Objectives: In the field of sleep medicine, there is a paucity of evidence-based curriculum development strategies in the literature. We chose to determine the number of polysomnograms (PSG) necessary to be scored by sleep fellows in order to reasonably approximate sleep scoring by a Diplomate of the American Board of Sleep Medicine (DABSM).
Design: The fifth PSG scored by two sleep fellows during the 12 consecutive months of training was chosen for analysis. A DABSM not involved in the training of fellows scored sleep on each of the selected PSG with replication of montage and filter settings. Epoch by epoch comparison of sleep stage scoring is described as the frequency of concordance between fellow and DABSM (f correct).
Measurements and results: The mean (SD) f correct for all PSG for each fellow was 0.83 (0.06) and 0.83 (0.08) (p = 0.93). Concordance between sleep fellow and DABSM approached 0.8 after scoring between 20-30 PSG. This milestone was reached after the fourth month of training. F correct was highest for stage 2 sleep and REM sleep and most variable for slow wave sleep and stage 1 sleep. The variability in f correct for these stages was in part related to the relative paucity of these sleep stages.
Conclusions: Scoring of sleep becomes reasonably proficient after scoring approximately 20-30 PSG and/or four months of dedicated sleep disorders training. A standard measure of concordance that corrects for epoch sample size may be helpful for use in future similar investigations.