The data

View the most recent version.

Estimates of seniors’ medication use were based on the latest cycles of the National Population Health Survey (NPHS) that collected information on medication use in the two days before their interview. Detailed documentation on the NPHS can be found on Statistics Canada’s website, and descriptions of the survey design, sample, and interview procedures are available in published reports.^10, 11

The NPHS household file covers household residents in all provinces, except persons living on Indian reserves, on Canadian forces bases, and in some remote areas. The data in this analysis are from 1998/1999 (cycle 3), which has a household response rate of 87.6% and a selected person response rate of 98.5%. Eleven percent of the senior household sample (317) relied on proxy reporters (Appendix Table A).

The NPHS health institutions file covers people living in hospitals, nursing homes, and facilities for people with disabilities. The data in this analysis are from 1996/1997 (cycle 2), which has institutional and individual response rates of 100% and 89.9%, respectively. Fifty-nine percent of the institutionalized respondents (1,013) relied on proxy reporters—49% were family members, and 10% were staff of the institutions.

Demographic distributions of the samples and populations used in this analysis are presented in Appendix Table A.

The primary outcome variables in this study are medication use and multiple medication use. Medication use refers to prescription and OTC medications including natural and alternative medicines. Household residents, who were usually interviewed by telephone, were asked to gather their medications and read the names from the containers. For institutionalized respondents, staff members of the institutions provided this information; these medications would all be classified as “prescribed,” because seniors in institutions usually do not have the option of self-medicating. Current users were those who had taken medication in the two days before their interview.

The terms multiple medication use and polypharmacy are sometimes used interchangeably. The latter has been defined in the literature in relative terms (for example, the administration of an excessive number of drugs) and in absolute terms, ranging from two to more than six simultaneous medications.^5, 12-15 In this study, preference is given to the term, multiple medication use, defined as currently taking five or more different medications. The threshold of five is relatively conservative alongside other absolute definitions of polypharmacy, and is consistent with an earlier Statistics Canada study based on the NPHS.¹⁶

NPHS respondents reporting current medication use were asked the names of their medications; data were recorded for a maximum of 12 medications. Drugs were listed in the order that they were reported, and so could not be ranked according to strength or importance. The drugs were coded using the Canadian edition of the Anatomical Therapeutic Chemical (ATC) Classification System for Human Medications.

Self-perceived health was based on the question, “In general, would you say your health is: ….” The five response categories were combined into two groups: good/very good/excellent health was defined as “positive” self-perceived health, and fair/poor health as “negative” self-perceived health.

Chronic pain was defined as a response of “no” to the question, “Are you usually free of pain or discomfort?”

The presence of chronic conditions was established by asking respondents if they had been diagnosed by a health professional with a long-term chronic health condition—one that had lasted, or was expected to last, at least six months. Respondents were read a list of conditions. Individual conditions included in this study were incontinence, arthritis, diabetes, heart disease, stroke, Alzheimer’s disease or other dementia, and cataracts. For the institutional population, incontinence included urinary incontinence and difficulty controlling bowels, but for the household population, was limited to urinary incontinence. A more comprehensive list of chronic conditions was used to estimate the overall number of chronic conditions each respondent experienced (Appendix Table B). The count of chronic conditions was categorized into three groups: none or 1, 2, and 3 or more.

The analysis was based on independent samples from households and institutions. Data were weighted to reflect the age and sex distribution of the appropriate target populations. Weighted frequencies and cross-tabulations were used to estimate the proportion of people who had used medication/multiple medications in the past two days by selected characteristics. Logistic regression was used to model associations between indicators of ill health (chronic pain and number of chronic conditions) and multiple medication use while controlling for sex, age, education, and proxy reporting status. To account for survey design effects, standard errors and coefficients of variation were estimated with the bootstrap technique.^17-19

The current study has a number of limitations. The data on institutions are from the 1996/1997 NPHS, whereas the household data are from the 1998/1999 NPHS. These surveys are the most recent from which multiple medication use can be established. The count of chronic conditions may vary between household and institutional residents, in part, because the lists of conditions were not identical in the two surveys (Appendix Table B). As well, chronic conditions were self-reported and were not verified by any other source.

Respondents who reported medication use in the past two days were limited to providing the names of 12 different drugs. Seven of the 2,851 household sample and 27 of the 1,711 institutional sample reported more than 12 different medications. For an additional 18 household and 73 institutional respondents, data on the number of different medications taken in the past two days were missing.

It is possible that respondents may not consider certain OTC products such as vitamins and natural/herbal products to be drugs, in which case the true number of medications taken would be under-reported.

A substantial share of the respondents—11% of the household sample and 59% of the institutional sample—relied on proxy reporters. However, excluding these respondents (the most seriously ill or cognitively impaired seniors) would have biased the results.