Feasibility of a Contraception Intervention for Hospitalized Adolescents and Young Adults Free

We conducted a pilot study of AYAs assigned female sex at birth and aged 14 to 21 years who were admitted to the general pediatric service at a Midwestern tertiary children’s hospital from October 1, 2020 to March 31, 2022. Consistent with state law, the Institutional Review Board waived parental consent for minors.²⁷  We adapted a general SRH intervention previously developed by our team based on the Theory of Planned Behavior (TPB) with Motivational Interviewing principles to guide counseling.^28,29 

We developed our original intervention for sexually-experienced adolescents, including those identifying as gender diverse (ie, nonbinary, transgender), in the pediatric emergency department (ED). We adapted content to focus on contraceptive counseling for hospitalized AYAs and used a computerized decision support system to identify need for contraception and estrogen contraindications (eg, migraine with aura). A health educator administered the interactive, tablet-based intervention that included open-ended questions about sexual behaviors, risk-reduction counseling, family planning goals, and contraception. Participants received condom training via video and counseling on dual protection. Finally, the educator reviewed personalized contraception education and potential medications if desired. The educator offered all participants condoms, an SRH resource handout, adolescent medicine clinic referral, and/or initiation of hormonal contraception unless medically contraindicated (ie, pill, patch, vaginal ring, injection, subdermal implant, emergency contraception, clinic referral for IUD placement). The educator communicated the participant preferences to the inpatient medical and research teams to coordinate prescription. AYAs who started contraception were offered a pregnancy test to use 2 weeks after initiation. The intervention was pilot tested for comprehension with 5 AYAs included in this analysis; we made minor revisions to improve clarity on sexual behavior assessment and methods to identify estrogen contraindications and/or teratogen exposure.

We engaged with HCPs to address previously identified barriers to SRH care.²³  We provided contraception education for resident physicians, pediatric hospitalists, and inpatient staff nurses, which included oral presentations and publicly available resources (online and printed). The principal investigator met with administrative and nursing leaders before study initiation to provide education on the intervention and facilitate buy-in.

We trained 2 female Master’s-level health educators to deliver the intervention (minimum 12 hours training). The educators received an intervention manual that included a detailed description of the intervention, screen shots, and sample scripts. They received training to develop and practice Motivational Interviewing techniques as well as review of recorded role-playing sessions and direct observation with our study team to ensure proficiency before interacting with participants.

AYA participants were eligible if they spoke English and reported past or anticipated sexual activity. We excluded participants who were developmentally delayed, too ill to participate (per medical team), pregnant, currently using LARC, or receiving care for sexual assault. To assess feasibility and acceptability measures, we also approached AYA parents and guardians and health care providers (HCPs), both of whom were eligible if they spoke English.

We reviewed the electronic medical record (EMR) to identify a convenience sample of participants based on research team availability (eg, approximately 2 weekdays per week). After approval by the medical team, we approached eligible participants who provided electronically signed informed consent (adults) or assent (minors). When a parent was present, we provided a brief study overview and requested verbal parental permission to approach the participant. Family members and visitors were asked to leave the room during study procedures. If family members declined to leave, the participant was deemed ineligible. For AYAs who declined, we collected self-reported age, race and ethnicity declination reason, and insurance type from the EMR.

AYAs completed an electronic survey to report demographic information, medical history, and health behaviors, which linked to our previously developed computerized decision support system to generate tailored recommendations.²⁸  AYAs also completed a survey immediately postintervention ($20 remuneration) and were contacted 3 months later via web link or telephone to complete a follow up survey ($25 remuneration) assessing health behaviors and contraception use. Data were collected using Research Electronic Data Capture.^29,30  After the intervention, the team approached parents and guardians and nurses who provided verbal informed consent to complete a computerized survey.

For participants requesting pill, patch, vaginal ring, or injection, our team engaged with inpatient physicians to prescribe a 3 month supply or a single injection. For participants requesting the implant, trained study clinicians performed the insertion after obtaining signed parental procedure consent (per institutional policy). All study-related medications, including subdermal implants and placement, were provided before discharge at no charge. We included information about contraception initiation in the EMR and discharge summary to notify the primary care physician. If the participant requested confidential care, we suppressed this information from the EMR.

We assessed sociodemographic factors and health behaviors using validated items from a national survey (Youth Risk Behavior Surveillance).³⁰  The educator reviewed the EMR to assess medical insurance, completion of clinic referral and identify teratogenic medications (based on previous Food and Drug Administration categorization as “D” or “X”) and identify underlying medical conditions linked with pregnancy complications.^31,32 

To assess feasibility, we computed the proportion of enrolled AYAs who completed the intervention. We calculated intervention duration using start and stop times captured electronically via Research Electronic Data Capture. We assessed the impact on hospital care using 3, 5-point Likert scale items with AYAs (“Being in the health program to talk about birth control did not make it harder to get care for my main medical problem”), 1 item with parents and guardians (“Meeting with the educator to discuss health behaviors was minimally or not at all disruptive to my child’s regular hospital care”), and 1 item with nurses (eg, “When your patient was talking with the health educator about contraception, how much interruption to your workflow did you experience?”). We dichotomized Likert scale responses (eg, strongly agree or agree, minimally or not at all disruptive) and calculated the proportion of respondents who indicated feasibility.

To evaluate acceptability, we computed the proportion of AYAs rating the intervention as “somewhat or very” satisfactory on a 5-point Likert scale. We also evaluated intervention acceptability with an item designed for parents and guardians (eg, “It was acceptable to approach my child to participate in the health program”). We evaluated acceptability among nurses by assessing agreement with the statement “… increasing access to contraception by offering this care in the hospital fits within the goals of the hospital”.

For efficacy, we assessed health service uptake as the proportion of AYA participants who accepted ≥1 health service (ie, condoms, SRH resource handout, adolescent medicine clinic referral, initiation of hormonal contraception, home pregnancy test) at study enrollment. At 3 months, we assessed sexual health behaviors, condom use and moderately or highly effective contraception use or continuation (ie, oral contraception pill [OCP], patch, ring, injection, intrauterine device, implant).

Informed by the TPB, we gathered data on pregnancy prevention attitudes (eg, behavioral beliefs) and perceived behavioral control (eg, confidence, self-efficacy). We assessed pregnancy prevention attitudes by asking “Which of the following best describes your thoughts about pregnancy?” Response options were “Wouldn’t mind getting pregnant,” “Not sure,” “Wouldn’t mind avoiding pregnancy,” and “Trying to avoid pregnancy.” We measured perceived behavioral control pre and postintervention by asking, “Overall, how confident would you say you are in your ability to protect yourself against any unwanted pregnancy?” and with the question “How likely are you to start birth control within the next 30 days?”

We assessed patient-centeredness among AYA participants immediately postintervention by computing the proportion of participants agreeing “somewhat” or “strongly” on a 5-point Likert scale that the educator “… made me feel comfortable;” “…listened to what was important to me;” and “… helped me make the best choice for me.”

We used means with SD or medians with interquartile ranges to summarize normally and nonnormally distributed continuous data, respectively. We presented categorical data as proportions. We included participants even if they had missing data from skipped questions and noted data missing in frequency calculations. We used independent samples t tests and χ² tests to compare groups on continuous and categorical variables, respectively. All analysis were completed using SAS Version 9.4 (SAS Institute, Cary, NC).

We enrolled 25 AYAs (38% of those eligible; Fig 1). The mean age was 16.4 years (SD 1.5), 14 (56%) had either public or no insurance, 17 (68%) were white, and 2 (8%) identified as gender diverse (Table 1). The 2 groups (ie, those who participated and declined) were comparable in terms of age or race and ethnicity. There were more AYAs without insurance in the enrolled group compared with those who declined (21% vs 7%, P < .001). Of those who declined, the most common reasons were lack of interest (n = 34, 83%) and not up to it (eg, tired, sick; n = 5, 12%). We approached 1 parent or guardian and 1 inpatient staff (eg, nurse) for each enrolled AYA and enrolled 8 parents and guardians (response rate 32%) and 11 nurses (response rate 44%).

Twenty participants (87%; n = 2 missing data) reported any past vaginal sexual activity (Table 1). Most (n = 15, 60%) participants reported they were “trying to avoid pregnancy”. Five participants (20%) were using a hormonal birth control method (all using OCPs). OCP adherence challenges were common, with 2 (40%) reporting they started their OCP pack late and 2 (40%) reporting a recently missed pill. Eleven (50%, n = 3 missing) reported they used a birth control method every time they had vaginal intercourse in the past 30 days, whereas the other 11 reported they were unsure or did not use a birth control method.

Eight (32%) participants had a teratogenic medication exposure but only 3 (38%) reported being informed they were taking potentially harmful medication. Five reported a history of migraine headaches with aura, 4 reported a personal or family history of blood clots, and 4 reported diabetes.

Of 25 AYAs who enrolled, 100% completed the intervention. The median intervention duration was 32 (Interquartile range 25–45) minutes. Seventeen (71%) participants strongly or somewhat agreed that the intervention did not make it harder to receive hospital care. All 8 parents and guardians reported that the intervention was not at all or minimally disruptive to their child’s hospital care. Nine (82%) nurses reported the intervention was not at all or minimally disruptive to their workflow.

The 24 AYA participants (missing data n = 1) who responded to the acceptability item were very or somewhat satisfied with the intervention. Of 8 parent and guardian participants, 7 (88%) reported it was acceptable for the educator to meet with their child in private. All nurse respondents strongly agreed or agreed that increasing access to contraception in the hospital fits within the goals of the hospital.

Twenty-three participants (92%) received condoms (ie, external or male) and 11 (44%) started contraception during their hospitalization (Table 2). Of those who started contraception, 7 (64%) started the subdermal implant, 2 (18%) started the patch, 1 (9%) chose the OCP, and 1 (9%) started the vaginal ring.

Postintervention, all 24 participants who responded were extremely or very confident in their ability to protect themselves against unwanted pregnancy compared with seventeen (68%) preintervention. Twenty-two (91%; missing data n = 1) were somewhat or very likely to start birth control within the next 30 days (including those who were continuing or changing their contraception method).

The 24 participants (missing data n = 1) who responded strongly agreed they were treated as an equal by the educator. Similarly, 24 participants (missing data n = 1) somewhat or strongly agreed the educator made them feel comfortable, listened to what was important, and helped them make the best choice for them.

Eight participants (32%) completed the 3 month follow-up survey. Seven initiated contraception at enrollment, all of whom received the subdermal implant. None stopped or changed their method since enrollment. Three (38%) reported medication side effects related to menstrual changes. Since enrollment, 6 (75%) reported they had vaginal intercourse with a male and 4 (50%) reported condom use at last sex.

In this pilot study of AYAs in a pediatric hospital, we found evidence of feasibility and acceptability of our interactive intervention designed to increase contraception access. Most participants accepted at least 1 health service (eg, contraception, condoms) and many initiated contraception, most commonly choosing a LARC (ie, implant). Additionally, nearly all AYA participants, parents and guardians, and HCPs rated the intervention as acceptable with minimal or no disruptions to care. In light of increasing limitations to reproductive care access, efforts to increase access to contraception for historically marginalized populations, including AYAs, are critically important to offset risk for negative health outcomes.

We are among the first to develop and evaluate an intervention to offer contraception counseling and prescription, including LARC, for AYAs in this setting. McFadden et al demonstrated that EMR changes and e-mail prompts can increase documentation of a sexual history in hospitalized adolescents, although no meaningful change in health service provision was seen.³³  Work by our team and others demonstrated strong feasibility and acceptability of SRH interventions in other settings (ie, pediatric ED).^28,34  The high rates of intervention feasibility and acceptability and health service uptake in our study sample are similar to past studies of AYAs in other nontraditional settings (ie, ED).^28,34–36  We overcame provider-level barriers by using an educator, providing multilevel HCP contraception training, and collaborating with inpatient HCPs for contraception prescription. Future research can expand on our findings to increase AYA contraception access in nontraditional settings using a similar multilevel approach and thus offset risk for negative health outcomes in this population.

Among participants who started contraception, nearly two-thirds chose a LARC (ie, subdermal implant). There are no comparative studies in the pediatric hospital, however this rate is similar or higher compared with LARC uptake in ED- and primary care-based studies.^9,28,34  Because of institutional regulations, we required parental consent for the implant procedure, which may have been a barrier for some AYAs. Although we were prepared to offer EC, no participants reported unprotected intercourse in the previous 5 days, which differs from ED-based studies where 14% of females reported unprotected intercourse within the past 5 days.³⁷  Future work could expand on methods to protect confidentiality and to identify patients who may be in need of time-sensitive EC counseling.

Although a majority of our sample was sexually active and reported they wanted to avoid pregnancy, most were not using effective contraception. Participants reported high rates of sexual intercourse with the same or both sexes and an average of >4 lifetime sexual partners, which are independently associated with increased risk for unintended pregnancy.^1,38  Additionally, hospitalized AYAs commonly have underlying medical conditions or teratogenic medication exposure, which increases risk for complications during pregnancy.^14,25  Nearly one third of participants had teratogenic exposure, although most were unaware these medications could be harmful during pregnancy. Given the high risk for negative health outcomes related to unintended pregnancy, provision of SRH care, including pregnancy prevention, in this population is critically important.

Our findings should be viewed in light of the following limitations. We enrolled a small number of participants because of issues related to the coronavirus disease 2019 pandemic (ie, institution-wide research enrollment halt), 1 of our educators departing during the study, and access to a limited number of inpatient teams. Our findings may not be generalizable to other hospital settings or AYA populations. We did not offer IUD placement inpatient because of lack of credentialled providers. Future studies should expand upon our work and enroll a larger sample size across multiple sites. There is risk for social desirability bias, although we attempted to mitigate this using electronic data collection. We did not exclude inpatient nurses based on previous participation; thus, it is possible for bias if duplicated responses occurred. Also, because our enrollment criteria required parents and guardians to agree to leave the room while we interacted with their AYA, our sample of parents and guardians and their responses about the intervention acceptability may have been biased. Many AYAs were lost to follow up, limiting our ability to evaluate longer-term outcomes.

As AYAs face numerous and increasing barriers to accessing SRH care, innovative methods to offer contraception education and services at every clinical encounter are critical. Our intervention for AYAs in the hospital was feasible, acceptable, and demonstrated promise of efficacy with high rates of contraception initiation. Future efforts should expand on this work to increase SRH care provision for AYAs in the hospital to offset adverse health outcomes related to unintended pregnancy.