Role of medication beliefs on medication adherence in hypertensive Middle 1 Eastern refugees and migrants in Australia.

28 Background : Adherence to medication is essential in some patients for achieving treatment 29 control in hypertension. Medication beliefs is one of the personal modifiable factors that has 30 been recognised to influence medication adherence in different populations. However, there 31 is no published research to confirm the relationship between medication beliefs and 32 medication adherence in Middle Eastern refugees and migrants in Australia. These two 33 different groups may develop different beliefs about their medications that lead to different 34 medication taking behaviours. Understanding the possible differences in beliefs may have a 35 significant impact on enhancing medication adherence in these groups. 36 37 Design : 320 Middle Eastern refugees and migrants with hypertension were approached via 38 various social groups in Australia and asked to complete Arabic versions of the Beliefs about 39 Medicine Questionnaire (BMQ) and the Medication Adherence Questionnaire. BMQ scores 40 ( necessity and concerns scales) were classified as "accepting", "indifferent", "ambivalent" or 41 "skeptical". Multiple mediation modelling was applied to examine the role of necessity and 42 concerns scales as mediators between migration status and medication adherence. 43 Results : There were significant associations between medication adherence and medication 45 beliefs scores ( necessity and concerns scales) ( p =0.0001). Necessity, and concern were 46 mediators in the relationship between migration status and medication adherence. Significant 47 differences were found between refugees and migrants for medication adherence and 48 medication beliefs. Refugees were likely to have less necessity , and more concern beliefs 49 than migrants. They were also less likely to adhere to medications. Almost 30% of refugees 50 could be classified as skeptical and 40% as ambivalent. In contrast, 50% of migrants had 51 accepting beliefs, and around 35% held ambivalent beliefs. Refugees and migrants with “accepting” beliefs reported the highest adherence to medication and those holding “skeptical” 53 beliefs reported the lowest adherence. 55 Conclusion : Medication beliefs are potentially modifiable and are reasonable targets for 56 clinical interventions designed to improve medication adherence. Understanding these beliefs 57 and the likely differences between refugees and migrants is crucial to provide specific and 58 targeted advice to each group independently in order to improve medication adherence and 59 overall health.


65
Essential hypertension is an epidemic affecting approximately a quarter of all adults worldwide 66 (Chen, Tsai, & Chou, 2011) with high mortality and morbidity when not treated or adequately 67 controlled, especially in vulnerable populations (Pesantes et al., 2015). In 2015, 5.8% of the 68 total burden of disease in Australia was due to high blood pressure (AIHW, 2019). The 69 effective control of hypertension requires patients to adopt and maintain a healthy lifestyle 70 and take medication regularly (Shahin, Kennedy, Cockshaw, & Stupans, 2020). However, 71 medication non-adherence constitutes a significant obstacle to hypertension care worldwide 72 with a prevalence between 20% and 50% (Nafradi, Galimberti, Nakamoto, & Schulz, 2016). In 73 Australia, the mean non-adherence to antihypertensive therapy is about 15% (Carvalho & 74 Santos, 2019). Poor adherence is considered a major problem and is associated with 75 suboptimal clinical outcomes, increased emergency-room visits, and hospitalizations all of 76 which contribute to an increased burden on the health care system (Lemay, Waheedi, Al-77 Sharqawi, & Bayoud, 2018). 78 79 Effective strategies for the treatment of hypertension should include a good understanding of 80 the barriers to medication adherence. Patient-related characteristics such as health literacy, 81 health beliefs and satisfaction with health care are potentially modifiable factors that may 82 influence patient adherence to medication (Al-Ruthia et al., 2017). 83 84 Theoretical models of patient behaviour can be useful in designing interventions to improve 85 medication adherence (Kucukarslan, 2012). One model that has shown promise for identifying 86 potential targets for interventions is the Necessity-Concerns Framework (Tibaldi et al., 2009), 87 which is a multidimensional theory that posits relationships between two separate 88 dimensions-patients' necessity beliefs and concerns regarding medication, and an outcome 89 (medication adherence) (Phillips, Diefenbach, Kronish, Negron, & Horowitz, 2014) .This 90 5 suggests that patients weigh up their perceived personal need for treatment against their 91 concerns about potential adverse effects of treatment when deciding whether or not to adhere 92 to medications (Tibaldi et al., 2009). Thus, medication adherence will be greater when the 93 difference between patients' beliefs in the necessity of the medication exceed their concerns 94 (Phillips et al., 2014). The Beliefs about Medicines Questionnaire (BMQ-specific) has two 95 subscales to assess patients' perceived need for treatment (necessity) and their concerns 96 about potential adverse effects (concerns) (Jimenez et al., 2017;Tibaldi et al., 2009). A 97 relatively recent meta-analysis (Rob Horne et al., 2013) , has described how beliefs about 98 medicines determined through the BMQ are correlated to medication adherence in a number 99 of chronic illnesses, including hypertension, and reported that medication adherence was 100 correlated positively with necessity beliefs, and negatively with concern, across the majority 101 of included studies. 102

103
The management of hypertension adds a new demand to already existing daily life tasks, 104 especially in vulnerable populations (Pesantes et al., 2015), such as refugees under stress 105 after fleeing from their home countries to seek safety. Over the past years conflicts in some of 106 Middle Eastern countries, such as Lebanon, Algeria, Sudan, Libya, Iraq, and Syria have 107 contributed to the many hundreds of thousands of refugees seeking safety in neighbouring 108 states and in more distant countries, such as Australia which has ranked as one of the top 109 three resettlement countries in the world (Fábos, 2015). 110 6 population, and considered under the same umbrella (Shahin, Stupans, & Kennedy, 2018). 119 However, each has an explicit legal definition that distinguishes one from the other (Mumford, 120 2016). Refugees forced to leave their countries of origin because of a direct threat of 121 persecution or death, and can't safely return home, whereas migrants face no such 122 impediment to returning to their country of origin as they choose to move mainly to improve 123 their lives by finding work, or in some cases for education, family reunion, or other reasons 124 (Edwards, 2016). 125 126 It has been reported in the literature that migrants and refugees have an elevated prevalence 127 of medical diseases, such as hypertension. Although, both these populations may have similar 128 difficulties during the resettlement processes, a number of factors distinctly differentiate the 129 social and personal lives of refugees and migrants. These factors may include: the intentions 130 and motivations for migration, the impediments to returning back to home countries, having a 131 control over their lives through migration, and taking into account the damaging effect of 132 persecution on their education, employment, and socioeconomic status (Shahin et al., 2020). 133 Therefore, these two different populations might evolve different beliefs about their 134 medications. Thus, it is important to have a well-founded understanding of how Middle Eastern 135 refugees and migrants perceive their prescribed medicine, and how these beliefs about 136 medicines may impact medication adherence. 137

138
Medication non-adherence in the Middle Eastern population was addressed in a systematic 139 review, and estimated to be 48% in chronic illnesses, such as hypertension, diabetes and 140 chronic obstructive pulmonary disease (Al-Qasem, Smith, & Cliffor, 2011) . 141 According to the findings of the meta -analysis described above (Rob Horne et al., 2013), 142 only two studies examined the association between medication adherence and medication 143 beliefs in Middle Eastern population (Aflakseir, 2012;Fawzi et al., 2012). However, neither of 144 these two studies examined adherence and medication beliefs in hypertension, also neither 145 evaluated refugees' population, or the differences between refuges and migrants. Medication 146 adherence was correlated positively with necessity beliefs, and negatively with concern in 147 these two studies (Aflakseir, 2012;Fawzi et al., 2012). 148

149
To date, there has not been any research that has investigated the role of medication beliefs 150 on medication adherence of Middle Eastern hypertensive refugees and migrants in Australia 151 or indeed in other countries, or that has assessed the differences between these two groups 152 regarding their beliefs about medications. The aim of this study was to explore the relationship 153 between beliefs about medicine and adherence in Middle Eastern refugees and migrants in 154 Australia, and also to assess the differences between both groups with regards to medication 155 beliefs and adherence. 156 157

159
This study is a part of a larger cross-sectional study that was conducted in several non-profit 160 organisations supporting refugees and migrants in Melbourne, Australia. The settings for the 161 study were selected because they specialise in supporting Middle Eastern refugees and 162 migrants. Facebook was also used to recruit refugees and migrants through Arabic community 163 groups that included large populations of Middle Eastern refugees and migrants who were 164 sharing their experiences and interests via these groups. The recruitment process, and data 165 collection were previously reported (Shahin et al., 2020). Ethics approval for this study was 166 obtained from RMIT University Ethics Committee, (SEHAPP 53-18). 167

Study participants
168 Study participants have been reported and published previously. Throughout the 10-month 169 recruitment period, attendees at Adult Migrant English Program centres and the community 170 groups, were approached and invited to consider participating in the study. A poster including 171 the survey link was published in some Facebook Arabic interest gathering groups in Australia 172 (Shahin et al., 2020). Participants were invited to take part in the study if they were aged 30 173 years or older, diagnosed with hypertension, and had migrated to Australia from one of the 22 174 countries in the Middle East, as refugees or migrants. Refugee or migrant status was identified 175 through a survey question which asked participants to describe how they arrived to Australia. 176 The choices were: "refugee," "work," "studying," "economic reasons" "any other reason". 177 Migrants were defined as those participants who selected any choices of other than "refugee". 178 179

Development of questionnaire 180
The self-administered questionnaire consisted of 21 items divided into four sections. The first 181 section comprised socio-demographic information including; age, gender, place of birth, 182 migration status, education level, and occupation. In the second section, participants were 183 asked to identify whether they had one or more than one of eight significant chronic conditions 184 as categorised by the Australian Institute of Health and Welfare. These conditions included; 185 arthritis, asthma, back pain and associated problems, cancer, cardiovascular disease (such 186 as hypertension, coronary heart disease and stroke), chronic obstructive pulmonary disease 187 (COPD), diabetes and mental health conditions (Australian Institute of Health and Welfare, 188

2016). 189
In sections three and four, validated and reliable questionnaires were used to assess 190 medication adherence (Morisky, Green, & Levine, 1986) and medication beliefs (R. Horne & 191 Weinman, 1999). Medication adherence was measured using the four-item Medication 192 Adherence Questionnaire (MAQ) that assesses both intentional and unintentional non-193 adherence. The scale is scored 1 point for each "no" and 0 points for each "yes". Patients were 194 described as adherent (if the total score was four) or non-adherent (if the total score was less 195 than 4) (Khan, Shah, & Hameed, 2014). This questionnaire was selected because it has been 196 used across many chronic illnesses, such as cardiovascular disease, and has demonstrated 197 adequate predictive validity and good reliability (Mann, Ponieman, Leventhal, & Halm, 2009). 198 The psychometric properties have been reported to range from adequate (Brooks et al., 1994;199 Morisky et al., 1986) to high (Erickson, Coombs, Kirking, & Azimi, 2001). 200 Beliefs regarding medications were measured using Beliefs about Medicine Questionnaire 201 (BMQ). The BMQ-specific is a 10-item questionnaire that comprised of two scales; a 5-item 202 necessity scale that assesses perceived personal need for the medication (necessity), and a 203 5-item concerns scale that assesses common concerns about potential adverse effects such 204 as dependence, adverse effects, or accumulation effects (concerns). Participants indicate how 205 much they agree with a series of statements on a five point scale from strongly agree to 206 strongly disagree. Subscales scores were calculated as mean item scores. Higher scores 207 indicate a stronger endorsement of the construct being measured (Tibaldi et al., 2009). 208 Following the necessity concerns framework posited by Horne and colleagues, the difference 209 between necessity and concern subscale scores was calculated in order to evaluate the 210 patient-perceived, cost-benefit, analysis of medication adherence (Jimenez et al., 2017). 211

212
The questionnaires were available in English language and were translated to Arabic language 213 by a bilingual researcher whose first language was Arabic, and then they were back-translated 214 to English by another bilingual researcher. The original questionnaires were compared with 215 the back-translated version by two researchers whose first language was English. No 216 significant differences or changes in meaning were detected and hence the translated 217 questionnaires were used in the study.  variables. Socio-demographics characteristics for this sample have been published previously 236 (Shahin et al., 2020), and are reproduced here in Table 1. Associations of independent and 237 dependent variables for the two groups were examined using Chi-square tests, and 238 differences between the two groups were examined using t-tests. Bivariate associations for 239 dependent variables were examined using Pearson's correlations (r). A two-tailed significance 240 level of p<0.05 was used to evaluate all statistical analyses. 241 A multiple mediation model in which necessity and concern mediate the association between 242 migration status (refugee or migrant) and medication adherence was tested as presented in 243 Figure 1. In order to analyse the model and determine the confidence interval for the indirect 244 effect, bootstrapping (5,000 samples) was applied using the SPSS PROCESS macro (Hayes, 245 2012). This process does not require the indirect effect to be normally distributed, therefore is 246 preferred to the Sobel's test (Sobel, 1986) (Preacher & Hayes, 2008). If the 95% bias-247 corrected confidence interval does not include zero, the indirect effect is statistically significant. 248 Confounding factors that were significantly associated with medication adherence were 249 entered in the mediation analysis as covariates. 250 Based on the empirical and theoretical framework (Aikens, Nease, Nau, Klinkman, & 251 Schwenk, 2005), cluster analysis was conducted to examine the differences between refugees 252 and migrants medication beliefs, and to examine the association of these clusters with 253 medication adherence. Firstly, four groups were generated by splitting the scores from 254 necessity and concerns items at the median; subsequently, participants were categorized, 255 according to Horne and his colleagues (R. Horne, Parham, Driscoll, & Robinson, 2009), into 256 one of four subgroups according to their attitudes towards medication; Skeptical (low 257 necessity, high concerns), Ambivalent (high necessity, high concerns), Indifferent (low 258 necessity, low concerns), and Accepting (high necessity, low concerns). Following this, 259 associations between adherence rates and the four belief groups were examined in both 260 groups, and differentiated using Chi-square tests, and analysis of variance. 261 262

Participants demographics and clinical characteristics 263
A total of 320 participants were recruited: 168 refugees, and 152 migrants. All participants 264 were born in the Middle East, and there were slightly more women than men in both groups. 265 The highest proportion of refugees were from Iraq and Syria. Significant differences between 266 refugees and migrants regarding demographics and clinical characteristics were found (Table  267 1), which have been discussed elsewhere (Shahin et al., 2020). Broadly, differences reflect 268 the characteristics expected in these groups. Refugees had lower levels of education 269  (Table 2). Refugees had 276 substantially lower adherence than migrants with a large effect size (p=0.0001, d= 0.81), and 277 they reported a significantly lower level of necessity beliefs with a medium effect size 278 (p=0. 0001, d=0.48). Refugees also demonstrated significantly higher concern beliefs about 279 medicines than migrants with a large to very large effect size (p=0.0001, d=1.04). Importantly, 280 the association between necessity and concern beliefs was low (r=-0.20, p=0.02), 281 indicating that concern beliefs are not strongly contingent upon necessity beliefs. This 282 confirms that necessity and concerns beliefs lie on different continua. Medication beliefs were correlated with adherence in both groups (Table 3)  After adjusting for employment, comorbidity and educational level, the relationship between 300 migration status (refugee or migrant) and medication adherence was mediated by both 301 13 concern about and necessity of taking medication. The standardized regression coefficient 302 between migration status was statistically significant with necessity (p = 0.004), and concern 303 (p =0.0001). Also, the standardized regression was statistically significant between status of 304 migration and medication adherence (p =0.0001). We tested the significance of this indirect 305 effect using bootstrapping procedures. The unstandardized indirect effect for necessity was 306 0.08, and the 95% confidence interval ranged from 0.03 to 0.12, and for concern 0.16 and the 307 95% confidence interval ranged from 0.10 to 0.23 Thus, the indirect effect was statistically 308 significant (see Table 4 & Figure 1). 309 310

322
This is the first study to explain the relationship between medication adherence and 323 medication beliefs in Australia using multiple mediation modelling, and also to differentiate Following from the findings of our study, interventions to improve medication adherence need 341 to focus on the 'skeptical' and 'ambivalent' clusters. The two clusters constitute of more than 342 70% of refugees, and 40% of migrants who have high concern beliefs about medicines. 343

344
The patients in the skeptical cluster were found to have the lowest level of medication 345 adherence, making the patients in this cluster a target for interventions that enhance 346 medication adherence. Concern beliefs about medicine need to be reduced in both refugee 347 and migrant groups and necessity beliefs need to be increased in refugees. 348 349 Though both skeptical and ambivalent patients are non-adherent, the type of interventions 350 needed for each group is different. Understanding the characteristics of each these clusters 351 for both refugees and migrants by healthcare providers may lead to appropriate interventions 352 for improving medication taking behaviours (Unni & Shiyanbola, 2016). 353

354
The findings of the current study are also consistent to studies cited in the literature that have 355 reported that having high accepting beliefs, and low skeptical beliefs is associated with higher 356 medication adherence (Mann et al., 2009;Tibaldi et al., 2009). 357

358
Studies from Middle East are limited regarding medication beliefs and medication adherence. 359 Our findings are consistent with those from a study conducted in Kuwait (Lemay et al., 2018), 360 that reported low adherence was associated with high concerns beliefs. 361 Medication beliefs may be modified with health care interventions and advice. These beliefs 362 also, are potentially influenced by various personal characteristics that may derive from culture 363 and religion (Al-Ruthia et al., 2017). Patients' cultural beliefs about medication-taking are also 364 factors contributing intentional medication non-adherence (Bussell, Cha, Grant, Schwartz, & 365 Young, 2017). Health care providers should be encouraged to recognize confusion and 366 misconceptions about medications in patients from different cultures and to provide sensitive 367 care to people from diverse ethnic backgrounds to achieve better medication adherence 368 (Shahin, Kennedy, & Stupans, 2019). 369 Taking prescribed medications regularly is imperative to maintaining adequate blood pressure 370 control in most hypertensive patients, especially for refugees who have high levels of mortality 371 due to chronic illnesses (Amara & Aljunid, 2014). Refugees in this study reported suboptimal 372 adherence levels, highlighting the need for attention that may improve the overall quality of 373 life for these vulnerable individuals who arrive in Australia. Most refugees in this study came 374 from countries that are currently involved in war or conflict. These countries experience severe 375 disruptions of their health systems resulting in a high degree of uncertainty regarding the 376 safety of seeking healthcare services (Shahin et al., 2020). Moreover, patients are often 377 reluctant to share intentions to not take medications and concerns with health care providers, 378 and therefore providers need to create an encouraging, blame-free environment to allow 379 patients to describe their medication-taking behaviour (Bussell et al., 2017). 380 Social support plays an important role in determining treatment uptake, recovery and 381 adherence. Refugees who have been taken away from their friends and families, lack social 382 support and thus, lower medication adherence and poorer recovery is to be expected (Shahin 383 et al., 2018). Our previously reported data (Shahin et al., 2020) are consistent with literature 384 which show that Middle Eastern refugees have low educational level, a low socioeconomic 385 status (Gil-González et al., 2015;Hjelm, Bard, Nyberg, & Apelqvist, 2003;Hjelm, Nyberg, 386 Isacsson, & Apelqvist, 1999;Njeru et al., 2016) and consequently a wide range of factors that 387 affect the quality of healthcare. These factors include education level, health literacy, income 388 level, employment status, insurance status and access to care (Roldan, Ho, & Ho, 2018). There is evidence that indicates educational interventions change migrants' and refugees' 406 concerns about medications and that this increases knowledge about hypertension and its 407 treatment (Unni & Shiyanbola, 2016). This study highlights also, the importance of 408 understanding the differences between refugees and migrants, and how each group has 409 different beliefs about their medications. Acquiring an awareness of each population's beliefs 410 about medicine may help healthcare providers to identify gaps in their own understanding and 411 the expectations of refugees and migrants about treatments. This may lead to the provision of 412 more optimal health care that meets the needs and expectations of each population (Shahin 413 et al., 2020). 414

415
This study has some limitations, due to the cross-sectional design, and self-report measures 416 used to assess both medication adherence and medication beliefs. Self-reporting adherence 417 might not be accurate due patients' poor memory and overestimation of adherence. However, 418 more than 50% of the participants in this study reported low levels of adherence suggesting 419 that overestimation was not a major limitation in this study. The assessment of the validity of 420 the Arabic versions of the questionnaires was beyond the scope of the current study. However, 421 the English versions have been validated (Robert Horne et al., 1999)  and concern beliefs about medicines, and adherence prior to providing counselling, and 429 medical advice. This study highlights the need for tailored interventions for each cluster of 430 beliefs, and to understand the characteristics of these clusters, to provide optimal healthcare 431 advice and counselling. This study also gives an insight to the need for future intervention 432 studies to promote medication adherence amongst vulnerable patients, by reducing concerns 433 and increasing necessity beliefs. 434

Ethics approval and consent to participate 436
Ethics approval for this study was obtained from RMIT University Ethics Committee, . 438

Consent for publication 439
Not applicable. 440

Availability of data and materials 441
The datasets used and/or analysed during the current study are available from the 442 corresponding author on reasonable request. 443