An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Cynthia H. Chuang, MD, MSc, Carol S. Weisman, PhD, Diana Velott, MPA, MS, Erik Lehman, MS, Vernon Chinchilli, PhD, Erika B. Francis, MS, Richard Legro, MD, Christopher Sciamanna, MD, MPH, Merry-K. Moos, BSN, FNP, MPH, and Christopher Armitage, PhD.
Author Information and AffiliationsUnintended pregnancies occur when no contraception is used, is used inconsistently, or contraceptive failure occurs. In August 2012, the Affordable Care Act (ACA) required most private insurers cover FDA-approved contraceptive methods without cost-sharing. Without cost barriers, privately insured women may be well positioned to respond to interventions designed to assist them with individualized contraceptive decision making. Potential interventions include reproductive life planning (RLP) and contraceptive action planning. Recommended by the Centers for Disease Control and Prevention (CDC) and the American College of Obstetricians and Gynecologists (ACOG), RLP involves setting goals for having or not having children and making a plan to achieve those goals. Contraceptive action planning guides users to identify solutions ahead of time for challenges commonly encountered when using contraception. Whether web-based RLP or contraceptive action planning affects, contraceptive use has not been evaluated in controlled trials.
The MyNewOptions study was a randomized trial designed to test whether web-based RLP alone or in combination with contraceptive action planning (RLP+) affects patient-centered contraceptive outcomes in insured women.
We invited insured women residing in Pennsylvania, 18 to 40 years of age, and not intending pregnancy in the next year to join MyNewOptions, a fully online study. After completing a baseline survey, women were randomized to 1 of 3 arms: RLP; RLP+; or an information-only control group. Women completed follow-up surveys and revisited the study website every 6 months during the 24-month study. The primary outcomes were proportion of the study period that women self-reported any contraceptive use, prescription contraceptive use, and high method adherence when at risk for unintended pregnancy, as well as the switch to a more-effective contraceptive method by the end of the study. Secondary outcomes included self-reported method satisfaction and contraceptive self-efficacy. We compared the longitudinal outcome measures by the study group using binomial logistic regression.
The MyNewOptions trial randomized 984 women between April and July 2014. The proportion of the study period that women in the control, RLP, or RLP+ group were at risk for unintended pregnancy and using any method (95%, 94%, and 95%, respectively; P = .15), a prescription method (64%, 61%, 65%, respectively; P = .03), or reported high adherence (76%, 73%, 72%, respectively; P = .99) did not differ by group allocation. The proportion of the study period in which women were highly satisfied with their method (58% control, 58% RLP, 57% RLP+; P = .86) and had high contraceptive self-efficacy also did not differ by group allocation. Switching to a more-effective method during the study also did not differ by study group (24% control, 26% RLP, 24% RLP+; P = .85). Contraceptive use and long-acting reversible contraceptive use increased significantly from 88.5% and 8.4% at baseline to 95.8% and 19.4% at 2 years, respectively, but we observed no differences by group allocation.
Our results suggest that a web-based reproductive life planning tool may not result in privately insured women using more-effective contraceptive methods.
The null findings may be due to lack of intervention intensity, the web-based intervention format, an unintended intervention effect of the control condition, or a ceiling effect in our well-educated, privately insured sample. We also observed no intervention effects in post hoc stratified analyses by age group (<26 years and ≥26 years).
In 2011, a total of 2.8 million unintended pregnancies occurred in the United States.1 Unintended pregnancies happen when no contraception is used, when it is not used continuously or perfectly, or when contraceptive failure occurs, which is more common with less-effective methods. Barriers to effective contraceptive use are multifactorial, attributable in part to cost, use of methods that are not well aligned with personal needs and preferences, and lack of effective strategies for improving adherence. Discontinuation and poor adherence commonly occur—a decision analysis estimated that 20% of the unintended pregnancies that occur each year in the United States are attributable to poor adherence or discontinuation of oral contraceptives.2 Discontinuation occurs when side effects arise that possibly could have been addressed with individualized contraceptive counseling aimed at strategies for managing side effects or assisting women with the selection of another method.3
Prior research has demonstrated that when contraception is provided at no cost and accompanied by contraceptive information or dedicated contraceptive counseling, women are more likely to choose more-effective and more-expensive methods over less-effective, less-expensive methods.4-9 The Affordable Care Act (ACA) required that most private insurers cover all FDA-approved contraceptive methods without cost-sharing in all new health plans effective after August 2012. Since most plans are updated and reissued with each calendar year, most plans were subject to the contraceptive coverage requirement as of January of 2013. By removing the financial barrier to contraception, the contraceptive coverage mandate allowed privately insured women to be well positioned to respond to interventions designed to encourage personalized contraceptive decision making. Although many contraceptive counseling tools have been developed and studied, they have usually been designed and tested in clinical settings (eg, women seeking contraceptive care in family planning clinics). The MyNewOptions study tested 2 interventions for assisting women with contraceptive choice and adherence: (1) reproductive life planning (RLP) and (2) RLP in combination with contraceptive action planning (RLP+). Early studies evaluated these interventions in clinical settings—we aimed to study these tools in an insured sample not currently seeking care.
Reproductive life planning is recommended by the CDC10 and the American College of Obstetricians and Gynecologists (ACOG).11 RLP is a Title X Family Planning Program Priority12 and is included as a component of the CDC and the Office of Population Affairs Quality Family Planning Guidelines.13 RLP involves setting goals for having or not having children and making a plan to achieve those goals. In clinical settings, face-to-face RLP-based counseling has been evaluated in several small studies that have shown increased knowledge among participants about reproductive life planning and contraception.14,15 We know of only one study that has evaluated whether RLP-based counseling affects contraceptive behavior. Bommaraju et al found that RLP counseling among non–pregnancy-seeking women using Title X services did not affect contraceptive use behavior.16 There have been no larger trials (and thus no systematic reviews) on whether RLP affects contraceptive behaviors or helps women achieve their reproductive goals. In the MyNewOptions study, we evaluated a web-based RLP intervention based on the CDC tool which we further enhanced to be interactive and personalized. Enhancements also provided individualized contraceptive information and basic assistance with seeking a contraceptive provider if needed.
Another promising strategy for contraceptive counseling is action planning. In a UK study by Martin and colleagues, contraceptive action planning for oral contraceptive and condom users was shown to improve contraceptive adherence and reduce unintended pregnancy.17,18 Action planning interventions guide individuals to make a specific plan—ie, what they will do when faced with common barriers that make it difficult to perform a desired behavior (ie, “If situation X is encountered, then I will initiate goal-directed behavior Y.”). By specifying when, where, and how one will act, action planning passes control of behavior to future environmental cues, reducing the need for cognitive control and effort.19,20 These interventions have been highly effective for losing weight,21-23 reducing cigarette smoking24,26 and alcohol intake,27-29 and increasing physical activity30,31 and fruit and vegetable consumption.23,32,33 Action planning interventions may be well suited for contraceptive adherence because effective contraceptive use often requires a series of behaviors that are repeatedly met with certain barriers (eg, obtaining a prescription, buying condoms, filling/refilling the prescription from a pharmacy, taking a pill every day, using a condom at each instance of intercourse).
The MyNewOptions study tested whether RLP alone or in combination with contraceptive action planning (RLP+) would change patient-centered contraceptive outcomes in insured women. The specific aims were the following:
The impact of these aims is widespread, as nearly 60% of adults under age 65 years have private health insurance.34
The MyNewOptions study included patients and clinicians in all phases of study design and execution. We gathered suggestions and feedback from 3 separate groups of stakeholders during the course of the study: (1) listening sessions, (2) a patient advisory group (PAG), and (3) a clinician advisory group (CAG). A lead patient partner was part of the research team, attended the monthly MyNewOptions research team meetings, and served as the coordinator of the PAG. All of the women in the listening sessions and the PAG were reproductive-age women who had private health insurance coverage, were sexually active, were not currently pregnant or trying to get pregnant in the next year, and had not had a hysterectomy or tubal ligation. They included women who had not had children as well as women who had completed childbearing.
During study proposal development, 12 privately insured women of reproductive age were recruited to form 2 listening groups comprising 6 members each. To recruit women meeting these specifications, the research team used word of mouth and snowball sampling, as well as posted flyers in our health system facility where they were visible to patients, staff, and students. Each listening session group met in person with the principal investigator for close to 2 hours. The purpose of the listening sessions was to learn how to adapt the CDC reproductive life plan tool into a patient-centered, interactive web-based format.
Women in the listening sessions were involved in the prestudy conceptual discussion only. Although their time involvement was limited, their input was critical to the design and execution of the study interventions in 3 specific ways. First, they participated in a candid discussion about their personal challenges with contraceptive adherence and offered concrete examples of how to make reproductive planning realistic for themselves. In particular, they were adamant that making a “life” plan was not practical but that making a 5-year plan was something they thought they could do. Second, the challenges to contraceptive adherence that the participants shared were added to the action intervention as barriers to contraceptive planning. Finally, women suggested strategies for reporting information summaries to the website user (research participant) after completing their intervention exercises online. We incorporated these suggestions into the protocol, allowing women to email and/or print a copy of their information directly from the website.
Once the study began, we hired a patient partner who became a member of the research team and attended monthly team meetings. The patient partner then worked to form a 12-member PAG consisting of reproductive-age women who mirrored the inclusion and exclusion criteria of research participants. The patient partner used a volunteer list maintained from previous research studies of women who had indicated interest in helping with future research studies. Following this step, the patient partner used word of mouth and a snowballing approach until the PAG was formed. In addition to serving as a patient partner on the study research team, she served as the PAG coordinator. The PAG met in person twice a year for the duration of the study (6 meetings). Each meeting lasted 2 hours.
The PAG provided patient stakeholder input during all stages of study design and execution. Their guidance was particularly evident during the intervention design work, the recruitment and retention work, and in discussions about findings and next steps. The PAG provided substantive input to the research team on intervention refinement. For example, they carefully reviewed and pilot tested prototype action plans for birth control pills, condoms, and natural family planning to ensure they were realistic, woman-centered, and viewed as helpful. Following their review and suggestions, we revised the action plans to include additional challenging scenarios that could occur with specific method use as well as additional solutions.
The involvement of our patient advisors contributed substantially to our recruitment and retention efforts. Working within very narrow constraints, the PAG advised the recruitment material design research team to optimize a woman-friendly feel to the materials. Because of their work, we used a cherry blossom to represent the study. We used this symbol as the theme of the website layout as well as on all recruitment material. The PAG also made suggestions about participant follow-up when surveys were missed. They cautioned us about too much contact with participants, and we included their guidance in our protocol and the procedures we followed.
At each of the PAG meetings, patients were updated on the status of the study to ensure transparency of the research process. Near the end of the study, the research team shared early analysis indicating that many study participants who experienced migraines with aura were using birth control pills containing estrogen, even though this was contraindicated. PAG members were adamant that the information be shared with research participants. We published this information in a peer reviewed journal33 and subsequently included it in a newsletter sent to all research participants. When the null results of the MyNewOptions study were shared with PAG members, they asked the research team to consider conducting the same, or a very similar study, in a population of uninsured or underinsured minority women. They expressed their belief that this was an excellent study with potential to help women who were not highly insured or might not be highly educated, as these groups may be less likely to have access to health education or health care providers. PAG members were pleased they had learned a great deal about contraception by advising this study. They expressed that many women did not have this information and should be given the opportunity to participate in a similar study.
In the second year of the study, a clinician advisory group was formed, at the invitation of the principal investigator, to provide clinical stakeholder input. Although we intentionally designed the study interventions to take place outside the clinical setting, we recognized that women seeking prescription methods ultimately need to see health care providers for additional counseling and prescriptions or procedures for contraception. Therefore, the group included physicians and midlevel providers who routinely provide primary and reproductive care for adult reproductive-age women. Clinical providers represented primary care (internal medicine and family medicine) and obstetrics and gynecology in both community and academic practices. The 10-member CAG met for a 2-hour meeting in each of the last 2 years of the project.
The CAG did not participate in intervention design or in recruitment and retention efforts because it was important that the materials reflect a patient-centered perspective. However, we informed members of the clinician group at every stage of the study in the interest of full transparency. The provider meetings did not result in any concrete effects, but group members provided thought-provoking opinions and suggestions that we will seriously consider in future work. Specifically, members of the CAG shared concerns that barriers to contraceptive adherence might exist that we had not included in our model, that a paper-based intervention process or an in-person consultation with a provider might be better than an online format, and that it was possible the research participants in the control group also received a form of intervention (ie, access to the ACA policy and a library of method choices) and might, therefore, have diluted our results.
The MyNewOptions study was a 2-year, randomized controlled, single-blind trial with a 3-arm parallel group design to compare the effectiveness of reproductive life planning alone, reproductive life planning plus contraceptive action planning, and an information-only control group in a sample of insured women. The randomized controlled trial study design provided high-quality evidence comparing the effectiveness of reproductive life planning interventions. We included a control group since it is unknown whether any reproductive life planning is beneficial in nonclinical settings. We collected data through self-report (ie, surveys). The IRB at the Penn State College of Medicine approved the study.
The MyNewOptions study was a randomized controlled trial intentionally designed to occur online and outside the clinical setting so it would be easy to disseminate if effective. The conceptual framework for this study, shown in Figure 1, identifies variables hypothesized to influence women's contraceptive choices and method adherence. Basic to reproductive life planning is consideration of pregnancy beliefs and attitudes—we considered pregnancy intentions regarding both the time frame in which the woman wished to avoid pregnancy and how she would have felt if she accidentally became pregnant. In measuring these constructs, we recognize that many women are ambivalent both about the time frame and about their feelings if they were to become pregnant; thus, we specifically included ambivalence in our measures. In addition, a woman-centered approach requires taking into account how a woman's pregnancy history (children she already has, prior unintended pregnancies) and current pregnancy risk exposure (including her current partner status) influence her desire for pregnancy avoidance and her contraceptive choices.
Influences on Contraceptive Outcomes.
In addition, this framework incorporates personal contraceptive requirements and preferences (eg, effectiveness, safety in relationship to personal health history, adverse effects, ease of use, interference with sexual pleasure, financial cost to the woman)36; women's preferences for different attributes, which are known to vary37; and information about these attributes that was provided in the intervention to show the benefits and risks for each method. Finally, we noted dimensions of health care access (health insurance coverage for contraception and perceived access to providers for provision of a method or removal of a method) because they could affect the impact of the intervention.
The target population was reproductive-age women with private health insurance and, thus, guaranteed no-cost contraceptive coverage under the ACA. We identified the sampling frame through Highmark Health, a large private health insurer based in Pennsylvania. Highmark identified members with prescription medication coverage who met the following criteria: women 18 to 40 years of age who resided in Pennsylvania; were not part of an employer group with a religious exemption from contraceptive coverage; and had no previous claim for tubal ligation, hysterectomy, or infertility-related service.
In order to achieve 90% power to detect a 10% difference in any contraceptive use between study groups using a 2-sided 0.025 significance level test (Bonferroni correction), we targeted a sample size of 972 (324 participants per study group). We assumed women in the intervention and control groups would use contraception during 75% and 65% of the 24-month follow-up period, respectively. We extrapolated these estimates from national statistics indicating that 86% of unsterilized women at risk of unintended pregnancy report contraceptive use in the past month.38 Based on a prior research study that recruited Highmark members for participation, we estimated that 15 women would need to be in the sampling frame to achieve 1 enrollee.39 Thus, we identified a random sample of 15 000 members who met the sampling frame criteria. Study invitations, which we mailed between April and July 2014, invited women to visit the MyNewOptions website to learn more about the study and to enroll. The website provided detailed information and directed interested women through the eligibility criteria and consent process. Inclusion criteria included having internet access, being sexually active with a male partner within the past 6 months or expected in the next 6 months, and not planning to get pregnant within the next 12 months. Exclusion criteria were tubal ligation, hysterectomy, or a current partner with a vasectomy and not able to read and write English.
Women identified as eligible on the website were directed to the baseline survey in the Research Electronic Data Capture (REDCap) database. After completing the baseline survey, REDCap linked participants back to the MyNewOptions website, where they created a unique user ID and password. Once women were authenticated by replying to a website-generated email, they were randomized using a permuted-block algorithm to achieve 1:1:1 allocation into 1 of the following groups: (1) RLP, (2) RLP+, or (3) information-only control. The study team was blinded to participant group allocation until the end of the trial. Participants then viewed different website content specific to their group allocation. Regardless of group assignment, all MyNewOptions participants began their website visit by viewing information about the ACA requirement for health plans to cover all FDA-approved contraceptive methods without cost-sharing. Regardless of group allocation, all participants ended their intervention at the library page, which contained information about different contraceptive methods. When women were finished reviewing the library, they chose their preferred gift card from options displayed on the screen. The website recorded the exact time gift cards were selected. All website material was written at a sixth- to seventh-grade reading level.
We adapted the MyNewOptions RLP intervention from the CDC's Reproductive Life Plan,40 which guides women to determine whether they are intending pregnancy based on their current goals for school, marriage/partnership, job/career, finances, and other important life circumstances. We further enhanced the MyNewOptions RLP to assist individual decision making, such as risks associated with older reproductive age (eg, decreased fertility, increased risk of chromosomal abnormalities) and benefits of pregnancy spacing. The RLP tool allowed women to enter answers in an online form, specifying if they planned to have children in the next 2 years, how many children they planned to have, when they planned on having them, or whether they were unsure. The RLP then assessed the woman's personal requirements for a contraceptive method—that is, reversibility, effectiveness, frequency of use, side effects that are or are not personally acceptable, medical conditions or health risks that may contraindicate a particular method, whether protection from sexually transmitted infections was needed, and whether noncontraceptive benefits of certain methods were important.36
Similar to the stoplight approach used by Garbers et al,41 the MyNewOptions RLP tool then displayed a summary of which contraceptive method(s) met all or most of the woman's requirements (green zone), some of her requirements (yellow zone), or none of her requirements (red zone) so she could identify the methods that best suited her contraceptive needs (or allowed her to confirm that her current method was an appropriate choice). After completing the RLP tool, the participant could print, save, or email herself her personal reproductive life plan, specifying her goals for future pregnancy, when she wanted to be pregnant, how far apart she wanted her children, and how she planned to prevent pregnancy until she was ready to be pregnant. If a woman decided to change her method because of the MyNewOptions RLP intervention, she was directed to contact her health care provider if the change required a prescription or procedure. If she did not have a current health care provider, the RLP intervention directed her to the Highmark member services site that would help find a provider in her area. Women's selections in the RLP tool were not available to the research team.
Women in the RLP group subsequently received an RLP booster email after completing the 6-, 12-, and 18-month surveys (no booster at 24 months). The RLP booster email instructed the participant to view and confirm her previous reproductive life plan or to develop a new plan.
The RLP+ online intervention was the MyNewOptions RLP intervention described above plus an additional contraceptive action planning step aimed at improving contraceptive adherence. The study team developed the MyNewOptions contraceptive action planning tool with the assistance of the MyNewOptions PAG.42 Each woman in the RLP+ group was directed to complete an action plan for the contraceptive method she was using (or an action plan for women using no method). The user was shown a list of situations that might challenge her ability to use her birth control method perfectly. For each situation, she was shown a list of possible strategies for managing the scenario and asked to pick one or to come up with her own strategy. Participants could view and complete action plans for other methods if desired.
Women assigned to the RLP+ group subsequently completed an RLP+ booster after the 6-, 12-, and 18-month surveys (there was no booster at 24 months). In the RLP+ booster, the participant viewed and confirmed her previous reproductive life plan, or she could develop a new plan. She also completed the contraception action plan again. Women still using the same birth control method were asked to complete an action plan for that same method, while women using a different birth control method were asked to complete an action plan for the new method.
Women in the control group (and in the other study groups) were able to view standard information about all FDA-approved contraceptive methods on the MyNewOptions website43 using patient education materials from ACOG,44 the National Campaign to Prevent Teen and Unplanned Pregnancy,45 and the Association of Reproductive Health Professionals.
Current contraceptive use was assessed at each survey and categorized into 4 groups representing highest to lowest effectiveness: (1) long-acting reversible contraceptives (LARCs) or sterilization, (2) other prescription methods (ie, oral contraceptives, injectable, patch, vaginal ring), (3) nonprescription methods (ie, barrier methods, spermicide, withdrawal, natural family planning), and (4) no method. The primary outcome variables were any contraceptive use, prescription contraceptive use, contraceptive use congruent with pregnancy intention, contraceptive adherence, and switch to a more-effective contraceptive method. We defined any contraceptive use, prescription contraceptive use, contraceptive use congruent with pregnancy intention, and contraceptive adherence as the proportion of the 24-month follow-up period in which a participant reported using any contraception, any prescription contraception, contraception congruent with pregnancy intention, and contraception with high method adherence, respectively. We included in the assessment of these longitudinal outcomes only follow-up periods in which a participant was at risk for unintended pregnancy (ie, periods when not intending pregnancy in the next year). We did not include in these outcome assessments women who reported intending pregnancy in the next 12 months at every follow-up survey (n = 17) or who had missing intention data at every follow-up survey (n = 49).
We created the variable contraceptive method congruent with pregnancy intention as a more patient-centered outcome describing whether a woman's contraceptive method was aligned with her future pregnancy intentions, taking into account whether she intended a future pregnancy, during what time frame (in 1-2 years, in 2-5 years, in ≥5 years), if she was ambivalent (not sure if she intended future pregnancy), and how she would feel if she accidently became pregnant based on her response to the question, “Imagine that you have just found out today that you are pregnant. How would you feel?” The response categories were very upset, a little upset, a little pleased, very pleased, or not sure. We coded women who were intending a future pregnancy but not in the next 12 months AND responded they would be very upset or a little upset if they became pregnant today as having intention-congruent contraceptive use if they were using a LARC or other prescription method; otherwise we considered them intention incongruent. We considered women who indicated they would be a little pleased, very pleased, or not sure how they would feel to be using a method congruent with their pregnancy intentions.
The contraceptive adherence outcome measure was determined by self-report and categorized as high or low at each survey based on the method of contraception used. We automatically defined all sterilized women, women whose current partner had a vasectomy, and LARC users as having high adherence. For oral contraceptive users, we defined high adherence as missing no more than 1 pill per month, based on the World Health Organization and the CDC recommendation that missing >1 active pill per cycle increases the risk of pregnancy.46,47 For injectables, patches, and vaginal rings, we considered women to have low adherence if they were at least 1 week late for their last injection or at least 1 day late with applying/inserting 1 or more patches/rings in the past 3 months. For coitus-dependent methods (ie, condoms, spermicides, withdrawal, sponges, diaphragms, fertility awareness methods), we considered women to have high adherence if they had used their method with every sexual encounter in the past 6 months. We defined the contraceptive adherence outcome as the proportion of the follow-up period during which the participant reported high adherence.
Switch to more-effective method was a dichotomous (yes/no) outcome indicating whether the participant switched to a method in a more-effective category at any follow-up survey than the method she was using at baseline (LARC users at baseline were not included in this outcome assessment).
Secondary outcomes included satisfaction with contraceptive method and contraceptive self-efficacy. At each survey, women who reported using a birth control method were asked their overall method satisfaction on a 5-point Likert scale, ranging from very satisfied to strongly dissatisfied. We defined contraceptive satisfaction as the proportion of follow-up surveys in which a participant indicated being satisfied or very satisfied with her current method.
A novel measure of contraceptive self-efficacy assessed confidence in one's ability to obtain and use birth control correctly throughout the 2-year study. Participants agreed or disagreed on a 5-point Likert scale with 8 statements, including “I am confident in my ability to use birth control correctly” and “I am embarrassed to talk about birth control with my doctor or health care provider.” The measure of self-efficacy is a summated rating with possible scores ranging from 8 to 40 and dichotomized at the median score of 34 to define high vs low contraceptive self-efficacy. We defined contraceptive self-efficacy as the number of follow-up periods in which a participant reported high contraceptive self-efficacy.
The participants' only required interactions with the study were through the study website. Study staff were available by phone or email if questions or concerns arose. We intentionally chose to have the study and interventions occur outside the clinical setting so it would be easy to disseminate if effective.
The study interventions and data collection occurred during a 2-year period. We chose this time frame to allow adequate time to observe any contraceptive changes that women made over time.
Baseline and follow-up surveys were completed using REDCap, a secure, web-based application designed exclusively to support surveys for research studies.48 Using standard survey measures, the baseline survey collected information about sexual and contraceptive history, pregnancy history, pregnancy intentions,49,50 current contraceptive use and adherence,51,52 relationship history,53,54 and sociodemographic49; it took approximately 20 minutes to complete.
An email was sent to participants at 6, 12, 18, and 24 months, directing them to complete the follow-up surveys in REDCap by clicking on the link included in the email. Surveys measured any interval pregnancies, current pregnancy intentions, current contraceptive use, and current contraceptive adherence. Women who reported that they had become pregnant or had a hysterectomy since the prior survey were not eligible to complete subsequent surveys since they were no longer capable of pregnancy. Following the completion of each follow-up survey, all participants were directed to complete boosters of their respective conditions, according to their randomization allocation. We used reminder calls and emails to increase retention. Participants received a $25 gift card incentive for completing the baseline survey and each of the 4 follow-up surveys, for a total of $125 for perfect retention throughout the 24-month study.
We collected no data from the website other than user IDs, timestamps, and gift card selections. We estimated the time to complete the website intervention to be the elapsed time between baseline survey completion (REDCap completion timestamp) and baseline gift card selection (website timestamp for gift card selection). Women were not able to select their gift card until they had finished their baseline survey, were randomized, and had completed their intervention process.
The baseline sociodemographic, pregnancy history, and contraceptive characteristics of the study sample were presented and compared by study group using chi-square tests. We anticipated that 10% of the enrolled sample would be completely lost to follow-up per year (ie, not complete any additional follow-up surveys) while other women might skip surveys. We used multiple imputation using baseline data variables to handle missing data in both cases, and used sensitivity analyses to compare main study results using only collected data with imputed data.
The longitudinal outcome measures were compared by study group using binomial logistic regression. We compared the change from baseline to 24 months in contraceptive self-efficacy score among study groups using a fixed-effects model including an adjustment for the baseline measure. We conducted multivariable analysis of each study outcome using binomial logistic regression analysis or generalized estimating equation (GEE). Subgroup analyses to evaluate heterogeneity of treatment effect were not prespecified; however, post hoc analyses evaluated the main effect by age group (<26 years and ≥26 years).
We compared the contraceptive method categories used by study participants for the entire sample and by study group at baseline and 24 months using a multinomial or binomial GEE, depending on the composition of the outcome variable, and included factors for survey (baseline vs 24 months), study group (control, RLP, RLP+), and interaction between survey and study group.
The full study protocol is presented in Appendix A, and an abbreviated version has been published.42 The original proposal and study protocol specified that one of the main study outcomes (aim 2) was continuity of contraception. This was based on prior research indicating that contraceptive method switches were associated with gaps in contraceptive use and increased risk of unintended pregnancy.55-57 However, we designed the MyNewOptions interventions to assist women in identifying contraceptive methods best suited to their individual needs and requirements. If that method(s) was different from what she was using then a method switch would have been appropriate; thus, we determined that method continuity was not an appropriate outcome for the MyNewOptions study. We created an alternative outcome measure—contraceptive method congruent with pregnancy intention—which was designed to describe whether a woman's contraceptive method was aligned with her future pregnancy intentions, as described above.
An overview of enrollment, randomization, and follow-up is shown in Figure 2.
MyNewOptions Enrollment, Randomization and Survey Completion.
A total of 15 000 study invitations, mailed between April and July 2014, invited women to visit the MyNewOptions website to learn more and enroll. The website provided detailed information about the study and directed interested women through the eligibility criteria and consent process. There were 2011 women who accessed the study website; women were excluded if they did not complete or meet the inclusion criteria (n = 863), did not complete the baseline survey (n = 137), did not proceed to the randomization step (n = 20), or were ineligible based on their baseline survey responses (n = 7), leaving 984 women randomized to the MyNewOptions study. We attained the target sample size after 4 months of enrollment, after which the website did not allow new visitors. Additional details of the sampling, enrollment, and consent process have previously been described.42
Compared with the women in the sampling frame who were not enrolled in the study, randomized women were statistically younger (mean age 27.1 vs 28.0 years; P < .0001) but representative in terms of the Pennsylvania region and rural/urban residence.42
We compared the time spent interacting with the MyNewOptions website by study group. The amount of time between completing the baseline survey and selecting the baseline gift card on the website (“web visit”) was categorized as <5 minutes, 5 to 20 minutes, or 21 to 60 minutes. Of participants, 10% spent more than an hour on the website following survey completion and were not included in the following calculations as sometimes hours, days, or even weeks had elapsed before the process was completed, which we assumed represented women who may have left their computers and returned to finish later. About one-third of participants (34% [n = 335]) completed their web visit in less than 5 minutes; of those, 86.9% were in the control group, 12.4% were in the RLP group, and fewer than 1% (0.8%) were in the RLP+ group. The control group was not required to interact with the website at all, while the RLP required interaction and the RLP+ required the longest period of interaction. Most participants (85%) completed their web visit in 20 minutes or less. Of those participants, the average time spent on the website was 3.56, 8.23, and 10.41 minutes for the control, RLP, and RLP+ group, respectively. Of participants, 5% took between 20 minutes and 1 hour to complete the web visit (51% in the RLP+ group, 29.4% in the RLP group, and 19.6% in the control group).
As reported in Table 1, out of the 984 randomized women, 939 (95.4%) completed the 6-month survey. Due to the censoring of women reporting interval pregnancies and hysterectomies, 960, 924, and 877 women were eligible to complete the 12-, 18-, and 24-month surveys, respectively. The survey response rates were 95.1%, 93.5%, and 93.7% for the 12-, 18-, and 24-month surveys, respectively. One woman withdrew from the study because she moved out of the country and 20 women (3 RLP, 6 RLP+, 11 controls) did not complete any follow-up surveys. The mean number of surveys completed by eligible participants was slightly greater in the information control group (3.55 out of 4.00) compared with the intervention groups (3.48 for RLP and 3.32 for RLP+; P = .038 using Kruskal-Wallis test).
Number of MyNewOptions Participants Eligible to Complete Surveys, Number Who Completed Each Survey, and Percentage of Surveys Completed by Intervention Group.
The baseline characteristics by study group allocation are shown in Table 2. There were no statistically significant or clinically meaningful differences in sociodemographic, pregnancy history, and contraceptive characteristics between study groups at baseline. Most of the participants were in a relationship, non-Hispanic White, college graduates, and employed. Most women were nulliparous (67.5%), and 17.5% had a prior unintended pregnancy or abortion. At baseline, 88.5% of women were using a contraceptive method: 8.4% were using LARCs, 49.7% were using other prescription methods, 30.4% were using nonprescription methods, and 11.5% reported using no method.
Baseline Characteristics of MyNewOptions Study Participants, N = 984.
Table 3 compares the contraceptive use outcomes by study group. Logistic regression models indicated no difference between any of the study groups in terms of the odds of any of the prespecified outcomes other than a small difference in prescription contraceptive use between RLP+ and RLP (P = .03). The mean proportion of the follow-up period that women were using any contraceptive method was 95%, 94%, and 95% in the control, RLP, and RLP+ groups, respectively. The mean proportion of time that women were using a prescription contraceptive method was 64%, 61%, and 65% in the control, RLP, and RLP+ groups, respectively. The mean proportion of the follow-up period that women were using a method congruent with their pregnancy intention was 88%, 86%, and 86% in the control, RLP, and RLP+ groups, respectively. There were also no differences between any of the study groups in terms of the odds of high contraceptive adherence during the follow-up period. A subgroup analysis of women who were not using a LARC at baseline showed there was no difference between study groups in the odds of switching to a more-effective method at some point during the follow-up period compared with baseline (P = .85). There were 24%, 26%, and 24% of women in the control, RLP, and RLP+ groups, respectively, who switched to a more-effective method during the 24 months.
Comparison of Contraceptive Use Outcomes Over 2 Years (2014-2016) by Study Group, n = 918 Women Not Intending Pregnancy.
To handle missing data for women who were completely lost to follow-up after the baseline survey (n = 20) or who missed surveys (N = 76), we performed a sensitivity analysis with multiple imputation using baseline data variables. There was no difference in the main results or conclusions when using multiple imputation; thus, results were presented using data from completed surveys alone.
The proportion of the follow-up period during which contraceptive satisfaction was high did not differ significantly between study groups (P = .86), as shown in Table 3. Contraceptive self-efficacy scores increased from baseline in all 3 study groups (P < .001) but there was no statistically significant difference in the magnitude of this change between groups (P = .37).
Comparison of contraceptive method categories at baseline and 24 months by study group are shown in Table 4. Overall, contraceptive use increased during the follow-up period; any contraceptive use increased from 88.5% to 95.8% (P < .001) and LARC use increased from 8.4% to 19.4% (P < .001), while use of non-LARC prescription methods and nonprescription methods remained unchanged. These increases did not differ significantly by study group.
Comparison of Contraceptive Method Categories at Baseline and 24 Months by Study Group, N = 984.
Subpopulation analyses for evaluation of heterogeneity of treatment effects were not prespecified in the original study proposal. Although we observed no intervention effects, we were interested in whether there may have been an intervention effect in younger women, as the ACA expanded insurance coverage to dependents up to age 26 years. We conducted post hoc stratified analyses by age group (<26 years and ≥26 years) but observed no intervention effects.
Reproductive life planning has been recommended by the CDC, ACOG, and the Office of Population Affairs for use in preconception and family planning counseling.10,11 To our knowledge, the MyNewOptions study is the first prospective trial to evaluate the effectiveness of reproductive life planning on contraceptive outcomes. MyNewOptions was designed to test the hypotheses that a web-based reproductive life planning intervention alone or in combination with contraceptive action planning could increase effective contraceptive use among insured women at risk for unintended pregnancy. In comparing the intervention and control groups, there were no differences in any of the contraceptive outcomes studied in this population of privately insured women in Pennsylvania.
The null results of the MyNewOptions study leave us to consider 2 possible conclusions—either reproductive life planning does not work or it did not work in the context of our study conditions. Reproductive life planning is grounded in the idea that establishing explicit, well-informed, and preferably unambivalent pregnancy intentions will lead to more-effective contraceptive choices and a reduction in unintended pregnancies. However, this assumes that forming pregnancy intentions is a realistic concept for women that, in turn, leads to planning behaviors in line with those intentions. While this paradigm has framed how researchers have approached strategies for unintended pregnancy reduction for many years, it has recently been questioned.58 The traditional paradigm of pregnancy intention oversimplifies women's perceptions of pregnancy because, in reality, many unintended pregnancies are welcomed and are not associated with increased risk of adverse pregnancy outcomes. Reproductive life planning may not resonate with women at risk for unintended pregnancy, and alternative approaches to helping women achieve contraceptive decision making must be considered.59
Alternatively, it is also possible that reproductive life planning was ineffective in the context of our study conditions. First, we created the MyNewOptions RLP and RLP+ interventions to be web-based, interactive versions of the CDC reproductive life planning tool that delivered consistent information to all users that would be low cost and easy to disseminate outside clinical settings. The MyNewOptions interventions contained more decision support information than the CDC RLP tool alone, in that they provided individualized information about pregnancy planning and contraception depending on responses to health information and preferences about contraceptive attributes. It is possible that the web-based approach did not provide adequate intervention intensity or was not a suitable substitute for in-person individualized counseling. Also, the MyNewOptions study sample was privately insured women in Pennsylvania, most of whom were White and highly educated with high incomes. We deliberately conducted the MyNewOptions study in insured women to test a reproductive life planning intervention in the context of no-cost contraception under the ACA. While we reasoned that insured women would have fewer barriers to switching to more-effective contraceptive methods and thus be responsive to the intervention, there may have been a ceiling effect. The study sample already had relatively high contraceptive use at baseline (88.5%) with high contraceptive method satisfaction. However, more than 30% of the baseline sample of women at risk for unintended pregnancy were using nonprescription methods or no method at all, and we expected to observe at least a modest increase in effective method use.
What implications do the null results of the MyNewOptions study have on current CDC, ACOG, and Title X position statements recommending reproductive life planning? It is important to recognize that we conducted MyNewOptions in a nonclinical study sample, so determining the effectiveness of reproductive life planning in clinical samples was beyond the scope of this study and remains to be seen. While the MyNewOptions study did not detect effects on contraceptive behavior outcomes, reproductive life planning could have different effects in the face-to-face clinical setting, perhaps related to knowledge, attitudes, and patient– provider engagement. Although reproductive life planning may be time consuming, it is unlikely to be associated with significant harm.
While we did not observe a difference in contraceptive outcomes by study group, there were increases in overall contraceptive use (88.5% to 95.8%) and LARC use (8.4% to 19.4%) during the follow-up period across all study groups. According to National Survey of Family Growth data, LARC use increased nationally from 2.4% of all contraceptive users in 2002 to 11.6% in 2011-2013.60 The 2014 National Survey of Family Growth indicated that, among women at risk for unintended pregnancy (and not sterilized), 17.5% were using LARCs, 38.9% were using other prescription methods, 30.7% were using coital methods, and 12.9% were using no method.61 It is also possible that the contraceptive information provided to all MyNewOptions study groups in conjunction with contraceptive coverage without cost-sharing under the ACA may have been responsible for this significant increase independent of the MyNewOptions interventions.
We conducted the MyNewOptions study in a sample of privately insured women with no-cost contraceptive coverage. Our privately insured sample was largely White, well-educated women and limited to a single state, and we cannot generalize our study findings to other groups of women, as discussed below (see the “Study Limitations” section). However, even the generalizability of these results to other privately insured women will depend on the durability of the contraceptive coverage requirement. In October 2017, the Trump administration issued new rules that would greatly expand the ability of employers to be exempt from the contraceptive coverage requirement based on employers' religious beliefs or moral objections.62 A preliminary injunction was subsequently issued preventing the rules from going into effect in a lawsuit filed by the Commonwealth of Pennsylvania against the Trump administration, citing that the rules contradicted the ACA and would cause irreparable harm to women.63 The Trump administration has not yet indicated a response.
The MyNewOptions study was the first trial to evaluate the comparative effectiveness of web-based reproductive life planning, which is currently recommended by the CDC and ACOG. At this time, the null results do not provide evidence to support web-based reproductive life planning or contraceptive action planning for altering contraceptive outcomes in privately insured women.
We found that the study participants were highly engaged in participating in a web-based intervention study, with high survey completion and retention throughout the study period. Web-based intervention studies are thus feasible and may be an alternative to clinic-based intervention studies.
The strength of the MyNewOptions study was that it used a randomized controlled trial design to test a reproductive life planning intervention on contraceptive use outcomes. It was the first study to evaluate the effectiveness of reproductive life planning in a controlled trial. We had excellent participant retention throughout the 24-month study, thus minimizing threats to internal validity.
We intentionally conducted the study in privately insured women with no-cost contraceptive coverage who would have fewer barriers to switching to more-effective contraceptive methods. However, our privately insured sample was largely White, well educated, English-speaking, and limited to a single state, and we cannot generalize our study findings to other groups of women. We expect that women without private insurance have more barriers to receiving patient-centered reproductive health care and could possibly be more responsive to a web-based intervention if they are unable to access care elsewhere. Thus, the question remains whether the MyNewOptions RLP/RLP+ or other reproductive life planning interventions might impact contraceptive use in women who do not have access to contraception without cost-sharing.
Our study sample was representative of the sampling frame based on basic demographic data (age, geographic region, rural/urban residence). However, women more engaged in their health may have been more attracted to enroll in a research study about women's health. There may have been a ceiling effect with our well-educated study sample who may have already been knowledgeable about their contraceptive options. Internet access and willingness to participate in a web-based study were necessary to participate in the MyNewOptions study. Although study participants reported the web-based format was easy to use and desirable, we may have selected out women less willing to participate in an interactive web-based study.
The study results were based on self-reported survey measures, which introduces the potential for recall and social desirability biases. Measurement and comparison of longitudinal contraceptive behaviors is challenging, not only due to the need to rely on self-reported measures but also because of the need to compare across a variety of different methods. We defined our longitudinal contraceptive outcomes based on whether the measured contraceptive outcome (ie, any contraceptive use, high contraceptive adherence, high method satisfaction) was present during the survey periods during which the participant reported she was not trying to become pregnant and thus at risk for unintended pregnancy. We acknowledge that measuring outcomes in 6-month time frames is arbitrary and, in reality, pregnancy intentions can change much more frequently.
The MyNewOptions study did not provide evidence to support reproductive life planning (with or without contraceptive action planning) when delivered in a web-based format to privately insured women. For some of the reasons outlined above, it is possible that reproductive life planning may still be an effective tool if delivered in a different manner or in a different study population. Future investigations to evaluate reproductive life planning interventions in alternative settings and populations, such as publicly insured and uninsured, low-income, and more race and ethnically diverse groups are warranted.
In conclusion, evidence for CDC-, ACOG-, and Title X family planning–supported reproductive life planning remains lacking. We designed the MyNewOptions study on the heels of the passage of the ACA to potentially assist privately insured women with no-cost coverage for contraception. The MyNewOptions study was the first trial to evaluate a web-based reproductive life planning intervention, but the null results indicate the MyNewOptions web-based RLP/RLP+ interventions were not effective in this sample of privately insured women. Although the vast majority of the study participants (88.5%) were using some form of contraception at the start of the study, nearly 42% were using a less-effective method (ie, barrier methods, withdrawal, natural family planning) or no method at all, which did not significantly change at the end of the 2-year study. Thus, we need to find better ways to help women achieve their reproductive goals in the context of contraceptive cost considerations and accommodate privately insured women as well as publicly insured and uninsured women who may have varying levels of contraceptive access and out-of-pocket costs. These interventions may include retooled reproductive life planning approaches or alternative strategies given that RLP has not been found to significantly change contraceptive behaviors to date. Most importantly, these efforts are needed to provide reproductive-age women with access to contraceptive methods that best meet their individual needs and preferences.
We thank Suzanne Stoner, Vanessa Adjei, Lindsay Confer, and Kayla Confer for their assistance with participant recruitment; John Webster and Robert Stouffer of The John Webster Company for web design; and the members of the MyNewOptions Patient Advisory Group and Clinician Advisory Group for their contributions to study design and data interpretation.
We would like to thank Highmark Health for assistance with participant recruitment. The findings and conclusions presented are solely those of the authors and do not represent the views of Highmark Health.
Study data were collected and managed using REDCap tools hosted at the Penn State Milton S. Hershey Medical Center and College of Medicine. REDCap is supported by the Penn State Clinical & Translational Science Institute, Pennsylvania State University CTSI, NIH/National Center for Advancing Translational Sciences (NCATS) grant UL1 TR000127. The contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH/NCATS.
Research reported in this report was [partially] funded through a Patient-Centered Outcomes Research Institute® (PCORI®) Award (#CD-1304-6117) Further information available at: https://www.pcori.org/research-results/2013/using-reproductive-life-planning-website-and-action-plan-help-women-choose-and
The MyNewOptions Study Protocol (PDF, 842K)
Figure 1. Improving Contraceptive Use through Self-Regulation (PDF, 185K)
Figure 1. Excerpt from Contraceptive Action Plan for Oral Contraceptives (PDF, 67K)
Table 2. Categorization of contraceptive efficacy (PDF, 138K)
Table 3. Definition of Contraceptive Adherence (PDF, 267K)
Figure 2. Governance Plan (PDF, 79K)
Website Screenshots (PDF, 7.6M)
Chuang CH, Weisman CS, Velott D, et al. (2019). Using a Reproductive Life Planning Website and Action Plan to Help Women Choose and Use Birth Control. Patient-Centered Outcomes Research Institute (PCORI). https://doi.org/10.25302/9.2019.CD.13046117
The [views, statements, opinions] presented in this report are solely the responsibility of the author(s) and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute® (PCORI®), its Board of Governors or Methodology Committee.
This book is distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License which permits noncommercial use and distribution provided the original author(s) and source are credited. (See https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
