Background
Almost one-half of women having an abortion in the United States have had a previous procedure, which highlights a failure to provide adequate preventive care. Provision of intrauterine devices and implants, which have high upfront costs, can be uniquely challenging in the abortion care setting.
Objective
We conducted a study of a clinic-wide training intervention on long-acting reversible contraception and examined the effect of the intervention, insurance coverage, and funding policies on the use of long-acting contraceptives after an abortion.
Study Design
This subanalysis of a cluster, randomized trial examines data from the 648 patients who had undergone an abortion who were recruited from 17 reproductive health centers across the United States. The trial followed participants 18-25 years old who did not desire pregnancy for a year. We measured the effect of the intervention, health insurance, and funding policies on contraceptive outcomes, which included intrauterine device and implant counseling and selection at the abortion visit, with the use of logistic regression with generalized estimating equations for clustering. We used survival analysis to model the actual initiation of these methods over 1 year.
Results
Women who obtained abortion care at intervention sites were more likely to report intrauterine device and implant counseling (70% vs 41%; adjusted odds ratio, 3.83; 95% confidence interval, 2.37–6.19) and the selection of these methods (36% vs 21%; adjusted odds ratio, 2.11; 95% confidence interval, 1.39–3.21). However, the actual initiation of methods was similar between study arms (22/100 woman-years each; adjusted hazard ratio, 0.88; 95% confidence interval, 0.51–1.51). Health insurance and funding policies were important for the initiation of intrauterine devices and implants. Compared with uninsured women, those women with public health insurance had a far higher initiation rate (adjusted hazard ratio, 2.18; 95% confidence interval, 1.31–3.62). Women at sites that provide state Medicaid enrollees abortion coverage also had a higher initiation rate (adjusted hazard ratio, 1.73; 95% confidence interval, 1.04–2.88), as did those at sites with state mandates for private health insurance to cover contraception (adjusted hazard ratio, 1.80; 95% confidence interval, 1.06–3.07). Few of the women with private insurance used it to pay for the abortion (28%), but those who did initiated long-acting contraceptive methods at almost twice the rate as women who paid for it themselves or with donated funds (adjusted hazard ratio, 1.94; 95% confidence interval, 1.10–3.43).
Conclusions
The clinic-wide training increased long-acting reversible contraceptive counseling and selection but did not change initiation for abortion patients. Long-acting method use after abortion was associated strongly with funding. Restrictions on the coverage of abortion and contraceptives in abortion settings prevent the initiation of desired long-acting methods.
Almost one-half of the women in the United States who have an abortion have had a previous procedure, which highlights a failure to provide adequate preventive care. As is the case for all women wanting to prevent pregnancy, abortion patients stand to benefit from receiving information and access to a range of Food and Drug Administration–approved contraceptives that include long-acting reversible contraceptives (LARCs). Intrauterine devices (IUDs) and the subdermal implant are the most effective reversible contraceptives and are safe to initiate on the day of an aspiration abortion. LARC use is low in the United States compared with other developed countries, at approximately 7% of reproductive-aged women, which may contribute to the high unintended pregnancy rate.
Providing LARC methods in the abortion care setting has particular challenges. Although very cost-effective over time, LARC methods have high upfront costs. They can be unaffordable for women without health insurance or when devices or insertion fees are not fully covered. There are also financial barriers to offering contraception during an abortion visit in some settings, including strict regulations regarding Title X funding or prohibitions against the use of state family planning funds. Some abortion facilities face difficulties billing insurance for contraceptive services, given poorly defined coverage or need for preauthorization. Others face obstacles with LARC counseling, stocking, and placement because of resource shortages.
Although approximately two-thirds of US obstetrician-gynecologists agree that IUDs can be placed immediately after abortion, only 27% of those who offer abortions provide postabortion IUDs. For LARC methods to be offered after the abortion, provider knowledge about the methods and patient eligibility, as well as clinical training, are required. Lack of training can contribute to lower provision, as can patient misconceptions about IUD and implant safety and use after the abortion.
This analysis examines postabortion contraceptive care with data from a large cluster randomized trial with Planned Parenthood health centers across the United States. The trial evaluated the impact of a clinic-wide provider training about IUDs and implants on women’s access to the methods and unintended pregnancy. Primary analyses indicated that the intervention reduced pregnancy rates among women in family planning care by almost one-half; however, in the abortion care setting, high pregnancy rates persisted over the next year. This subanalysis assesses the role of health insurance and funding policies in access to postabortion LARC. Understanding coverage factors that impede contraceptive uptake can help identify policy changes and the interventions that are needed to support women’s reproductive health.
Materials and Methods
Study design and procedures
We conducted a cluster randomized trial of 40 Planned Parenthood health centers that served diverse, low-income women. Study details are described elsewhere. Briefly, eligible clinics had ≤20% LARC use, a patient volume of ≥400 annually, no ongoing LARC interventions or special funding programs, and no shared staff with other study clinics. The study randomly allocated clinics to receive LARC training (intervention, 20 clinics) or provide standard of care (control, 20 clinics) and concealed allocation until study initiation. Of the 40 participating sites, 23 sites recruited clients who were seeking general reproductive health services; the other 17 sites recruited women who were having abortions. This subanalysis uses data from the participants at the 17 sites located in a range of states (CA, CO, CT, FL, ID, MN, NC, OH, PA, WA) that provided abortion care.
Staff members at intervention clinics participated in a continuing medical education–accredited training session. The training emphasized updated LARC evidence, eligibility, counseling, and provision skills, which included same-day insertion when possible. The training included patient-centered counseling skills such as open-ended questions, reproductive life planning, ethical issues specific to LARC that included removal when desired, and incorporation of the World Health Organization tiered contraceptive effectiveness chart. Clinicians received hands-on IUD training with models, and we facilitated implant trainings with the manufacturer. All sites maintained usual costs for contraceptives.
After the training at intervention sites, we recruited a cohort of women from all study clinics between May 2011 and March 2012 and followed participants for 1 year. Eligible women were 18-25 years old, sexually active, not desiring pregnancy within a year, and were receiving contraceptive counseling. Women at the 17 abortion care sites were eligible to enroll on the day of an aspiration abortion or when mifepristone medication abortion was initiated. After providing informed consent and completing the enrollment visit, participants filled out a self-administered questionnaire that covered sociodemographics, pregnancy attitudes, contraceptive history, and methods discussed and selected at the visit. Providers completed a visit summary that documented abortion type, gestational age, and sources of payment for abortion.
Participants who underwent phone or online follow-up questionnaires quarterly for 1 year received $20 remuneration for each questionnaire. Investigators conducted medical record reviews at the end of 1 year. Clinic managers at each site completed surveys at baseline and study completion regarding clinic abortion and contraceptive care practices. The University of California, San Francisco, Committee on Human Research and the Allendale Investigational Review Board approved the study.
Measures
Outcomes
We assessed 3 outcomes to capture a woman’s access to LARC. To measure LARC counseling, we used a question on the baseline participant survey about whether a counselor, nurse, or doctor had discussed the IUD or implant during the abortion visit. We asked participants which method of birth control they decided to use at the visit or in the last week, if any, and created a dichotomous variable for selecting whether to use a LARC method. Finally, we assessed the actual initiation of a LARC method over 1 year using follow-up surveys and medical records to document IUD and implant insertions. We also used 5 questions about LARC effectiveness and traits to create a knowledge scale (range, 0-5, α = .68).
Patient funding
Participants reported their health insurance type (public [Medicaid, other state program], private, no insurance, don’t know). The visit summary indicated payment sources for the abortion (state Medicaid, private insurance, self or donated funds).
Funding policies
Guttmacher Institute data were used to indicate whether the clinic was in a state with the following policies: state Medicaid covers abortion care; abortion facilities can receive state family planning funds; Medicaid family planning expansion program exists; and private health insurance is mandated to cover contraceptives. Policy data aligned with dates of participant contact. We also examined data from the clinic manager survey on whether the site provided immediate postaspiration abortion LARC. Finally, to address the possibility that policy associations with LARC outcomes were not merely due to social climate around contraception and abortion at the site, we assessed 2 funding variables that we hypothesized would not be associated with LARC use. We included a measure of whether the site provided reduced-cost contraceptive care through the Title X family planning program, which is regulated strictly in the abortion setting, from the manager survey. We used Guttmacher Institute data to indicate whether the site was in a state with a mandated waiting period before abortion.
Clinic intervention and control variables
All analytic models included the study arm (intervention, control). The following control variables were selected a priori as being associated with LARC counseling and use: age (in years), self-reported race/ethnicity (white, black, Hispanic, or other), parity (nulliparous, parous), and LARC/hormonal contraceptive use in the 3 months before enrollment. We also assessed abortion type (aspiration, medication) and attitudes about pregnancy within a year (very unhappy/unhappy; happy/very happy).
Analysis
The analysis population included participants who were enrolled into the trial (n = 648) from sites that provided abortion care (N = 17). Intent-to-treat analyses were conducted, and the outcomes assessor (C.H.R.) was blinded to study arm. We assessed differences in participant characteristics by arm using regression with generalized estimated equations to account for clustering, with robust standard errors. The model link depended on the measure of the characteristic (eg, a logit link was used for dichotomously coded characteristics).
To estimate the effect of the training intervention on LARC counseling and selection, we used logistic regression with generalized estimated equations. We repeated analyses including health insurance and control variables. For LARC initiation, we used life-table analysis to estimate rates and Cox proportional hazards models with shared frailty for clustering to examine associations with time to LARC insertion. Women contributed observation time to the analysis until they initiated LARC, became pregnant, or exited the study. Initiation analyses excluded 1 participant who had an implant before the abortion. We estimated Schoenfeld residuals to check proportionality assumptions.
We individually introduced each funding variable into models for each LARC outcome. We fit a separate model for each variable because of correlation between them. We assessed interactions between funding policy and intervention to determine whether the effect of the provider training on participant LARC outcome differed by policy environment.
To estimate the increase in the proportion of women who selected LARC that would have been able to initiate it if, hypothetically, funding policies were universal, we used a population intervention model approach to calculate the causal attributable risk. This approach produced a marginal, causal effect estimate of what LARC use would be in a counterfactual population with the same covariate structure as the study population.
Analyses were conducted with Stata 14 software (Stata Corporation, College Station, TX), with multiple imputation for missing data (<1% for any variable).
Results
In total, 648 women (intervention, 322; control, 326) enrolled in the study from 8 intervention and 9 control clinics that were providing abortion care ( Figure 1 ). Participants were on average 21.6 years old ( Table 1 ). Thirty-six percent of the participants had public insurance; 35% had private insurance, and 27% were uninsured. Twenty-seven percent of the women used state Medicaid to pay for the abortion; only 12% used private health insurance. Most women (62%) thus paid for the abortion procedure themselves or with donated funds. Characteristics did not differ by study arm other than the proportion that had medication abortions (intervention, 16%, vs control, 36%). No funding policy variables differed by study arm. Approximately three-quarters of both intervention and control clinics offered LARC initiation on the day of an aspiration abortion.
| Variable | Intervention | Control |
|---|---|---|
| Participant characteristic a | ||
| Mean age (n = 648), y ± SD | 21.5 ± 2.3 | 21.6 ± 2.0 |
| Race/ethnicity (n = 648), n (%) | ||
| White | 169 (52.5) | 182 (55.8) |
| Black | 68 (21.1) | 65 (19.9) |
| Latina | 71 (22.1) | 52 (16.0) |
| Other | 14 (4.4) | 27 (8.3) |
| Never married (n = 641), n (%) | 286 (90.2) | 281 (86.7) |
| Nulliparous (n = 641), n (%) | 189 (59.4) | 193 (59.8) |
| Abortion type (n = 644), n (%) | ||
| Aspiration | 229 (84.4) | 206 (63.8) |
| Medication | 49 (15.6) | 117 (36.2) |
| Most effective contraceptive used in previous 3 months (n = 643), n (%) | ||
| Condom/barrier or no method | 215 (67.6) | 219 (67.4) |
| Hormonal (short-acting/depot medroxyprogesterone acetate) | 96 (30.3) | 99 (30.5) |
| Long-acting reversible contraceptive (intrauterine device or implant) | 7 (2.2) | 7 (2.2) |
| Happiness if pregnant in next year (n = 639), n (%) | ||
| Unhappy or very unhappy | 269 (85.1) | 273 (84.5) |
| Happy or very happy | 47 (14.9) | 50 (15.5) |
| Health insurance (n = 641), n (%) | ||
| Public insurance | 117 (36.7) | 114 (35.4) |
| Private insurance | 112 (35.1) | 109 (33.9) |
| None | 83 (26.0) | 88 (27.3) |
| Don’t know | 7 (2.2) | 11 (3.4) |
| Source of abortion payment (n = 637), n (%) | ||
| State Medicaid | 68 (21.5) | 104 (32.4) |
| Private insurance | 33 (10.4) | 40 (12.5) |
| Self or donated funds | 215 (68.0) | 177 (55.1) |
| Funding policy and provision practice, n (%) b | ||
| Medicaid covers abortion care | 3 (37.5) | 5 (55.6) |
| Abortion providers may receive state family planning funds | 6 (75.0) | 7 (77.8) |
| Medicaid family planning expansion program is in place | 4 (50.0) | 7 (77.8) |
| Private health insurance is mandated to cover contraception | 6 (75.0) | 5 (55.6) |
| Provides reduced-cost contraceptive care through Title X | 4 (50.0) | 5 (55.6) |
| Mandated abortion waiting period is in place | 3 (37.5) | 3 (33.3) |
| Provides long-acting reversible contraception on the day of aspiration abortion | 6 (75.0) | 7 (77.8) |
a Intervention, 322 patients; control, 326 patients
Women at intervention clinics were far more likely than women at control clinics to report that contraceptive counseling included the IUD or implant (70% vs 41%; odds ratio, 3.32; 95% confidence interval [CI], 2.11–5.23). Knowledge about LARC methods was also higher among intervention participants (mean score, 2.6 vs 2.0 on 0-5 scale; β = .58; 95% CI, 0.23–0.93). Similarly, women at intervention sites were twice as likely to select a LARC method after the abortion (36% vs 21%; odds ratio, 2.04; 95% CI, 1.29–3.24). Most women who selected LARC were “very sure” (80%) or “sure” (12%) that they would use the method for 1 year. All women who selected LARC reported making the decision either themselves (89%) or together with the provider (11%). However, the actual initiation of LARC within the year was no different between arms (22/100 person years each; hazard ratio, 0.96; 95% CI, 0.53–1.74), with only 51% of women who selected LARC overall ever having placed. Intervention effects were similar in multivariable models; the training was associated with increased LARC counseling and selection, but it did not affect the actual initiation after the abortion ( Figure 2 ).
Patient health insurance was important for LARC selection and use ( Table 2 ; Figure 3 ). Compared with women with no health insurance, women with public insurance had higher LARC selection and initiation (adjusted hazard ratio [aHR] for initiation, 2.18; 95% CI, 1.31–3.62). Having private insurance, however, was not significant. Notably, low proportions of insured women used insurance to pay for their abortions; among women with public insurance, 58% used it to pay for the abortion. Among women with private insurance, 28% used their coverage. Women who used public or private insurance to pay for the abortion initiated LARC at about twice the rate as women who paid out-of-pocket or with donated funds (aHR public, 2.29; 95% CI, 1.44–3.61; aHR private, 1.94; 95% CI, 1.10–3.43).
