Objective. To investigate the association between psycho-
tropic medication use and fractures in elderly women.
Design. Prospective cohort study.
Setting and participants. Acohort of 8127 white women aged
65 years and older was recruited from population-based list-
ings from Baltimore, Minneapolis, Portland, OR, and Monon-
gahela Valley, PA. Patients who were unable to walk at base-
line evaluation or had a history of bilateral hip replacement
Methods. Patients were interviewed between 1992 to 1994
and 1995 to 1996 to collect information on their use of central
nervous system (CNS) active medications, including benzo-
diazepines, narcotics, antidepressants, and anticonvulsants.
Reported use was verified by examination of pill bottles in
the 86% of the cohort who were seen in the study clinic or at
a home visit. Observation period began after first medication
assessment and continued until 1999.
Main outcome measures. Incident nonspine and hip frac-
tures as reported by participants every 4 months and con-
firmed by radiology reports.
Main results. After a mean 4.8 years of follow-up,1256 (15%)
of women experienced at least 1 nonspine fracture, including
288 (4%) with first hip fractures. When compared to medica-
tion nonusers, narcotics users were more likely to develop
nonspine fractures with a multivariate hazard ratio (HR) of
1.40 (95% confidence interval [CI], 1.06–1.83). Antidepres-
sant users also were more likely to develop nonspine frac-
tures (multivariate HR, 1.25 [95% CI, 0.99–1.58]) and hip frac-
ture (multivariate HR, 1.65 [95% CI, 1.05–2.57]). Women
taking tricyclic antidepressants or using selective serotonin
reuptake inhibitors had similar fracture rates. No indepen-
dent relationship was detected between benzodiazepine and
anticonvulsant use and fractures.
Conclusion. Narcotic and antidepressant use are associated
with nonspine fractures, and antidepressant use also is asso-
ciated with hip fractures in community-dwelling elderly
women. Neither benzodiazepines nor anticonvulsants are
associated with fractures.
Psychotropic medications, particularly benzodiazepines,
have been associated with both falls and fractures in multiple
epidemiologic investigations . The reported pooled odds
ratio for use of any psychotropic drugs and falling is 1.7 ,
and this finding is reasonably consistent across psychotropic
medication classes. However, few prospective studies have
examined these associations, and most retrospective studies
have not been able to adjust for differences in dosage or dura-
tion of therapy. Additionally, the narrow spectrum of data
available in administrative databases used by previous stud-
ies does not allow adjustment for all confounders, partic-
ularly confounding by indication (ie, the conditions being
treated by the medication also are responsible for the poor
outcomes) or confounding by comorbid conditions. A gold
standard randomized trial would likely be unethical or sig-
nificantly limited by selection criteria and would not settle all
issues of confounding and generalizability. Thus, a prospec-
tive cohort study with long follow-up, careful and thorough
measurement of exposure and outcome, and little attrition is
the most definitive study design to address adverse events
related to psychotropic medications.
This latest report satisfies many of the above criteria: the
data collection is prospective, participant reports of psy-
chotropic use are confirmed by examination of pill bottles,
and fractures are confirmed by radiology reports. Loss to
follow-up is 7%, compared with an overall event rate of 15%,
so it is unlikely that complete ascertainment of the data
would change the conclusion. The finding that narcotics and
antidepressant use are associated with nonspine fractures is
consistent with previous research. However, the univariate
relationship between benzodiazepines and hip fractures so
often reported in other studies is no longer significant after
308 JCOM June 2003 Vol. 10, No. 6
OUTCOMES RESEARCH IN REVIEW
Central Nervous System Active Medications and the Risk of
Fracture in Older Women
Ensrud KE, Blackwell T, Mangione CM, et al. Central nervous system active medications and risk for fractures in older
women. Arch Intern Med 2003;163:949–57.
(continued on page 313)
Vol. 10, No. 6 June 2003 JCOM 313
OUTCOMES RESEARCH IN REVIEW
adjustment for femoral neck bone density. This finding im-
plies that women with osteoporosis are more often being
prescribed benzodiazepines, and it is the osteoporosis rather
than the benzodiazepine that is increasing the fracture risk.
Is it possible that benzodiazepines are contributing to lower
bone density by inhibiting weight-bearing activities that are
known to ameliorate osteoporosis? The authors do not ad-
dress this possibility. In fact, untangling the risk associated
with medication use from the risk associated with comorbid
conditions is still very difficult, even in a high-quality study
such as this one. Ultimately, it is the responsibility of the
treating physician to ensure the benefit of these medications
outweighs the risk of a low-frequency but potentially devas-
tating adverse drug event.
Applications for Clinical Practice
Opiates and antidepressants are associated with fractures in
elderly women and should be avoided in the absence of
clear indications for their use.
–Review by Josh F. Peterson, MD, MPH
1.Wang PS, Bohn RL, Glynn RJ, et al. Zolpidem use and hip
fractures in older people. J Am Geriatr Soc 2001;49:1685–90.
Leipzig RM, Cumming RG, Tinetti ME. Drugs and falls in
older people: a systematic review and meta-analysis: I. Psy-
chotropic drugs. J Am Geriatr Soc 1999;47:30–9.
(continued from page 308)
Copyright 2003 by Turner White Communications Inc., Wayne, PA. All rights reserved.