Objectives To describe factors associated with initiating antipsychotics and patterns of persistence to antipsychotic therapy in a large cohort of adults with intellectual and developmental disabilities.
Design Population-based cohort study.
Setting Ontario, Canada.
Participants Adults with intellectual and developmental disabilities (IDD) in Ontario.
Outcome measures We used multivariable logistic regression to investigate patient characteristics associated with antipsychotic initiation. Patient characteristics studied included sociodemographic characteristics, measures of clinical comorbidity and health service use.
Results Among 39 244 individuals eligible for this study, 6924 (17.6%) initiated an antipsychotic over the accrual window, of whom 1863 (26.9%) had no psychiatric diagnosis in the prior 2 years. A number of factors were significantly associated with antipsychotic initiation, including male gender, residence in a group home, prior use of benzodiazepines, antidepressants or cognitive enhancers, a recent emergency department visit or mental health hospitalisation and a visit to a psychiatrist or family physician in the prior 90 days. In a secondary analysis, the association between antipsychotic initiation and age, prior diagnosis of diabetes or myocardial infarction and polypharmacy differed slightly on the basis of whether an individual had a previously diagnosed psychiatric disorder.
Conclusions Factors associated with the initiation of an antipsychotic differ according to the presence of a psychiatric diagnosis. Given the long duration of antipsychotic use in this population, future research is needed to understand the appropriateness of antipsychotic initiation among adults with IDD and the safety implications of long-term use of these products.
- antipsychotic agents
- developmental disabilities
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
This study uses linked population-based data allowing us to study medication use and health services utilisation among a large cohort of adults with intellectual and developmental disabilities (IDD), regardless of how they interact with the health system.
We relied on diagnostic codes from administrative claims databases in the prior 2 years to define the presence of psychiatric diagnoses. Although this definition is used regularly in Ontario, it has not been validated in our population of adults with IDD, and therefore it is possible that some misclassification occurred.
We cannot confirm whether antipsychotics that were dispensed to individuals without a psychiatric diagnosis were being prescribed to manage behavioural concerns, and therefore, cannot draw definitive conclusions about the use of this class of medications for this indication in our study.
We determined residence in a group home setting using data from 2009 and assumed that this did not change over the study period. Although residence in these settings is relatively stable in Ontario, it is possible that living arrangements for some members of our cohort could have changed over the study period.
We are unable to study some variables that may be associated with antipsychotic initiation, including severity of intellectual disability, history of challenging behaviour and family and paid caregiver level of stress and attitudes towards medication use.
There is a growing recognition that use of antipsychotics is high among adults with intellectual and developmental disabilities (IDD) internationally, with rates ranging between 21% and 45%.1–7 In Canada, a recent study found that 39% of adults with IDD are treated with these medications.8 For some, antipsychotics are prescribed to treat psychiatric disorders, but in many cases, they are used as a means to manage behaviour in the absence of a comorbid mental health disorder,3 despite limited evidence of their efficacy for this purpose.3 9 Long-term use of antipsychotics has been associated with adverse effects including sedation, extrapyramidal side effects, anticholinergic effects, cardiometabolic problems including diabetes, hypertension, thromboembolism, stroke and, rarely, sudden death.3 10 11 Given these risks, the high rate of antipsychotic use among adults with IDD is concerning, and even more so because this population has high rates of comorbid health issues, and may have difficulty reporting side effects, thus delaying medical intervention.12 13
Despite many studies documenting high rates of antipsychotic use in this population, few have focused specifically on patient factors associated with initiation of these medications. Given the frequent use of antipsychotics in this population, concerns around their safety and off-label use for behavioural management, more evidence is needed on patient characteristics associated with the initiation and duration of use of these medications among adults with IDD. The objective of this study was to describe factors associated with initiating antipsychotics and patterns of persistence to antipsychotic therapy in a large cohort of adults with IDD.
Material and methods
We conducted a population-based cohort study that followed a cohort of 66 484 Ontario adults with IDD aged 18–64 years as of 1 April 2009 who received disability support benefits. All residents of Ontario have universal, government-funded access to physician and hospital services. Furthermore, this population has access to publicly funded medications through the disability support programme. IDD was defined to be consistent with Ontario legislation to include terms such as mental retardation/intellectual disability, autism and other pervasive developmental disorders, fetal alcohol syndrome and various chromosomal disorders known to be aetiological factors of IDD. This cohort was generated by linking data from provincial health and social services data in Ontario, Canada, and defined IDD on the basis of: (1) the presence of IDD diagnostic codes in two or more physician claims since database inception (1991), (2) the presence of IDD diagnostic codes on one or more hospital claims since database inception (1988) or (3) IDD recorded as a diagnosis leading to eligibility for disability benefits from social services.13 This project was completed as part of the Health Care Access Research and Developmental Disabilities Program and the Ontario Drug Policy Research Network.
We used the Registered Persons Database to determine patient demographics and vital statistics and identified all prescription drugs dispensed from community pharmacies using the Ontario Drug Benefit claims database. We used the Canadian Institute for Health Information’s Discharge Abstract Database, Same Day Surgery database, National Ambulatory Care Reporting System and Ontario Mental Health Reporting System to identify diagnoses and procedures from inpatient hospitalisations, same-day surgeries, emergency department visits and mental health hospitalisations, respectively. We used the Ontario Health Insurance Program claims database to determine outpatient health services utilisation, including primary care and specialist visits and related billings. Finally, we used three validated databases, the Ontario Diabetes Database,14 the Ontario Hypertension Database15 and the Ontario Myocardial Infarction Database16 to determine past diagnoses of diabetes, hypertension and myocardial infarction, respectively. These datasets were linked using unique, encoded identifiers, were analysed at ICES (www.ices.on.ca) and are used regularly to assess the drug utilisation, safety and policy in the Ontario healthcare system.8 17–19 More details on these data sources can be found in online supplementary appendix etable 1. ICES is a prescribed entity under section 45 of Ontario’s Personal Health Information Protection Act (PHIPA), which enables analysis and compilation of statistical information related to the management, evaluation and monitoring of, allocation of resources to, and planning for the health system. Section 45 authorises health information custodians to disclose personal health information to a prescribed entity, like ICES, without consent for such purposes. Therefore, this study did not require informed consent.
Supplementary Appendix 1
From within the cohort of adults with IDD, we identified our primary outcome as individuals who were newly prescribed an antipsychotic (online supplementary appendix etable 2) between 1 April 2010 and 31 March 2016. We defined antipsychotic initiation as any antipsychotic prescription filled during the study period with no antipsychotic prescribed in the prior year. Individuals not prescribed an antipsychotic over our study period or in the year prior were considered unexposed. The first antipsychotic prescription date was the index date for the exposed cohort. For the unexposed, we randomly assigned an index date following the same distribution as of dates observed among antipsychotic users.
We explored sociodemographic characteristics, measures of clinical comorbidity and health service use as potential factors associated with antipsychotic initiation in this population. Sociodemographic characteristics included age, sex, income quintile, rural location of residence and whether the individual resided in a group home. We defined prior psychiatric diagnoses using methods published previously.20 21 Specifically, we defined a psychiatric diagnosis as either a major mental disorder (schizophrenia, non-schizophrenia psychotic disorder, bipolar disorder or major depressive disorder) or other mental disorder (other depressive disorders, anxiety, somatoform, dissociative, psychosomatic, personality disorders, adjustment disorders and disorders of conduct or impulsivity) diagnosed in the 2 years up to and including the index date using inpatient and outpatient datasets (see online supplementary appendix etables 3 and 4).20 21 We used the Charlson Comorbidity Index as a measure of overall medical comorbidity using hospital claims in the prior 3 years22 and reported the presence of any prior diagnosis of diabetes, hypertension or myocardial infarction using validated databases described above. Measures of medication use in the prior year included total number of drugs dispensed, and any exposure to antidepressants, benzodiazepines, lithium or antiepileptic drugs, stimulants or cognitive enhancers. We included several measures of health services utilisation, including inpatient hospitalisations (medical and psychiatric) and emergency department visits in the prior 2 years, number of physician visits in the prior 1 year and recent visits to a specialist (psychiatrist or neurologist; prior 90 days).
Duration of antipsychotic use
Among new antipsychotic users, we characterised prescriber specialty for the initial antipsychotic prescription (psychiatrist, family physician, other specialist and unknown) and the initial antipsychotic dose (in chlorpromazine equivalents). In cases where individuals were initiated on multiple antipsychotics, the total dose was calculated as the sum of the doses on all prescriptions dispensed at initiation. We also characterised a period of ongoing use of antipsychotics among new users of these drugs to investigate patterns of drug discontinuation over a 1-year follow-up period. We defined ongoing antipsychotic use on the basis of a prescription refill within 180 days of a previous prescription. If no such refill was found, the individual was flagged as having discontinued antipsychotic therapy, and their discontinuation date was set as the date on which the last prescription filled would have been completed (ie, date of dispensing plus days supply). In this analysis, we followed people forward until the first of drug discontinuation or a censoring event (death, end of observation period (31 March 2016) or 1 year of maximum follow-up).
In the primary analysis, we compared baseline characteristics of exposed and unexposed individuals using standardised differences, which are used in large population-based studies where the size of the population can influence p values. A standardised difference above 0.10 is considered meaningful.23 We used multivariable logistic regression to investigate patient characteristics associated with antipsychotic initiation. We included all baseline covariates described above in the model. All continuous variables were first modelled as such. If a non-linear relationship between these variables and the log-odds of the outcome was suspected, then we categorised these variables and modelled the categorical variables instead. Regression models were created for the overall cohort and stratified into individuals with and without a psychiatric diagnosis at baseline. Missing values were categorised separately in the models.
For antipsychotic users only, we generated Kaplan-Meier estimates for time to drug discontinuation and reported the median time to discontinuation and the proportion of the population who remained on therapy for at least 6 and 12 months. All analyses were conducted overall and stratified by psychiatric diagnosis at baseline. We used the log-rank test to measure differences in time to discontinuation and χ2 tests to compare initial antipsychotic prescription characteristics between those with and without a psychiatric diagnosis at baseline. We defined statistically significant results as those with a type 1 error rate less than 5%.
Patient and public involvement
Patients and the public were not involved in the design or conduct of this study.
Among the 66 484 adults with IDD in Ontario, 20 316 were prescribed an antipsychotic over the accrual period, among whom 6924 met eligibility criteria and were new antipsychotic users. The remaining 46 168 did not have a record of an antipsychotic prescription through the public drug programme over the accrual period, among whom 32 320 met our eligibility criteria (see online supplementary appendix efigure 1 for full details of exclusions). In general, adults with IDD initiating antipsychotics were similar to the unexposed group with respect to mean age (41.0 vs 40.7 years), sex (56.1% vs 56.7% male), neighbourhood income (32.7% in lowest income quintile in both groups), rurality (84.2% vs 82.0% residing in urban areas) and residence in a group home setting (11.1% vs 9.1%; table 1). There were several important differences between these groups in prior medication use, prior health services utilisation and clinical comorbidities. For example, adults initiating antipsychotics were more likely to have recently been prescribed antidepressants (43.2% vs 15.1%) and benzodiazepines (28.8% vs 11.4%), were more likely to have visited an emergency department in the past 2 years (68.8% vs 42.6%) and had a higher prevalence of major mental illness (31.3% vs 2.6%) compared with the unexposed group (table 1).
In the multivariable model, all continuous variables were modelled as categories, with the exception of the number of visits to a physician in the prior year (c-statistic for model, 0.862). Important factors associated with antipsychotic initiation were being male (adjusted OR (aOR) 1.21, 95% CI 1.13 to 1.29), residence in a group home (aOR 1.20, 95% CI 1.08 to 1.34), past use of antidepressants (aOR 1.96, 95% CI 1.82 to 2.12), benzodiazepines (aOR 1.60, 95% CI 1.46 to 1.74) and cognitive enhancers (aOR 5.10, 95% CI 3.78 to 6.87), an ED visit (aOR 1.43, 95% CI 1.33 to 1.54) or mental health hospitalisation (aOR 1.84, 95% CI 1.61 to 2.09) in the past 2 years and a visit to a psychiatrist (aOR 4.82, 95% CI 4.31 to 5.40) or family physician (aOR 1.67, 95% CI 1.55 to 1.79) in the previous 90 days (table 2). As expected, both prior diagnosis of a major mental disorder (aOR 5.18, 95% CI 4.65 to 5.77) and other mental disorder (aOR 3.18, 95% CI 2.96 to 3.41) were strongly associated with antipsychotic use. Several factors were associated with a lower odds of antipsychotic initiation in the multivariable model, including being aged 30 to 45 (aOR 0.84, 95% CI 0.77 to 0.92 compared with those aged 18–29 years), past use of lithium or antiepileptic drugs (aOR 0.84, 95% CI 0.74 to 0.94) and each of diagnoses of diabetes mellitus (aOR 0.85, 95% CI 0.76 to 0.94), hypertension (aOR 0.76, 95% CI 0.69 to 0.83) or a history of myocardial infarction (aOR 0.54, 95% CI 0.37 to 0.79).
Among the 6924 individuals who initiated an antipsychotic, 1863 (26.9%) had no evidence of a psychiatric diagnosis in the 2 years preceding the index prescription. In analyses stratified by presence or absence of a psychiatric diagnosis in the 2 years preceding the index date, the multivariable models were similar (c-statistics, 0.753 and 0.762, respectively). Among those with a psychiatric diagnosis, older individuals (aged 46 and older) were less likely than those aged 18–29 to start an antipsychotic (aOR 0.87, 95% CI 0.77 to 0.97). The opposite pattern was observed for those without a psychiatric diagnosis, whereby older adults were more likely to start antipsychotics (aOR 1.19, 95% CI 1.03 to 1.38). Furthermore, those without a psychiatric diagnosis were less likely to start antipsychotics if they had a history of diabetes or myocardial infarction; however, this was not the case among those with a psychiatric diagnosis. Finally, there were important differences in the relationship with polypharmacy measures between groups. Among individuals with no psychiatric diagnosis, those taking two or more medications in the previous year were more likely to initiate antipsychotics compared with those with no medication use in the prior year (range of aOR 1.23–1.24). In contrast, among those with a psychiatric diagnosis, higher degrees of polypharmacy were protective against initiating an antipsychotic (range of aOR 0.62–0.78; table 3).
We also observed differences in initial antipsychotic dose and prescriber specialty between those with and without a psychiatric diagnosis (table 3). In general, those without a psychiatric diagnosis were more likely to receive their first antipsychotic prescription from a family physician (65.3% vs 47.4%; p<0.001) and less likely to receive their first prescription from a psychiatrist (9.6% vs 36.3%; p<0.001). Initial antipsychotic dose was similar between patient groups (median 80 mg chlorpromazine equivalents; IQR 40–200 mg), but those with a psychiatric diagnosis escalated to a higher maximum antipsychotic dose compared with those with no such diagnosis (150 mg vs 110 mg chlorpromazine equivalents; p<0.001).
Overall, the median time to antipsychotic discontinuation among adults with IDD was 354 days; however, this also differed significantly according to prior psychiatric diagnosis (p<0.001). Specifically, those with a prior psychiatric diagnosis were more likely to continue therapy for 6 months (60.8% vs 50.7%, p<0.001) and 12 months (50.7% vs 41.8%, p<0.001) compared with those with no such diagnosis (table 3).
In this population-based study of nearly 40 000 previously untreated adults with IDD residing in Ontario, Canada, we found that approximately 1 in 6 initiated an antipsychotic medication over our 6-year study period. Furthermore, over one-quarter of those starting an antipsychotic had no indication of a psychiatric condition being diagnosed in the prior 2 years and had a lower maximum dose achieved. This suggests that antipsychotics may be used frequently in adults with IDD for off-label indications, such as managing behavioural issues, sedation or sleep. In general, independent factors associated with antipsychotic initiation included being male, residing in a group home, prior use of psychiatric medications, high health services utilisation and previous psychiatric diagnoses, while having diabetes, hypertension or a previous myocardial infarction reduced the chance that an antipsychotic would be initiated. Although there was little difference in variables associated with antipsychotic initiation when considering those with and without psychiatric diagnoses separately, one important finding is that people without a psychiatric diagnosis who had received multiple other medications in the year prior were more likely to be started on an antipsychotic, whereas such polypharmacy was protective against antipsychotic initiation among people with psychiatric diagnoses. Although the specific drivers of this finding are unknown, future research should investigate the differential role of polypharmacy in antipsychotic initiation among those with and without a concurrent psychiatric diagnosis.
The patterns observed in this study are generally consistent with previously published research exploring antipsychotic use in the IDD population. For example, a study by Sheehan et al7 similarly found that major mental illness was associated with increased likelihood of initiating antipsychotics among adults with IDD, although they also identified a subgroup of the population who initiated antipsychotics in the absence of a psychiatric diagnosis. Furthermore, several studies have highlighted psychotropic use generally24 and antipsychotics specifically4 5 25 being more commonly prescribed to those in residential care settings compared with those living with family or independently. This may speak to increased patient complexity (eg, higher rates of psychiatric disorders or challenging behaviour) leading to residential care but may also be associated with greater stressors in those environments or prescribing attitudes of formal versus family caregivers.6 Prior studies have suggested that older adults are more likely to be prescribed antipsychotics than younger adults,1 2 7 26 but this was influenced by the way in which age is grouped in each study. Our study suggests that this relationship is impacted by concurrent psychiatric diagnosis. In particular, adults over age 45 years were more likely to be prescribed antipsychotics in the absence of a psychiatric diagnosis but less likely to be prescribed these medications when there was a psychiatric diagnosis. This may be influenced by the timing at which psychiatric illnesses are commonly diagnosed, with this diagnosis typically occurring in younger adults who will therefore be initiated on antipsychotics at a younger age. It could also be suggestive of antipsychotics being used off-label for sleep disturbances in older populations as has been reported in studies of older adults with IDD.27
Our findings have important implications for clinical practice in this vulnerable population. First, we found that the likelihood of antipsychotic initiation is lower among individuals who have a prior diagnosis for a metabolic or cardiovascular condition (eg, diabetes, hypertension and myocardial infarction), which suggests that clinicians are aware of the literature regarding the potential risk factors for adverse events when prescribing this class of medications in this population28 29 and demonstrate more caution when diabetes or hypertension are already present. However, our findings also suggest that individuals taking a large number of other medications and those with a high degree of comorbidity are more likely to initiate antipsychotics and that nearly half of this population will continue their medication for at least 1 year. This is concerning given the risks associated with long-term use of antipsychotics, particularly among individuals who may have challenges communicating with providers when experiencing adverse events. Finally, among individuals with no psychiatric diagnosis, nearly two-thirds of antipsychotics were initiated by a family physician, which is much higher than rates observed among those with psychiatric diagnoses in our study (47%). Although this could be due to different patterns of access to specialists between these populations, this finding is consistent with those discussed in research by Holden et al, which also found that family physicians were less likely to provide psychosocial interventions and monitor medication side effects in this population compared with psychiatrists.1 Therefore, the findings of our study suggest that future research is needed to better understand indications for antipsychotic initiation and rigour of long-term medication monitoring by family physicians in this vulnerable population.
Strengths and limitations
A main strength of this study is its use of linked population-based data allowing us to study medication use and health services utilisation among a large cohort of adults with IDD, regardless of how they interact with the health system. However, there are several limitations that merit emphasis. First, we relied on diagnostic codes from administrative claims databases in the prior 2 years to define the presence of psychiatric diagnoses. Although this definition is used regularly in Ontario, it has not been validated in our population of adults with IDD, and therefore, it is possible that some misclassification occurred. However, given the differences observed in our stratified models, the findings from this study suggest that our definition of psychiatric diagnoses successfully identified two unique subpopulations in our cohort. Second, we cannot confirm whether antipsychotics that were dispensed to individuals without a psychiatric diagnosis were being prescribed to manage behavioural concerns, and therefore, we cannot draw definitive conclusions about the use of this class of medications for this indication in our study. However, the fact that individuals without a psychiatric diagnosis were also more likely to have a lower maximum dose achieved suggests that this may be the case. Furthermore, it is possible that antipsychotics are also being used off-label to manage challenging behaviour among people with psychiatric diagnoses; however, we are unable to determine the indication for use in this population. It is also important to note that guidelines recommend the use of antipsychotics to manage challenging behaviour when evidence-based alternatives have been unsuccessful, and when the risk to the person or others is severe.30 Using data available to us, we are unable to determine the alignment of antipsychotic use with these recommendations in this analysis. However, the high antipsychotic initiation rate, often with several other medications, and variability in use across comorbid profiles and practice settings suggest that considerable risk is being introduced to adults with IDD due to their use. Future research should investigate further the appropriateness of this prescribing and its association with harm. Third, our cohort was restricted to adults aged 18–64 years, and therefore, these findings may not be generalisable to youth or older adults. Fourth, we defined new antipsychotic use on the basis of no antipsychotic prescription in the prior year. Therefore, it is possible that some individuals in our cohort had more remote prior use of antipsychotics. Fifth, we are unable to study some variables that may be associated with antipsychotic initiation, including severity of intellectual disability, history of challenging behaviour and family and paid caregiver level of stress and attitudes towards medication use. Furthermore, we are unable to determine the degree of intellectual disability or to specifically investigate the subgroup of individuals within the cohort with an autism diagnosis. Future research should consider further exploration into antipsychotic initiation within subsets of the IDD population. Finally, we determined residence in a group home setting using data from 2009 and assumed that this did not change over the study period. Although residence in these settings is relatively stable in Ontario, it is possible that living arrangements for some members of our cohort could have changed over the study period. Several of these limitations could be addressed through the ongoing collection and linkage of provincial health and social services data which would provide researchers with accurate estimates of IDD prevalence, living arrangements, and eligibility for public drug programmes.
The findings of this study suggest that, not only is antipsychotic prescribing high among adults with IDD, but factors associated with the initiation of an antipsychotic differ according to the presence of a psychiatric diagnosis and certain comorbidities. Over one-quarter of antipsychotic initiation occurred in the absence of any psychiatric diagnosis, suggesting a noteworthy degree of off-label antipsychotic use to manage challenging behaviour in a population with a high comorbidity and medication burden. However, in contrast, physicians appear to be more cautious in their antipsychotic prescribing to patients with pre-existing metabolic conditions, suggesting an appreciation of the potential metabolic complications associated with their use. Given the long duration of antipsychotic use in these individuals, future research is needed to understand the appropriateness of antipsychotic initiation among adults with IDD, and the safety implications of long-term use of these products. Until such evidence is developed, opportunities must be found to improve clinician knowledge regarding guideline recommendations for use and monitoring of antipsychotics in this vulnerable population to ensure that these medications are not being used to broadly manage behavioural issues without adequate knowledge of their harmful effects.
Contributors TG was responsible for conception and design of the study, data acquisition and interpretation and drafting of the manuscript. WK was responsible for design of the study, data acquisition, analysis and interpretation and critical revisions of the manuscript. MT, SV and VC were responsible for design of the study, data interpretation and critical revisions of the manuscript. YL was responsible for conception and design of the study, data acquisition and interpretation and drafting of the manuscript. All authors provided final approval of the manuscript.
Funding This study was funded by grants from the Province of Ontario. It was also supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). We would like to thank IMS Brogan Inc for use of their Drug Information Database. Parts of this material are based on data and information compiled and provided by the Canadian Institute for Health Information (CIHI).
Disclaimer The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding source and those of CIHI. No endorsement by ICES, the Ontario government, or CIHI is intended or should be inferred.
Competing interests None declared.
Ethics approval This study was approved by the research ethics board of Sunnybrook Health Sciences Centre, Toronto and conducted at ICES in Toronto, Canada using deidentified data.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The data set from this study is held securely in coded form at ICES. While data sharing agreements prohibit ICES from making the data set publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS. The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the programmes may rely on coding templates or macros that are unique to ICES.
Patient consent for publication Not required.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.