Better understanding the influence and complexity of beliefs on medication adherence in asthma

Better understanding the influence and complexity of beliefs on medication adherence in asthma

Accepted Manuscript Title: Better understandingPlease check the author indicator for “Adam La Caze”.–> the influence and complexity of beliefs on medi...

378KB Sizes 0 Downloads 1 Views

Accepted Manuscript Title: Better understandingPlease check the author indicator for “Adam La Caze”.–> the influence and complexity of beliefs on medication adherence in asthma Authors: Holly Foot, Adam La Caze, Peter Baker, Neil Cottrell PII: DOI: Reference:

S0738-3991(18)30874-7 https://doi.org/10.1016/j.pec.2018.10.010 PEC 6093

To appear in:

Patient Education and Counseling

Received date: Revised date: Accepted date:

16-5-2018 4-10-2018 11-10-2018

Please cite this article as: Foot H, La Caze A, Baker P, Cottrell N, Better understanding the influence and complexity of beliefs on medication adherence in asthma, Patient Education and Counseling (2018), https://doi.org/10.1016/j.pec.2018.10.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1. Title Page Better understanding the influence and complexity of beliefs on medication

Authors Holly Foot1; Adam La Caze; Peter Baker2; Neil Cottrell1

SC RI PT

adherence in asthma

1

School of Pharmacy, The University of Queensland, Brisbane, Australia

2

School of Public Health, The University of Queensland, Brisbane, Australia

Corresponding Author at:

U

Holly Foot

N

Pharmacy Australia Centre of Excellence

Level 4, 20 Cornwall St, Woolloongabba, QLD, 4102, Australia

A

Phone: +617 3346 1900

M

Fax: +617 3346 1999

A

CC

EP

TE

D

Email: [email protected]

1

Highlights There is a need to better understand non-adherence in individuals with asthma.



A multivariate regression model predicted 39% of variance in medication adherence.



The inclusion of interaction effects in the final model improved model prediction.



Beliefs interrelate with each other to change the relationship with adherence.

SC RI PT



2. 3. Abstract Objective

U

The aim was to better understand how beliefs influence medication adherence in asthma.

N

Methods

A

All participants were prescribed an inhaled corticosteroid for a diagnosis of asthma. Each

M

participant completed a survey consisting of: Beliefs about Medicines Questionnaire (BMQ), Brief-Illness Perception Questionnaire (B-IPQ) and Multi-dimensional Health Locus of

D

Control Scale (MHLCS). Adherence to inhaled corticosteroids was elicited using the

TE

Medication Adherence Report Scale (MARS). Multiple linear regression with interaction effects was used to identify significant predictors of medication adherence and interactions

Results

EP

between beliefs.

CC

A total of 198 participants completed the survey. The mean(±SD) MARS score was 19.2(±4.5). A multivariable model (adjusted R2=0.39) predicted adherence using: age, asthma

A

hospitalisation, timeline (B-IPQ) subscale, necessity and concern (BMQ) subscales, doctor (MHLCS) subscale and the two interaction effects (concerns [BMQ] moderated by chance [MHLCS] and treatment control [B-IPQ] moderated by understanding [B-IPQ]). Conclusion

2

The findings of this study contribute to a better understanding of the role of beliefs in medication adherence in asthma. Certain beliefs meaningfully interrelate and change the relationship they have with medication adherence.

SC RI PT

Practice Implications If these beliefs are causally related to medication adherence and can be intervened upon, the findings are useful for providing targets to personalise adherence support.

U

Keywords: beliefs; medication adherence; asthma; illness perceptions; locus of control beliefs;

A

CC

EP

TE

D

M

A

N

beliefs about medicines

3

4

4. 1. Introduction Adhering to medication is a complex health behaviour that involves a number of different factors.[1, 2] An individual’s beliefs about their medicines and their health informs the

SC RI PT

individual’s medication-taking behaviour.[3-5] Interventions that change an individual’s beliefs have been demonstrated to improve adherence[6] and health outcomes.[7, 8]

Theoretical models of health behaviour guide research in medication adherence.[9, 10] The

most commonly applied models include The Health Belief Model,[11] The Common Sense

U

Model[12] and The Locus of Control Model.[13] Each model assumes that individuals make

N

decisions through a deliberative and systematic process based on the beliefs or perceptions an

A

individual’s holds. The Beliefs about Medicines Questionnaire (BMQ),[14] Brief Illness Perception Questionnaire (B-IPQ)[15] and the Multi-dimensional Health Locus of Control

M

Scale (MHLCS)[16] have been developed to elicit an individual’s beliefs about medicines,

D

illness perceptions and beliefs in who or what controls their health outcomes, respectively.

TE

The BMQ consists of two subscales that seek to elicit an individual’s beliefs in the necessity and concerns of medicines.[14] Individuals who believe their medicine is necessary for

EP

current and long-term health are more likely to adhere to their medicine than those who

CC

do not endorse these beliefs.[4] Individuals who perceive fewer concerns about their medicine, such as believing their medicines cause little or no disruptions to daily life, are

A

also more likely to adhere.[4] The B-IPQ elicits an individual’s beliefs about their condition, with respect to the perceived consequences, how long they expect their condition to last (timeline), their ability to control (personal control) and treatment to control (treatment control), how many symptoms they perceive to be associated with their condition (identity) and their perceived understanding of their condition. The B-IPQ also

4

5 elicits an individual’s emotional illness representations, specifically how concerned (emotional representation 1) and emotionally affected (emotional representation 2) they are about their condition. Lastly, the MHLCS is divided into four subscales: internal (individual control); chance; doctor and other (powerful) people. The internal subscale

SC RI PT

elicits the degree to which an individual believes that their condition becoming better or worse is their own responsibility and due to their own behaviour. The chance, doctor and other people subscales represent measures of external locus of control. The chance subscale elicits the degree to which an individual believes that chance controls whether their condition gets better or worse. The doctor subscale elicits beliefs about the role of

the other people subscale elicits an

U

the doctor in controlling their condition and

N

individual’s beliefs about the role of other people, such as family and friends, in

A

determining whether their condition gets better or worse. Each of these scales have been

M

associated with measures of medication adherence across a number of different chronic

D

conditions. [17-21]

TE

A significant shortcoming of the adherence literature is that many researchers have relied on a single model of health behaviour to explain medication-taking behaviour. Attempting to

EP

explain adherence behaviour using only one belief scale relies on an oversimplified model of adherence. Generally, a single scale does not predict a large portion of variance in medication

CC

adherence.[10, 18, 19, 22] For example, a meta-analysis found that the necessity and concern beliefs are associated with adherence, but the correlation is weak (BMQ necessity, r=0.17;

A

BMQ concerns, r= -0.18).[18] A bivariate approach to adherence cannot account for the interrelationships between an individual’s beliefs about their medicines, illness and who or what controls their health. Adherence decisions are likely to be made through the interplay of an individual’s beliefs, leading the individual to make a decision about whether they should adhere to their medication.

5

6 Studies investigating the relationships between the BMQ, B-IPQ and MHLCS suggest that the constructs examined by these scales are interrelated, but little work has been conducted to determine whether these relationships meaningfully relate to medication adherence. Scores from the BMQ have been found to correlate with scores from the B-IPQ[9, 23, 24] and

SC RI PT

MHLCS.[25] Specifically, the necessity (BMQ) subscale positively correlates with the consequences (B-IPQ) subscale and timeline (B-IPQ) subscale in asthma.[23] In chronic pain,

the necessity (BMQ) subscale was likewise correlated with consequences (B-IPQ) but researchers also reported that the concerns (BMQ) subscale was correlated with consequences

(B-IPQ) and illness emotion (B-IPQ).[24] In individuals taking lipid-lowering medication, the

U

powerful others (MHLCS) subscale was correlated with both the necessity (BMQ) and

N

concerns (BMQ) subscale.[25]

A

These interrelationships are important to consider as they may change the way different beliefs

M

influence medication adherence. To provide one example, Hong et al. investigated the way

D

perceived barriers to taking medication (e.g. side effects, forgetting to take medication) were

TE

associated with medication adherence in patients taking anti-hypertensives. This study found that the way an individual’s perceived barriers influenced adherence depended on that

EP

individual’s locus of control beliefs.[26] When participants reported having minimal medication barriers, those with an internal locus of control had better adherence than those who

CC

did not. However, when medication barriers were reported to be high, those with an internal locus of control had the worst adherence.[26] The authors proposed that those with an internal

A

locus of control prefer to be in control and when medication barriers are high, these individuals may become overwhelmed, fear they are losing control and therefore regain control by changing the way they take their medication.[26] We propose that better understanding the ways in which beliefs about medicines, illness perceptions and locus of control beliefs

6

7 collectively influence adherence will provide improved insight into medication-taking behaviour. There is a need to better understand non-adherence in individuals with chronic conditions. One

SC RI PT

such condition is asthma, where poor medication adherence contributes to increased symptoms, healthcare utilisation and mortality.[27-31] Adherence rates of less than 80% to inhaled

corticosteroids have been associated with poor asthma control (nocturnal symptoms, exacerbations and limitation on physical activity)[32, 33] and the need for ventilation.[34] Less-than-perfect (<100%) adherence has been shown to increase the risk of asthma-related

U

hospitalisation.[35] To support individuals with asthma in taking their medicines, a better

N

understanding of the role of beliefs in medication-taking behaviour is needed.

A

The aim of this study was to better understand how beliefs influence medication adherence in

M

asthma.

D

5. 2. Methods

TE

A cross-sectional study was conducted in individuals with asthma. Participants were recruited through community pharmacies in Brisbane, Australia and online through an advertisement on

EP

a national asthma organisation website and an university online newsletter (sent to all current and past staff and students) between May 2014 and May 2015. Patients at one of two

CC

community pharmacies who presented a prescription for an asthma medication when the researcher was present or were known to the pharmacy to be on a preventative asthma

A

medication were approached by the researcher while their medication was being dispensed. All Individuals were invited to participate in the study if they met the following: 16 years of age or older; responsible for taking their own medication; had a diagnosis of asthma and were prescribed an inhaled corticosteroid, with or without long acting beta2-agonist (LABA) for at 7

8 least three months. All participants provided written consent. This study was approved by The School of Pharmacy Ethics Committee, The University of Queensland (2014/8). All participants completed a survey, composed of participant characteristics and previously

SC RI PT

validated questionnaires to elicit their medication adherence to and health beliefs about their inhaled corticosteroid. Self-report measures, rather than objective measures, were used in this

study to elicit both the extent and reasons behind an individual’s adherence behaviour.[36] Participants recruited through community pharmacy completed a paper based survey at

the pharmacy and handed it to the researcher once completed. Participants recruited

U

online completed the survey through a web-based survey system.

N

Participant characteristics collected were: age, gender, level of education, duration of

A

asthma, hospital admissions in the past two years because of asthma, prescribed

M

preventer and frequency of use, short-acting beta-agonist (SABA) use, concession card (eligible to receive medication at a lower cost) and number of other prescribed

D

medication.

TE

An individual’s adherence to their inhaled corticosteroid was assessed using the Medication Adherence Report Scale (MARS).[23, 37] This consists of five statements regarding adherence

EP

to medication that are answered on a 5-point Likert scale with an overall score ranging from 5

CC

to 25, with a high MARS score indicating better adherence. In asthma, MARS correlates with objective measures of adherence such as prescription refills (r=0.46),[38] MEMS

A

(r=0.48)[39] and pill counts (r=0.53).[40] MARS has good test-retest reliability (r=0.74)[40] and has shown good reliability in the way participants complete it (Cronbach’s alpha=0.75 – 0.83).[40-43] The BMQ Specific was used to elicit a participant’s beliefs about their inhaled corticosteroid.[14] Participants indicated their level of agreement with ten statements through 8

9 a five-point Likert scale, with scores being summed to give a necessity and concerns subscale score (5 – 25). The BMQ was originally validated in six different chronic illness groups, including asthma.[14] In asthma, the BMQ subscales had good internal consistency and test-retest reliability.[14] Criterion validity was demonstrated through positive

SC RI PT

correlations with similar subscales on the Illness Perception Questionnaire, Sensitivity

Soma scale and the Reported Adherence to Medication scale.[14] Discriminant validity

was shown through BMQ scores being different in different illness and treatment groups.[14]

U

The B-IPQ elicits the cognitive and emotional representations an individual holds about their

illness.[15] In the questionnaire, the term ‘illness’ was replaced with ‘asthma’ and ‘treatment’

N

was replaced by ‘preventative inhaler’ as suggested by the authors to make it more relevant for

A

participants.[15] All subscales are measured on an 11-point scale (0-10), with each subscale

M

measuring a different construct (e.g. perceived timeline of asthma). The B-IPQ has good test-

D

retest reliability, with all subscales showing significant correlations when repeated in

TE

participants at a three and six week time period.[15] Furthermore, it has shown satisfactory concurrent validity by having strong correlations with similar subscales of

EP

the BMQ, validated self-efficacy scales and predicting patients HbA1c.[15] The B-IPQ was also able to distinguish between different illnesses, demonstrating discriminant

CC

validity.[15]

A

Form C of the MHLCS was used in this study to elicit an individual’s beliefs in the controllability of their health.[16] Participants indicated their level of agreement on a 6-point Likert scale. The internal and chance subscale scores can range from 6 – 36 and the other people and doctor subscales range from 3 – 18.

The MHLCS displays satisfactory

psychometric properties. Test-retest reliability was satisfactory for all subscales except for the other people subscale which had generally low stability in the initial 9

10 development.[16] Importantly, locus of control beliefs are thought to change over time, so a very high test-retest reliability is not expected. Form C correlated well with similar subscales of form B of the MHLCS and did not correlate with dissimilar subscales of form

SC RI PT

B indicating acceptable concurrent and discriminate validity.[16] 2.1 Data Analysis

Data analysis was conducted in R.[44] Descriptive statistics were used to describe the sample population. A multiple linear regression line was calculated to predict self-reported medication

adherence (MARS score) based on participant demographics, clinical variables and scores from

U

the BMQ, B-IPQ and MHLCS. Categorical variables (e.g. gender) were dummy-coded so that

each category was entered in as a dichotomous variable (e.g. female = 1, male = 0). All

A

N

variables were checked for normal distribution and multi-collinearity.

M

Variable and model selection was based on the Akaike Information Criterion (AIC) measure.[45] A preliminary model was first identified without any interaction effects

D

added. Once the preliminary model was developed, identification of significant interaction

TE

effects between subscales was conducted in two steps. Firstly, interactions were explored between different belief questionnaire subscales (e.g. necessity [BMQ] and doctor [MHLCS]

EP

subscales) and between belief subscales, demographics and clinical characteristics using linear regression. Interactions that were identified in this step as being statistically significant (p-

CC

value of the t-statistic <0.05) were then added into a preliminary model and backwards stepwise

A

linear regression was used to identify the final model. All predictor variables and interaction terms were standardised (z scoring) and centred (subtracted the data mean score from the individual data point, then divide each point by the standard deviation from all data points) to reduce multi-collinearity problems associated with interaction effects.[46, 47] Standardising variables significantly reduces the correlation between the two terms.[48, 49]

10

11 Effect sizes for each predictor variable were reported as standardised β-coefficients, which correspond to a unit increase (decrease if negative) in MARS score for every one standard deviation change in the predictor variable. Standardised β-coefficients are comparable to each other. Semi-partial correlations (squared) were reported as a method of assessing the relative

SC RI PT

importance of each variable in determining the MARS score.[49] Variance Inflation Factors

(VIF) were used as an index of multi-collinearity between variables.[49] Generally, if a variable has a VIF larger than 10, the variable is considered to have significant multicollinearity.[49]

U

Model fit parameters were examined through visual inspection of four different graphs. A

Residuals vs Fitted plot and Scale-Location plot was produced to check the assumption of

N

homoscedasticity. A Q-Q plot was produced to examine the distribution of residuals and

A

inspect for outliers in the data. A Residuals vs Leverage plot was produced to identify any data

M

points that may have a large influence on the model fit.[50]

D

6. 3. Results

TE

The characteristics of the sample are described in Table 1. One hundred and ninety eight participants completed the survey. The majority of participants were recruited through a

EP

national asthma organisation website (n=126). In addition, 49 participants were recruited

CC

through the university newsletter advertisement and the remaining 23 participants were recruited in community pharmacy. For participants recruited in community pharmacies

A

by the principal investigator (HF), two individuals declined to enter into the study and three individuals were recruited and then excluded as they did not meet the inclusion criteria. Reasons for declining were time and not wanting to be involved. This information was unable to be captured in the online setting. The bivariate analysis of this cohort has been reported previously.[51]

11

12 Table 1: Participant characteristics and association with MARS score. Characteristics Mean (±SD) age, years, n=195 Gender, n (%) Female Male

value 39.8 (±12.7) 161 (81.7) 36 (18.3)

Level of Education, n (%) Grade 10 Grade 12 Diploma/Certificate University Further Degree Other

SC RI PT

16 (8.2) 29 (14.8) 22 (11.2) 81 (41.3) 46 (23.5) 2 (1.0)

47 (24.2) 24.4 (±14.2)

M

N

A

Main prescribed inhaled corticosteroid ± LABA, n (%) Fluticasone/salmeterol Budesonide/eformoterol Fluticasone Others

U

Concession card, n (%), n=189 Mean (±SD) asthma duration, years, n=184

65 (45.8) 77 (54.2)

D

SABA use, n (%) < 3 times per week ≥ 3 times per week

87 (44.6) 72 (36.9) 12 (6.2) 24(12.3)

EP

TE

Hospitalisation in last 2 years because of asthma, n (%) No Yes

153 (78.1) 43 (21.9) 78 (40.8) 97 (50.8) 13 (6.8) 3 (1.6)

Mean (±SD) MARS Score, n= 174

19.2 (±4.5)

A

CC

Other prescribed medicines, n (%) None 1-4 5-9 ≥10

LABA, long-acting beta-agonist; SABA, short-acting beta-agonist; SD, Standard Deviation The timeline (B-IPQ) subscale was found to be severely skewed towards higher scores and for this reason, scores were dichotomised into forever (10) and not forever (<10). Asthma duration was also positively associated with concession card (p=0.035) and age (p<0.001), which could 12

13 cause multicollinearity if all were added into the model. To avoid multicollinearity, age was found to be the strongest predictor of MARS score and was the only one of these variables entered in the final model.

SC RI PT

The inclusion of interaction effects improved model prediction (concern (BMQ) subscale moderated by chance (MHLCS) subscale and treatment control (B-IPQ) subscale

moderated by understanding (B-IPQ) subscale). (The preliminary model without interaction effects is shown in Supplementary File 2. )

The final regression model with main effects and interaction terms, explained 39% of the

U

variance (F (8,121) = 11.35, p<0.001, Adjusted R2 = 0.39) in MARS score. MARS score

0.82





0.52











0.90







1.18















EP





TE

0.56



1.56

D

1.61 1.02



A

19.0

M



N

was found to be predicted by the following equation:

Where hospitalisation has been dummy-coded and is coded 1 for yes and 0 for no and timeline (B-IPQ) score has been coded 1 for not forever (score of <10) and 0 for forever (score of 10).

CC

U is an error term that represents variables that were not measured.

A

* represents an interaction effect

Overall, model fit graphs showed that the final model fits the data well and that the assumptions of homoscedasticity (variance in residuals does not change with different values of the predictor variables) and normally distributed residuals are met (Supplementary File 1).

13

14 Table 2 displays the effect size for each variable. The doctor (MHLCS) subscale uniquely accounts for 12% of the variance in MARS score (semi-partial correlation squared=0.12). All variables in the model had a VIF of less than 2, indicating little evidence of multi-collinearity.

Supplementary File 3.

SC RI PT

An effect plot for each of the independent predictors in the final model is shown in

Table 2: Adjusted standardised β-coefficient, standard error (SE), p-value, semi-partial correlation squared and variance inflation factor (VIF) for each predictor variable and interaction effects on MARS score (n=130) Adjusted

SE

0.82

0.30

0.01

0.032

1.04

1.56

N

Semi-partial correlation squared

Standardised β

0.08

0.015

1.16

0.74

0.11

0.012

1.06

0.52

0.34

0.13

0.011

1.25

1.61

0.32

<0.001

0.12

1.10

Concern (BMQ) subscale

-0.90

0.34

0.008

0.034

1.23

Concerns (BMQ) subscale *Chance (MHLCS) subscale

-1.02

0.30

<0.001

0.055

1.07

Treatment control (B-IPQ) subscale * Understanding (BIPQ)subscale

-0.56

0.24

0.02

0.026

1.11

Necessity (BMQ) subscale

EP

TE

D

Doctor (MHLCS) subscale

1.18

M

HospitalisationYes Timeline (B-IPQ) subscale– Not forever

0.89

A

Age

p-value

U

Variable

VIF

CC

Variables entered but not retained in final model: Prescribed more than 1 preventer and Other People (MHLCS) subscale. * represents an interaction effect

A

The interaction terms that were found to be significant in the model are presented graphically in Figure 1 and 2. In Figure 1, the relationship between MARS scores and concerns (BMQ) scores at different scores on the chance (MHLCS) subscale is presented. When an individual has a score of the mean or above on the chance (MHLCS) subscale (long dashed, dotted and

solid line) (hold some belief in chance to control their asthma outcomes), there is a significant 14

15 negative correlation between the concern (MHLCS) subscale score and MARS score (Pearson’s correlation= -0.46, p<0.001). Conversely, when an individual scores lower than the mean on the chance (MHLCS) subscale (dot-dash and dashed line) (do not hold strong beliefs in chance to control their asthma), there is no statistically significant correlation between the

SC RI PT

concern (BMQ) subscale and MARS score (Pearson’s correlation= -0.001, p=0.99). This

indicates that the chance (MHLCS) subscale is moderating the correlation between concerns (BMQ) score and MARS score.

The second significant interaction observed was between the treatment control (B-IPQ)

U

subscale and understanding (B-IPQ) subscale. Figure 2 shows that the relationship between

treatment control (B-IPQ) scores and MARS scores is different depending on the score of the

N

understanding (B-IPQ) subscale. Specifically, for those who have a score lower than the mean

A

on the understanding (B-IPQ) subscale (long dashed, dot-dash and dashed line) (do not hold

M

strong beliefs that they understand their asthma), there is a positive correlation between

D

treatment control (B-IPQ) scores and MARS scores (Pearson’s correlation=0.29, p=0.01). For

TE

those with a score of the mean or higher on the understanding (B-IPQ) subscale (dotted and solid line) (strong understanding of their asthma), there is no statistically significant correlation

EP

between treatment control (B-IPQ) scores and MARS scores (Pearson’s correlation= -0.06, p=0.57). This indicates that the understanding (B-IPQ) subscale is moderating the correlation

CC

between treatment control (B-IPQ) scores and MARS scores in this population.

A

7. 4. Discussion and Conclusion 4.1 Discussion This study highlights the complex ways in which beliefs influence medication adherence in asthma. Age, hospitalisation in the last two years due to asthma, necessity and concern beliefs,

timeline illness perceptions and doctor locus of control beliefs all predicted self-reported medication adherence. Further, some of the relationships between beliefs and self-reported 15

16 medication adherence were moderated by other beliefs. Specifically, participants who scored high on the concerns (BMQ) subscale reported less adherence if they also scored high on the chance (MHLCS) subscale, and participants who rated their disease understanding (B-IPQ) as poor were more likely to report less adherence if they scored low on the treatment control (B-

SC RI PT

IPQ) subscale. The final model accounted for 39% of the variance seen in participants’ selfreported medication adherence, which is significantly more than seen in the literature using only the BMQ and IPQ (Aflakseir [R2=0.25],[52] Rajpura and Nayak [R2=0.21][53] and Morgan et al. [R2=0.20][54]) and the BMQ and MHLCS (Berglund et al. [R2=0.06][25]).

U

In common with previous research, beliefs were a predictor of medication adherence in the sample.[1, 23, 55] Belief that the doctor can control asthma outcomes (elicited by the doctor

N

[MHLCS] scale) accounted for the most variance (12%) in the regression model. This was ten

A

times greater than the predictive power of the necessity (BMQ) subscale and nearly four times

M

greater than the concerns (BMQ) subscale. This finding contributes to the literature as the

D

relationship between doctor locus of control beliefs and medication adherence is rarely

TE

investigated.[56] Other facets of the patient-doctor relationship have been found to be important for medication adherence. These include physician communication,[5, 57] patient

EP

attitudes towards doctors[58] and trust in the doctor.[59, 60] Across different chronic diseases, one study reported that negative attitudes towards doctor were the most important factor

CC

associated with medication adherence.[58] It is possible that the doctor locus of control scale is eliciting the same (or part of the same) constructs that relate to an individual’s beliefs around

A

other aspects of the patient-doctor relationship. Interactions effects have not been widely explored in the adherence literature or described in theoretical models, but were important predictors in the final model. Different scores on the chance (MHLCS) scale were found to moderate the relationship between concern beliefs and

medication adherence. The correlation between the concerns (BMQ) subscale and MARS score 16

17 for individuals with a high chance (MHLCS) subscale score was stronger (r=-0.46, p<0.001) compared to those who did not have a high chance (MHLCS) subscale score (r=-0.001, p=0.99). This finding is in line with health behaviour models. Individuals who believe their health outcomes are ultimately due to chance may be less likely to endure concerns regarding

SC RI PT

the side effects of medication and therefore stop taking their medication.[61] This interpretation suggests that chance locus of control beliefs changes the way individuals react when they

experience concerns, such as perceived side effects of their medicine.[61] Individuals who do not believe that chance controls their asthma may be more likely to continue to take their medicines even when they have concerns, as they may believe it is the only way to control their

U

asthma. If this finding is a true effect, it suggests that chance locus of control beliefs should be

N

taken into account when considering the relationship between concern beliefs and medication

A

adherence. In our study population of individuals with asthma, this suggests that addressing

M

concern beliefs may be more relevant in those individuals who also have a strong belief in chance to control their illness. Particular concern beliefs that may be relevant to individuals

D

who also hold concomitant beliefs that their asthma is due to chance, may be around potential

TE

long-term side effects and dependence on their inhaled corticosteroids.

EP

The second interaction showed that an individual’s score on the understanding (B-IPQ) subscale moderated the correlation between the treatment control (B-IPQ) subscale and MARS

CC

scores. In individuals with low scores on the understanding (B-IPQ) subscale, treatment

control (B-IPQ) scores were positively correlated to MARS scores. This means that only in

A

individuals who self-reported a poor understanding of asthma, were beliefs about treatment to control their asthma correlated to medication adherence. In individuals with a high score on the understanding (B-IPQ) subscale, there was no correlation between treatment control (B-

IPQ) scores and MARS scores. One interpretation of this finding is that individuals with good understanding of their asthma, answers questions regarding their treatment control differently. 17

18 In individuals with poor understanding of their asthma, whether or not they believe treatment controls their asthma is an important predictor their adherence behaviour. In individuals who perceive to have a good understanding of their asthma may interpret the questions about treatment control in a different way; i.e. does the treatment always control my asthma? These

SC RI PT

participants might recognise that their treatment doesn’t always control their asthma while at the same time recognise the importance of taking their treatment. This interaction effect has

not been well described previously. While holding strong beliefs in treatment to control asthma (elicited by the treatment control (B-IPQ) subscale) has been previously correlated with

different measures of medication adherence in asthma[62] and other chronic conditions,[63-

U

71] we believe this is the first study to find that the relationship is moderated by perceived

N

understanding of asthma. Assuming there is a causal relationship, this can be interpreted that

A

addressing treatment control beliefs and improving patient understanding of their illness can

M

improve medication adherence in a population of individuals with asthma.

D

This study is not without its limitations. The findings of this study should be tested in a larger

TE

asthma population to ensure robustness and to test the predictability of the regression model. In particular, it would be important to validate the negative association between timeline (B-

EP

IPQ) perceptions and MARS score as this was an unexpected finding. The timeline (B-IPQ) subscale was not a significant correlate of MARS in the bivariate analysis[51] and previous

CC

work suggests that believing a chronic illness will last forever is associated with better adherence.[62] Our findings are limited by the skewed results and having to dichotomise the

A

timeline (B-IPQ) subscale which may have influenced the findings. We acknowledge self-

report methods of adherence can be prone to recall bias and social conformity. Future research should explore whether the regression model predicts adherence using other methods, such as prescription refills. Also, this study did not elicit all the beliefs that are likely to influence medication-taking behaviour in this sample. The focus on beliefs 18

19 derived from the Health Belief Model, Common Sense Model and Locus of Control Model was based on empirical support and ability to measure beliefs related to the models in a convenient way. Patient characteristics that were not elicited may have influenced our results. Disease severity was not objectively measured in the study. People with severe

SC RI PT

asthma may have very different reasons for holding certain beliefs, compared to someone

who has never had a severe exacerbation of their asthma. Disease severity is also thought

to be positively correlated with medication adherence in less serious diseases such as asthma.[72] Indirect measures of severity used in the study were asthma-related

hospitalisations in the last two years and SABA use. Responses to the indirect measures

U

were typical for an asthma population and did not reflect a sample with severe asthma.

N

In the future it would be useful to elicit oral corticosteroid use in participants as a better

A

indication of asthma severity, including previous asthma exacerbations.

M

The key contribution of this research is that better understanding the interplay of different

D

beliefs and medication adherence provides deeper insight the factors that influence medication-

TE

taking behaviour. Interaction effects may explain the weak correlations previously reported in the literature. To be able to make this contribution, the interpretation of the findings of this

EP

study are bound by a number of assumptions. We assume that the self-report methods used (BMQ, B-IPQ and MHLCS) are measuring the important constructs described by the Health

CC

Belief Model, Common Sense Model and Locus of Control Model. Further, that the constructs measured by the BMQ, B-IPQ and MHLCS are based on specific beliefs than an individual

A

holds and that, to some extent, modifying these specific beliefs will cause a corresponding change in the scale score. Adherence research is limited by the difficulty in assessing whether some variables associated with adherence are causes or consequences of adherence (they may also be both). To be able to inform interventions we must begin to identify which variables may be amenable to an intervention to improve medication adherence. 19

20 This requires giving the regression model a causal interpretation. This paper proposes that if these beliefs are modified in individuals, we assume we will see a change in the scales measuring those beliefs and, if the causal assumptions are correct, we will see a change in

4.2 Conclusion

SC RI PT

adherence behaviour as measured by MARS score.

The results of the regression analysis have contributed to and improved our understanding of

the relationships between health beliefs and medication-taking behaviour in individuals with asthma. Health beliefs elicited from validated questionnaires were found to predict self-

reported medication adherence. Certain beliefs meaningfully interrelate with each other and

N

U

change the relationship they have with medication adherence. 4.3 Practice Implications

A

This research opens up new avenues of inquiry for understanding medication adherence and

M

developing interventions to better support medication-taking behaviour. If these beliefs are causally related to medication adherence and can be intervened upon, the findings of this study

TE

EP

requirements.

D

are useful for providing targets to better personalise adherence support to the individual’s

A

CC

Conflict of Interest: None

20

21

References 8. [1] E. Sabaté, Adherence to long-term therapies: evidence for action, World Health Organisation, Switzerland, 2003. 9. [2] H. Leventhal, L. Cameron, Behavioral theories and the problem of compliance, Patient

SC RI PT

Educ. Couns. 10 (1987) 117-38. http://dx.doi.org/10.1016/0738-3991(87)90093-0. 10. [3] E.J. Unni, O. Shiyanbola, K.B. Farris, Medication adherence: a complex behavior of medication

and

illness

beliefs,

Aging

Health

http://dx.doi.org/10.2217/ahe.13.41.

9

(2013)

377-87.

11. [4] R. Horne, J. Weinman, Patients' beliefs about prescribed medicines and their role in

N

http://dx.doi.org/10.1016/S0022-3999(99)00057-4.

U

adherence to treatment in chronic physical illness, J. Psychosom. Res. 47 (1999) 555-67.

A

12. [5] E. Vermeire, H. Hearnshaw, P. Van Royen, J. Denekens, Patient adherence to treatment:

M

three decades of research. A comprehensive review, J. Clin. Pharm. Ther. 26 (2001) 33142. http://dx.doi.org/10.1046/j.1365-2710.2001.00363.x.

D

13. [6] T.M. Nguyen, A. La Caze, N. Cottrell, Validated adherence scales used in a

TE

measurement-guided medication management approach to target and tailor a medication adherence intervention: a randomised controlled trial, BMJ open 6 (2016) e013375.

EP

http://dx.doi.org/10.1136/bmjopen-2016-013375. 14. [7] K.J. Petrie, L.D. Cameron, C.J. Ellis, D. Buick, J. Weinman, Changing illness

CC

perceptions after myocardial infarction: An early intervention randomized controlled trial, Psychosom. Med. 64 (2002) 580-6.

A

15. [8] K.J. Petrie, K. Perry, E. Broadbent, J. Weinman, A text message programme designed to modify patients' illness and treatment beliefs improves self-reported adherence to asthma preventer

medication,

Br.

J.

Health.

http://dx.doi.org/10.1111/j.2044-8287.2011.02033.x.

21

Psychol.

17

(2012)

74-84.

22 16. [9] R. Horne, Representations of medication and treatment: Advances in theory and measurement, in: K. Petrie, J. Weinman (Eds.), Perceptions of health and illness: Current research and applications, Harwood Academic Publishers, Amsterdam, 1997, pp. 155-88. 17. [10] E.A. Holmes, D.A. Hughes, V.L. Morrison, Predicting adherence to medications using

SC RI PT

health psychology theories: a systematic review of 20 years of empirical research, Value Health 17 (2014) 863-76. http://dx.doi.org/10.1016/j.jval.2014.08.2671.

18. [11] M.H. Becker, The Health Belief Model and Sick Role Behavior, Health Educ. Behav. 2 (1974) 409-19. http://dx.doi.org/10.1177/109019817400200407.

19. [12] H. Leventhal, D.R. Nerenz, J. Purse, Illness representations and coping with health

N

Associates, Hillsdale, NJ, 1984, pp. 219-52.

U

threats, in: A. Baum, J. Singer (Eds.), A Handbook of Psychology and Health, Eribaum

A

20. [13] J.B. Rotter, Generalized expectancies for internal versus external control of

M

reinforcement, Psychol. Monogr. 80 (1966) 1-28. http://dx.doi.org/10.1037/h0092976. 21. [14] R. Horne, J. Weinman, M. Hankins, The beliefs about medicines questionnaire: The

Psychol

TE

medication,

D

development and evaluation of a new method for assessing the cognitive representation of Health

14

(1999)

1-24.

http://dx.doi.org/10.1080/08870449908407311.

EP

22. [15] E. Broadbent, K.J. Petrie, J. Main, J. Weinman, The Brief Illness Perception J.

Psychosom.

Res.

60

(2006)

631-7.

CC

Questionnaire,

http://dx.doi.org/10.1016/j.jpsychores.2005.10.020.

A

23. [16] K.A. Wallston, M.J. Stein, C.A. Smith, Form C of the MHLC scales: a conditionspecific measure of locus of control, J. Pers. Assess. 63 (1994) 534-53. http://dx.doi.org/10.1207/s15327752jpa6303_10. 24. [17] R. Horne, S.C. Chapman, R. Parham, N. Freemantle, A. Forbes, V. Cooper, Understanding patients' adherence-related beliefs about medicines prescribed for long-term

22

23 conditions: a meta-analytic review of the Necessity-Concerns Framework, PLoS ONE 8 (2013) e80633. http://dx.doi.org/10.1371/journal.pone.0080633. 25. [18] H. Foot, A. La Caze, G. Gujral, N. Cottrell, The necessity-concerns framework predicts adherence to medication in multiple illness conditions: A meta-analysis, Patient

SC RI PT

Educ. Couns. 99 (2016) 706-17. http://dx.doi.org/10.1016/j.pec.2015.11.004.

26. [19] S.N. Kucukarslan, A review of published studies of patients’ illness perceptions and

medication adherence: Lessons learned and future directions, Research in social & administrative

pharmacy

:

RSAP

8

http://dx.doi.org/10.1016/j.sapharm.2011.09.002.

(2012)

371-82.

U

27. [20] E. Unni, O.O. Shiyanbola, Clustering medication adherence behavior based on beliefs

Med.

Res.

Opin.

A

Curr.

N

in medicines and illness perceptions in patients taking asthma maintenance medications, 32

(2016)

113-21.

M

http://dx.doi.org/10.1185/03007995.2015.1105204. 28. [21] E.L. O’hea, K.B. Grothe, J.S. Bodenlos, E.D. Boudreaux, M.A. White, P.J. Brantley,

D

Predicting medical regimen adherence: The interactions of Health Locus of Control beliefs,

TE

J. Health. Psychol. 10 (2005) 705-17. http://dx.doi.org/10.1177/1359105305055330. 29. [22] K. Brandes, B. Mullan, Can the common-sense model predict adherence in chronically patients?

A

EP

ill

meta-analysis,

Health.

Psychol.

Rev

8

(2013)

129-53.

CC

http://dx.doi.org/10.1080/17437199.2013.820986. 30. [23] R. Horne, J. Weinman, Self-regulation and self-management in asthma: Exploring the

A

role of illness perceptions and treatment beliefs in explaining non-adherence to preventer medication,

Psychol.

Health.

http://dx.doi.org/10.1080/08870440290001502.

23

17

(2002)

17-32.

24 31. [24] L.B. Nicklas, M. Dunbar, M. Wild, Adherence to pharmacological treatment of nonmalignant chronic pain: The role of illness perceptions and medication beliefs, Psychol. Health. 25 (2009) 601-15. http://dx.doi.org/10.1080/08870440902783610. 32. [25] E. Berglund, P. Lytsy, R. Westerling, Adherence to and beliefs in lipid-lowering

concern

framework,

Patient

Educ.

SC RI PT

medical treatments: A structural equation modeling approach including the necessityCouns.

http://dx.doi.org/10.1016/j.pec.2012.11.001.

9

(2013)

105-12.

33. [26] T.B. Hong, E.Z. Oddone, T.K. Dudley, H.B. Bosworth, Medication barriers and anti-

hypertensive medication adherence: the moderating role of locus of control, Psychol.

U

Health Med. 11 (2006) 20-8. http://dx.doi.org/10.1080/14786430500228580.

N

34. [27] M.R. Partridge, T. van der Molen, S.-E. Myrseth, W.W. Busse, Attitudes and actions

A

of asthma patients on regular maintenance therapy: The INSPIRE study, BMC Pulm. Med.

M

6 (2006) 1-9. http://dx.doi.org/10.1186/1471-2466-6-13. 35. [28] M.J. Mäkelä, V. Backer, M. Hedegaard, K. Larsson, Adherence to inhaled therapies,

D

health outcomes and costs in patients with asthma and COPD, Respir. Med. 107 (2013)

TE

1481-90. http://dx.doi.org/10.1016/j.rmed.2013.04.005. 36. [29] Australia Centre for Asthma Monitoring., Asthma in Australia 2011: with a focus

EP

chapter on chronic obstructive pulmonary disease, AIHW, Canberra, 2011.

CC

37. [30] S. Suissa, P. Ernst, S. Benayoun, M. Baltzan, B. Cai, Low-Dose Inhaled Corticosteroids and the Prevention of Death from Asthma, N. Engl. J. Med. 343 (2000)

A

332-6. http://dx.doi.org/doi:10.1056/NEJM200008033430504.

38. [31] M. Engelkes, H.M. Janssens, J.C. de Jongste, M.C. Sturkenboom, K.M. Verhamme, Medication adherence and the risk of severe asthma exacerbations: a systematic review, Eur. Respir. J. 45 (2015) 396-407. http://dx.doi.org/10.1183/09031936.00075614.

24

25 39. [32] L. Lasmar, P. Camargos, N.S. Champs, M.T. Fonseca, M.J. Fontes, C. Ibiapina, C. Alvim, J.A. Moura, Adherence rate to inhaled corticosteroids and their impact on asthma control, Allergy 64 (2009) 784-9. http://dx.doi.org/10.1111/j.1398-9995.2008.01877.x. 40. [33] L.K. Williams, E.L. Peterson, K. Wells, B.K. Ahmedani, R. Kumar, E.G. Burchard,

SC RI PT

V.K. Chowdhry, D. Favro, D.E. Lanfear, M. Pladevall, Quantifying the proportion of severe asthma exacerbations attributable to inhaled corticosteroid nonadherence, J. Allergy Clin. Immunol. 128 (2011) 1185-91. http://dx.doi.org/10.1016/j.jaci.2011.09.011.

41. [34] A.C. Murphy, A. Proeschal, C.E. Brightling, A.J. Wardlaw, I. Pavord, P. Bradding,

R.H. Green, The relationship between clinical outcomes and medication adherence in

U

difficult-to-control asthma, Thorax 67 (2012) 751-3. http://dx.doi.org/10.1136/thoraxjnl-

N

2011-201096.

A

42. [35] L.K. Williams, M. Pladevall, H. Xi, E.L. Peterson, C. Joseph, J.E. Lafata, D.R. Ownby,

M

C.C. Johnson, Relationship between adherence to inhaled corticosteroids and poor outcomes among adults with asthma, J. Allergy Clin. Immunol. 114 (2004) 1288-93.

D

43. http://dx.doi.org/10.1016/j.jaci.2004.09.028.

TE

44. [36] W.F. Gellad, C.T. Thorpe, J.F. Steiner, C.I. Voils, The myths of medication adherence, Pharmacoepidemiol. Drug Saf . 26 (2017) 1437-1441. http://dx.doi.org/10.1002/pds.4334.

EP

45. [37] R. Horne, M. Hankins, The Medication Adherence Report Scale (MARS), University

CC

of Brighton, Brighton, 2004. 46. [38] T.T. Menckeberg, M.L. Bouvy, M. Bracke, A.A. Kaptein, H.G. Leufkens, J.A.

A

Raaijmakers, R. Horne, Beliefs about medicines predict refill adherence to inhaled corticosteroids, J. Psychosom. Res. 64 (2008) 47-54.

47. [39] L.A. Phillips, H. Leventhal, E.A. Leventhal, Assessing theoretical predictors of longterm medication adherence: Patients’ treatment-related beliefs, experiential feedback and

25

26 habit

development,

Psychol.

Health

28

(2013)

1135-51.

http://dx.doi.org/10.1080/08870446.2013.793798. 48. [40] R. Ohm, L.S. Aaronson, Symptom perception and adherence to asthma controller medications, J. Nurs. Scholarsh. 38 (2006) 292-7. http://dx.doi.org/10.1111/j.1547-

SC RI PT

5069.2006.00116.x.

49. [41] M. Axelsson, The function of medication beliefs as mediators between personality traits and adherence behavior in people with asthma, Patient Prefer. Adherence 7 (2013) 1101-9. http://dx.doi.org/10.2147/PPA.S49725.

50. [42] M. Emilsson, I. Berndtsson, J. Lötvall, E. Millqvist, J. Lundgren, Å. Johansson, E.

treatment,

Prim.

Care

J.

20

(2011)

141-7.

A

http://dx.doi.org/10.4104/pcrj.2011.00005.

Respir.

N

asthma

U

Brink, The influence of personality traits and beliefs about medicines on adherence to

M

51. [43] H. Jonsdottir, S. Opjordsmoen, A.B. Birkenaes, J.A. Engh, P.A. Ringen, A. Vaskinn, T.O. Aamo, S. Friis, O.A. Andreassen, Medication adherence in outpatients with severe

D

mental disorders: relation between self-reports and serum level, J. Clin. Psychopharmacol.

TE

30 (2010) 169-75. http://dx.doi.org/10.1097/JCP.0b013e3181d2191e. 52. [44] R core team., R: A Language and Environment for Statistical Computing, in: R

EP

Foundation for Statistical Computing. (Ed.) Vienna, Austria, 2014.

CC

53. [45] H. Akaike, A new look at the statistical model identification, IEEE Trans. Autom. Control 19 (1974) 716-23. http://dx.doi.org/10.1109/TAC.1974.1100705.

A

54. [46] L.S. Aiken, S.G. West, Multiple regression: Testing and interpreting interactions, Sage Publications, Newbury Park, CA, 1991.

55. [47] P.A. Frazier, A.P. Tix, K.E. Barron, Testing Moderator and Mediator Effects in Counseling

Psychology

Research,

J.

Couns.

http://dx.doi.org/10.1037/0022-0167.51.1.115.

26

Psychol.

51

(2004)

115-34.

27 56. [48] S.G. West, L.S. Aiken, J.L. Krull, Experimental personality designs: Analyzing categorical by continuous variable interactions, J. Pers. 64 (1996) 1-48. 57. [49] J. Cohen, Applied multiple regression/correlation analysis for the behavioral sciences, L. Erlbaum Associates, Mahwah, N.J, 2003.

SC RI PT

58. [50] S. Weisberg, Applied Linear Regression, Wiley, Hoboken, 2014.

59. [51] H. Foot, A. La Caze, N. Cottrell, Identifying the relationship between beliefs and medication

adherence

in

asthma,

Ann.

Allergy

http://dx.doi.org/10.1016/j.anai.2017.06.012.

Asthma

Immunol.

(2017).

60. [52] A. Aflakseir, Role of illness and medication perceptions on adherence to medication

U

in a group of Iranian patients with type 2 diabetes, J. Diabetes 4 (2012) 243-7.

N

http://dx.doi.org/10.1111/j.1753-0407.2012.00183.x.

A

61. [53] J.R. Rajpura, R. Nayak, Role of illness perceptions and medication beliefs on

M

medication compliance of elderly hypertensive cohorts, J. Pharm. Pract. 27 (2014) 19-24. http://dx.doi.org/10.1177/0897190013493806.

D

62. [54] C. Morgan, J. McBeth, L. Cordingley, K. Watson, K.L. Hyrich, D.P. Symmons, I.N.

TE

Bruce, The influence of behavioural and psychological factors on medication adherence over time in rheumatoid arthritis patients: a study in the biologics era, Rheumatology

EP

(Oxford) 54 (2015) 1780-91. http://dx.doi.org/10.1093/rheumatology/kev105.

CC

63. [55] S.A. AlGhurair, C.A. Hughes, S.H. Simpson, L.M. Guirguis, A systematic review of patient self-reported barriers of adherence to antihypertensive medications using the world

A

health organization multidimensional adherence model, J. Clin. Hypertens. 14 (2012) 87786. http://dx.doi.org/10.1111/j.1751-7176.2012.00699.x.

64. [56] K. Wells, M. Pladevall, E.L. Peterson, J. Campbell, M. Wang, D.E. Lanfear, L.K. Williams, Race-Ethnic Differences in Factors Associated with Inhaled Steroid Adherence

27

28 among Adults with Asthma, Am. J. Respir. Crit. Care Med. 178 (2008) 1194-1201. http://dx.doi.org/10.1164/rccm.200808-1233OC. 65. [57] K.B. Zolnierek, M.R. Dimatteo, Physician communication and patient adherence to treatment: a meta-analysis, Med. Care 47 (2009) 826-34.

a strong predictor, Aust. Fam. Physician 41 (2012) 815-8.

SC RI PT

66. [58] H. Craig, B. Wright, Nonadherence to prophylactic - negative attitudes toward doctors

67. [59] J. Clatworthy, R. Bowskill, T. Rank, R. Parham, R. Horne, Adherence to medication

in bipolar disorder: a qualitative study exploring the role of patients’ beliefs about the condition

and

its

treatment,

Bipolar

Disord.

(2007)

656-64.

U

http://dx.doi.org/10.1111/j.1399-5618.2007.00434.x.

9

N

68. [60] E.J. Mills, J.B. Nachega, D.R. Bangsberg, S. Singh, B. Rachlis, P. Wu, K. Wilson, I.

A

Buchan, C.J. Gill, C. Cooper, Adherence to HAART: A Systematic Review of Developed

M

and Developing Nation Patient-Reported Barriers and Facilitators, PLoS Med. 3 (2006) e438. http://dx.doi.org/10.1371/journal.pmed.0030438.

D

69. [61] K.A. Wallston, B. Strudler Wallston, R. DeVellis, Development of the

TE

Multidimensional Health Locus of Control (MHLC) Scales, Health Educ. Behav. 6 (1978) 160-70. http://dx.doi.org/10.1177/109019817800600107.

EP

70. [62] D.C. Jessop, D.R. Rutter, Adherence to asthma medication: The role of illness Psychol.

Health

18

(2003)

595-612.

CC

representations,

http://dx.doi.org/10.1080/0887044031000097009.

A

71. [63] U. Zugelj, M. Zupancic, L. Komidar, R. Kenda, N.M. Varda, A. Gregoric, Selfreported adherence behavior in adolescent hypertensive patients: the role of illness representations

and

personality,

J.

http://dx.doi.org/10.1093/jpepsy/jsq027.

28

Pediatr.

Psychol.

35

(2010)

1049-60.

29 72. [64] V.L. Morrison, E.A.F. Holmes, S. Parveen, C.O. Plumpton, W. Clyne, S. De Geest, F. Dobbels, B. Vrijens, P. Kardas, D.A. Hughes, Predictors of Self-Reported Adherence to Antihypertensive Medicines: A Multinational, Cross-Sectional Survey, Value Health 18 206-16. http://dx.doi.org/10.1016/j.jval.2014.12.013.

SC RI PT

73. [65] S. Ross, A. Walker, M.J. MacLeod, Patient compliance in hypertension: role of illness perceptions and treatment beliefs, J. Hum. Hypertens. 18 (2004) 607-613.

74. [66] S.H. Lo, J.P. Chau, J. Woo, D.R. Thompson, K.C. Choi, Adherence to

Antihypertensive Medication in Older Adults With Hypertension, J. Cardiovasc. Nurs. 31 (2016) 296-303. http://dx.doi.org/10.1097/JCN.0000000000000251.

U

75. [67] S.-L. Chen, J.-C. Tsai, W.-L. Lee, The impact of illness perception on adherence to

N

therapeutic regimens of patients with hypertension in Taiwan, J. Clin. Nurs. 18 (2009)

A

2234-44. http://dx.doi.org/10.1111/j.1365-2702.2008.02706.x.

M

76. [68] S.M. Mosleh, M.M. Almalik, Illness perception and adherence to healthy behaviour in Jordanian coronary heart disease patients, Eur. J. Cardiovasc. Nurs.

(2014) 1-8.

D

http://dx.doi.org/10.1177/1474515114563885.

TE

77. [69] A. Searle, P. Norman, R. Thompson, K. Vedhara, A prospective examination of illness beliefs and coping in patients with type 2 diabetes, Br. J. Health Psychol. 12 (2007) 621-

EP

38. http://dx.doi.org/10.1348/135910706X164935.

CC

78. [70] K. Griva, L.B. Myers, S. Newman, Illness perceptions and self efficacy beliefs in adolescents and young adults with insulin dependent diabetes mellitus, Psychol. Health 15

A

(2000) 733-50. http://dx.doi.org/10.1080/08870440008405578.

79. [71] S.T. Ashur, S.A. Shah, S. Bosseri, D.E. Morisky, K. Shamsuddin, Illness perceptions of Libyans with T2DM and their influence on medication adherence: a study in a diabetes center

in

Tripoli,

Libyan

http://dx.doi.org/10.3402/ljm.v10.29797.

29

J.

Med.

10

(2015)

29797.

30 80. [72] M.R. DiMatteo, K.B. Haskard, S.L. Williams, Health beliefs, disease severity, and

A

CC

EP

TE

D

M

A

N

U

SC RI PT

patient adherence: A meta-analysis, Med. Care 45 (2007) 521-8.

30

31

81. Legends Figure 1: Effect plot for concerns (BMQ) score on MARS score being moderated by chance (MHLCS) subscale. Each line on the graph represents a different standard deviation away from the mean score (mean [±SD] = 16.8 [±5.2]) for the chance (MHLCS) subscale.

A

CC

EP

TE

D

M

A

N

U

SC RI PT

Figure 2: Effect plot for treatment control (B-IPQ) score on MARS score being moderating by understanding (B-IPQ) subscale. Each line on the graph represents a different standard deviation away from the mean score (mean [±SD] = 7.2 [±2.4]) on the understanding (B-IPQ) subscale.

31