Antipsychotic Drugs and Risk of Hip Fracture in People Aged 60 and Older in Norway

Marit S. Bakken, MD, PhD; Jan Schjøtt, MD, PhD; Anders Engeland, MSc, PhD; Lars B. Engesæter, MD, PhD; Sabine Ruths, MD, PhD


J Am Geriatr Soc. 2016;64(6):1203-1209. 

In This Article


Older people in Norway had a twice the risk of sustaining a hip fracture during antipsychotic drug exposure than during nonexposure. Previous case–control studies showed similar associations between antipsychotic drug use and risk of hip fracture based on data from 1987 to 2002 (primarily including FGAs)[23,41,42] and 2005 to 2008 (including FGAs and SGAs).[13] After adjusting for possible confounders such as psychiatric diagnoses[41] and concomitant drug use,[13] an association between antipsychotic agents and higher risk of hip fracture was still evident.

FGAs and SGAs

Although SGAs have fewer adverse side effects than FGAs in terms of sedation and parkinsonism, these drugs are associated with greater risks of cerebrovascular and cardiovascular events and mortality;[14] in the last decade, associations have also been found with hip fractures.[13,42,43] The current results suggest that SGAs are not necessarily safer than FGAs with regard to hip fracture. In a large self-controlled case series[13] that included 8,234 individuals with hip fracture, greater risk of hospitalization for hip fracture was identified with short- and long-term (>12 weeks) use of FGAs and SGAs. The risk associated with SGAs was highest during the first week after initiation and declined with prolonged use (albeit still significantly elevated), whereas the risk associated with FGAs persisted at the same level.[13] Firm conclusions about whether FGAs or SGAs affect the risk of hip fracture more cannot yet be drawn; randomized controlled trials are lacking, and previous observational studies generally included few SGA users.[23,41–43]

Potential Effects on Bone Tissue

Antipsychotic drug use is an established risk factor for falls. Furthermore, all antipsychotics have some prolactin-elevating potential, which may affect bone metabolism. Hyperprolactinemia is a commonly reported adverse side effect, and it has been proposed that prolactin-sparing antipsychotics should be preferred for people at high risk of sustaining a hip fracture.[44] In support of this suggestion, the incidence of hip fracture was three times as great in individuals treated with antipsychotics and experiencing high prolactin levels.[43] The current study showed twice the risk of hip fracture associated with using antipsychotics with high or intermediate risk of increasing prolactin levels but an even higher excess risk associated with using antipsychotics with a low risk of increasing prolactin levels. These results suggest that other qualities of the drugs could be important. It has been suggested that blockade of 5-HT2B and α1-adrenoceptors may affect osteoblast proliferation and differentiation when using the SGA risperidone,[31] and recent preclinical studies have shown direct serotonergic effects on bone homeostasis.[28–30] To the knowledge of the authors of the current study, no clinical studies have investigated the associations between serotonergic and adrenergic effects on bone tissue of antipsychotics and risk of hip fracture.

Complex effects on transmitters and respective receptors, including dopamine, serotonin, and adrenergic pathways; indirect effects through prolactin and sex hormones; and metabolic (fat and sugar metabolism), sedative, and cardiovascular side effects (e.g., postural hypotension and arrhythmias) of the various antipsychotics probably influence the risk of hip fracture. A clear dose–response relationship is lacking for many proposed mechanisms of action of antipsychotic drugs on bone metabolism, and even prolactin-elevating effects are not always dose dependent.[27] Thus, a complex interplay between direct and indirect effects of the drugs may affect the risk of falls and hip fracture. The design of the current study does not allow for conclusions to be reached on mechanisms (changes in bone, bone mineral density, falls, or other factors) or causality (the use of antipsychotic drugs or their indication (psychosis)).

Age and Sex

Generally, the excess risk of hip fracture during antipsychotic drug exposure was most prominent in the youngest birth cohorts. Prescriptions for nursing homes residents are not included in the Norwegian Prescription Database, leading to systematic misclassification of approximately 40,000 people at any time as drug nonusers. Thus, the estimated associations between antipsychotic drug use and hip fractures in the oldest birth cohorts are probably conservative. Exposure to antipsychotic drugs was associated with a higher excess risk of hip fracture in men than in women. Because clinical information was lacking, it is not known whether antipsychotic drug use affecting, for example, fall risk and bone metabolism differently in men than in women or if confounders are differently distributed in men and women caused this difference. Thus, these results should be interpreted with caution. Nevertheless, a previous study that adjusted for comorbidities identified similar trends.[42]

Recently Started Drug use

The risk of hip fracture was higher in drug users who had started recently. Other studies[13,23] have found that antipsychotic drug initiation is associated with greater risk of fracture, but the mechanisms involved are unknown. Orthostatic hypotension is the most frequent vascular side effect of antipsychotic drugs, affecting approximately 40% of users.[45] This could be a problem in older adults at high risk of falling. Orthostatic hypotension is a particular concern during the early stages of antipsychotic treatment, and the development of tachyphylaxis, a sudden decrease in drug response, reduces the risk. Orthostatic hypotension is associated with the blockade of peripheral α1-adrenoceptors.[46]

The current results suggest that men who have recently initiated an antipsychotic drug are at higher risk than women who are new users, but the numbers are small and should be interpreted with caution.

Methodological Considerations

The national health registries provided a unique opportunity to link complete data on antipsychotic drugs purchased by a nationwide unselected community-dwelling older population with all primary hip fractures registered in Norway. The 6-year follow-up period revealed large numbers of cases, and the design prevented selection and information bias. The conservative definition of exposed person-days, not allowing for nonadherence (treatment gaps), yielded conservative risk estimates.

The sparse amount of clinical information available is the most important limitation of the study. SIRs were adjusted for major confounders (age and sex), but confounding by indication, by participant clinical characteristics, and by other medication use cannot be excluded. The health registries in Norway include neither complete and validated diagnostic information nor data on falls, bone mineral density, body mass index, or lifestyle factors such as alcohol and smoking. Thus, it is unknown whether people purchasing prescriptions for antipsychotics were actually diagnosed with psychosis or not. Low-dosage antipsychotics are also prescribed for behavioral and psychiatric symptoms in people with dementia and for pain.[14] Confounding by indication may affect the results, because psychoses, other clinical conditions, and antipsychotic drug use itself may increase the risk of falls and fractures through altered activity level, psychomotor function, and bone mineral density. In a large case–control study in the United Kingdom,[41] the positive associations between antipsychotic drug use and risk of hip fractures remained when adjusting for mental disorders, lifestyle factors, and concomitant drug use—indicating that the antipsychotic drug use itself affects fracture risk.

It is unknown whether purchased antipsychotic drugs were actually consumed, but the large proportion redeeming more than one prescription supports assumed adherence. Widespread nonadherence (misclassification of nonexposed person-time as exposed person time) would possibly have led to overestimation of the association between antipsychotics and hip fracture. Several strategies (strict definition of exposure and time-varying exposure) were applied to minimize, but could not exclude, misclassification. Information on other medication use is available from the Norwegian Prescription Database, but with the time-varying exposure used in the analysis, matching exposure periods for other medications was not possible. The time-varying exposure is a major strength of the study, because fixed exposure would have led to extensive and unmeasurable misclassification, yielding unreliable results.

The Norwegian Hip Fracture Registry includes 90% of all hip fracture operations in Norway,[33] with somewhat lower completeness during the first 3 years. It was assumed that underreporting was not systematically biased because of individual factors related to exposure (antipsychotic drug use) or outcome (hip fracture), which could have affected the results of the study.

Organizational factors in the healthcare system (mainly public in Norway), substance availability, treatment traditions, and prescribing patterns must be considered in transferring these results to other countries.

In conclusion, the results of this study support the hypothesis that antipsychotics are a risk factor for hip fracture. No evidence was found that SGAs and prolactin-sparing antipsychotics were safer than FGAs and prolactin-elevating antipsychotics with regard to hip fracture. This large registry-based study with limited clinical information does not allow for conclusions on mechanisms (effect on bone quality or falling) or causality of the observed association between antipsychotic drug use and hip fracture, and confounding by indication, comorbidity, other drugs used, lifestyle factors, or a combination of these cannot be excluded. Thus, clinical studies are needed to further explore these questions, although the observed association between antipsychotic drug use and hip fracture is relevant for clinical practice because both factors are prevalent in vulnerable older individuals.