Abstract
Background: In Canada, many patients face substantial out-of-pocket costs for prescription medication, which may affect their ability to take their medications as prescribed. We sought to conduct a comprehensive analysis of the burden and predictors of cost-related nonadherence in Canada.
Methods: Using pooled data from the 2015, 2016, 2018, 2019, and 2020 iterations of the Canadian Community Health Survey, we calculated weighted population estimates of the burden of cost-related nonadherence in the preceding 12 months and used logistic regression models to measure the association of 15 demographic, health, and health system predictors of cost-related nonadherence overall and stratified by sex.
Results: We included 223 085 respondents. We found that 4.9% of respondents aged 12 years or older reported cost-related nonadherence. Those who self-identified as female, belonging to a racial or ethnic minority group, or bisexual, pansexual, or questioning were more likely to report cost-related nonadherence. Younger age, higher disease burden, poorer health, non-employer prescription drug coverage, and not living in the province of Quebec were associated with cost-related nonadherence.
Interpretation: Our nationally representative findings reveal inequities that disproportionally affect marginalized people at the intersections of sex, race, age, and disability, and vary by province. This foundational understanding of the state of cost-related nonadherence may be used to inform potential expansion of public drug coverage eligibility, premiums, and cost-sharing policies that address financial barriers to medication adherence.
Prescription medications are necessary for disease prevention and management, particularly for people with chronic conditions; however, they are not always taken as prescribed. The costs of long-term medication use may impose considerable economic hardship to patients and lead to suboptimal care choices.1 Cost-related nonadherence refers to skipping doses, reducing dosages, delaying refilling prescriptions, or not filling a prescription because of out-of-pocket costs.2,3
In Canada, most public health insurance is administered through 13 independent provincial and territorial systems that cover the costs of all medically necessary hospital and physician visits for citizens and permanent residents, in accordance with the Canada Health Act.4,5 Although prescription medications are the second most expensive component of health care spending, costing $43 billion and accounting for 13% of annual health care expenditure in 2021,6 they are not universally covered under Canada’s public health insurance systems.4,5 In 2021, Canadian households paid an estimated $7.4 billion out of pocket for prescription medications, accounting for 17.3% of total prescription drug expenditures, with out-of-pocket spending forecasted to continue to increase at a faster rate than other health expenditures.7
Studies have identified several individual (e.g., female sex, belonging to a racial or ethnic minority group, younger age, low household income, province of residence), health (e.g., poor health status, high disease burden), and health care system (e.g., high out-of-pocket expenses, lack of drug coverage) factors associated with increased cost-related nonadherence in Canada.8,9 However, generalizing these findings remains challenging as studies have largely used single-year data, been restricted to specific populations (e.g., older adults, people with chronic conditions), and lacked a sample size sufficient to evaluate several complex predictors.8,9 To address limitations of previous research, we aimed to conduct a comprehensive analysis of the burden and predictors of cost-related nonadherence in Canada.
Methods
Study design and data sources
We conducted an analysis of pooled data from several cycles of the Canadian Community Health Survey (CCHS), a national telephone survey administered annually by Statistics Canada that collects self-reported data on health, health care utilization, and determinants of health. The CCHS is representative of 97% of the community-dwelling household population aged 12 years or older living in Canada.10 We pooled data from the 2015, 2016, 2018, 2019, and 2020 cycles (average response rate of 59%) of the CCHS as they included questions on cost-related nonadherence and used comparable sampling designs and population representation targets. We accessed confidential microdata master files through Statistics Canada’s secure research facility, the Research Data Centre. We restricted our sample to respondents who were asked about cost-related nonadherence in the preceding 12 months.
Primary outcome
We coded the occurrence of cost-related nonadherence as a binary variable based on participants’ responses to the question, “During the last 12 months, was there a time when you did not fill or collect a prescription for your medicine, or you skipped doses of your medicine because of the cost?,” with affirmative responses coded as the presence of cost-related nonadherence. We coded the remaining responses (no, not applicable [no medication prescription in the last 12 months], don’t know, refusal, not stated) as the absence of cost-related nonadherence.
Predictors
We chose predictors based on our conceptual knowledge and empirical evidence from previous studies on cost-related nonadherence in Canada.8,9,11,12 Demographic variables included sex, race or ethnicity, sexual orientation, age, education, marital status, home ownership, annual income, and province of residence. Health and health care system variables included number of chronic health conditions, having a regular provider, medication insurance coverage, perceived health status, and life satisfaction.
Statistical analysis
We used Statistics Canada’s survey weights to calculate weighted population estimates and bootstrapping weights to calculate confidence intervals (CIs).13 We used logistic regression to estimate the adjusted odds ratios (ORs) for the association between cost-related nonadherence and its predictors for the study population. We tested for effect modification between variables, specifically sex and race or ethnicity, and reported interaction terms with p values less than 0.05. We also reported adjusted ORs stratified by sex to provide findings consistent with national reporting on drug expenditures from the Canadian Institute for Health Information.14,15 We employed a multistep process to impute missing values for all independent variables, wherein we created 5 imputed data sets and then synthesized the resultant 5 regression models to fully incorporate the necessary variance adjustments from multiple imputations.16,17 We describe the composition of the logistic regression models — including CCHS questions, variables, and derived categories for analysis — in Appendix 1, Table S1, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.241024/tab-related-content. We conducted all analyses using SAS Studio 9.4 (SAS Institute).
Ethics approval
This study was approved by the University of British Columbia Behavioural Research Ethics Board and Statistics Canada’s Data Access Division (H22-00125).
Results
Our study population consisted of 233 085 respondents (Figure 1), which represented 133 378 497 (95% CI 133 271 702–133 485 293) weighted person-years. Of the total sample, 4.9% (95% CI 4.7%–5.0%) reported cost-related nonadherence. The prevalence of cost-related nonadherence in each cycle of the CCHS and among respondents who received a prescription in the preceding 12 months is reported in Appendix 1, Table S2. The raw and weighted frequencies of respondents’ descriptive characteristics are shown in Table 1. The distributions of descriptive characteristics in respondents with and without cost-related nonadherence are presented in Appendix 1, Table S3.
Selection of study population. See Related Content tab for accessible version. Note: CCHS = Canadian Community Health Survey, CRNA = cost-related nonadherence.
Characteristics of study population derived from the 2015, 2016, 2018, 2019, and 2020 annual components of the Canadian Community Health Survey
Predictors of cost-related nonadherence
Predictors of cost-related nonadherence in the whole and stratified study population are presented in Table 2.
Multivariable modelling of predictors of cost-related nonadherence
In analyses of the whole study population, demographic predictors associated with cost-related nonadherence were sex, race or ethnicity, sexual orientation, age, education, marital status, home ownership, annual income, and province of residence. Females showed 1.44 times the odds of cost-related nonadherence than males. Indigenous, Latin American, multiracial, West Asian, Arab, and Black respondents had 1.20–1.67 times the odds of cost-related nonadherence than White respondents, whereas East Asian and Southeast Asian respondents showed 0.71 to 0.84 times the odds of cost-related nonadherence. The interaction terms between sex and identifying as Indigenous (p = 0.009) or South Asian (p = 0.02) were both statistically significant. Bisexual, pansexual, or questioning respondents had 1.43 times the odds of cost-related nonadherence than heterosexual respondents. Younger age was consistently associated with higher odds of cost-related nonadherence, most prominently among respondents aged 18–34 years, who showed 8.99 times the odds of cost-related nonadherence than those aged 75 years or older. Respondents with lower levels of education and those with lower incomes consistently showed higher odds of cost-related nonadherence, particularly those making less than $40 000 annually. Respondents who rented showed 1.24 times the odds of cost-related nonadherence than those who resided in a home owned by a member of their household. Finally, compared with Quebec residents, respondents residing in the other provinces showed 1.06–1.58 times the odds of cost-related nonadherence.
Health and health care system predictors associated with cost-related nonadherence were the number of chronic health conditions, having a regular provider, medication insurance coverage, perceived health status, and life satisfaction. Greater disease burden was consistently associated with greater odds of cost-related nonadherence, with respondents with 1 or more chronic health conditions showing 1.91–3.41 times the odds of cost-related nonadherence compared with those who did not have any comorbidities. Respondents with a regular health care provider had 1.30 times the odds of cost-related nonadherence. Compared with having employer insurance coverage for prescription medication, having no coverage, government insurance, or associate or private insurance showed 2.76, 1.43, and 1.35 times the odds of cost-related nonadherence, respectively. Respondents who reported poorer perceived health and lower life satisfaction consistently showed higher odds of cost-related nonadherence.
In stratified analyses, the magnitude of the association between predictors and cost-related nonadherence was similar for most predictors among both males and females (Table 2). Exceptions were race or ethnicity, sexual orientation, and immigration status.
Interpretation
Our study provides nationally representative estimates of the burden and predictors of cost-related medication nonadherence in Canada. We found that almost 1 in 20 respondents aged 12 years or older reported cost-related nonadherence and that females had 44% higher odds of reporting cost-related nonadherence than males. Our findings highlight the complex influence of demographic, health, and health care system factors on cost-related nonadherence in Canada, which exhibited independent effects even after accounting for other predictors in analyses of the overall study population and those stratified by sex.
Previous studies have shown that some racial and ethnic minority groups experience financial and structural barriers to health care access that are likely to affect adherence to prescription medications, but their sample sizes were too small to allow detailed analysis of many distinct racial identities, and they often grouped heterogeneous racial identities into a single category (e.g., Asian).18,19 In our study, compared with White respondents, Indigenous, Latin American, multiracial, West Asian, Arab, and Black respondents had 20%–67% higher odds of reporting cost-related nonadherence, whereas East Asian and Southeast Asian respondents had 16%–29% lower odds. We also evaluated the effect of racial identity on cost-related nonadherence among females and males and found effect modification between sex and Indigenous and South Asian identities. When stratified by sex, Indigenous females had 35% higher odds of cost-related nonadherence, whereas Indigenous males had 3% lower odds. Similarly, South Asian females had 20% higher odds of cost-related nonadherence, whereas South Asian males had 21% lower odds. These findings suggest an intersectional relationship between sex and race or ethnicity that may influence experiences of cost-related nonadherence and warrant further research, particularly from a narrative paradigm, to explore the individual, cultural, and systematic drivers behind these differences.
Our findings contribute evidence on the intersectional effect of sex and sexual orientation on cost-related nonadherence. Specifically, we found that bisexual, pansexual, or questioning respondents had 43% higher odds of reporting cost-related nonadherence than heterosexual respondents. Although the magnitude of this association among the female and male populations differed (i.e., 48% and 25% higher odds of reporting cost-related nonadherence, respectively), this finding was consistent with those of a 2022 study using CCHS data, which found that this disparity was particularly pronounced among bisexual females.21 Our study showed associations between homosexuality and cost-related nonadherence, with respondents who identified as lesbian having 11% lower odds of reporting cost-related nonadherence, whereas those who identified as gay had 16% higher odds; a previous study did not detect these differences.20
Previous Canadian studies have identified the effect of younger age on cost-related nonadherence;2,19 we found that respondents aged 18–34 years, who are often affected by an age-related transition of drug coverage (e.g., no longer being eligible for parents’ coverage), had 9 times greater odds of reporting cost-related nonadherence than older adults (aged ≥ 75 yr) who receive benefits under all provincial drug insurance plans.4,5
Our analysis showed that respondents living in Quebec were the least affected by cost-related nonadherence, consistent with previous Canadian research.19 As prescription medications are not universally covered under Canada’s public health insurance systems,4,5 each jurisdiction has independently developed its own drug insurance program.5,21 The lack of national standards for these programs has led to interprovincial disparities in public drug coverage related to eligibility, premiums, and cost-sharing policies (e.g., deductibles, co-payments, out-of-pocket limits) and has created the need for financing of prescription drugs via private insurance and out-of-pocket costs incurred by patients.5,22,23 Finally, our study provided nationally representative measures of the effect of several predictors identified in other Canadian studies, including change in marital status,20 lower household income,2,20 higher disease burden,2,18 lack of employer prescription drug coverage,2,18–20,24 and poor health status.2,18–20
Limitations
The variables we studied are largely non-modifiable (e.g., age, sex, province of residence); further research is needed to address how our findings may be used to inform policy decisions addressing inequities in access to pharmaceutical care. We pooled data from independent cross-sectional surveys; the observed associations between predictors and cost-related nonadherence do not imply causality. Moreover, our results are limited by the inability to capture treatment changes over time.25,26 Finally, as we used self-reported data, the accuracy of our findings is subject to both recall and social desirability bias, which may be further influenced by variations in respondents’ sociocultural characteristics.
Conclusion
In a nationally representative study, 4.9% of people in Canada reported cost-related nonadherence. Certain subsets of the population were particularly affected; this information can be used to inform eligibility for public drug coverage, premium amounts, and cost-sharing policies to reduce the need for private financing (i.e., private insurance, out-of-pocket costs), thereby addressing the financial barriers to prescription medication adherence in Canada.
Footnotes
Competing interests: Michael Law reports funding from the British Columbia Ministry of Health; consulting fees from Health Canada, Canada’s Drug Agency, and iTAD Limited; and payment for expert testimony from the Society of United Professionals, the Vancouver Firefighters’ Union, and the Durham Police Association. No other competing interests were declared.
This article has been peer reviewed.
Contributors: Nevena Rebić, Michael Law, Jacquelyn Cragg, Lori Brotto, and Mary De Vera contributed to the conception and design of the work. Nevena Rebić, Lucy Cheng, Michael Law, and Mary De Vera contributed to data acquisition. All of the authors contributed to data analysis and interpretation. All of the authors drafted the manuscript, revised it critically for important intellectual content, gave final approval of the version to be published, and agreed to be accountable for all aspects of the work.
Funding: Michael Law, Jacquelyn Cragg, Lori Brotto, and Mary De Vera are supported by the Canada Research Chairs Program.
Data sharing: The data supporting the findings of this study were obtained from Statistics Canada and are governed by a Microdata Research Contract, which strictly limits their use to the statistical and research purposes of this study.
- Accepted October 21, 2024.
This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) licence, which permits use, distribution and reproduction in any medium, provided that the original publication is properly cited, the use is noncommercial (i.e., research or educational use), and no modifications or adaptations are made. See: https://creativecommons.org/licenses/by-nc-nd/4.0/
References
In this issue
Article tools
Jump to section
Related Articles
Cited By...
- Impacts of British Columbias free contraception policy on out-of-pocket payments and contraception costs: an interrupted time series analysis with other provinces as controls
- Effect of universal no-cost coverage on use of long-acting reversible contraception and all prescription contraception: population based, controlled, interrupted time series analysis
- Des fondations a consolider pour lassurance medicaments au Canada
- The first, fragile foundations of national pharmacare in Canada
More in this TOC Section
Similar Articles
Collections
Latest Jobs
-
Kingston, Ontario
-
Oshawa, Ontario
-
Halifax, Nova Scotia
-
Sarnia, Ontario
-
Vancouver Island, British Columbia
-
Richmond Hill, Ontario
