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


This was a nationwide study based on merged data from the Norwegian Prescription Database,[32] the Norwegian Hip Fracture Registry,[33] and the Central Population Registry.[34] The study lasted from January 1, 2005, to December 31, 2010.

Data Sources

The Norwegian Prescription Database, starting from January 2004, contains detailed information on all prescription drugs purchased at all pharmacies in Norway.[32] The data extracted for this study comprise all prescriptions of antipsychotics (Anatomical Therapeutic Chemical (ATC) system code[10] N05A) dispensed from January 2004 (prescriptions dispensed during 2004 necessary to identify current drug users when the study period started) until December 2010 according to each item's generic name, ATC code, and defined daily dose (DDD).[10] The Norwegian Prescription Database lacks individual information on medication dispensed to people staying in the hospital (~12,000 at any time) and in nursing homes (~40,000 at any time).

The Norwegian Hip Fracture Registry, starting from January 2005, contains national data (injury, fracture, surgery) on people who undergo surgery for hip fracture at all 55 hospitals in Norway performing such surgery.[33] For the purpose of this study, the date of first (primary) hip fracture registered for the period January 2005 until December 2010 was extracted. Even though hip fractures occurring during a hospital or nursing home stay are included in the Norwegian Hip Fracture Registry, these groups could not be identified in the dataset.

The Central Population Registry contains demographic information on the entire population of Norway. The data extracted for this study comprise birth year, sex and date of death or emigration if applicable.

The variables selected from these three registries were linked using the unique 11-digit personal identity number assigned after 1960 to everyone living in Norway.

Study Population

The study population included everyone born before 1945 and living in Norway on January 1, 2005. All individuals in this cohort were followed until the day of any first hip fracture, emigration or death, or the end of the study period on December 31, 2010.

Medications Studied

The following medications were included in this study: ATC code N05A, antipsychotics:

N05AA, phenothiazines with aliphatic side-chain (chlorpromazine, levomepromazine)

N05AB, phenothiazines with piperazine structure (dixyrazine, fluphenazine, perphenazine, prochlorperazine, trifluoperazine)

N05AC, phenothiazines with piperidine structure (thioridazine, pipotiazine)

N05AD, butyrophenone derivatives (haloperidol, melperone)

N05AE, indole derivatives (sertindole, ziprasidone)

N05AF, thioxanthene derivatives (flupenthixol, chlorprothixene, zuclopenthixol)

N05AG, diphenylbutylpiperidine derivatives (pimozide, penfluridol)

N05AH, diazepines, oxazepines, thiazepines, oxepines (clozapine, olanzapine, quetiapine)

N05AL, benzamides (sulpiride, tiapride amisulpride)

N05AX, others (risperidone, aripiprazole)

Although lithium is classified as N05A in the ATC system (N05AN01), its main indication as a mood stabilizer differs from that of all other N05A drugs, and it was excluded.[35]

For the purpose of subgroup analysis, the antipsychotic drugs were classified according to generation (first, second)[36] and prolactin effects (high or intermediate risk of increasing prolactin levels (prolactin elevating), low risk of increasing prolactin levels (prolactin sparing).[9,27,36,37] (See Table 1 for details.)


The DDD is the assumed average maintenance dose per day for a drug used for its main indication in adults.[10] Prescribed daily dose (PDD) and actual drug consumption vary within a population. The Norwegian Prescription Database does not include information on whether or when purchasers consumed the dispensed drugs. Thus, assumptions had to be made about drug exposure. For any antipsychotics, calculations were performed for 0.25, 0.5, and 1.0 DDDs, respectively. The average DDD is probably closest to 0.25 in this study population.[38,39] Quite similar results were found when calculating standardized incidence ratios (SIRs) for 0.25 and 0.5 DDDs (Table 1); to avoid misclassifying antipsychotic drug nonusers as users, 0.5 was chosen as the best proxy for medication exposure for all subgroup analysis. It was assumed that people started using the drugs on the day they were purchased and that they continued using them on the consecutive days corresponding to the number of 0.5 DDDs prescribed. An individual could possibly switch between exposure and nonexposure one or more times during the 6-year period.

Overall and recently started use of antipsychotic drugs was investigated. Overall use was defined as any exposure to antipsychotics within the study period, including all exposure periods. Recently started use was defined as the first 14 days of exposure to the drug in question after a 360-day washout period.

Statistical Analysis

Incidence of primary hip fracture during the person-days exposed and unexposed to antipsychotics in the study period was compared by calculating the SIR.[40] Standardization was indirect and accounted for sex, birth cohort, and time period (divided into 2-month intervals). The magnitude of two different estimates (e.g., the SIRs for FGAs and SGAs or the SIRs for women and men) cannot be directly compared. A SIR greater than 1 indicates greater risk of hip fracture associated with antipsychotic drug exposure.

For SIR values based on fewer than 100 observed primary hip fractures in exposed people, exact 95% confidence intervals (CI) were calculated assuming a Poisson distribution of the observed number of hip fractures (O) in exposed people, estimating the mean using the expected number of hip fractures in the exposed people. When the observed numbers of hip fractures in exposed people exceeded 100, 95% CIs were approximated using the following formula: (SIR · exp (−1.96√O), SIR · exp (1.96√O)).

To calculate the attributable risk of exposure to antipsychotic drugs on hip fracture, the observed number of fractures minus the expected number of fractures during the number of person-days exposed to antipsychotic drugs was divided by the observed number of fractures in the study population.

Ethics and Approval

The Regional Committee for Medical and Health Research Ethics (138/07) and the Norwegian Data Inspectorate (08/00133) approved the study. The Norwegian Directorate of Health granted an exemption from the duty of confidentiality (08/1843).