Campbell 2005.

Study characteristics
Methods	Study design: RCT Number randomized: 391 (100 in environmental group, 97 in behavioral group, 98 in the environmental and behavioral group, 96 in the social visit group) Number analyzed: 391 Number of centers: 2 Date of first enrollment: January 2002 Length of follow‐up: 12 months Sample size estimation: the authors used the rate of falls in those aged ≥75, 35% reduction in falls achieved, a power of 0.80, and a two sided significance of 0.05. The authors allowed for the Poisson distribution of falls and a 20% dropout rate.
Participants	Country: New Zealand Age: mean 84 years (range 75 to 96 years) Sex: 32% male, 68% female Key inclusion criteria: having poor vision (visual acuity of 6/24 or worse in the better eye after the best possible correction), and living in the community. Key exclusion criteria: unable to walk around their own residence, receiving physiotherapy at time of enrollment, could not understand the study requirement Comparability of baseline characteristics: comparable
Interventions	Environmental intervention #1: occupational therapists visited the person at home and assessed home safety using a checklist. They identified hazards and initiated a discussion with the participant about any items that could lead to falls. The therapists and participant agreed on which recommendations to implement.The therapists listed these recommendations in a follow‐up letter to the participant. They facilitated provision of equipment and payment from usual sources depending on the price and type of the home modification. Referrals were made to the Royal New Zealand Foundation of the Blind. A second home visit was needed to sign off the equipment installed by some providers. Behavioral intervention #2:physiotherapists initiated a one‐year Otago exercise program (specific muscle strengthening and balance retraining exercises). It was modified for those with severe visual acuity loss, and with vitamin D supplementation. The physiotherapists prescribed the exercises during five home visits at weeks one, two, four, and eight and a booster visit after six months. The degree of difficulty of the exercise and the number of 1, 2, and 3kg ankle cuff weights were used for muscle strengthening and increased at each visit as appropriate. Audio tapes of the exercises in four different levels of difficulty were available for those who could not see the exercise instruction sheets. Participants were expected to exercise at least three times a week (about 30 minutes a session) and to walk. The physiotherapists delivered vitamin D tablets if needed. For the months with no scheduled home visit, the physiotherapists telephoned to encourage the person to maintain motivation and discuss any problems. Environmental and behavioral intervention #3: received both the environmental and behavioral intervention. Control intervention #4: research staff made two home visits lasting an hour each during the first six months of the trial to participants who were not randomized to either environmental or behavioral intervention groups.
Outcomes	Primary outcome: number of fallers at 12 months Secondary outcomes: (1) falls per person year, (2) number of injurious falls, (3) injurious falls per person year, (4) incidence rate ratio for falls, (5) incidence rate ratio for injurious falls.
Notes	Funding sources: Health Research Council of New Zealand Statistical analyses: appropriate Subgroup analyses: reported Registration: ISRCTN ID: 15342873
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A statistician developed the group allocation schedule using computer generated random numbers
Allocation concealment (selection bias)	Low risk	The schedule was held by an independent person at a separate site and was accessed by a research administrator for the study, who telephoned after each baseline assessment was completed
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants were informed of their allocated treatment group following randomization.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Falls were monitored for one year for each person using prepaid, addressed, tear‐off monthly postcard calendars. The independent assessors in each center telephoned participants to record the circumstances of the falls and any resulting injuries or use of resources. They remained blind to group allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	361 out of 391 (92%) participants completed one year of follow‐up. The mean and total follow‐up time were equal by four arms. Reasons for not being follow‐up were reported and similar across four arms.
Selective reporting (reporting bias)	Low risk	Outcomes specified in the study protocol were reported.
Other bias	Low risk	No other sources of bias identified.