First, four bilingual translators who were Brazilian-Portuguese speakers and English-proficient independently translated the original version of SMAQ into Brazilian-Portuguese. Two of them did not know about the study theme, and the other two were informed about the objectives and content of the study. The translators were selected by convenience and invited by email. A 30-day deadline was agreed for returning the translations. To synthesize the results, an expert committee formed by three research pharmacists who had good Portuguese knowledge and English proficiency compared the four scale translations (SMAQ versions A1, A2, A3, and A4).

Ambiguities or discrepancies in translated words were resolved by consensus, leading to a single translated version of the instrument (version A0).

In this step, the Brazilian-Portuguese version A0 was back-translated into the original English by three other bilingual translators whose native language is English and who operated in a Brazilian-Portuguese domain. The back translators did not receive information about the study objectives or concepts. Therefore, we hired English-native individuals, two of whom were resident in Brazil, who work in English–Portuguese and Portuguese–English translation to do the back-translations. All three back-translated SMAQ versions (versions RA1, RA2, and RA3) were again compared by an expert committee formed by the previous three researchers and another research pharmacist with a master’s degree in pharmaceutical science and who was bilingual (Brazilian-Portuguese and English fluent). Ambiguities or discrepancies were resolved by consensus, leading to the instrument’s back-translation (version RA0). These back-translation versions were submitted to the instrument’s original version developers to obtain semantic equivalence to the translated versions. The expert committee confirmed idiomatic, experiential, and conceptual equivalence (prefinal versions).

We conducted two cycles of pretesting. In the first, 15 individuals were recruited and answered the prefinal SMAQ version that the expert committee previously defined. In the second cycle, we applied the scale once more with five participants.

The pretest and psychometric property evaluation steps were conducted following the descriptive cross-sectional cohort study model.

This was a multicentric study. Data collection of the pretest phase was conducted in the cardiology clinic of a teaching hospital in Aracaju, Sergipe. For the validation phase, as well as the already mentioned site, the study was simultaneously conducted in two units of family health and two public state institutions that provide pharmaceutical care in Maceió, Alagoas.

The study population included patients with confirmed systemic arterial hypertension aged over 18 years, who were using antihypertensive drugs for at least 30 days. We adopted, as exclusion criteria, the use of antihypertensive drugs to treat other health conditions.

The data were collected by researchers, pharmacy graduate students, and properly trained pharmacists through patient interviews and blood pressure measures. Pretest data were collected between September and November 2018. Data for the validation phase were collected between January and July 2019. The interviews were based on four questionnaires: the first instrument collected sociodemographic data (name, register number, age, sex, and schooling) and history of antihypertensive drug use; the instruments following were the SMAQ pharmacotherapy adherence scale in the final translated versions; a Brazilian-Portuguese version of the Morisky–Green–Levine (MGL) scale to concurrent validation (Morisky et al., 1986; BEN, 2011); a Brazilian version of the Marlowe–Crowne social desirability scale (MC-SDS-BR) (Crowne and Marlowe, 1960; Ribas et al., 2004). Systolic (SBP) and diastolic (DBP) blood pressure values were obtained using calibrated manual sphygmomanometers, taking the average of two measures with at least a 1-min interval between measures, as advocated by the seventh Brazilian Guideline for Arterial Hypertension (Malachias, 2016). In the present study, a complete case (CC) analysis was performed via casewise deletion.

Responses to the SMAQ items were coded analogous to the original version (Knobel et al., 2002). Patients were considered non-adherent if they gave positive responses to any of the qualitative questions (1, 2, 3, and 5), had missed more than two doses over the past week, or had missed over 2 days of total non-medication during the past 3 months.

The SMAQ contains items that elicit information on i) specific medication-taking behaviors (e.g., dose taken and dose frequency) and ii) barriers to adherence (e.g., forgetfulness and side effects) (Nguyen et al., 2014). According to Krousel-Woord et al (2021), these explicit attitudes influence health behavior that is consciously chosen or premeditated, and they are usually assessed through self-report surveys that focus on patients’ conscious attitudes toward disease and its treatment.

In this step, the researcher and two previously trained research pharmacist collaborators applied the prefinal versions of the translated instrument to a 20-individual group from the target population using the methodological adjustment proposed by Ferreira et al. (2014). We thus conducted two cycles of pretesting. In the first, individuals were recruited and answered the prefinal SMAQ version that the expert committee had previously defined. In the second cycle, we applied the scale again.

All individuals who consented to participate in the study had been informed about the objectives and nature of the research and were asked to sign the Consent Informed Test, in accordance with CNS resolution nº 466/2012 (BRASIL, 2012).

The pretest was conducted as a pilot study, according to the item (data collection) and the guide to apply the SMAQ adherence scale. If any individual in this phase had a question about or difficulty understanding the instrument items, the question would be registered in the corresponding scale formulary to be discussed again by the expert committee and possibly reformulated.

To verify reliability by internal consistency, we used Cronbach’s alpha, which indicates if each item of a scale is appropriated for evaluating the nominated concept (Cronbach, 1951). On the other hand, α is a function of the interrelation of items with the number of items, so the number of items is a factor which affects the α coefficient. In simple terms, it is a reliability statistical parameter from 0 to 1 where values above 0.6 may be considered satisfactory and those exceeding 0.8 may indicate high internal consistency (Taherdoost et al., 2014; Ventura-León and Peña-Calero, 2021; Zakariya, 2022). In this study, Cronbach’s alpha values under 0.5 were considered unacceptable. We also evaluated the corrected item–total correlation and Cronbach’s alpha if the item was excluded to verify the contribution of each item to the general reliability coefficient of each scale.

The criterion validation (concurrent validation) of the SMAQ scale was evaluated through the Morisky–Green–Levine (Morisky et al., 1986; BEN, 2011) test association to determine the scales’ correlation coefficient. R-values above 0.5 were considered satisfactory and above 0.7 a strong correlation (Drost, 2011; SCHMIDT and DANTAS, 2011; Souza et al., 2017). The considered significance level was p < 0.05. The instrument has four questions structured as dichotomous “yes/no” answers. The worst adherence is considered to be many “yes” answers; only all negative answers classify the patient as an “adherer.”

In the analysis of per-item correlation, the association between each individual item of the SMAQ scale and the total score of the MGL criterion was evaluated. This involved calculating Pearson’s correlation coefficient for each item of the SMAQ scale in relation to the total MGL score.

Exploratory factor analysis (EFA) was conducted on the SMAQ tool. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were used to determine database eligibility. Visual inspection for an inflection point in scree plots was utilized to identify initial factors. EFA was conducted using principle components analysis with direct Oblimin rotation; factor loadings >0.30 were considered valid (Taherdoost et al., 2014).

In this evaluation, we tested the capability of a measure to make distinctions among groups of individuals that differ according to some known factor. The construct validation by known groups was conducted by associating the evaluated scales with blood pressure control (SBP <140 mmHg and DBP <90 mmHg) using the chi-squared and Student’s t-tests and considering that patients with low adherence scores also present low BP control. The significance level was considered to be p < 0.05.

The sensitivity, specificity, and positive and negative predictive value properties were determined to verify whether the Brazilian-Portuguese version of the SMAQ scale would serve as a screening tool to identify patients with low BP control.

In addition in this step, social desirability was compared among adherent and non-adherent patients using the Marlowe–Crowne Social Desirability Scale (Crowne and Marlowe, 1960; Ribas et al., 2004).

Table 1 compares the translated versions and the synthesis of the SMAQ instrument translations. In the first two versions (A1 and A2), translators who did not know about the study theme translated the scale using less formal English than versions A3 and A4, which were made by translators who received guidance about the study objectives. However, in version A1, it is possible to observe more frequently the use of technical terms, common to the health field, such as “medicamento” (medicine) instead of “remédio” (pills) when the translator was a health professional (nutritionist).

In the translation conciliation process, the expert committee decided to mostly use the term “medicamento” (medicine), judging it to be technically more appropriate in a global context and preserving the original meaning. Nevertheless, they decided to use the term “remédio” (pills) in item 5 to approximate it to patients’ informal language, considering that one of the aims of the study is to make the scale accessible to users of Brazil’s public health system, who usually have low health literacy.

In the next step of the SMAQ instrument adaptation—the back-translation—the RA0 version was translated back into English by another three translators, independent of each other, to verify whether the Brazilian versions had any important mistakes or translation inconsistencies that could make the actual content different from the original. As foreseen in the methods, the back-translators were not previously informed about the study objectives.

After this process, a new version conciliation step was conducted by an expert committee which included a fourth expert who was English-proficient to increase the synthesis reliability.

Table 2 shows the original SMAQ scale followed by the back-translations and the version synthesis. The original scale author, Dr. Hernando Knobel, confirmed the semantic equivalence without needs of adjustment.

The pretest phase was the last step of the semantic equivalence process and was the pilot model of the validation process. The 20 participants attended individually. Their mean age was 57 years (22–79 years, standard deviation = 14.6); 65% were female, half did not complete middle school, 85% affirmed having some comorbidity, and 65% had had an arterial hypertension diagnosis for more than 5 years. As for the number of antihypertensive drugs in use, half of participants affirmed using three or more drugs, which should be an alert for possible resistant hypertension.

Only two participants agreed to self-apply the scales, which may be associated with the fact that 50% of participants had low health literacy, which led the researchers to apply the scales by interviewing participants as follows:

In the first of two cycles of pretesting (with 15 and 5 participants, respectively), approximately 30% of the initial sample had some difficulty understanding the term “desatento” (“careless”) in the SMAQ scale item 2. This issue was brought to the attention of the expert committee, who decided to replace the term with “descuidado” (“inattentive”). In the second cycle, after adjustments, the lack of clarity with the new term still remained. Therefore, the expert committee reverted to the original term (“careless”) and approved the final version for further reliability and validation tests to determine whether additional adjustments were necessary.

The Brazilian-Portuguese final version of the SMAQ scale was applied to 117 outpatients with arterial hypertension diagnosis. The mean age was 56.6 years (standard deviation = 10.7 years); 70.9% of the patients were over 50 years, most were female (72.6%), and the self-declaration of race and/or skin color was mostly “brown” (60.7%). As to schooling levels, 73 patients did not reach high school, and 12% (n = 14) were not literate. The mean number of prescribed antihypertensives per patient was 1.87 (standard deviation = 0.87), and 22 respondents (18.8%) reported using three or more medicines to control their blood pressure. The most reported medicines were losartan (n = 85) and hydrochlorothiazide (n = 49), which 72.6% and 41.9% of patients had used, respectively. The medicines that had these drugs combined were not included in this calculus. Regarding existing comorbidities, 79 patients reported having at least one: diabetes mellitus (n = 41; 35%), dyslipidemia (n = 39; 33.3%), and cardiovascular complications (n = 12; 10.2%).

As to scale adherence classification, 79.5% of patients were considered non-adherent to treatment (n = 86). The mean time spent to complete the instrument was 66.6 s. Table 3 shows the study sample characteristics of SMAQ adherence classification.

The Cronbach’s alpha coefficient of the SMAQ scale was 0.631, which is considered satisfactory.

Table 4 shows the values of corrected item–total correlation and Cronbach’s alpha if the item was excluded from this instrument.

In addition to the adapted instruments, this study also measured the internal consistence of the MGL scale, Brazilian-Portuguese version, used as criteria validity, and we obtained alpha value 0.627.

The Brazilian-Portuguese version of the MGL pharmacotherapy adherence self-report was used as an external criterion to the concurrent validation process as it is a worldwide instrument and has already been adapted to different health conditions.

We obtained a positive Spearman’s correlation coefficient of 0.56 between the total score of the adapted SMAQ version and the external criterion of the total MGL, with a significance level of p < 0.001. This is considered a satisfactory correlation between the test and the external criterion. The results of total scores and score per item are shown in Table 5.