Development and validation of a medication non-adherence scale for Malaysian hypertensive patients: a mixed-methods study

Background Non-adherence to anti-hypertensive medications can lead to hypertension-related complications. One of the most effective preventive measures to mitigate these complications is to understand the underlying determinants of medication non-adherence using various scales. Unfortunately, existing scales for measuring non-adherence to anti-hypertensive medications have certain limitations, such as insufficient consideration of validity, dimensionality, and cultural adaptation. In response, the current study aimed to develop and validate a measure of non-adherence to anti-hypertensive medications—known as the Malaysian Anti-hypertensive Agent Non-Adherence Scale (MAANS)—for use in local hypertensive patients. Methods A two-phase mixed-methods approach was used. Phase 1 involved qualitative interviews with hypertensive patients from two health clinics in Kuala Lumpur, Malaysia. The themes extracted from these interviews were used to generate items for the MAANS. In Phase 2, data from 213 participants were analysed using exploratory factor analysis (EFA) to establish the scale’s factor structure, thereby created the modified version of the MAANS. Confirmatory factor analysis (CFA) was then conducted on a separate dataset of 205 participants to confirm the factor structure, resulted in the final version of the MAANS. The reliability of the final MAANS version was assessed using Cronbach’s alpha coefficient. The MAANS scores were used to predict subscales of the Malay version of the WHO Quality-of-Life (QOL) BREF, demonstrating the scale’s predictive validity. Results Ten qualitative interviews yielded 73 items. The EFA produced a modified MAANS with 21 items grouped into five factors. However, the CFA retained three factors in the final scale: Perceived Non-Susceptibility, Poor Doctor-Patient Relationship, and Unhealthy Lifestyle. The final 14-item, 3-factor MAANS demonstrated moderate reliability (Cronbach’s alpha coefficient = 0.64) and exhibited partial predictive validity, with the Poor Doctor-Patient Relationship and Unhealthy Lifestyle subscales significantly predicting Social QOL and Environmental QOL. Conclusion The MAANS is a reliable, valid, and multidimensional scale specifically developed to evaluate non-adherence to anti-hypertensive medications in local clinical settings with the potential to further the advancement of research and practice in sociomedical and preventive medicine.


Background
Non-adherence to anti-hypertensive medications is an alarming issue worldwide with a global prevalence ranged 27.0-40.0%[1].In Malaysia, the prevalence of nonadherence to anti-hypertensive medications was also high, ranged 24.2-51.3%[2,3].This not only has substantial economic implications to the country [4] but also carries the risk of uncontrolled blood pressure among the patients [5], ultimately leading to hypertension-related complications, such as cardiovascular diseases [6,7].As such, one of the most effective strategies for preventing these complications is to target the root cause, which is the issue of medication non-adherence [8].To achieve a satisfactory level of adherence to medication, it is essential for both healthcare providers and patients to gain a comprehensive understanding of the underlying social determinants for medication non-adherence [9].This understanding is a critical prerequisite for developing effective preventions to address the issue of medication non-adherence.
In the past decades, researchers have developed multiple scales to assess medical non-adherence among hyperten-sive patients [10].Notable examples of these scales include the Brief Medication Questionnaire (BMQ) [11], the Morisky Medication Adherence Scale (MMAS-4) [12], the Morisky Medication Adherence Scale (MMAS-8) [13], the Maastricht Utrecht Adherence in Hypertension (MUAH-25) [14], the Maastricht Utrecht Adherence in Hypertension (MUAH-16) [15], the Treatment Adherence Questionnaire for Patients with Hypertension (TAQPH) [16], the Hill-Bone Compliance Scale (HBSC) [17], and the Self-Efficacy for Appropriate Medication Use Scale (SEAMS) [18].While these adherence scales are widely used in different countries, they possess certain limitations, especially when employed in the specific context of hypertensive patients locally.Firstly, it should be noted that these scales were initially created for diverse populations, and some were not explicitly designed with hypertensive patients in mind [12,13].Consequently, it remains uncertain to what extent these scales accurately capture the adherence behaviours within the local hypertensive community [19].Secondly, existing literature indicates that nonadherence to medication-taking is a multidimensional construct [20].However, some of the currently employed scales, such as the MMAS-4, MMAS-8, and HBSC, are unidimensional in nature.These scales categorise patients solely as adherent or non-adherent, offering limited understanding of the underlying causes of non-adherence.Thirdly, when considering the reliability of these scales, none of the aforementioned scales demonstrate adequate internal consistency, particularly when applied to the local hypertensive community [21].Fourthly, many of the currently used scales lack sufficient reporting of content validity, criterion validity, and/or construct validity during their development process, casting doubt on the validity of these scales.Lastly, the existing hypertension-related scales generally lack the incorporation of theoretical frameworks in their development, hindering researchers and practitioners from ascertaining whether the constructs in these adherence scales comprehensively describe a person's adherence behaviour.
Recognising these gaps in both the literature and clinical practice, the objectives of the present study were threefold.Firstly, the study aimed to investigate the themes of non-adherence to anti-hypertensive medications among hypertensive patients receiving treatment at health clinics in Kuala Lumpur.Secondly, the study sought to develop a scale called the Malaysia Anti-hypertensive Agent Non-Adherence Scale (MAANS) based on the identified themes.Finally, the study aimed to assess the psychometric properties of the MAANS, including its factor structure, reliability, and predictive validity.

Methods
The study was conducted in two distinct phases, referred to as Phase 1 (i.e., a qualitative study to generate items for the MAANS) and Phase 2 (i.e., a quantitative study to validate the MAANS).The study protocol was published elsewhere [22].Ethical approval for the present study was granted by the Medical Research Ethics Committee (MREC), Ministry of Health Malaysia (NMRR- 19-2152-49365).Table 1 shows the scale development process of the MAANS.

Phase 1
In Phase 1 of the study, hypertensive patients who were attending follow-up appointments at two health clinics in Kuala Lumpur, Malaysia between September 2019 and September 2020 were purposively sampled.To be eligible for participation in the study, patients had to meet the following criteria: (a) be 18 years of age or older, (b) have a diagnosis of hypertension (defined as consistently elevated systolic blood pressure (SBP) ² 140 mmHg and/or diastolic blood pressure (DBP) ² 90 mmHg), (c) have been receiving anti-hypertensive medications for a minimum of six months, and (d) be able to read and communicate in either English or Malay language.
However, patients were excluded from the study if they met any of the following criteria: (a) diagnosed with secondary hypertension, (b) currently undergoing active cancer treatment, (c) experiencing severe mental health disorders or cognitive impairment at the time of recruitment, or (d) pregnant.
In-depth interviews were conducted using an interview schedule (in both English and Malay languages) from October 2020 to January 2021, until thematic saturation was achieved.Following each interview, the audio recordings were transcribed verbatim using NVIVO software.Interviews conducted in English language were transcribed directly by the interviewer, while interviews conducted in Ten in-depth interviews were conducted.Thematic analysis was performed.
Five themes were obtained.

2
Item pool was created based on the five themes.Seventy-three items were created and later translated into Malay.Content validity test was performed.
The preliminary version of the 67-item MAANS was obtained.Sixteen think-aloud sessions were conducted.
The pilot version of the 67-item MAANS was obtained.The calibration sample included 213 participants.Exploratory factor analysis was performed.
The modified version of the 5-factor model, 21-item MAANS was obtained.
The validation sample included 205 participants.Confirmatory factor analysis was performed.Reliability test was performed.Predictive validity test was performed.
The final version of the 3-factor model, 14-item MAANS was obtained.
Environmental Health and Preventive Medicine (2023) 28:75 Malay language were translated and transcribed into English language by the interviewer.Thematic analysis was conducted using a deductive coding approach, guided by the WHO Adherence Framework.

Initial item pool
In Phase 2, the themes of non-adherence to anti-hypertensive medications identified during Phase 1 of the study were utilised to create an initial item pool in the English language.Subsequently, a language expert translated the initial item pool into Malay language, as the intention was to construct the MAANS in Malay language for subsequent analyses.

Content validity
The initial item pool underwent a content validity test through an expert review.Three experts evaluated the representativeness and clarity of each item and subscale using several content validity indices, including the itemlevel content validity index for representativeness (I-CVI representativeness), item-level content validity index for clarity (I-CVI clarity), subscale-level content validity index for representativeness (S-CVI representativeness), and subscale-level content validity index for clarity (S-CVI clarity).The content validity indices and qualitative feedback provided by the experts was carefully reviewed and revisions were implemented.This created the preliminary version of the MAANS.

The preliminary version of the MAANS
In the preliminary version of the MAANS, a four-point Likert response scale ranging from 1 (strongly disagree) to 4 (strongly agree) was used for each item.For scoring purposes, higher total scores on the scale indicate a higher degree of non-adherence to anti-hypertensive medications.

The pilot version of the MAANS
Sixteen hypertensive patients were randomly selected to participate in the pilot study.These patients met the same inclusion criteria as mentioned earlier.During the pilot study, they were instructed to "think-aloud" and provide verbal feedback when they encountered any words or statements that were difficult to interpret.They were also asked to revisit any problematic items and offer suggestions for improving them.After the pilot study, the pilot version of the MAANS was finalised based on the feedback and input received from the participating patients.

The modified version of the MAANS
To determine the factor structure of the pilot version of the MAANS, an exploratory factor analysis (EFA) was conducted.A total of 213 hypertensive patients, who adhere to the same inclusion and exclusion criteria mentioned earlier, were randomly sampled from both health clinics.Upon agreement, they were invited to complete the pilot version of the MAANS.
The EFA stage was carried out from March 2021 to April 2021.The sociodemographic characteristics of the participants were collected, along with their scores for each item in the pilot version of the MAANS.The suitability of the data for conducting an EFA was assessed using the Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy and Bartlett's Test of Sphericity.For the KMO index, a value of at least 0.60 was considered appropriate for the EFA [23].Additionally, the Bartlett's Test of Sphericity needed to yield a significant result ( p < 0.05) to confirm the data's suitability for the EFA [23].Once the data's suitability for the EFA was confirmed, the subsequent steps involved factor rotation and extraction.Several methods were utilised to determine the optimal number of factors to retain in the model.Firstly, the Kaiser-Guttman criterion was applied, whereby factors with eigenvalues above 1 were retained [24].Secondly, the degree of over-determination was considered, and factors consisting of at least three items were retained [25][26][27].Thirdly, scree plots were examined, with the component number at the plot's breakpoint indicating the number of factors to be retained [28,29].Fourthly, items with communality values below 0.20 were removed based on communalities.Lastly, items with factor loadings below 0.40 and cross-loadings on two or more factors with loading values above 0.32 were eliminated based on the size of loading [25,30].To ensure the robustness of the factor structure, parallel analysis was conducted using the SPSS syntax by O'Connor [31].The EFA was performed using Statistical Package for the Social Sciences (SPSS) version 26.Following the completion of the EFA, the modified version of the MAANS was obtained.

The final version of the MAANS
To validate the factor structure of the modified version of the MAANS, a confirmatory factor analysis (CFA) was conducted.A sample of 205 hypertensive patients were randomly selected, employing the same inclusion and exclusion criteria as previously described.The CFA stage was conducted between April 2021 and May 2021.Upon agreement, they were invited to complete the modified version of the MAANS.The sociodemographic characteristics of the participants and their scores on all items in the modified version of the MAANS were collected.Additionally, to assess the predictive validity of the scale, an additional section consisting of 26 items from the Malay version of the World Health Organization Quality of Life (QOL) BREF was included in the research questionnaire.In the CFA, several goodness-of-fit indices were employed to evaluate the overall model fit without specific numbering.These indices include the chi-square statistic (» 2 ), relative chi-square (» 2 /df ), root mean square error of approximation (RMSEA), standardised root mean square residual (SRMR), and comparative fit index (CFI) [32].The chi-square statistic, » 2 , is considered a good fit if it is nonsignificant ( p > 0.05).The relative chi-square (» 2 /df ) value should be less than 2 to indicate a good model fit Environmental Health and Preventive Medicine (2023) 28:75 [33,34].The RMSEA value should be 0.06 or lower, while the SRMR value should be less than 0.08 for a good model fit [35].The comparative fit index (CFI) value should be greater than or equal to 0.95 [35], although a value between 0.90 and 0.95 is considered acceptable [36].The CFA was conducted using AMOS 24 software [37].Following the CFA, the final version of the MAANS was obtained.

The reliability and validity of the final version of the MAANS
The internal consistency of the final version of the MAANS was assessed to determine its reliability.Cronbach's alpha coefficient was used for this analysis, where values below 0.50 were considered insufficient, 0.50-0.69were moderate, 0.70-0.79were satisfactory, and ²0.80 were considered good [38].To evaluate the predictive validity of the final version of the MAANS, multiple regression analyses were conducted, regressing the WHO QOL BREF-Malay version on the MAANS final version.The WHO QOL BREF-Malay version comprises four subscales: physical QOL, psychosocial QOL, social QOL, and environmental QOL.The hypothesis was that the final version of the MAANS would be able to predict the QOL in these four subscales.Specifically, it was expected that poorer adherence to anti-hypertensive medication would be associated with QOL scores in the physical, psychosocial, social, and environmental subscales.

Phase 1
Ten in-depth interviews were conducted and revealed that non-adherence to anti-hypertensive medications was associated the patient-related, condition-related, therapyrelated, socioeconomic, and healthcare system-related themes.These five themes corresponded to the five constructs outlined in the WHO Medication Adherence Framework.Details of the qualitative findings were published elsewhere [39].

Initial item pool
In Phase 2, themes from Phase 1 were operationalised into 73 items.These items were originally in the English language and were later translated into Malay language by a language expert.

Content validity
Three experts were given response sheets containing 73 items in Malay language.After conducting the expert review, 7 items were removed due to inadequate item representativeness.Additionally, one double-barrelled item was divided into two separate items.No new items were introduced to the scale.Consequently, a preliminary version of the MAANS consisting of 67 items was obtained.

The pilot version of the MAANS
Following the 16 think-aloud sessions, all participants found the instructions clear to address various aspects of non-adherence to anti-hypertensive medications.However, feedback from 14 participants indicated that the scale was too lengthy and time-consuming to complete in a clinic setting.Notably, three items caused confusion among the patients and were subsequently revised.No items were removed from the scale after the pilot study.As a result, a pilot version of the MAANS consisting of 67 items was obtained.

The modified version of the MAANS
The EFA (calibration sample) included a total of 213 participants.Table 2 displays the sociodemographic characteristics of the participants in the EFA.
The KMO value for the 67 items was 0.70 and the Bartlett's test of sphericity yielded a significant result (» 2 = 4636.83,p < 0.001), indicated the suitability of the data for factor analysis.As the assumption of normality was not violated in the EFA data, a promax (oblique) factor rotation and Maximum Likelihood extraction method were employed.The results of the EFA indicated that a 24-factor solution, as suggested by the Kaiser-Guttman criterion, explained 52.80% of the total variance among all items.This 24-factor structure was also supported by the Catell's scree plot.However, after applying additional criteria such as size of loading, degree of over-determination, and communalities, a total of 46 items were deleted.Among these, 20 items had factor loadings below 0.40, and 26 items loaded on factors consisting of less than three items.No items were removed based on communalities < 0.20 and crossloadings > 0.32.As a result of the item deletions, a 5factor solution with 21 remaining items emerged.Parallel analysis, conducted using the SPSS syntax developed by O'Connor [40], also supported the retention of five factors.Therefore, the 5-factor solution with 21 items was deemed the most parsimonious and conceptually relevant.The factors were named Perceived Non-Susceptibility, Poor Doctor-Patient Relationship, Unhealthy Lifestyle, Perceived Barriers, and Limitations of Healthcare Facilities.This constituted the modified version of the MAANS.

The final version of the MAANS
The CFA (validation sample) included a total of 205 participants.Table 3 displays the sociodemographic characteristics of the participants in the CFA.
In CFA, the hypothesized 5-factor model demonstrated a moderate overall fit.The chi-square statistic, » 2 (179, N = 205) = 285.89,p < 0.001, was found to be significant.The » 2 /df of 1.49, RMSEA of 0.05, and SRMR of 0.07 suggested a good fit of the 5-factor model.However, the CFI value of 0.78 was below the cut-off point, indicating that the model did not have a good fit according to that particular index.A closer inspection of the hypothesized 5factor model (see Fig. 1) revealed that the factor loadings of all items were statistically significant ( ps < 0.05), except for Life1, Hea2, Bar1, and Bar2 ( ps > 0.05).In order to improve the model-fit of the 5-factor model, these four items were removed.Consequently, the Perceived Barriers and Limitations of Healthcare Facilities factors, which consisted of less than three items after the removal, were also eliminated.This decision aligns with the degree of over-determination criterion, which stipulates that each factor must have at least three items.
After conducting the second CFA, the final version of the MAANS was determined to be best characterized as a 3-factor model consisting of Perceived Non-Susceptibility, Poor Doctor-Patient Relationship, and Unhealthy Lifestyle factors.This final version comprised a total of 14 items.

Reliability of the final version of the MAANS
The reliability of the final version of the MAANS was assessed using internal consistency, specifically Cronbach's alpha.The Cronbach's alpha for the final version of the MAANS total scale was found to be 0.64.The Cronbach's alpha estimates for the Perceived Non-Susceptibility, Poor Doctor-Patient Relationship, and Unhealthy Lifestyle subscales were 0.54, 0.64, and 0.41, respectively.Overall, the reliability evidence for the final version of the MAANS can be considered moderate.

Predictive validity of the final version of the MAANS
In order to assess the predictive validity of the final version of the MAANS, multiple regression analysis was conducted to examine the relationship between the three factors (subscales) of the final version of the MAANS and the four subscales of the WHO QOL-BREF Malay version (i.e., Physical QOL, Psychosocial QOL, Social QOL, and Environmental QOL).The results of the multiple regression analyses, depicting the relationship between the WHO QOL-BREF Malay version and the final version of the MAANS, are presented in Table 4.As presented in Table 4, the analyses revealed that none of the subscales of the final version of the MAANS emerged as significant independent predictors of Physical QOL and Psychosocial QOL.However, when Social QOL was regressed on the final version of the MAANS subscales, both Poor Doctor-Patient Relationship (¢ = ¹0.27,p = 0.00) and Unhealthy Lifestyle (¢ = ¹0.22,p = 0.00) emerged as the significant predictors of Social QOL.Perceived Non-Susceptibility did not exhibit statistical significance within this model.Furthermore, when regressing Environmental QOL on the subscales of the final version of the MAANS, Poor Doctor-Patient Relationship (¢ = ¹0.26,p = 0.00) and Unhealthy Lifestyle (¢ = ¹0.22,p = 0.00) emerged as significant predictors of Environmental QOL.Perceived Non-Susceptibility did not demonstrate statistical significance in this model.Therefore, it was observed that the final version of the MAANS demonstrated only partial predictive validity when regressed against the WHO QOL-BREF Malay version.It is noteworthy that the Perceived Non-Susceptibility subscale did not show a significant prediction for any of the subscales in the WHO QOL-BREF Malay version.

Discussion
The present study validated a newly developed scale for assessing non-adherence to anti-hypertensive medications among local hypertensive patients as derived from qualitative themes.The final version of the MAANS comprises 14 items.Although the scale demonstrated moderate reliability and partial predictive validity, it offers a multidimensional, patient-centered approach to measuring medication non-adherence in the context of hypertension.

Factor structure of the MAANS
The MAANS has three factors that capture perceived susceptibility, unhealthy lifestyle, and doctor-patient relationship.Firstly, perceived susceptibility refers to an individual's subjective belief regarding their likelihood of acquiring a disease or facing negative consequences due to specific behaviours [41].Patients with a strong perception of susceptibility demonstrate higher medication adherence [42][43][44].On the other hand, when patients perceive themselves as not susceptible to hypertensive complications, assuming they are unlikely to experience such consequences due to non-adherence, they may consciously or unconsciously neglect taking their anti-hypertensive medications.This lack of perceived susceptibility can diminish their motivation to    [45], influenced by factors like traffic congestion, air pollution, limited parks or walkways, and inadequate sports or leisure facilities [46].Furthermore, sedentary activities such as television viewing, video watching, and excessive cell phone usage [47] contribute to an unhealthy lifestyle.These behaviours significantly impact adherence to anti-hypertensive medications [48][49][50][51], consequently affecting the overall health of individuals with hypertension.Lastly, the doctor-patient relationship refers to a mutual and consensual partnership between a patient and their physician, where the patient seeks the physician's assistance, and the physician willingly accepts them as a patient [52].It is widely recognized that the quality of the doctor-patient relationship has a significant impact on patient satisfaction, medication adherence, and overall health outcomes [53].A local qualitative study highlighted that hypertensive patients in Malaysia tend to discuss medicationrelated concerns with volunteers from non-governmental organizations (NGOs) who offer close support, rather than their doctors or pharmacists [54].Additionally, another study revealed that patients perceived nurses to be friendlier and more attentive compared to other healthcare providers.Consequently, a poor doctor-patient relationship has been associated with lower levels of medication adherence among hypertensive patients, which is a commonly reported finding in previous literature [50,55,56].By comprehending the underlying determinants nonadherence factors, healthcare providers can strategically design interventions that center around sociomedical and preventive medicine, thus enhancing the management of hypertension and preventing hypertension-related complications.For instance, in addressing the factor of perceived susceptibility, healthcare providers can elucidate to patients the significance of adhering to their medication regimen.Meanwhile, tackling the poor doctor-patient relationship factor may entail enhancing communication between doctors and patients, while addressing the unhealthy lifestyle factor could involve educating patients about the importance of adopting healthy habits.

Reliability of the MAANS
The final version of the 14-item MAANS demonstrated moderate reliability, as indicated by Cronbach's alpha statistics.Specifically, two subscales, namely Perceived Non-Susceptibility and Poor Doctor-Patient Relationship, exhibited moderate reliability.However, the reliability of the Unhealthy Lifestyle subscale was deemed insufficient.It is worth noting that reliability values tend to increase as the test length increases.Nevertheless, in the current scale, increasing the number of items could potentially lead to decreased participation and respondent fatigue due to the length of the scale.

Predictive validity of the MAANS
Considering that medication adherence is often evaluated in relation to quality of life (QOL) [57,58], the present study aimed to assess the predictive validity of the MAANS by examining its association with QOL measured using the WHO QOL-BREF Malay version.It is worth noting that both the Poor Doctor-Patient Relationship and Unhealthy Lifestyle subscales emerged as significant independent predictors of Social QOL [59,60].This can be explained owing to the fact that poor doctor-patient relationship can impede patients' capacity to effectively manage negative emotions, thus creating a barrier to establishing social connections with others [61].Meanwhile, an unhealthy lifestyle characterized by an unhealthy diet contributes to weight gain, decreased energy levels, and compromised mental well-being [62,63], can affect both selfconfidence and the motivation to establish new social relationships [64,65].Of note, both the Poor Doctor-Patient Relationship and Unhealthy Lifestyle subscales were also found to predict Environmental QOL, which include physical safety, home environment, and engagement in recreational or leisure activities [66].A poor doctor-patient relationship can hinder thorough assessment by physicians and contribute to patient distrust in diagnosis and treatment, ultimately leading to reduced adherence to medical advice and increased healthcare expenses [67], potentially limiting the patient's involvement in recreational activities.Adopting unhealthy lifestyle such as a sedentary lifestyle, smoking, excessive alcohol consumption, and an unhealthy diet can contribute to adverse health outcomes, disability, and even mortality [68].Consequently, individuals or their family members may face considerable income loss, thereby affecting the quality of their home environment and compromising their physical safety and security.

Strengths and limitations
The present study has several notable strengths.Firstly, the development of the MAANS involved a rigorous validation process incorporating both qualitative and quantitative methods, distinguishing it from existing literature, medication adherence scales, and established conceptual frameworks.This unique approach ensures that the MAANS is specifically designed to capture the pertinent aspects of medication adherence in the local context of hypertension.Secondly, the content validity of the MAANS was established through an expert panel.The experts provided both quantitative and qualitative feedback during the review process.They evaluated the content validity indices at the item and subscale levels, and also had the opportunity to provide qualitative feedback on each item.This thorough expert review enhances the confidence in the content validity of the scale.These strengths contribute to the robustness and practicality of the MAANS as a measurement tool for assessing medication adherence in clinical and research settings.
The present study acknowledges several limitations that should be taken into consideration.Firstly, the sample Environmental Health and Preventive Medicine (2023) 28:75 population included in the EFA and CFA may not fully represent the entire Malaysian population.The study was conducted in two health clinics located in an urban setting, which may limit the generalizability of the findings to other regions or populations with different socioeconomic characteristics.Therefore, caution should be exercised when extrapolating the results to the broader population.Secondly, in contrast to scales that include a neutral response as a midpoint, the deliberate exclusion of a neutral option in the MAANS' four-point Likert-type response format was made to address potential issues associated with neutral responses, which can arise from factors such as ignorance, uncooperativeness, reading difficulty, reluctance to answer, or inapplicability [69,70].It is worth noting that the absence of a neutral option may affect participants' response patterns.

Conclusion
In summary, 14-item, 3-factor MAANS is a reliable, valid, and multidimensional scale for evaluating non-adherence to anti-hypertensive medications in local clinical settings.It captures the underlying factors of non-adherence and has the potential to be a valuable tool for advancing research and practice in sociomedical and preventive medicine.

Fig. 1
Fig. 1 Path diagram of the hypothesised 5-factor model

Fig. 2
Fig. 2 Path diagram of the hypothesised 3-factor model

Table 1
The scale development process of the MAANS

Table 2
Sociodemographic characteristics of participants in the EFA

Table 3
Sociodemographic characteristics of participants in the CFA

Table 4
Multiple regression analysis of the MAANS on the WHO QOL-BREF Malay version