Oral contraceptive use among women aged 15 to 49: Results from the Canadian Health Measures Survey
by Michelle Rotermann, Sheila Dunn and Amanda Black
Oral contraceptives (OCs) have been available in Canada for more than 50 years and are the most commonly used method of reversible contraception.Note 1,Note 2 They are also among the medications most frequently used by Canadian women,Note 3 three-quarters of whom take OCs at some point in their lives.Note 4 OCs include combined hormonal contraceptive pills, which contain both estrogen and a progestin, and progestin-only pills.
Although primarily indicated for pregnancy prevention, OCs have non-contraceptive benefits, including cycle regulation; less dysmenorrhea; fewer ovarian cysts; improved perimenopausal, vasomotor, moliminal and endometriosis symptoms; decreased menstrual flow, acne and hirsutism; and decreased risk of endometrial and ovarian cancer.Note 5-9
For the vast majority of healthy, non-smoking women, OCs are safe.Note 1,Note 4,Note 8 However, like many medications, OCs are associated with side effects and risks,Note 7,Note 10-14 including cardiovascular events such as venous thromboembolism (VTE), myocardial infarction, and stroke. Prospective cohort studies have found that the risk of VTE is two times higher in OC users than in non-users (9 or 10/10,000 woman-years versus 4 or 5/10,000 woman-years).Note 15,Note 16 Even so, the risk of VTE during pregnancy and the postpartum period is much higher than that associated with OC use.Note 17 OCs may also increase the risk of breast cancer,Note 12,Note 18,Note 19 although 10 years after use ceases, the risk returns to baseline.Note 18
OCs have evolved over time, with decreasing estrogen doses, new progestins, and different dosing regimens. As the estrogen dose has decreased, the risk of cardiovascular events and certain cancers appears to have lessened.Note 12,Note 20 The use of different progestins may confer different risk-benefit profiles.Note 9,Note 10,Note 13,Note 21
Despite widespread exposure of Canadian women to these medications, detailed information about OC use is lacking. Administrative dispensary or billing data contain very basic sociodemographic information and tend to be available only for certain provinces (for example, British ColumbiaNote 22,Note 23). Some national surveys have included OC-specific content, but are either datedNote 24 or pertain only to youth,Note 25 and none have captured information about OC formulations. For instance, the Canadian Contraception Survey lacked data on OC type and other health indicators.Note 2
Based on results for 2007 to 2011 from the Canadian Health Measures Survey (CHMS), the present study estimates the prevalence of OC use among non-pregnant reproductive-aged women; profiles OC users by sociodemographic characteristics and cardiovascular risk factors; and identifies the OCs used, by estrogen dose and progestin type.
The CHMS, which is conducted by Statistics Canada in partnership with the Public Health Agency of Canada (PHAC) and Health Canada, produces nationally representative estimates.Note 26 Data collection occurs in two steps: an interview at the respondent’s home and a subsequent visit by the respondent to the CHMS mobile examination centre where physical measures and blood and urine samples are taken. The CHMS excludes people living on reserves and other Aboriginal settlements in the provinces, full-time members of the Canadian Forces, the institutionalized population, and residents of some remote regions (less than 4% of the target population).Note 27 Ethics approval for the CHMS was obtained from Health Canada’s Research Ethics Board.Note 28
Of households selected for CHMS cycles 1 (2007 to 2009) or 2 (2009 to 2011), 72.7% agreed to participate, 89.3% of whom completed the household questionnaire. After adjustments for the sampling strategy, the final response rate for people age 6 to 79 for the two cycles combined was 53.5%.Note 27
Cycle 1 took place from March 2007 through February 2009, and collected information from 5,604 respondents aged 6 to 79 living in private households in 15 locations across Canada. Cycle 2 took place from August 2009 through November 2011, and collected data from 6,395 respondents aged 3 to 79 living in private households in 18 locations. The combined sample for the two cycles was 11,999. Cycle 1 respondents were not eligible to participate in cycle 2. Nearly 95% (11,387/11,999) of cycle 1 and 2 respondents completed the mobile examination centre component. One record was dropped because all prescription medication-related fields were missing.
The present study combined data for 15- to 49-year-old women from each cycle for a total 2,790. The analysis excluded 85 respondents because of self-reported pregnancy (n = 66) or missing/unknown pregnancy status (n = 19), yielding a final sample of 2,705.
Drug identification numbers (DINs) were collected from the packaging of medications (maximum of 15) during the household interview and verified during the mobile examination centre visit. Only medications that respondents reported taking in the month before the household interview were used. A computer-assisted reference tool facilitated medication capture when required. Each DIN was associated with Anatomical Therapeutic Chemical (ATC) classification codes assigned by Health Canada.Note 29 In accordance with Statistics Canada’s Directive on Sensitive Information, which includes the protection of brands, the specific medications identified based on the DINs are presented as aggregate ATC codes.Note 30
The OCs used by study participants correspond to 13 level-7 ATC codes. Oral contraceptive users (Appendix A) were compared with non-users. Non-users were defined as women who did not report using OCs or other hormonal contraceptive formulations in the month before the household interview. Non-oral hormonal contraceptives included transdermal, intravaginal, and injectable contraceptives and correspond to three level-7 ATC codes (Appendix A). Identification and removal of users of these other hormonal contraceptive formulations helped ensure that the groups were as distinct as possible and were not taking similar hormones delivered non-orally.
For several reasons, use of hormonal IUDs was likely underestimated by CHMS respondents. Respondents may not have considered the device a medication, and it may have been inserted several years previously, thereby affecting recall. The small number of respondents (fewer than 5) who reported hormonal IUDs were considered “non-users” in the analysis, largely because the non-user group likely contained other respondents who did not report use of a similar device. Preliminary analyses suggested that excluding these cases altogether or including them in the non-user or other hormonal category did not substantively change the results.
Descriptive statistics were used to present OC use by age group, marital status, parity, immigrant status, household income, sexual activity, and having a regular doctor. Multiple logistic regression was used to model relationships between these variables and OC use. Selection of covariates was based on the literature and data availability.
Age groups were established according to the respondent’s age at the time of the household interview. Marital status was categorized as married, living common-law, previously married (including separated, divorced and widowed), and single (never married). Parity (number of live births) was categorized as nulliparous (none) and parous (one or more). Respondents were classified as being sexually active if they answered “yes” to the question: “In the past 12 months, have you had sexual intercourse?” Immigrant status was classified as immigrant or Canadian-born based on country of birth and citizenship. Household income, adjusted for household size, was classified into two categories: higher and lower. Households were classified as higher income if their past-year adjusted total earnings were in the highest income quartile (top 25% of households), and lower income if they were in the lower three quartiles. Respondents were asked if they had "a regular medical doctor.”
Descriptive statistics were used to estimate OC use by cardiovascular risk factors overall and by age. Cardiovascular risk factors were: daily smoking, body mass index (BMI) based on measured height and weight (BMI) based on measured height and weight, and a composite measure of “heart disease, stroke or hypertension.”
Those aged 18 or older with a BMI of 25 or more were classified as overweight/obese; those whose BMI was 15 to 29.9 were classified as overweight; and those whose BMI was 30 or more were classified as obese. For 15- to 17-year-olds, the Centers for Disease Control BMI categorization was used. Adolescents of the same age and sex with height/weight corresponding to the 86th percentile or higher were classified as overweight/obese; those in the 86th to 95th percentile were classified as overweight; and those above the 95th percentile were classified as obese.
The composite measure of heart disease, stroke or hypertension was based on respondents’ reporting that a health professional had diagnosed them with heart disease, heart attack, high blood pressure (BP) or stroke. Use of cardiovascular-related medications—anti-hypertensives, lipid-modifying agents (statins), thrombolytics (anti-coagulants) or anti-platelets—was also considered to indicate the presence of these conditions (Appendix B). Respondents’ BP was taken six times; those with an average systolic BP of 140 mmHg or more and/or an average diastolic BP of 90 mmHg or more were considered to have hypertension. The three cardiovascular risk factors (daily smoking, overweight/obesity, and the composite measure or heart disease, stroke and/or hypertension) were combined into a single dichotomous cardiovascular risk factor variable; respondents were categorized as having at least one risk factor or no risk factors. Data on other medical conditions that are contraindications for OC use, such as migraine headache with aura and personal history of breast cancer,Note 31 were either not available, or the sample with the condition was too small to analyze (fewer than 10).
OC use was categorized by the dose of ethinyl estradiol (EE) (less than 30 micrograms (mcg) or 30 or more mcg ), and by progestin type (levonorgestrel, norgestimate, drosperinone, desogestrel or other). Information about active ingredients and EE dosages was taken from Health Canada’s Drug Product Database.Note 32
To account for survey design effects, coefficients of variation and 95% confidence intervals were estimated with the bootstrap technique.Note 33 Differences between prevalence estimates were calculated with t-tests. All analyses were conducted in SUDAAN v.11 (RTI International, Research Triangle Institute, NC, USA), using weighted data and DDF = 24 in the procedure statements to account for the degrees of freedom of the combined datasets. Details about the CHMS, including sampling, quality assurance and combining cycles, are available elsewhere.Note 26,Note 27
Sociodemographic and behaviour characteristics
According to combined data from the 2007 to 2009 and 2009 to 2011 CHMS, 1.3 million (16%) non-pregnant women aged 15 to 49 used OCs in the previous month (Table 1). OC use decreased from 30% among 15- to 19-year-olds to 3% among 40- to 49-year-olds. The average age of OC users was 26 years, compared with 35 years for non-users (data not shown). OC use was significantly higher among single compared with married/previously married women, among nulliparous compared with parous women, and among Canadian-born compared with immigrant women. Differences in the prevalence of OC use by sexual activity, household income, and having a regular medical doctor were not significant.
Because sociodemographic and behaviour characteristics are not independent of each other, multiple logistic regression analysis was performed to account for the simultaneous effects of these factors (Table 2). The adjusted odds of OC use were higher for women aged 15 to 39 than those aged 40 to 49. As well, the adjusted odds were significantly higher in women who were nulliparous, sexually active in the past year, or Canadian-born. Marital status, household income, and having a regular doctor were not significantly associated with OC use.
Cardiovascular risk factors
All cardiovascular risk factors except smoking were less prevalent among OC users than among non-users aged 15 to 49. OC users were significantly less likely than non-users to be overweight (19% versus 25%), obese (15% versus 22%), or have heart disease, stroke and/or hypertension (5% versus 10%) (Table 3). Differences in smoking rates between the user and non-user groups were not significant. When these cardiovascular risk factors were combined into a single dichotomous measure, OC users were significantly less likely than non-users to have at least one (43% versus 58%).
At older ages, cardiovascular risk tends to increase, while OC use declines. To examine the prevalence of cardiovascular risk factors among OC users and non-users independent of age differences, results were stratified by age group (15 to 34 versus 35 to 49). In both age groups, the prevalence of each cardiovascular risk factor tended to be lower among OC users, but only one difference was statistically significant: at ages 35 to 49, non-users were more likely to have at least one cardiovascular risk factor (67% versus 48%).
Oral contraceptive formulation
Almost all (99%) OC users took OCs containing a combination of ethinyl estradiol (EE) and a progestin (data not shown). More than half (56%) took OCs containing EE in combination with levonorgestrel or norgestimate (Table 4). Desogestrel- and drospirenone-containing formulations were taken by 17% and 16% of OC users, respectively.
During the study reference period, the combination OCs available in Canada contained 20 to 50 mcg of EE. Two-thirds of users (66%) took an OC containing 30 or more mcg of EE (Table 5), with most (99%) taking 30- or 35-mcg-EE formulations (data not shown). Women aged 15 to 24 were significantly less likely than those aged 35 to 49 to take OCs containing 30 or more mcg of EE: 60% versus 79%.
This is the first nationally representative analysis of OC use by sociodemographic characteristics, cardiovascular risk factors, and medication formulation.
Estimates of OC use in Canada (16% of women aged 15 to 49) were comparable to survey-based estimates for 2006 to 2010 from the United States (17% of women aged 15 to 44)Note 34 and a 1996/1997 Canadian estimate (18% of women aged 15 to 49).Note 24 It is difficult to compare CHMS data with those of studies from countries such as DenmarkNote 35 and Australia,Note 4,Note 36 which used non-specific annual or bi-annual reference periods and/or excluded some populations.
In this study, OC use decreased with age, and nulliparous women were more likely than parous women to use OCs. These findings are similar to other Canadian studies and to American, Danish, and Australian research.Note 2,Note 4,Note 24,Note 34-36 Possible explanations may be that at older ages, women prefer IUDs, permanent contraceptive methods, or barrier methods. Health care providers may also be hesitant to prescribe OCs for women older than 40 because of the risk of adverse events or because older women are more likely to have medical conditions that are contraindications for OC use.Note 37
Cultural factors may affect attitudes toward contraception and the use of hormones. According to CHMS results, Canadian-born women were more likely than immigrant women to use OCs. Other studiesNote 34,Note 38,Note 39 have also found lower use of hormonal contraceptives by immigrant and/or minority women, who may prefer condoms, intrauterine devices, and permanent methods.Note 34,Note 38,Note 40 Preference for non-hormonal methods may also stem from concerns about weight gain and compromised future fertility, and a belief that OCs are unhealthy.Note 39,Note 41
Certain cardiovascular risk factors―uncontrolled hypertension, smoking after age 35, and heart disease―are contraindications for OC use.Note 37 Other cardiovascular risk factors, such as obesity, are not contraindications, but may be associated with an increased risk of adverse events. Although the CHMS data demonstrated a lower prevalence of cardiovascular risk factors among OC users than non-users, the former were considerably younger. It is not clear if the difference in cardiovascular risk can be attributed solely to the age gap between the groups, because the CHMS sample is not large enough to attain statistical significance for most calculations. As more CHMS cycles become available, sample sizes and statistical power will increase.
Potential side effects and adverse events associated with OCs may be influenced by estrogen dose and progestin type. For example, cancer risk or adverse cardiovascular events may be related to estrogen dose,Note 12,Note 20,Note 42 so reducing EE may lower rates of these adverse events. After 2010, OCs containing 50 mcg or more of EE ceased to be available in CanadaNote 32; currently, only OCs containing 35 mcg or less of EE are available. CHMS data covering the 2007-to-2011 period show that virtually all (99%) Canadian OC users were taking formulations with 35 mcg or less of EE: one third were using formulations with less than 30 mcg. Younger women, who presumably would be healthier and have fewer cardiovascular risk factors, were more likely than women aged 35 to 49 to be taking OCs with less than 30 mcg of EE. Similarly, other research showed that newer products, which tend to contain less EE, are more often taken by younger womenNote 35; older women, with potentially higher risk profiles, continue to take higher EE-dose formulations.
The impact of different progestins on risk and side effect profiles is controversial. Some experts consider levonorgestrel to be the safest,Note 10,Note 11 with the lowest risk of OC-associated VTE. Others contend that no appreciable difference exists in VTE risk between any of the progestins.Note 15,Note 16,Note 43
According to the CHMS data, the progestins most commonly taken were levonorgestrel and norgestimate. The CHMS findings by progestin type are comparable to administrative prescription data from British Columbia,Note 22 but are less consistent with prescribing patterns in other countries.Note 44
The results of this analysis should be interpreted in the context of several limitations. The data are self-reported, and so are susceptible to recall bias. Logistical and budgetary constraints restricted the number of CHMS collection sites and sample sizes.Note 27 Consequently, this analysis occasionally used more general covariate categories than desirable. As well, the CHMS does not collect data on some relevant covariates, such as pregnancy intentions, recent pregnancy/birth, use of other contraceptive methods, duration of OC use, indications for OC use, breastfeeding, and insurance coverage. Small sample sizes may result in the analysis being underpowered to detect statistically significant differences. Although survey weights ensured that the sample is representative of the target population, bias may exist if the use of medications by respondents and non-respondents differed systematically.
This analysis provides a basis for examining trends in the use of OCs by Canadian women. The results also allow for international comparisons. This is the first national, population-based study to provide detailed sociodemographic and cardiovascular risk information about Canadian OC users and the types of OCs they take. As additional CHMS data become available in the future, it will be possible to combine cycles for a more detailed and ongoing analysis of OC use in Canada.
- Date modified: