Article Text


Procalcitonin as a prognostic marker in children with meningococcal septic shock
  1. F Leclerc,
  2. S Leteurtre,
  3. O Noizet,
  4. A Dorkenoo,
  5. A Sadik,
  6. R Cremer,
  7. C Fourier
  1. Paediatric Intensive Care Unit, University Hospital of Lille, Lille, France
  1. Correspondence to:
    Professor Leclerc, Hôpital Jeanne de Flandre, 59037, Lille, France;

Statistics from

Carrol and coworkers1 confirm the findings from Karabocuoglu et al who reported that procalcitonin (PCT) was higher in children with severe meningococcaemia (fever, petechia or purpura, and hemodynamic instability) than in children with systemic meningococcal infection without shock (291.29 ± 167 v 19.7 ± 23 ng/ml; p<0.001).2

Unfortunately, information is lacking in the report of Carrol et al,1 namely: a clear definition of severe MCD (defined in their paper as a Glasgow Meningococcal Septicaemia Prognostic Score ≥8) and median PCT values of survivors and non-survivors and comparison in term of prediction of outcome between PCT level and generic or specific severity scoring systems. We report that admission PCT level is an accurate predictor of mortality in the subgroup of children with meningococcal septic shock (MSS). We prospectively investigated 35 children (median age: 16 months; Q1:9–Q3:45) with MSS (defined as ecchymotic or necrotic purpura with shock, needing fluid expansion (median for the first 24 hrs: 90 ml/kg; Q1–Q3: 48–120) and catecholamine infusion) admitted to our PICU between July 1999 and May 2002. We estimated the accuracy in predicting death of PCT, C reactive protein (CRP: nephelometry)3 on admission, and the Pediatric Risk of Mortality (PRISM) score4 within 24 hrs of admission or at the time of death. Sensitivity, specificity, positive and negative predictive values, and percentage of well classified children were calculated at the following cutoff values: PCT >130 ng/ml (the best cutoff value of the PCT level was determined by χ2 optimisation (Fisher’s test: p=0.0004)), CRP <100 mg/l 3, PRISM value >20 and PRISM probability of death >50 %.5 For each severity index, we calculated the area under the ROC curve (AUC) and the standard error (SE)6 and determined the significance of comparisons.7

Eleven of 35 children died (31%); predicted mortality with the PRISM score was 15.6 (standardised mortality ration: 0.71; 95 % confidence interval: 0.35–1.26). The median (Q1–Q3) PCT and CRP levels and PRISM value and probability of death were the following: (survivors v nonsurvivors) PCT 73 (15–210) v 277 (208–606) ng/ml (p=0.001); CRP 92 (44–160) v 72 (41–109) mg/l (p=0.25); PRISM value 17 (8–22) v 33 (26–37) (p <10−3); PRISM probability 19 (4–42) v 88 (63–95) % (p <10−3). Performance characteristics and AUC ± SE of PCT, CRP, and PRISM score are given in the table and the figure.

In our study, PCT on admission was as accurate as the PRISM value and PRISM probability of death calculated within 24 hrs of admission or at the time of death, and more accurate than the CRP level in classifying survivors and nonsurvivors of MSS. These results accord with those of Hatherill et al who observed, in 37 children with MSS, that admission PCT level (values not indicated) was higher in nonsurvivors (11%) than in survivors (p=0.04) and related to the severity of organ failure (p=0.02); however, in the whole group of children with septic shock whatever the causative organism, admission PCT functioned worse than the PRISM score (AUC 0.73 (0.59–0.88) v 0.83 (0.71–0.93); statistical comparison not performed).8

The PRISM score is accepted in PICUs worldwide and has been reported to accurately predict outcome of meningococcal disease.9,10 However, as it needs a 24 hour observation period, it cannot be used as an inclusion criterion for clinical trials. Admission PCT could represent a good alternative tool if further studies confirm its ability to predict mortality.

Table 1

Performane characteristics of PCT, CRP, and PRISM score in 35 children with MSS

Figure 1

ROC curves (AUC±SE) for PCT, CRP, and PRISM score in 35 children with MSS (PCT v PRISM value, p=0.45; PCT v PRISM probability, p=0.31; PCT v CRP, p=0.06; CRP v PRISM value, p<10−2; CRP v PRISM probability, p<10−2).


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.