Defining, measuring, and addressing medication non-adherence in cardiovascular disease

1. Background

Cardiovascular medications that can increase life expectancy, reduce hospitalizations, and improve quality of life are more available than ever [1]. Therefore, we are faced with a new challenge: we know which therapies we should be using, but how do we enable and empower people to use them? This question centers around the concept of therapy adherence, which is defined as “the extent to which a person’s behavior – taking medication, following a diet, and/or executing lifestyle changes – corresponds with agreed recommendations from a health care provider” [2]. Adherence is a complex, multi-factorial collection of behaviors and environmental factors that contribute to how an individual engages with their prescribed therapy [3]. Accordingly, estimates of the burden of non-adherence, and descriptions of its antecedent causes vary widely in the literature [3]. Given the chronic nature of cardiovascular disease, international guidelines contain recommendations to monitor and improve long-term adherence to sustain the clinical benefits of medications [1].

Much like how the effectiveness of medications vary between individuals and their disease phenotypes, interventions addressing non-adherence also have varying benefits depending on patients’ profiles of medication non-adherence [3]. Therefore, an understanding of the causes and contributors to suboptimal adherence is required to select the most appropriate intervention.

1.1. Definitions of medication non-adherence

There are two main types of medication non-adherence: primary and secondary non-adherence [3]. Primary non-adherence is where medications are prescribed but never actually dispensed to patients, meaning patients never initiate therapy [3]. Secondary non-adherence refers to having the medications dispensed, but not using them as intended (e.g., taking too little or too much). These two types of medication non-adherence are distinct and can be measured specifically at the patient- and the population-level.

The behaviors driving primary and secondary medication non-adherence can also be classified as intentional or unintentional [3]. Intentional non-adherence is behavior-based, where a person purposefully chooses to not engage with therapy [3]. Unintentional non-adherence relates to non-rationalized behaviors, such as cognitive, sensory, or dexterity issues [3]. Both primary and secondary non-adherence can be due to either intentional or unintentional non-adherence [3]. These features of non-adherence are important when measuring adherence and choosing the right intervention to implement.

1.2. How adherence can be measured

Adherence is typically measured by personal observation (such as self-reporting or observation) or by administrative data (such as prescribing or dispensing data). Self-reported measures are often less reliable and subject to responder bias [3]. However, examples of studies on post-acute coronary syndromes have shown self-reporting to align well with administrative data [4]. Administrative data can be an excellent tool for population-level adherence estimation (e.g., proportion of days covered), but many assumptions are required, which reduces comparability across therapeutic areas and jurisdictions [5]. Pill-counting is considered the gold standard in clinical trials but is laborsome in routine clinical practice, especially when patients are taking multiple medicines. Currently, there is no standardized and accepted method for clinicians to screen for medication non-adherence in the clinic.

1.3. Burden of medication non-adherence

Clinical guidelines are largely based on evidence from randomized control trials (RCTs), where adherence is greater than the real-world experience. For example, in RCTs of statins, non-adherence ranged from 1% to 2%, while real-world figures exceed 10–45% [6]. The financial burden of non-adherence in cardiovascular disease may reach as much as $USD20,000 per individual in preventable health care costs [7]. This is largely attributable to increases in direct costs to the health system due to major adverse cardiovascular events, and indirect costs due to productivity loss [7]. Overcoming barriers to non-adherence is crucial in improving individual outcomes as well as the wider economic burden of cardiovascular disease.

1.4. How to practically identify barriers

While patients are often arbitrarily considered as ‘adherent’ or ‘nonadherent,’ for the clinician, the ‘why’ and ‘how’ are critical to tailor appropriate interventions [8,9]. Risk factors for non-adherence have been extensively researched and include patient-related (age, cognition, socioeconomic status), clinician-related (experience, communication), or health-system (costs, access) factors [8,10]. Furthermore, patient perceptions of therapy can also affect adherence, such as the improvements seen in the SAMSON trial, demonstrating nocebo effects present in a sample of previously statin intolerant individuals [11]. Indeed, assumptions made by clinicians about their patient’s level of adherence are often incorrect. If we consider that only 50% of clinicians ask about adherence, it is not surprising that barriers are not commonly identified and addressed [8,12].

Considering medication adherence is a dynamic and continuous process, clinicians should systematically assess adherence as part of routine history taking in the clinic [8]. Validated instruments are available to more reliably evaluate adherence, covering medication access (to understand primary and secondary adherence) and aspects of behaviors and attitudes toward medications (to understand intentional and unintentional adherence) [3]. These tools should not replace an open conversation about experiences and perceptions of medications to ensure expectations of therapy are aligned between clinician and patient [8].

In addition to conversations with patients, medication dispensing records can often be reviewed for non-adherence. This might involve monitoring dispensing records to rule out primary non-adherence, with secondary non-adherence requiring higher fidelity tools such as monitoring devices (e.g., electronic pill bottles) [8,10].

2. How to address barriers in nonadherence

Once we understand adherence barriers that exist for an individual, we can recommend specific direct and indirect interventions, both of which require active participation by the patient and clinician.

2.1. Indirect interventions

2.1.1. Text messaging and digital ‘nudges’

A Cochrane review of 18 studies that used text messaging interventions to improve adherence demonstrated 10 studies with benefits but with low overall quality of evidence [13]. This was due to the heterogeneity of text message content and a wide spectrum of health literacy and digital engagement exhibited by participants [13]. There are many different models utilizing outreach services and follow-up messaging when patients fail to fill prescriptions within a specified timeframe. Barriers to success include alert fatigue, sustainability, and lack of bi-directional engagement [8,13]. One of the most robust and recent examples of digital nudges was the EPIC-HF trial, where the use of a 3-min video and medication checklist increased therapy intensification for people with heart failure in 49.0% of the intervention compared to 29.7% in the control group [14].

2.1.2. Indirect – monitoring and notification systems

With increases in data linkage and timely availability, monitoring and notification systems are a fast-growing area to address adherence [8,10]. An example could be the use of dispensing data to flag primary non-adherence and calculate secondary adherence [8]. In healthcare environments with limited resourcing, dashboards presenting these data could serve as a method for prioritizing medication adherence interventions. There are also examples of more direct monitoring through electronic pill bottle recorders, however these are laborsome and costly. There is no gold standard method for remote monitoring of medication adherence [8].

2.1.3. Indirect – cost subsidy and incentivisation

Medication cost often creates challenges in accessing medication. In the United States, it was reported that 18% (360/2005) of surveyed individuals aged ≥65 years cited cost as the reason for non-adherence [15]. Additionally, registry data from 8,285 Australians aged ≥45 years found medication subsidization was associated with improved persistent use of medications at 12 months (Relative Risk 1.29 [95%CI 1.16–1.44]) [16]. While medication costs are not in the remit of an individual clinician, or single health institution, selection of therapies (or reasonable alternatives or generics) that are less costly may improve intentional adherence.

2.2. Direct interventions

2.2.1. Tailored medication review and behaviour management

Tailored medication review processes have been shown to enhance adherence and patient outcomes [8,9,17]. A systematic review highlighted multiple pharmacist interventions to address medication adherence and cardiovascular disease outcomes across diverse ambulatory settings [17]. When compared to a matched cohort, tailored medication reviews delivered by specialist pharmacists to patients with acute coronary syndrome increased adherence to optimal medical therapy by 13% (31% vs 44%, p = 0.04) [4]. Another intervention to consider is the uptake of motivational interviewing and self-management education, with evidence of marginal benefits reported in a recent meta-analysis of 17 studies (Relative Risk 1.13 [1.01–1.28]) [9]. In the same way all cardiovascular guidelines point toward multi-disciplinary approaches, the same philosophy can be applied in the context of designing and delivering interventions for medication adherence. Successful interventions often involve utilizing pharmacists, nurses, and other health professionals [8,10]. When using these data-driven and collaborative models to enhance adherence, it is imperative that care is delivered in a patient-centered and respectful manner [4].

2.2.2. Medication modification

The use of multiple classes of medications to treat cardiovascular disease is common, such as four-pillar therapy in heart failure [1]. This is reflected in real-world data, with 92% of people starting on antiplatelets also using other cardiovascular medicines [18]. Polypills offer a way of simplifying multiple medication regimens into a single dose formulation to improve adherence and outcomes [19]. Economic modeling has shown polypills, when compared to using multiple monotherapies, are effective in improving adherence and are cost-effective [19]. The last decade of drug development has given rise to significant advancements in therapies for cardiometabolic diseases, such as dyslipidaemia, diabetes and obesity [8]. While less frequent administration seen with these agents (weekly through to bi-annually) should favor adherence, newer issues such as cold chain management, needle phobia, and tolerability, all pose newer barriers to adherence.

2.2.3. Dose administration aids (DAA’s)

DAA’s include blister packs, sachet systems or automated dispensing devices. These directly address unintentional non-adherence when a person has difficulty taking the right amount of their medication at the right frequency [20]. They are commonly prepared by pharmacies or via validated packing robots. Similar to many of the interventions discussed, the population who will likely benefit most from DAAs are heterogenous and not well defined. DAAs can improve secondary non-adherence, providing patients are appropriately trained and able to take the medications from the pack [20].

3. Future directions in managing adherence

An array of tools to assess and address non-adherence exist, yet there has been little progress toward universal strategies to improve adherence. System-level changes that integrate prescribing and dispensing data may assist in identifying suboptimal adherence and to monitor the impact of interventions. Funding arrangements need to incentivize multi-disciplinary value-based care, inclusive of medication adherence, rather than funding shorter and isolated health care visits. Clear definitions of adherence, and transparency of methods for measurement, in addition to an agreed taxonomy of intervention types are imperative to allow comparison of effectiveness and feasibility of implementation.

Author contributions

AL and AJN contributed to the concept and initial design of manuscript.

LD contributed to writing and reviewing the manuscript.

SN contributed to reviewing the manuscript.

All authors reviewed and approved the final version of the manuscript.

Disclosure statement

AL. Consulting for Sanofi, Boehringer Ingelheim, Novartis LD none.

SJN. Research support from AstraZeneca, Amgen, Anthera, CSL Behring, Cerenis, Cyclarity, Eli Lilly, Esperion, Resverlogix, New Amsterdam Pharma, Novartis, InfraReDx and Sanofi-Regeneron and is also a consultant for Amgen, Akcea, AstraZeneca, Boehringer Ingelheim, CSL Behring, Cyclarity, Daiichi Sankyo, Eli Lilly, Esperion, Kowa, Merck, Takeda, Pfizer, Sanofi-Regeneron, Vaxxinity, CSL Seqirus, and Novo Nordisk.

AJN. Research support from AstraZeneca, Amgen, Eli Lilly, Novartis and is a consultant for Amgen, AstraZeneca, Boehringer Ingelheim, CSL Sequiris, Eli Lilly, GSK, Sanofi Pasteur and Novo Nordisk.

Funding

No formal funding was provided for the development of this editorial. LD is supported by a Postdoctoral Fellowship (107125) from the National Heart Foundation of Australia.