Objective To determine if a targeted and tailored intervention based on a discussion informed by validated adherence scales will improve medication adherence.
Design Prospective randomised trial.
Setting 2 community pharmacies in Brisbane, Australia.
Methods Patients recently initiated on a cardiovascular or oral hypoglycaemic medication within the past 4–12 weeks were recruited from two community pharmacies. Participants identified as non-adherent using the Medication Adherence Questionnaire (MAQ) were randomised into the intervention or control group. The intervention group received a tailored intervention based on a discussion informed by responses to the MAQ, Beliefs about Medicines Questionnaire-Specific and Brief Illness Perception Questionnaire. Adherence was measured using the MAQ at 3 and 6 months following the intervention.
Results A total of 408 patients were assessed for eligibility, from which 152 participants were enrolled into the study. 120 participants were identified as non-adherent using the MAQ and randomised to the ‘intervention’ or ‘control’ group. The mean MAQ score at baseline in the intervention and control were similar (1.58: 95% CI (1.38 to 1.78) and 1.60: 95% CI (1.43 to 1.77), respectively). There was a statistically significant improvement in adherence in the intervention group compared to control at 3 months (mean MAQ score 0.42: 95% CI (0.27 to 0.57) vs 1.58: 95% CI (1.42 to 1.75); p<0.001). The significant improvement in MAQ score in the intervention group compared to control was sustained at 6 months (0.48: 95% CI (0.31 to 0.65) vs 1.48: 95% CI (1.27 to 1.69); p<0.001).
Conclusions An intervention that targeted non-adherent participants and tailored to participant-specific reasons for non-adherence was successful at improving medication adherence.
Trial registration number ACTRN12613000162718; Results.
- medication adherence
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Strengths and limitations of this study
The adherence intervention was targeted by identifying participants who were non-adherent to their medication prior to inclusion in the trial.
The use of validated adherence scales provided insight into a person's adherence and can be used in a similar way to electronic monitoring in a measurement-guided medication management approach to improve adherence.
The study would have been improved by addition of a reliable objective measure of adherence.
This study had a relatively small sample size and was not powered to measure clinical outcomes.
Improving adherence to medication has been identified as one of the most cost-effective and achievable opportunities for improving health outcomes.1 ,2 Many interventions have been implemented to improve adherence to medications, including: reminder systems (text reminders, dose administration aids); behavioural counselling (motivational interviewing); social support (peer support therapy); cognitive-educational interventions (verbal information) and measurement-guided medication management.3 ,4 While many of these interventions have been successful in improving adherence in specific trials, no intervention has conclusively demonstrated effectiveness in improving adherence and clinical outcomes.5–8 The few interventions that have been successful in improving adherence and clinical outcomes in well-conducted randomised trials have been multifaceted, complex interventions that are difficult to replicate in practice.7
There are a number of issues that may account for these results, but perhaps the two most pertinent are: lack of assessment of participants' adherence prior to enrolment,7 ,9–13 and using an intervention that may not specifically address reasons for the participants' non-adherence.6 ,8 ,14 ,15 Most studies introduce an intervention into an unselected population and employ an intervention that may or may not address participant-specific reasons for non-adherence. Targeting non-adherent participants and tailoring interventions to specific reasons for non-adherence has been suggested to improve the effectiveness of medication adherence interventions,1 ,16–18 but few studies to date have adopted this approach.19–21
There are many objective and subjective measures of adherence that can provide information in relation to a patient's medication-taking behaviour albeit with limitations specific to each method.6 ,22–25 Objective measures of adherence include electronic monitoring of medication administration (eg, Medication Event Monitoring System, MEMS), prescription records and dose counts. These measures are often good at measuring medication-taking behaviour, but can be expensive, impractical and do not provide information on reasons for behaviour. Subjective measures of adherence include physician reports, self-report and adherence scales. Subjective measures are prone to recall and social desirability bias, but they are often easy to administer and provide the opportunity to explore why the patient may be non-adherent. Self-report adherence scales are relatively easy to administer and elicit different information: medication-taking behaviour, barriers to adherence and beliefs associated with adherence.26
MEMS has been used in a measurement-guided medication management approach to identify non-adherence and inform discussion between the patient and their health professional about potential barriers to adherence. This approach has been successful in improving adherence in several studies.3 We believe the measurement-guided medication management approach could be adopted using adherence scales that are strategically selected to identify non-adherence and key reasons for non-adherence.
We conducted a randomised trial to determine if a measurement-guided medication management-approach based on a discussion informed by validated adherence scales, would improve adherence to a recently initiated cardiovascular or oral hypoglycaemic medication. We hypothesised that randomising participants assessed to be non-adherent and tailoring an intervention based on a discussion informed by adherence scales would improve adherence at 3 months as measured by the Medication Adherence Questionnaire (MAQ). We also tested whether any improvements in adherence at 3 months would be sustained at 6 months.
This was a randomised controlled trial recruiting participants who recently initiated a medicine for chronic cardiovascular disease or type 2 diabetes. The recruitment of potential participants occurred between the 25 March 2013 and 24 July 2013. Participants were followed for 6 months from recruitment, with the last participant contact occurring on the 10 February 2014. This trial is registered on the Australian New Zealand Clinical Trials Registry, which can be accessed at http://www.anzctr.org.au/ using trial ID ACTRN12613000162718.
Potential participants presenting a prescription for a medicine to manage hypertension, type 2 diabetes, dyslipidaemia or other cardiovascular diseases were identified and recruited by the principal investigator (TN) who is a registered pharmacist. Potential participants were recruited from two community pharmacies in Brisbane, Australia. The two pharmacies were selected on the basis of convenience. The researcher had worked in both of the pharmacies. The pharmacies serviced a broad range of middle working class patients with chronic diseases. These community pharmacies do not provide adherence interventions as a routine service. These pharmacies were approached by the researcher and were provided with information on the study. Once the pharmacies agreed to the study taking place, the dates for participant recruitment were organised. Participants were interviewed in the semiprivate counselling area of the pharmacy.
Individuals who were over 18 years of age and started a new medication for hypertension, type 2 diabetes, dyslipidaemia or other cardiovascular diseases (myocardial infarction, heart failure, hypertension, arrhythmia and stroke) within the past 4–12 weeks were approached to participate in the study. Specific medications included ACE inhibitors, angiotensin II receptor antagonists, calcium channel blockers, lipid-lowering agents or oral hypoglycaemic drugs. If multiple medications were prescribed within the past 4–12 weeks, then the most recently initiated medication was selected. This standardises the sample as all participants would be in the implementation phase27 of taking their medicine, and would have had the opportunity to have some experience with their medicine. Individuals who were unable to complete the survey tool were excluded from the study.
All interviews were conducted by the principal investigator (TN), who is a registered pharmacist. The survey instruments used in the interview included: the MAQ, Beliefs about Medicines Questionnaire-Specific (BMQ-S) and Brief Illness Perceptions Questionnaire (BIPQ). These scales were selected following a systematic review of the literature.26
The MAQ was used to assess adherence behaviour to the recently initiated medicine of interest.28 Participants identified as adherent (score of 0) using the MAQ were enrolled and followed for 6 months. Participants identified as non-adherent (score of 1–4) using the MAQ were randomised into either the intervention or control group, using block randomisation and followed for 6 months. The random allocation sequence was generated by an internet-based randomisation software (Research Randomiser). The block size was 10, providing an allocation ratio of 1:1 (eg, ABBABABAAB). The intervention group received a tailored intervention to improve medication adherence. Owing to the nature of the intervention, neither the researcher nor the participants were blinded to the allocation at the baseline interview. No data analysis occurred prior to completion of the study.
Baseline demographics of the participants were also collected. All participants were followed for 6 months. Participants were asked to complete the same three validated adherence scales (MAQ, BMQ-S and BIPQ) at 3 and 6 months over the telephone. No further interventions were conducted at the 3 and 6-month time points.
Medication Adherence Questionnaire
The four-item MAQ was selected because it has been well-validated to identify adherence behaviour in a number of chronic cardiovascular disease populations and scores have been shown to correlate well with objective adherence measures and clinical outcomes, such as blood pressure, lipid levels and blood glucose control.28–31 The MAQ has also been used to explore reasons for non-adherence.28 Specifically, MAQ has been used to identify unintentional non-adherence, intentional non-adherence or a mix of both.32
Participants were asked to respond to the MAQ in relation to the recently initiated medication of interest. Participants answering ‘no’ to all items of the MAQ (MAQ score=0) were identified as adherent to their medicine.29 ,33–35 These participants were followed for 6 months in the ‘adherent’ group (the results of this participant group will be reported elsewhere). Participants answering ‘yes’ to at least one of the MAQ items (MAQ score=1–4) were identified as ‘non-adherent’ and were randomised to either the intervention or control groups. This cut-off has been used in the literature, and provides a highly sensitive tool for identifying medication non-adherence.29 ,33–35 Responses to the MAQ were also used to identify adherence behaviour and identify the likely type of non-adherence, for instance: unintentional non-adherence due to being forgetful or careless, or intentional non-adherence by ceasing their medicines when they felt better or worse, and a mix of both types.
Beliefs about Medicines Questionnaire—Specific
The BMQ-S elicits an individual's beliefs about their medicines in the domains of necessity of medicines and concerns about medicines. The BMQ-S has been validated in many disease populations.26 In general, individuals who have strong concerns about their medicines or believe their medicines are not necessary tend to be less adherent.36–38
All participants were interviewed using the BMQ-S to measure perceived necessity of and concerns about medicines.36 The BMQ-S consists of 10 statements about medicines: five of the statements are related to beliefs about the necessity of medicines and the remaining five statements are related to concerns that individuals may have about their medicines.
Brief Illness Perception Questionnaire
Illness representations identified in the BIPQ have been closely associated with medication adherence.39 ,40 The BIPQ consists of nine items that assess the cognitive and emotional representations of illness.39 This questionnaire provided insight into a participant's perceptions and understanding of their illness and treatment.
The intervention took place at a single time point, immediately following randomisation and focused only on the recently initiated medication of interest. For participants randomised to the intervention group, the investigator used participant responses to the adherence scales to prompt further discussion regarding the participant's adherence and the factors that supported or impeded them to take their medicine. The intervention used the measurements provided by the validated adherence scales to tailor an adherence support strategy for each participant (it is in this sense that the intervention is a form of measurement-guided medication management). The investigator and participant then selected and implemented a strategy from an ‘evidence-based toolkit’ to support the participant's adherence based on the information discussed in the interview.
Reminder systems (dose administration aids, dosette boxes, alarm clock reminders, text reminders, treatment simplification);
Cognitive-educational interventions (verbal information, written information);
Behavioural-counselling interventions (reinforcing behaviour, empowering individuals to actively participate in their healthcare and problem-solving);
Social support interventions (family member support);
Multifaceted interventions (reminder systems coupled with cognitive-educational interventions).
For example, some participants who stated they forget to take their medicine on the MAQ may be asked: How often they forget? Where they store their medicines? Or why they think they forget to take their medicine? This information helped determine if the participant would benefit from a reminder and the specific type of reminder strategy. If participants indicated they had a poor perceived understanding of their illness or their treatment on the BIPQ, participants may be asked what they knew about their illness and/or medicine to help individualise the education provided in a cognitive-educational strategy to support adherence. If participants had a low necessity score and/or a high concerns scores on the BMQ-S, the specific beliefs the participant held that led to these scores were explored with the participant. These discussions focused on identifying and discussing any non-veridical beliefs held by the participant about their medicine (eg, strong concerns about an adverse effect that is very unlikely or can be mitigated with appropriate monitoring). These participants received individualised education or a behavioural-counselling strategy to support their adherence.
Participant responses to the MAQ, BMQ-S and BIPQ were collected at baseline, 3 and 6 months. The primary outcome was the difference in the mean MAQ score between the intervention and control groups at 3 months. An intention-to-treat analysis was used for the primary outcome. Secondary outcomes included the difference in the mean MAQ score between the intervention and control groups at 6 months. A post hoc analysis was conducted to assess whether changes in survey responses were consistent with the specific adherence intervention employed.
Baseline demographics of the intervention and control groups were compared using t-tests for continuous data and Fisher's exact tests for categorical data.
A one-sided independent-samples t-test was conducted to compare the mean MAQ score of the intervention and control group, based on the intention-to-treat population using R (V.3.0.2) statistical software, at 3 and 6 months.
Changes in the questionnaires scores at 3 and 6 months were also visually observed in different strategy types in the intervention group.
The study was powered to observe a difference in mean MAQ scores between intervention and control of 0.683. This difference in mean MAQ was observed in a trial of an education intervention to improve adherence.41 This improvement in mean MAQ was associated with a clinically significant improvement in blood pressure control. Forty-one participants per group (intervention and control) provided 80% power to detect a statistically significant change in adherence at a level of 0.05. Taking into account anticipated dropouts, our target sample size was 60 participants per group (intervention and control).
A total of 408 individuals were assessed for eligibility, of which 152 participants (112 recruited from first pharmacy and remaining 40 from the second pharmacy) were enrolled into the study (figure 1). 120 participants were identified as non-adherent and randomised 1:1 to intervention or control. At 6 months, there were 55 participants remaining in the intervention group and 45 participants in the control group. The movement of participants throughout the study is shown in figure 1.
Participant baseline demographics
The participants identified as non-adherent using the MAQ had a mean age of 63.5 years (table 1). Of these participants, 66 (55%) were women and 98 (81.7%) had attained secondary school qualifications or higher. There were no significant differences in the demographics between the intervention and control groups. The mean MAQ score at baseline in the intervention and control groups were similar: 1.58: 95% CI (1.38 to 1.78) and 1.60: 95% CI (1.43 to 1.77), respectively.
The mean length of the baseline interview for the intervention group was 13.5±2.9 min (including implementation of strategy) and control group was 11.8±2.8 min.
The tailored strategies that were implemented are shown in table 2. Reminder systems accounted for 45% of the implemented strategies.
The intervention improved adherence as measured by the MAQ at 3 months. Mean MAQ score in the intervention and control group were: 0.42: 95% CI (0.27 to 0.57) vs 1.58: 95% CI (1.42 to 1.75); p<0.001 (lower MAQ scores reflect better adherence to treatment). The lower MAQ score in the intervention group compared to control was sustained at 6 months (0.48: 95% CI (0.31 to 0.65) vs 1.48: 95% CI (1.27 to 1.69); p<0.001). This represents a statistically significant improvement in the primary end point at 3 and also at 6 months (p<0.001; figure 2).
On a more individual level, we identified 53 of the 60 (88.3%) participants in the intervention group as adherent at 3 months. The greatest individual improvement in the MAQ score was from 4 to 0, in the intervention group. In the control group, only 7 of the 60 (11.7%) participants were identified as adherent at 3 months.
Changes in adherence scale scores
The changes in mean BMQ-S scales and BIPQ scales for the intervention and control groups are provided in tables 3 and 4. Figure 3 provides the changes in the mean scores of the MAQ, BMQ-S and two items of the BIPQ (‘treatment’ ‘control’ and ‘coherence’) for participants in each of the tailored strategy groups at 3 months. Changes observed in the BMQ-S and BIPQ scores reflect the type of intervention implemented. Minimal changes in the BMQ-S and BIPQ scores were visually observed at 3 months in the group that received a reminder intervention. In the group that received a cognitive-educational intervention, we observed an increase in the mean BIPQ treatment coherence score, reflecting an increase in perceived understanding of their illness. Participants in the behavioural-counselling intervention group underwent a brief version of health coaching, which resulted in an increase in the BMQ-S necessity score and a decrease in BMQ-S concerns score over time. These changes reflect stronger necessity beliefs towards medicine and weaker concerns beliefs about their medicine. The visually observed changes on the BMQ-S and BIPQ scores were sustained at 6 months.
A measurement-guided medication management approach using validated adherence scales to inform a targeted and tailored intervention improved adherence to a recently initiated medication for chronic disease at 3 and 6 months.
Interventions that have been successful tend to be multifaceted, complex and involve repeated follow-up.14 Despite these results, the outlook for adherence research may not be quite so bleak. Few studies included in the review were sufficiently powered to observe improvements in clinical outcomes. The lack of studies consistently demonstrating benefits in clinical outcomes says more about the size of the trials than the success or otherwise of the intervention. Furthermore, many of the studies included in the review neither targeted a non-adherent population nor tailored the intervention to the individual's reasons for non-adherence. There is increasing evidence that studies that target a non-adherent population and tailor the intervention to individual-specific reasons for non-adherence are more effective for improving adherence.42 ,43
Three key components contributed to the success of the intervention employed in this study. First, trial participants were identified as non-adherent using a well-validated adherence scale (MAQ). Second, participant responses to validated adherence scales (MAQ, BMQ-S and BIPQ) were used to provide insight into the likely reasons behind the participant's medication non-adherence. This permitted targeting the adherence support strategy to the participant. Third, the discussion between the investigator and participant led to a shared decision on the most appropriate tailored strategy to support the participant's adherence to their medication.
Assessing an individuals' adherence status would seem an obvious first step prior to implementing a strategy to support adherence, particularly if no intervention is required because the individual is adherent. The improvement in adherence observed in our study is consistent with other studies that enrolled a non-adherent sample for an intervention to support their adherence.20 ,44
The findings of our study suggest that it may be possible to achieve the benefits observed from complex, multifaceted interventions with a much simpler intervention providing that the intervention is targeted to a non-adherent population and tailored to the individual's specific reasons for non-adherence. The intervention employed in this study was easy to administer and quick enough that it could be incorporated into day-to-day practice. The improvement in adherence observed in our study is consistent with other studies that targeted an intervention to a non-adherent sample,20 ,44 and tailored an adherence strategy to the participant-specific reasons for non-adherence.11 ,45–48 Determining the reasons for medication non-adherence facilitated the introduction of interventions that would be more likely to improve medication adherence.1 We used the MAQ to distinguish whether non-adherence to medication was unintentional, intentional or a mix of both, along with the BMQ-S to elicit beliefs about medicines and the BIPQ to identify illness representations, to identify and explore participant-specific reasons for non-adherence. The participant's responses to these tools were clarified with further discussion, and the investigator and participant selected and implemented an individualised, evidence-based strategy to support adherence.
The success of key aspects of the intervention, such as, accurately identifying participant-specific reasons for non-adherence and effectively implementing appropriate adherence support strategies, are supported by the changes that were observed in the participant's responses to the adherence scales at 3 and 6 months. The changes to adherence scale responses are consistent with those that would be expected from successfully implementing specific adherence support strategies. A behavioural counselling strategy was employed in participants with significant concerns about their medicines and a limited belief in their necessity. Following implementation of the strategy, participants reported improved adherence and expressed less concerns and a stronger belief in the necessity of their medicines at 3 and 6 months. Similarly, a cognitive-educational strategy was employed in participants who expressed a limited understanding of their disease on the BIPQ treatment coherence scale. Following implementation of the strategy, participants reported improved adherence and that they felt they had a much better understanding of their disease. Finally, those participants who identified forgetfulness about taking their medication did not have large differences in their response to the BMQ-S or BIPQ, but did report improved adherence and less forgetfulness on the MAQ at 3 and 6 months in response to implementation of a reminder strategy.
The association between beliefs that medicines are necessary and concerns towards medicines and medication adherence has been well-established in the literature.18 ,34 ,37 ,38 ,46 ,49 ,50 Further, BIPQ treatment coherence and treatment control scales have been related to non-adherence in previous studies in patients with hypertension and type 2 diabetes.39 ,51 However, no studies have linked strategies used to support medication adherence with improvement in specific measures included in the BMQ-S or BIPQ. This approach provides an avenue for further research to explain how an intervention may have impacted adherence.
The study would have been improved by the addition of a reliable objective measure of adherence. The study recruited participants who had recently initiated one of several medications to reduce cardiovascular risk or manage type 2 diabetes. While this is a benefit of the study, one consequence is that it makes electronic monitoring of medication adherence (such as via products like MEMS) impractical. Prescription refill counts from the participating pharmacies were not a reliable alternative because participants were free to refill their prescriptions at pharmacies not participating in the trial. Most of the medicines participants were taking were subsidised on Australia's Pharmaceutical Benefits Scheme. This national pharmaceutical claims database provides the best prospects for a reliable objective measure of adherence. However, at the time of the study it was not possible to receive individual-level pharmaceutical use data in a timely or cost-effective manner. We hope to rectify this in future studies.
The MAQ is a very well-validated measure of medication-taking behaviour. While self-report measures are prone to overestimating adherence,52 the more likely problem in this study was that MAQ results identified some participants as non-adherent when an objective measure would have identified the participant as adherent. Using a cut-off of a MAQ score >0, ∼80% of the enrolled population were identified as non-adherent. If the MAQ incorrectly identified participants as non-adherent, this would be expected to reduce rather than increase the effects of the intervention.
The process of following up participants at 3 and 6 months may have influenced adherence to the medication independently of the intervention. Whether or not this effect occurred is hard to judge, but any effect would be small and affect both the control and intervention group. MAQ scores in the control group did not change to a statistically significantly degree during the follow-up. It should also be noted that the use of adherence scales (MAQ, BMQ-S, BIPQ) to inform and then assess tailored interventions is preliminary. These scales have been validated at single time points. Further research is needed to assess the reliability of these scales in measuring ‘changes’ in the participant's beliefs about their medicines and health.
Some studies have shown that improving adherence to medications, improves clinical outcomes, such as blood pressure control, blood glucose levels and lower lipid levels.53–55 This study had a relatively small sample size and was not powered to measure clinical outcomes. We hope to conduct this study in a larger cohort to show the effect of the intervention on clinical outcomes.
We believe that the intervention could be successfully employed in a wide range of pharmacies. It needs to be recognised, however, that the intervention was examined in only two pharmacies that service the middle working class. Further work is needed to assess whether aspects of the intervention or outcomes are influenced by factors relating to differences in the types of pharmacies and the communities that they serve.
The interview was performed by a sole pharmacist. Different pharmacists conducting the interview may result in different results. We hope to explore this in a larger study using a number of different pharmacists who have undergone training.
A measurement-guided medication management adherence intervention using validated adherence scales successfully improved adherence in non-adherent patients. This intervention was easy to administer and quick enough that it could be incorporated into day-to-day practice. If this targeted and tailored intervention proves successful in larger studies that assess clinical outcomes, it has the potential for widespread implementation.
Contributors T-M-UN, AL and NC, designed the research and wrote the manuscript. T-M-UN performed the research and analysed the data.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Ethics approval was obtained from the School of Pharmacy Ethics Committee, University of Queensland (approval number 92013/5).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement No additional data are available.