Abstract

Suboptimal medication adherence is a widespread problem in ambulatory care of chronic diseases, with deviations in either direction from the prescribed dosing regimen. For the non-vitamin K antagonist oral anticoagulants (NOACs), such deviations occur and can lead to bleeding or clotting, as suboptimal adherence involves temporary periods of either overdosing or underdosing. In this expert review, we discuss: (a) the proper definition of adherence in terms of its three elements: initiation, implementation, and discontinuation; (b) how adherence is reliably and accurately measured and (c) successfully enhanced, to achieve and maintain safe and effective levels of NOAC-based anticoagulation. We also discuss the comparative effects of prescribing the same total daily dose, given either once-daily or as half-strength twice-daily doses. Because NOACs have plasma half-lives of ∼12 h, the twice-daily dosing regimen is less prone than the once-daily dosing regimen to hazardously high peaks or hazardously low troughs in anticoagulant concentrations and associated actions. As in other fields of oral drug treatment, the continuity of drug action is greater with twice-daily than with once-daily dosing, despite the fact that a few more doses are skipped with twice-daily than with once-daily dosing. This paradox is explained by the disproportionately greater impact on drug action of skipping a once-daily than a twice-daily dose. Integration of these principles into real-world medication management is the next step in the improvement of oral anticoagulation.

Introduction

From patient compliance to medication adherence

In 1997, the American Heart Association issued a visionary statement in which patient compliance (the extent to which medical recommendations are followed as defined) was described as ‘a complex behavioural process, strongly influenced by the environment in which the patients live, how healthcare providers practice, and how healthcare systems deliver care’.1 This definition acknowledges that a patient's compliance is also influenced by system-level factors, e.g. the provider of healthcare, healthcare organizations, economics, and other aspects of the overall system of healthcare.

Later, the term ‘patient compliance’ has been increasingly replaced by ‘medication adherence’. The shift from ‘compliance’ to ‘adherence’ reflects a fundamental change in understanding relationships between patients and practitioners2 as the term ‘adherence’ implies cooperation between patient and prescriber, rather than the patient's passive obedience to the physician's instructions. Patient adherence to medications (i.e. the process by which patients take their medications as prescribed) is now recognized to consist of three components: (1) the process starts with initiation of the treatment, then (2) continues with the implementation of the dosing regimen, and eventually (3) discontinuation marks the end of therapy, with persistence being the length of time between initiation and discontinuation.3 In the context of life-long therapy like anticoagulation, discontinuation may occur in response to the prescriber's decision to halt the treatment, but most often it is the result of a unilateral action by the patient, without the knowledge of the prescriber. Still, the decision to discontinue anticoagulation therapy after a major bleed will likely be the result of a joint decision from the patient and physician, although patient views may temper the final decision on whether the anticoagulation therapy is restarted or not. When assessing adherence in studies, it may thus be difficult to assess the subtleties of whether the decision to discontinue therapy is taken by the patient, prescriber, or both.

Suboptimal adherence in anticoagulation therapy

Adherence to warfarin is suboptimal

The consequences of non-adherence are particularly problematic for drugs with a narrow therapeutic window, as is the case for warfarin: anticoagulation levels below the target range of International Normalised Ratio (INR) values are associated with increases in thrombotic risk, while levels above the therapeutic INR level increase the bleeding risk;4 both can be life-threatening or result in major morbidities. In addition to the risk of non-adherent patients being assumed anticoagulated when they are actually not, non-adherence to anticoagulation therapy can also lead to withdrawal symptoms with a prothrombotic state in the non-adherent patient. In both the ROCKET-AF5 and ARISTOTLE6 trials, an excess of thrombo-embolic events has been observed at the end of the study period, during the transitioning phase to vitamin K antagonist (VKA) therapy.

Kimmel et al. have assessed adherence to the VKA warfarin7 using the Medication Event Monitoring System (MEMS®) that records the time and date of each pill bottle opening, a ‘medication event’ that is tightly linked to dose ingestion.8 In that study, non-adherence to warfarin was shown to be widely prevalent.7 Among 136 patients observed for a mean period of 32 weeks, 92% had at least one missed or extra dose, while 36% missed more than 20% of their prescribed doses and 4% had more than 10% extra doses. Moreover, this poor implementation of the dosing regimen was shown to have a considerable effect on anticoagulation control: missing one to two doses a week (20–30% missed doses) was associated with up to a two-fold increased odds of sub-therapeutic INR, while taking extra doses was associated with an increased risk of a too-high INR value.7

Could adherence improve with non- vitamin K antagonist oral anticoagulants?

The various well-known drawbacks of warfarin therapy, including its narrow therapeutic window, its many potential drug interactions, and the considerable time required for onset and offset of its therapeutic effect, have spurred the development of non-VKA oral anticoagulants (NOACs). Up to now, no prospective adherence studies on NOACs have been reported; adherence to apixaban in patients with non-valvular atrial fibrillation is currently under assessment in the ‘Assessment of an Education and Guidance Programme for Eliquis Adherence in Non-Valvular Atrial Fibrillation’ (AEGEAN) study (ClinicalTrials.gov: NCT01884350). The AEGEAN study will not compare adherence of NOACs to VKA, but randomizes patients started on apixaban between ‘regular care’ and ‘regular care with structured education and follow-up by a virtual clinic’. Adherence is measured via an electronic smart package, but the results from these measurements are not used for feedback during the trial.

While the population that will be taking the NOACs is the same as those taking warfarin, differences in adherence to NOACs compared with VKA can nevertheless be expected in terms of initiation, implementation, and discontinuation. Patients are often reluctant to use warfarin because of a fear of bleeding; thus, they may prefer NOACs and therefore be more inclined to initiate them. Moreover, patients tend to discontinue VKA therapy due to poor maintenance in the INR range and associated side effects; thus, persistence may improve when using NOACs instead of VKA. Indeed, a review of a healthcare claims database for propensity-matched cohorts of atrial fibrillation patients newly initiated on rivaroxaban or warfarin showed that treatment persistence was significantly better with rivaroxaban than with warfarin (hazard ratio (HR) = 0.66; 95% confidence interval: 0.60–0.72, P < 0.0001).9

In contrast, NOACs have a much shorter half-life of around 12 h, meaning that the anticoagulation effect will rapidly decline when scheduled doses are not taken. Non-vitamin K antagonist oral anticoagulants may thus require a stricter implementation, although the consequences of suboptimal adherence will depend on the dosing schedule and the forgiveness of the specific drug as detailed below in this review.

For patients with suboptimal adherence that is not caused by VKA side effects or other VKA-related difficulties, simply switching to a NOAC will not necessarily improve adherence. Hence, a comprehensive analysis of the reasons for a patient's non-adherence would be needed, in order to provide the patient with tools and support to improve his or her level of medication adherence.

Management of patient adherence

For many years, medication adherence has been considered solely a patient problem. However, it is now becoming clear that patient adherence is responsive to certain management methods that can improve continuity of patients' exposure to prescribed medications, as detailed below.10 There is, however, no single solution to achieve that continuity; simply changing the dosing frequency, sending text messages, or providing an informative website to a patient will not solve adherence problems. The solution must be based on reliable quantification of the dosing errors incurred by patients, not limited to average quantities of doses missed. Reliable data on when lapses in dosing occur, for how long, and the dosing patterns that prevail as patients resume correct dosing can reveal patient-specific causes of non-adherence and point to individualized solutions. Such detailed and reliable information forms the foundation for new models of integrated care.

Measurement of adherence

In addition to using a transparent and well-defined taxonomy, it is critical to appropriately quantify the three elements of medication adherence (initiation, implementation, and discontinuation). Apt quantification of adherence should provide researchers, clinicians, and patients with meaningful metrics which should be reliable, allow tracking over time, be coherent with the three elements of medication adherence, and be implementable on a large scale.

Various methods are currently used to measure adherence, each with specific advantages and limitations. Blaschke et al.11 have described these advantages and disadvantages in detail, but did not distinguish between the three elements of patient adherence. Below, we summarize the main characteristics of different adherence measurement methods according to the three components of adherence (Table 1).

Table 1

Advantages and disadvantages of methods to monitor adherence

 1. Initiation 2. Implementation 3. Discontinuation 
Self-report Desirability bias Recall bias Desirability bias 
Pill counts Easily censored by patient Only aggregate summary Easily censored by patient 
Direct methods (PK/PD) Requires sampling after prescription Sampling is too sparse Subject to white coat adherence 
Prescription and refill databases Gold standard if both databases are combined Only aggregate summary Gold standard but retrospective 
Electronic monitoring Gold standard in clinical trials needs activation Gold standard Gold standard in clinical trials needs patient engagement 
 1. Initiation 2. Implementation 3. Discontinuation 
Self-report Desirability bias Recall bias Desirability bias 
Pill counts Easily censored by patient Only aggregate summary Easily censored by patient 
Direct methods (PK/PD) Requires sampling after prescription Sampling is too sparse Subject to white coat adherence 
Prescription and refill databases Gold standard if both databases are combined Only aggregate summary Gold standard but retrospective 
Electronic monitoring Gold standard in clinical trials needs activation Gold standard Gold standard in clinical trials needs patient engagement 

PK, pharmacokinetics; PD, pharmacodynamics.

One popular method for quantifying adherence is self-reporting, using, for example, diaries or retrospective questionnaires. Self-reports may be useful for assessing very recent drug use; self-report for cardiovascular drugs taken in the last 24 h has been reported to correlate well with the presence of the drug in the blood.12 However, in other studies assessing self-report methods to assess adherence over longer time periods, these methods have been shown to be biased towards overestimation of drug exposure.13,14

The counting of returned, untaken doses (‘pill counts') has also been repeatedly discredited since the first reliable test of their validity by chemical marker methods in 1989;15 electronic monitoring methods have confirmed and extended the dynamic range of the results obtained through chemical marker methods.8,11,16 Nevertheless, clinical researchers still continue to report pill-count data, despite well-documented unreliability. One recent example is the failed STABILITY study that relied on discredited criteria and methods for assessing in-trial adherence to the dosing regimen of darapladib, a selective oral inhibitor of lipoprotein-associated phospholipase A2.17

Methods that are based on sampling (e.g. blood sampling to measure drug levels in the blood) performed during a clinical visit for therapeutic drug monitoring are subject to the phenomenon of ‘white-coat adherence’, which is a short-term escalation in adherence during a few days prior to the scheduled visit to the clinic or laboratory.18–21 Many drugs, but not all, can be restored to their therapeutic concentration ranges by 1–3 days of correct dosing.

And while electronic prescription and dispensing databases provide objective data, they provide no information on when doses were taken or missed. Temporal sequence is an essential measurement in causal inference, as one seeks to understand the consequences of intermittently omitted doses, of recurrent first-dose effects when dosing suddenly resumes after a period of interruption, and when hazardous rebound effects occur in sequence with interrupted dosing.

Of the currently available options, automatic compilation of dosing histories using electronic detection of package entry (smart packages) or direct detection of pills in the stomach (smart pills) is the only reliable and sufficiently richly sampled method to estimate the three elements of patient adherence. Electronic methods for compiling drug dosing histories are often used as a standard for quantifying adherence, the parameters of which support model-based, continuous projections of drug actions and concentrations in plasma that are confirmable by intermittent, direct measurements at single timepoints.11 While electronic detection of package entry is an indirect method of estimating when and how much drug is administered, it can accurately project the time-course of drug concentrations in the plasma.8

Smart packages deliver a reliable detailed assessment of dosing history data over time, allowing the identification of different patterns of adherence (Figure 1). MEMS® is a smart package in the form of a cap that contains an electronic chip registering the time and date of each pill bottle opening, and has been widely used in clinical trials, as illustrated by the more than 650 available peer-reviewed publications (www.iAdherence.org). Smart packages are safe, un-intrusive (easy to use), long lasting (battery), the resulting adherence data are reliable and detailed, and can easily be adapted to any form factor (blister packs,23 inhalers,24 injectable,25 etc.) or for polymedication.23

Figure 1

Varying patterns of adherence. The same amount of drug intake over a given time period may be the result of different intake patterns, each requiring specific approaches for optimization. Electronically compiled drug dosing histories are shown for four patients who each took 75% of their prescribed doses during a 3 months period. The blue dots represent electronically captured dosing times; the vertical grey bars mark omitted doses. The first three patients display sub-optimal implementation: patient A missed mainly evening doses, patient B missed both evening and morning doses, and patient C displays a drug holiday (i.e. three or more days without a dose). Patient D initially had a high level of adherence, but discontinued the treatment prematurely. Figure adapted from Vrijens et al., Expert Rev Clin Pharmacol 2014.22

Figure 1

Varying patterns of adherence. The same amount of drug intake over a given time period may be the result of different intake patterns, each requiring specific approaches for optimization. Electronically compiled drug dosing histories are shown for four patients who each took 75% of their prescribed doses during a 3 months period. The blue dots represent electronically captured dosing times; the vertical grey bars mark omitted doses. The first three patients display sub-optimal implementation: patient A missed mainly evening doses, patient B missed both evening and morning doses, and patient C displays a drug holiday (i.e. three or more days without a dose). Patient D initially had a high level of adherence, but discontinued the treatment prematurely. Figure adapted from Vrijens et al., Expert Rev Clin Pharmacol 2014.22

One more direct method to measure adherence is the electronic detection of pills in the stomach by formulating the drug in question with an electronic ‘chip’ included directly in each pill.26,27 Once the chip-containing pill is swallowed, a signal is briefly generated via the creation of a short-lived voltage difference through ionic differences between gastric fluid and the electronic ‘chip’. The signal is detected by an antenna integrated into a dermal ‘patch’, worn on the patient's torso (which in the future may be integrated into other implanted devices like an implantable cardioverter-defibrillator, pacemaker or implantable loop recorder). The signal is then amplified for transmission to a nearby mobile device, and transmitted to a server. However, the chip-in-the-pill method has potential drawbacks such as the risk of failed detection (∼5%),27 the need to reformulate and revalidate stability characteristics of the chip-containing medications, and, so far, the lack of information on the cost of using this method.

Once- vs. twice-daily dosing: lessons from electronic monitoring

The importance of timing of drug dose intake

Reducing the frequency of dosing has long been widely believed to improve medication adherence. This belief arises from the focus on percentages of doses taken irrespective of when doses were or were not taken; the percentage of doses taken is generally higher with less frequent dosing regimens. A recent analysis of a large claims database assessing 10 697 patients with atrial fibrillation showed that the adherence and persistence for intake of antidiabetics, antihypertensives, calcium channel blockers, or diuretics (no anticoagulants were studied) was significantly higher for drugs with a once-daily intake regimen than for those with a twice-daily regimen.28

However, drug actions are both dose- and time-dependent. To illustrate, having doses clustered at odd times with long intervals between those clusters, can create an impressively high percentage of doses taken, but does not provide continuity of drug action. To predict real therapeutic consequences of non-adherence, it is not sufficient to only focus on the percentage of doses taken; it is crucial to also take into account the distribution of exactly when doses were taken.

Comprehensive assessment of the risks of a patient's non-adherence to a medication should therefore combine detailed dosing history data obtained from electronic monitoring, as well as information on the effect of non-adherence on the drug concentrations in the patient's plasma (pharmacokinetics) and the effect on the actions of the drug (pharmacodynamics).11 Such integrated analyses are an exercise in systems pharmacology, taking into account the fact that several interacting factors are operative and must be considered simultaneously.

Superior therapeutic coverage with twice-daily dosing regimens

Analyses that combine patients' dosing history data and the pharmacokinetic properties of the assessed drug have clearly shown that a twice-daily dosing regimen maintains a better continuity of drug plasma levels than once-daily dosing for drugs with a half-life of ∼12 h. For instance, an integrated analysis has suggested superior therapeutic coverage with twice-daily compared with once-daily protease inhibitors for the treatment of HIV-infected patients.29 This model-based finding has then been supported by an outcome study showing that patients with a viral load of more than 100 000 copies/mL had a greater probability of a sustained viral response on a twice-daily than on a once-daily regimen.30

A second example comes from the superior inhibition of platelet aggregation (IPA) with twice-daily administered ticagrelor compared with once-daily clopidogrel. Model-based simulations using observed patient dosing histories showed that both average and trough IPA levels remained significantly higher for ticagrelor than for clopidogrel, despite somewhat lower percentages of prescribed doses taken with twice-daily ticagrelor than with once-daily clopidogrel.31

The clinical benefits of twice-daily ticagrelor compared with once-daily clopidogrel have also been confirmed in the Platelet Inhibition and Patient Outcomes (PLATO) trial:32 in patients with acute coronary syndrome (ACS), with or without ST-segment elevation, treatment with ticagrelor as compared with clopidogrel significantly reduced the rate of death from vascular causes, myocardial infarction, or stroke without an increase in the rate of overall major bleeding but with an increase in the rate of non-procedure-related bleeding. This result is consistent with the twice-daily regimen's ability to maintain a greater degree of continuity of drug action than was achieved by the once-daily regimen, notwithstanding the fact that a higher percentage of prescribed doses were taken with the once-daily than with the twice-daily regimen. Nonetheless, there are other factors that may have contributed to the observed differences between ticagrelor and clopidogrel, such as the fact that both drugs are not identical in their action.

These two examples show that while once-daily dosing may be seen as an option to simplify the dosing regimen and increase patient adherence, it in fact may require near-perfect adherence to achieve its intended pharmacodynamic and clinical results, whereas the twice-daily dosing is, depending on the drug's pharmacokinetics, more forgiving of variations in dose-timing or occasionally missed doses. The real therapeutically relevant question is the impact of suboptimal adherence on the pharmacologic effects of the drug.

It is of paramount importance to investigate these elements also in detail in NOAC patients, as the consequences of suboptimal pharmacologic effects are so severe (bleeding or thrombotic events, both of which may be fatal). Clearly, the above-mentioned findings cannot be just extrapolated to NOAC therapy; not only may the consequences of non-adherence differ depending on the specific characteristics of the drug, also the patients taking NOAC are different from those taking HIV medication and may therefore have specific issues due to for example ageing or dementia. Thus, it would be useful to obtain electronic dosing histories and related consequences of non-adherence from patients on NOAC therapy.

A simulation of the consequences of non-adherence with once- or twice-daily dosing

In both examples, the pharmacokinetic equivalent of a single missed once-daily dose was two to three sequentially omitted twice-daily doses. This finding is illustrated by the simulation in Figure 2, which shows the typical pharmacokinetic profile for a drug with a half-life of about 12 h, similar to NOACs.

Figure 2

Once-daily vs. twice-daily dosing: difference between intake and predicted biological impact in general. Different patterns of non-adherence lead to different exposition to ‘risk’ between once-daily and twice-daily drugs. These graphs illustrate the theoretical pharmacokinetic profiles of a dose X administered once-daily (QD), and a dose X/2 administered twice-daily (BID), for a drug with a half-life of about 12 h and a Tmax of 3 h. (A) the peak-to-trough ratio is much smaller for the BID than the QD dosing. (B) The concentration after a single missed BID dose (red dot) is similar to the expected trough concentration of QD dosing, suggesting that missing a single dose of a twice-daily dosing regimen should not be therapeutically critical. (C) The pharmacological equivalent of missing a single dose in a once-daily regimen (blue dot) is missing three consecutive doses (red dots) of a twice-daily dosing regimen. (D) Taking an extra dose results in a much higher peak for the QD than for the BID dosing regimen.

Figure 2

Once-daily vs. twice-daily dosing: difference between intake and predicted biological impact in general. Different patterns of non-adherence lead to different exposition to ‘risk’ between once-daily and twice-daily drugs. These graphs illustrate the theoretical pharmacokinetic profiles of a dose X administered once-daily (QD), and a dose X/2 administered twice-daily (BID), for a drug with a half-life of about 12 h and a Tmax of 3 h. (A) the peak-to-trough ratio is much smaller for the BID than the QD dosing. (B) The concentration after a single missed BID dose (red dot) is similar to the expected trough concentration of QD dosing, suggesting that missing a single dose of a twice-daily dosing regimen should not be therapeutically critical. (C) The pharmacological equivalent of missing a single dose in a once-daily regimen (blue dot) is missing three consecutive doses (red dots) of a twice-daily dosing regimen. (D) Taking an extra dose results in a much higher peak for the QD than for the BID dosing regimen.

In Figure 2A, one can see that when the pharmacokinetic characteristics are the same, the peak-to-trough ratio is much smaller if the drug is given at half the dose twice-daily (in red), than if the dose is given in a single dose once-daily (in blue). The theoretical pharmacokinetic profile thus shows much less variability over time with twice-daily dosing compared with once-daily dosing.

Figure 2B illustrates, in red, the pharmacokinetic profile of the twice-daily dosing modified after a single missed dose. The resulting concentration is similar to the expected trough concentration of once-daily dosing, suggesting that missing a single dose of a twice-daily dosing regimen should not be therapeutically critical.

Figure 2C shows that the pharmacological equivalent of missing a single dose in a once-daily regimen is missing three consecutive doses of a twice-daily dosing regimen. Thus, instead of assessing solely the percentage of missed doses, it is more cogent to compare the probability of two to three sequentially missed twice-daily doses vs. the probability of missing one once-daily dose.

Figure 2D shows the pharmacological consequence of an extra dose, which results in a much higher peak for the once-daily than for the twice-daily dosing regimen.

Once- vs. twice-daily dosing of non-vitamin K antagonist oral anticoagulants

The half-lives of NOACs are dependent on age and renal function of the patient, ranging from ∼9 to 14 h in young patients with normal renal function, and increasing to 9 to 17 h for factor Xa inhibitors and even 28 h for dabigatran in patients with moderate-to-severe renal dysfunction.33 Even with this variability, all NOACs are rapidly absorbed and usually have half-lives well below 24 h; nevertheless, different dosing regimens (once-daily or twice-daily) have been selected, depending on the drug (often based on phase 2 trial results) and on the particular indication. The simulation in Figure 2 indicates that, as with the two examples detailed in the previous section, twice-daily dosing of NOACs could be beneficial for maintaining continuity of drug action when there is variable drug exposure from suboptimal adherence.

The potential advantage of twice-daily dosing for NOACs is further illustrated in Figure 3, where model-based concentrations (upper plot) are projected using the dosing chronology (lower plot). The red projection in the upper plot assumes a hypothetical prescribed dosing regimen that is perfectly implemented for a dose X prescribed once daily (left) and half the dose X/2 prescribed twice daily (right). One can see that the peak-to-trough variability is larger for once-daily dosing, which could be related to increased risks of bleeding or thrombotic events, respectively. The blue lines in the upper plots show the model-projected continuous time-course of the concentrations resulting from a dosing history in which 15% of doses were missed. Omitted doses are indicated by vertical grey bars, each of which indicates a single omitted dose. The right side shows a similar plot for 15% of missed doses of a twice-daily drug. In both figures, lapses in dosing lead to lower-than-usual projected concentrations of drug, and extra doses lead to higher-than-usual projected drug concentrations. Relative to the hypothetical therapeutic window used in this example, one can see that the twice-daily dosing regimen is more forgiving for a missed dose or an extra dose than the once-daily dosing regimen for drugs with a half-life of 12 h. Therefore in practice, once-daily dosing may require more vigilance34 for single missed or extra doses and thus closer management of patient adherence.

Figure 3

Once-daily vs. twice-daily dosing: predictions for NOAC drugs. The red projection in the upper plot assumes a hypothetical prescribed dosing regimen that is perfectly implemented for a dose X prescribed once daily (QD; left plot) and half the dose X/2 prescribed twice daily (BID; right plot), for a drug with a half-life of about 12 h and a Tmax of 3 h. The blue lines in the upper plots show the model-projected continuous time-course of the concentrations resulting from a dosing history in which 15% of doses were missed (omitted doses are indicated by the grey bars, while the blue dots represent electronically captured dosing times). Relative to the hypothetical therapeutic window used in this example, one can see that the twice-daily dosing regimen is more forgiving for a missed dose or an extra dose than the once-daily dosing regimen.

Figure 3

Once-daily vs. twice-daily dosing: predictions for NOAC drugs. The red projection in the upper plot assumes a hypothetical prescribed dosing regimen that is perfectly implemented for a dose X prescribed once daily (QD; left plot) and half the dose X/2 prescribed twice daily (BID; right plot), for a drug with a half-life of about 12 h and a Tmax of 3 h. The blue lines in the upper plots show the model-projected continuous time-course of the concentrations resulting from a dosing history in which 15% of doses were missed (omitted doses are indicated by the grey bars, while the blue dots represent electronically captured dosing times). Relative to the hypothetical therapeutic window used in this example, one can see that the twice-daily dosing regimen is more forgiving for a missed dose or an extra dose than the once-daily dosing regimen.

It remains to be proven in how far these projected differences also reflect in clinical outcomes with NOACs (as was exemplified by the HIV- and ACS-trials above), but it is clear that research in this field is needed to better inform clinicians about the possible impact of the prescribed dosing scheme on the drug's therapeutic effect. The findings presented in this review show the importance of considering a twice-daily dosing regimen instead of automatically assuming that once-daily dosing would be better due to the higher percentage of doses taken. On the other hand, it should also be clear that there will not be one all-encompassing answer on which dosing regimen is best for NOACs; this question will need to be assessed for each NOAC and each patient separately.

Managing adherence to non-vitamin K antagonist oral anticoagulant therapy in clinical practice

Interventions to improve adherence

Healthcare systems and providers play a role in the management of adherence by monitoring and supporting their patients' correct medication intake.3 A recent literature review assessing interventions to improve patient adherence to medications, found that electronic monitoring (EM)-feedback was the biggest factor influencing adherence (8.8% more effective than interventions without EM feedback; P < 0.01).10 Electronic monitoring feedback interventions are designed to provide feedback on patients' personal dosing histories compiled from electronic medication event monitors. Using the patient's own dosing history, the healthcare practitioner can discuss with the patient the reason why there was suboptimal adherence at specific timepoints, and propose solutions to prevent this from happening in the future. Such adherence-improving interventions also include elements of education (to increase the patient's knowledge on his or her disease and on the importance of adherence), and motivation (to increase the patient's self-efficacy).22 Further proposals to support patients' adherence are discussed below, and compiled in Table 2.

Table 2

Proposals to improve adherence

Electronic monitoring feedback Providing feedback provided based on the patient's own detailed electronic dosing history 
Patient education Providing the patient with sufficient education about his/her disease, dosing regimen, and the importance of adherence 
Structured patient follow-up Follow-up of each patient at fixed timepoints (e.g. at therapy initiation, after 1, 3, and 6 months, etc.) 
Nurse-led interdisciplinary care programmes Interdisciplinary care programmes could be set up, with specialized nurses as the main communication point for the patient 
Telemonitoring of medication intake The provided dosing histories can be used to assessed whether intervention is needed to improve a patient's adherence 
Improving physician adherence to guidelines Guideline adherence has been shown to be higher in nurse-led care programmes 
Electronic monitoring feedback Providing feedback provided based on the patient's own detailed electronic dosing history 
Patient education Providing the patient with sufficient education about his/her disease, dosing regimen, and the importance of adherence 
Structured patient follow-up Follow-up of each patient at fixed timepoints (e.g. at therapy initiation, after 1, 3, and 6 months, etc.) 
Nurse-led interdisciplinary care programmes Interdisciplinary care programmes could be set up, with specialized nurses as the main communication point for the patient 
Telemonitoring of medication intake The provided dosing histories can be used to assessed whether intervention is needed to improve a patient's adherence 
Improving physician adherence to guidelines Guideline adherence has been shown to be higher in nurse-led care programmes 

The need for a structured follow-up for patients on non-vitamin K antagonist oral anticoagulant therapy

The process of gaining information on patient's adherence and taking preventive or corrective action will require structured care around the patient. The European Heart Rhythm Association has proposed a framework for structured follow-up of patients on NOAC therapy.33 This proposal includes follow-up at various timepoints (i.e. at initiation of the therapy, and subsequently after 1, 3,and 6 months, etc.) at which the patients need to meet with their healthcare practitioners, who can be cardiologists, general practitioners or specialist nurses. These meetings could be used to assess the patient's adherence and provide feedback, education, and motivation where needed. Regular meetings throughout the entire duration of the therapy may be needed to ascertain long-term therapy. It will be an extremely important challenge to devise strategies that improve long-term adherence. The European Heart Rhythm Association also proposed a NOAC patient card to integrate follow-up information among the different healthcare workers.33

Moreover, for atrial fibrillation, a proposal has been made for interdisciplinary atrial fibrillation expert programmes to structure daily practice.35 In such programmes, clinical nurse specialists have an important role, which can even be a central coordinating one, because they have more time for patients than physicians and are more readily accessible; they can thus function as the main point of communication. After a comprehensive physician-supervised assessment of the patient, the nurse can coordinate an interdisciplinary guideline-based management plan.

Such a structure should well serve the follow-up of NOAC therapy. Anticoagulation clinics, now focused on VKA therapy, could redefine their role for the future in this respect.36 Moreover, such a programme could also incorporate electronic and even remote monitoring of adherence, in analogy with remote follow-up of implanted devices.37 A pilot study assessing telemonitoring of medication intake in patients with chronic heart failure, with interventions initiated by healthcare providers in case of absence of transmissions, indicated that patients' acceptance of the tool was high and that telemonitoring might prove to be an effective method to improve medication intake.38

Based on the patient's dosing history, the nurse can assess if intervention is required and provide specific support tailored to the needs of the patient (Figure 4). As with device follow-up, specific server software could be developed that analyses the patients' dosing history and provides filtered alerts on a per-patient basis, or could even provide direct feedback to the patient about his or her inadequate adherence and the possible consequences. Such virtual clinic software would also benefit healthcare practitioners outside of an integrated care network. If the effectiveness of such integrated systems can be proven, appropriate reimbursement is warranted to provide incentives to set up this form of care.

Figure 4

Schematic representation of potential NOAC care models integrating monitored drug intake. (A) Various centers with integrated care can each have their own specific set-up. Within such a center, the electronic monitoring technology can be integrated as part of their overall approach. (B) Healthcare practitioners who do not belong to an integrated care center can receive support from ‘virtual clinic’ software. The electronic monitoring of the patient's adherence can be linked to this software, so that feedback to the healthcare practitioner and patient can be provided when needed.

Figure 4

Schematic representation of potential NOAC care models integrating monitored drug intake. (A) Various centers with integrated care can each have their own specific set-up. Within such a center, the electronic monitoring technology can be integrated as part of their overall approach. (B) Healthcare practitioners who do not belong to an integrated care center can receive support from ‘virtual clinic’ software. The electronic monitoring of the patient's adherence can be linked to this software, so that feedback to the healthcare practitioner and patient can be provided when needed.

Adherence and patient education

Suboptimal adherence may be due to the fact that the proper dosing regimen and its importance were not adequately described to the patient, or due to the fact that the patient did not properly understand the need for specific dosing requirements. This part of the non-adherence problem could be remedied by providing the patients with sufficient education and training. As mentioned earlier, interventions found to improve adherence often include elements of education.22 Interdisciplinary nurse-led programmes have a strong focus on educating the patient, even repeatedly during regular clinic visits or phone contacts. Atrial fibrillation patients receiving nurse-led care have been reported to have a significantly higher level of knowledge about their disease and its management, when compared with patients receiving regular care.39,40

Physician adherence to guidelines

While this review focuses on the process of patient adherence to medications, it is important to note that there is also a preceding process concerning the adherence of physicians to the guidelines for prescription of anticoagulant medication. On top of patient-related reasons, medications may not be initiated due to physician-related reasons, such as an exaggerated fear of bleeding relative to the fear of stroke. A physician is only confronted with the bleeds caused by the medication but not with the strokes that are prevented. Interdisciplinary nurse-led programmes have been demonstrated to improve physician adherence to guideline recommendations.39

Conclusion

There is an urgent need for research on adherence-optimizing technology and interventions in the NOAC field, because suboptimal adherence is widely prevalent and has severe consequences including bleeding or increased thrombotic risk, which can be fatal.

Once-daily dosing may increase absolute adherence, but twice-daily dosing regimens may be more forgiving in patients with sub-optimal adherence. Prospective clinical evaluation is needed, also to relate outcome factors of drug regimen to the type of patient: one regimen may not suit all.

Also the effect of structured care on NOAC adherence needs to be evaluated as in the currently ongoing AEGEAN trial (ClinicalTrials.gov: NCT01884350). Electronic monitoring of patients' adherence promises to be a useful tool in such integrated care systems, both for assessing adherence as well as for improving adherence through feedback based on the patient's dosing history.

MEMS is a registered trademark of MeadWestvaco Corporation.

Funding

Medical writing and editorial support for this article was funded by Bristol-Myers Squibb.

Acknowledgements

The authors would like to thank John Urquhart for his input and thorough review of this manuscript, and Joke Vandewalle and Melissa McNeely (XPE Pharma & Science) for writing assistance and coordination of manuscript development.

Conflict of interest: Bernard Vrijens is an employee of MWV Healthcare. Hein Heidbuchel is a member of scientific advisory boards for Boehringer-Ingelheim, Bayer, BMS-Pfizer, Daiichi-Sankyo and Sanofi-Aventis, and received lecture fees from these companies.

References

1
Miller
NH
Hill
M
Kottke
T
Ockene
IS
.
The multilevel compliance challenge: recommendations for a call to action. A statement for healthcare professionals
.
Circulation
 
1997
;
95
:
1085
90
.
2
Lutfey
KE
Wishner
WJ
.
Beyond “compliance” is “adherence”. Improving the prospect of diabetes care
.
Diabetes Care
 
1999
;
22
:
635
9
.
3
Vrijens
B
De Geest
S
Hughes
DA
Przemyslaw
K
Demonceau
J
Ruppar
T
et al
A new taxonomy for describing and defining adherence to medications
.
Brit J Clin Pharmacol
 
2012
;
73
:
691
705
.
4
Hylek
EM
Skates
SJ
Sheehan
MA
Singer
DE
.
An analysis of the lowest effective intensity of prophylactic anticoagulation for patients with nonrheumatic atrial fibrillation
.
N Engl J Med
 
1996
;
335
:
540
6
.
5
Patel
MR
Mahaffey
KW
Garg
J
Pan
G
Singer
DE
Hacke
W
et al
Rivaroxaban versus warfarin in nonvalvular atrial fibrillation
.
N Engl J Med
 
2011
;
365
:
883
91
.
6
Granger
CB
Alexander
JH
McMurray
JJ
Lopes
RD
Hylek
EM
Hanna
M
et al
Apixaban versus warfarin in patients with atrial fibrillation
.
N Engl J Med
 
2011
;
365
:
981
92
.
7
Kimmel
SE
Chen
Z
Price
M
Parker
CS
Metlay
JP
Christie
JD
et al
The influence of patient adherence on anticoagulation control with warfarin: results from the International Normalized Ratio Adherence and Genetics (IN-RANGE) Study
.
Arch Intern Med
 
2007
;
167
:
229
35
.
8
Vrijens
B
Tousset
E
Rode
R
Bertz
R
Mayer
S
Urquhart
J
.
Successful projection of the time course of drug concentration in plasma during a 1-year period from electronically compiled dosing-time data used as input to individually parameterized pharmacokinetic models
.
J Clin Pharmacol
 
2005
;
45
:
461
7
.
9
Laliberte
F
Cloutier
M
Nelson
WW
Coleman
CI
Pilon
D
Olson
WH
et al
Real-world comparative effectiveness and safety of rivaroxaban and warfarin in nonvalvular atrial fibrillation patients
.
Curr Med Res Opin
 
2014
;
30
:
1317
25
.
10
Demonceau
J
Ruppar
T
Kristanto
P
Hughes
DA
Fargher
E
Kardas
P
et al
Identification and assessment of adherence-enhancing interventions in studies assessing medication adherence through electronically compiled drug dosing histories: a systematic literature review and meta-analysis
.
Drugs
 
2013
;
73
:
545
62
.
11
Blaschke
TF
Osterberg
L
Vrijens
B
Urquhart
J
.
Adherence to medications: insights arising from studies on the unreliable link between prescribed and actual drug dosing histories
.
Annu Rev Pharmacol Toxicol
 
2012
;
52
:
275
301
.
12
Glintborg
B
Hillestrom
PR
Olsen
LH
Dalhoff
KP
Poulsen
HE
.
Are patients reliable when self-reporting medication use? Validation of structured drug interviews and home visits by drug analysis and prescription data in acutely hospitalized patients
.
J Clin Pharmacol
 
2007
;
47
:
1440
9
.
13
Liu
H
Golin
CE
Miller
LG
Hays
RD
Beck
CK
Sanandaji
S
et al
A comparison study of multiple measures of adherence to HIV protease inhibitors
.
Ann Intern Med
 
2001
;
134
:
968
77
.
14
Smith
SR
Wahed
AS
Kelley
SS
Conjeevaram
HS
Robuck
PR
Fried
MW
et al
Assessing the validity of self-reported medication adherence in hepatitis C treatment
.
Ann Pharmacother
 
2007
;
41
:
1116
23
.
15
Pullar
T
Kumar
S
Tindall
H
Feely
M
.
Time to stop counting the tablets?
Clin Pharmacol Ther
 
1989
;
46
:
163
8
.
16
Rubio
A
Cox
C
Weintraub
M
.
Prediction of diltiazem plasma concentration curves from limited measurements using compliance data
.
Clin Pharmacokinet
 
1992
;
22
:
238
46
.
17
STABILITY Investigators
,
White
HD
Held
C
Stewart
R
Tarka
E
Brown
R
et al
Darapladib for preventing ischemic events in stable coronary heart disease
.
N Engl J Med
 
2014
;
370
:
1702
11
.
18
Cramer
JA
Scheyer
RD
Mattson
RH
.
Compliance declines between clinic visits
.
Arch Intern Med
 
1990
;
150
:
1509
10
.
19
Feinstein
AR
.
On white-coat effects and the electronic monitoring of compliance
.
Arch Intern Med
 
1990
;
150
:
1377
8
.
20
Kass
MA
Gordon
M
Meltzer
DW
.
Can ophthalmologists correctly identify patients defaulting from pilocarpine therapy?
Am J Ophthalmol
 
1986
;
101
:
524
30
.
21
Podsadecki
TJ
Vrijens
BC
Tousset
EP
Rode
RA
Hanna
GJ
.
“White coat compliance” limits the reliability of therapeutic drug monitoring in HIV-1-infected patients
.
HIV Clin Trials
 
2008
;
9
:
238
46
.
22
Vrijens
B
Urquhart
J
White
D
.
Electronically monitored dosing histories can be used to develop a medication-taking habit and manage patient adherence
.
Expert Rev Clin Pharmacol
 
2014
;
7
:
633
44
.
23
Arnet
I
Walter
PN
Hersberger
KE
.
Polymedication Electronic Monitoring System (POEMS) – a new technology for measuring adherence
.
Front Pharmacol
 
2013
;
4
:
26
.
24
Chan
AH
Reddel
HK
Apter
A
Eakin
M
Riekert
K
Foster
JM
.
Adherence monitoring and e-health: how clinicians and researchers can use technology to promote inhaler adherence for asthma
.
J Allergy Clin Immunol Pract
 
2013
;
1
:
446
54
.
25
Lugaresi
A
.
RebiSmart (version 1.5) device for multiple sclerosis treatment delivery and adherence
.
Expert Opin Drug Deliv
 
2013
;
10
:
273
83
.
26
Belknap
R
Weis
S
Brookens
A
Au-Yeung
KY
Moon
G
DiCarlo
L
et al
Feasibility of an ingestible sensor-based system for monitoring adherence to tuberculosis therapy
.
PloS One
 
2013
;
8
:
e53373
.
27
Eisenberger
U
Wuthrich
RP
Bock
A
Ambuhl
P
Steiger
J
Intondi
A
et al
Medication adherence assessment: high accuracy of the new Ingestible Sensor System in kidney transplants
.
Transplantation
 
2013
;
96
:
245
50
.
28
Laliberte
F
Nelson
WW
Lefebvre
P
Schein
JR
Rondeau-Leclaire
J
Duh
MS
.
Impact of daily dosing frequency on adherence to chronic medications among nonvalvular atrial fibrillation patients
.
Adv Ther
 
2012
;
29
:
675
90
.
29
Comte
L
Vrijens
B
Tousset
E
Gerard
P
Urquhart
J
.
Estimation of the comparative therapeutic superiority of QD and BID dosing regimens, based on integrated analysis of dosing history data and pharmacokinetics
.
J Pharmacokinet Pharmacodyn
 
2007
;
34
:
549
58
.
30
Blaschke
TF
.
Variable adherence to prescribed dosing regimens for protease inhibitors: scope and outcomes
.
Curr Opin HIV AIDS
 
2008
;
3
:
603
7
.
31
Vrijens
B
Claeys
MJ
Legrand
V
Vandendriessche
E
Van de Werf
F
.
Projected inhibition of platelet aggregation with ticagrelor twice daily vs. clopidogrel once daily based on patient adherence data (the TWICE project)
.
Brit J Clin Pharmacol
 
2014
;
77
:
746
55
.
32
Wallentin
L
Becker
RC
Budaj
A
Cannon
CP
Emanuelsson
H
Held
C
et al
Ticagrelor versus clopidogrel in patients with acute coronary syndromes
.
N Engl J Med
 
2009
;
361
:
1045
57
.
33
Heidbuchel
H
Verhamme
P
Alings
M
Antz
M
Hacke
W
Oldgren
J
et al
European Heart Rhythm Association Practical Guide on the use of new oral anticoagulants in patients with non-valvular atrial fibrillation
.
Europace
 
2013
;
15
:
625
51
.
34
Heidbuchel
H
Verhamme
P
Alings
M
Antz
M
Hacke
W
et al.  
all authors of the EPG
Authors’ response: from monitoring to vigilance about patient adherence to new oral anticoagulants
.
Europace
 
2014
;
16
:
149
50
.
35
Berti
D
Hendriks
JM
Brandes
A
Deaton
C
Crijns
HJ
Camm
AJ
et al
A proposal for interdisciplinary, nurse-coordinated atrial fibrillation expert programmes as a way to structure daily practice
.
Eur Heart J
 
2013
;
34
:
2725
30
.
36
Verhamme
P
Heidbuchel
H
.
Organizing patient-centred anticoagulant therapy
.
Acta Cardiol
 
2013
;
68
:
413
5
.
37
Heidbuchel
H
.
Telemonitoring of implantable cardiac devices: hurdles towards personalised medicine
.
Heart
 
2011
;
97
:
931
9
.
38
Schmidt
S
Sheikzadeh
S
Beil
B
Patten
M
Stettin
J
.
Acceptance of telemonitoring to enhance medication compliance in patients with chronic heart failure
.
Telemed J E-health
 
2008
;
14
:
426
33
.
39
Hendriks
JM
de Wit
R
Crijns
HJ
Vrijhoef
HJ
Prins
MH
Pisters
R
et al
Nurse-led care vs. usual care for patients with atrial fibrillation: results of a randomized trial of integrated chronic care vs. routine clinical care in ambulatory patients with atrial fibrillation
.
Eur Heart J
 
2012
;
33
:
2692
9
.
40
Hendriks
JM
Vrijhoef
HJ
Crijns
HJ
Brunner-La Rocca
HP
.
The effect of a nurse-led integrated chronic care approach on quality of life in patients with atrial fibrillation
.
Europace
 
2014
;
16
:
491
9
.