Agreement between self-reported “ideal” and currently used contraceptive methods among women Veterans using the Veterans Affairs healthcare system

Abstract

Background:

Women Veterans who use the Veterans Affairs (VA) healthcare system theoretically have access to the full range of contraceptive methods. This study explores match between currently-used and self-reported “ideal” methods as a potential marker of contraceptive access and preference matching.

Methods:

This mixed methods study uses data from a nationally representative survey of reproductive-aged women Veterans who use VA for primary care, including 979 participants at risk of unintended pregnancy. Women reported all contraceptive methods used in the past month and were asked, “If you could choose any method of contraception or birth control to prevent pregnancy, what would be your ideal choice?” and selected a single “ideal” method. If applicable, participants were additionally asked, “why aren’t you currently using this method of contraception?” We used adjusted logistic regression to identify patient-, provider-, and system-level factors associated with ideal-current method match. We qualitatively analyzed open-ended responses about reasons for ideal method non-use.

Results:

Overall, 58% were currently using their ideal method; match was greatest among women selecting an IUD as ideal (73%). Non-white race/ethnicity (aOR:0.68; 95% CI:0.52–0.89) and mental illness (aOR:0.69; 95% CI:0.52–0.92) were negatively associated with ideal-current match in adjusted analyses; presence of a gynecologist at the primary care site was associated with increased odds of match (aOR:1.35; 95% CI:1.03–1.75). Modifiable barriers to ideal method use were cited by 23% of women, including access issues, cost concerns, and provider-level barriers; 79% of responses included non-modifiable reasons for mismatch including relationship factors and pregnancy plans incongruent with ideal method use, suggesting limitations of our measure based on differential interpretation of the word “ideal.”

Conclusions:

Many women Veterans are not currently using the contraceptive method they consider ideal. Results emphasize the complexity of contraceptive method selection and of measuring contraceptive preference matching.

INTRODUCTION

The efficacy of contraceptive methods in preventing pregnancy is an important consideration for the majority of women (Callegari et al., 2017b; Madden, Secura, Nease, Politi, & Peipert, 2015; Marshall, Guendelman, Mauldon, & Nuru-Jeter, 2016). However, additional factors are highly influential to decisions about contraceptive use and method type, including the perceived importance of avoiding pregnancy (Jones, Tapales, Lindberg, & Frost, 2015); emotional orientations toward pregnancy (Aiken, 2015; Wolgemuth et al., 2018); personal and cultural attitudes toward pregnancy and birth control (Moreau, Trussell, & Bajos, 2013; Woodsong, Shedlin, & Koo, 2004); and preferences for method characteristics such as frequency of use, sexual satisfaction, and side effect profiles (Callegari et al., 2017b; Higgins & Smith, 2016; Jackson, Karasek, Dehlendorf, & Foster, 2016; Madden et al., 2015; Marshall et al., 2016; Newton & Hoggart, 2015). Although measures of contraceptive efficacy, such as the proportion of women using highly effective or prescription methods, remain predominant in family planning research, calls are increasing to create more patient-centered measures of contraceptive access (Callegari, Aiken, Dehlendorf, Cason, & Borrero, 2017a; Dehlendorf, Bellanca, & Policar, 2015). Assessing agreement between preferred methods and current contraceptive use aligns with a human-rights based framework prioritizing individual preferences, and is one potential avenue for measure development (Callegari et al., 2017a; Potter et al., 2019; Ross, 2017).

Emerging data suggest that many women experience discordance between their current contraceptive methods and methods they would prefer to use or feel would be best for them (He, Dalton, Zochowski, & Hall, 2017; Hopkins et al., 2018; Potter et al., 2017). Prior work has highlighted system-level barriers such as method costs and healthcare access inequities as the primary correlates of preference-use discordance (He et al., 2017; Hopkins et al., 2018; Potter et al., 2017). However, provider biases in contraceptive counseling (Dehlendorf, Grumbach, Vittinghoff, Ruskin, & Steinauer, 2011; Gomez & Wapman, 2017), the extent to which preferred method features correspond with available methods (Jackson et al., 2016), and other factors may also impact women’s ability to obtain and use methods that align with their goals and preferences. Improved understanding of reasons for preference-use discrepancies is needed to address potential disparities and better support women in achieving reproductive autonomy.

Women Veterans enrolled in the Veterans Affairs (VA) healthcare system are required to have access to a primary care provider (PCP) proficient in contraceptive management (VHA DIRECTIVE 1330.01(1), 2017) and have access to the full range of FDA-approved contraceptive methods at low or no cost through a centralized pharmacy and device formulary (US Department of Veterans Affairs, 2019). However, the extent to which Veterans’ current contraceptive methods match their preferences for an “ideal” method remains unknown. We therefore sought to examine agreement between self-reported ideal and currently used contraceptive methods in a population of women Veterans with access to an integrated healthcare system. Using mixed methods, we aimed to evaluate agreement between ideal and current contraceptive methods, to identify characteristics associated with ideal-current method match, and to describe reasons for non-use of stated ideal methods.

MATERIALS AND METHODS

Study design and population

Data are from the Examining Contraceptive Use and Unmet Need among women Veterans (ECUUN) study, which was approved by the institutional review boards of [redacted] (Borrero et al., 2017). ECUUN recruited a nationally representative sample of 2302 women Veterans, ages 18–45, who had at least one VA primary care visit within the past 12 months. Participants provided verbal consent and completed a cross-sectional, computer-assisted telephone-based survey administered by trained interviewers between April 2014 and January 2016. The overall response rate was 28%, with a completion rate of 83% among enrolled participants. Participants were similar to non-participants in terms of age, race/ethnicity, marital status, income, geographic region, and presence of medical and mental illness, suggesting that the sample is representative of reproductive-aged women Veterans who use VA for primary care (Borrero et al., 2017).

This analysis includes women identified as at risk for unintended pregnancy (n=992), defined as sexually active with a man within the past month; not currently pregnant, trying to conceive, or up to 6 weeks postpartum; and with no history of infertility or hysterectomy. We additionally excluded 13 women who did not report an ideal method (n=2), reported “other” ideal methods not consistent with contraception method options on the survey (n=7), or had missing data on current methods (n=4), for a study sample of 979 women.

Measures

Current contraceptive use was defined as use of a method within the past month, per standard definitions (“2013–2015 NSFG: Public Use Data Files, Codebooks, and Documentation,” 2017). Participants were asked about use of 17 contraceptive methods, including no method or “other” (non-listed) methods; women could report use of multiple methods. Following assessment of current methods, participants were asked, “If you could choose any method of contraception or birth control to prevent pregnancy, what would be your ideal choice?” Participants were read the same 17 method options, and asked to select a single method. If participants responded prior to hearing the entire list, the interviewer confirmed that the participant considered that method to be ideal among all available methods, and offered to read all options to be sure. Participants who did not report current use of their stated ideal method were additionally asked a single question: “why aren’t you currently using this method of contraception?” Interviewers recorded open-ended responses verbatim, with no maximum length.

We assessed patient-, provider-, and system-level characteristics as potential covariates based on theoretical or empirical associations with contraceptive preferences or use. Patient-level variables included age, race/ethnicity (non-Hispanic white vs. non-white), marital status, education, annual household income, parity, body mass index (BMI), self-reported history of at least one medical condition (hypertension, coronary artery disease, thromboembolic disease, breast cancer, stroke, liver disease, HIV/AIDS, diabetes, migraines, lupus, or seizure disorder) or mental health disorder (depression, anxiety, post-traumatic stress disorder, bipolar disorder or schizophrenia), history of military sexual trauma (MST), whether participants were ever deployed during their military service, and whether participants had additional, non-VA insurance. Provider-level variables were VA PCP gender and whether the participant sees their VA PCP for almost all medical care or for routine gynecologic care such as Pap smears. System-level factors included participant perspectives on the presence of a designated women’s health clinic at their VA primary care site and whether they receive care in that clinic, and availability of an on-site gynecologist if they were to need specialized gynecologic care. Responses of “no” and “don’t know” were combined for system-level variables, as we hypothesized that affirmative knowledge about women’s health-specific services availability is most likely to drive differences in seeking such care, and consequent impact on ideal method match. Census region of the VA primary care site was also assessed. Survey data was used for all variables except census region, which was determined using administrative data.

Data Analysis

Frequencies and percentages were generated to describe overall sample characteristics and ideal method distribution. Ideal method type was described by sample characteristics, and differences in proportions tested using Chi-square or Fisher exact tests if expected counts were less than 5. Our primary outcome was agreement between ideal and currently used methods. Any use of the stated ideal method in the past month was considered a match, regardless of additional methods used. The number and percentage of women with ideal-current method match was calculated for the total sample and by stated ideal method. We used unadjusted logistic regression to test bivariate associations between sample characteristics and match. Adjusted logistic regression was used to identify factors associated with match while adjusting for other pertinent predictors; variables associated with match at the p<0.2 level in bivariate analyses were included in the adjusted model. Stata 14.2 (StataCorp, College Station, TX) was used for all quantitative analyses.

Among the subset of women with ideal-current method mismatch, we analyzed open-ended responses to evaluate and classify reasons for ideal method non-use, identifying potentially modifiable versus non-modifiable reasons from a systems-level perspective. Codes were created as they arose from the data. The first author read all responses, created a codebook of 19 discrete codes with definitions and representative quotations, and coded all responses using the final codebook. The second author independently coded all responses using the final codebook, with the opportunity for additional iterative code creation. Multiple codes could be applied to each response. Cohen’s kappa was calculated using the full dataset as a measure of inter-coder reliability. The average overall kappa was 0.88, with individual codes ranging from 0.74 to 1. Following kappa calculation, the coders discussed discrepancies and coded all responses to consensus; a total of 48 responses (14%) required a consensus discussion regarding the presence or absence of one or more codes. We summarized content associated with each code and grouped codes into modifiable versus non-modifiable reasons for ideal method non-use. Microsoft Excel was used for qualitative data management and coding.

RESULTS

Sample Characteristics

Among 979 women Veterans at risk of unintended pregnancy, 55% were non-Hispanic white and the median age was 34 (range 21–45) (Table 1). The majority (52%) had a bachelor’s degree or higher, were married or cohabitating (63%), were parous (71%), and reported at least one medical (54%) or mental (65%) illness. Over half (53%) reported a history of military sexual trauma.

Table 1.

Sample characteristics of female VA users at risk of unintended pregnancy

Characteristic	n (%)*
Patient-level
Age
20–29	222 (23)
30–34	320 (33)
35–39	251 (26)
40–45	186(19)
Race
Non-Hispanic white	538 (55)
Non-white	441 (45)
Marital status†
Single, never married	142 (15)
Married or cohabitating	621 (63)
Formerly married	215 (22)
Education
Bachelor’s degree or higher	507 (52)
Annual household income†
< $20,000	174 (18)
$20,000-$59,999	506 (52)
>= $60,000	289 (30)
Parous (≥1 live birth)†	698 (71)
Body mass index†
Underweight/normal (<25)	328 (34)
Overweight (25 to <30)	324 (33)
Obese (≥ 30)	324 (33)
Medical illness	530 (54)
Mental illness	640 (65)
History of military sexual trauma	514 (53)
Ever deployed†	537 (55)
Has additional (non-VA) insurance†	520 (53)
Provider-level
VA PCP is female†	760 (79)
Sees VA PCP for almost all care†	772 (80)
Sees VA PCP for gynecologic care†	558 (58)
System-level
Primary care in VA WHC
No WHC at site, or don’t know‡	312 (32)
WHC at site, not seen in WHC	215 (22)
WHC at site and seen in WHC	452 (46)
On-site gynecologist§	601 (61)
Census region
Northeast	82 (8)
Midwest	188(19)
South	503 (51)
West	206 (21)

Abbreviations: VA, Veterans Affairs; PCP, primary care provider; WHC, women’s health clinic.

^*N=979. Percentages may not add to 100% due to rounding.

^†Missing data: marital status (n=1), annual household income (n=10), parity (n=2), BMI (n=3), deployment (n=1), additional insurance (n=1), VA PCP gender (n=12), sees VA PCP for all care (n=10) and gynecologic care (n=17).

^‡n=103 women (11%) responded “don’t know” to a question regarding the presence of a women’s health center at the VA site where they receive primary care.

^§n=163 women (17%) responded “don’t know” to a question regarding the presence of on-site gynecology at the VA site where they receive primary care.

“Ideal” contraceptive methods and match with current methods

Participants reported a range of contraceptive methods that they considered “ideal” to prevent pregnancy (Table 2). IUDs were the most frequently cited ideal method by 215 women (22%), followed by partner vasectomy (19%), birth control pills (15%), and tubal ligation (14%). Differences in ideal method type were observed according to numerous patient-level demographic characteristics, but not by provider- or system-level factors (Appendix).

Table 2.

Ideal methods and percent match with current method(s)

Method	Identified as “ideal” method	Percent currently using their stated ideal method
	n (column %)	n (row %)
Intrauterine device (IUD)	215 (22)	156 (73)
Partner’s vasectomy	189 (19)	69 (37)
Birth control pills	142 (15)	95 (67)
Tubal ligation	133 (14)	94 (71)
Male condoms	83 (8)	45 (54)
Depo-Provera injections	55 (6)	27 (49)
Contraceptive implant	44 (4)	25 (57)
Vaginal ring	43 (4)	30 (70)
Natural family planning	41 (4)	15 (37)
Withdrawal	14 (1)	9 (64)
Patch	8 (1)	2 (25)
No method	7 (1)	2 (29)
Female condom	2 (0)	0 (0)
Spermicides	1 (0)	1 (100)
Sponge/diaphragm/cap	1 (0)	0 (0)
Emergency contraception	1 (0)	0 (0)
Total	979 (100)	570 (58)

Overall, 570 women (58%) reported current use of their stated ideal method. A single participant reported spermicide as both her ideal and current method, resulting in 100% match. Otherwise, match was greatest among women who selected an IUD as their ideal method (73% currently using), followed by tubal ligation (71% currently using).

Factors associated with ideal-current method match

In bivariate analyses, non-white women were less likely to report ideal-current method match compared to non-Hispanic white women (54% match vs. 62%, respectively; p=0.02), as were women with a history of at least one mental health disorder compared to women with no history of mental illness (55% match vs. 64%, respectively; p=0.01) (Table 3). Presence of a gynecologist at the VA primary care site was associated with increased match (61% match vs. 54% with no on-site gynecology, p=0.045).

Table 3.

Associations of sample characteristics with ideal-current method match

	Match of Ideal and Current Method; n= 570 (58.2%)
Characteristic	% Match	Unadjusted OR* (95% CI)	p-value	Adjusted OR† (95% CI)	p-value
Patient-level
Age			0.60		-
20–29	58	Ref.		-
30–34	60	1.09 (0.77, 1.54)		-
35–39	55	0.90 (0.62, 1.29)		-
40–45	61	1.14 (0.76, 1.69)		-
Race			0.02		0.004
Non-Hispanic white	62	Ref.		Ref.
Non-white	54	0.74 (0.57, 0.96)		0.68 (0.52, 0.89)
Marital status			0.66		-
Single, never married	55	Ref.		-
Married or cohabitating	59	1.16 (0.81, 1.68)		-
Formerly married	60	1.21 (0.79, 1.85)		-
Education			0.59		-
Less than college degree	59	Ref.		-
Bachelor’s degree or higher	57	0.93 (0.72, 1.20)		-
Income			0.70		-
< $20,000	58	Ref.		-
$20,000-$59,999	59	1.06 (0.75, 1.51)		-
>= $60,000	56	0.94 (0.64, 1.37)		-
Parity			0.37		-
Nulliparous	61	Ref.		-
Parous (≥1 live birth)	57	0.88 (0.66, 1.17)		-
Body mass index			0.62		-
Underweight/normal (<25)	60	Ref.		-
Overweight (25 to <30)	56	0.86 (0.63, 1.18)		-
Obese (≥ 30)	59	0.97 (0.71, 1.32)		-
Medical illness			0.17		0.45
Yes	56	0.84 (0.65, 1.08)		0.90 (0.69, 1.17)
No	61	Ref.		Ref.
Mental illness			0.01		0.01
Yes	55	0.70 (0.54, 0.92)		0.69 (0.52, 0.92)
No	64	Ref.		Ref.
History of military sexual trauma			0.67		-
Yes	58	0.95 (0.73, 1.22)		-
No	59	Ref.		-
Ever deployed			0.64		-
Yes	59	1.06 (0.82, 1.37)		-
No	57	Ref.		-
Additional (non-VA) insurance			0.48		-
No	57	Ref.		-
Yes	59	1.10 (0.85, 1.41)		-
Provider-level
VA PCP is female			0.96		-
Yes	58	1.01 (0.74, 1.37)		-
No	58	Ref.		-
Sees VA PCP for almost all care			0.28		-
Yes	59	1.19 (0.87, 1.63)		-
No	55	Ref.		-
Sees VA PCP for gynecologic care			0.25
Yes	60	1.17 (0.90, 1.51)		-
No	56	Ref.		-
System-level
Primary care in VA WHC			0.28		-
No WHC or don’t know	57	Ref.		-
Yes WHC, not seen in WHC	55	0.93 (0.65, 1.32)		-
Yes WHC and seen in WHC	61	1.19 (0.88, 1.59)		-
On-site gynecologist			0.045		0.03
Yes	61	1.31 (1.006, 1.69)		1.35 (1.03, 1.75)
No/don’t know	54	Ref.		Ref.
Census region			0.51		-
Northeast	66	Ref.		-
Midwest	57	0.68 (0.40, 1.18)		-
South	57	0.69 (0.43, 1.13)		-
West	59	0.74 (0.43, 1.26)		-

Abbreviations: VA, Veterans Affairs; PCP, primary care provider; WHC, women’s health clinic. Bolded cells indicate results that are statistically significant at the p<0.05 level.

^*Unadjusted logistic regression models with outcome of match vs. no match by sample characteristics. n=979 except for variables with missing data, as noted in Table 2.

^†Adjusted logistic regression model with outcome of match vs. no match; variables associated with match in bivariate analyses at the p<0.2 level were included in the adjusted model. n=979 (no missing data for included variables).

In a model adjusting for race/ethnicity, mental illness, medical illness, and on-site gynecology, ideal-current match remained significantly negatively associated with both non-white race (aOR 0.68; 95% CI: 0.52, 0.89) and mental illness (aOR 0.69; 95% CI: 0.52, 0.92). On-site gynecology remained positively associated with match (aOR 1.35; 95% CI:1.03, 1.75). Medical illness was not associated with match in the adjusted model.

Reasons for ideal-current method mismatch

Among 409 women with ideal-current method mismatch, 340 (83%) provided open-ended reasons for non-use of their stated ideal method. Due to survey skip patterns, 63 women who reported no current contraceptive use and 5 women who reported “no method” as ideal were not asked the open-ended question. One woman was asked about reasons for non-use but did not respond. Qualitative analysis of open-ended responses revealed varied reasons for mismatch, which were classified into modifiable and non-modifiable reasons for ideal method non-use (Table 4). Responses were assigned multiple codes based on the number of reasons cited; 82% (n=278/340) of responses were assigned a single code that sufficiently described the content; 17% (n=57) had two codes and 2% (n=5) had three.

Table 4:

Women Veterans’ reasons for non-use of a stated ideal method

Reasons for non-use of ideal method	Frequency (%)*
Modifiable	78 (23)
Access issues	38 (11)
Cost	19 (6)
Provider barrier	14 (4)
Need (more) information	20 (6)
Non-modifiable	267 (79)
Using another method	95 (28)
Pregnancy plans/goals conflict with	62 (18)
ideal method use
Partner influence	55 (16)
Concern for side effects	23 (7)
Contraindication to ideal method	14 (4)
Ideal method inconvenient	11 (3)
Ideal method not necessary in current relationship context	11 (3)
Lack of permanent sexual partner	7 (2)
Not sexually active (enough)	7 (2)
General fear re: ideal method	5 (2)
Perceived subfertility	1 (0.3)
Non-specific reason	14 (4)
In process of obtaining ideal	11 (3)

^*n=340 women with ideal-current mismatch who provided an open-ended reason for current non-use of their stated ideal contraceptive method (83% of n=409 women with mismatch). Percentages do not add to 100% because codes were not mutually exclusive; 278 responses were assigned a single code (82%), 57 responses were double-coded (17%), and 5 responses had three codes (2%).

Modifiable barriers to ideal method use

Overall, 23% of responses (n=78/340) included a potentially modifiable barrier to current use of their ideal method, including healthcare access issues; concerns about cost; and provider-level barriers. Access barriers were described by 38 women (11%), with 23 mentioning barriers specific to VA. Several cited scheduling difficulties or unavailability of VA providers as reasons for ideal method non-use, explaining, “the VA usually has a month wait to get my birth control pills,” or “it takes a while to get in to a VA OBGYN.” Eleven women claimed their ideal method was not offered by VA, with most describing misinformation about VA benefits or services. Regarding an IUD, one woman explained, “I don’t know if it was available, I heard rules that you had to be married in order to get it from the VA.” A few women described personally experiencing unavailability of a desired method, such as one who explained not using an IUD, “because I went to the VA and they told me that they only give you pills.” Another woman whose ideal method was Depo-Provera explained, “it wasn’t offered when I got on birth control. Or they didn’t have it at the VA, or something like that.”

Cost concerns were cited as a reason for ideal method non-use by 19 women (6%). While cost-related comments were often terse (“It’s expensive;” “It costs too much”), most concerns about cost stemmed from perceived lack of insurance coverage for ideal methods, for example by a woman who said, “my insurance won’t cover it and it’s really expensive” regarding an IUD. While such concerns generally reflected misconceptions about VA’s contraceptive coverage, five women identified lack of coverage for a partner’s vasectomy as a gap in their VA benefits. One explained, “we don’t have the money for me to talk him into doing that, and that’s not covered by the VA, because he’s not the Veteran.”

Provider-level barriers to use of an ideal method were described as a reason for mismatch by 14 women (4%). Several described receiving misinformation from a provider about their ideal method, such as one who explained, “I was told I had to have a baby first, or it’s easier to get implemented after a birth” regarding an IUD. Others perceived provider biases against particular methods, such as a woman whose ideal method was Depo Provera, who said, “I actually asked for it from my doctor, and he gave me a list of side effects. He recommended that I stick with the pill.” Another woman described the impact of conscientious objection from a VA provider, explaining, “The VA doctor said it was against his religion and wouldn’t give [the pill] to me. So they’ve put me on a list for another doctor.” Half of women reporting provider barriers cited tubal ligation as their ideal method (n=7/14). One explained, “I just turned 25 and my doctor didn’t want to do it when I was young.” Others described providers’ refusal to even discuss sterilization as an option, such as one who explained, “Usually my doctor says that it’s too extreme of a choice.”

Finally, 20 women cited needing additional information about their ideal method as a potentially modifiable barrier to use. Most expressed needing to talk with their doctor about prescription or procedural method features or side effects prior to initiation, or needing information about VA’s provision or coverage of the method. However, several women described needing more information about non-prescription methods such as female condoms (“I wasn’t very aware that there were any”) or natural family planning (“I just don’t know enough about it. There’s a lot of learning that goes into it”).

Non-modifiable or personal factors related to ideal method non-use

Seventy-nine percent of responses (n=267/340) included non-modifiable or personal reasons for ideal method non-use, including partner influences; pregnancy plans incongruent with ideal method use; perceived contraindications or side effects; and current use of another method. While some non-modifiable reasons reflect circumstantial or medical barriers to ideal method use, others highlight the complexity of contraceptive method selection and suggest limitations to our question asking about “ideal” methods.

The direct influence of an intimate partner was cited as a reason for ideal method non-use by 55 women (16%), of whom 52 cited partner vasectomy as their ideal method. Women explained non-use of vasectomy in terms of their partners’ decisions, for example, “Because it’s not my choice. It’s my partner’s choice.” Numerous women described a partner’s fears about the procedure, such as, “he’s scared and won’t get it done,” while several depicted partners’ concerns about diminished masculinity with vasectomy, saying, for example, “my husband said he’d feel like less of a man.” For others, absence of a permanent partner precluded use of vasectomy as an ideal method. One woman explained, “I’m not in a serious long-term relationship, so I feel like I don’t have the right to ask that of someone.” Aside from vasectomy, a few women described their partner as the primary reason for non-use of another method, such as one woman who reported, “my husband doesn’t like [condoms].”

Sixty-two women (18%) described incompatibility of their stated ideal method with current or future childbearing goals, suggesting that many may have interpreted the word “ideal” in the abstract, rather than within their current life contexts. Having selected an IUD as ideal, one woman explained, “We’re going to start trying again in April … we’re fine with using condoms until April. After the next baby we’ll pick the IUD again.” Multiple women who selected tubal ligation or vasectomy as their ideal method indicated that permanent methods were not currently appropriate due to future childbearing plans. One explained, “We’re just waiting a few more years just to be sure we don’t want more kids.” Conversely, numerous women cited completion of childbearing as a reason for non-use of reversible ideal methods, such as one who explained relying on her tubal ligation rather than her stated ideal method of Depo-Provera, saying, “I decided I never want to have any more kids, and I made a permanent choice instead of a temporary choice.” Such responses nearly always suggested that the already obtained permanent method was actually superior to the stated ideal method.

Other reasons for ideal-current mismatch that suggest that women interpreted the word “ideal” without relation to their current life contexts include contraindications and side effects, inconvenience of the ideal method, and current use of a different method. Fourteen women (4%) indicated that their stated ideal method was not appropriate due to perceived contraindications. One woman explained she cannot use the pill, “because I smoke, and I’m over 35,” while another described her contraindication to male condoms, stating, “I have an allergy to latex and to spermicide.” Similarly, 23 women (7%) described potential or experienced side effects as reasons for non-use, such as one who said, “I had a lot of bad side effects from hormones before” to explain non-use of the vaginal ring. Others reported that their stated ideal method was actually inconvenient within their life context. Regarding the pill, one woman explained, “my life is too busy, I forget to take them,” while another reported she could not practically use her ideal method of natural family planning, “because it’s not a perfect world, and I don’t have the ability to think about that at the time.” Finally, 95 women (28%) described current use of a different contraceptive method as a reason for ideal method non-use (e.g. “I have my tubes tied,” “I chose the IUD”), with the implication that the current method was working for them at this time, or was in fact superior to the stated ideal method.

DISCUSSION

Despite engagement in an integrated healthcare system offering low or no cost access to the full range of contraceptive methods, only 58% of female VA enrollees at risk of unintended pregnancy reported current use of a stated “ideal” contraceptive method, with non-white women and women with mental illness having reduced odds of ideal-current method match and women with a gynecologist at their VA site having increased odds of match. Potentially modifiable barriers to ideal method use were reported by 23% of women with mismatch; however, 79% of responses included non-modifiable or personal reasons for non-use, suggesting that many women may have interpreted the word “ideal” from perspectives other than their current life contexts.

Our finding of substantial levels of mismatch between currently used and stated ideal methods aligns with emerging work examining contraceptive preference matching in other settings. Among a recent national sample of US women, 36% were not “currently using the type of birth control that [they] would most like to use” (He et al., 2017). Similarly, stated preferences for prescription methods were found to exceed actual use among samples of college and postpartum women in Texas (Hopkins et al., 2018; Potter et al., 2017). These studies highlight method costs and inadequate healthcare access as primary reasons for preference-use discordance (He et al., 2017; Hopkins et al., 2018; Potter et al., 2017). Such system-level barriers were unsurprisingly less prevalent among our sample of women Veterans, who have uniform access to VA health care. Nevertheless, women in our study identified barriers to VA care that may have contributed to ideal method non-use, such as long wait times and provider unavailability. Our finding that an on-site gynecologist was associated with increased ideal-current method match is consistent with known associations between on-site gynecology and access to reproductive health services, particularly provision of contraceptive methods that require a procedure, such as IUDs (Katon et al., 2013; Seelig, Yano, Bean-Mayberry, Lanto, & Washington, 2008). This suggests that proximity to specialized gynecologic care may contribute to overall convenience of contraceptive method acquisition. Misconceptions about contraceptive coverage highlight a need to improve patient awareness of VA contraceptive benefits; however, some Veterans may indeed experience financial barriers to contraceptive use, for instance because a significant subset incur copayments for contraceptive medications (US Department of Veterans Affairs, 2019). Furthermore, as multiple women pointed out, VA does not cover vasectomy for Veterans’ male partners, precluding access to this method for some based on financial means.

Women in our study also reported provider-level barriers to use of ideal methods. Several responses indicated gaps in provider knowledge despite VA’s policy of requiring access to providers proficient in contraceptive management (VHA DIRECTIVE 1330.01(1), 2017). Other provider-level obstacles were similar to those identified in the general population, such as perceived pressure to use specific methods (Amico, Bennett, Karasz, & Gold, 2016; Gomez et al., 2017); discouragement from using a method of choice, particularly in the case of female sterilization (Borrero et al., 2009; Kimport, Dehlendorf, & Borrero, 2017); and provider refusal to provide contraception based on personal beliefs (American College of Obstetricians and Gynecologists, 2007). Our findings highlight the ongoing need across healthcare settings to promote patient-centered, non-biased contraceptive counseling strategies that elicit patient preferences and respect their primacy in method selection (Callegari et al., 2017a; Dehlendorf, Grumbach, Schmittdiel, & Steinauer, 2017).

Our finding of reduced ideal-current match among non-white women and women with mental illness is concerning, and likely reflective of broader healthcare disparities. Differences in contraceptive use by race/ethnicity are well established, with non-white women being less likely to use any contraception and prescription methods in both the general US population (Daniels, Daugherty, Jones, & Mosher, 2015) and among our sample of women Veterans (MacDonald et al., 2017). Reasons for such discrepancies are multifactorial, including healthcare access disparities (Dehlendorf, Rodriguez, Levy, Borrero, & Steinauer, 2010a), disparities in contraceptive knowledge (Rosenfeld et al., 2017), and differential preferences for specific method features (Callegari et al., 2017b; Jackson et al., 2016). Women of minoritized race/ethnicity and other socially marginalized identities are also particularly vulnerable to biased or coercive contraceptive counseling (Dehlendorf et al., 2010b; Gomez et al., 2017; Higgins, Kramer, & Ryder, 2016), which could impede their ability to use prescription methods aligned with their preferences. Prior research found that perceived race-based discrimination while seeking VA health care is associated with reduced odds of using prescription contraceptive methods independent of race/ethnicity, suggesting that negative interactions with providers or the healthcare system may directly impact contraceptive use (MacDonald et al., 2017).

Prior analyses using this dataset revealed no differences in overall contraceptive use or method effectiveness between women with and without mental health disorders, which was interpreted as evidence of equitable service delivery regardless of mental health status (Judge-Golden, Borrero, Zhao, Mor, & Callegari, 2018). Our novel finding of reduced ideal-current method match among women Veterans with mental illness underscores the potential promise of measures based on preference matching to better assess contraceptive access and equity. Further research is needed to better understand reasons for preference-use mismatch in this vulnerable population.

Despite the potential advantages of measuring contraceptive preferences, methodological challenges remain. Preferences for specific method features do not necessarily align with or predict chosen contraceptive methods (Marshall et al., 2016; Walker et al., 2019), and a single method meeting all desired criteria may not exist for all women. Similarly, our qualitative results highlight the complexity of contraceptive decision-making and suggest that many women interpreted the word “ideal” from variable perspectives not necessarily aligned with their current life circumstances. Other studies examining preference-use matching have asked women to identify the method they would “most like to use … regardless of cost of other difficulties” (He et al., 2017), or “if you could use any birth control method you wanted” (He et al., 2017; Hopkins et al., 2018). However, this type of language is also subject to variable interpretation. For instance, in He et al.’s study of U.S. women (2017) “lack of perceived/actual need” for the preferred method was the second most commonly cited reason for preference-use mismatch, after cost and insurance concerns. Other reasons for non-use of the stated preferred method included “pregnancy ambivalence” and “fear of side effects and health concerns” (He et al., 2017). These reasons suggest that identified mismatches are not necessarily suboptimal within respondents’ current life contexts, and are similar those identified in our study.

Strengths of this study include a large sample size, our ability to examine contraceptive preference-use matching in a population with homogeneous access to health care, and our use of mixed methods to explore a complex topic. There are also several limitations. First, small sample sizes for qualitative reasons for mismatch preclude examination of statistical differences of mismatch reasons across patient characteristics. Next, due to survey skip patterns, we are missing qualitative data for women not using contraception (n=63) or who stated that their ideal method was no method (n=5). These women may have had unique reasons for ideal method non-use, which we are unable to explore. Finally, our qualitative results suggest that asking about an “ideal” method may have encouraged women to consider contraceptive preferences outside of their current life contexts. The evident variability in participants’ interpretation of the word “ideal” is a significant limitation of our work, and suggests that this measure may not be a valid marker of access or disparity. While we relied on participant-reported perceptions for provider- and system-level covariates, the relationship between contraceptive preference matching and actual service availability (e.g. PCP contraceptive competencies, care in a women’s health clinic and specific care models) is a different, and important, area of inquiry that should be explored in future work. In addition to refining measurement of preference-use mismatch, future work should examine specific patterns of contraceptive preference-use mismatch to evaluate for potential coercion (e.g. people using IUDs, implants or sterilization who prefer user-controlled methods) in addition to access inequities.

IMPLICATIONS FOR PRACTICE AND/OR POLICY

Healthcare-related barriers to use of preferred contraceptive methods persist even among women Veterans receiving care through an integrated healthcare system. Continued efforts are needed to ensure Veterans are informed of their insurance benefits covering the full range of contraceptive options, and that providers deliver accurate, person-centered and non-biased counseling. However, the factors that drive women’s ability to use a perceived ideal method at a given time are complex, and include personal and contextual elements. This work contributes to a growing evidence base demonstrating that individual preferences are vital to contraceptive selection and use, and that the language used by providers and researchers to ask about reproductive goals and desires impacts interpretation and disclosure of needs and preferences. While our qualitative findings suggest limitations to asking about an “ideal” contraceptive method, they emphasize the ways in which traditional measures of contraceptive efficacy are insufficient as markers of access and reproductive autonomy. Our results underscore the ongoing need for new measures that can more fully assess women’s abilities to select, access, and use contraception in concordance with their goals and preferences.

Author Biographies

Colleen Judge-Golden, PhD is completing her MD/PhD at the University of Pittsburgh. She studies patient-, provider-, and system-level factors impacting contraceptive access and use, particularly among women Veterans.

Tierney Wolgemuth, BS, is a medical student at the University of Pittsburgh School of Medicine. She is currently studying patient- and system-level factors influencing contraceptive use in the female veteran population.

Xinhua Zhao, PhD is a biostatistician at the Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System.

Maria Mor, PhD is an Adjunct Research Professor of Biostatistics, University of Pittsburgh and Director, Pittsburgh Biostatistics and Computing Core, Center for Health Equity Research and Promotion, VA Pittsburgh Healthcare System.

Sonya Borrero, MD, MS is a Professor of Medicine and Clinical and Translational Sciences and Director, Center for Women’s Health Research and Innovation (CWHRI), University of Pittsburgh School of Medicine. She is the principal investigator of the ECUUN study.