Forty studies were included.
No randomised controlled trials were included. Only one published study has addressed the RR of falls for patients taking newer antidepressants. Areas of difference between studies included settings; sample size; characteristics of population; years of data collection; study design; medication ascertainment; and temporal relationship between medication ascertainment, fall ascertainment, and the index fall. 30% of studies did not provide a definition of a fall.
For community-dwelling older people (7 prospective studies) the annual incidence of falls averaged 36% (range 29% to 53%) with psychotropics being prescribed to 34% of subjects (range 28% to 52%).
For one or more falls:
Any psychotropic use, drugs not specified (19 studies): OR = 1.73 (95%CI: 1.52, 1.97), heterogeneity P = 0.001; Neuroleptic use (22 studies) OR = 1.50 (95%CI: 1.25, 1.79), heterogeneity P = 0.01.
Sedative/hypnotic (22 studies) OR = 1.54 (95%CI: 1.40, 1.70), heterogeneity P = 0.01. Any antidepressant (27 studies) OR = 1.66 (95%CI: 1.41, 1.95), heterogeneity P = 0.55.
Tricyclic anti-depressants (12 studies) OR = 1.51 (95%CI: 1.14, 2.00), heterogeneity P = 0.62.
Any benzodiazepine (13 studies) OR = 1.48 (95%CI: 1.23, 1.77), heterogeneity P = 0.60.
Short-acting benzodiazepine (9 studies) OR = 1.44 (95%CI: 1.09, 1.90), heterogeneity P = 0.10.
Long-acting benzodiazepine (9 studies) OR = 1.32 (95%CI: 0.98, 1.77), heterogeneity P = 0.84.
Psychiatrically hospitalized patients with neuroleptic use OR = 0.41 (95%CI: 0.21, 0.82); recently discharged patients, community dwelling subjects and long-term care subjects with neuroleptic use OR = 1.66 (95%CI: 1.38, 2.00), heterogeneity P = 0.21. Few studies adjusted for confounders. Pooled crude OR for antidepressants: 1.67 (95%CI: 1.15, 2.40) with adjusted OR = 1.85 (95%CI: 1.20, 2.85); crude OR for psychotropics = 1.95 (95%CI: 1.55, 2.46) with adjusted OR = 1.69 (95%CI: 1.25, 2.27).
Apart from psychiatrically hospitalized patients with neuroleptic use, no effect on OR was noted after stratification for the following factors: subject residence in the community or an institution; percentage of fallers 35% or greater in community studies; mean age of subjects 75 years or over; ascertainment of medications and falls at the time of the fall; and study design. None of the weighted correlation coefficients between size of OR and either the mean age of subjects or the initial year of data collection were statistically significant. Trend for psychotropics towards reduction in size of OR over time 9 = - 0.40, P = 0.09).
Psychotropics RR = 1.35 vs OR = 1.66; Antidepressants RR = 1.27 vs OR = 1.62; Neuroleptics RR = 1.31 vs OR = 1.90; Sedatives/ hypnotics RR = 1.12 vs OR = 1.25; Benzodiazepines RR = 1.2 vs OR 1.4; Tricyclic antidepressants RR = 1.16 vs OR = 1.40. Increased falls occurred in patients taking more than one psychotropic drug.