Perceptions and Behaviors Related to Contraceptive Use in the Postpartum Period Among Women with Pregestational Diabetes Mellitus

Abstract

Objective:

To describe perceptions and behaviors related to contraception and preconception care and to test the association between these perceptions and contraceptive use in the postpartum period among women with pregestational diabetes mellitus.

Design:

Cross-sectional, descriptive survey.

Setting:

Three high-risk obstetric clinics in the Southeastern United States.

Participants:

Fifty-five women who were 18 years or older with pregestational type 1 or type 2 diabetes mellitus.

Methods:

Between 4 and 8 weeks after birth, we used investigator-developed items and psychometrically validated scales to measure participants’ perceptions and behaviors related to contraception and preconception care. We dichotomized use of contraception in the postpartum period as procedure/prescription or non-prescription/no method. We tested the hypothesis that perceptions are associated with contraceptive use using multiple logistic regression.

Results:

When data were collected 4 to 8 weeks after birth, almost half (49%, n = 27) of the participants had resumed sexual activity; however, most (95%, n = 52) did not want another pregnancy in the next 18 months. Fifty-six percent (n = 31) of participants used procedure/prescription contraception, and 44% (n = 24) used non-prescription/no method. Those who perceived contraception use and preconception care to be beneficial were more likely to use procedure/prescription contraception (adjusted odds ratio 1.52, 95% confidence interval, 1.07 – 2.17).

Conclusion:

When caring for women in the postpartum period, providers should be mindful that women’s perceptions of the benefits of contraception and preconception care may have implications for whether their use aligns with their reproductive goals and optimizes outcomes for future pregnancies.

Precis

Among women with pregestational diabetes mellitus, use of a highly effective contraceptive method following childbirth is more likely when they view contraception as beneficial.

The incidence of pregestational diabetes (henceforth referred to as “diabetes” in this article) during pregnancy is rising in the United States (Admon et al., 2017; Bardenheier et al., 2015; Correa, Bardenheier, Elixhauser, Geiss & Gregg, 2014). Between 2000 and 2010, the prevalence of diabetes increased from 0.65 to 0.89 per 100 births (Bardenheier et al., 2015). When blood glucose is elevated during pregnancy, women with type 1 diabetes mellitus (T1DM) or type 2 diabetes mellitus (T2DM) experience higher rates of adverse outcomes, including perinatal loss and fetuses with congenital malformations (Feig et al., 2014; Timar et al., 2014). The American Diabetes Association (ADA, 2019) recommended preconception care, including use of contraceptives, until women with diabetes are ready for pregnancy, ideally with glycated hemoglobin (A1C) levels below 6.5% to minimize the risks of obstetric complications. An estimated $767 million in direct medical costs and $3.6 billion in lost productivity could be saved by preconception care for all women diagnosed with diabetes (Peterson et al., 2015; Wahabi et al., 2010). However, among women surveyed in 10 states of for the 2009 – 2010 Pregnancy Risk Assessment Monitoring System, only 53% of women with diabetes reported that they obtained preconception care when they were asked “Before you got pregnant with your new baby, did you talk with a doctor, nurse, or other healthcare worker to prepare for a healthy pregnancy and baby?” (Kachoria & Oza-Frank, 2014).

During the postpartum period, contraception can prevent unintended pregnancy and support healthy pregnancy intervals while women with diabetes can work to establish euglycemia (Thiel de Bocanegra, Chang, Howell, & Darney, 2014). In non-population-based samples, uptake of contraception after childbirth by women with diabetes ranged from 52% (Schwarz et al., 2017) to 77% (Perritt, Burke, Jamshidli, Wang, & Fox, 2013). Better understanding of behaviors and perceptions that can affect unintended pregnancy risk, including when women resume sexual activity after childbirth or whether they view contraception favorably, can also support efforts to achieve high quality, patient-centered, maternity care.

To identify potentially modifiable targets for nursing interventions, we conceptualized contraceptive use in the postpartum period as a health behavior that could be influenced by three constructs from Pender’s Revised Health Promotion Model. Pender (2011, p. 4) defined perceived benefits of action as “perceptions of the positive or reinforcing consequences of undertaking a health behavior”; perceived barriers to action as “perceptions of the blocks, hurdles, and personal costs of undertaking a health behavior”; and perceived self-efficacy as “judgment of pe rsonal capability to organize and execute a particular health behavior; self-confidence in performing the health behavior successfully.” In this study, we collectively refer to Pender’s three constructs as perceptions. Among women with diabetes who have not recently given birth, high self-efficacy about the use of contraception and obtaining preconception care were associated with their perceived usefulness (Grady & Geller, 2016) and care seeking behavior (Komiti et al., 2013). In qualitative research studies, women with diabetes who perceived limited benefits or substantial barriers to obtaining contraception or preconception care were inclined to obtain neither (Charron-Prochownik et al., 2015; Chuang et al., 2010; Earle et al., 2017; McCorry, Hughes, Spense, Holmes, & Harper, 2012; Murphy et al., 2010; O’Higgins, Mcguire, Mustafa, & Dunne, 2014; Shawe, Smith, & Stephenson, 2011). Therefore, we conducted a survey of women with diabetes in the postpartum period to describe perceptions and behaviors related to contraception and preconception care and to test the association between these perceptions and contraceptive use in the postpartum period among women with diabetes.

Methods

Design

We used a cross-sectional, descriptive survey to measure perceptions and behaviors related to contraception and preconception care. We obtained informed consent from all participants. The Institutional Review Board of the University of North Carolina at Chapel Hill approved all study procedures.

Participants and Setting

We surveyed adult women who had diabetes before recent pregnancies. We used the following inclusion criteria for participation: age 18 or older, diagnosis of T1DM or T2DM before pregnancy, and consent to complete a survey in English between 4 and 8 weeks after birth. Exclusion criteria were fetal demise, newborn with major malformation, or instructions in the electronic health record or from a provider indicating that the woman should not be contacted.

Participant recruitment and study procedures occurred at three high-risk obstetric clinics in a public not-for-profit integrated health care system that is associated with a Southeastern academic medical center. These clinics provide care regardless of insurance status. Nurse practitioners and obstetricians, including maternal-fetal medicine specialists, provide care during and after pregnancy. Site partners estimated that they served approximately 150 women who met eligibility criteria in the previous year.

We conducted a power analysis to determine our target enrollment. Assuming, as reported by Schwarz, Maselli, & Gonzales (2006), that approximately 30% use procedure/prescription contraception and 70% non-prescription/none, 90 participants would allow for 80% power to detect a medium-large standardized mean difference of d=.65 between the two groups, in a two-tailed t-test with a significance level of .05.

Procedures

We identified potentially eligible women in the system-level, electronic health record that integrates out-patient and in-patient care. We made initial contact in person (at the clinic) or remotely (by phone or email) between 37 weeks gestation and 8 weeks after birth to describe the study objectives and procedures. When a woman expressed interest, we confirmed eligibility, described the study in more detail, answered questions, and confirmed preferred communication mode (phone or email). We obtained informed consent in person or through an online informed consent form on a secure Qualtrics platform; study personnel were available for questions. On site, participants could complete the survey on a study iPad or on paper, or they could follow a link to complete the survey on the Qualtrics platform from their own devices. Participants could also provide consent remotely via a hyperlink. We confirmed the date of birth in the electronic health record to send links to the survey between 4 and 8 weeks after birth. We gave participants who completed the survey $20.

Measures

The survey contained 35 questions, including 13 questions about participant characteristics and 22 items/scales (general perceptions and behaviors; pregnancy planning; perceived self-efficacy, barriers, and benefits of contraception and preconception care; and contraceptive use). There were also two free text fields where participants could provide qualitative responses. We used investigator-developed items unless indicated otherwise.

Participant characteristics.

Participants provided demographic data, including age, race, ethnicity, educational attainment, religion, parity, and health insurance status. Diabetes characteristics included type and age of diagnosis from which we calculated duration.

General perceptions and behaviors.

We asked participants whether they resumed sexual activity since the birth and if so, when and how often. We asked them about their desire for future childbearing and the timing of pregnancy using wording modeled on the National Survey of Family Growth (CDC, 2016). Participants answered a series of single-item questions about perceptions of contraception and preconception care, including whether (a) diabetes influenced their choice of contraception, (b) diabetes caused complications in their recent pregnancies, (c) contraception was safe for women with diabetes, (d) contraception was effective for women with diabetes, (e) women with diabetes needed contraception as much as women without diabetes, and (f) diabetes made it harder for women with diabetes to get pregnant. Items were analyzed individually. Women could agree, disagree, or indicate they did not know. In an open field at the end of the survey, we asked participants if they wanted researchers to know anything else about their experience.

Pregnancy planning.

We used the London Measure of Unplanned Pregnancy (LMUP) scale to measure the degree to which the recent pregnancy was planned (Barrett, Smith & Wellings, 2004). There is an ongoing debate about what affective and cognitive constructs constitute pregnancy intention (Aiken, Borrero, Callegari & Dehlendorf, 2016; Kavanaugh & Schwarz, 2009; Santelli, Lindberg, Orr, Finer & Speizer, 2009). Diabetes can further complicate women’s feelings about trying to become pregnant and using preconception care (McCorry et al., 2012; Murphy et al., 2010; Paiva, 2016). The LMUP elicits multiple dimensions of pregnancy intentions through six items, scored 0 to 2; possible total scores range from 0 to 12. The authors of the LMUP noted that the scale should be used as a continuous variable because cutoffs for a categorical variable had not been adequately validated (G. Barrett, personal communication, February 7, 2017). Higher values indicate that the recent pregnancy was more planned. Initially, the authors of the LMUP developed the scale in Britain with women who were pregnant or recently pregnant and recruited participants at antenatal, abortion, and general practitioner clinics. The Cronbach’s alpha coefficient estimate of internal consistency reliability was 0.92, and other assessments of validity and reliability are described elsewhere (Barrett, Smith & Wellings, 2004). Morof et al. (2012) made minor modifications to the LMUP to adapt the scale to the United States., and reliability and validity were evaluated as sufficient (including Cronbach’s alpha coefficient of 0.78, all item-total correlations above 0.2, and weighted Kappa of 0.72 for the English-language version). In our sample, the standardized Cronbach’s alpha coefficient for the LMUP was 0.88 (n = 54).

Perceived self-efficacy, barriers, and benefits of contraception and preconception care.

We measured perception of self-efficacy, barriers, and benefits of contraception and preconception care with psychometrically validated scales from the theory-based Reproductive Health Attitudes and Behaviors (RHAB) questionnaire designed the scale for use with adolescent women with T1DM (Charron-Prochownik, Wang, Sereika, Kim, & Janz, 2006).

The RHAB Benefits Scale contains four items with total scores that range from 4 to 20, the RHAB Barriers Scale contains five items with total scores that range from 5 to 25, and the RHAB Self-Efficacy Scale contains six items with scores that range from 0 to 60 (Charron-Prochownik, et al., 2006). A higher score indicates a greater endorsement of the three scales’ concepts. On the RHAB Barrier Scale, participants can indicate that an item is not applicable, in which case we used mean imputation to calculate the final score, an analytic approach which the creator of the RHAB approved (D. Charron-Prochownik, personal communication, May 7, 2019). We noted that several items on the RHAB scales reference preconception care, which was defined as “achieving normal blood sugars, obtaining preconception counseling, and using effective birth control” (Charron-Prochownik et al., 2006, p 211).

In the original study, Cronbach’s alpha coefficient estimates of internal consistency were 0.65, 0.72, and 0.65 for the RHAB Benefits Scale, RHAB Barriers Scale, and RHAB Self-Efficacy Scale respectively (Charron-Prochownik et al., 2006). In our sample, the standardized Cronbach’s alpha coefficients were 0.55 (n = 54) for the RHAB Benefits Scale, 0.80 (n = 52) for the RHAB Barriers Scale, and 0.81 (n = 53) for the RHAB Self-Efficacy Scale.

Use of contraceptives.

We queried which contraceptive method(s), if any, participants were using. The procedure/prescription methods included female sterilization (tubal ligation, tubes tied or blocked); male sterilization (vasectomy); intrauterine device without hormones (Paragard); intrauterine device with hormones (Mirena, Skyla, Liletta); etonogestrel implant (Norplant, Implanon, or Nexplanon); birth control pills, patch, or ring; the Depo-Provera injection; diaphragms; or emergency contraception. The non-prescription methods included male or female condoms; withdrawal; vaginal sponge; contraceptive film, suppository, or crème; and lactational amenorrhea. Non-users could provide multiple reasons for using no method. For hypothesis testing, participants were categorized into two groups: prescription/procedure or non-prescription/no method. Participants who selected more than one method were categorized according to the most effective method selected. Non-prescription/no method were categorized together because they are less effective than procedure/prescription methods (except for perfectly executed lactational amenorrhea, which was beyond the scope of this study) and do not require interfacing with the health care system (Hatcher et al., 2018).

Data Analysis

We performed all analyses in SAS, version 9.4 (SAS Institute, Cary, NC, USA). To describe perceptions and behaviors related to contraception and preconception care, we provided descriptive statistics for perceptions and behaviors and tested bivariate associations with contraceptive use. For continuous variables, we reported means/standard deviations, except when not approximately normally distributed, we reported medians/interquartile ranges. For categorical variables, we reported counts/percentages. When we analyzed responses to our questions about general perceptions, participants who agreed were compared to those who disagreed or did not know. To test the bivariate associations for continuous variables, we used two-sample t-tests or, in the case of non-normality, the non-parametric Wilcoxon test of association. For categorical variables, we used the χ² test, or Fisher’s exact test if 25% or more of the cells contained fewer than 5 observations.

To test the associations among perceptions and contraceptive use in the postpartum period, we built 3 models for procedure/prescription contraceptive use with multiple logistic regression using scores for RHAB Benefits Scale, RHAB Barriers Scale, and RHAB Self-Efficacy Scale as the predictors. We tested the hypothesis that contraceptive use was associated with perceived self-efficacy, barriers, and benefits of contraception and preconception care. As determined a priori, we controlled for demographic characteristics (age, race and ethnicity, educational attainment, insurance and type of diabetes), whether sexual activity had been resumed following birth, and LMUP score. We also planned to control for any study variables that were bivariately associated with contraceptive use at the 0.05 significance level. Because of our sample size, we collapsed race and ethnicity, education, and insurance into two-level categories for hypothesis testing. We compared non-Hispanic White participants to participants of all other races and ethnicities based on the data about racial and ethnic disparities in diabetes outcomes and quality of care (Ali, McKeever Bullard, Imperatore, Barker, & Gregg, 2012).

Results

Participant Characteristics

Between June 2017 and September 2018, we identified 120 women in the electronic health record who met the eligibility criteria (Figure 1). We approached or contacted 96 women, of whom 61 agreed to participate. We did not meet our a priori sample size estimate. Fifty-five women completed the survey at a median of 6 weeks postpartum. Participants ranged from 21 to 44 years old, with a median age of 32 years (Table 1). Most participants were multiparous (58%, n = 32), non-Catholic Christians (72%, n = 40), and had completed some college or vocational training (75%, n = 41). Most participants were non-Hispanic Black (35%, n = 19), non-Hispanic White (33%, n = 18), or Hispanic (24%, n = 14). Most participants had insurance (81%, n = 45).

Figure 1.

Flow diagram of participant recruitment. We initially had Institutional Review Board approval to approach women in person, but this was not possible when they did not attend their postpartum appointments. After we obtained approval to contact women by phone or email, we were still unable to contact 29 women who did not respond to email, texts, or calls. We also did not contact women if the nurses providing their care instructed us not to approach them or similar instructions were included in the electronic health record.

Table 1:

Participant Characteristics (N = 55)

Categorical Variable	n	Percent
First birth	23	41.8
Race and Ethnicity
Hispanic, any race	14	25.5
Non-Hispanic, Asian	2	3.6
Non-Hispanic, Black	19	34.5
Non-Hispanic, Other or More than One	2	3.6
Non-Hispanic, White	18	32.7
Religion
Catholic	10	18.2
Christian, not Catholic	40	72.7
Other or None	5	9.1
Educational Attainment
Less than high school	2	43.6
High school graduate	12	21.8
Some college or vocational training	17	30.9
College graduate or more	24	43.6
Health Insurance
Private insurance	33	60.0
Medicaid	12	21.8
None/Don’t Know	10	18.2
Type of Diabetes
Type 1	15	27.3
Type 2	40	72.7
Continuous Variable	Median	Interquartile Range
Age	32	28–37
Age at diagnosis	26	16–31
Years since diagnosis	7	2–13

Diabetes was diagnosed between ages 5 and 44 years, with a median age of 26 years. The median duration since diagnosis was seven years (range less than 1 to 31 years). Almost threequarters (73%, n = 40) of the participants had T2DM, and the remainder (27%, n = 15) had T1DM. Demographic characteristics did not significantly vary by diabetes type, except that participants with T1DM were more likely to be non-Hispanic White (p = 0.02). Diabetes characteristics varied by type. Participants with T1DM were diagnosed younger than those with T2DM (p < 0.0001). Participants received T1DM diagnosis between the ages of 5 and 21 years, whereas participants with T2DM were diagnosed between the ages of 13 and 44 years. Participants with T1DM had diabetes longer than those with T2DM, with a median of 18 years versus 3 years (p < 0.001).

Description of Perceptions and Behaviors related to Contraception and Preconception Care

General perceptions and behaviors.

In Table 3, we report univariate findings of perceptions and behaviors in the postpartum period. Forty-nine percent of participants had resumed sexual intercourse between 4 and 8 weeks after childbirth (n = 27). Resumption occurred between 1 and 7 weeks, with a median of 5 weeks, and frequency was once per week or less for 80% (n = 22) of sexually active participants. Most participants did not want another child (27%, n = 15) or were uncertain (40%, n = 22). Of the 15 participants who did want another child, most (n = 9) desired to conceive within 2 to 5 years.

Table 3:

Perceptions and Behaviors of Women with Pregestational Diabetes in the Postpartum Period (N = 55)

Categorical Variable	n	Percent
Has had sex since the childbirth	27	49.1
Frequency of sex was once a week or lessa	22	81.5
Want to be pregnant again ever
No	18	32.7
Yes	15	27.3
Uncertain	22	40.0
Wants to be pregnant againb
Within 18 months	3	20.0
18 months–2 years	2	13.3
2–5 years	9	60.0
More than 5 years	1	6.7
Thinks about diabetes when making decisions about birth control	20	36.4
Believes diabetes caused complications/problems in index pregnancy	29	52.7
Believes birth control is less safe for women with diabetes	9	16.4
Believes birth control is less effective for women with diabetes	4	7.3
Believes women with diabetes need birth control less than other women	2	3.6
Believes diabetes makes it harder for women to get pregnant	21	38.2
Diabetes management part of preconception care for index pregnancy	9	16.4
Continuous Variable	Median	Interquartile Range
Weeks when sex resumed after childbirth, median a	5	4–6
London Measure of Unintended Pregnancy, median	9	6–11

^aAmong the 27 women who had sex since the birth.

^bAmong the 15 women who wanted to become pregnant again. For hypothesis testing, women who wanted pregnancies within 2 years compared to women who wanted pregnancies in more than 2 years.

Most participants reported that they did not think about diabetes when making decisions about birth control (64%, n = 35) and that they believed diabetes caused complications in their most recent pregnancy (53%, n = 29). Participants who believed they experienced diabetes-related complications were no more likely to think about diabetes when making contraceptive decisions than participants who experienced no complications (p = 0.79). Few participants thought that birth control was less safe (16%, n = 9) or less effective (7%, n = 4) for women with diabetes, or that women with diabetes need birth control less than women without (4%, n = 2). However, 38% (n = 21) thought diabetes makes it harder for women to get pregnant.

Participants were given an open field into which they could share any final thoughts. One participant shared, “Great precounseling [sic] and care before pregnancy contributes to a great pregnancy when diabetic.” Another woman wrote, “Insulin and dieting played a major part in preventing my baby from getting any bigger than he was at birth.” Another described how difficult it was for her to manage her condition financially:

Not having money for my insulin makes everything harder when it’s $200 a vile [sic] for the cheapest one. It’s ridiculous I can’t take care of myself. Not everyone has a family to help them and sometimes things get really hard but I’d do anything for my baby. I try my best everyday [sic] to keep her healthy.

Two participants became pregnant unexpectedly when starting metformin and losing weight, while in contrast, another participant had an unplanned pregnancy at a particularly unhealthy time: “Unfortunately, when I did get pregnant I was in one of my health slumps in which I was not really watching what I was eating and ultimately neglecting my diabetes. This pregnancy was very much unexpected.”

Pregnancy planning.

The median LMUP score was 9, indicating that pregnancies were typically planned; nonetheless, 44% (n = 24) of participants responded that they did nothing when asked, “Before you became pregnant, did you do anything to improve your health in preparation for pregnancy?” Among those who reported preparations, the most common behavior was taking folic acid or prenatal vitamins (33%, n =18). In an open text field, participants described promoting general wellness (e.g., “decreased stress as much as possible”), healthcare-seeking (e.g., “saw an endocrinologist for my type 2 diabetes to discuss getting pregnant again”), and improving diabetes control (e.g., “get A1C as low as possible”). Among the 9 (16%) participants who described improving diabetes control before pregnancy, this preparation was more common among those with T1DM (n = 6) than T2DM (n = 3, p = 0.01).

Perceived self-efficacy, barriers, and benefits of contraception and preconception care.

Participants generally endorsed high perceived self-efficacy (median score 50 out of 60), low perceived barriers (median score 5 out of 25), and high perceived benefits (median score 16 out of 20) concerning contraception and preconception care (Table 4). None of these scores were normally distributed.

Table 4:

Odds Ratios for Using Procedure/Prescription Contraception Associated with Each Additional ½ Standard Deviation Increase in Perceptions About Contraception Among Women with Pregestational Diabetes in the Postpartum Period

Reproductive Health Attitudes and Behaviors Scale	Median (Interquartile Range)	P valuea	Odds ratios for using procedure/prescription contraceptionb	Adjusted odds ratios for using procedure/prescription contraceptionc
Benefits	16 (14–18)	0.01	1.46 (1.06 – 2.01)	1.52 (1.07 – 2.17)
Barriers	5 (4–8)	0.10	0.83 (0.63 – 1.11)	0.79 (0.57 – 1.09)
Self-Efficacy	50 (44–56)	0.32	1.18 (0.89 – 1.57)	1.13 (0.80 – 1.58)

^aNon-parametric Wilcoxon test used to test for significant differences in the mean scores between women who used procedure/prescription contraception and women who used non-prescription/no contraception.

^bScaled to represent the change in odds ratio for an increase in the score equivalent to the value of one-half standard deviation for each scale from the Reproductive Health Attitudes and Behaviors questionnaire (Charron-Prochownik, Wang, Sereika, Kim, & Janz, 2006).

^cAdjusted for age, race and ethnicity (dichotomized as non-Hispanic White or not), educational attainment (high school or less vs. some college or more), insurance (Private vs not), type of diabetes, whether this was their firstborn child, intendedness of index pregnancy (LMUP score), and whether they had resumed sexual activity since childbirth.

Use of contraceptives use.

Fifty-six percent of the participants (n = 31) used procedure/prescription contraception; hormonal intrauterine devices and pills were the most common methods (Table 2). Only 11 participants (20%) used non-prescription methods. Among participants who used contraceptives, two said they used contraception “sometimes.” Twenty-four percent (n = 13) used no contraception since giving birth. Reasons for non-use included desires to get pregnant, chooses not to use contraceptives, husband or partner does not support contraceptive use, fears potential side effects, and plans to start a contraceptive method.

Table 2.

Contraceptive method used by women with pregestational diabetes between 4 and 8 weeks after birth (N = 55)

Contraceptive Method	n	Percent
Procedure/Prescription
Female Sterilization	6	11
Hormonal Intrauterine Device	10	18
Implant	3	5
Pills	10	18
Injection	2	4
Non-Prescription/No Method
Withdrawal	1	2
Male condom	9	16
Lactational Amenorrhea	1	2
Nothing	13	24

Bivariate findings.

No participant characteristics (Table 1) or perceptions or behaviors (Table 3) had a significant bivariate association with procedure/prescription contraceptive use. There was no difference in contraceptive use based on how soon after childbirth participants took the survey (p = 0.22). Participants using procedure/prescription contraception had significantly higher mean RHAB Benefits scores than those using non-prescription/no contraception (p = 0.01) (Table 4). Mean scores for RHAB Barriers and RHAB Self-Efficacy were not significantly different between participants who did and did not use procedure/prescription contraceptives.

Associations among Perceptions and Contraceptive Use in the Postpartum Period

In the multiple logistic regression models constructed to evaluate associations, we enhanced comparability by reporting the change of log odds of using procedure/prescription contraception for a one-half standard deviation increase in each independent variable because the RHAB scales had different ranges. In the unadjusted logistic model, for every additional half of a standard deviation increase in RHAB Benefit score, the odds of using procedure/prescription contraception increased by 1.46 (95% CI 1.06−2.01), and then by 1.52 (95% CI 1.07−2.17) when we controlled for participant characteristics (Table 4). In the models tested, one half standard deviation increase in RHAB Barriers and RHAB Self-Efficacy were not associated with a significant increase in the odds of using procedure/prescription contraception with or without adjustments.

Discussion

Guided by Pender’s Revised Health Promotion model (Pender, 2011 we provided novel findings about perceptions and behaviors in the postpartum period of the growing population of women with diabetes. More than 75% of our participants used contraception in the 4 to 8 weeks after childbirth, and most used highly effective methods. These findings are similar to the rate of postpartum contraceptive use by women with diabetes in the Maryland Pregnancy Risk Assessment Monitoring System (77%; Perritt et al.,2013) but higher than the rate in California’s Medicaid program (52%; Schwarz et al., 2017)

Participant Characteristics

We identified a few notable characteristics in our participants. The median age was 32 years; therefore, many participants who were considered to be of advanced maternal age (35 and older) during their recent pregnancies or would be in future pregnancies. However, as demographic trends indicate that T2DM is emerging earlier in life, health care providers should anticipate caring for more pregnant women with diabetes in younger cohorts as well (Britton et al., 2018; Geiss et al., 2014).

Use of contraceptives did not vary by race and ethnicity in our participants. These findings diverged from our previous work about non-pregnant young adult women with diabetes, among whom Hispanic women were significantly more likely to use non-prescription methods and significantly less likely to use no method (Britton et al., 2019). Nonetheless, because diabetes disproportionately affects women of color during their reproductive years (Britton et al., 2018), improving maternity care for women with diabetes may be a strategy that can contribute to greater racial and ethnic equity in reproductive health outcomes.

General Perceptions and Behaviors

Our participants resumed coitus as early as 1 week after childbirth, and almost half resumed coitus by 4 to 8 weeks postpartum. In other hospital-based samples, 43% (Sok, Sanders, Saltzman, & Turok, 2016) to 51% ( Rogers, Borders, Leeman, & Albers, 2009) of women resumed sexual activity by 6 weeks after birth. Women who are not yet sexually active may benefit from establishing a satisfactory contraceptive method before resumption of coitus because women who are not breastfeeding may ovulate as early as 4 weeks after childbirth (Jackson & Glasier, 2011).

In previous qualitative research in the United Kingdom women with diabetes expressed the beliefs that do not need contraception (Earle et al., 2017; Murphy et al., 2010) or that contraception is unsafe or ineffective for women with diabetes (Murphy et al., 2010; Shawe et al., 2011). Largely, our participants did not endorse these beliefs, underscoring the importance of conducting research directly with women with diabetes in the United States to understand their concerns and needs.

Pregnancy Planning

In our sample, most participants planned their recent pregnancies, which is similar to a finding of a study by Perritt et al. (2013) in which women with diabetes were no more likely to have mistimed or undesired pregnancies than women without medical conditions. In contrast, Chor, Rankin, Harwood, & Handler (2011) found that women with chronic illnesses may be more likely to have unplanned pregnancies than their healthier peers. The prevalence of unintended pregnancies among with diabetes is unknown but the occurrence is periodically noted in the qualitative research (Collier et al., 2011; Earle et al., 2017; Mersereau et al., 2011; Murphy et al., 2010; Spence Alderice, Harper, McCance & Holmes, 2010). For women who desire more children, the postpartum period may constitute the preconception period for a future pregnancy. Supporting pregnancy planning and health optimization for women with chronic illnesses is important for their reproductive self-determination and well-being.

Despite the fact most planned their pregnancies, few of our participants tried to improve glycemic control before conception. Qualitative researchers found that women with diabetes may not fully understand the effects of diabetes on pregnancy (Chuang et al., 2010; McCorry et al., 2012) Murphy et al., 2010; O’Higgins et al., 2014), that preconception care can reduce risks (McCorry et al., 2012; Murphy et al., 2010; O’Higgins et al, 2014), or that prenatal care cannot reverse the effect of elevated blood glucose on embryogenesis (McCorry et al., 2012).

Among our participants, pregnancy planning (LMUP score) was not associated with the use of contraceptives; this finding differs from the finding in one study that suggested that women with mistimed or unwanted pregnancies may be more likely to adopt highly effective contraception (Guzzo & Hayford, 2017). Our participants endorsed a variety of reasons for not using contraception (fear of side effects, partner does not support contraceptive use, and planning to start later). Nurses and other providers can tailor care to assure that women and their partners receive the relevant counseling needed to support their decision-making related contraception.

Perceived Self-Efficacy, Barriers, and Benefits of Contraception and Preconception Care

The association between effective use of contraceptives and perceived benefits of contraception and preconception care was consistent with Pender’s model (2011). Women may not seek contraception or preconception care if they do not believe those actions will have any value, which is also consistent with Pender’s model (2011) and behaviors described in the qualitative literature (Chuang et al., 2010; Murphy et al., 2010; O’Higgins et al., 2014). The implications for communication are that women with diabetes may be more receptive to positive information about preconception care rather than scare tactics (Collier et al., 2011; McCorry et al., 2012). Providers face the challenge of conveying worrisome or unwelcome information because women may become frustrated with being told not to have children (Spence et al., 2010) or to delay pregnancy until their blood glucose levels are lower (Lavender et al., 2010). Women with diabetes may delay prenatal care when they fear that providers will disapprove of their pregnancies (Collier et al., 2011; Edwards, Speight, Bridgman, & Skinner, 2016; Murphy et al., 2010; Spence et al., 2010). Because our data were collected at a single time point, we do not assert that modifying perceptions about contraceptives would change contraceptive use. Future researchers should explore how promoting the benefits of contraction and preconception care might also increase uptake and how to optimize positive health communication.

Although our findings did not support a relationship between the perception of barriers or self-efficacy and use of contraceptives, these relationships may be worth future study because other studies suggested possible links, and our sample size was small. Higher self-efficacy scores were associated with perceived usefulness of preconception counseling, birth control (Grady & Geller, 2016), and preconception care seeking behaviors (Komiti et al., 2013). The expense of diabetes supplies and medications present barriers to optimizing blood glucose levels in the preconception period (Mersereau et al., 2011); this concern was articulated by one of our participants in an open response. Providers may alienate women by not acknowledging these barriers. Collier et al. (2011) found that participants delayed prenatal care because they anticipated providers would criticize them for poor glycemic control when they could not afford to comply with their treatment plans. Further, barriers to contraception might be particularly challenging for women as they meet the biological, psychological, and social demands of the postpartum period (Verbiest, Tully, & Stuebe, 2017).

Limitations and Recommendations for Research

We sought to enroll 90 women but did not meet our target because women with diabetes in the postpartum period proved to be a hard-to-reach population. Although we engaged with site partners to assess our recruitment plan, the sites served fewer English-speaking women than projected. Recruitment increased when we obtained approval from the Institutional Review Board to contact women by phone or email; this also diminished the bias in our findings against women who did not use procedure/prescription contraception which require interaction with a provider. A larger sample would clarify if non-significant associations were related to sample size or a true lack of association. Nonetheless, our sample size limitations did not undermine our findings of a statistically and clinically significant relationship between perception of benefits and use of contraceptives.

Although we found it challenging to contact potentially eligible participants, our study content and format were largely received positively. Only one woman explicitly declined to participate. While 29 women did not respond to our initial attempts at contact, we do not know if we had correct contact information. Nor do we know why some women who consented decided not to complete the survey. We were concerned about study participation being burdensome in the postpartum period, but an advantage of the online survey was that women could participate when their schedules permitted. We also had no missing data for contraceptive use and resumption of sexual activity, which suggests that these were acceptable topics for the survey.

The RHAB scales and LMUP scale were not previously used with adult women with diabetes during the postpartum period and performed moderately well, although our sample size was small for good estimation of Cronbach’s alpha coefficients (Rouquette & Falissard, 2011). Our values for the Cronbach’s alpha coefficient was lower for RHAB Benefits scale, suggesting lower internal consistency in our sample than in the sample used for validation. In contrast, the values were similar for RHAB Barriers and higher for RHAB Self-Efficacy and LMUP.

To determine if certain perceptions predicted use of contraceptives, we controlled for numerous demographic variables. However, we did not control for every issue which may have affected women’s choices. Also, our sample size was small for additional statistical controls. Notably, it was beyond the scope of this study to assess breastfeeding goals, mental health, or women’s preference for contraceptive features; studying those dimensions in the future may be fruitful. Additionally, while we controlled for type of diabetes and confounding by sociodemographic characteristics, there may be meaningful differences between women with T1DM and T2DM for which larger, separate samples would be needed to address.

These findings should not be generalized to women who were not included in our sample: women who have severe complications, including fetal demise and significant malformations; those who do not speak English; and those who did not obtain perinatal care from an academic medical center. We collected data from cisgender women and our findings may not address the needs of transgender or gender non-confirming individuals with diabetes.

Conclusion

Improving care in the postpartum period is critically important (Lowe, 2019), particularly for women managing chronic illnesses. Our findings highlight potential opportunities to improve maternity care for women with diabetes, which has promise for reducing future risks of adverse outcomes and supporting women to achieve their personal pregnancy goals.

Call outs.

1. After birth, effective contraception may be underutilized by women with pregestational diabetes, a condition that elevates the risk of obstetric complications.

2. Most participants with pregestational diabetes used contraception after birth.

3. Providers should be aware that use of contraceptives in the postpartum period is more likely when women with pregestational diabetes view contraception and preconception care as beneficial.