Reliability of telemedicine examination.

OBJECTIVES To assess the reliability of telemedicine examination and identify the issues to be addressed if the conduct of physical examination and the reading of images and tracings by telemedicine are to be as reliable as conventional examination and reading. METHODS Patients were examined both conventionally and by telemedicine in 12 clinics, and the results were compared. There were 1826 matched pairs of observations. Cardiac auscultation, echocardiography, electrocardiography, electroencephalography, obstetric ultrasonography, ophthalmologic examination, physical therapy assessment, pulmonary auscultation, and the reading of chest radiographs with telemedicine cameras and monitors were studied. The main outcome measure was agreement between the telemedicine findings and a criterion standard. RESULTS For ophthalmology, physical therapy, and cardiac auscultation, 91.2% of the conventional findings and 86.5% of the telemedicine findings were identical or similar to the criterion standard. The kappa coefficient on matched-pair analysis was 0.66. For pulmonary auscultation and reading of chest films with a telemedicine camera and monitor abnormalities were suppressed at default settings but subsequently revealed with extensive manipulation of system settings. For tracings and images, both conventional and telemedicine findings showed 92% reliability, with a kappa coefficient of 0.87. CONCLUSIONS On the basis of these observations and the methods used, reliability varied with the type of examination, clinician experience with telemedicine, and participant knowledge of system limitations. Clinicians without experience or knowledge of system limitations missed findings of clinical importance. Improvements in equipment since the clinics were conducted in 1994 may have resolved some of these problems. Our findings raise doubts about the reliability of occasional telemedicine consultations by clinicians inexperienced in the technology.