Objective
The objective of the study was to assess the adherence to hormonal contraception (pill, patch, ring, or injectable) among women veterans and examine the relationships between race/ethnicity and the months of contraceptive supply dispensed with contraceptive adherence.
Study Design
We conducted a retrospective analysis of the Department of Veterans Affairs (VA) national databases to examine the adherence to hormonal contraception over 12 months among women aged 18-45 years who had hormonal contraceptive coverage during the first week of fiscal year 2008. We examined several adherence indicators including gaps between refills and months of contraceptive coverage. Descriptive statistics and multivariable models were used to examine the associations between race/ethnicity and contraceptive supply dispensed with adherence.
Results
Our cohort included 6946 women: 47% were white, 6% were Hispanic, 22% were black, and 25% were other race or had missing race information. Most women (83%) received a 3 month supply of contraception at each fill. More than 64% of women had at least 1 gap in coverage of 7 days or longer. Only 22% of women received a full 12 months of contraception without any gaps (perfect adherence). Compared with whites, Hispanics were significantly more likely to experience gaps (64% vs 70%; P = .02), and Hispanics and blacks received fewer months of contraceptive coverage (9.3 vs 8.9 and 9.0, P < .001). Compared with women receiving 3 month supplies, those receiving 1 month supplies had a higher likelihood of a gap (63% vs 72%, P < .001), fewer months of coverage (9.3 vs 6.9, P < .001), and a lower likelihood of perfect adherence (22% vs 11%, P < .001).
Conclusion
Adherence to hormonal contraception among women veterans is poor. Efforts to improve contraceptive adherence and lower risk of unintended pregnancy are needed; dispensing more months of supply for hormonal contraception may be a promising strategy.
See Journal Club, page 153
Unintended pregnancies account for nearly 50% of all pregnancies among US women and confer potentially significant adverse consequences for women, their children, and society. Unintended pregnancy can be averted by the consistent use of effective contraception. Most US women use a method of contraception with adherence requirements, such as oral contraceptives, and consistent use of these contraceptives is challenging. In fact, the disparity in effectiveness between perfect use and typical use of hormonal contraceptives reflects the high rates of nonadherence and discontinuation. In one study using a nationally representative sample of women at risk for unintended pregnancy, 30% of women had a lapse in contraceptive coverage for at least 1 month over the course of a year. Furthermore, racial and ethnic minorities in the United States demonstrate higher rates of inconsistent contraceptive use than white women as well as higher rates of unintended pregnancy and abortion.
As the number of women of reproductive age continues to increase in the Department of Veterans Affairs (VA) Healthcare System, reproductive health care including contraceptive provision has become a growing priority, yet there is a dearth of published data about contraceptive use patterns and unintended pregnancy among women veterans.
The VA pharmacy program has a number of unique policies that may enhance medication adherence, including mail-out prescriptions, various methods for refilling prescriptions (online, phone, mail-in refill slips), up to 90 day dispensing, fixed copays of $9 for all medications, and no co-pays for veterans who have been recently discharged from the service in Afghanistan or Iraq (Operation Enduring Freedom [OEF] and Operation Iraqi Freedom [OIF], respectively) or who are disabled by an injury or illness that was incurred during active military service (ie, service connection). In contrast, nearly 50% of women in the United States get only a 1 month supply of their prescription contraception, and most pay substantial out-of-pocket costs ($16 per pack of pills on average).
Recent studies have examined the impact of number of contraceptive packs dispensed on method continuation, and 13 month dispensing is associated with not only greater continuation rates but also lower pregnancy and abortion rates at 1 year.
Little is known about contraceptive use and adherence in the VA Healthcare System and whether observed racial differences in adherence in the larger population are also seen within VA. In a previous analysis, we examined the overall prevalence of contraception among the 103,950 women veterans of reproductive age who used VA for primary care in fiscal year (FY) 2008. In the current analysis, we assessed contraceptive adherence among those women who were using hormonal contraception (pill, patch, ring, and injectable) and also examined the relationships between race/ethnicity and months of contraceptive supply dispensed with contraceptive adherence.
Materials and Methods
Data sources
We conducted a retrospective analysis of data from national VA administrative databases including the Pharmacy Benefits Management (PBM) Database and the Medical SAS Datasets. VA uses a universally applied electronic medical record that captures pharmacy and clinical data on all care performed within VA. The PBM database is the national electronic system for tracking medications within VA and was used to obtain information on all prescription contraceptives in VA. The Medical SAS Datasets are the central repository for VA inpatient and outpatient clinical data and were used to obtain information on patient demographic information; mental health and medical diagnoses according to the International Classification of Diseases , ninth revision; patient utilization of VA clinics; receipt of contraceptive procedures (ie, intrauterine device [IUD] and implant placement and sterilization); and facility-level information.
Study sample
The study cohort included all women veterans between the ages of 18 and 45 years who made at least 1 visit to a VA primary care clinic (PCC) or women’s health clinic (WHC) during our study time frame (FY 2008: Oct. 1, 2007, to Sept. 30, 2008) and who had hormonal contraceptive (pills, patch, vaginal ring, or injection) coverage during the first week of the study time frame. Women who were sterilized (ie, had a hysterectomy or tubal sterilization procedure) or also used a long-acting, highly-effective reversible contraceptive method (ie, IUD or implant) in FY 2008 were excluded from analysis. We limited our sample to those women who had had at least one outpatient primary care visit to help discern those VA enrollees who use VA for primary care from those who do not.
Outcome variables
We constructed several patient-level indicators of contraceptive adherence over FY 2008. These indicators included the time between refills and the total months of contraceptive coverage and whether the woman had contraceptive coverage during the last week of FY 2008. For time between refills, we included only women who had filled at least 1 more prescription during the study time frame. Consistent with another recently published study examining contraceptive prescription adherence rates in a non-VA setting, we considered a gap of 7 days or more between refills as a marker of nonadherence. Granting a grace period of 7 days allowed for some variability in the initial date of starting the prescription. We examined whether a woman had any gaps of 7 days or longer between refills, how many gaps she had, and the time to the first gap. For the outcome of having contraceptive coverage during the last week of FY 2008, we distinguished between coverage with gaps and continuous coverage over the year (perfect adherence).
Independent variables
Our primary predictors of interest were race/ethnicity and months of contraceptive supply dispensed. Race/ethnicity, as captured in the administrative patient record, was categorized as non-Hispanic white, Hispanic, non-Hispanic black, non-Hispanic other, and unknown/missing. In general, VA prescriptions may be given as a 1, 2, or 3 month supply at 1 time, as per the clinician discretion. In this analysis, a 3 month supply indicated that the woman received a 3 month supply of contraception at each fill over the course of the year. Likewise, a 1 month supply indicated that the woman received only a 1 month supply at each fill. Anyone who received a 2 month supply or received varied months of supply over the course of the year was categorized as receiving other supply.
We examined as covariates other patient demographic information (eg, age, marital status, income, religion, non-VA insurance status, and OEF/OIF status) as well as clinical variables including presence of mental illness (depression, bipolar disorder, schizophrenia, and posttraumatic stress disorder); drug or alcohol abuse or dependency; and medical conditions that could influence use of estrogen-containing contraception (hypertension, history of venous thromboembolic disease, coronary artery disease, breast cancer, stroke, tobacco use among women aged 35 years old or older, and diabetes with complications).
We also examined whether the veteran received primary care in a WHC, a traditional PCC, or both in FY 2008; the number of outpatient primary care visits (to either a PCC or WHC); whether the veteran had a VA gynecology visit; the clinic site’s geographic region; and whether the site was a hospital-based or community-based clinic. The type of contraceptive method used (pill only, patch only, ring only, injectable only, or switched methods over the year) as well as whether the woman was an established contraceptive user (defined as having received 1 or more prescriptions for contraception in the 6 months preceding the study time frame) were also examined as covariates. Contraceptive pills included both combined estrogen-progestin pills as well as progestin-only pills.
Statistical analysis
Patient demographic and clinical variables, utilization of VA care, VA facility information, and the type of contraceptive method were described overall as well as by race/ethnicity and months of supply dispensed. χ 2 tests were used to compare these categorical variables across race/ethnicity and months of supply status. We summarized the indicators of contraceptive adherence (gaps of 7 days or longer, total months of contraceptive supply, and contraceptive coverage during the last week of FY 2008) and compared the differences across the key independent variables (race/ethnicity and months of supply dispensed). We calculated the number of gaps per person-year on contraception to account for differential time on contraception. Differences between groups were tested using Wilcoxon scores (rank sums) tests for the continuous indicators and χ 2 tests for the categorical indicators.
We also examined the length of all the gaps experienced (7-30 days, 31-90 days, and more than 90 days); group differences in the length of gaps were tested using a random-effect ordinal logistic regression model to account for women with more than 1 gap. We used Kaplan-Meier survival curves to illustrate time to first gap of 7 or more days between consecutive refills, with between-group differences tested using the log-rank test. Women were censored on the last day of their prescription supply in the study time frame. A Cox regression model was used to examine the adjusted hazard ratios of experiencing the first gap while controlling for all covariates.
We also used a random-effects logistic regression model to examine the adjusted odds ratios for perfect adherence (having continuous contraceptive coverage for the whole year). All models included a random effect for facility to account for the nonindependence of outcomes from women seen at the same facility.
We recognize that the use of the injectable may be associated with different adherence patterns because it inherently provides 3 months of contraceptive protection and also because clinics may have reminder systems in place to help women return for timely repeat injections. However, the injectable still requires adherence on the patient’s part and is also used more commonly by black women.
Because we were interested in examining the relationship between race/ethnicity and adherence, we did not want to exclude any user-dependent methods that are differentially used by race/ethnicity. Therefore, women using the injectable were included in the main analyses, and type of method was examined as a covariate. We did, however, conduct sensitivity analyses in which we excluded women using the injectable from the multivariable models. In all analyses, we used a separate category for missing or unknown race/ethnicity. To assess the impact of missing race/ethnicity on the model estimates, we conducted a sensitivity analysis using multiple imputation. The results of both of these sensitivity analyses (excluding the injectable and multiple imputation of missing race) were similar to the results from the main analysis (all model estimates changed less than 5% with 1 exception of 7%, data not shown). As such, we report only the results from our main analysis below.
All analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC) and Stata 12 (StataCorp, College Station, TX). This study was approved by the VA Pittsburgh Institutional Review Board.
Results
Our study cohort included the 6946 women veterans who had contraceptive coverage during the first week of FY 2008, had at least 1 primary clinic visit, had not been sterilized, and did not use a highly effective reversible contraceptive method (ie, IUD or implant) during the year. In the total cohort, 46.9% of the women were white, 5.7% were Hispanic, 22.2% were black, 3.1% were of other race, and 22.1% had missing race information. Most women (83.5%) received a 3 month supply at each fill, 3.6% received a 1 month supply, and 12.9% received other supply (0.73% received a 2 month supply and 12.2% received varied months of supply over the course of the year). Months of contraceptive supply did not differ by race/ethnicity ( Table 1 ).
| Characteristic | Total (n = 6946) | White (46.9%) | Hispanic (5.7%) | Black (22.2%) | Other (3.1%) | Unknown/missing (22.1%) | 1-month supply (3.6%) | 3-month supply (83.5%) | Other supply (12.9%) |
|---|---|---|---|---|---|---|---|---|---|
| Age, y | |||||||||
| 18-25 | 18.2 | 21.0 | 27.4 | 14.6 | 19.3 | 13.2 | 21.5 | 18.0 | 18.6 |
| 26-35 | 54.0 | 54.8 | 55.8 | 51.2 | 60.8 | 53.7 | 57.0 | 53.4 | 57.1 |
| 36-45 | 27.8 | 24.2 | 16.8 | 34.2 | 19.8 | 33.1 | 21.5 | 28.7 | 24.3 |
| Marital status | |||||||||
| Married | 26.1 | 28.1 | 25.1 | 20.3 | 25.5 | 27.9 | 28.3 | 26.1 | 25.1 |
| Divorced/separated/widowed | 25.8 | 27.1 | 22.9 | 24.4 | 26.9 | 24.9 | 26.7 | 25.4 | 27.7 |
| Never married | 47.2 | 44.5 | 51.3 | 53.9 | 47.2 | 45.3 | 43.4 | 47.5 | 46.4 |
| Unknown | 1.0 | 0.3 | 0.8 | 1.4 | 0.5 | 2.0 | 1.6 | 1.0 | 0.8 |
| Adjusted individual annual income >$20,000 | 35.1 | 36.4 | 34.7 | 34.2 | 36.3 | 33.2 | 32.7 | 35.3 | 34.7 |
| Mental illness a | 37.8 | 43.0 | 37.9 | 32.9 | 36.3 | 31.6 | 41.4 | 36.8 | 42.9 |
| Drug/alcohol abuse | 4.5 | 6.0 | 4.8 | 2.5 | 4.2 | 3.3 | 4.4 | 4.5 | 4.8 |
| Comorbidity b | 16.7 | 14.9 | 10.8 | 24.2 | 11.3 | 14.9 | 17.5 | 16.8 | 15.6 |
| Primary care clinic type | |||||||||
| Seen in PCC only | 36.8 | 41.3 | 30.7 | 27.2 | 33.0 | 38.9 | 56.6 | 36.2 | 35.1 |
| Seen in WHC only | 19.5 | 16.1 | 20.9 | 24.5 | 20.3 | 21.3 | 12.0 | 20.2 | 17.3 |
| Seen in both | 43.7 | 42.5 | 48.5 | 48.3 | 46.7 | 39.7 | 31.5 | 43.6 | 47.7 |
| Number of visits | |||||||||
| 1 | 17.7 | 17.4 | 14.3 | 14.5 | 21.7 | 21.9 | 23.5 | 17.7 | 16.3 |
| 2 | 19.4 | 19.3 | 20.6 | 17.3 | 13.2 | 22.1 | 17.1 | 19.4 | 19.7 |
| 3 | 16.3 | 16.8 | 15.6 | 15.3 | 11.8 | 16.9 | 23.1 | 16.1 | 15.5 |
| 4 | 11.8 | 12.3 | 9.0 | 12.6 | 14.2 | 10.3 | 10.4 | 11.7 | 12.4 |
| 5 | 9.7 | 9.3 | 9.0 | 10.7 | 12.7 | 9.1 | 4.8 | 9.8 | 10.0 |
| ≥6 | 25.2 | 24.9 | 31.4 | 29.6 | 26.4 | 19.7 | 21.1 | 25.2 | 26.2 |
| Seen in gynecology clinic | 26.3 | 24.8 | 26.9 | 32.0 | 26.9 | 23.5 | 27.1 | 25.4 | 31.7 |
| Religion | |||||||||
| Protestant | 46.0 | 40.2 | 17.8 | 69.6 | 37.3 | 43.1 | 45.8 | 46.3 | 44.3 |
| Catholic | 19.3 | 20.7 | 55.0 | 5.6 | 23.6 | 20.4 | 21.1 | 18.9 | 21.8 |
| Other | 8.7 | 9.5 | 6.5 | 6.7 | 11.8 | 8.9 | 7.6 | 8.7 | 8.5 |
| Unaffiliated | 26.0 | 29.6 | 20.6 | 18.1 | 27.4 | 27.6 | 25.5 | 26.1 | 25.4 |
| Has non-VA insurance | 31.2 | 32.2 | 22.9 | 32.2 | 23.1 | 31.3 | 31.1 | 31.7 | 27.6 |
| Hospital-based clinic | 92.2 | 90.4 | 96.7 | 96.4 | 93.4 | 90.7 | 90.4 | 92.1 | 93.9 |
| Geographic region c | |||||||||
| South | 47.5 | 35.9 | 40.5 | 73.1 | 38.2 | 49.4 | 41.4 | 47.9 | 46.4 |
| Midwest | 20.3 | 27.5 | 8.3 | 13.5 | 13.7 | 15.9 | 27.5 | 20.3 | 18.2 |
| Northeast | 10.2 | 13.6 | 10.3 | 5.2 | 7.5 | 8.4 | 14.3 | 9.8 | 11.7 |
| West | 22.0 | 23.0 | 41.0 | 8.2 | 40.6 | 26.2 | 16.7 | 21.9 | 23.7 |
| OEF/OIF | 0.1 | 0.0 | 1.5 | 0.1 | 75.9 | 74.2 | 80.9 | 76.0 | 78.3 |
| Established contraceptive user | 76.3 | 77.8 | 73.1 | 72.7 | 75.0 | 77.6 | 84.1 | 75.6 | 78.4 |
| Contraceptive type d | |||||||||
| Pill only | 74.0 | 75.7 | 73.6 | 68.2 | 72.2 | 76.5 | 89.6 | 71.3 | 86.4 |
| Ring only | 3.4 | 3.6 | 3.5 | 3.2 | 4.2 | 3.2 | 3.2 | 3.7 | 1.7 |
| Patch only | 2.0 | 1.7 | 1.8 | 2.3 | 5.2 | 1.8 | 6.4 | 1.7 | 2.7 |
| Injectable only | 15.9 | 14.5 | 13.6 | 20.9 | 15.1 | 14.5 | 0.0 | 19.0 | 0.0 |
| Switched methods e | 4.8 | 4.5 | 7.5 | 5.4 | 3.3 | 4.0 | 0.8 | 4.2 | 9.2 |
| Race/ethnicity | |||||||||
| White | 46.9 | 55.8 | 46.4 | 47.8 | |||||
| Hispanic | 5.7 | 4.4 | 5.8 | 5.8 | |||||
| Black | 22.2 | 15.1 | 22.4 | 23.1 | |||||
| Other | 3.1 | 4.0 | 3.0 | 3.2 | |||||
| Missing race | 22.1 | 20.7 | 22.4 | 20.2 | |||||
| Months of supply | |||||||||
| 1 | 3.6 | 4.3 | 2.8 | 2.5 | 4.7 | 3.4 | |||
| 3 | 83.5 | 82.5 | 84.2 | 84.1 | 81.6 | 84.8 | |||
| Other | 12.9 | 13.2 | 13.1 | 13.4 | 13.7 | 11.8 | |||
a Mental illness includes depression, bipolar disorder, schizophrenia, or posttraumatic stress disorder
b Comorbidities include medical conditions that could impact the use of estrogen-containing methods: hypertension, history of venous thromboembolic disease, coronary artery disease, breast cancer, stroke, tobacco use among women aged 35 years old or older, and diabetes with complications
c Geographic region is based on US Census regions
d Pill prescriptions include progestin-only pills, which accounted for 2.8% of all pills
e Refers to use of more than 1 type of method over the year, including 2.3% of women who used pills and the injectable, 1.3% who used pills and the ring, and 1.2% with other combinations.
Of the 6946 women in our study cohort, 6761 (97.3%) filled at least 1 more prescription in the study time frame and comprised the subcohort for the patient-level gap indicators in Table 2 (any gap and median number of gaps). Overall, 64.2% of the women had at least 1 gap during the year. Compared with white women, Hispanics were significantly more likely to have a gap (63.7% vs 69.6%, P = .02) and had more gaps (median 1.1 vs 1.3 gaps per person-year on contraception, P < .001). Similarly, women who received 1 month supplies were significantly more likely than women who received 3 month supplies to have any gap (71.5% vs 63.2%, P = .008) and to have more gaps (median 2.2 vs 1.1 gaps per person-year on contraception, P < .001). Compared with whites, Hispanic and blacks were more likely to experience gaps lasting more than 90 days (11.5% vs 14.4% and 16.6%, P < .001).
| Variable | Total (n = 6946) | White (47%) | Hispanic (6%) | Black (22%) | P value | 1-month supply (4%) | 3-month supply (83%) | Other supply (13%) | P value |
|---|---|---|---|---|---|---|---|---|---|
| Any gap between refills, % of women (n = 6761) | 64.2 | 63.7 | 69.6 | 63.4 | .06 | 71.5 | 63.2 | 68.8 | < .001 |
| Number of gaps per person-year on contraceptives, median (interquartile range) | 1.1 (2.2) | 1.1 (2.2) | 1.3(2.5) | 1.1 (2.2) | < .001 | 2.2 (5.6) | 1.1 (2.2) | 1.1 (2.4) | < .001 |
| Length of gaps (n = 5900 gaps), d | < .001 | < .001 | |||||||
| 7-30 | 59.3 | 61.4 | 54.8 | 55.4 | 70.1 | 58.1 | 61.6 | ||
| 31-90 | 27.9 | 27.2 | 30.9 | 28.0 | 23.9 | 28.2 | 28.0 | ||
| >90 | 12.8 | 11.5 | 14.4 | 16.6 | 6.0 | 13.7 | 10.4 | ||
| Months covered by contraceptives (% of women) | < .001 | < .001 | |||||||
| 1-3 | 9.4 | 8.4 | 9.5 | 12.0 | 32.7 | 8.7 | 7.0 | ||
| 4-6 | 14.4 | 14.2 | 18.8 | 14.7 | 15.1 | 15.0 | 10.6 | ||
| 7-9 | 21.0 | 21.1 | 23.1 | 21.3 | 15.9 | 21.1 | 21.8 | ||
| 10-12 | 55.2 | 56.3 | 48.5 | 52.0 | 36.3 | 55.2 | 60.6 | ||
| Contraceptive supplies during the last week of FY 2008 (% of women) | .28 | < .001 | |||||||
| No | 34.4 | 34.8 | 35.2 | 36.0 | 57.0 | 34.5 | 27.6 | ||
| Yes, with gap in coverage | 43.8 | 43.6 | 47.0 | 41.9 | 32.3 | 43.3 | 50.7 | ||
| Yes, no gap (perfect adherence) | 21.7 | 21.5 | 17.8 | 22.1 | 10.8 | 22.2 | 21.7 |
Women had a mean of 9.2 months of contraceptive coverage over the year, with Hispanics and blacks having fewer months of coverage than whites (8.9 and 9.0 months vs 9.3 months, respectively, P < .001). Women who received 1 month contraceptive supplies had significantly fewer months of coverage than those who received 3 month supplies (6.9 vs 9.3 months; P < .001).
Of the full cohort, 65.6% had contraceptive coverage the last week of the fiscal year, but the majority of these women (66.9%) had experienced 1 or more gaps over that time period. Only 21.7% of women received a full 12 months of contraception without any gaps between refills (perfect adherence). In bivariate analysis, there were no racial/ethnic differences with regard to the likelihood of having contraceptives on hand during the last week of FY 2008. Compared with women who received 3 month supplies, women who received only 1 month supplies were significantly less likely to be on a method the last week of the fiscal year (65.5% vs 43.0%, P < .001) and to have achieved perfect adherence than those who received 3 month supplies (22.2% vs 10.8%, P < .001).
Kaplan-Meier curves demonstrating the time to the first gap of 7 or more days between refills showed that Hispanics experienced a shorter time to first gap compared with white and black women ( P = .02; Figure 1 ). Women who received 1 month supplies also experienced a shorter time to first gap than women who received 3 month supplies ( P < .001). In multivariable Cox regression analysis, Hispanics remained significantly more likely to experience a gap than white women (hazard ratio [HR], 1.18, 95% confidence interval [CI], 1.03–1.34; Table 3 ). Women who received 1 month supplies also remained more likely to experience a gap than women who received 3 month supplies (HR, 1.62, 95% CI, 1.39–1.90).
