Persistence of Antipsychotic Treatment in Elderly Dementia Patients: A Retrospective, Population-Based Cohort Study

We conducted a population-based, retrospective cohort study among patients with dementia aged 66 years or older residing in Ontario, Canada, and who were initiated on treatment with an antipsychotic between 1 January 2009 and 31 December 2012. This study was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre, Toronto, ON, Canada.

We used linked administrative healthcare databases for cohort identification, co-morbidity assessment, and outcome ascertainment. We used the Ontario Drug Benefit (ODB) database, a computerized pharmacy records system that records prescription drugs dispensed to all Ontario residents over 65 years of age to obtain prescription data. The Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD) was used to obtain records of inpatient hospitalizations, including diagnostic and procedural information, while the CIHI National Ambulatory Care Reporting System (CIHI-NACRS) was used to obtain details of emergency department visits. Furthermore, the CIHI Ontario Mental Health Reporting System (CIHI-OMHRS) was used to obtain information about mental health diagnoses. Physician billing information and demographic information were obtained from the Ontario Health Insurance Plan (OHIP) database and the Ontario Registered Persons Database (RPDB), respectively. Data from the OHIP database was coupled with the ICES Physician Database (IPDB) to identify the specialty of antipsychotic prescribers. These datasets were linked using unique, encoded identifiers and analyzed at the Institute for Clinical Evaluative Sciences (ICES).

We identified users of antipsychotics based upon prescription records for any atypical or typical antipsychotic within the accrual period (January 2009 to December 2012). We defined the date of each patient’s first prescription as their cohort entry date and restricted the cohort to patients with dementia by means of a validated dementia definition, based on a hospitalization or physician visit with a dementia diagnosis in the previous 5 years (using OHIP or CIHI-DAD diagnostic codes) or a cognitive enhancer prescription dispensed in the year prior [16]. We excluded patients less than 66 years old at cohort entry and those who were prescribed any antipsychotic in the year prior to the cohort entry date in order to restrict the cohort to elderly individuals newly initiated on antipsychotic treatment. Finally, individuals were excluded if they were only dispensed one prescription, as it is highly likely that these individuals never initiated therapy. This is an approach used regularly in research leveraging administrative claims databases investigating medication persistence [17, 18].

Patients were followed forward from the date of their first prescription until they discontinued therapy, died, had 2 years of follow-up, or reached the end of the study period (31 December 2013). For each patient, a period of continuous drug use was defined as receipt of a subsequent prescription refill within 180 days of the previous prescription in order to account for varied prescription regimes and minor lapses in treatment. Patients who switched between different antipsychotics were still considered to be persistent. In a sensitivity analyses, we used an alternate, stricter definition for continuous use, whereby an individual was persistent if a prescription refill was identified within 1.5 times the days supplied of the previous prescription.

We summarized the baseline characteristics of all individuals included in the study, including demographic information and a variety of relevant clinical characteristics. We also examined the prevalence of schizophrenia and bipolar co-diagnoses among patients in our cohort by looking for relevant diagnostic codes from CIHI-OMHRS and OHIP databases [19]. We conducted a competing risk analysis to determine time to discontinuation of any antipsychotic, accounting for death as a competing risk as a potentially high mortality rate was hypothesized for our cohort. We reported all analyses overall and stratified by the antipsychotic dose initiated to observe the effect of dosage on persistence. Doses were divided into three categories defined using tertiles of the dose initiated—low dose (<26 mg/day), medium dose (26–74 mg/day), and high dose (>74 mg/day)—normalized into milligrams of chlorpromazine equivalent dose [20]. Standardized differences were calculated to compare the characteristics of the low- and medium-dose groups against the high dose group, with significance achieved if the standardized difference was greater than 0.1 [21]. Gray’s test was used to check for the significance of any differences in the time to discontinuation between dose groups in the competing risk analysis [22]. All analyses were conducted using SAS Enterprise Guide (version 6.1, Cary, NC, USA).