Early diagnostic suggestions improve accuracy of GPs: a randomised controlled trial using computer-simulated patients

Background Designers of computerised diagnostic support systems (CDSSs) expect physicians to notice when they need advice and enter into the CDSS all information that they have gathered about the patient. The poor use of CDSSs and the tendency not to follow advice once a leading diagnosis emerges would question this expectation. Aim To determine whether providing GPs with diagnoses to consider before they start testing hypotheses improves accuracy. Design and setting Mixed factorial design, where 297 GPs diagnosed nine patient cases, differing in difficulty, in one of three experimental conditions: control, early support, or late support. Method Data were collected over the internet. After reading some initial information about the patient and the reason for encounter, GPs requested further information for diagnosis and management. Those receiving early support were shown a list of possible diagnoses before gathering further information. In late support, GPs first gave a diagnosis and were then shown which other diagnoses they could still not discount. Results Early support significantly improved diagnostic accuracy over control (odds ratio [OR] 1.31; 95% confidence interval [95% CI] = 1.03 to 1.66, P = 0.027), while late support did not (OR 1.10; 95% CI = 0.88 to 1.37). An absolute improvement of 6% with early support was obtained. There was no significant interaction with case difficulty and no effect of GP experience on accuracy. No differences in information search were detected between experimental conditions. Conclusion Reminding GPs of diagnoses to consider before they start testing hypotheses can improve diagnostic accuracy irrespective of case difficulty, without lengthening information search.