This cross-sectional, online, repeated measures study was reported according to the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) recommendations^{( 18 )} and the taxonomy for the following measurement properties: content validity (the degree to which the content of a PROM is an adequate reflection of the construct to be measured), structural validity (the degree to which the scores of a PROM are an adequate reflection of the dimensionality of the construct to be measured), internal consistency (the degree of the interrelatedness among the items), measurement invariance (the degree to which the performance of the items on a translated or culturally adapted PROM are an adequate reflection of the performance of the items of the original version of the PROM), test-retest reliability (the extent to which scores for patients who have not changed are the same for repeated measurement over time), and measurement error (the systematic and random error of a patient’s score that is not attributed to true changes in the construct to be measured).^{( 19 )} Floor and ceiling effects were also assessed.

The study protocol was reviewed and approved by the Research Ethics Committee of Universidade Federal de Santa Maria (CAAE: 56633222.5.0000.5346; #5.489.920). All procedures complied with the Declaration of Helsinki and relevant national regulations. Informed consent was obtained electronically from all participants before their participation in the study. All anonymized data generated and analyzed during this study were stored on a secure institutional server and will be preserved for at least 5 years. Access to the datasets is restricted to the research team; however, they may be made available upon reasonable request in compliance with Brazil’s General Data Protection Law. No personally identifiable information will be shared or published.

The participants were invited to participate in the study through Facebook^®, Instagram^®, WhatsApp^®, and emails from universities via a Google Forms link. No paid advertisements were used in this study. Brazilian university students aged 18-25 years who could speak, read, and write in Brazilian Portuguese were included. We included individuals in this age group because the SRBS was developed for this population.^{( 3 )} Moreover, we included people of all sexes from different geographic regions of Brazil to increase sociocultural variability. Those who were not university students were excluded.

The sample size for structural validity and measurement invariance was calculated according to Kline’s recommendations:^{( 20 )} 5 to 10 people per item of the PROM; however, more than 100 people were sufficient. Therefore, the minimum sample size required for structural validity was 100. The sample size required for test-retest reliability was calculated using the formula for two repeated measures, α=0.05, power=80%, expected intraclass correlation coefficient (ICC)=0.8, lower limit of acceptable ICC=0.6, and expected dropout rate=20%.^{( 21 )} The total expected sample size was 49.

The translation and content validity of the SRBS were assessed in five steps. First, two independent translators translated the original version of the SRBS into Brazilian Portuguese. Both translators were Brazilian and had no prior knowledge of the SRBS. In the second step, the two translated versions of the SRBS were merged into a single version and independently back-translated into English by two other Native American translators. The two back-translators were not familiar with the original instrument. In the third step, discrepancies between the back-translations were resolved by consensus. In the fourth step, the content validity of the SRBS translated into Brazilian Portuguese was assessed by an expert committee composed of seven professionals who were experts in human sexuality and one professional with experience in studying the measurement properties of measurement instruments. Finally, the content validity of the SRBS translated into Brazilian Portuguese was assessed using the target population and was culturally adapted into Brazilian Portuguese.^{( 22 )}

The content validity of the SRBS, translated into Brazilian Portuguese, was assessed by an expert committee and the target population through semi-structured interviews using Google Meet^®. The interviews were conducted by a trained interviewer using an interview guide that included topics related to the SRBS structure, missing concepts, and suggestions for modifying the PROM. The interviews were recorded, transcribed verbatim, and analyzed by two other researchers. An interviewer asked the expert committee and target population about the relevance and comprehensibility of each item, instructions, recall period, response options, and comprehensiveness of the SRBS items. A spreadsheet was used to control response saturation after the interviews were transcribed. The content of the transcribed interviews was analyzed using content analysis with thematic category coding, and the main categories of concepts and words were described.^{( 23 )}

We used 12 to 22 days as the time interval between the test and retest to assess test-retest reliability and measurement error. To ensure that participants had not received any sexual therapy or intervention at the retest, they were asked to answer “yes” or “no” to the question “From the time you first participated in this research to the present, have you received any sexual or sexuality therapy?” Participants who answered “no” to this question on the retest were included in the analyses of test-retest reliability and measurement error. The invitation to participate in the retest was sent by email to participants who provided an email address during the test.

Participants were characterized using a questionnaire consisting of the following items: age, region of residence in Brazil (Southeast, Northeast, South, North, or Midwest), biological sex (male or female), gender identity (cisgender, transgender, or non-binary), sexual orientation (heterosexual, gay/lesbian, bisexual, or other), clinician-diagnosed anxiety and/or depression, and relationship (no sexual partner or with sexual partner). For sexual orientation, we included pansexual, asexual, and demisexual in the “other” category.

The SRBS was developed to assess the frequency of risk of engaging in SRBs in the past month. Items on the SRBS were answered on a scale from zero (never) to four (very often), with higher scores indicating a higher risk of engaging in SRBs. The six items included: “how often you have anal sex without a condom,” “how often you have vaginal sex without a condom,” “oral sex without protection,” “sex while under the influence of alcohol,” “sex while under the influence of drugs or substances,” and “sex without a condom with someone you have just met.” In the development study, the measurement properties of this PROM showed a single factor with sufficient internal consistency (Cronbach’s alpha=0.76) and invariance across sexes.^{( 3 )}

Content validity was assessed by the responses of the expert committee and the target population through content analysis, with selective coding of transcripts and grouping of codes into thematic categories. In the content analysis, we noted the frequency of concepts and words contained in the data. Finally, the categories of concepts and words were described, and participants responded to whether the instrument required revision.^{( 23 )}

Structural validity was assessed using a confirmatory factor analysis (CFA). In the CFA, we used Weighted Least Squares with Mean and Variance adjusted as the estimator, Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Squared Residual (SRMR), Comparative Fit Index (CFI), and Tucker-Lewis Index (TLI). RMSEA and SRMR <0.08 and CFI and TLI >0.90 were considered appropriate models.^{( 24 )} Items with modification indices (MI) >30,000 showed error covariance. Additionally, we used the Bayesian Information Criterion (BIC) and the Sample-size Adjusted Bayesian Information Criterion (SABIC) to compare the one- and two-factor models. The model with the lowest BIC value was considered the most appropriate.^{( 25 )} Internal consistency was assessed using McDonald’s Omega. Values of McDonald’s Omega ≥0.70^{( 26 )} were considered sufficient.

Measurement invariance with Weighted Least Squares with Mean and Variance adjusted as the estimator was assessed using multigroup CFA between females and males according to biological sex. We compared the configural, metric, and scalar levels. For configural invariance, the factorial structure must be similar across groups. In metric invariance, the factor loadings of configural invariance are fixed and indicate the magnitude of the factor loadings across groups. Scalar invariance was evident when the item loadings and intercepts were parallel across groups. Invariance was observed when ΔCFI was ≤0.01 and ΔRMSEA was ≤0.015.^{( 27 )}

We assessed test-retest reliability using the ICC_agreement with a two-way mixed-effects model with interaction for absolute agreement between mean measurements. An ICC_agreement ≥0.7 was considered adequate.^{( 28 )} Measurement error was calculated using the Standard Error of Measurement (SEM^agreement) [SD^difference/√2], the Smallest Detectable Change (SDC^agreement) at the individual level [SEM × 1.96 ×√2],(29) and the Limits of Agreement (LoA) [d-±(1.96 × SD^difference)]. The d- and SD^difference are the mean and standard deviation (SD) of the differences between the test and retest scores, respectively.^{( 29 )}

The distribution of the SRBS scores was assessed based on floor and ceiling effects. Floor or ceiling effects of less than 15% were considered appropriate.^{( 22 )} Missing values (sample characterization variables) were excluded from the analyses. Responses to the SRBS were required, and there were no missing values. All analyses were conducted using the lavaan package in R (version 4.2.1; R Core Team, Vienna, Austria) and RStudio version 2024.04 (Posit Software, PBC, Boston, MA, USA).

The content validity, including face validity, of the SRBS was judged to be adequate by an expert committee, with minor modifications. The expert committee suggested the inclusion of “and/or vaginal sex / e/ou sexo vaginal” and “condom / camisinha” in item 1 (How often have you had vaginal sex without a condom?) to provide broader coverage of the types of sex and to improve comprehension of the term “condom.” Thus, items 1 (How often have you had vaginal sex without a condom?) and 2 (How often have you had anal sex without a condom?) were merged. In addition, the wording of items 2, 3, and 4 was modified to improve comprehension. In the new item 2 [How often have you performed oral sex without protection (condom or dental dam)?], the term “dental dam” was replaced with “condom or other method / preservativo ou outro método” because it is often used instead of “dental dam.” In item 3 [How often have you had sex while under the influence of alcohol (i.e., drunk)?], the term “influence of alcohol (drunk)” was replaced by “influence of alcoholic beverage (drunk) / influência de bebida alcoólica (bêbado).” Moreover, in item 4 (How often have you had sex while under the influence of drugs or substances?), “alter sensations, awareness, and emotional state / alteram as sensações, consciência e estado emocional” was added. After the modifications, a new round was conducted with the expert committee, which judged the modified version of the instrument to be appropriate.

The mean age of participants in the cognitive interview step was 22.4 years (SD=1.7), and the majority were from the southern region of Brazil (n=6; 42.9%), had higher education (n=10; 71.4%), and self-identified as cisgender (n=11; 78.6%). Participants found the SRBS instructions, recall time, items, and response options comprehensive, relevant, and comprehensible. The translation of the SRBS into Brazilian Portuguese were carried out with formal authorization from the original author Emanuele Fino, available in the Table 1S, Supplementary Material.

A total of 362 (57.2%) women and 271 (42.8%) men participated in the study (mean age=22.8, SD=2.0 years). Table 1 presents the participants’ sociodemographic characteristics. In the total sample, most participants lived in Southern Brazil (n=268; 42.3%), were cisgender (n=624; 98.6%), heterosexual (n=314; 49.6%), and had sexual partners (n=355; 56.1%).

Table 1

	Female (n = 362)	Male (n = 271)	Total sample (n = 633)
Age (years), mean±SD	22.1±2	22.3±1.9	22.8±2
Geographic regiona, n (%)
South	166 (45.9)	102 (37.6)	268 (42.3)
Southeast	82 (22.7)	61 (22.5)	143 (22.6)
Midwest	17 (4.7)	15 (5.5)	32 (5.1)
North	21 (5.8)	15 (5.5)	36 (5.7)
Northeast	74 (20.4)	78 (28.8)	152 (24.0)
Gender identity, n (%)
Cisgender	362 (100)	262 (96.7)	624 (98.6)
Transgender	0	02 (0.7)	02 (0.3)
Non-binary	0	07 (2.6)	07 (1.1)
Sexual orientationa, n (%)
Heterosexual	172 (47.5)	142 (52.4)	314 (49.6)
Gay/lesbian	14 (3.9)	54 (19.9)	68 (10.7)
Bisexual	0	53 (19.6)	53 (8.7)
Other	175 (48.3)	21 (7.8)	196 (31)
Relationship, n (%)
No sexual partner	120 (33.1)	158 (58.3)	278 (43.9)
With sexual partner	242 (66.9)	113 (41.7)	355 (56.1)
Anxiety, n (%)	145 (40.1)	71 (26.2)	216 (34.1)
Depression, n (%)	69 (19.1)	38 (14)	107 (16.9)

Missing value.

For both models tested (one- and two-factor), all items had factor loadings (>0.40). Residual correlations were added to the items with the highest MI in the one-factor model: 3 to 4 (MI=102.71), 1 to 2 (MI=95.02), 1 to 3 (MI=34.95), and 1 to 4 (MI=31.63). According to the BIC and SABIC, the one-factor model was recommended. The final one-factor model [χ²(df)=588.84(10)] showed adequate goodness of fit indices for CFI (0.995), TLI (0.96), SRMR (0.01), and RMSEA (0.06, 90% confidence interval (95%CI) [0.00 - 0.14]). McDonald’s Omega was sufficient for the one-factor model (0.723). Tables 2 and 3 present the item factor loadings and goodness-of-fit indices of the SRBS, respectively.

Table 2

Items	One-factor model	Two-factors model
		F1	F2
1. Anal sex and/or vaginal sex without a condom	0.496	0.752
2. Oral sex without protection	0.593	0.956
3. Sex while under the influence of alcohol	0.593		0.820
4. Sex while under the influence of drugs or substances	0.419		0.768
5. Sex without a condom with someone you have just met	0.431		0.534

F1: Factor 1. F2: Factor 2.

Table 4 presents the MGCFA indices for the configural, metric, and scalar invariances for females and males. According to the results, scalar invariance was observed (ΔCFI=0.028). Thus, there may be differences in the factor loadings of the SRBS between females and males.

Table 4

Measurement invariance	CFI	RMSEA (90%CI)	ΔCFI	ΔRMSEA (90%CI)
Females versus males
Configural invariance	0.938	0.118 (0.088 - 0.149)	-	-
Metric invariance	0.923	0.111 (0.085 - 0.138)	0.015	0.007 (-0.003 - 0.011)
Scalar invariance	0.895	0.115 (0.092 - 0.138)	0.028	0.004 (-0.007 - 0.000)

CFI: Comparative Fit Index; ΔCFI: Difference in Comparative Fit Index; CI: Confidence Interval; RMSEA: Root Mean Square Error of Approximation; ΔRMSEA: Difference in Root Mean Square Error of Approximation.

In the current study, the total score for the SRBS was 1.22 (SD=0.82) for the total sample, 1.27 (SD=0.77) for females, and 1.16 (SD=0.88) for males. A floor effect was observed in the total sample, with 17.43% of participants scoring the minimum and 0.20% scoring the maximum.

One hundred and sixty-nine (26.70%) participants responded to the retest. Of these, 143 (84.61%) did not receive any sexual therapy or intervention between the test and retest, and responded to the retest between 12 and 22 days. The test-retest reliability value was considered sufficient (ICC^agreement=0.87, 95%CI [0.767 - 0.924], and the mean difference (d-) between the test and retest was 1.06 points. For the measurement error, the SEM_agreement, SDC_agreement, LoA_inf, and LoA_sup were 1.30 (95%CI [1.09 - 1.52]), 3.65 (95%CI [3.02 - 4.20]), -0.51, and 2.62 points, respectively.