Prior medication adherence of participants and non participants of a randomized controlled trial to improve patient adherence in cardiovascular risk management

Cardiovascular risk reduction is predominantly based on lipid and blood pressure lowering treatment, inhibition of platelet aggregation, smoking cessation and control of obesity [1]. A limitation in lowering cardiovascular risk is poor adherence to prescribed medication [2] which may consequently can lead to increased morbidity, mortality and costs [3‐6]. Nevertheless, a recent review with mostly cohort studies, showed that only 60% of people who use cardiovascular medication, were adherent to their cardiovascular medication [7]. In view of that there is a need for interventions to improve medication adherence in this population. Although there is a considerable amount of research in the field of interventions to improve medication adherence in cardiovascular patients, they often show only small effects [8]. It is suggested that patient recruitment methods in randomized controlled trials (RCT) to improve patient adherence to medication may influence outcome [8, 9]. An important observation is that patients participating in RCTs generally have higher adherence rates at baseline than could be expected based on observational studies [10‐14]. It is conceivable that the informed consent procedure results in a selection of patients with higher adherence rates [12]. Willingness to participate is positively influenced by patients’ engagement with their medical condition, high level of education and the influence of an important person [11, 15]. These characteristics are also considered as positive determinants for medication adherence [16]. Although a recent review showed that the inclusion of non adherent patients was the single feature significantly associated with effective adherence interventions, most studies seem to include patients because they are willing to participate not because they are poor adherent [11]. It is suggested that patients participating in these RCTs already have a pre-existing high adherence level at baseline [10, 12‐14]. Selection of participants with high levels of adherence at baseline, makes it difficult to show an improving intervention effect (ceiling effect) [8]. More understanding about the medication adherence of participants as well as non-participants before the start of these RCT’s may contribute to a better understanding of why so many studies show no improvement in medication adherence. One possible explanatory determinant for (non) adherent behaviour is medication beliefs. Personal beliefs about need for treatment (necessity beliefs) and concerns about several potential adverse consequences (concern beliefs) could explain a large part of (non) adherent behaviour [16‐18]. If patients perceive that the need for medication outweighs the concerns, they are more likely to be adherent to their medication(s) [19].

In total, 900 patients with a new cardiovascular event between October 2011 and October 2013 were eligible for participation into the MIRROR trial. Of these, 419 agreed and 481 refused participation. Of all the non-participants who received a letter for another study, 220 did not respond. Consequently, 261 non-participants were classified as responders (Fig. 1).

To our knowledge this is the first study exploring the differences in medication adherence in patients who did or did not consent to participate in an RCT evaluating the effect of an intervention to improve medication adherence. Our study showed that patients willing to participate in an RCT evaluating the effect of an intervention to improve medication adherence, have a comparable adherence level to patients who declined participation. Even by further exploring the non-participant group in responders and non-responders, we did not observe differences in adherence between the groups. Consequently, the results of this study suggest that a population representative in adherence level participated in an RCT evaluating the effect of an intervention to improve medication adherence.

Previous studies suggested that patients not participating in RCTs to improve medication adherence have a different pre-existing adherence level from patients who participate [10‐14]. This was supported by the observed differences in adherence levels between these RCTs and observational studies [10‐14]. Typically, adherence levels among patients in RCTs were higher than in observational studies. Although not different among participants and non-participants, adherence in this study was also high. An explanation for the high adherence rate in both groups could be that we started inclusion for the RCT within six weeks after the cardiovascular event. For cardiovascular patients who just had an event, the need for adherent behaviour is emerging [31, 32]. Yet, as the event fades and symptoms subside, adherence levels can also decline [33]. Research with a long follow up is needed to establish if there will be a difference in adherence between participants and non-participants over time.

In all groups, we observed significant differences in patient characteristics. Compared to non-participants, participants were younger and more were highly educated. This was also observed among responders and non-responders. These are known as prognostic characteristics for patients who are willing to participate in a clinical trial [15] and for a high adherence level [34, 35]. Although the relationship between socio-demographic variables and adherence is mainly weak and inconsistent [34, 36, 37] it was expected that these characteristics could have been an explanation for the assumed higher adherence rates in the participant groups. However, we could not support this hypothesis. Also, next to the high adherence rate, a high mean NCD score was present in all groups. This only confirmed the adherent behaviour in both groups [19]. It is also congruent with earlier studies showing that medication beliefs can be a more powerful predictor of medication adherence than clinical and socio-demographic factors [19, 38]. However, we did not observe a relationship between NCD and trial participation. We did observe a significant difference in the beliefs about medication groups in the (non) responder groups. More patients of the non-responder group were also in the indifferent and skeptical group. Non-responders of surveys are known for more negative evaluations of healthcare [39]. This could be associated with higher concern beliefs in medication as these are partly influenced by the prescriber-patient relationship in healthcare [40]. The number of patients who differed in these groups was very small and the NCD did not differ. More research is needed to draw any conclusions on this point.

This study had some limitations. We had to deal with missing data especially in the self reported questionnaires BMQ and MMS®. There were fewer missing in the participants group compared to the non-participants group. The questionnaires were just implemented in the screening program. As the questionnaires were also part of the MIRROR trial, more attention could have been paid to participants for documenting these questionnaires. So there were more patients in the non-participant group who did not fill out the MMS®. These patients could very well be non-adherent [21]. There are different methods available to measure adherence. Each method has advantages and disadvantages. The MMS® [22, 24, 41] is a validated questionnaire that can be easily applied to large populations. As MMS® is a subjective measure, adherence levels may be higher than what is expected in real life. Refill data from the out-patient pharmacy on the other hand has been used extensively to provide insight into drug acquisition and dispensing [42]. However, to use the pharmacy refill data we need an informed consent from patients. This study however used data from patients collected only in standard care because we wanted to include patients who declined participation in a RCT. Other methods, such as MEMS or pill count, seem to influence patient’s behavior through direct confrontation. Moreover, application of MEMS is relatively expensive, especially when applied in standard care [42]. The BMQ was used because, to our knowledge, is the only validated questionnaire that evaluates patients’ beliefs, necessity and concerns the patient has according to his illness and prescribed medication. This in contrast to other validated adherence questionnaires that measure specific medication-taking behavior of patients [26, 43]. The high NCD score confirmed the prediction of adherent behaviour in both groups [19, 38] .

This study showed no differences in medication adherence between participants and non-participants prior to the inclusion of the MIRROR trial. A representative group seems to have participated in this randomized controlled trial designed to improve medication adherence [20].