Is single observer identification of wound infection a reliable endpoint?

The majority of wound studies over the past 20 years have relied on single observer determination of infection presence or absence as a study endpoint. Nevertheless, there are few data on the reliability of those determinations. If single observer determinations are not reliable, the validity of studies using this endpoint should be questioned. Two physicians independently evaluated wounds for presence or absence of the primary endpoint of infection and the secondary endpoints of purulence, tenderness, warmth, and erythema. Kappa values were calculated as measures of inter-observer agreement, which was used as a measure of reliability. Of 125 patients enrolled, 115 wounds were available for analysis. Kappa =.57 [95%CI.39 to .75] for the primary endpoint of infection, .84 for purulence, .72 for tenderness, .48 for warmth, and .48 for erythema. Agreement on two features of infection and overall agreement on the presence or absence of wound infection was only 'moderate.' Single observer determination of wound infection by inexperienced observers using imprecise definitions, though commonly used, is an unreliable measure for scientific study.