A 3-year Retrospective Review of Contraceptive Initiation, Continuation, Switching, and Pregnancy Among Adolescents and Young Adults

Abstract

Objectives:

To understand contraceptive use patterns (initiation, switching, discontinuation) as well as associations with pregnancy in adolescents and young adults attending a teen family planning clinic.

Study Design:

We performed a chart review of adolescent and young adult patients (ages 12–20) attending a teen family planning clinic in Atlanta, GA between January 1, 2017 and December 31, 2019. Using a standardized abstraction form with quality controls, we collected available data on contraceptive methods used and pregnancy test results during the 3-year period. We analyzed contraceptive use patterns descriptively. We calculated and compared pregnancy incidence according to different contraceptive switch patterns.

Results:

Our sample included 2798 individuals who initiated 2358 prescribed methods. The most commonly prescribed methods of contraception were the contraceptive injection (28.3%), etonogestrel implant (23.5%) and combined hormonal pill (23.2%). There were 599 discontinuations of prescribed methods; side effects like bleeding and headache were the most cited reasons for discontinuation. Most (75.8%) initiated a moderately or highly effective method after discontinuing a moderately or highly effective method. The incidence rate of pregnancy was highest for those who had discontinued an intrauterine device or implant and started a shorter-acting contraceptive method.

Conclusions:

Employing patient-centered contraceptive counseling that incorporates contraceptive experiences in addition to facts and allows for exploration and change may be valuable for young people. Successful navigation of contraceptive switches may require additional attention, education, and strategy, which could include hypothetical problem solving, close follow-up, and telehealth or virtual care.

Implications:

Contraceptive initiation and switching were common in this sample. The incidence rate of pregnancy was highest for those who had discontinued an intrauterine device or implant and started a shorter-acting contraceptive method. Clinicians should employ patient-centered contraceptive counseling and consider additional strategies to support successful navigation of contraceptive switches.

Introduction

Over the past three decades, the United States (US) has seen dramatic declines in teen pregnancy and teen births. The teen birth rate was at its peak in 1991 with 61.8/1,000 15–19-year-old females. By 2020, the birth rate had dropped to 15.4/1,000 15–19-year-old females.(1) This is an important public health success because the large majority of teen births are unplanned and undesired and they have substantial impact on the individual, family and societal levels.(2) Declines in teen birth have been seen across all racial and ethnic groups.(1) However, disparities remain; Hispanic and non-Hispanic Black teens continue to experience a birth rate that is twice that of white teens.(1) More than race and ethnicity, the teen birth rate is associated with social determinants of health and socioeconomic factors, as both predictors and outcomes.(2, 3)

Some of the early decline in teen birth rate can be attributed to delay in initiation of sexual activity.(4) However, a majority of the decline through 2002 and nearly all of the decline thereafter has been attributed to increased contraceptive use and use of more effective contraceptive methods.(5, 6) Some national surveys provide high-level information on contraceptive use by adolescents. The US National Survey for Family Growth (NSFG), 2015–2017, reports that among females 15–19 years who have ever had sex, 53% had ever used pills, 19% had ever used the injection, 15% had ever used an implant and 5% had ever used an intrauterine device (IUD).(7) The 2019 Youth Risk Behavior Surveillance Survey (YRBS), a school-based survey, found that of high school students who were currently sexually active, 5% used an IUD or implant and 25% used a pill before last sex.(8) However, rich detail about contraceptive initiation, switching and discontinuation is lacking from these sources.

Little is known about patterns of contraceptive use among young people, particularly in those who disproportionately experience higher birth rates, including minoritized adolescents and those from lower income households. Increased understanding of contraceptive patterns, including initiation, switching, discontinuation, and factors associated with these patterns and their outcomes, is necessary. The primary aim of this study was to describe contraceptive use patterns of females attending a teen family planning clinic in Atlanta, GA over a three-year period.

Methods

This study was performed at a teen family planning clinic that receives Title X funding in Atlanta, GA. The clinic is located within a large safety net hospital system which includes full-service care (emergency, primary care, labor and delivery, etc.) which uses a single electronic medical record (EMR). The teen family planning clinic serves patients through age 20 after which they are seen in the adult clinics. Most patients speak English and have public insurance or no insurance. Title X funding provides financial support for family planning services for people who do not have another mechanism for payment. Title X support can be particularly important for young people who may need to use the program to access family planning services confidentially. We offer full scope contraceptive care with teen-friendly practices.(9–11) Our clinic encourages parental involvement, but most young people attend the clinic without a parent or guardian. We conducted a chart review of all visits for female patients from January 1, 2017 to December 31, 2019. This study window was selected to avoid impact from the COVID-19 pandemic. Using the EMR, we abstracted data on patient characteristics, contraceptive methods, and pregnancy tests. Patients were included if they were under 20 years of age at their first visit during the study window. Gender was ascertained from the demographics section in the EMR. We included contraceptive care or pregnancy testing from other settings (e.g. emergency room, labor and delivery, other outpatient clinics) during the study period if these visits were in the EMR. Our hospital offers immediate postpartum contraception, and review of discharge summaries allowed for us to capture contraceptive initiations that happened in labor and delivery. This study was approved by Emory University’s Institutional Review Board and Grady Health System’s Research Oversight Committee.

Data abstraction processes were based upon best practices from Vassar and Holzmann.(12) We created a database in REDCap, a HIPAA-compliant web-based application designed to support data capture for research studies. We collected the following elements: date of birth, date of service, race, ethnicity, payor information, reason for visit, gravidity, contraceptive method(s) at the beginning and end of the visit, complaints, contraceptive side effects, and results of pregnancy tests.

We developed standardized REDCap forms for data abstraction based on the study protocol. Two researchers performed pilot abstraction and revised the forms in an iterative fashion until all variables of interest were appropriately captured. Abstraction forms were formatted to reduce data entry errors by minimizing open text, creating parameters for realistic values, using drop-down menus, and leveraging branching logic. We created standard operating procedures to guide the abstraction and ensure consistency when reviewing and abstracting the charts. A senior team member trained all other members prior to abstraction. There were 11 individuals who did any chart review, and the large majority (~90%) was completed by 4 individuals. For quality control, 10% of visits were abstracted by two people. We checked data quality monthly, identified, and corrected discrepancies, and retrained team members as needed.

Descriptive statistics were compiled for sociodemographic characteristics, appointments, and variables of interest. We calculated age using the time between date of birth to date of service of the first visit. We assessed the primary contraceptive method at the beginning and end of each visit and used this information to examine initiation, switching, and discontinuation. We considered the use of a contraceptive method at the end, but not the beginning of the visit, as an initiation of a prescribed contraceptive method (for an intrauterine device (IUD), implant, or conceptive injection, pill, ring, or patch). Sexual activity or intention of use of contraceptive method was not considered. Because non-prescribed contraceptive methods are readily available, do not require interaction with the clinic, and were inconsistently documented in the EMR, we did not consider these methods as part of the contraceptive initiation variable. If a person was given a second prescribed method (for example, combined hormonal pills to manage breakthrough bleeding), that was also not considered an initiation. We categorized methods into efficacy tiers: Tier 1 which included the etonogestrel implant, levonorgestrel IUD, and copper IUD; Tier 2 which included contraceptive injection, patch, ring, and pills; and Tier 3 which included non-prescribed contraceptives (e.g. condoms, withdrawal, emergency contraception)(9). Discontinuation occurred when a person had a prescribed method at the start of the visit and a different method or no method at the end of the visit. Discontinuation could be initiated by the provider or patient, and patients do not need to justify their desire for discontinuation. We assessed length of use of prescribed methods in days from the date of initiation until switch, discontinuation, or positive pregnancy test. If none of these were identified, the length of continuation for Tier 1 methods was measured up to the end of study period. For a contraceptive injection, length of continuation was considered 12 weeks (84 days) unless there were future visits for repeat injections. For contraceptive pills, patches, and rings, we considered continuation 12 weeks (84 days) because the clinic routinely dispenses a minimum of 3-month supply at initiation (and more when feasible) with refills for one year. Because we cannot know how much of the product given to the patient was actually taken, we also repeated this assessment attributing just one month (28 days) for pills, patch, and rings. Tests of association were conducted to assess differences of initiation and length of use by age. Data elements that were missing or not documented were coded as such, respectively.

We calculated pregnancy incidence rates per 100 person-years for individuals by type of contraceptive switch. While we did not consider Tier 3 methods for contraceptive initiation, we did consider switching from a Tier 3 method to a Tier 1 or Tier 2 method in describing contraceptive switches. We compared the incidence rates across switch types and assessed for statistical significance with and without controlling for gravidity and age using established statistical techniques.(13) We chose to control for gravidity and age because these factors may be developmentally and experientially relevant to the analysis and were available in the dataset. Analyses were performed using SAS and IBM SPSS Statistics, v. 28. A p value of <0.05 was considered statistically significant.

Results

Between January 1, 2017 and December 31, 2019, n=2798 adolescent and young adult female patients attended our clinic, and the characteristics are presented in Table 1. Patients were primarily non-Hispanic African American and had public insurance or were uninsured. The mean age at the first visit was 18.0. The mean number of non-obstetric visits was 2.9 in the 3-year period (range: 1–21, median: 2). Contraceptive care and STI care were the most frequently sought services.

Table 1.

Characteristics of female adolescent and young adult family planning patients, in Atlanta, GA, 2017–2019

Patient Characteristics	N=2798 (%)
Age* (12–20) (mean, SD)	18.0 (1.4)
Gravidity*
0	799 (28.6)
1	843 (30.1)
2 or more	444 (15.9)
Missing	712 (25.4)
Race and ethnicity
Non-Hispanic African American	2528 (90.4)
Hispanic	169 (6.0)
Non-Hispanic White	37 (1.3)
Other	64 (2.3)
Payor*
Private (Blue Cross, Commercial)	217 (7.8)
Public (Medicaid, Medicaid Managed Care, Medicaid Pending)	1540 (55.0)
Self-pay (or Null)	1039 (37.1)
Missing	2 (0.1)
Location of visits abstracted (n=8786)
Teen Clinic	5791 (65.9)
Family Planning	2407 (27.4)
Labor and Delivery	530 (6.0)
Other	58 (0.7)
Services sought at visits** (n=11,552)
STI concern/test/care	2963 (25.6)
Change/start/stop/problem with contraception	2545 (22.0)
Pregnancy concern/test/care	2414 (20.9)
Check-up, annual exam or other	1948 (16.9)
Refill or repeat contraceptive injection	1682 (14.6)

^*At time of first visit

^**Multiple services possible, total services mentioned N=11,552.

During the study period, n=2358 prescribed contraceptive methods were initiated in over 70% of the sample (n=1986) (Figure 1). The most initiated methods were the contraceptive injection, implant, and combined pill. Short-acting contraceptives accounted for over two thirds of initiations. Seventy-six percent of patients (n=1508) initiated one method, 20.3% (n=403) initiated two methods, 3.4% (n=67) initiated three, and 0.4% (n=8) initiated four methods. The average number of prescribed contraceptive methods initiated was 1.28 per person (SD ±0.543, median 1.0, range 1–4). Individuals 18 years of age and older were more likely to initiate a copper IUD, levonorgestrel IUD, contraceptive ring, or progestin only pill compared to those under 18 years.

Figure 1.

Distribution of prescribed contraceptive methods initiated among a sample of adolescent and young adult female patients in Atlanta, GA, 2017–2019.

(N=2358 methods, one person could have initiated more than one method over the study)

Length of continuation of prescribed methods initiated during the study window are outlined in Table 2. Tier 1 methods had longer length of continuation and was over one year on average for each method. Three quarters of those placed during the study period were still in use at the end of the study (82% of levonorgestrel IUDs, 70% of copper IUDs, and 76% of etonogestrel implants). This analysis did not include length of continuation of methods that were initiated prior to the study period. The length of use for the contraceptive pill, patch, and ring was widely variable. The median length of use defaulted to our attributed value (for both 84 and 28 days) as not having further contact with the clinic after method initiation was common (25.6%, n=312). Length of use of Tier 1 and Tier 2 methods did not vary by age category.

Table 2.

Length of continuation (in days) after initiation of prescribed contraceptive methods among a sample of adolescent and young adult female patients in Atlanta, GA, 2017–2019

	Days of use	Days of use
	Mean (SD)	Median (Min, Max)
Copper IUD (n=43)	490 (317)	573 (7, 1020)
Levonorgestrel IUD (n=136)	458 (302)	430 (15, 1080)
Etonogestrel implant (n=554)	450 (296)	413 (2, 1090)
Any Tier 1 (n=733)	454 (298)	417 (2, 1090)
Contraceptive Injection (n=668)	175 (160)	84 (7, 1060)
Combined Contraceptive Pill (n=548)	122 (101)	84 (2, 968)
Progestin-only Pill (n=93)	87 (58.6)	84 (9, 509)
Contraceptive Patch (n=247)	136 (130)	84 (9, 939)
Contraceptive Ring (n=69)	117 (101)	84 (5, 545)
Any Tier 2 (n=1625)	144 (134)	84 (2, 1060)
All methods (n=2358)	240 (246)	107 (2, 1090)

Length of use in days was calculated from date of prescription, injection, or insertion to date of removal or known discontinuation.

IUDs and implants placed during the study period that did not have a visit for removal or a subsequent visit that indicated the device was no longer in place were considered to still be in place at the end of the study period.

For unknown date of discontinuation, the contraceptive injection was assigned 84 days of use.

For the contraceptive pill, patch, and ring, 84 days of use was assumed as that is the minimum amount dispensed at the clinic.

Tier 1: Etonogestrel implant, levonorgestrel intrauterine device (IUD), or copper intrauterine device (IUD)

Tier 2: Contraceptive injection, contraceptive patch, contraceptive ring, combined pills, or progestin only pills

Over the course of the study, 599 prescription contraceptive methods were discontinued during a clinic visit. Patients generally switched from one Tier 1 or Tier 2 method to another Tier 1 or Tier 2 method (75.8%, n=454), while one-quarter chose no method or a Tier 3 method. Eighty percent (478/599) of discontinuations had a reason documented in the medical record (Figure 2). Side effects were the leading reason why an individual chose to discontinue a prescribed contraceptive method. Bleeding concerns (irregular, heavy, absent, painful) were the most commonly listed reasons for discontinuation, followed by vaginal discharge or other vulvar complaints, weight gain, and headaches.

Figure 2.

Primary reasons documented for N=599 discontinuations of a prescribed contraceptive method among adolescent female patients in Atlanta, GA, 2017–2019

There were 202 pregnancies diagnosed after a documented contraceptive switch to a Tier 1 or Tier 2 method. Pregnancy incidence rates per 100 person-years for various switching patterns are shown in Table 3. There was no difference in incidence of pregnancy after switching from a Tier 2 or Tier 3 method to a Tier 1 (p=0.46). However, switching to a Tier 1 method was associated with the lowest incidence of pregnancy (1.5–3.7 pregnancies per 100 person-years), and all other pairwise comparisons were statistically significant (p<0.001). The highest incidence rate occurred after switching from a Tier 1 to a Tier 2 method (62.2 pregnancies per 100 person-years).

Table 3.

Unadjusted pregnancy incidence rates among adolescents and young adults by most recent contraceptive switch among those who had initiated contraception in Atlanta, GA, 2017–2019

Contraceptive tier at beginning of visit	Contraceptive tier at end of visit	No. of Observations	No. of pregnancies	Person-years (PY)	Incidence rate Per 100 PY [95% CI]
2	1	101	2	130.8	1.5 [0.0 3.7]
3	1	142	6	163.6	3.7 [0.7 6.6]
3	2	405	43	159.3	27.0 [18.8 35.1]*
1	2	156	39	62.7	62.2 [42.7 81.8]*

^*p<0.001 for all pair-wise comparisons except for 2 to 1 vs. 3 to 1 switches

Tier 1: Etonogestrel implant, levonorgestrel intrauterine device, or copper intrauterine device

Tier 2: contraceptive injection, contraceptive patch, contraceptive ring, combined pills, or progestin only pills

Tier 3: non-prescribed methods including condoms, withdrawal, emergency contraception

Table 4 shows pregnancy incidence rate comparisons of switching up to a Tier 1 and down from a Tier 1 to a Tier 2 method, adjusted separately for gravidity and age. Incidence rates of pregnancy for those who switched up were low across gravidity and age strata. Incidence rates of pregnancy for those who switched down from a Tier 1 to a Tier 2 were statistically and clinically significantly higher in those with previous pregnancies (p<0.001) and those over 18 years (p<0.001).

Table 4.

Pregnancy incidence rates among adolescents and young adults by most recent contraceptive switch stratified by gravidity and age among those who had initiated contraception in Atlanta, GA, 2017–2019

	Contraceptive switch up (from Tier 2 or Tier 3 to Tier 1)			Contraceptive switch down (from Tier 1 to Tier 2)
	Pregnancies	Person-years (PY)	Incidence rate per 100 PY [95% CI]	Pregnancies	Person-years (PY)	Incidence rate per 100 PY [95% CI]
Gravidity =0	2	138.2	1.6 [0–3.5]**	4	24.2	16.7 [0.33–32.7]**
Gravidity >0	6	129.0	4.4 [0.9–8.4]**	35	36.6	95.0 [63.9–127.3]**
Age* <=17	2	118.1	1.7 [0–4.0]**	15	31.4	47.7 [23.6–71.9]**
Age>=18	6	176.3	3.4 [0.7–6.1]**	24	31.3	76.8 [46.0–107.4]**

^*Age was rounded (floored) to whole number

^**P<0.001 for comparisons of switching up vs. switching down, adjusting for gravidity, and adjusting for age

Discussion

Of prescribed methods, our patients initiated the contraceptive injection, implant, and combined hormonal pills most frequently. This is higher injection and implant use than has been reported in the NSFG and YRBS.(7, 8) This may be because our sample came from clinics that provided the full scope of contraceptive care in a teen-friendly environment that had Title X support, which allowed for both financial coverage and confidential care. This may underlie some of the differences compared to the more general populations surveyed by the NSFG and YRBS. Studies have also demonstrated that familiarity with contraceptive methods enhances their acceptability.(14) This may factor into our contraceptive method mix because young people developmentally rely heavily on others, and having a friend or family member who was using a method may reduce hesitancy and increase use of that method.

As in other studies, individuals in our study tended to continue IUDs and implants for longer than shorter acting methods. This is consistent with a systematic review and meta-analysis of IUD and implant use (twelve studies including nearly 5,000 females under 25 years) demonstrating a high continuation rate at one year (84% overall; 74% IUD and 84% implant).(15)

Contraceptive switching was common in our sample and is not in itself problematic. Indeed, one analysis suggests that contraceptive decision-making is a journey and not necessarily a straight line between contraceptive goals and intention and contraceptive use.(16) Several factors have been associated with contraceptive discontinuation in young people. The most frequent reasons for dissatisfaction resulting in request for removal are pain and bleeding for the IUD and bleeding for the implant.(17–21) Lunde and colleagues found that adolescents and young women recall being counseled on the potential side effects of the contraceptive implant, but they felt unprepared when they experienced them. Importantly, half of participants in that study who discontinued the implant did not initiate another method despite a desire not to become pregnant.(20) In our sample, discontinuation due to a stated desire to seek pregnancy was rare. Most people initiated a highly effective method of contraception after discontinuation of a different a highly effective method. However, the incidence of pregnancy after switching from a Tier 1 method to a Tier 2 method was high and was higher than when switching from a Tier 3 to a Tier 2. This may be due to increased burden to use, take, or get the method. Individuals may not be prepared to problem solve logistical challenges or navigate side effects for a new method. Sharing this information with healthcare professionals and patients, alike, may be valuable and serve as a conversation starter for how to optimize use of a new method after switching.

Pregnancy desires among young people are complex and are influenced by developmental, emotional, cultural, and structural factors. While young people are not monolithic in their goals for pregnancy, the large majority do not want to become pregnant soon.(22) This substantiates that our patients, who are largely individuals of color and lower income, disproportionately experience unwanted or mistimed pregnancy, birth, and birth outcomes. In our study, individuals over 18 years and those who had had a previous pregnancy had a higher rate of pregnancy after switching a contraceptive method. It is unclear if this reflects different pregnancy goals with regards to life course or unique challenges to contraceptive access or use. Regardless, it underscores the need to better understand and support young people in achieving their reproductive goals. This may include enhancing support of prescribed methods as well as proactively encouraging effective use, and possibly combination, of non-prescribed methods. This has always been a critical need, but it is even more urgent now in the setting of restrictive reproductive health policies.

Insights gained from this study exploring the natural history of contraceptive use in adolescent populations may help clinical providers to offer nuanced and patient-centered contraceptive counseling and care to young people. Patient-centered contraceptive counseling (PCCC) is assessed from the patient perspective and includes domains like interpersonal connection, adequate information, and decision support.(23) It reflects the “how” of contraceptive counseling rather than tangible outcomes like method selection. While this chart review was unable to assess PCCC measures objectively, PCCC measures are now available, could be incorporated into routine care and documentation in the medical record, and would be a valuable addition to future assessments of contraceptive patterns in young people. Information from this study could be incorporated into PCCC by creating a connection with young people when they are initiating and/or discontinuing contraceptive methods, seeking to understand their goals, perceptions, and realities, and partnering with them to help them create solutions that can work for them. Attentive, non-coercive counseling that provides information about what the data show and simultaneously incorporates patient priorities, experiences and preferences may help young people navigate their contraceptive journeys in a way that aligns with their goals.

There are several limitations to consider when interpreting these findings. First, as a chart review, we are limited to the documentation provided in the medical record, which may or may not be complete. Documentation is subject to the biases of the person documenting, the patient-medical professional power imbalances, and social desirability bias. We must also rely on the prescription or medication fill information that is in the medical record. Due to the window of review, individuals contributed different amounts of time during the study period which is important in interpreting these results. Further, ours is not a closed medical system, so patients could have received contraceptive care or pregnancy diagnoses in settings that would not have been available in our review. Next, the documentation of condom, other non-prescription, and behavioral methods was variable, which limited description of these methods as primary or dual methods. Finally, we do not have measures of sexual activity, and the extent to which abstinence was employed by our patients is unknown.

There are several important clinical take-aways from this study. Employing patient-centered counseling that incorporates contraceptive experiences, facts, and space for personalization and change may be valuable for young people. Clinicians should support young people in self-reflection and exploration and accept that the contraceptive journey for everyone may look unique.(16) Importantly, successful navigation of contraceptive switches may require additional attention, education, and strategy. This could include anticipatory hypothetical problem solving, close follow-up, and telehealth or virtual care. Taken together, these steps may help young people achieve their reproductive health goals and work towards equity in reproductive health outcomes.