|| Checking for direct PDF access through Ovid
We examined the effect of interview characteristics (ie, recall interval, interview version) on estimates of health-related lost productive work time (LPT). Three versions of a telephone interview were administered using 7-day and 4-week recall periods. In a population-based survey, 7674 workers randomly were assigned to one of six interviews at contact; 615 participants received a follow-up interview. We found strong evidence of under-reporting using a 4-week recall period and a not significant trend in over-reporting LPT using a 7-day recall period. Of the three interviews, version 3 could be administered most quickly, on average, and yielded the most discriminating estimates of LPT by health condition (ie, headache, allergic rhinitis, and cold/flu). Our data suggest that variation in relatively short recall periods influences estimates of health-related LPT. A 2-week recall period may be optimal for minimizing overall reporting error but requires additional research to verify.