Diagnosing pneumonia in primary care: implementation of C-reactive protein point-of-care testing in daily practice

Cough is a very common symptom in the community, and lower respiratory tract infections (LRTIs) occur very frequently in all seasons. The primary goal in the diagnostic work-up of LRTI is to differentiate serious conditions like community acquired pneumonia (CAP) from self-limiting conditions and to prevent unnecessary prescribing of antibiotics. To improve diagnostic accuracy additional diagnostic testing is potentially useful. A promising additional test is C-reactive protein (CRP). A point-of-care (POC) test in capillary blood yields CRP results within a few minutes, and can thus be applied in decision making (whether or not to prescribe antibiotics) during consultation. POC CRP testing was introduced in Dutch primary care after incorporation in the guideline LRTI of the Dutch College of General Practice in 2011. So far, little is known about the compliance of GPs with these guidelines and the impact of POC CRP introduction on prescription of antibiotics for LRTI. This thesis aimed to evaluate optimal use of the point-of-care (POC) C-reactive protein (CRP) in primary care and its impact in day-to-day practice. More specifically our aims were to determine: [1] the (added) diagnostic value of CRP in the diagnostic work-up of community acquired pneumonia (CAP) in primary care, [2] the suitability of POC CRP test devices for primary care and [3] the impact of the POC CRP test result on the general practitioner’s (GP) decision to prescribe antibiotics. The first part of this thesis showed that a prediction model including coryza, dyspnoea, crackles, diminished breath sounds, fever, and tachycardia, had the best diagnostic performance in diagnosing CAP. Moreover, adding CRP to signs and symptoms in the diagnostic workup of CAP improved both discrimination and risk classification. Nevertheless, in the majority of patients uncertainty about the diagnosis remained after adding CRP, because their predicted risk of CAP after application of the risk model remained in the ‘intermediate’ range. The second part of this thesis showed that five different POC CRP test devices show considerable variation in analytical performance and agreement with a laboratory reference test. These differences do not translate into differences in accuracy of predicting (radiographic) CAP, both as single test as well as in combination with available clinical findings. In the third part of this thesis we concluded that implementation of POC CRP testing did not reduce overall antibiotic prescribing for patients presenting with acute cough. Moreover, only a minority of patients who underwent POC CRP testing had an indication for testing according to guideline recommendations. The POC CRP test result did influence the GPs’ decision making to prescribe antibiotics in nearly one third of all cases. Finally, only a quarter of all eligible patients were recruited in our implementation study, which resulted in overestimation of POC CRP test use, antibiotic prescribing and in an overestimation of the effects of POC CRP test use on antibiotics prescribing.