Out-of-pocket spending on drugs and pharmaceutical products and cost-related prescription non-adherence among Canadians with chronic disease
by Deirdre Hennessy, Claudia Sanmartin, Paul Ronksley, Rob Weaver, Dave Campbell, Braden Manns, Marcello Tonelli, and Brenda Hemmelgarn.
Prescription drugs are essential in the management of chronic diseases—an estimated 90% of individuals with major chronic conditions report taking one or more medications.Note 1Note 2 Unlike hospital and physician care, prescription medications (other than those provided in hospital) are not covered by a universal insurance plan. Although all Canadian provinces provide some publicly funded drug insurance for certain segments of the population such as seniors and people receiving social assistance, provinces differ in their coverage of prescription medications for other groups and in the provision of catastrophic drug coverage.Note 3Note 4Note 5Note 6 For many Canadians, prescription drugs are financed primarily through private insurance and out-of-pocket expenditures.Note 7Note 8 According to the Canadian Institute for Health Information, out-of-pocket spending made up 34.3% of private expenditures on prescription medications in 2010.Note 7
In 2011, Canadian households spent an average of $476 out-of-pocket on prescription medications.Note 9 Average annual expenditures were higher at older ages, exceeding $600 for households with heads older than 55.Note 10 In 2010, 20% of households spent more than 1% of their after-tax income on prescription medications, and 3% spent more than 5%.Note 11 The economic burden of out-of-pocket expenditures related to medications and health care in general has grown over time, particularly among the lowest income households.Note 8Note 12Note 13
Inconsistent public insurance coverage for prescription drugs and high out-of-pocket spending have raised concern that some people may forego medications owing to cost.Note 14Note 15 Law et al. observed that about one Canadian in 10 reported cost-related non-adherence to prescriptions.Note 16 For individuals with chronic diseases, non-adherence can have serious consequences, including increased mortality, use of emergency departments, and need for in-patient services.Note 17Note 18Note 19Note 20Note 21Note 22 Based on the same source as the present analysis, Campbell et al. found that patients facing financial barriers to care were more likely to visit the emergency department or to be admitted to hospital for their chronic conditions and less likely to use guideline-recommended therapies like statins.Note 6
Previous research has identified factors associated with cost-related prescription non-adherence (including out-of-pocket spendingNote 12Note 23), but no Canadian study has examined the relationship between the level of out-of-pocket spending and the likelihood of cost-related non-adherence. North American studies on health care expenditures overall (including prescription medications) have used an out-of-pocket threshold of 10% of income to indentify high-spending households.Note 12Note 24 In Canada, the 2002 Kirby Report recommended that households pay no more than 3% of their after-tax income for prescription medications,Note 25 although this level was not empirically associated with a higher likelihood of cost-related non-adherence. While a cut-point for high out-of-pocket spending on medications has not been identified, national agencies currently use 5% of household income as a health indicator.Note 11Note 26
This study analyzed results from the survey on Barriers to Care for People with Chronic Health Conditions to determine if spending 5% or more of household income out-of-pocket on drugs and pharmaceutical products was associated with cost-related prescription non-adherence.
The survey on Barriers to Care for People with Chronic Health Conditions (BCPCHC) was designed by the Interdisciplinary Chronic Disease Collaboration (www.ICDC.ca)Note 27 and administered by Statistics Canada. The sampling frame consisted of respondents aged 40 or older to the 2011 Canadian Community Health Survey (CCHS) who resided in British Columbia, Alberta, Saskatchewan or Manitoba and who reported having heart disease, stroke, diabetes, or hypertension. Based on these criteria, 2,400 respondents were identified for participation in the BCPCHC; 2,316 were still in scope at the time of survey administration (February 1 to March 31, 2012). The response rate was 80%, yielding a final sample of 1,849 individuals. Respondents were asked about care that they received for their chronic condition(s) and potential barriers to care that they experienced. With participant consent, responses to the BCPCHC were linked to their 2011 CCHS data to provide more detailed sociodemographic information, including household income. In accordance with Statistics Canada procedures, the BCPCHC underwent pilot testing to ensure that potential participants understood the questions.
This study was approved by the Conjoint Health Research Ethics Board of the University of Calgary and the Health Research Ethics Board of the University of Alberta.
Cost-related non-adherence to prescription medications, the primary outcome in this analysis, was determined from responses to two sets of questions. In the first set, respondents were asked: “In the past 12 months, how often did you not get the necessary services, equipment or drugs you needed due to cost?” Those who answered “always,” “often,” or “sometimes” were asked: “What type of services, equipment or medication?” Those who specified “prescription medication” were flagged as reporting cost-related non-adherence. In the second set of questions, respondents were asked: “Over the past 12 months, have you ever stopped taking one or more of your drugs as prescribed for a week or more?” Those who answered affirmatively were asked: “What are the reasons that you did not take your medication as prescribed - Cost?” Respondents who answered “yes” to either of these questions were classified as reporting cost-related non-adherence.
Out-of-pocket spending as a percentage of household income, which has been used as a health indicator,Note 10Note 11Note 26 was the primary exposure of interest. Respondents were asked to estimate their out-of-pocket spending on drugs and pharmaceutical products during the previous 12 months; those who could not recall the past 12 months were asked about the previous 3 months, and their annual expenses were extrapolated. About 11% of respondents did not reply to this question, and were excluded from the main analysis. Self-reported out-of-pocket spending on drugs and pharmaceutical products was divided by self-reported household income and multiplied by 100 to determine the percentage of total household income allocated to these expenditures. Of the study sample, 70% provided their annual household income; for the remainder, income was imputed using a nearest-neighbor technique including postal code, median income, household size, and approximate household income range.Note 28
The variables in the analysis included sex, age group (40 to 64, 65 to 74, and 75 or older) and household income (less than $30,000, $30,000 to $54,999, $55,000 to $94,999, and $95,000 or more). Participants were chosen for the BCPCHC based on a positive response to questions about physician-diagnosed heart disease, stroke, diabetes, or hypertension in the 2011 CCHS. For descriptive analysis, respondents were categorized as having one versus two or more of these chronic diseases. Respondents were asked how many medications they took on an ongoing basis, and if insurance covered all or part of their prescription medications costs.
Out-of-pocket spending on drugs and pharmaceutical products as a percentage of household income was grouped into three categories: 0%, more than 0% to less than 5%, and at least 5%. A descriptive analysis provided a profile of respondents in each spending category. The 5% cut-off was chosen because national agencies use it as a health indicator.Note 11Note 26 A sensitivity analysis, with out-of-pocket spending at 3% of household income as the cut-off, was also completed.
Log-binomial regression models were used to calculate unadjusted and adjusted prevalence rate ratios (PRR) of the association between out-of-pocket spending on drugs and pharmaceutical products and cost-related prescription non-adherence. The purpose was to determine if out-of-pocket spending of more than 5% of household income was associated with cost-related non-adherence to prescribed medication, not to identify factors contributing to cost-related non-adherence, which have previously been described for Canada.Note 12Note 23
For all tests, p < 0.05 was considered statistically significant. The analyses were weighted to reflect the population aged 40 or older with cardiovascular-related conditions in the four western provinces using sampling weights provided by Statistics Canada. Bootstrapping techniques with 500 replications were used to calculate standard errors and confidence intervals (CI) around the estimates. To determine the reliability of the reported estimates, the coefficient of variation (CV) was calculated.Note 29 As per Statistics Canada guidelines, estimates that did not meet minimum cell-size requirements and had CVs greater than 33.3% were not reported (denoted by F). Estimates that met cell-size requirements, but had a CV greater than 33.3%, were reported as being less than the upper limit of the 95% confidence interval. All analyses were conducted using STATA 11.0 (Statacorp, College Station, TX).
Fully 90% of the 1,849 respondents to the survey on BCPCHC answered the questions about out-of-pocket spending on drugs and pharmaceutical products. Of these individuals, 19% reported no out-of-pocket spending; 76% reported spending more than 0% but less than 5% of household income; and 4.8% reported 5% or more (Table 1). The overall mean out-of-pocket expenditure was $539. The mean among those spending less than 5% of household income was $517, but among for those spending 5% or more, the mean was $3,021 (Table 1).
Compared with people who spent less than 5% of household income on drugs and pharmaceutical products, those spending 5% or more were older, and more likely to live in households receiving less than $30,000 annually (53% versus 18%) and to have at least two cardiovascular-related conditions (52% versus 31%) (Table 1). They also used significantly more medications on average (6.9 versus 3.9). As well, 25% of those spending more than 5% did not have supplemental insurance for prescription medications, compared with 15% of those spending less than 5%; although substantial, this difference was not significant.
Overall, 4.1% (95% CI: 2.6 - 6.3) of individuals reported cost-related prescription non-adherence (Table 2). As expected, people without any out-of-pocket expenses on drugs and pharmaceutical products did not report cost-related non-adherence. However, 5.2% of those with out-of-pocket greater than 0% but less than 5% of household income reported cost-related non-adherence. Because of a large coefficient of variation, it was not possible to report the exact prevalence of cost-related prescription non-adherence among individuals whose out-of-pocket spending amounted 5% or more of their household income; nonetheless, the upper bound of the 95% confidence interval of the estimate was 21.5%.
Log binomial modelling of association between out-of-pocket spending on drugs and pharmaceutical products at 5% and 3% of household income and cost-related prescription non-adherence, household population age 40 or older with cardiovascular-related conditions, Manitoba, Saskatchewan, Alberta, and British Columbia, 2012
According to results of the unadjusted log binomial regression model, the prevalence rate ratio (PRR) for cost-related prescription non-adherence among individuals whose out-of-pocket spending on drugs and pharmaceutical products was 5% or more of household income was 2.2 (95% CI: 0.8 - 5.6), which was not significantly different from 1.0. Age- and sex-adjusted results revealed that those spending 5% or more were significantly more likely to report cost-related non-adherence than were those spending less than 5% (PRR = 2.6; 95% CI: 1.0 - 6.4) (Table 2). A sensitivity analysis showed that cost-related prescription non-adherence was not significantly related to out-of-pocket spending at 3% of household income, in either the unadjusted or adjusted models (PRR = 2.0; 95% CI: 0.9 - 4.9 in adjusted model) (Table 2).
About 80% of western Canadians aged 40 or older with cardiovascular-related chronic diseases reported out-of-pocket spending on drugs and pharmaceutical products; for almost 5% of them, this spending amounted to at least 5% of their household income. These individuals were older, more often lived in households with annual income less than $30,000, and more often reported multiple morbidities, compared with people whose out-of-pocket spending on drugs and pharmaceutical products was less than 5% of their household income. On average, the out-of-pocket amounts reported by the higher-expenditure group were about six times those of respondents whose spending was less than 5% of household income. In addition, after adjustment for age and sex, spending more than 5% of household income out-of-pocket on drugs and pharmaceutical products was significantly associated with a higher likelihood of cost-related prescription non-adherence, compared with those whose spending totalled less than 5%.
The overall prevalence of cost-related prescription non-adherence in this study (4.1%) was lower than the almost 10% found by Law et al.Note 11 The results were closer to those of Kennedy and Morgan, who, in a cross-national study of cost-related prescription non-adherence in Canada and the United States, reported an overall rate of 5.1%.Note 23 The study by Law et al. was based on a national survey representing Canadians aged 12 or older; Kennedy and Moore sampled people aged 18 or older. By contrast, the sample in the present study consisted only of people aged 40 or older with specific cardiovascular-related diseases, among whom prescription non-adherence would be less common because the consequences are potentially life-threatening and because a larger percentage of respondents (due to older age) would have provincial drug plan coverage.
The factors associated with cost-related prescription non-adherence in the present analysis are generally consistent with Canadian and international literature. In Quebec, Tamblyn et al. found that 31.3% of first-time prescriptions were not filled within nine months, with the risk of primary non-adherence being greater for higher-cost medications and among patients with higher drug co-payments.Note 30 Internationally, Helin-Salmivaara et al. showed that cholesterol medications that required smaller out-of-pocket payments promoted adherence to ongoing therapy.Note 31 In British Columbia, Hanley et al. demonstrated that uptake of recommended therapies after an AMI was less than optimal among individuals of lower socioeconomic status, possibly because of small co-payments for prescription medications in that province.Note 32 Similarly, Kennedy and Morgan found that cost-related non-adherence was associated with lower income and with lack of prescription medication insurance, which indirectly reflects out-of-pocket costs.Note 23 In a review of patient-borne costs, non-adherence, and health outcomes,Note 33 Eaddy et al. concluded that patient cost-sharing was associated with decreasing adherence across a range of therapies, which, in turn, was associated with poorer health outcomes.
While the association between out-of-pocket spending and cost-related prescription non-adherence is well established, this study adds empirical information about an upper threshold of spending linked to cost-related non-adherence. In the present analysis, individuals whose out-of-pocket spending on drugs and pharmaceutical products was at least 5% of their household income were more likely to report cost-related non-adherence. While this suggests that some level of out-of-pocket spending can be tolerated before costs negatively affect adherence, the threshold may actually be lower. Evidence from Canada and the United States indicates that for certain populations, even small co-payments increase the risk of prescription non-adherence.Note 30Note 34Note 35 A larger dataset would be required to determine the exact level of out-of-pocket spending at which cost-related non-adherence occurs among Canadians with chronic conditions.
The small sample size of this study limited the ability to more precisely define the level (rather than a range) of out-of-pocket spending associated with cost-related non-adherence. Additional research is needed to confirm these findings. As well, the data were self-reported, and therefore, are subject to recall and social desirability biases. For example, 11% of respondents did not know or could not estimate the amount they spent out-of-pocket on drugs and pharmaceutical products. Also, the results are based on out-of-pocket spending of individuals, and may underestimate total out-of-pocket spending of households. Finally, the generalizability of the results is limited to people with heart disease, stroke, diabetes, or hypertension.
Sub-optimal adherence to prescriptions for medications threatens both individual and societal health.Note 36 Canadians with cardiovascular-related conditions in the four western provinces whose out-of-pocket spending on drugs and pharmaceutical products was at least 5% of their household income were almost three times as likely to report cost-related prescription non-adherence as were those who spent less than 5%. While the results of this study point to an upper threshold, more data are required to determine if even lower levels of spending put individuals at risk of cost-related non-adherence.
These findings may be relevant for policy makers interested in measuring and tracking out-of-pocket spending on prescription medications as a health indicator, and underline the importance of collecting similar data to confirm the findings of the current study.
- Date modified: