Analysis of recruitment in a pragmatic observational study on C-reactive protein point-of-care testing in primary...

Failure to recruit all eligible study patients can lead to biased results. Little is known on selective patient recruitment in studies on implementation of diagnostic devices.

The aim of this observational study was to measure recruitment of patients in an implementation study in primary care on use of point-of-care (POC) C-reactive protein (CRP) and to evaluate recruitment bias and its impact on the study endpoint.

In a cross-sectional observational study on POC CRP implementation and related antibiotics prescribing, we compared included patients with all eligible patients to assess the representativeness of the included subjects. Eligible patients were adults presenting with acute cough in primary care between March and September 2012. The frequency of POC CRP testing and the proportion of prescribed antibiotics were compared between recruited and non-recruited patients. As measure of bias, odds ratios (ORs) with accompanying 95% confidence intervals (CIs) for the association between CRP level (<20 mg/l or not) and antibiotic prescribing were computed.

Of all 1473 eligible patients 348 (24%) were recruited. In recruited patients, POC CRP tests were conducted and antibiotics prescribed more frequently as compared to non-recruited patients (81% versus 6% and 44% versus 29%, respectively). The ORs were 18.2 (95%CI: 9.6-34.3), 30.5 (95%CI: 13.2-70.3) and 3.8 (95%CI: 0.9-14.8) respectively in all eligible patients, the recruited and the non-recruited patients.

Selective recruitment resulted in an overestimation of POC CRP test use and antibiotic prescribing.