A review of some of the latest low-resources emergency medicine evidence

Point of care CRP to reduce antibiotic prescribing without leading to worse outcomes.

Safety and Efficacy of C-reactive Protein–guided Antibiotic Use to Treat Acute Respiratory Infections in Tanzanian Children: A Planned Subgroup Analysis of a Randomized Controlled Noninferiority Trial Evaluating a Novel Electronic Clinical Decision Algorithm (ePOCT)

Kristina KeitelJosephine SamakaJohn MasimbaHosiana TembaZamzam SaidFrank KagoroTarsis MlaganileWilly SanguBlaise GentonValerie D’Acremont


To establish if CRP testing can improve clinical outcomes AND decrease antibiotic prescribing in children under 5. 


Fever and cough are common presenting complaints in low-resource settings; it is very difficult to reliably diagnose those with pneumonia who need antibiotics and those who don’t. The WHO advice on using respiratory rate lacks both specificity and sensitivity. There is a need for additional tools to help differentiate bacterial and non-bacterial aetiology. 


Planned subgroup analysis of larger open-label randomised non-inferiority trial. 


9 Outpatient clinics in Sar el Salam, Tanzania.


1726 consecutive children aged 2 months to 5 years who presented with a temperature >37.5 and a cough.

Excluded those with signs of severe illness (including Severe acute malnutrition)


All children were managed with a validated algorithm to manage febrile children. This including POC malaria testing and treatment if positive, and non treatment if respiratory rate <50 or chest indrawing. Salbutamol given if wheezing. 


In the intervention group a CRP was undertaken and antibiotics were only prescribed if >80. If CRP>80 salbutamol was also given.


Primary outcome

Proportion of children with clinical failure at day 7 (broad definition)

Secondary outcomes

Proportion of children prescribed antibiotics at DO and at D1-D6.


2.9% of patients experienced clinical failure in the CRP group vs 4.8 % in the control group.

2.3% of children receive antibiotics on D0 in the group vs 40/4% in control group


Author's Discussion

They state the two step algorithm improved clinical outcomes and led to less antibiotic prescribing.

They draw attention to the fact that very few children were above the CRP threshold therefore questioning the need for testing or antibiotic treatment at all in this population. 


Exclusion of malaria patients increases the utility of this tool in endemic areas.  

Broad definition of treatment failure increased validity of safety. 

Intention to treat analysis



External validity in populations without Hib vaccine coverage questionable, if higher incidence of pneumonia may change safety profiel

Unsure why salbutamol was given to all children in the CRP>80 group. 

Use of RR <50 in exclusion group despite suggesting this lacks sensitivity in their introduction. 

Urban location in a single city. 

This is a non-inferiority design so I don’t think the author’s should be drawing conclusions about improving outcomes 


 My humble opinion

The decision for which children can be a difficult one, finding the few patients who need it amongst the large volume of children with cough and fever is a challenge. 

In my experience, the use of malaria RDT is very useful in empowering clinicians to give no treatment to well patients. 

I can see a POC CRP being used similarly.

I would not object to add a point of care CRP being used as a tool to decrease antibiotic prescribing, until then I will continue to not prescribe antibiotics to well children with no signs of severe disease, however would still prescribe to those with a high respiratory rate in settings without vaccine coverage and for those who returning to a clinic is difficult.





Written by Daniel Roberts

Any questions or comments @poor_evidence on twitter, or by email me directly.