Intended for healthcare professionals

Practice Rational Testing

Role of C reactive protein and procalcitonin in the diagnosis of lower respiratory tract infection in children in the outpatient setting

BMJ 2021; 373 doi: (Published 11 June 2021) Cite this as: BMJ 2021;373:n1409
  1. Andrés Pérez-López, assistant professor of clinical pathology and laboratory medicine1 2,
  2. Adam Irwin, senior lecturer in paediatric infectious diseases3 4,
  3. Carlos Rodrigo, associate professor of paediatrics5 6,
  4. Cristina Prat-Aymerich, assistant professor of medical microbiology7 8 9
  1. 1Divison of Microbiology Sidra Medicine, Doha, Qatar
  2. 2Weill Cornell Medical College in Qatar, Doha, Qatar
  3. 3UQ Centre for Clinical Research, The University of Queensland, Herston, Queensland, Australia
  4. 4Children’s Health Queensland Hospital and Health Service, South Brisbane, Queensland, Australia
  5. 5Department of Pediatrics, Hospital Universitari Germans Trias I Pujol, Badalona, Spain
  6. 6Autonomous University of Barcelona, Badalona, Spain.
  7. 7Department of Microbiology, Hospital Universitari Germans Trias I Pujol, Badalona, Spain
  8. 8CIBER Enfermedades Respiratorias, University Medical Center Utrecht, Utrecht University, Utrecht, the Netherlands
  9. 9Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, the Netherlands
  1. Correspondence to: A Pérez-López aperezlopez{at}

What you need to know

  • The difficulty of discriminating between viral and bacterial lower respiratory tract infection (LRTI) in children using clinical features alone often leads to overprescription of antibiotics

  • Biomarkers such as C reactive protein (CRP) and procalcitonin (PCT) have a limited capacity to rule in bacterial pneumonia in children in ambulatory settings where the prevalence of bacterial pneumonia is low. (CRP and PCT have limited diagnostic value in severely ill children who meet criteria for pneumonia or sepsis and who are candidates for broad spectrum antibiotic therapy)

  • There is growing evidence that antibiotic therapy can be safely withheld in children who are not severely ill with equivocal clinical presentation and low CRP (<20 mg/L) and PCT (<0.5 μg/L) levels

A previously healthy and fully vaccinated (including 13-valent pneumococcal conjugate vaccine) 22 month old boy is brought to the emergency department because of a 12 hour history of high fever (up to 40°C). He had had low grade fever, runny nose, cough, and decreased oral intake for the past two days. On examination, he did not look severely ill but was febrile (38.3°C). His respiratory rate was 45 breaths/minute (normal range 25-40 breaths/min at 18-24 months old), heart rate was 140 beats/minute (normal range 98-135 beats/min at 18-24 months), and blood oxygen level was 95%. Although breath sounds were not decreased, some bibasilar crackles were noted on chest auscultation. A chest x ray was interpreted as having bilateral peribronchial infiltrates and haziness in the right lower lobe. To aide their decision whether to initiate antibiotic therapy, clinicians requested blood tests, which revealed a white blood cell count of 22.5×109/L (60.0% neutrophils), a CRP of 30 mg/L (normal <5 mg/L), and a PCT of 0.25 μg/L (normal <0.5 μg/L).

Lower respiratory tract infections (LRTIs) in childhood are commonly of viral aetiology. Distinguishing viral from bacterial LRTI in children—and thus appropriately prescribing antibiotics—solely based on a medical history and physical examination can be challenging.1 In these circumstances, an accurate marker for bacterial pneumonia would be useful in order to prevent return to medical care (if bacterial infection was not treated) and to avoid antibiotic use and adverse effects (when the underlying cause is viral). The National Institute for Health and Care Excellence (NICE) recommends point-of-care testing of C reactive protein (CRP) to guide antibiotic therapy for adults with symptoms of LRTI and diagnostic uncertainty after a clinical assessment (antibiotic treatment should be offered to patients with CRP levels >100 mg/L and avoided for CRP levels <20 mg/L).2 Although CRP and procalcitonin lack sufficient sensitivity and specificity to rule in bacterial pneumonia in children in ambulatory care,34 data generated over the past few years suggest that both biomarkers could help clinicians to reduce diagnostic uncertainty and unnecessary antibiotic prescriptions in a subset of children with LRTI and equivocal clinical features.

What are C reactive protein and procalcitonin?

C reactive protein (CRP) and peripheral white blood cell count are the most common biomarkers for infection in clinical practice worldwide. CRP, which is primarily produced by the liver in response to inflammation, plays a major role in inducing complement activation and facilitating phagocytosis by macrophages.56 Procalcitonin (PCT) is a precursor peptide of the hormone calcitonin, which is secreted by a wide range of parenchymal cells in response to systemic inflammation. Although both biomarkers have a good negative predictive value to rule out serious bacterial infections, PCT is increasingly used to identify severe bacterial infections in children such as urinary tract infection and meningitis and to determine the risk of serious bacterial infection in infants with fever of unknown source and oncology patients with neutropenic fever, since it shows a more specific increase in response to bacterial infection, becomes elevated faster, and decreases earlier in response to appropriate antibiotic therapy than CRP (table 1).578 PCT testing is mostly performed in emergency care settings in middle-high income settings because of its higher cost and longer turnaround time than CRP testing; CRP is widely used in primary care, including in some low income countries, because of its affordability and fast turnaround time (table 1).5678

Table 1

Comparison between C reactive protein and procalcitonin as biomarkers for bacterial infection

View this table:

The optimal cut-off values for CRP and PCT to rule in or rule out bacterial LRTI in ambulatory care have not been established. Nevertheless, among febrile children assessed in acute care settings with intermediate (5.0-20.0%) to high (>20.0%) prevalence of serious bacterial infections, a CRP value <20 mg/L or a PCT value <0.5 μg/L makes a serious bacterial infection improbable, whereas it should be suspected if the CRP level is >80 mg/L or PCT value is >2 μg/L.9

Can CRP and PCT improve diagnosis of bacterial LRTI in children?

PCT and CRP lack diagnostic sensitivity to rule in bacterial (compared with viral) LRTI in children in primary care settings, since the prevalence and pre-test probability of serious bacterial infections is low in this scenario, at least in high income countries.

Studies evaluating the ability of PCR and PCT to predict bacterial pneumonia caused by typical microorganisms in children who present to emergency services found variable sensitivities, ranging from 44.0% to 94.0%, using optimal cut-off values (>1.5 μg/L for PCT and >65 mg/L for CRP).10111213 However, such high serum concentrations are often observed in patients who are sufficiently ill to require hospitalisation and in whom bacterial pneumonia may be efficiently diagnosed by a typical clinical assessment. In other words, for ill-appearing children who meet traditional clinical criteria for bacterial pneumonia, CRP and PCT at the thresholds that best predict bacterial pneumonia do not seem to provide additional information beyond a comprehensive clinical evaluation. Few studies have compared the diagnostic performance of both biomarkers with clinical diagnosis. For example, in a prospective cohort study with 75 children hospitalised with community acquired pneumonia, including 37 patients with presumed pneumococcal aetiology, PCT ≥1.5 μg/L was 94% sensitive for pneumococcal pneumonia, with a 1.99 positive likelihood ratio. Based on a pre-test probability of 49%, PCT increased post-test probability to 65%.13 The added clinical value of this post-test probability—that is, the proportion of patients who could benefit from PCT testing—is unclear without knowing the post-test probability of a medical history and physical exam. Further, specificities and positive predictive values for both biomarkers rarely reach 80.0%, indicating a substantial number of viral infections (false positives) among patients with high serum CRP and PCT concentrations.10111213

Few studies evaluate the performance of CRP and PCT to predict bacterial pneumonia in children in primary care. A Finnish group measured CRP and PCT in 193 and 190 serum samples of children with radiologically confirmed pneumonia managed in primary care. There were no significant differences in mean CRP and median PCT concentrations among children with serological evidence of pneumococcal infection compared with those with atypical and viral pneumonia. In fact, it was found that mean CRP and median PCT concentrations were <30 mg/L and <0.5 μg/L regardless of the aetiology.1415

Can CRP and PCT improve antimicrobial prescribing in children with LRTI in the outpatient setting?

Although most ambulatory children with LRTI do not need antibiotic treatment, there is likely some added clinical value of a low CRP or PCT level to rule out bacterial pneumonia and reduce antibiotic use in well-appearing children in whom the distinction between bacterial and viral LRTI infection is unclear after a thorough clinical assessment. Evidence suggests that PCT <0.5 μg/L can accurately identify adults and children with low risk of typical bacterial pneumonia, particularly Streptococcus pneumoniae. For instance, a multicentre, population-based, prospective, active surveillance study that analysed PCT levels in 532 hospitalised children with radiologically confirmed pneumonia found a 96.0% negative predictive value for typical bacteria among 242 children (45.0% of the cohort) with PCT values <0.25 μg/L. In fact, none of the 120 children with PCT values <0.1 μg/L had typical bacteria detected.16 These and other results suggest that antibiotic therapy can be safely withheld in children who appear well with low PCT levels and equivocal clinical presentation.713161718

Low CRP concentrations may similarly reduce the likelihood of bacterial LRTI and, in turn, reduce unnecessary antibiotic prescribing, but the evidence is more mixed. Currently, point-of-care CRP testing is routinely used in the diagnostic work-up of adults with LRTI in primary care in several high income countries, where low CRP concentrations combined with clinical assessment have successfully reduced antibiotic prescription.619 This benefit has not always been observed in children; this may be related to poor adherence to CRP-guided prescribing guidelines plus an overestimation of the positive predictive value of a moderately elevated CRP level.202122 In the Netherlands, for example, where national guidelines on the management of LRTI are similar to NICE guidelines,23 a trial that randomised 309 non-seriously ill children aged 3 months to 12 years with fever and cough from 28 primary practices to receive either clinical assessment plus point-of-care CRP testing or clinical assessment only, did not find a significant difference in antibiotic prescription rates between both groups. Notably, among 170 children who had CRP measured, only 4% had CRP >100 mg/L. However, 14.0% with CRP <10 mg/L and 44.0% with CRP between 10 and 100 mg/L had antibiotics prescribed, suggesting an overprescription of antibiotics in a fraction of patients with low and intermediate CRP values.22

By contrast, a study performed in nine primary care practices in Tanzania enrolled 1726 non-seriously ill febrile children aged 2-59 months with cough, who were randomly allocated to two groups. The intervention group received antibiotic treatment based on sequential use of the World Health Organization clinical criteria to define childhood pneumonia (tachypnoea and chest indrawing) and point-of-care CRP testing. Patients meeting the clinical criteria plus a CRP >80 mg/L were deemed to have a bacterial LRTI and had oral antibiotics prescribed (20/865 patients; 2.3%). The control group was treated with antibiotics based on clinical criteria only (345/854 patients; 40.4%). Notably, antibiotic prescription was almost 20-fold lower in the intervention group (risk ratio 0.06, 95% confidence interval 0.04 to 0.09) and secondary hospital admissions and deaths were also significantly lower compared with the control group (risk ratio 0.30, 0.10 to 0.93).24

While CRP and PCT lack sensitivity to definitively diagnose bacterial LRTI, there is growing evidence that both biomarkers can be used to reduce unnecessary antibiotic exposure in well-appearing children in whom the distinction between bacterial and viral LRTI is not possible after a thorough clinical assessment in the outpatient setting.

Case outcome

Since the patient was clinically stable and his PCT level was low, his mother was reassured that he probably had a viral infection that would not benefit from antibiotic treatment. Consequently, the child was discharged home without antibiotics. A follow-up visit three days later with his general practitioner showed that he had been completely apyrexial for the past 24 hours and was feeding better. No further follow-up was advised.

How patients were involved in the creation of this article

Testing for C reactive protein (CRP) and procalcitonin (PCT) are available 24 hours a day in our hospital. We sought feedback from parents of a group of children with fever and respiratory symptoms assessed in our emergency department in whom PCT and CRP testing were used in combination with clinical judgment to withhold antibiotics. The clinical scenario presented was elaborated based on these experiences. Interestingly, all the parents believed that antibiotics are overprescribed in acute respiratory infections in children, which led us to emphasise the value of low CRP and PCT concentrations to improve clinical decision making in children with lower respiratory tract infections and diagnostic uncertainty in ambulatory care.

Rational testing into practice

  • Think about how many children with lower respiratory tract infections (LRTI) are prescribed antibiotics due to diagnostic uncertainty in clinical practice. Do you think that point-of-care testing for C reactive protein (CRP) or procalcitonin (PCT) could help optimise antibiotic prescribing in LRTI in children in your practice?

  • High serum concentrations of CRP and PCT are often found in seriously ill children with pneumococcal pneumonia assessed as emergency department outpatients. However, what proportion of these patients would be missed by a meticulous clinical assessment?

How this article was made

We reviewed British national guidelines on the assessment and management of children and adults with lower respiratory tract infections. Relevant articles were searched using PubMed and the following keywords: C-reactive protein (CRP), procalcitonin (PCT), biomarkers, lower respiratory tract infections, pneumonia, children. Furthermore, the manuscript was reviewed by four internationally renowned specialists in the fields of paediatric infectious diseases and emergency medicine whose comments and views contributed to produce this review.


  • Contributors: APL conceived the paper and wrote the manuscript with input from AI. CR and CP critically revised and helped finalise the manuscript. All authors contributed to the intellectual content and approved the final version of the manuscript.

  • Competing interests: We have read and understood the BMJ policy on declaration of interests and have no relevant interests to declare.

  • Provenance and peer review: Commissioned; externally peer reviewed.