Beliefs about medications when treating hypertension in primary health care: results from “PERson-centredness in hypertension management using information Technology (PERHIT)”

Abstract

Purpose

Hypertension is a major global health concern. Despite of efficient antihypertensive medications a low percentage of patients reach a blood pressure (BP) of <140/90. Nonadherence is a great concern in hypertension treatment and patients’ beliefs about medications has been shown to have a strong impact on adherence. The objective of this study is to examine beliefs about medications and its impact on BP treatment in a group of Swedish primary healthcare patients treated for hypertension with or without an E-health platform.

Materials and method

In a randomised unblinded controlled trial, 949 patients with hypertension from Swedish primary health care centres were included. The intervention group used a web-based system to support self-management of hypertension for eight weeks. Beliefs about medication questionnaire (BMQ) were administered to all patients at inclusion, 8-week follow up and 1-year follow up.

Results

Data were collected from the 862 patients who completed the trial. No statistically significant difference was found in BMQ-scores between the intervention and the control group. An association between lower scores in the BMQ subsection ‘General-Harm’ and achieving target BP of <140/90 mmHg were noted (p = 0.021).

Conclusion

This study shows a significant association between beliefs about medication and BP levels, on hypertensive patients in the Swedish primary care setting, in only one out of four subsections of the BMQ. The intervention did not have a significant effect on changing patients’ beliefs about medication. Further emphasis on patients’ beliefs about medications could be useful in the clinical setting.

Plain language summary

• What is the context? Insufficient treatment effect of high blood pressure is a major global health concern, even though there are several different effective medications. Patients not taking their medications, as they have been prescribed, is a well-known contributing factor. There are associations between underlying beliefs about medications and how strict patients adhere to their prescriptions.

• What is new? In this study data was collected from 862 patients with high blood pressure. The participants were randomised into two groups, one group got treatment as usual and the other group used a web-based interactive information technology system for 8 weeks, in addition to their medications. All participants answered questionnaires about their beliefs about medications. It was shown that the beliefs about medications had limited significant associations to blood pressure levels. Furthermore, the intervention seemed to have no effect upon patients’ beliefs about medications.

• What is the impact? This study provides further evidence that patients’ beliefs about medications might be a possible factor to take into consideration when aiming to treat high blood pressure. The intervention used in this study had no impact on patients’ beliefs about medications.

Keywords:

• Blood pressure
• BMQ
• E-health
• hypertension
• information technology
• primary healthcare

Introduction

The largest preventable risk factor for cardiovascular disease and all-cause mortality worldwide is hypertension [1].

Even though there is a multitude of pharmaceutical treatments available for hypertension there is still a high percentage of patients that does not reach a well-controlled blood pressure (BP). In a study with over 40 000 Swedish primary healthcare patients with hypertension it was reported that no more than 37% of patients reach a target BP lower than 140/90 mmHg [2].

Nonadherence to treatment is identified by The World Health Organisation (WHO) as one of the most serious problems in treatment of chronic diseases [3]. Nonadherence is of great concern in the treatment of hypertension; almost 50% of patients prescribed an antihypertensive drug discontinues the treatment within a year [4]. In recent years awareness of adherence as a treatment factor has increased but nonadherence continues to be a major concern [5].

In order to increase adherence and improve hypertension treatment several new treatment plans have emerged under the banner of e-health. A meta-analysis of the effectiveness of self-management of hypertension using mobile health interventions showed the effectiveness of e-health solutions and correlates them to significant BP reduction, 16 out of the 24 studies included reported improved adherence to medications [6].

Furthermore, significant associations have been described between nonadherence to prescribed medications and beliefs about medications, measured with the beliefs about medications questionnaire (BMQ), in patients with various diseases, including hypertension [7–11]. Beliefs about medications affects adherence to treatment, adherence to treatment affects treatment effect.

It therefore stands to reason that a greater understanding of patients’ beliefs about medication, and the possible impact of e-health solutions upon these beliefs, could lead to a better chance to improve adherence to medications, which in turn could lead to improved treatment results of hypertension, a major global health concern.

The present study is part of the PERson-centredness in Hypertension management using Information Technology (PERHIT)-project, a randomised controlled trial designed to investigate the effect of an e-health platform and a person-centered approach on BP treatment. A significant increase in the proportion of participants reaching target BP of <140/90 mmHg compared to the control group was shown [12].

Objectives

The objective of this study is to examine the beliefs about medications, with the BMQ, and its impact on the ability to reach target BP in primary healthcare patients with hypertension treated with or without a self-report information system.

Materials and methods

The PERHIT study was a multi-centre unblinded randomised controlled trial, with a study length of twelve months. The primary outcome of the study was the number of patients who reached target BP after eight weeks and 12 months; the results from the intervention group was compared to the results in a control group. Patients were included from primary healthcare centres (PHCC), both public and private, located in four different counties (Skåne, Västra Götaland, Östergötland and Jönköping) in Sweden. The study design of PERHIT has previously been described in other publications [12,13].

Inclusion criteria were as follows

• Age 18 or older.

• Diagnosed with hypertension.

• Prescribed one or more antihypertensive medication.

• Understanding of Swedish to be able to make use of the interactive web-based self-management support system using the mobile phone and to provide informed consent.

Exclusion criteria were as follows

• Secondary hypertension, according to medical records.

• Pregnancy induced hypertension.

• Cognitive impairment.

• Terminal illness.

• Psychotic disease.

• Vision impairment (not able to read messages on the mobile phone).

Eligible patients were invited by their nurse or physician to their PHCC for a baseline assessment and study inclusion. After the assessment, patients were randomised in a 1:1 ratio to intervention group or control group. The randomisation was made via a block randomisation approach.

For a full summary of the data collected at the baseline assessment see the PERHIT study protocol [13]. The data collected of interest for this study was the following:

age, education, gender, smoking (current, former, never), duration of hypertension diagnosis, congestive heart failure, diabetes, marital status, BP (systolic and diastolic) and BMI. Several questionnaires were used in PERHIT, Beliefs about Medicines Questionnaire (BMQ) being the one of interest for this study [14].

The patients randomised into the intervention group were instructed on how to use the system and then asked to self-report once daily in the evening, via their own mobile phone for eight consecutive weeks. On each occasion, the patients first answered the questions and then measured their BP (mean of three readings) and pulse. The recordings were made within a window for answers 5 p.m.–12 p.m. in the self-report system.

The system used in the study includes four components

• A module for self-reporting: symptoms, well-being, lifestyle, medication intake, and side effects of medication.

• A validated BP monitor for daily home BP and pulse measurements.

• Weekly motivational messages to encourage lifestyle changes, personalised and tailored to the patient.

• A web-based dashboard to enable physicians and nurses as well as the patients themselves to examine graphs for visualisations of the patients BP in relation to the self-reports, including medication intake and physical activities.

No intervention was made in the control group, they received care as usual. Both groups had three study visits at their PHCC (baseline, after 8 weeks and after 1 year) where BP was measured and a set of questionnaires was answered. Study data were collected and managed using REDCap electronic case report forms (eCRF) hosted at Clinical Studies Sweden – Forum South, Region Skåne [15]. The surveys during the study, including the BMQ, was administered at the visits to the primary healthcare centres via the survey function in REDCap. Other visits to the PHCC were not regulated in the study, for either group.

Beliefs about medications questionnaire

The ‘Beliefs about Medication Questionnaire’ (BMQ) showcases patients’ attitudes towards medications (Appendix 1). It consists of two parts; the BMQ-general and the BMQ-specific, and each part is divided into subsections [14]. The BMQ-general focuses on attitudes towards all medications in general and has two subsections (‘Harm’ and ‘Overuse’) with four questions in each subsection. The BMQ-specific has two subsections (‘Necessity’ and ‘Concern’) with five questions each and focuses on attitudes towards the medications the patient is prescribed. It is a validated [16] and well used questionnaire. However not all studies use all the subsections, and the number of questions included as well as the language may be altered [16–19]. This study used the subsections BMQ-general ‘Harm’ and ‘Overuse’ and BMQ-specific ‘Concern’ and ‘Necessity’. Furthermore this study includes the necessity concern differential (‘NCD’) which aims to quantify the interaction between the perceived necessity of medicines (which can lead to improved adherence) and concern about the therapy (which can lead to poor adherence) [20–22]. A high NCD-score would mean that the patient considers the need for their medication to override the concern for its potential harm. The ‘NCD’ is a stronger predictor of adherence then factors like demographics and type of illness [22]. For patients missing more than two answers in a subsection the corresponding subscale was excluded, otherwise a mean value of the subscale was calculated from the retrieved answers and assigned to the missing values.

Data analysis

Statistical analyses were made with IBM SPSS Statistics for Windows, Version 29.0. Released 2022. Armonk, NY: IBM Corp.

The answers to the BMQ were graded on a 5-point Likert scale (from 1 strongly disagree to 5 strongly agree). Cronbach’s alfa values and mean inter-item correlations was calculated to secure homogeneity in the instrument and are presented in Appendix 1. Changes in mean values of BMQ scores, for all the subsections, at both the 8-week and the 1-year follow-up were calculated and compared between the intervention and the control group using Student’s t-test. Nominal data was analysed with Pearson’s Chi-2 test and numerical data with unpaired Student’s t-test. The data in the large study population was considered to satisfy the condition of normality with the central limit theorem. P-values are presented with 95% CI.

Ethical considerations

The PERHIT study has been approved by the Regional Ethical Review Board in Lund, Dnr 2017/311 (17 May 2017) and Dnr 2019/00036 (11 January 2019).

The study was registered at ClinicalTrials.gov [NCT03554382].

Results

Patient characteristics are described for all patients, as well as for the intervention group and the control group respectively, in Table 1.

Table 1. Baseline data – descriptive intervention/control.

	All patients n = 862	Intervention n = 442	Control n = 420	Missing n (%)
Age in years	62.7 ± 9.7	62.6 ± 9.6	62.8 ± 9.8	0 (0)
Gender: Female (%)	364 (42.2)	179 (40.5)	185 (44.0)	0 (0)
Education				11 (1.3)
Up to high school	199 (23.1)	92 (20.8)	107 (25.5)
High school	394 (45.7)	202 (45.7)	192 (45.7)
University	258 (29.9)	142 (32.1)	116 (27.6)
Body mass index	28.9 ± 4.6	28.8 ± 4.4	29.0 ± 4.7	0 (0)
Years with hypertension	10.1 ± 9.7	9.6 ± 8.9	10.7 ± 10.4	75 (8.7)
Marital status				10 (1.2)
Unmarried or widowed	146 (16.9)	70 (15.8)	76 (18.1)
Married or cohabiting	691 (80.2)	356 (80.5)	335 (79.8)
Other	15 (1.7)	10 (2.3)	5 (1.2)
Diabetes	104 (12.1)	56 (12.7)	48 (11.4)	0 (0)
Congestive heart failure	16 (1.9)	9 (2.0)	7 (1.7)	1 (0.1)
Smoking	44 (5.1)	22 (5.0)	22 (5.2)	14 (1.6)
Patients with target BP* at inclusion	298 (34.6)	154 (34.8)	144 (34.3)	0 (0)
Blood pressure levels at inclusion
Systolic	144.8 ± 16.4	144.2 ± 16.2	145.5 ± 16.6	0 (0)
Diastolic	85.3 ± 9.9	84.9 ± 9.0	85.7 ± 10.7	0 (0)

*Target BP – Target blood pressure <140/90.

The test results for the entire study population from the BMQ administered at inclusion, the 8-week follow-up and the 1-year follow-up is presented in Table 2, with comparisons between the group that reached target BP and those that did not, regardless of if they were in the intervention group or control group. The values in the subsection ‘Harm’ in the ‘General’ part of the questionnaire, at the 8-week follow-up, was significantly lower in the group that reached target BP then the group that did not (p-value 0.021). Although the results from the other subsections did not show any statistical significance the descriptive statistics indicated higher values in the positively aligned subsections i.e. ‘Necessity’ and ‘NCD’ and lower values in the negatively aligned subsections i.e. ‘Concern’ and ‘Overuse’ in the group that reached target BP as compared to the group that did not reach target BP. The same trend is visible in Table 3, demonstrating the BMQ results from the 8-week and the 1-year follow-up for the intervention and the control group respectively. Separating the intervention and the control group, there was no statistically significant difference in any of the subsections between the group that reached target BP and the group that did not.

Table 2. ‘Belief about medication questionnaire’ – BP < 140/90 vs BP ≥ 140/90, entire study population.

	Inclusion			8-week follow-up			1-year follow-up
	BP < 140/90 n = 298	BP ≥ 140/90 n = 564	p-value (95% CI)	BP < 140/90 n = 384	BP ≥ 140/90 n = 478	p-value (95% CI)	BP < 140/90 n = 380	BP ≥ 140/90 n = 482	p-value (95% CI)
BMQ-Specific
Necessity	17.8 ± 4.1 (1)	17.5 ± 3.6 (3)	0.29 (–0.85–0.25)	18.2 ± 3.7 (4)	17.9 ± 3.7 (5)	0.38 (–0.73–0.28)	18.0 ± 3.7 (6)	18.3 ± 3.3 (10)	0.23 (–0.18–0.78)
Concern	9.6 ± 3.6 (2)	9.9 ± 3.9 (3)	0.17 (–0.16–0.90)	9.6 ± 3.9 (4)	10.0 ± 3.9 (5)	0.18 (–0.17–0.89)	9.5 ± 3.8 (6)	9.8 ± 3.9 (10)	0.30 (–0.25–0.80)
NCD	8.3 ± 5.3 (2)	7.6 ± 5.1 (3)	0.08 (–1.40–0.07)	8.6 ± 5.1 (4)	8.0 ± 5.2 (5)	0.10 (–1.29–0.11)	8.6 ± 5.2 (6)	8.6 ± 5.2 (10)	0.95 (–0.68–0.73)
BMQ-General
Harm	8.4 ± 2.5 (1)	8.4 ± 2.6 (3)	0.96 (–0.37–0.35)	8.1 ± 2.6 (4)	8.5 ± 2.7 (7)	0.02 (0.06–0.78)	8.1 ± 2.7 (5)	8.1 ± 2.6 (10)	0.91 (–0.33–0.38)
Overuse	10.6 ± 2.5 (1)	10.5 ± 2.6 (3)	0.75 (–0.41–0.30)	10.4 ± 2.6 (4)	10.6 ± 2.6 (8)	0.35 (–0.19–0.52)	10.2 ± 2.8 (6)	10.1 ± 2.6 (10)	0.83 (–0.41–0.33)

BP: Blood pressure; NCD: Necessity concern differential; (x): number of missing.

Table 3. ‘Belief about medication questionnaire’ – 8-week and 1-year follow-up, BP < 140/90 vs BP ≥ 140/90 in the intervention group and control group respectively.

	8-week follow-up						1-year follow-up
	Intervention group			Control group			Intervention group			Control group
	BP < 140/90 n = 214	BP ≥ 140/90 n = 228	p-value (95% CI)	BP < 140/90 n = 170	BP ≥ 140/90 n = 250	p-value (95% CI)	BP < 140/90 n = 208	BP ≥ 140/90 n = 234	p-value (95% CI)	BP < 140/90 n = 172	BP ≥ 140/90 n = 248	p-value (95% CI)
BMQ-Specific
Necessity	18.2 ± 3.7 (2)	18.1 ± 3.6 (0)	0.85 (–0.75–0.62)	18.2 ± 3.8 (2)	17.8 ± 3.9 (5)	0.33 (–1.13–0.37)	18.1 ± 3.5 (4)	18.3 ± 3.3 (4)	0.45 (–0.40–0.90)	18.0 ± 3.9 (2)	18.3 ± 3.4 (6)	0.34 (–0.37–1.08)
Concern	9.8 ± 3.9 (2)	10.2 ± 4.0 (0)	0.28 (–0.33–1.15)	9.4 ± 3.9 (2)	9.8 ± 3.9 (5)	0.33 (–0.39–1.15)	9.7 ± 3.8 (4)	9.7 ± 4.0 (4)	0.98 (–0.73–0.75)	9.2 ± 3.8 (2)	9.8 ± 3.7 (6)	0.12 (–0.15–1.33)
NCD	8.4 ± 5.3 (2)	7.9 ± 5.3 (0)	0.35 (–1.47–0.53)	8.8 ± 4.8 (2)	8.0 ± 5.1 (5)	0.13 (–1.73–0.22)	8.3 ± 5.3 (4)	8.6 ± 5.2 (4)	0.64 (–0.76–1.24)	8.8 ± 5.0 (2)	8.6 ± 5.1 (6)	0.64 (–1.23–0.76)
BMQ-General
Harm	8.1 ± 2.7 (2)	8.5 ± 2.7 (1)	0.09 (–0.07–0.94)	8.1 ± 2.5 (2)	8.5 ± 2.6 (6)	0.12 (–0.10–0.91)	8.1 ± 2.6 (3)	8.1 ± 2.5 (3)	0.79 (–0.55–0.42)	8.1 ± 2.8 (2)	8.2 ± 2.6 (7)	0.70 (–0.43–0.63)
Overuse	10.3 ± 2.6 (2)	10.5 ± 2.6 (2)	0.29 (–0.22–0.75)	10.6 ± 2.6 (2)	10.6 ± 2.6 (6)	0.89 (–0.48–0.56)	10.1 ± 2.7 (3)	10.3 ± 2.8 (3)	0.51 (–0.35–0.69)	10.3 ± 2.8 (2)	10.0 ± 2.5 (7)	0.33 (–0.79–0.27)

BP: Blood pressure; NCD: Necessity concern differential.

Table 4 shows that the entire study population changed their beliefs about medication during the year the trial lasted. The mean value in all subcategories for the entire study population, as well as the intervention group and control group individually, changed equal to or exceeding 0.2, with ‘Necessity’ and ‘NCD’ increasing and ‘Concern’, ‘Harm’ and ‘Overuse’ decreasing. When comparing the changes in mean values, for all the subsections, between the intervention and the control group at both the 8-week and the 1-year follow-up no statistically significant difference was noticeable.

Table 4. ‘Belief about medication questionnaire’ – change in BMQ-score over time.

	At Inclusion^a	At 8-week check-up^a	At 1 year check-up^a
BMQ-specific
Necessity
Total population	17.6 ± 3.8 (4)	18.0 ± 3.7 (9)	18.2 ± 3.5 (16)
Intervention	17.6 ± 3.7 (3)	18.2 ± 3.6 (2)	18.2 ± 3.4 (8)
Control	17.7 ± 3.9 (1)	17.9 ± 3.8 (7)	18.2 ± 3.6 (8)
Concern
Total population	9.8 ± 3.8 (5)	9.8 ± 3.9 (9)	9.6 ± 3.8 (16)
Intervention	9.9 ± 3.8 (4)	10.0 ± 4.0 (2)	9.7 ± 3.9 (8)
Control	9.7 ± 3.9 (1)	9.6 ± 3.9 (7)	9.5 ± 3.7 (8)
NCD
Total population	7.8 ± 5.2 (5)	8.2 ± 5.2 (9)	8.6 ± 5.2 (16)
Intervention	7.7 ± 5.1 (4)	8.1 ± 5.3 (2)	8.5 ± 5.3 (8)
Control	8.0 ± 5.2 (1)	8.3 ± 5.0 (7)	8.7 ± 5.1 (8)
BMQ-General
Harm
Total population	8.4 ± 2.6 (4)	8.3 ± 2.6 (11)	8.1 ± 2.6 (15)
Intervention	8.4 ± 2.6 (3)	8.3 ± 2.7 (3)	8.1 ± 2.5 (6)
Control	8.4 ± 2.6 (1)	8.4 ± 2.6 (8)	8.2 ± 2.7 (9)
Overuse
Total population	10.5 ± 2.6 (4)	10.5 ± 2.6 (12)	10.2 ± 2.7 (16)
Intervention	10.6 ± 2.5 (3)	10.4 ± 2.6 (4)	10.2 ± 2.7 (7)
Control	10.5 ± 2.6 (1)	10.6 ± 2.6 (8)	10.1 ± 2.6 (9)

(x): number of missing; NCD: Necessity concern differential.

^aTotal population n = 862, Intervention n = 442, Control n = 420.

Discussion

In this study with Swedish primary healthcare patients diagnosed with hypertension, beliefs about medications – as calculated by the validated questionnaire BMQ – was significantly associated with lower BP in the subsection ‘General harm’. Though there were no statistically significant differences in other subsections, a trend was noticeable in all the subsections.

The strengths of this study are its size and its study population. It was a randomised, large study with primary healthcare patients from several different counties and with different socioeconomic status. As such it’s a study population that accurately reflects the real-life setting of hypertension care.

The major limitation of this study is that BMQ-scores were not the primary outcome for which the PERHIT study was designed. The primary outcome was to examine the interventions effect on BP levels. Selection bias is another potential limitation in this study as it seems more likely that patients with a lower belief in medication might have a lower belief in healthcare in general and could be less likely to agree to participate in studies. This limitation would be very hard to fully circumvent. However, it is possible that a study design with beliefs about medications as primary focus could reduce the impact of this limitation. This assumption is supported by the fact that the mean BMQ-scores at inclusion in this study was substantially lower in the negatively aligned subsections and somewhat higher in the positively aligned subsections than the scores noted in other similar studies with beliefs about medication and adherence as primary focus [16,23].

The PERHIT-project aims to increase the understanding of, and efficiency in, treatment of hypertension, a diagnosis with major global implications. Specifically, to evaluate the effect of adding a mobile platform (self-management support system) to care as usual.

This project aimed to evaluate the effect of patient beliefs of medications upon the success rate on reaching target BP in patients with hypertension with or without an information technology system.

In a study by M. Sjölander et al. [23], a change in BMQ score of a subsection by 0.2 points is considered to be clinically significant. In this study, during its yearlong study period, both the intervention group and the control group had changes in their BMQ-scores that were equal to or exceeded 0.2 in all subcategories. The positively aligned subsectors i.e. ‘Necessity’ and ‘NCD’ increased and the negatively aligned subsectors i.e. ‘Concern’, ‘Harm’ and ‘Overuse’ decreased. A trend was noticeable where the change in mean values in the intervention group was somewhat greater than the changes in the control group in ‘Necessity’, ‘NCD’ and ‘Harm’ but equal in ‘Concern’ and ‘Overuse’ at the one-year follow-up, however, this trend was not statistically significant.

A significant increase in the proportion of participants reaching target BP compared to the control group in the PERHIT-study was shown in a previously published article [7]. It was argued that this change might be, in part, due to increased adherence to treatment. As there is a known link between adherence to medications and beliefs about medications it was hypothesised that increased adherence might be due to a change in beliefs about medications gained by increasing the patient’s understanding about hypertension, and its correlation to medications and lifestyle habits. It was hypothesised that the intervention might have an impact on patients’ beliefs about medications. This study cannot support that hypothesis, since there was no statistically significant difference between the changes in BMQ-scores between the intervention group and the control group. Furthermore, this study found the association between BP lower than 140/90 mmHg and beliefs about medications to be significant in only one subsection. In this aspect our findings differ from those presented in a multitude of other studies [8,24–26], where BMQ-scores are shown to have a stronger correlation, in multiple subsections, to adherence to treatment and therefore in extension to BP levels. The reason for the discrepancies between the number of statistically significant subsections in this study compared to others are not clear, it’s a large study with heterogeneity in the study population. Possible explanations are the study design being focussed on BP changes rather than BMQ changes, and the comparably high BMQ levels at inclusion in the entire study population.

The hypothesis that the intervention might have an impact on patients’ beliefs in medications is supported by the fact that the BMQ-scores improved in the intervention group but is rejected due to lack of a statistically significant difference. The lack of a difference could however be due to the fact that the BMQ-scores improved in the control group as well. The reason for this is unclear. The extra visits to the healthcare centre because of the study, or simply the fact of participating in a study about hypertension, may have had an impact.

Further research concerning beliefs about medications impact on BP, through its effect on adherence, might be useful to determine the BMQ’s usefulness in the clinical setting. If the BMQ is administered when patients are diagnosed with hypertension and the test indicates a low belief in medications it could be clinically motivated to further explore reasons for this and offer an educational intervention. If so, further research concerning if these educational interventions could effectively be done with the assistance of e-health tools would be beneficial.

Conclusion

This study, conducted in the Swedish primary healthcare, shows a significant association between patients’ beliefs about medication and their ability to reach target BP, in one out of four subsections examined with the BMQ. No significance was found when examining the effect of using an information technology system on patients’ beliefs about medication.

Acknowledgement

The authors would like to thank Patrick O’Reilly for his assistance with English-language editing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author on reasonable request. The data are not publicly available due to ethical restrictions.

Additional information

Funding

This work was supported by the Kamprad Foundation under Grant 20170102, the Heart and Lung foundation under Grant 20170251, and the Swedish Research Council under Grant 2018-02648.

Appendix 1.

‘Beliefs about medication questionnaire’ (BMQ) – instrument description and cronbach’s alfa scores

Table

BMQ subscales	Measures	Number of questions	Score range	Cronbach’s alfa at inclusion (Mean Inter-item Correlations)	Cronbach’s alfa at 8w (Mean Inter-item Correlations)	Cronbach’s alfa at 1y (Mean Inter-item Correlations)
BMQ-Specific
Necessity	The personal need for drugs to maintain or improve own health	5	5–25	0.79 (0.42)	0.80 (0.45)	0.76 (0.39)
Concern	The perceived concern about the negative effects of their own use of drugs	5	5–25	0.79 (0.42)	0.81 (0.47)	0.81 (0.46)
NCD*	Obtained by subtracting the concerns score from the necessity score.		−20 to 20
BMQ-General
Harm	The harmful nature of drugs in general	4	4–20	0.59 (0.27)	0.64 (0.31)	0.66 (0.33)
Overuse	The belief that doctors overuse or put too much trust in drugs	4	4–20	0.58 (0.26)	0.63 (0.29)	0.64 (0.31)

*NCD – Necessity concern differential.