Context Despite frequent complaints that biomedical knowledge is quickly forgotten after it has been learned, few investigations of actual long-term retention of basic science knowledge have been conducted in the medical domain. Objectives Our aim was to illuminate the long-term retention of basic science knowledge, particularly of unrehearsed knowledge. Methods Using a cross-sectional study design, medical students and doctors in the Netherlands were tested for retention of basic science knowledge. Relationships between retention interval and proportion of correct answers on a knowledge test were investigated. Results The popular notion that most of basic science knowledge is forgotten shortly after graduation is not supported by our findings. With respect to the full test scores, which reflect a composite of unrehearsed and rehearsed knowledge, performance decreased from approximately 40% correct answers for students still in medical school, to 25-30% correct answers for doctors after many years of practice. When rehearsal during the retention interval is controlled for, it appears that little knowledge is lost for 1.5-2 years after it was last used; from then on, retention is best described by a negatively accelerated (logarithmic) forgetting curve. After >= 25 years, retention levels were in the range of 15-20%. Conclusions Conclusions about the forgetting of unrehearsed knowledge in this study are in line with findings reported in other domains: it proceeds in accordance with the Ebbinghaus curve for meaningful material, except that in our findings the 'downward' part appears to start later than in most other studies. The limitations of the study are discussed and possible ramifications for medical education are proposed.
