The modified 6-item BGA risk stratification predicted length of hospital stay at a better rate, when compared to a priori risk stratification using history of falls as an item. Recall bias relative to the history of falls may be considered a limitation when using the a priori BGA. Falls are usually underreported because of the cognitive decline of fallers, who forget to report them. This bias may underestimate their predictive value for length of hospital stay. The use of a walking aid, which has similar value in terms of gait and/or balance impairment marking, is more objective when compared to history of falls and, thus, may more efficiently detect the highest-risk inpatients, as suggested by our results.