The use of c-reactive protein in acute oncology admissions


Session type:

Kirsty Cavanagh1,Mark Baxter2,Ellie Dow3,Judith A Strachan4,Gillian Smith5,Michelle Ferguson6
1School of Medicine, University of Dundee, Ninewells Hospital & Medical School,2Division of Molecular & Clinical Medicine, School of Medicine, University of Dundee, Ninewells Hospital & Medical School,3Department of Clinical Biochemistry, Ninewells Hospital & Medical School,4Department of Blood Sciences, Ninewells Hospital & Medical School,5Division of Cellular Medicine, School of Medicine, University of Dundee, Ninewells Hospital & Medical School,6Tayside Cancer Centre, Ninewells Hospital & Medical School



Cancer represents a significant NHS economic burden.  The Chief Medical Officer acknowledges changes are necessary to cope with increasing healthcare demands, highlighting the need to reduce unnecessary harm and waste of resources.  Laboratory tests influence approximately 70% of clinical decisions and are a key target for improvement, particularly the overuse of certain investigations.

C-Reactive Protein (CRP) is a serum inflammatory marker commonly requested in suspected infection.  Its value has been questioned due to low specificity, particularly in oncology, as cancer is an inflammatory process.  We aimed to investigate the use of CRP in acute oncology admissions.


Patients who had CRP requested on admission to Ninewells oncology department between 1/7/18-31/8/18 were identified retrospectively.  Clinical data was collated from electronic records and admission blood results recorded.  Notes were reviewed and final diagnosis recorded.  Comparison of median laboratory values between clinically-defined infectious and non-infectious diagnoses was conducted.  Receiver-operated characteristic (ROC) analysis was used to understand the diagnostic value of CRP.  Binary logistic regression was then performed.  Local ethical approval was obtained.


142 patients were included in the study.  The median value of CRP in clinician-determined infectious (n=47) vs non-infectious (n=95) diagnoses were 95mg/L and 47mg/L respectively (p=0.002).  CRP AUC was 0.684 95% CI (0.580-0.789) on ROC analysis.  Evaluation of ROC curve co-ordinates revealed AUC was maximal at 76mg/L.  Binary logistic regression identified CRP as the only independent serological marker with predictive value (p=0.013, HR 1.029 95% CI (1.006-1.053)).


Requesting of CRP in an acute setting is common regardless of suspicion of infection.  In our study, a CRP cut-off value of 76mg/L was identified as providing increased specificity for the diagnosis of infection.  A larger study could aim to identify a cut-off for use in clinical practice.