at different time points
AUC% (95% CI)
Sensitivity% (95% CI)
Specificity % (95% CI) +LR (95%
CI)
Post test Probability (%)
<12hN = 74 (*2.1mg/dL)
76 (63-88) 72 (52-87) 77 (64-86) 3.1 (1.8-5.5) 76 (62-89)
>12-24hN=67 (*6mg/dL)
81 (69-92) 68 (48-83) 83 (69-92) 4.2 (2-8.4) 80 (63-96)
>24-48h N=51 (*10.7mg/dL)
87 (77-96) 68 (47-84) 90 (73-96) 6.8 (2.1-20) 87 (62-99)
>48h N=98 (*12.6mg/dL)
90 (84-97) 80 (64-90) 94 (85-97.5)
13.3 (4.8-33) 93 (82-99)
Pre-test probability = 27%* Cut off
Are the results of the trial valid?• Sensitivity, specificity and likelihood ratios presented,
including confidence intervals, but unclear how these were calculated
• How sure are we about the results?
• Positive post test probability from this study suggests that a CRP >16 in the first 24h is always a bacterial infection
Are the Results of the Trial Valid?
• Only a small proportion of eligible pts enrolled– Was there a significant difference between those who had
bloods done & those that did not– Higher rate of bacterial infection than expected
• How accurate was the differentiation between bacterial & viral infection?– Definition of viral infection includes recovery without antibitoics,
therefore are those with severe infection presumed bacterial?
Will the Results Help Locally?• Can the results be applied to our patients / population?
– Unclear. – What is their threshold for doing blood tests on children?– Likelihood is that these are the children that we have diagnostic
uncertainty about
• Can the test be applied to our population?– Yes
• Will knowledge of the test result improve patient well being or lead to a change in management?– Unlikely
Bottom Line Conclusion
• A high CRP can be used to rule in bacterial infection• A low CRP does not rule out bacterial infection,
especially if it is done early on in the illness
• The addition of time from onset of fever adds little to the management of a child, with a single CRP
• Knowledge of the expected rise in CRP over the first 24 hours is a useful reminder not to be too reassured by an early low CRP
Summary & Conclusion