CONTRACEPTIVE USE AND PREGNANCY IN CYSTIC FIBROSIS: SURVEY FINDINGS FROM 10 CYSTIC FIBROSIS CENTERS

Abstract

Background:

Reproductive life planning is key, now that people with cystic fibrosis (pwCF) may live into their 60s. This study explores contraceptive use, pregnancy trends, and whether concomitant cystic fibrosis transmembrane conductance regulator (CFTR) modulator therapy reduces contraceptive effectiveness.

Methods:

Females with CF aged 18–45 years from 10 U.S. CF centers completed a self-administered web-based questionnaire. Pregnancy rates were calculated by year from age of menarche to survey completion, and linear-mixed models with a logit link detected associations with contraception and modulator use.

Results:

A total of 561 pwCF (median age of 29 years [IQR 24.9–35.8] years) completed the survey. Most participants (n=499, 89%) used modulators, and almost all (n=555, 99%) used contraception. Condoms (n=448, 80%) and oral contraceptive pills (n=363, 65%) were the most prevalent methods used. One-third (n=189, 34%) reported ever being pregnant. Of those reporting pregnancies (n=319), about half (n=151, 48%) were unintended. Pregnancy was significantly associated with age (20–29 years or 30–39 years), partner cohabitation (aOR 21.5, 95% CI 5.1 to 91.1), and non-hormonal contraceptive use (aOR 5.1, 95% CI 1.1 to23.0). Among pwCF cohabitating with a partner, modulator use was positively associated with pregnancy (OR 1.8, 95% CI 1.3 to 2.6) (p = 0.0008).

Conclusions:

Despite almost universal contraceptive use, unintended pregnancy among pwCF is common. Likelihood of pregnancy is further increased among CFTR modulator users who are partnered. CFTR modulators themselves do not appear to decrease hormonal contraceptive effectiveness. Patient education about contraception is an increasingly critical aspect of CF care.

1. Introduction

Since the United States (U.S.) Food and Drug Administration (FDA) approved elexacaftor/tezacaftor/ivacaftor (ETI) in late 2019, variant-specific cystic fibrosis transmembrane conductance regulator (CFTR) modulator therapies are used by approximately 89% of variant-eligible people with cystic fibrosis (pwCF) [1]. Such medical advances have led to predicted life expectancy for pwCF into their 60s, making reproductive health care a top priority. Studies have established that pwCF initiate sexual activity at similar rates to the general population, but with a lower likelihood of utilizing contraception compared to their healthy peers [2, 3]. In CF, unintended pregnancy is associated with increased adverse outcomes, such as more pulmonary exacerbations for the pregnant person and potential complications in their infants [4, 5]. Historically, up to 35% of the CF female patient population has been considered subfertile, compared to 5–15% in the general population [6]. However, recent case series in which spontaneous conception occurred in people with a history of infertility within 2 months suggest that variant-specific modulator therapy normalizes fecundity due to multifactorial mechanisms (including potentially through correction of the underlying CFTR defect in the cervix) [7, 8]. Nevertheless, CF is still a complex medical condition in which pregnancy planning is recommended to optimize nutrition and lung function, screen for pre-existing CF-related diabetes, and offer genetic testing and counseling [9].

Recent surveys among pwCF and CF providers suggest a need for more research on pregnancy and contraception to better empower pwCF around their reproductive autonomy [10–12]. Contraceptive methods have been traditionally listed according to their effectiveness, with male and female sterilization and long-acting reversible contraceptives, such as intrauterine devices (IUDs), as the most effective, followed by moderately effective hormonal methods such as oral pills, transdermal patches and vaginal rings, and less effective non-hormonal methods, such as condoms, spermicides and coitus interruptus [13]. U.S. national contraceptive guidance suggests that hormonal contraceptives are safe and effective in CF [4, 14, 15]. However, evidence regarding the safety of hormonal contraception with concomitant modulator use is lacking. Ivacaftor/lumacaftor may reduce hormonal contraceptive effectiveness [16]. ETI has not been associated with reduced hormonal contraceptive effectiveness, but pharmacokinetic modeling regarding drug interactions is limited to drug product labeling [17].

This multi-center study expands on a prior study we conducted at three CF centers [18]. We explore contraceptive use, pregnancy trends, and whether concomitant variant-specific modulator therapy is associated with reduced hormonal contraceptive effectiveness among pwCF at 10 CF centers. Understanding these trends is essential to meeting the family planning needs in the CF population.

2. Methods and Materials

2.1. Study design

We conducted a multisite, retrospective, observational cohort, web-based survey between January, 2021 and April, 2022 that measured self-reported contraceptive method use and lifetime incidence of pregnancy. For quality control purposes, we randomly selected approximately 10% of the total number of participants to complete a follow-up research staff-administered interview to better inform imputation procedures of missing data.

2.2. Study Population

Participants included reproductive-aged women recruited from 10 separate CF centers in the U.S., including Boston Children’s Hospital/Harvard University, Johns Hopkins University, Mount Sinai Beth Israel, National Jewish Health, University of Alabama at Birmingham, University of California San Diego, University of Pennsylvania, University of Pittsburgh, University of Texas Southwestern, and the University of Washington. Participants were considered eligible for the study if they identified as female, were between the ages 18–45 years at the time of consent, had a CF diagnosis, and were enrolled as participants of the Cystic Fibrosis Foundation Patient Registry (CFFPR). We excluded persons who were unable to read or understand English, did not have a working email address, or had a history of lung transplantation. CF Center research coordinators identified potential participants using electronic medical records (EMR) and approached them through email, phone, or in-person during clinic appointments. The participants received study information about the study purpose, procedures, risks, benefits, and study team contact information. This study was reviewed by the research ethics committees at each of the participating institutions and was granted expedited approval or deemed exempt.

2.3. Data collection methods

2.3.1. REDCap

We collected and managed all study data using the electronic data capture tools hosted at the University of Washington (UW). Data Capture (REDCap) is a secure, web-based application designed to support data capture for research studies [20]. We pretested the contraception and pregnancy survey questions among 50 women ages 18–45 years with and without CF using a 3-tiered approach, which has been described elsewhere [21].

Interested participants were sent a unique REDCap survey link via email. Before completing the survey, participants responded to an eligibility survey to verify they met the study criteria. Each participant electronically provided informed consent after reviewing the study information. The web-based survey had six sections, including medical, medication, surgical, menstrual, contraceptive, and pregnancy histories. The questionnaire, which has been published previously, asked about smoking and blood clots, CFTR modulator use, menarche and current menstrual status, reproductive surgical history, contraceptive method use, and prior pregnancies [21]. Questions related to contraception included pictures and brand names for each method as prompts and queried lifetime use, followed by a breakout of specific time intervals (month and year of start and stop dates) to determine total duration and amount of overall between methods. Pregnancy history questions included whether the respondent had ever been pregnant, how many times and the outcome of each pregnancy and whether the pregnancy was intended or not. We defined an unintended pregnancy as one that occurred either too soon or later than wanted or was unwanted [22]. Participants received a $30 electronic gift card for completing the online survey.

2.3.2. Cohort comparisons

We used data from the CFFPR to compare 561 participants and 982 non-participants to understand representativeness of our study sample (Table 1). Non-participants consisted of females with CF age 18–45 years at the same 10 CF centers included in this study in the year 2020 and are reported in aggregate. We also used the CFFPR to capture and merge some characteristics from each year (2008 to 2019) for survey study participants: sociodemographics, BMI, and pregnancy in prior reporting year. Additionally, in Table 3, we compared survey participant pregnancy outcomes to published 2019 estimates from the National Survey of Family Growth [(NSFG), a survey of the U.S. population aged 15–49 years] pregnancy history data, which is the most recent year for which pregnancy outcomes were reported and captures the years (1996–2022) [23].

Table 1:

Participant and non-participant demographic and clinical characteristics at baseline at 10 CF Centers^*

Sociodemographic Characteristics	Participants (n = 561)	Non-participants (n=982)
Age
Median age in years (IQR)	29.9 (24.9, 35.8)	29.7 (25.1, 35.2)
<20 years	21 (3.7%)	N/A
20–29 years	262 (46.7%)	N/A
30–39 years	216 (38.5%)	N/A
40+ years	62 (11.1%)	N/A
Race/ethnicity
Race: White	510 (90.9%)	930 (94.7%)
Ethnicity: Hispanic	28 (5%)	48 (4.9%)
Any other race/ethnicity	6 (1.1%)	51 (5.2%)
Mixed race	13 (2.3%)	N/A
Insurance
Private	401 (71.5%)	650 (66.2%)
Public (Medicare, Medicaid, Indian Health Service, state, TriCare, other military	188 (33.3%)	362 (36.9%)
Other or no insurance	11 (2.0%)	30 (3.1%)
Parents’ insurance plan	117 (20.9%)	155 (15.8%)
Patient assistance program	190 (33.9%)	335 (34.1%)
Education
Some College, High school graduate or GED, Some high school or less	194 (34.6%)	415 (42.3%)
College/graduate degree	366 (65.2%)	477 (48.6%)
Prefer not to answer	1 (0.2%)	N/A
Employment
Full time employment	203 (36.2%)	370 (37.7%)
Any other employment	331 (59.0%)	630 (64.2%)
Marital Status (at time of survey completion)
Married/partnered/living together	270 (48.1%)	478 (48.7%)
Any other marital status	283 (50.4%)	504 (51.3%)
Prefer not to answer	8 (1.4%)	N/A
Medical History
Body Mass Index (BMI)
Annualized median BMI (IQR)	23.2 (21.4, 25.9)	22.6 (20.8–25.3)
Percent Predicted Forced Expiratory Volume (FEV1pp)
Annualized median FEV1pp (IQR)	83.8 (66.3, 99.7)	78.5 (58.6–95.5)
CFTR Genotype (Delta F508)
Homozygous F508	255 (45.5%)	N/A
Heterozygous F508	227 (40.5%)	N/A
Other mutation	72 (12.8%)	N/A
CFTR Modulator use
Current CFTR modulator use (any)	499 (88.8%)	N/A
Kalydeco = Ivacaftor (VX770)	18 (3.2%)	N/A
Orkambi = Ivacaftor (VX770) + Lumacaftor (VX 809)	< 5	N/A
Symdeko = Ivacaftor (VX770) + Tezacaftor (VX661)	9 (1.6%)	N/A
Ivacaftor(VX770) + Tezacaftor (VX661) + Elexacaftor (VX445)	472 (84.1%)	N/A
Unknown	63 (11.2%)	N/A
Reproductive Health History
Menarche (median age, IQR)	13 (12, 14)	N/A
Ability to become pregnant at the time of survey		N/A
Currently regularly menstruating**	407 (72.6%)	N/A
Tubal ligation or hysterectomy	18 (3.2%)	N/A
Same sex partner or partner with vasectomy	73 (13.0%)	N/A
Pregnancy History
Pregnant in reporting year	45 (8.2%)	82 (8.4%)
Pregnancy: live birth outcome	26 (4.6%)	41 (4.2%)
Smoking History
Have smoked	53 (9.4%)	N/A
Never	508 (90.6%)	N/A

^*CFF Patient Registry was used for variables comparing participants and non-participants. Non-participants consist of females with CF age 18–45 years at same 10 CF centers enrolled in this study.

^**Over 80% of the women not menstruating reported being currently pregnant or using birth control, with no persons identifying as post-menopausal. Data of participants in which non-participant information is “N/A” are from the self-respondent survey.

Table 3.

Pregnancy history of study participants / females with CF, N=561 compared to general U.S. population of females, aged 15–44 years in 2019

	Participant survey Responses n (% of 561)	General U.S. population -Females aged 15–44 years in 2019
Pregnancy history among participants
Women who were ever pregnant (N=190)	189 (33.7%)
1 pregnancy	108 (19.3%)
2 pregnancies	51 (9.1%)
3 or more pregnancies	30 (5.3%)
Women who had at least one unintended pregnancy	105 (18.7%)
Pregnancy outcomes among 189 participants who had at least one pregnancy in their lifetime, N=319 pregnancies	N (% of 319 pregnancies)	% of 5.5 million pregnancies in 2019
Live birth	203 (63.8%)	66.9%
Stillbirth	0	20.0%*
Miscarriage	67 (21.1%)
Ectopic or tubal pregnancy	3 (0.9%)
Abortion	37 (11.6%)	13.1%
Prefer not to answer	8 (2.5%)	N/A
Unintended pregnancies (too soon or later than wanted or unwanted)	151 (47.5%)	41.6%

^*Includes miscarriage, ectopic pregnancy and stillbirth.

2.4. Quality Control Interviews

For quality control purposes, we randomly selected ~10% of participants (n=45) to participate in a follow-up interviewer-administered, HIPAA-compliant phone interview two weeks after completing the self-respondent, web-based survey. Quality control questions repeated the same surgical history and contraception as in the self-respondent web-based survey to examine whether recall might be impacted by the presence of an interviewer [21]. Participants who completed the follow-up interview received an additional $20 electronic gift card.

2.5. Statistical analysis

The focus of the current analysis was on contraceptive use and pregnancy trends in the setting of widespread CFTR modulator use. We first described “ever use” of each contraceptive, including medians and interquartile ranges, counts, and proportions, when appropriate. For the >90% of persons who were not selected for quality control interviews, we used multiple imputations to extrapolate the non-observed responses to the interviewer-administered questionnaire. We compared confidence intervals for the proportion of persons who might have used each contraceptive method to those obtained from the self-administered survey. We used multiple-imputation using chained equations (MICE) to estimate prevalence of contraceptive use using both logistic regression and predictive-mean matching [24]. All analyses were performed in R (version 4.4.1).

For purposes of assessing risk of pregnancy, we described contraceptive method use as currently occurring in each year of potential use, starting from menarche. This approach was used as some persons reported that their dates of use were approximate (only known to the year). We associated each year of potential pregnancy with the contraceptive use in the current and prior year: (1) on hormonal contraceptives in either year, (2) on non-hormonal contraceptives only during both years, or (3) not using contraceptives during either year. We included years at risk from reported age of menarche to year of survey completion, and also recording pregnancy outcome (e.g. live birth, termination, miscarriage, or ectopic). We considered being “at risk for pregnancy” as the time period in which a person was potentially fertile, regardless of contraceptive use (since no method of contraception is considered 100% protective against pregnancy, including sterilization) [13]. We considered “not being at risk for pregnancy” as persons who had undergone a hysterectomy, practiced abstinence, or were menopausal. Of the potentially fertile years, we removed any reporting years with no risk of pregnancy. For the portion at risk of pregnancy over time, we removed individuals for whom no CFFPR data were available (N=9).

We performed generalized linear mixed modeling with a logit link to determine whether the pregnancy odds were associated with age, CFTR modulator use, partner status, or contraceptive use in the current or previous year. We examined the models for all potential two-way interactions between CFTR modulator use, partner status, and contraceptive methods. We included year of age in decades, but made no assumptions regarding linearity, and we categorically included decade of age. Unknown partner status mainly refers to years prior to 2008, a period for which we did not request CFFPR data.

3. Results

A total of 561 participants from 10 CF centers completed the self-respondent, web-based survey, and 45 (8% of respondents) completed a quality-control survey (Figure 1). Sociodemographic characteristics were largely similar to the population of persons participating in the CFFPR;, however, we performed no statistical testing (Table 1).

Figure 1:

CASE.4.CF Study flow diagram

Almost all survey participants (n=555, 99%) reported “ever” using at least one type of contraceptive method, although hormonal contraceptive use was less prevalent than non-hormonal contraceptive use (74% and 96%, respectively) (Table 2). Over the fertile lifespan, combined oral pill use was the most prevalent hormonal contraceptive (n=363, 65%; 95% CI for ever-use: 53% to 78%), and its median use in years appeared longest (4.0 years, IQR 2–8.4 years). Hormonal emergency contraception pills were the second most commonly used method (38%), followed by the hormonal IUD (26%). Condom use was the most prevalent non-hormonal method (n=448, 80%; 95% CI for ever-use 68% to 95%), and its median use in years appeared longest among all non-hormonal methods (6.1 years, IQR 3.1–11.1 years). Coitus interruptus (withdrawal) closely followed condom use, with a prevalence of 58% (95% CI 40% to 85%), with a median use of 4.6 years (IQR 1.7 to 9.7 years). The confidence limits in Table 2 for the imputed proportion of persons using each contraceptive method using interviewer-administered surveys includes the reported proportions from the self-administered surveys. This suggests we found no evidence of social acceptability or recall bias, although with only 8% of the cohort interviewed, our confidence intervals are relatively wide.

Table 2.

Contraceptive use among study participants between first period and 2021, N=561

				Interview respondents (N=45)
	Ever use N (%)	Number of episodes of use (N, % with exact duration)*	Length of use (median (IQR) in years)	Ever use N (%)	Imputed** 95% confidence interval for ever use
Contraception used
Any use of some form of contraception	555 (98.9%)			43 (95.6%)
Contraceptive method used
Any hormonal method	417 (74.3%)	--	--	38 (84.4%)	--
Combined pills	363 (64.7%)	464 (142, 30.6%)	4.0 (2.0,8.4)	33 (73.3%)	(52.9%, 77.6%)
Mini-pill	63 (11.2%)	73 (20, 27.4%)	1.5 (0.9, 3.0)	4 (9.1%)	(2.7%, 27.4%)
Emergency contraceptive pills	213 (38.0%)	--	--	17 (37.8%)	--
Hormonal IUD	148 (26.4%)	159 (107, 67.3%)	3.0 (1.4, 5.3)	9 (20.0%)	(19.9%, 37.3%)
Arm implant (Nexplanon)	41 (7.3%)	42 (23, 54.8%)	2.0 (0.7, 3.0)	3 (6.7%)	(0.1%, 81.8%)
Contraceptive patch	37 (6.6%)	37 (10, 27.0%)	1.7 (1.0, 2.5)	2 (4.4%)	(1.4%, 19.4%)
Vaginal ring	78 (13.9%)	85 (19, 22.4%)	2.0 (1.0, 5.8)	5 (11.1%)	(0.1%, 79.7%)
Depo shot	82 (14.6%)	90 (27, 30.0%)	2.0 (1.0, 5.0)	7 (15.6%)	(8.1%, 37.4%)
Any non-hormonal method	536 (95.5%)	--	--	42 (93.3%)	--
Condoms	448 (79.9%)	532 (142, 26.7%)	6.1 (3.1, 11.1)	39 (86.7%)	(68.3%, 94.7%)
Copper IUD	63 (11.2%)	67 (39, 58.2%)	3.9 (1.9, 5.6)	2 (4.5%)	(0.1%, 50.6%)
Female sterilization (tubal ligation or Essure)	7 (1.2%)		7.7 (7.2, 13.0)	1 (2.2%)	(0.1%, 22.4%)
Partner sterility	73 (13.0%)	82 (39, 47.6%)	4.1 (1.2, 7.3)	11 (24.4%)	(8.2%, 41.0%)
Diaphragm/cervical cap	7 (1.2%)	8 (2, 25.0%)	4.0 (3.2, 6.8)	0 (0.0%)	--
Spermicide/sponge	23 (4.1%)	23 (5, 21.7%)	2.5 (2.0, 8.2)	2 (4.5%)	(1.3%, 5.3%)
Withdrawal	324 (57.8%)	345 (72, 20.9%)	4.6 (1.7, 9.7)	31 (68.9%)	(40.1%, 84.8%)
Fertility awareness method	155 (27.5%)	162 (47, 29.0%)	3.0 (1.4, 6.5)	14 (31.1%)	(8.6%, 64.2%)
Breast feeding	25 (4.5%)	35 (29, 82.9%)	1.0 (0.5, 1.7)	0 (0.0%)	--
Abstinence	161 (28.7%)	180 (49, 27.2%)	6.0 (2.0, 12.8)	8 (18.2%)	(0.9%, 86.5%)

^*Among number of episodes of use across all persons, the number and proportion with start and stop date known to the month and year are provided; and episodes ongoing at time of survey were included if start month and year were known.

^**Imputed confidence interval for ever use in interview is computed from imputed datasets on use in the full 561 participants. Multiple imputation from chained equations (MICE) is performed using parametric (logistic regression) and non-parametric methods (predictive-mean matching) and the widest endpoints of the two intervals were retained.

The youngest age at which a pregnancy occurred was 15 years, and the oldest was 42 years, with the oldest participant being 45 years of age at the time of survey completion. One in five participants reported having at least one unintended pregnancy. The pregnancy outcomes of 189 participants (34%) who had at least one pregnancy in their lifetime resulted in a total of 319 pregnancies (Table 3). Of the 319 pregnancies, 203 (64%) resulted in live births, and 151 (48%) were reported as unintended. The percentages of live births and unintended pregnancies in our survey participants were similar to 2019 U.S. national data estimates (Table 3). Appendix A highlights trends in pregnancy among participants, whether assisted reproductive technology (ART) was used or pregnancy was unintended.

Overall, in 9572 person-years at risk of pregnancy, the prevalence of pregnancy appeared to be higher at older ages, among those with a partner, for those using CFTR modulators, and for those using either hormonal or non-hormonal contraceptives. (Table 4). Compared to pwCF aged <19 years, pwCF who were between age 20–29 or 30–39 years had significantly about 5 times higher adjusted odds for pregnancy. Those not using any contraception in the current or previous year [who were younger (median age 17 versus 23) and less likely to be partnered (9.6% versus 22.8%) than those using contraception]had the lowest incidence of pregnancy 1.4% (n=51/3568).

Table 4.

Regression results for occurrence of pregnancy, over all reported history from date of first period, including 9572 person-years at risk. Both CFTR modulator use and contraceptive use were differently associated with pregnancy depending on the partner status of the participant. So the multivariable adjusted regression results include their associations for each level of partner status, with reference groups separated by horizontal lines.

Characteristic	Prevalence of pregnancy	Associations in logistic regression
	% (n/years at risk for pregnancy)	Odds ratio (95% CI), p-value		Adjusted* Odds Ratio (95% CI), p-value
Age
Age first period to 19	0.5% (17/3527)	Ref		Ref
Age 20 to 29	3.8% (160/4199)	8.5 (5.1, 14.10), p<0.0001		5.3 (3.1, 9.2), p<0.0001
Age 30 to 39	6.9% (114/1658)	16.6 (9.8, 28.1), p<0.0001		5.6 (3.1, 10.2), p<0.0001
Age 40 to 45	2.3% (4/171)	5.9 (1.9, 18.2), p=0.0021		1.7 (0.5, 5.6), p=0.35
Partner status
No partner	0.9% (25/2672)	Ref		Ref
Partner	8.6% (167/1949)	9.6 (6.2, 14.8), p<0.0001		21.5 (5.1, 91.1), p<0.0001†
Unknown	2.1% (103/4951)	2.1 (1.3, 3.2), p=0.0016		4.3 (1.0, 18.6), p=0.053+
CFTR modulator use **
None	3.1% (185/6787)	Ref
Some use	6.1% (102/1810)	2.7 (2.1, 3.6), p<0.0001
		Partner status CFTR mod use
		No partner	None	Ref
		No partner	Some use	0.6 (0.2, 1.7), p=0.30
		Partner	None	Ref
		Partner	Some use	1.9 (1.3, 2.7), p=0.0005
		Unknown	None	Ref
		Unknown	Some use	1.5 (0.9, 2.4), p=0.10
Contraceptive use **
None	1.4% (51/3568)	Ref
Hormonal use	3.3% (118/3546)	2.5 (1.8, 3.6), p<0.0001
Only Non-hormonal use	5.1% (126/2458)	4.1 (2.9, 5.8), p<0.0001
		Partner status Contraceptive use
		No partner	None	Ref
		No partner	Hormonal	3.2 (0.7, 14.5), p=0.12
		No partner	Only non	5.1 (1.1, 23.0), p=0.034
		Partner	None	Ref
		Partner	Hormonal	0.7 (0.4, 1.1), p=0.16
		Partner	Only non	0.9 (0.6, 1.5), p=0.81
		Unknown	None	Ref
		Unknown	Hormonal	2.2 (1.2, 3.9), p=0.010
		Unknown	Only non	3.0 (1.7, 5.6), p=0.0003

95% CI =95% Confidence Interval

^*The adjusted model includes all other variables present.

^**Use was considered ‘yes’ if it occurred in the same year or the previous year as the date the pregnancy ended.

^†Persons with a partner or with unknown partner status had higher prevalences of pregnancy than those without a partner at every level of modulator use and contraceptive use, though interactions are omitted for space and simplicity. The odds ratios shown are for those neither on modulators or contraception.

We subsequently found (through adjustment and by including interactions) that associations with pregnancy differed by whether the participant had a married or cohabitating partner. In persons without partners or with unknown partner status, contraceptive users tended to have higher pregnancy rates; while in married persons this relationship was not found. For example, among persons without partners, non-hormonal only method users had significantly higher adjusted odds of pregnancy compared to non-users (aOR 5.1, 95% CI 1.1 to 23.0), whereas among married persons, hormonal contraceptive users had a trend of lower adjusted odds of pregnancy (aOR 0.7, 95% CI 0.4 to 1.1, p=0.16). Similarly, CFTR modulator use was found to be associated with pregnancy within persons who were married or cohabitating (aOR 1.9, 95% CI 1.3 to 2.7), but conclusive association was not determined for persons of other partner statuses (Table 4). Concomitant CFTR modulator and current hormonal contraceptive use did not result in more pregnancies (Appendix B).

4. Discussion

This large retrospective survey study found that almost all (99%) females with CF report having used contraception at least one time during their fertile years. Considering this commonality and its importance in CF care, members of the CF community should consider adding contraceptive use to the CFFPR. The most popular methods among participants in this study includes condoms (80%), combined estrogen and progestin oral pill (65%), and coitus interruptus (withdrawal) (58%). Compared to the U.S. general population of females aged 15–49 years who had ever had sexual intercourse with a male partner and were interviewed between 2015–2019 NSFG (similar years in which our participants reported their contraceptive use), we found that the top three “ever use” were the same (condoms, followed by combined oral pill and coitus interruptus) [25]. Given the underlying medical complexity of CF disease, there are some differences between the CF and general population that are worth noting. While only 1 in 4 in the general U.S. population use highly effective long-acting reversible contraceptive (LARC) methods, almost 1 in 2 in the CF population are ever using these methods, suggesting that some pwCF may be following the requirements for contraception during CFTR modulator trials or some are getting comprehensive contraceptive counseling and are choosing methods that help them avoid unwanted pregnancy in light of their complex disease. On the other hand, more pwCF report ever using emergency contraception (38.1%) compared to the general population (23.5%), suggesting that this population may have difficulty accessing their chosen contraceptive method. Additionally, very few pwCF reported female sterilization as a method (1.2%) compared to the general population (21.3%). This stark difference in sterilization between pwCF and the general population may suggest that most females with CF may consider themselves already subfertile or reflect the reluctance of healthcare providers to perform the procedure in those with complex disease [26]. Because of the relatively new phenomenon of ease of achieving pregnancy and longer life expectancies in the era of CFTR modulators, female sterilization trends as a chosen method may rise among pwCF.

Among pwCF, pregnancy was more prevalent in persons aged 20–39 years, and married/cohabitating persons. Using contraceptives was associated with pregnancy in unpartnered persons, which is consistent with the U.S. general population [13, 27]. Although our findings may not necessarily be novel, they are to our knowledge, the first to report population-based contraceptive use among pwCF in the U.S.. Consistent with previous findings in this population, we found that contraceptive use among participants is a marker of exposure to sexual activity [19]. While this study did not find that concomitant use of CFTR modulators and hormonal contraceptives reduces contraceptive effectiveness, we did find that when CFTR modulators are used among pwCF with higher exposures to sexual activity (i.e. cohabiting with a sex partner), the odds of pregnancy almost doubled. Of concern, almost half of all pregnancies (47%) reported by our survey participants were unintended.

National trends are moving away from categorizing a pregnancy as either intended or unintended, as pregnancy experiences can be vastly different [22]. Nonetheless, pregnancy in CF was historically considered high risk. Case series and retrospective studies suggest pregnancy contributes to lower lung function, increased infections, and CF-related diabetes compared to pwCF who do not become pregnant [28]. Although more females with CF desire to have a child and prognosis for pregnancy has improved for many, preconception planning is considered essential part of CF care [2,9]. Thus, in the era of prevalent CFTR modulator therapy use, CF centers should recognize the importance of reproductive autonomy by integrating reproductive life planning discussions that center patient values and preferences for pregnancy prevention or planning in the context of their complex disease [11]. Starting discussions early in adolescence is important in this population, especially in light of findings suggesting that the mean age at first sexual intercourse is 18 years old, similar to that in the general population [2]. This body of research supports the increased need for improved continuing education for CF clinicians regarding family planning and pregnancy-related topics.

This study has several strengths, including pre-tested and quality-control survey questions, thereby increasing the reliability of our findings [21]. This study is also one of the largest and most comprehensive studies evaluating contraceptive use and pregnancy history in this population and compares participants and non-participants, which strengthens its generalizability.

The study also has limitations. Participants reported contraceptive use going as far back as 13 years, which may have contributed to recall bias. We attempted to reduce recall bias by including pictures and brand names of many contraceptive methods. Nevertheless, it is possible that the methods available in 2008 were not included in our visual list of currently available pills. The timing of the contraceptive method use relative to the period of conception is also approximated, as the majority of contraceptive use episodes were recalled only to the year, and not the month. However, the consistency between self-administered and interviewer-administered survey results demonstrates some validity. Due to the abrupt onset of the COVID-19 pandemic as we initiated study enrollment, recruitment occurred mostly electronically, rather than in-person, and thus we enrolled fewer participants than anticipated. Our study population has less racial diversity and higher education, socioeconomic status (as indicated by private insurance) and health status (higher FEV1 and BMI) than the general population of pwCF, possibly limiting applicability to all pwCF in the U.S.

This study adds to the growing evidence that suggests the importance of addressing reproductive health among females with CF starting before age 18 years and optimizing contraceptive counseling. Patient-centered comprehensive contraceptive counseling and provision should be a routine part of CF care. Further research should evaluate whether such care should be implemented at the CF center level or whether better coordination is needed between CF care centers, family planning, and primary care providers.

Highlights.

• 99% of participants report having used contraception at least once

• Despite having access to contraception, nearly half of pregnancies were unintended

• Likelihood of pregnancy is further increased among CFTR modulator users who are partnered

• CFTR modulator use was not associated with reduced hormonal contraceptive effectiveness