Validity of 12-month falls recall in community-dwelling older women participating in a clinical trial

To compare 12-month falls recall with falls reported prospectively on daily falls calendars in a clinical trial of women aged >/=70 years. Methods. 2,096 community-dwelling women at high risk of falls and/or fracture completed a daily falls calendar and standardised interviews when falls were recorded, for 12 months. Data were compared to a 12-month falls recall question that categorised falls status as no falls, a few times, several, and regular falls. Results. 898 (43 ) participants reported a fall on daily falls calendars of whom 692 (77 ) recalled fall(s) at 12 months. Participants who did not recall a fall were older (median 79.3 years versus 77.8 years, P = 0.028). Smaller proportions of fallers who sustained an injury or accessed health care failed to recall a fall (all P <0.04). Among participants who recalled no fall, 85 reported zero falls on daily calendars. Few women selected falls categories of several times or regular (4.1 and 0.4 , resp.) and the sensitivity of these categories was low (30 to 33 ). Simply categorising participants into fallers or nonfallers had 77 sensitivity and 94 specificity. Conclusion. For studies where intensive ascertainment of falls is not feasible, 12-month falls recall questions with fewer responses may be an acceptable alternative.