Association of C-reactive protein with cause, duration and severity of lower respiratory infections in primary care: a prospective cohort study.

Introduction: The association of C-reactive protein (CRP) with cough duration and severity has not been evaluated in a contemporary US population of patients with acute lower respiratory tract infection (LRTI).

Methods: We identified outpatients with less than 2 weeks of cough plus at least one lower respiratory or systemic symptom. The duration of illness, demographics, signs, symptoms, CRP and PCR for 46 pathogens were assessed. Duration and severity of cough were tracked using a paper diary plus responses to text messages. We classified CRP as low (<5 mg/L), moderate (5-19 mg/L) and high (≥20 mg/L).

Results: A total of 575 patients had valid CRP data and 503 had both valid CRP and PCR data. Patients with a CRP ≥20 mg/L were more likely (p<0.001) to have any bacterial detection (34.4%) or a mixed detection (37.7%) than viral infection (19.7%) or no detection (7.2%). Patients with elevated CRP were also more likely to have a likely bacterial pathogen detected (61.2% vs 37.8%, p<0.001) and were more likely to have a clinician label their illness as moderate or severe (53.7% vs 26.0%, p<0.001). Patients with a CRP ≥20 mg/L had a longer duration of illness than those with a lower CRP (18.5 vs 16.1 days, p=0.026) as well as a greater overall severity of cough (29.9 vs 23.0 points, p=0.001). In multivariable analysis, CRP ≥20 mg/L was the strongest independent predictor of a likely bacterial pathogen (adjusted OR 3.21, 95% CI 1.61 to 6.40).

Discussion: CRP ≥20 mg/L has a strong, independent association with the presence of bacterial pathogen in patients with acute LRTI and predicts longer duration and severity of illness. Further research is warranted to understand the impact of CRP on antibiotic prescribing and patient outcomes.