Background: Fall incidence differs considerably between studies and countries. Reasons may be differences between study samples or different assessment methods. The aim was to derive estimates of fall incidence from two population-based studies among older community-living people in Germany and compare retrospective and prospective falls data collection methods.

Methods: Data were derived from the 2008–11 wave of the German health interview and examination survey for adults (DEGS1), and the Activity and Function of the Elderly in Ulm study (ActiFE-Ulm). Data collection took place in community facilities (DEGS1) or participants’ homes (ActiFE-Ulm). Participation rates were 42% (newly recruited) and 64% (panel component) in DEGS1 and 19.8% in ActiFE-Ulm. Self-report retrospective fall data covering the previous 12 month period in DEGS1 and ActiFE-Ulm were collected, but only ActiFE-Ulm used prospective 12 month fall calendars. The incidence of ‘any fall’ and ‘recurrent falls’ were calculated for both methods.

Results: Fall rates increased with age in men but not women. The ActiFE-Ulm prospectively assessed incidence (95% confidence interval) in women and men aged 65- < 90 years were 38.7 (36.9-40.5) and 29.7 (28.1-31.3) fallers/year and 13.7 (12.5-14.9) and 10.9 (9.9-12.0) recurrent fallers/year, respectively. Retrospective and prospective fall incidence in ActiFE-Ulm did not differ.The retrospectively assessed incidence of ‘any fall’ among persons 65- < 80 years were significantly lower in DEGS1 than ActiFE-Ulm (women: 25.7% (22.4-29.2) versus 37.4% (34.8-39.9); men: 16.3% (13.6-19.3) versus 28.9% (26.6-31.1). Retrospective incidence estimates of recurrent falls were similar in both studies for women (10.4% (8.3-12.9) versus 10.2% (8.5-11.8)) and men (6.1% (4.3-8.5) versus 8.4% (7.1-9.8)).

Conclusion: Both studies were population-based, but retrospective self-reported fall incidence differed between studies. Study design influences retrospective reported fall incidence considerably. Costly collection of prospective data gives similar rates to the cheaper retrospective report method.