eLetters to:

ARTICLE: Annick Galetto-Lacour, Samuel A. Zamora, and Alain Gervaix Bedside Procalcitonin and C-Reactive Protein Tests in Children With Fever Without Localizing Signs of Infection Seen in a Referral Center Pediatrics 2003; 112: 1054-1060 [Abstract] [Full text] [PDF]

eLetters: Submit a response to this article

eLetters published:

Procalcitonin Study: Methadological errors

Sudhin Thayyil   (22 January 2004)

Re: Procalcitonin Study: Methadological errors

Alain GERVAIX, Annick Galetto-Lacour, Samuel A. Zamora   (25 February 2004)

Procalcitonin Study: Methadological errors 22 January 2004
Sudhin Thayyil, Specialist Registrar in Neonates RVI, Newcastle upon Tyne, UK Send letter to journal: Re: Procalcitonin Study: Methadological errors sudhints{at}doctors.org.uk Sudhin Thayyil	Yet another success story of procalcitonin by Galetto-Lacour was an interesting read. Several studies have shown that procalcitonin is a good marker of bacterial infection when comparing severely ill children (eg meningococcal septic shock) with normal controls. However, clinically more useful would be studies from emergency settings such as this, where incidence of serious bacterial infection is (supposedly) low. Does procalcitonin really add to clinical judgement when a diagnostic dilemma exists ie clinician is faced with a relatively well febrile child and no focus of infection. To treat or not to treat. Galetto-Lacour and team addresses this question; however there are some major methodological errors in the study. One of the criteria that make any diagnostic study valid is that there should be an independent, blind comparison of the diagnostic test with the reference standard and the reference standard should be applied regardless of the diagnostic test result. In this study the ‘Gold standard’ (blood and urine cultures) were done only if leukocytes or band counts were elevated or if there were pus cells in urine. This could therefore invalidate the results of the study. More over pyuria may not be a good indicator of UTI 1. Incidence of serious bacterial infection in this cohort has been surprisingly high (29%), despite a lower entry temperature of 38 0C. Chest X rays were used to diagnose bacterial pneumonia in this study. However radiological differentiation between bacterial and viral pneumonias has been shown to be inaccurate 2, which again questions the ‘Gold standards’. Rapid procalcitonin testing (PCT Q, Brahmas) requires plasma or serum and cannot be done on full blood as far as I am aware. How do authors propose to separate serum from few drops of whole blood at bedside? Or was it actually done on whole blood. References 1. Kumar RK, Turner GM, Coulthard MG Don't count on urinary white cells to diagnose childhood urinary tract infection. BMJ 1996; 312(7042): 1359. 2. Drummond P, Clark J, Wheeler J, Galloway A, Freeman R, Cant A Community acquired pneumonia--a prospective UK study. Arch Dis Child. 2000; 83(5): 408-12.
Re: Procalcitonin Study: Methadological errors 25 February 2004
Alain GERVAIX, MD Dpt of Pediatrics, University Hospital of Geneva, Switzerland, Annick Galetto-Lacour, Samuel A. Zamora Send letter to journal: Re: Re: Procalcitonin Study: Methadological errors alain.gervaix{at}hcuge.ch Alain GERVAIX, et al.	In reply- We thank Mr Sudhin Thayyil for his comments regarding the methodology of our clinical study. We fully agree that the evaluation of any new diagnostic test should ideally be done against a gold standard test applied to the entire study population. Blood and urine cultures can be used as gold standard tests when investigating bacteremia and urinary tract infection, respectively, but to our knowledge no single gold standard test can be applied for all severe bacterial infections (SBI). The best tool we have to classify children as having or not a SBI in our current practice is clinical judgment following recommended guidelines (Baraff et al. Pediatrics 92, 1993) and a thorough follow-up of patients. We followed the recommended guidelines and felt unethical to perform blood/urine cultures systematically on all children if this was not recommended by experts. Furthermore as all subjects in our study had a complete follow-up the potential for misclassification bias is small. If we only consider the 82 children who had both blood and urine cultures (83% of the study population) the results of our study remain valid. PCT in this subgroup has the following diagnostic properties: sensitivity of 96%, specificity of 73%, positive predictive value of 63% and negative predictive value of 98%. As mentioned in the “method” section, rapid procalcitonin testing was performed on plasma and a small table centrifuge was used to separate plasma from red cells.