Objective
The purpose of this study was to understand the relationship between daily contextual factors and oral contraceptive (OC) adherence among students who attend college or graduate school.
Study Design
Data on OC adherence, demographics, contextual factors, and side-effects were collected as part of the acceptability of the NuvaRing (Merck & Co, Whitehouse Station, NJ) vs OC study, in which students were assigned randomly to the contraceptive vaginal ring or to a low-dose OC. We performed bivariate and multivariable analyses to create an explanatory model for nonperfect OC adherence (missed at least 1 pill during 3 months of use).
Results
In a multivariable predictive model, missing a pill was associated positively with high perceived stress (odds ratio [OR], 3.16; P = .007), having ≥10 hours per week of paid employment (OR, 2.13; P = .075), and living with a partner (OR, 9.92; P = .040).
Conclusion
Stressful and hectic lives contribute to poor OC adherence. When counseling women about contraception, clinicians should consider the influence of daily life on contraceptive adherence.
Despite the availability of highly effective forms of contraception, in the United States, an estimated 3.1 million unintended pregnancies occur each year. Poor contraceptive adherence is a major contributor to unintended pregnancy rates because roughly one-half of the pregnancies occurred during a month when contraception was used. Although pregnancy can occur even with perfect contraceptive use, it is estimated that, for every 1 pregnancy caused by a failure of perfect use, there are 9 pregnancies caused by incorrect or inconsistent use. This disparity between the efficacy (pregnancy prevention with perfect use) and the effectiveness (pregnancy prevention with typical use) of contraceptive methods makes it imperative to understand which factors influence contraceptive adherence.
Each year in the United States, >1 million unintended pregnancies are related to oral contraceptive (OC) use, misuse, or discontinuation; thus, OC adherence requires attention. Young women are particularly important to consider because of the popularity of OCs in this age group (women 18-24 years old are most likely to use OCs), poor adherence patterns (women 18-24 years old have the highest rate of inconsistent OC use), and high rates of unintended pregnancy (women 20-24 years old have the highest rate of unintended pregnancy).
A number of factors have been associated with poor OC adherence: demographic factors (age, socioeconomic status, and race ), method-related factors (side-effects, method satisfaction), health service–related factors (satisfaction with one’s health care provider, health literacy ), and psychosocial factors (influence of the male partner, number of sex partners, perception of personal health, and degree of concern about pregnancy ). The relationship between daily-life context and OC adherence has been less well studied. It has been shown that women who lack an established pill-taking routine are more likely to miss pills. Work and school pressures are associated with missing pills. In a small study that used electronic pill packs to measure adherence, Oakley et al found that women with irregular daily schedules had more difficulty adhering to OCs than did women with a daily routine. To further examine the relationship between daily life context and OC use, we studied college and university students. This population provided homogeneity in education level and access to technology for an internet-based study. Students are also subject to stressors, long work hours, and potentially irregular sleep, all of which may impair their ability to adhere to a daily routine.
Ideally, adherence studies would identify factors that are associated with unintended pregnancy among OC users. But because unintended pregnancy is relatively infrequent, to identify these factors requires long study periods and very large groups of women. Instead, researchers often examine the relatively frequent outcome of “ineffective use,” a pattern of pill-taking that confers high risk of unintended pregnancy. Yet, definitions of “ineffective use” can vary by contraceptive method, study method, and study population. Because these definitions have not been established in an internet-based study in a college and graduate student population, we believe the most appropriate starting point is to examine the simple outcome of nonperfect use (ie, missing 1 pill). Here, we study factors that were associated with nonperfect OC adherence and hypothesize that students with a busier, more hectic life will be less likely to adhere to OCs, compared with those students with a less hectic life.
Materials and Methods
Participants
Between March 2006 and December 2008, participants were enrolled in the acceptability of the NuvaRing (Merck & Co, Whitehouse Station, NJ) vs OC study, a randomized, comparative contraceptive clinical trial that compared adherence with and acceptability of the contraceptive vaginal ring (CVR) vs OCs (Ortho Tri-Cyclen Lo; Ortho-McNeil Pharmaceutical, Raritan, NJ) among full-time undergraduate and graduate students. Outcome variables for the parent study included the following events: contraceptive adherence >3 months, side-effects, acceptability, intention to continue, and actual continuation at 6 months (3 months after study period). The current analysis focused on the pill users from the acceptability of the NuvaRing vs OC study.
Enrollment for the study occurred at the University of Chicago and the University of Illinois at Chicago. Participants were full-time undergraduate or graduate students with electronic mail (email) addresses and daily access to the internet. They were nonpregnant without pregnancy intention or with no contraindication to combined hormonal contraceptive use and without previous use of the CVR or recent use of injectable contraception. The study was reviewed and approved by the institutional review boards of both participating institutions, and all participants gave informed written consent before participation.
Procedures
After successful completion of enrollment data at a combined screening and enrollment visit, the participants were assigned randomly to use either the CVR or OCs for 3 consecutive 28-day cycles beginning with the next menses. Participants received a 3-month supply of their assigned medication along with packet instructions and method-related counseling. The Center for the Advancement of Distance Education at the University of Illinois at Chicago used an internet-based system to administer the surveys and collect data. Surveys were developed in collaboration with the Center for the Advancement of Distance Education and the Survey Research Lab at the University of Illinois at Chicago and were beta-tested by study staff before enrollment. After method initiation, the participant received an email notification containing a link to the baseline survey. The participant subsequently completed internet-based daily diaries (in response to an emailed reminder message from the Center for the Advancement of Distance Education) regarding adherence, side-effects, and satisfaction. Additional surveys were completed at 3 months (after the 3 cycles of method use) and 6 months after randomization (to determine whether the participant continued to use the method after study participation). Participants were given 2 weeks to complete diaries and surveys. For each week of completed daily dairies, participants received $5. After completing the 3- and 6-month online follow-up surveys, they received $25 for each survey.
Measures
Survey questions included known predictors of OC adherence and measures of adherence. The baseline survey included demographic variables (age, degree program, cohabitation status, living location, race/ethnicity, relationship status, and number of pregnancies). Additional data included sexual history (age at sexual debut and number of lifetime sexual partners). Contraceptive adherence questions included whether they missed a pill that day (yes/no), used condoms (yes/no), and had sexual intercourse in the past 24 hours (check number of times). At 3 months, the women were also asked whether they missed any pills during days 1-21 of any of the 3 pill-taking cycles (yes/no).
To capture daily contextual factors in the 3-month survey, we also queried participants regarding hours of weekly schoolwork, hours of paid employment, hours of extracurricular activities, and average hours of week night and weekend sleep. Perceived stress was measured also in the 3-month survey through a 4-question Perceived Stress Scale (internal consistency reliability 0.72 15 ) score that ranged from 0 (lowest stress) to 16 (highest stress). Perceived Stress Scale is a validated, widely accepted questionnaire that asks how often certain experiences of stress occurred in the last month (eg, in the last month, how often have you felt that you were unable to control the important things in your life?). The Perceived Stress Scale is a cognitive appraisal of stress and the participant’s perceived ability to cope, which correlates with outcomes such as depressive and physical symptoms, use of health care services, and smoking-reduction maintenance. In our study, stress was analyzed as a dichotomous variable with high stress defined as stress above the median stress score of the current sample.
Inquiries were made in each internet-based daily diary about the occurrence of side-effects (headaches, nausea, unscheduled bleeding and/or spotting, discharge, breast tenderness, bloating/weight gain, or vaginal discomfort). Unscheduled bleeding and/or spotting was defined as any bleeding or spotting that occurred while the woman was taking active hormone pills (except days 1-7 of the first cycle and days 1-4 of the second and third cycles). Side-effects were reported as the number of days of side-effects over the 84-day period; these values were normalized to the number of diaries that were completed.
Analysis
We excluded from analysis a small subset of participants with data that were considered to be unreliable. These women were those who were missing ≥20 daily diaries (precluding predictor categorization) or those who did not report any missed pills in daily diaries but failed to complete 5-19 diaries (precluding adherence categorization).
We used descriptive statistics to present demographic characteristics, sexual history, side-effects, and daily contextual variables. We converted most continuous variables to categoric variables. The primary outcome was nonperfect adherence to OCs , which was defined as reporting ≥1 missed pills on active pill-taking days during the 3-month study period. We performed bivariate logistic regression analysis and reported the results as percentages by the binary outcome (perfect vs nonperfect adherence) and as odds ratios (ORs) with 95% confidence intervals (CIs). Tables 1 and 2 indicate variables found to be significant to probability values of < .1 and < .05 in logistic regression. We also conducted a stepwise multivariable logistic regression using variables that were statistically significant ( P ≤ .05) or approaching significance ( P < .1) in bivariate analysis or were relevant based on the previous literature (age, race/ethnicity). We chose not to exclude a priori variables that approached significance only in the bivariate analysis but could potentially be significant in the multivariable analysis. Results from the multivariable regression are reported statistically as ORs with 95% CIs and probability values. We calculated a –2 Log Likelihood and a Nagelkerke R 2 for the multivariable model. Data were analyzed with SPSS statistical software (version 17.0; SPSS Inc, Chicago, IL).
| Demographic | Adherence | Crude odds ratio of nonperfect adherence a | 95% CI | |||
|---|---|---|---|---|---|---|
| Nonperfect (n = 62) | Perfect (n = 50) | |||||
| n | % | n | % | |||
| Age, y | ||||||
| 18-20 | 23 | 37.1 | 22 | 44.0 | 0.79 | 0.31–1.99 |
| 21-23 | 20 | 32.3 | 14 | 28.0 | 0.98 | 0.37–2.61 |
| ≥24 (reference) | 19 | 30.6 | 14 | 28.0 | 1.00 | |
| Degree program | ||||||
| Associate/Bachelor | 39 | 68.4 | 28 | 58.3 | 1.52 | 0.68–3.43 |
| Graduate/Professional (reference) | 18 | 31.6 | 20 | 41.7 | 1.00 | |
| Cohabitation | ||||||
| Lives with partner | 8 | 12.9 | 1 | 2.0 | 8.00 | 0.93–68.78 b |
| Lives with family, no partner | 17 | 27.4 | 8 | 16.0 | 2.83 | 0.96–8.37 b |
| Lives with roommate, no partner/family (reference) | 25 | 40.3 | 26 | 52.0 | 1.00 | |
| Lives alone | 12 | 19.4 | 15 | 30.0 | 0.80 | 0.31–2.05 |
| Living location | ||||||
| Lives off-campus (reference) | 48 | 77.4 | 35 | 70.0 | 1.00 | |
| Lives in dorm | 14 | 22.6 | 15 | 30.0 | 0.62 | 0.26–1.46 |
| Race/ethnicity | ||||||
| White (reference) | 29 | 46.8 | 24 | 48.0 | 1.00 | |
| Black | 15 | 24.2 | 7 | 14.0 | 1.98 | 0.66–5.92 |
| Asian | 5 | 8.1 | 8 | 16.0 | 0.50 | 0.14–1.72 |
| Hispanic | 3 | 4.8 | 2 | 4.0 | 1.19 | 0.18–7.73 |
| Other | 10 | 16.1 | 9 | 18.0 | 0.99 | 0.34–2.92 |
| Relationship status | ||||||
| In a committed relationship (reference) | 43 | 69.4 | 36 | 72.0 | 1.00 | |
| Not in a committed relationship | 19 | 30.6 | 14 | 28.0 | 1.04 | 0.46–2.38 |
| Previous pregnancy | ||||||
| No pregnancies (reference) | 51 | 91.1 | 43 | 93.5 | 1.00 | |
| ≥1 pregnancy | 5 | 8.9 | 3 | 6.5 | 1.31 | 0.29–5.80 |
| Sexual history | ||||||
| Age of sexual debut | ||||||
| ≤17 years | 31 | 55.4 | 23 | 50.0 | 1.30 | 0.59–2.88 |
| ≥18 (reference) | 25 | 44.6 | 23 | 50.0 | 1.00 | |
| Lifetime sexual partners | ||||||
| 0-3 (reference) | 25 | 44.6 | 26 | 56.5 | 1.00 | |
| ≥4 | 31 | 55.4 | 20 | 43.5 | 1.90 | 0.85–4.25 |
| Any sexual activity during study period | ||||||
| Some sexual activity (reference) | 60 | 96.8 | 43 | 86.0 | 1.00 | |
| No sexual activity | 2 | 3.2 | 7 | 14.0 | 0.19 | 0.04–0.96 c |
| Amount of sexual activity during study period | ||||||
| ≤9 acts of intercourse (reference) | 17 | 37.8 | 20 | 66.7 | 1.00 | |
| 10-24 acts of intercourse | 21 | 46.7 | 18 | 60.0 | 1.31 | 0.52–3.29 |
| ≥25 acts of intercourse | 24 | 53.3 | 12 | 40.0 | 2.12 | 0.81–5.52 |
| Contextual and behavioral characteristics | ||||||
| Weekly schoolwork, h | ||||||
| 0-9 (reference) | 16 | 25.8 | 12 | 24.0 | 1.00 | |
| ≥10 | 46 | 74.2 | 38 | 76.0 | 0.78 | 0.32–1.91 |
| Weekly paid employment, h | ||||||
| 0-9 (Reference) | 24 | 38.7 | 30 | 60.0 | 1.00 | |
| ≥10 | 38 | 61.3 | 20 | 40.0 | 2.55 | 1.17–5.56 c |
| Weekly extracurricular activities, h | ||||||
| 0-9 (reference) | 43 | 69.4 | 32 | 64.0 | 1.00 | |
| ≥10 | 19 | 30.6 | 18 | 36.0 | 0.78 | 0.35–1.74 |
| Average weeknight sleep, h | ||||||
| ≤6 | 28 | 45.2 | 21 | 42.0 | 1.11 | 0.52–2.39 |
| ≥7 (reference) | 34 | 54.8 | 29 | 58.0 | 1.00 | |
| Average weekend sleep, h | ||||||
| ≤7 | 18 | 29.0 | 16 | 32.0 | 0.79 | 0.35–1.79 |
| ≥8 (reference) | 44 | 71.0 | 34 | 68.0 | 1.00 | |
| Perceived stress | ||||||
| Low stress (reference) | 26 | 42.6 | 33 | 66.0 | 1.00 | |
| High stress | 35 | 57.4 | 17 | 34.0 | 2.43 | 1.10–5.35 c |
| Side-effects | ||||||
| Days of any side effects d (out of 84 days) | 43 e | 35 e | 1.01 f | 1.00 b | ||
| Days of unscheduled bleeding and/or spotting d (out of 84 days) | 3 e | 5 e | 0.95 f | 0.89–1.01 b | ||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
