• 
• 

Three Part Question

In [young children with fever without apparent focus on physical examination and history] is [measurement of C-reactive protein a good screening test] to [detect occult bacterial infection or bacteraemia]?

Clinical Scenario

You are a specialist registrar in a paediatric day assessment unit and often see young children with high fever but no obvious focus of infection on physical examination. You wonder if a screening test can be done in such patients to identify those with occult bacterial infection. You know that in some units C-reactive protein (CRP) is measured routinely in all children with high fever. However, you are not sure if CRP measurement is a good screening test to detect occult bacterial infection in a young child with fever without apparent focus. You decide to find out more.

Search Strategy

Medline (1951–2004) via Dialog DATA star
"C-reactive protein or acute phase reactant$ or acute phase protein$ or CRP AND fever without focus or fever of unknown origin or occult bacteraemia"
Pubmed and Embase: No further relevant articles were found.
Cochrane library and Bestbets website: No further relevant article was found.

Search Outcome

Outcome: A total of 58 articles were found. This was limited to 23 articles by selecting those in the English language and human studies related to children (up to 18 years of age). Each abstract was read and six relevant studies were found. One of these studies (Berger et al) was relevant but was designed to look at the use of CRP in differentiating bacterial and viral infection, and children with identifiable focus of infection were included in the study. Therefore this was excluded from the analysis. Five more relevant articles were obtained from the references of the above studies (Stanley, Bennish, Putto, McCarthy, Peltola). However, designs of these studies were similar to those of Berger et al, and all included children with identifiable focus of infection in their analysis. Subsequently none of them were included in the analysis here.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Kholi et al, 1993	100 children from 3 months to 3 years of age with temperature >38.5° of less than 3 days duration without an apparent focus on physical examination and history; normal CXR and a peripheral film negative for malarial parasites. Patients with underlying neoplastic or immunosuppressive condition or with a chronic liver, heart and renal disease and those who had already received antibiotics were excluded	Exploratory prospective cohort study (level IIb)	CRP, total and differential leucocyte count and ESR in children with laboratory or radiographically proven bacterial illness and non-bacterial illnesses	The sensitivity of serum CRP >4 mg/dl for diagnosis of bacteraemia was 95%, specificity 86% with and likely hood ratio of 6.8 None of the other acute phase reactants individually or in combination had better sensitivity or specificity than CRP The sensitivity of serum CRP >4 mg/dl for diagnosis of bacteraemia was 95%, specificity 86% with and likely hood ratio of 6.8 None of the other acute phase reactants individually or in combination had better sensitivity or specificity than CRP	High incidence of proven bacterial infection (25%) in the study population Children with possible but not proven bacterial infection (13%) were not included in statistical analysis Confidence interval of sensitivity and specificity values were not given
Pulliam et al, 2001	77 children from 1 to 36 months of age with temperature >39°C and clinically undetectable source of fever. Children with acute otitis media, acute pharyngitis, clinical pneumonia, acute respiratory tract infection, acute gastroenteritis, those with a history of antibiotic use during the past 7 days, a known underlying immunologic disease and who received vaccination during the previous two days were excluded	Exploratory prospective cohort study (level IIb)	WBC, absolute neutrophil count (ANC), Band count, CRP in children with laboratory or radiographically proven serious bacterial infections	Multivariate logistic regression analysis suggested CRP as an only predictor of serious bacterial infection (Beta = 0.76, 95% CI: 0.64, 0.89) A CRP value of >7 mg/dl had sensitivity of 79%, specificity of 91%, likelihood ratio 8.3 (95% CI: 3.8, 27.3) to detect occult bacterial infection CRP concentration of >9 mg/dl had a likelihood ratio of SBI of 9 CRP had better predictive value than WBC or ANC	The incidence of occult bacterial infection in the study population was high (18%) There was possibility of sampling bias as well appearing children were more likely to be discharged from emergency department without having had any laboratory studies than those who were more ill appearing
Bleeker et al, 2001	231 children aged 1–36 months who attended a paediatric emergency ward with fever >38°C without apparent focus of infection. Reviewing standardised medical records collected patient data. Laboratory results were retrieved from computer documented hospital information system	Retrospective case reviews	Using multivariate logistic regression and ROC area the diagnostic value of predictors for serious bacterial infection from history, examination and laboratory tests was judged, resulting in a risk stratification	Independent predictors for SBI from laboratory tests were WBC count, CRP and the presence of >70 WBC in urinalysis (ROC area: 0.83). The risk stratification for serious bacterial infection ranged from 6–92%	High incidence of serious bacterial infection in the study population (25%) Retrospective study
Isaccman et al, 2002	256 febrile children between 3–36 months of age without apparent focus. Patients who had taken antibiotics within 48 hours of presentation and those who were known case of bacteraemia during the previous 48 hours were excluded	Exploratory prospective cohort (level IIb)	CRP, absolute neutrophil count (ANC), WBC in children with laboratory or radiographically proven occult bacterial infection	The sensitivity, specificity, and likelihood ratio of CRP value of >4.4 mg/dl was 63% (95% CI 43–82), 81% (76–81), 3.3 (1.7–4.3) respectively in detecting occult bacterial infection ANC had better predictive value for detection of occult bacterial infection than CRP	High incidence of occult bacterial infection in the study population (11.3%) Good sample size
Galetto-Lacour et al, 2003	99 children between age 1 week to 3 years with fever >38°C and no localising sign of infection seen at the emergency department. Children with fever lasting longer than 7 days, with known immunodeficiencies and who were treated with antibiotics during the two previous days were excluded	Exploratory prospective cohort (level IIb)	Procalcitonin (PCT), CRP, IL-6 levels and WBC with differential counts in children with laboratory or radiographically proven serious bacterial infection	The sensitivity, specificity, and likelihood ratio of CRP value of >4 mg/dl was 76% (95% CI 60–92), 79% (67–88) and 3.6 (1.8–7.6) respectively in detecting serious bacterial infection PCT and CRP performed better than IL-6 and/or band count in predicting the occurrence of serious bacterial infection	High incidence of serious bacterial infection (29%) in the study population Neonates from 1 week to 1 month of age were also included In non-toxic appearing children blood culture was only done if their leucocytes count was >15 g/l or band count was >1.5 g/l

Comment(s)

The management of febrile young children without apparent source of infection remains controversial, because there has been no test available with adequate sensitivity and specificity required to distinguish children with occult bacterial infection from non-bacterial illness. Blood culture is the gold standard to detect occult bacteraemia; however, results are not quickly available. Five studies evaluating the use of various acute phase reactants in this clinical situation are appraised here. The diagnostic nature of the question determines that the best possible research studies would be validating prospective cohort studies, but four of them were exploratory cohort studies and one was a retrospective analysis, and all of them had methodological flaws in them. Three prospective studies (Kohli, Pulliam, Galetto-Lacour) showed that CRP has better predictive value than other acute phase reactants, while one study (Isaacman) found ANC to have better predictive value. Interestingly, in all of them mean CRP was significantly higher in children with serious bacterial infections compared with children with benign infections, and when taken in conjunction with other acute phase reactants, gave good probability of serious bacterial infection. One may imagine that trend in CRP over time may be more important than a single CRP value, and a single very high CRP may have very high specificity and sensitivity to detect serious bacterial infections. However, none of these studies gave enough data to answer these two questions, and no other studies are available looking at the serial measurements of serum CRP to detect occult bacterial infection in paediatric population. The incidence of serious bacterial infections was high in all the studies, ranging from 11.3% to 29%. The prevalence of occult bacteraemia in non-toxic appearing children between 3 and 36 months of age with temperatures higher than 39°C has declined to about 2% following the introduction of conjugate vaccine against Haemophilus influenzae type b (Lee, Alpern). Recently, conjugate pneumococcal vaccine has been introduced in a few countries such as the USA, and has been shown to substantially reduce the rate of invasive pneumococcal disease in immunised children (Black) so a screening test to detect occult bacterial infections in children attending emergency departments in these countries may be of little value, as the pre-test probability is much lower. On the basis of published evidence, it can be concluded that high CRP can only suggest the presence of serious bacterial infection. Nevertheless, taken in conjunction with other acute phase reactants, it can contribute towards decision making.

Clinical Bottom Line

Children with serious bacterial infections and occult bacteraemia are more likely to have high CRP than children with benign infections. (Grade B) A single CRP value gives a probability but never a certainty of presence or absence of serious bacterial infection. (Grade B) The use of CRP alone or with other factors may enhance clinicians' abilities in the early recognition of clinically undetectable serious bacterial infection, allowing for a more selective strategy for determining which children need additional diagnostic studies and antibiotic therapy.

References

1. Kohli V, Singhi S, Sharma P. et al. Value of serum C-reactive protein concentrations in febrile children without apparent focus. Ann Trop Paediatr 1993;13:373–8.
2. Pulliam PN, Attia MW, Cronan KM. C-reactive protein in febrile children 1 to 36 months of age with clinically undetectable serious bacterial infection. Pediatrics 2001;108:1275–9.
3. Bleeker SE, Moons KG, Derksen-Lubsen G. et al. Predicting serious bacterial infection in young children with fever without apparent source. Acta Paediatr 2001;90:1226–32.
4. Isaacman DJ, Burke BL. Utility of the serum C-reactive protein for detection of occult bacterial infection in children. Arch Pediatr Adolesc Med 2002;156:905–9.
5. Galetto-Lacour A, Zamora SA. Gervaix A. Bedside procalcitonin and C-reactive protein tests in children with fever without localizing signs of infection seen in a referral centre. Pediatrics 2003;112:1054–60.
6. Berger RM, Berger MY, van Steensel-Moll HA. et al. A predictive model to estimate the risk of serious bacterial infections infants. Eur J Pediatr 1996;155:468–73.
7. Stanley T, Barrett D, Salmond CE. et al. Viral and bacterial infection in childhood: the value of C-reactive protein. N Z Med J 1991;104:138–9.
8. Bennish M, Beem MO, Ormiste V. C-reactive protein and zeta sedimentation ratio as indicators of bacteremia in paediatric patients. J Pediatr 1984;104:729–32.
9. Putto A, Ruuskanen O, Meurman O. et al. C-reactive protein in the evaluation of febrile illness. Arch Dis Child 1986;61:24–9.
10. McCarthy PL. Jekel JF. Dolan TF Jr. et al. Comparison of acute phase reactants in paediatric patients with fever. Pediatrics 1978;62:716–20.
11. Peltola H, Jaakkola M. C-reactive protein in early detection of bacteremic versus viral infections in immunocompetent and compromised hosts. J Pediatr 1988;113:641–6.
12. Lee GM, Harper MB. Risk of bacteremia for febrile young children in the post Haemophilus influenzae type b era. Arch Pediatr Adolesc Med 1998;152:624–8.
13. Alpern ER, Alessandrini EA. Bell LM. et al. Occult bacteraemia from a paediatric emergency department: current prevalence, time to detection and outcome. Pediatrics 2000;106:505–11.
14. Black SB, Shinefield HR, Hansen J. et al. Postlicensure evaluation of the effectiveness of seven-valent pneumococcal conjugate vaccine. Pediatr Infect Dis J 2001;20:1105–7.