Measuring the Sexual Acceptability of Contraception: Psychometric Examination and Development of a Valid and Reliable Prospective Instrument

Abstract

Background:

People’s sexual experiences have a strong association with contraceptive satisfaction and continuation, but no measures exist to specifically assess contraceptive-related sexual acceptability.

Aim:

This study developed and examined the psychometric properties of reliability, separation, and item fit of a new Contraceptive Sexual Acceptability (CSA) instrument.

Methods:

Enrolled participants initiating a new contraceptive method from the HER Salt Lake longitudinal cohort study contributed baseline survey responses for scale development. The study included the Female Sexual Function Index, the New Sexual Satisfaction Scale, measures of physical and mood-related side effects, and self-reported perceptions of contraception’s sexual impacts. Items from these measures’ served as the basis for analyses. We analyzed responses using descriptive techniques and modeled using exploratory factor (EFA) and bifactor analyses (BFA). The Masters’ Partial Credit Rasch method modeled reliability, separation, and item fit statistics. Here we evaluate (i) the reproducibility of relative measure location on the modeled linear latent variable, (ii) the number of statistically unique performance levels that can be distinguished by the measure, and (iii) the discrepancy between item responses and expectations of the model. Psychometric findings and theoretical models informed item reduction and final scale development.

Outcomes:

We developed a 10-item Contraceptive Sexual Acceptability scale that exceeded the thresholds and sufficiently covered domains for use in contraceptive research and clinical settings.

Results:

Starting with data on 39-items from 4,387 individuals, we identified 10-items that best measured the CSA latent construct. The Rasch model included a total of 5 calibrations. We reduced items based on bifactor analysis and surpassed unidimensionality thresholds (OH = 0.84, ECV = 0.74) set a priori. The final items included questions with scaled responses about pleasure and orgasm (orgasm quality, orgasm frequency, giving partner pleasure), physical (arousal and function) and psychological (emotional connection, surrender) components, general questions of satisfaction and frequency, and a measure of perceived impact of contraception on sexual experiences in the previous 4 weeks.

Clinical Implications:

The 10-item CSA instrument covers physical and psychological aspects of contraceptive sexual acceptability and can be used in clinical settings.

Strengths & Limitations:

The unidimensional CSA instrument offers a brief, yet comprehensive assessment of sexual acceptability. Given the limited diversity of the sample, implementation of this scale in contraceptive research and clinical interactions should be evaluated and validated in more diverse settings.

Conclusion:

Attuning to sexual acceptability could ultimately help contraceptive clients find methods that better meet their needs and preferences.

INTRODUCTION

In healthcare settings, “acceptability” often refers to the extent to which individuals consider the medical or health intervention to be appropriate based on personal anticipated or experienced responses.¹ Contraceptive sexual acceptability refers specifically to, acceptability in a sexual context, or how individuals’ contraceptive method impacts their sexual experiences and to what extent these impacts are either desired or tolerated by the user.² Sexuality is an inherent part of human nature, and contraception is a widely used tool that gives people control over their reproductive lives while engaging in sexual activity. While an abundance of data support the preference-sensitive nature of contraception, sexual aspects of contraception remain secondary to medical models’ emphasis on safety and efficacy of pregnancy prevention.^3,4 Despite acknowledging the fundamental connection between contraception and sexual experiences, researchers and clinicians have relied on measurement tools that may not be complete or appropriate for healthy contraceptive-seeking populations.

The most common tool used to assess sexuality-related outcomes used in the medical literature is the Female Sexual Functioning Index (FSFI).^5,6 Key limitations surface when applying to this to contraceptive care. These include a strong focus on sexual dysfunction and validation limited to perimenopausal, postmenopausal, and diseased populations.^7-9 While the FSFI is useful in exploring sexual functioning across disease status and treatment exposures, these items alone may not be appropriate or complete for assessing contraceptive impact on healthy, young, contraceptive-seeking individuals.¹⁰ The New Sexual Satisfaction Scale responds to some of the limitations and provides an instrument that measures global sexual satisfaction, function, and includes partner as well as psychological domains.¹¹ The NSSS also expands beyond gender binary, using gender neutral terminology and considering sexual activity beyond penetrative sex. The NSSS can also be used regardless of sexual orientation.¹² However, this measure also has limitations in its applications as it includes questions about individuals’ ratings of partners’ sexual perceptions, which while essential to sexual experiences may be outside of the concept of sexual acceptability of a contraceptive method. Additionally, the NSSS 20-item scale may be cumbersome in clinical settings. Finally, neither of these measures explicitly asks about contraception. At present no instruments exist that specifically measure the way in which sexual-related factors feature in peoples’ contraceptive preferences and experiences. Thus, we set out to fill this gap and to develop a new measure to explicitly examine sexual attributes and impacts related to contraceptive sexual acceptability.^13,14

Despite a shift toward patient-centered contraceptive counseling and novel measurement of patient-reported outcomes, there is little guidance on standard measurement or how providers can approach patients’ sexual experiences in relation to contraceptive use. While some contraceptive methods are behavioral or self-managed, many hormonal methods and all long-acting-reversible methods are currently managed in a clinical setting. Contraceptive users feel more satisfied with their counseling and with their method when they feel involved in the decision and feel that their priorities are addressed.^15,16 Quality care requires ongoing assessment to assure that a chosen method continues to meet a person’s needs. Using tools that highlight user’s sexual experience prospectively may help the patient feel that their priorities are being addressed and invite more open conversations around sexual experiences with regards to contraception. Based on available research that describes the conceptual model of sexual acceptability, ^2,17 and highlights the importance of sexual acceptability in relation to contraception, we sought to (i) improve measurement of the impact of contraception on users sexual experiences, including but not limited to functioning and satisfaction^14,18; (ii) examine how these sexual items impact their contraceptive attitudes and behaviors ¹³; and (iii) realize this goal in a brief Contraceptive Sexual Acceptability (CSA) instrument that can be used for any contracepting individual, regardless of age, gender identity, sexual orientation, or sexual activity. The development of the CSA instrument, described herein, builds on existing scales and recent findings on perceived impact of contraception on sexual experiences and individuals’ prioritization of being able to have their desired sex life when starting a new method.^2,14,19

MATERIALS AND METHODS

Data Collection and Sample

The HER Salt Lake Study is a survey-based longitudinal cohort study nested in a quasi-experimental observational study and community intervention (The HER Salt Lake Contraceptive Initiative). A previous manuscript describes the larger initiative in detail, including information about the total population served in the initiative and the subset who enrolled in the study’s survey arm, along with their contraceptive uptake at enrollment.²⁰ Survey-arm recruitment and enrollment occurred from September 2015 to March 2017 with 3 year follow up surveys ending in March 2020. Individuals were eligible for enrollment if they were age 18–45, spoke English or Spanish, were receiving a new contraceptive service at 1 of 4 participating family planning health centers in Salt Lake County, Utah, and did not want to become pregnant for at least 1 year. The study initiated with a 6-month control period where clients navigated standard of care in accessing their desired contraception. The intervention period began in March 2016 when we removed all out-of-pocket cost for all reversible contraception. This analysis includes participants enrolled in both the control and intervention periods. Throughout the 3-years of follow up, clients had the option to discontinue or switch contraceptive methods at any point without cost. Participants received up to $100 in gift cards for completing all 9 surveys over the 36-month study period.

At their enrollment visit, participants (N = 4,425) selected the contraceptive method of their choice and completed the 39 sexual acceptability items. Participants repeated report of these measures at 1, 3, and 6 months via REDCap, an online electronic data capture instrument. The HER Salt Lake study is registered at The University of Utah institutional review board approved the HER Salt Lake Contraceptive Initiative study and trial registration occurred at ClinicalTrials.Gov (NCT02734199).

Measures

The baseline questionnaire collected extensive participant data. Follow-up questionnaires asked fewer questions but repeated all CSA-related measures for longitudinal assessment out to 6 months. We collected socio-demographic characteristics, including age, race and ethnicity, sexual orientation, country of birth, religious identity, highest level of education, federal poverty level (based on 2019 annual income, insurance status number of dependents), number of children under 5 living in the house, and obstetric history. The survey also collected information about relationship characteristics at baseline and each follow-up survey, including whether they engaged in any sexual activity in the previous 4 weeks. We also documented participants’ pregnancy intentions, which clinic they visited, and what method of contraception they selected at enrollment.

Development of the CSA Item Pool

We based our approach to instrument development on the conceptual model of the sexual acceptability of contraception published by Higgins and Smith (2016) and updated with qualitative input by Dalessandro (2021).^2,17 This model posits that the sexual acceptability of contraception depends on multiple factors at various levels, including the individual, interpersonal, and societal spheres.^2,17 The approach centers the interplay of factors contributing to an individual’s sexual experiences, contraceptive behaviors, and patterns. In defining the construct, we focus on individual physiological- and psychological- factors to explore how contraception may impact a user’s sexual experience. The study team systematically reviewed existing tools that measured sexual functioning and sexual satisfaction and identified key strengths and limitations.^6,11,21 In an attempt to explore reproducibility and to support the goal of adapting existing high-quality scales to this healthy contraceptive seeing population, we included 3 instruments and 3 subjective measures in the data collection phase:

1. Female Sexual Functioning Index-6 (FSFI-6).⁵ This scale is adapted from the FSFI-19,⁶ to provide a rapid screen for Female Sexual Dysfunction and asks about personal ratings of level of desire, level of arousal, frequency of lubrication, frequency of orgasm, frequency of pain and gender satisfaction during sexual activity in the preceding 4 weeks. A Receiver Operating Characteristic curve showed that women who scored ≤19 could be classified as having sexual dysfunction, the sensitivity and specificity of the abridged test were 0.93 and 0.94, respectively and maintained reliability, internal consistency and stability.

2. The New Sexual Satisfaction Scale (NSSS-20). ¹¹ The NSSS reports ratings from (0) not at all satisfied to (4) extremely satisfied. The NSSS assesses satisfaction ratings of the following items: “the intensity of my sexual arousal,” “the quality of my orgasms,” “my letting go and surrender to sexual pleasure during sex,” “my focus and concentration during sexual activity,” “the way I sexually react to my partner,” “my body’s sexual functioning,” “my emotional opening up in sex,” “my mood after sexual activity,” “the frequency of my orgasms,” “the pleasure I provide to my partner,” “the balance between what I give and receive in sex,” “my partner’s emotional opening up during sex,” “my partner’s initiation of sexual activity,” “my partner’s ability to orgasm,” “my partner’s surrender to sexual pleasure (letting go),” “the way my partner takes care of my sexual needs,” “my partner’s sexual creativity,” my partner’s sexual availability,” “the variety of my sexual activities,” and “the frequency of my sexual activity.”

3. Physical and mood-related side effects items.²² In qualitative research by this team and others, seemingly non-sexual side effects of hormones (both synthetic and menstrual cycle related), including breast tenderness, emotional affect and bleeding, can have sexual impacts. ¹⁹ Therefore, we used 10 questions drawn from the Menstrual Symptom Questionnaire about frequency of bleeding, cramping, headaches, bloating, breast tenderness, moodiness or irritability, acne, weight gain, weight loss, depressed mood, gastrointestinal distress to assess for baseline and changes and see how they related to CSA measures and changes in contraceptive use and satisfaction.

4. Subjective measures of sexuality and perceived sexual impact of contraceptive method. We also included a question to gauge individuals’ personal concerns about sexual functioning (“Do you have any concern about your sexual functioning?”). We also included a global rating of sex life using an adapted visual analog scale: “On a scale of 1 to 100, how might you rank your sex life right now?” Finally, we asked directly about perceived impact of contraception on participants’ sex lives by asking “In the last 4 weeks, would you say that your birth control or method to avoid pregnancy has (0) made my sex life a lot worse, (1) made my sex life a little worse, (2) had no effect on my sex life, (3) improved my sex life a little, or (4) improved my sex life a lot.”

Iterative Analyses, Conceptual Analysis and Winnowing

Given the absence of any specific measures that capture contraceptive sexual acceptability, we engaged in an early-stage approximation of the CSA as a latent construct using 39- items from extant measures described above, and took an exploratory approach to the data. These descriptive analyses were conducted in STATA 16.

EFA analysis identified multiple domains within the item pool containing the 3 validated measures of the FSFI-6, NSSS-20, and the menstrual/hormonal related side effects. The EFA investigation indicated a 5 factor solution according to visual inspection and the Kaiser criterion which retains factors with eigenvalues greater than 1.²³

CSA items were then calibrated to Masters’ Partial Credit Rasch model as implemented in Winsteps 4.7.0.0.²⁴ The Rasch model has 3 assumptions: unidimensionality (ie, single construct being measured), monotonicity (ie, item scores increase as total score increases), and local independence (ie, item associations are accounted for by the latent construct). Unidimensionality was evaluated through the calibration of data to the unidimensional Rasch model, and subsequently, through principal components analysis (PCA) of item residuals, or item variance not explained by the Rasch dimension (ie, item residuals) following model calibration. PCA of item residuals was evaluated conceptually and through parallel analysis, which compared PCA results of the HER Salt Lake data with data simulated to fit the Rasch model using the SIFILE procedure in Winsteps. Monotonicity was evaluated by increase, or lack thereof, in the average ability estimate of the individuals who respond to each given response option for each item. Local independence was evaluated using Yen’s Q3 statistic, a Pearson correlation coefficient between the residuals of a pair of items after partialling out the measured construct. Although unidimensionality, local independence and monotonicity are assumptions of the Rasch model, the assumptions were tested within the Rasch calibration to allow for the simultaneous consideration of these indicators with other measurement parameters.

Rasch model reliability, separation, and item fit statistics were used to evaluate, respectively: (i) the reproducibility of relative measure location on the modeled linear latent variable, (ii) the number of statistically unique performance levels that can be distinguished by the measure, and (iii) the discrepancy between item responses and expectations of the model. Conventional reliability and separation criteria (ie, cutoffs) of 0.8 and 2, respectively, were used in this study, which allowed for the detection of 3 statistically distinct levels of the latent trait [(4*separation+1)/3]. Mean square fit statistics were chosen over standardized fit statistics because of the size of the HER cohort sample.²⁵ Thresholds (ie, cutoffs) of 0.6 and 1.4 were used as the lower and upper bounds, respectively, of acceptable fit. Calibration of the item pool was conducted iteratively (below, as Calibrations 1–5) and the number of items reduced as model assumptions and parameters were evaluated.

RESULTS

Descriptive Analysis

This analysis focuses on the baseline measures of the survey-enrolled participants and only includes those with complete data on all 39 items (n = 3,626). Table 1 reports counts and percentages of the sociodemographic characteristics of sample. The majority of survey participants were under 25 years of age, married, cohabitating or in a committed relationship, while 20% were actively dating but not in relationship. 43% percent reported high school graduation as their highest level of education to date, about 44% reported incomes below the federal poverty level, and 50% did not have insurance at the time of enrollment. A quarter of the population identified as sexually expansive, while 73% identified as exclusively heterosexual. About two-thirds identified as white non-Hispanic while 21% identified as Hispanic or Latina, and 60% were not religious. These participants were all seeking a new method of contraception at a participating clinic and had indicated no desire for pregnancy for at least 1 year. Of the 4,425 individuals who enrolled in the survey arm of the study, 3,626 (82%) completed all items contributing to the CSA scale. Individuals who indicated they preferred not to answer or skipped any of the items were counted as missing data points and excluded from the reliability testing and the final description of the scale items in the development. Table 2 reports frequency of response category by item, item means and standard deviations, and interitem correlations, the Infit Mean-Square and the Outfit Mean-Square for the CSA item pool at baseline.²⁵

Table 1.

Sample demographics (N = 3,626)

	N	Percent
Age at enrollment (y)
18–19	717	20%
20–24	1,498	41%
25–29	830	23%
30–34	364	10%
35+	217	6%
Missing	0	0%
Relationship Status
Married	423	12%
Not married, but living together or in a committed relationship	1,772	50%
Actively dating, but NOT in a committed relationship	720	20%
Divorced/Separated/Widowed	130	4%
Single, not in a relationship	413	12%
Other	70	2%
I prefer not to answer or missing	98	3%
Highest Education
High school graduation or less	1,557	43%
Some college, vocational training, or more	2,044	56%
Missing	25	1%
Federal Poverty Level* (%)
0–100	1,543	43%
101–199	1,005	28%
200–299	635	18%
≥300	363	10%
Missing	80	2%
Race or Ethnicity
White, Non-Hispanic	2,292	63%
Hispanic or Latina	769	21%
Asian	120	3%
Native Hawaiian or PI	28	1%
American Indian or Alaska Native	57	2%
African American or Black	59	2%
Multiple Race or Self Describe	285	8%
Missing	16	0%
Religious Identity
Not Religious	558	60%
Christian (Protestant, Evangelical, Mainline, etc)	101	11%
Catholic	87	9%
Mormon	105	11%
Jewish	5	1%
Muslim	4	0%
Other	39	4%
Don’t know or prefer not to answer	38	4%
Missing Religious Identity	2,689	74%
Insurance Status
None	1,692	47%
Medicaid or Medicare	127	4%
Private	1,579	44%
Missing	228	5%
Sexual Activity (previous 4 wk)
No	354	10%
Yes	3,225	90%
Missing	37	1%
Sexual identity
Heterosexual	2,618	72%
Mostly Heterosexual	480	13%
Bisexual	421	12%
Mostly or exclusively Gay/Lesbian	16	0%
Other	29	1%
Missing	62	2%
Pregnancy History
None	2,211	61%
1 or more pregnancy	1,368	38%
Missing	47	1%
Number of Live Births
None	2,650	73%
1 of more live birth	913	25%
Missing	63	2%
Pregnancy Intentions (at enrollment)
Never wants to be pregnant	945	26%
In the next year	33	1%
2–5 y	888	25%
5–10 y	1,055	29%
Uncertain	689	19%
Missing	16	0%
Enrollment Location
Clinic A	504	14%
Clinic B	1,167	32%
Clinic C	1,221	34%
Clinic D	734	20%
Missing	0	0%
Enrollment Period
Standard of Care	592	16%
No cost Contraception	3,031	84%
Missing	3	0%
Primary Contraception Selected at Enrollment
Condoms, Diaphragm, Oral EC	53	1%
Cu IUD	446	12%
DMPA	456	13%
Implant	664	18%
LNG IUD	889	25%
Patch, Pills, Rings	1,094	30%
Missing	24	1%

About totals per categories not equaling 3,626 due to missing data from incomplete responses AND percentage totals for characteristics may not total 100% due to rounding.

^*Federal Poverty Level is based on 2016 U.S. Health and Human Services (HHS) Poverty Guidelines.

Table 2.

Baseline distribution of 10-item CSA

Tabulation of CSA scale items	Freq.	Percent	Yen’s Q3	(In, Out) MSQ
1. My contraception method (or lack of a method) has made my sex life…			0.43	(2.09, 2.10)
A lot worse	141	4
A little worse	649	18
No effect	1972	54
A little better	412	11
A lot better	452	12
2. Satisfaction with overall sex life			−0.43	(0.92, 0.98)
Not at all satisfied	150	4
A little satisfied	291	8
Moderately Satisfied	506	14
Very Satisfied	1278	35
Extremely Satisfied	1401	39
3. The frequency of my sexual activity			0.11	(0.99, 0.97)
Not at all satisfied	254	7
A little satisfied	414	11
Moderately Satisfied	857	24
Very Satisfied	1073	30
Extremely Satisfied	1028	28
4. My body’s sexual functioning			−0.12	(0.70, 0.68)
Not at all satisfied	189	5
A little satisfied	326	9
Moderately Satisfied	823	23
Very Satisfied	1311	36
Extremely Satisfied	977	27
5. My emotional opening up in sex			0.04	(0.82, 0.81)
Not at all satisfied	245	7
A little satisfied	418	12
Moderately Satisfied	808	22
Very Satisfied	1157	32
Extremely Satisfied	998	28
6. My level of arousal or how turned on I get			−0.47	(1.20, 1.32)
Not at all satisfied	102	3
A little satisfied	239	7
Moderately Satisfied	915	25
Very Satisfied	1093	30
Extremely Satisfied	1277	35
7. My ability to let go and surrender to sexual pleasure			0.22	(0.70, 0.69)
Not at all satisfied	272	8
A little satisfied	454	13
Moderately Satisfied	954	26
Very Satisfied	1123	31
Extremely Satisfied	823	23
8. The quality of my orgasms			0.30	(0.82, 0.78)
Not at all satisfied	387	11
A little satisfied	369	10
Moderately Satisfied	828	23
Very Satisfied	1192	33
Extremely Satisfied	850	23
9. The frequency of my orgasms			0.56	(0.85, 0.83)
Not at all satisfied	522	14
A little satisfied	523	14
Moderately Satisfied	843	23
Very Satisfied	942	26
Extremely Satisfied	796	22
10. The pleasure I provide to my partner			−0.65	(1.02, 0.92)
Not at all satisfied	97	3
A little satisfied	152	4
Moderately Satisfied	600	17
Very Satisfied	1327	37
Extremely Satisfied	1450	40
Total	3,626	100

The Yen’s Q3 Statistic is the interitem correlation after taking into account the latent trait. The IN MSQ is the infit mean-square, and OUT MSQ is the outfit mean-square standardized demonstrating fit diagnosis (Rasch, 1980).

Exploratory Factor Analysis

The factor solutions included items that measured the (i) psychological aspects of sexual acceptability primarily drawn from the NSSS, (ii) those that measured physiological aspects of the sexual experiences from the FSFI, (iii) items explicitly measuring orgasm from both FSFI and NSSS, (iv) partner related items, and (v) items specific to menstrual and hormonal related experiences or side effects. In examination of the total item function (TIF) curves, we were able to determine that the existing factors did a good job of measuring problems associated with contraceptive sexual acceptability but lacked items that measured the positive relationship between contraception and sexual experiences. The iterative exploration and addition of the subjective measures was driven by the need to may improve measurement of average to positive aspects of contraceptive sexual acceptability. Thus, for the Rasch Model calibrations we started with the 39 items and had a goal of creating a unidimensional measure that hung together while also being informed the multiple components of CSA.

Rasch Model Calibrations

Rasch model calibrations are shown in the flow cart in Figure 1. Calibration 1 included 39 items and 4,387 persons. In this iteration, separation and reliability were 2.73 and 0.88, respectively. As shown in Figure 1, the eigenvalue (ie, items worth of variance) of the first contrast in the HER Salt Lake collaborative (real data) was 9.70. This greatly exceeded the eigenvalue of the first contrast of 1.22 in the simulated data and indicated considerable multidimensionality (ie, more than 1 construct being measured), and as such, monotonicity, local independence and item fit were not evaluated. To try and remedy the multidimensionality (ie, measure just a single construct), 7 NSSS items that focused exclusively on partner’s experiences and satisfaction were excluded from subsequent analyses, while we retained interpersonal items and items that focused of the user’s individual experiences.

Figure 1. Rasch Model Calibrations of the HER Salt Lake contraceptive sexual acceptability items to report measure reliability and separation and amount of variance explained.

Note: The Eigenvalue is a measure of how much of the variance of the observed variables a factor explains. Any factor with an eigenvalue ≥1 explains more variance than a single observed variable.²⁸

Calibration 2 included the remaining 32 items. In this iteration, separation and reliability declined to 2.31 and 0.84, respectively. As shown in Figure 1, the eigenvalue of the first contrast in the real data was 8.95, again demonstrating considerable multidimensionality, and as such, monotonicity, local independence and item fit were not evaluated. Inspection of the pattern of loadings on the first contrast showed the 10 Menstrual/Hormonal Side Effects Questions and dichotomous question about concern with sexual functions with positive loadings and the other 22 items with negative loadings. These side-effects questions were removed and the remaining items recalibrated.

Calibration 3 included the remaining 22 items. In this iteration, model reliability and separation both improved to 3.03 and 0.90, respectively. Likewise, the eigenvalue of the first contrast improved precipitously, as it dropped from 8.95 to 2.67 (Figure 1). Whereas the eigenvalue of 2.67 in the real data exceeded the first contrast eigenvalue of 1.17 in the simulated data, the departure from unidimensionality was not egregious, and as such, item fit was evaluated. Appendix A shows that 6 items exceed fit statistic thresholds.

In the subsequent calibrations (4 and 5), we more closely approximated the eigenvalue in the simulated data, and thus, more closely approximated unidimensionality. As shown in Figure 1, the calibrations with fewer items (ie, 15- and 10-item) reduced reliability (0.88 and 0.81, respectively) and separation (2.67 and 2.09, respectively) estimates, though not egregiously. The number of misfitting items in calibrations 4 and 5 were 4 and 2, respectively (Appendix A). Moreover, in winnowing items to try and achieve unidimensionality, the item assessing the perceived impact of contraception on sex life was dropped. However, given its importance in cognitive interviewing, potential for increasing the measurement of average and more positive experiences, and the association with use patterns as described in previously published manuscripts this component was retained in the final calibration was conducted.

The final calibration included 10 items. In this iteration, model reliability and separation were 2.17 and 0.82, respectively. The eigenvalue of the first contrast was 1.98, exceeding the 1.19 in simulated data, indicating some degree of multidimensionality. As shown in Table 2, 9 of the 10 items demonstrated acceptable fit. All items’ scores increased monotonically with increases in the latent trait estimate for those choosing that response option. Yen’s Q3 statistic showed all item residual correlations were ∣0.30∣ except that between the quality and the frequency of orgasms, which was 0.50. The internal consistency of the 10-item CSA measure was (ie, KR-20; alpha equivalent) was 0.95, and ceiling (1.7%) and floor (0.5%) effects were minimal. Appendix B includes the Item Characteristic Curves to help visualize how each of the items discriminate. The final scale with recommended wording can be found in Figure 2. The unidimensional model simplifies use and scoring in a clinic environment (vs a multidimensional model). Clinicians can simply sum scores and use those as an ordinal measure (or convert those scores to an interval-level measure using the attached [Raw-to-Rasch] table included in the Appendix C.) Table 2 and Table 3 describe the distribution and the summary statistics of this sample in the final scale.

Figure 2. 10-item contraceptive acceptability scale.

Recommended Prompt: The following questions are about your sexual experiences over the past 4 weeks; these may include a variety of activities and is not limited to just intercourse.

*If all questions are complete, use sum score.

**If missing use average sum score/ number of completed questions.

Clinical Recommendation: We recommend adding an open-ended discussion prompt– Tell me a bit more about how your contraceptive method has impacted your sex life.

Table 3.

Summary of CSA baseline items and scale sum scores if using measures as continuous items

	N	mean	sd	min	max
CSA scale average	3,626	2.62	0.83	0	4
CSA scale sum	3,626	26.19	8.31	0	40
1.Impact of contraception on sex life	3,626	2.11	0.97	0	4
2.Satisfaction with overall sex life	3,626	2.96	1.1	0	4
3.The frequency of my sexual activity	3,626	2.61	1.21	0	4
4.My body’s sexual functioning	3,626	2.71	1.11	0	4
5.My emotional opening up in sex	3,626	2.62	1.19	0	4
6.My level of arousal or how turned on I get	3,626	2.88	1.05	0	4
7.My ability to let go and surrender to sexual pleasures	3,626	2.49	1.19	0	4
8.The quality of my orgasms	3,626	2.48	1.25	0	4
9.The frequency of my orgasms	3,626	2.27	1.34	0	4
10.The pleasure I provide to my partner	3,626	3.07	0.98	0	4

Summary statistics include sample size (N), mean, standard deviation (sd) minimum (min) and maximum scores (max).

Bolded values are the composite score sum and average. Numbered items, not bolded are the individual items summary statistics.

DISCUSSION

The development of the novel CSA instrument is intended to identify key elements of contraceptive sexual acceptability construct and measure changes in sexual experiences that users attribute to a new contraceptive method. The scale retained 2 items from the FSFI, 6 items from the NSSS, and 1 subjective measure. Each of the items that the scale addresses are critical to contraceptive sexual acceptability. The tool opens with an explicit question about the users perceived impact of their contraceptive method (or lack of a method) on their sex life and then attends to specific domains that make up the physical, psychological, and interpersonal components of contraceptive sexual acceptability. This scale provides a straightforward measure that asks the user to rate their recent satisfaction with physical domains including sexual functioning, arousal, and orgasm quality and frequency, psychological domain of ability to surrender and emotional openness, and interpersonal domains. In the winnowing of this scale, statistically informed decisions were made to remove items related to menstrual cycle related symptoms. These areas should still be addressed in clinical settings as impact of contraception on bleeding and pain associated with menstruation impact broader quality of life including sexual interests and availability. Additionally, while acknowledging the importance of partner experiences in the tolerance and acceptability of contraception, this brief measure excluded specific questions about partner perceptions that were originally in the NSSS in order to focus on personal satisfaction with each element of their sexual experience. We did retain the 1 item on personal satisfaction with the ability to give one’s partner pleasure, which came up repeatedly in both quantitative and qualitative work in this area.^18,26

We believe that this tool offers improvements to the existing measures for assessing the sexual acceptability of contraception. First, this tool was designed to accommodate people across the reproductive life course and of all genders and sexual orientations. The tool was also adapted to be used with individuals who are engaging in any sexual activity and not just including penetrative sex. Individuals who have only engaged in self-sex or masturbation can answer all questions with the exception of the last question which asks about the pleasure provided to their partner. As noted, this item was retained as partner pleasing emerged as a strong contributor to positive sexual experiences in relation to contraceptive use but can be left out when it is not applicable. In these cases, the mean score for the 9 relevant items, which also demonstrated unidimensional, can be used to examine changes in CSA and stimulate discussion in clinical settings.

Strengths of this study include the reliance on an established framework and the robust sample of contraceptive seeking individuals that participated in the HER Salt Lake Contraceptive Initiative. The unidimensional scale demonstrates high reliability and separation within the items. The final scale draws from both the NSSS and the FSFI-6 to measure contraceptive sexual acceptability changes overtime, as all of these items all work together toward explaining the latent concept of contraceptive sexual acceptability. Some limitations with the study population exist, including a sole geography location and a predominately White and English-speaking participant group. While there was close to 20% Latinx identified participants, a smaller proportion elected to take the surveys in Spanish, indicating a need to validate the scale in Spanish language at a later date. Additionally, when thinking about generalization and representation of the sample to the broader US population, the proportion of Latinx participants approximates the U.S. demographics, while Black and non-English speaking individuals are underrepresented. Additional validation of this scale in populations outside of the development cohort, especially among populations underrepresented in the HER Salt Lake cohort, will strengthen the validity and reliability of the scale across diverse experiences and settings. On the other hand, this study captured a sample who were socioeconomically and sexually diverse and large group of largely young and healthy contraceptive seekers, a population very consistent with contraceptive-seeking people across the country.

The use of the Rasch analysis is also a notable strength of the study. The Rasch analysis was helpful in the development of this 10-item measure as it prescribes a unidimensional model to which the data are fit rather than fitting the model to the data. The fit was imperfect, as reflected in the item-level misfit, local dependence, and dimensionality observed through parallel analysis of simulated data. That said, these violations of model assumptions and item misfit were relatively minor. Given the advantages that the Rasch model and other IRT models hold over classical test theory approaches (eg, item-level statistics, interval-level measurement), ²⁷ the measure developed in this project could demonstrate better psychometric performance than its “parent” measures, and even if only equivalent, the resulting measure is more efficient and represents a jumping off point for improvement. Future efforts could increase the precision of measurement by refining the construct definition, by designing items de novo to operationalize the construct definition, or through more thorough examination of item-level statistics.

The development of this scale and the psychometric examination of this limited-item measurement tool is an essential first step in demonstrating the utility and reliability of these items. Future research that expands on the validity of this scale in different populations and in different circumstances is essential, including populations with underlying medical conditions or people that have experienced sexual violence and trauma. That being said, the empirical evidence provided here supports the validity and reliability of the CSA instrument in this particular setting. Examining the scale at different timepoints in population subgroups will also increase our understanding of the utility of this scale. While some additional testing is needed, it is our hope that the brevity of the novel CSA instrument will have broad applicability. The instrument can be applied in contraceptive development or contraceptive clinical trials to incorporate user experience and provide additional information to researchers and developers. The instrument can also be administered in clinical settings to support person-centered conversations around contraception and sexual experiences. Incorporation of the CSA scale into broader patient reported outcomes and acceptability measures in clinical trials and clinical practice has the potential to improve user acceptability of contraceptive methods under development and more importantly improve individuals’ contraceptive experiences with contraceptive tools that better align with individuals’ positive sexual experiences.

APPENDIX A

Item theta (MZR; ie, latent trait) estimates and mean square infit (INFT; ie, inlier sensitive) and outfit (OUFT; ie, outlier sensitive) fit statistics for Calibrations 3, 4 and 5

Name	Calibration 3			Calibration 4			Calibration 5
	MZR	INFT	OTFT	MZR	INFT	OUFT	MZR	INFT	OUFT
On a scale of 1–100, how might you rank your sex life right now?	−0.51	1.20	1.36	−0.65	1.23	1.34
In the last 4 wk, would you say that your birth control or method to avoid pregnancy has:	0.32	2.05	2.07				0.80	1.60	2.19
Over the past 4 wk, how would you rate your level (degree) of sexual desire or interest?	−0.16	1.38	1.42	−0.28	1.38	1.41	0.11	1.21	1.36
Over the past 4 wk, how would you rate your level of sexual arousal (“turn on”) during sexual activity or intercourse?	−0.48	1.02	1.07	−0.63	1.01	1.04
Over the past 4 wk, how often did you become lubricated (“wet”) during sexual activity or intercourse?	−0.82	1.24	1.74	−0.98	1.22	1.63	−1.14	1.65	1.61
Over the past 4 wk, when you had sexual stimulation or intercourse, how often did you reach orgasm (climax)?	0.21	1.37	1.59	0.11	1.29	1.45
Over the past 4 wk, how satisfied have you been with your overall sexual life?	−0.43	0.89	0.95	−0.56	0.89	0.94
Over the past 4 wk, how often did you experience discomfort or pain during vaginal penetration?	−0.88	1.99	3.59
The intensity of my sexual arousal	−0.14	0.75	0.72	−0.24	0.74	0.70
The quality of my orgasms	0.22	0.81	0.77	0.13	0.78	0.72	0.34	1.05	1.00
My letting go and surrender to sexual pleasure during sex	0.15	0.67	0.66	0.06	0.72	0.70	0.33	0.70	0.68
My focus and concentration during sexual activity	0.16	0.69	0.69
The way I sexually react to my partner	−0.29	0.61	0.57				−0.18	0.57	0.53
My body's sexual functioning	−0.15	0.66	0.63	−0.26	0.72	0.68	−0.01	0.60	0.57
My emotional opening up in sex	−0.01	0.78	0.77	−0.11	0.88	0.86	0.14	0.82	0.79
My mood after sexual activity	−0.37	0.74	0.69
The frequency of my orgasms	0.45	0.86	0.82	0.37	0.84	0.79
The pleasure I provide to my partner	−0.63	0.92	0.83	−0.77	1.03	0.93	−0.60	0.90	0.80
The balance between what I give and receive in sex	−0.2	0.80	0.76
The variety of my sexual activities	−0.19	0.87	0.82
The frequency of my sexual activity	0.05	0.94	0.94	−0.05	1.03	1.03	0.21	1.09	1.05
Do you have any concern about your sexual functioning?	3.72	1.50	7.22	3.84	1.72	9.90

APPENDIX B

Item-characteristic curves for 10-items included in CSA Scale.

APPENDIX C

Raw-to-Rasch table option to convert CSA sum scores to interval level measures

Score	Measure	S.E.
0	−4.95E	1.82
1	−3.75	1.00
2	−3.06	0.71
3	−2.66	0.58
4	−2.36	0.51
5	−2.13	0.46
6	−1.93	0.43
7	−1.75	0.41
8	−1.59	0.39
9	−1.45	0.38
10	−1.31	0.36
11	−1.18	0.36
12	−1.06	0.35
13	−0.94	0.34
14	−0.82	0.34
15	−0.71	0.34
16	−0.59	0.34
17	−0.48	0.34
18	−0.36	0.34
19	−0.25	0.34
20	−0.13	0.34
21	−0.01	0.35
22	0.11	0.35
23	0.23	0.36
24	0.36	0.36
25	0.50	0.37
26	0.64	0.38
27	0.78	0.39
28	0.94	0.40
29	1.10	0.41
30	1.27	0.42
31	1.45	0.43
32	1.65	0.45
33	1.86	0.47
34	2.08	0.49
35	2.33	0.51
36	2.61	0.55
37	2.95	0.61
38	3.39	0.72
39	4.10	1.00
40	5.30E	1.82