Measuring, Analyzing, and Managing Drug Adherence in Resistant Hypertension
2013; Lippincott Williams & Wilkins; Volume: 62; Issue: 2 Linguagem: Inglês
10.1161/hypertensionaha.113.00687
ISSN1524-4563
AutoresMichel Burnier, Grégoire Wuerzner, Harry A.J. Struijker-Boudier, John Urquhart,
Tópico(s)Pharmaceutical Economics and Policy
ResumoHomeHypertensionVol. 62, No. 2Measuring, Analyzing, and Managing Drug Adherence in Resistant Hypertension Free AccessResearch ArticlePDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissions ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toFree AccessResearch ArticlePDF/EPUBMeasuring, Analyzing, and Managing Drug Adherence in Resistant Hypertension Michel Burnier, Gregoire Wuerzner, Harry Struijker-Boudier and John Urquhart Michel BurnierMichel Burnier From the Service of Nephrology and Hypertension, University Hospital, Lausanne, Switzerland (M.B., G.W.); Department of Pharmacology, Maastricht University, Maastricht, The Netherlands (H.S.-B.); AARDEX Group, S.A., Sion, Switzerland (J.U.); and Department of Bioengineering and Therapeutic Sciences, UCSF, San Francisco, CA (J.U.). , Gregoire WuerznerGregoire Wuerzner From the Service of Nephrology and Hypertension, University Hospital, Lausanne, Switzerland (M.B., G.W.); Department of Pharmacology, Maastricht University, Maastricht, The Netherlands (H.S.-B.); AARDEX Group, S.A., Sion, Switzerland (J.U.); and Department of Bioengineering and Therapeutic Sciences, UCSF, San Francisco, CA (J.U.). , Harry Struijker-BoudierHarry Struijker-Boudier From the Service of Nephrology and Hypertension, University Hospital, Lausanne, Switzerland (M.B., G.W.); Department of Pharmacology, Maastricht University, Maastricht, The Netherlands (H.S.-B.); AARDEX Group, S.A., Sion, Switzerland (J.U.); and Department of Bioengineering and Therapeutic Sciences, UCSF, San Francisco, CA (J.U.). and John UrquhartJohn Urquhart From the Service of Nephrology and Hypertension, University Hospital, Lausanne, Switzerland (M.B., G.W.); Department of Pharmacology, Maastricht University, Maastricht, The Netherlands (H.S.-B.); AARDEX Group, S.A., Sion, Switzerland (J.U.); and Department of Bioengineering and Therapeutic Sciences, UCSF, San Francisco, CA (J.U.). Originally published10 Jun 2013https://doi.org/10.1161/HYPERTENSIONAHA.113.00687Hypertension. 2013;62:218–225Other version(s) of this articleYou are viewing the most recent version of this article. Previous versions: January 1, 2013: Previous Version 1 IntroductionThe management of hypertension is based on 2 major approaches: a modification of lifestyle and the lifelong prescription of antihypertensive drugs.1,2 All hypertension guidelines recognize that "…the most effective therapy prescribed by the most careful clinician will control hypertension only if the patient is motivated to take the prescribed medication and to establish and maintain a health-promoting lifestyle." This statement clearly emphasizes the importance of supporting drug adherence (ie, adherence to the prescribed dosing regimens for prescribed medications) for patients to gain the maximal benefits of their therapy. Unfortunately, drug adherence in cardiovascular prevention remains low in the population. Thus, in a recent meta-analysis of data on 376 162 patients from 20 studies assessing drug adherence using prescription refill frequency for 7 preventive drug classes (aspirin, statins, and 5 antihypertensive drug classes) prescribed for primary and secondary prevention of cardiovascular diseases, mean adherence over all studies was only 57% after a median of 2 years. Furthermore, mean drug adherence was substantively and statistically significantly lower in primary than in secondary prevention.3The role of drug adherence is particularly relevant in clinical situations in which drug therapies do not provide the expected results. This is typically so in resistant hypertension, commonly defined as the failure to reach blood pressure goals in patients adhering to adequate or maximal doses of an appropriate 3-drug regimen that includes a diuretic. Obviously, this definition of resistant hypertension implies that patients fully adhere to their therapy. This might actually be one of the reasons why the true prevalence of patients with resistant hypertension in the population remains largely ambiguous.4,5 In clinical trials, 10% to 30% of patients are estimated to present with resistant hypertension.6 In the National Health and Nutrition Examination Survey conducted between 2003 and 2008, 8.9% of all US adults with hypertension and 12.8% of the treated hypertensive population met the criteria of resistant hypertension.7 A large Spanish survey has recently examined the clinical characteristics of patients with resistant hypertension and found it to be present in 12% of the treated population, although approximately one third of these patients had a white-coat hypertension during ambulatory blood pressure (BP) monitoring.8 These 2 studies included in their evaluation patients with controlled resistant hypertension (ie, patients controlled on ≥4 medications). The cardiovascular risk profile of resistant hypertensive patients was higher than in nonresistant patients.Interestingly, in none of these surveys was drug nonadherence considered as a potential cause of apparent resistant hypertension. In fact, in the discussion of his paper, the author of the National Health and Nutrition Examination Survey analysis admits that "the medication use questionnaire used in National Health and Nutrition Examination Survey does not distinguish whether medications were used consistently, only that they were used at all during the month before the examination. Some participants who were not fully adherent to their antihypertensive medication, discontinued a medication, or switched from one medication to another within the past month could be falsely classified as resistant."7 This same ambiguity has confounded attempts to estimate adherence in other clinical situations that call for long-term use of prescribed medications.9 Thus, the prevalence of nonadherence in resistant hypertension was very likely much higher than that actually published and was recently confirmed in a clinical study in which toxicological urine assessments were performed without patients' awareness of the test to evaluate drug intake in apparent resistant hypertension. In this study, the majority of resistant patients were either poorly adherent or totally nonadherent to their drug therapies.10The absence of consideration of drug adherence before establishing a diagnosis of resistance to treatment is a serious limitation that may have important clinical and financial implications, especially now that new, surgery-based therapeutic strategies, such as renal denervation or baroreceptor stimulation, are being developed and proposed to treat resistant hypertension.11,12 When dealing with such patients, physicians can evoke numerous reasons for not taking the proposed therapy, but the 2 most commonly encountered reasons are (1) nonadherence, often clinically unrecognized, to rationally prescribed drugs of proven efficacy and reliable bioavailability, and (2) pharmacological nonresponse to rationally prescribed, correctly administered drugs of proven efficacy and reliable bioavailability. The following sections review lines of evidence that support this conclusion.Nonadherence: A Topic Burdened by Poor Methods and Many MisconceptionsOne of the main reasons why drug adherence remains generally ignored, outside of clinical trials and prospective surveys, is the lack of reliable, easy to use, and economical methods to assess drug adherence in clinical practice. Since the 1980s, much effort has gone into devising methods for reliably quantifying ambulatory patients' adherence to prescribed medications, especially those intended for long-term use against various major chronic diseases (eg, hypertension, diabetes mellitus, osteoporosis, glaucoma, lipid disorders, asthma, posttransplant organ rejection, HIV-AIDS, chronic myelogenous leukemia, and others). This work and the methods that support it have recently been reviewed.13 The best of the available methods provide for the reliable capture, storage, analysis, and communication of dosing history data in ways that make it difficult or impossible for patients or trial staff to censor or otherwise manipulate the data. Methods that meet these criteria today include the following: (1) retrospective analysis of prescription refill records, (2) analysis of chemical markers of drug exposure, and (3) automatic electronic time-stamping and compilation of events more or less strongly linked to the act of medication (eg, package opening, dosage form dissolution). Other methods, such as questionnaires, interviews, and periodic counts of patients' returned, untaken doses, are subject to many uncertainties and easy manipulation by patients.13One of the most important results of adherence research in the past 3 decades has been the repeated finding, in a wide variety of clinical circumstances and prognoses, that substantive nonadherence to prescribed drug dosing regimens is much more commonly occurring than what many prescribers and clinical researchers had considered a priori or recognized clinically. This discrepancy has several sources: one of them is that prescribing physicians have been shown to have good ability to recognize good adherence but poor ability to recognize poor adherence unless patients admit that they are not taking their drug. So the net result is that prescribers' judgments about their patients' adherence have been aptly described as no better than a coin toss.14 On their side, clinical trialists continue to report counts of returned, untaken doses (pill counts) from subjects enrolled in trials, despite strong evidence for their unreliability and consistent overestimation of trial patients' adherence to protocol-specified dosing regimens. In the late 1980s, careful work by Pullar et al15 with a well-validated, low-dose chemical marker method showed that counting returned, untaken tablets or other dosage forms grossly overestimates adherence. Overestimation occurs because many patients discard or hoard untaken doses rather than returning them to the trial staff, thus biasing upward the estimate of the patient's adherence because the method is based on the assumption that all unreturned doses have been taken by the patient. That assumption is contradicted by 2 observations: first, what patients return when they have been given substantially more doses than needed for full adherence during the intervals between successive clinic visits and, second, the consistently higher estimates of adherence by returned tablet counts than by electronic monitoring methods.16Analysis of Chemical Markers of Drug ExposureChemical methods certify that drug has been ingested but they are labor intensive and tend to be costly. These methods also have some critical shortcomings: although they unequivocally document the ingestion of doses, they cannot provide information on when doses were taken or omitted. This limitation is important because of the basic pharmacological principle that the size of patients' responses to prescribed drugs is determined not only by the size of doses taken but also by the time intervals between successive doses.16 Much of the work on adherence research has neglected this fact and has, instead, used imprecise, often misleading surrogate measures of patients' exposure to prescribed drugs (eg, the percentage of prescribed doses taken, the proportion of patients who had, during an arbitrary time interval, taken more or less than an arbitrary percentage of prescribed doses). The proper specification of drug exposure should be robust enough to serve as input to appropriately formulated pharmacokinetic models for the ADME characteristics of the drug in question—absorption, distribution, metabolism, excretion, the 4 key parameters of pharmacokinetics—resulting in an accurate projection of the continuous time course of drug concentration in plasma,17–19 which is verifiable by intermittent, direct chemical measurements of the concentrations in plasma of the drugs in question.Drug measurements are also affected by the white-coat adherence phenomenon whereby patients tend to improve their adherence before and after clinical visits.20 White-coat adherence was identified in the first years of electronic monitoring of ambulatory patients' drug intake.20,21 In 1990, Cramer et al21 showed that the analogous white-coat dosing behavior occurred in seizure-prone patients prescribed antiepileptic drugs: those with frequent erratic dosing histories tended to revert to punctual oral dosing in the several days before a clinic visit. It is a phenomenon that has been repeatedly observed in other therapeutic fields, including in hypertensive patients (Figure 1). Its practical effect is to mask usually poor adherence by creating a false clinical impression that good adherence is maintained over long periods of time. This misimpression is reinforced by the fact that many drugs have pharmacokinetics that can bring the concentration of drug (or active metabolite) from zero up into the usual therapeutic range by the prior ingestion of only 1 or 2 doses of drug. Thus, patients who take 1 or 2 doses in a day or 2 before a scheduled clinical or laboratory visit can readily create false impressions of good adherence, provided that the pharmacokinetics and pharmacodynamics of the drug in question allow fast responses to fast changing dosing histories.Download figureDownload PowerPointFigure 1. An example of white-coat adherence in hypertensive patients receiving once-daily diuretic therapy. Note the progressive decline in drug adherence over weeks and the sudden improvement in drug adherence during the 3 days preceding the medical visit.Concentrations in plasma of antihypertensive drugs actually failed to find broad clinical use in the heyday of therapeutic drug monitoring, which occurred in the latter 1970s. Part of the problem has been the limitations of single-point assays for the concentration of drugs in plasma, many of which have large peak-to-trough swings within the intervals between properly timed, sequential doses. These big fluctuations preclude simple, intuitive recognition of concentration–effect relationships. The resulting lack of evident clinical explanatory power of single-point values of most antihypertensive drugs and the fact that not all antihypertensive drugs were detectable in urine or plasma closed that chapter in hypertension research.As mentioned previously, a recent study proposed single-point measurements of drug concentration in urine samples in patients with resistant hypertension.10 The authors found that a large proportion of samples had very low or zero concentrations of drug in this population. The problem posed by this approach is that, for ethical reasons, such measurements need to be done with informed consent of patients and thus tend to stimulate white-coat adherence. In overview, the use of spot checks of urinary drug or drug metabolite concentration requires one to reconcile the needs for (1) more than a single sample; (2) disclosure of the testing to patients; and (3) avoiding the upward bias created by white-coat adherence. It is doubtful that these needs can be simultaneously met well enough to result in a reliable test.The crucial variable in the measurement of patient adherence is the patient's drug dosing history, from which it is possible to project, noninvasively, the continuous time course of drug concentration in plasma. The accuracy of these projections is subject to verification by occasional direct measurements.17–19 But for the occasional direct sampling of blood for validation purposes, these projections are noninvasive. Temporary bursts of white-coat adherence or other transient biases may occur, but are set in the context of the patient's longer-run dosing history. Thus, the patient's exposure to the drug in question is not dominated by temporary aberrations in the long-running time series of doses taken and the complete time course, over days or weeks of drug dosing, of the drug's concentration in plasma. The capability of making these data-rich projections noninvasively puts the clinical correlates of patient adherence and pharmacokinetics into a long-term perspective and is a practical therapeutic consequence of the combined use of electronic medication event monitoring and modern pharmacometric methods of model-based analysis. Thus, ideally, drug measurements should be combined with a precise dosing history data enabling to interpret drug concentrations in relation to the dosing interval.Electronic Monitoring Methods of Assessing AdherenceBecause of their automaticity and precision of timing when patients take or omit doses, electronic monitoring methods have opened a new era in understanding the sequences of drug exposure and clinical events during prescribed pharmacotherapy for leading chronic diseases, including hypertension.22 Electronic monitoring methods have followed several approaches, starting with eye drop dispensing in the treatment of sight-threatening glaucoma and ocular hypertension23,24 and then into orally administered tablets or capsules. Two approaches to electronic monitoring of orally administered drugs have been developed: medication event monitoring system (MEMS; AARDEX Group, Ltd, Sion, Switzerland), which time stamps and stores times/dates of each opening of the drug package13,16 and the recent innovation by Proteus Digital Health Inc (Redwood City, CA), which incorporates into each dose of drug a specially designed microchip that, on ingestion and solubilization, emits a brief, low-strength, radio signal that is detected, amplified, and transmitted onward by an adhesive patch worn by the patient.25 The signal then passes, via a Bluetooth link, to a nearby wireless phone that communicates the dosing time data to a computer, which can be situated wherever there is a phone link. Error rates with both types of electronic monitoring have been reported to be ≈3%.13,25Of the foregoing methods, MEMS monitoring has had the most extensive use, having been marketed as a scientific product for use in drug trials since 1988, with >570 completed clinical research studies based on uses of MEMS monitoring and published in peer-reviewed journals. The references to these publications are accessible at www.iadherence.org. MEMS monitoring is based on an indirect measurement—packaging opening time—but this method has been extensively validated in the past 3 decades.17–19 Several studies have been conducted in the field of hypertension using the MEMS monitoring system, including in resistant hypertension.22,26–28What Lessons Have We Learned About Patient Adherence, its Detection, and its Analysis?The greatest catalysts for adherence research have been the succession of stark discrepancies between advances in pharmacological power and shortcomings in the clinical realization of therapeutic benefit from powerful drugs as observed today in resistant hypertension. During the last half-century, islands of special awareness of the importance of patient adherence have formed, as new and unprecedented, dose- and time-dependent actions, and therapeutic power was seen to be thwarted by patient nonadherence (eg, antibiotic and chemotherapeutic treatment of tuberculosis29; oral steroidal contraception30; antiretroviral treatment of HIV/AIDS31; immunosuppressant-based management of recipients of transplanted organs32 and of course treatment of hypertension).33,34The most recent studies on drug adherence have taught us several important clinical lessons. The most fundamental point is that adherence is not a therapeutic parameter that can be described by a single number, as usually reported in the literature. Adherence is essentially a dynamic process, with sometimes slow-to-change effects on drug actions of variable exposure to prescribed drugs. A major step toward defining the best metrics for properly quantified drug dosing histories is the recent publication of the ABC Taxonomy for Describing and Defining Patient Adherence to Medications.13 In this context, it became evident that setting arbitrary cutoffs, such as 80%, are of little clinical interest for many reasons. One of them is that an adherence of 80% can be achieved in many different ways, each with a very different clinical impact (eg, 1 missed dose every 5 days or 1 missed week of doses every 5 weeks). Furthermore, no one really knows what level of adherence is sufficient to obtain the full benefit of a drug because drugs were rarely investigated in this respect. Thus, depending on the pharmacological characteristics of the prescribed drug, 80% of prescribed doses taken may be sufficient or not for full therapeutic benefit. Thus, in resistant hypertension, we reported an increase in BP when drug adherence was <90% of prescribed doses being taken.26 Short-term drug omissions may have less impact on clinical parameters and cardiovascular end points, with drug having a long duration of action.35 This observation was the base of the concept of forgiving drugs.36 At this point, it is also important to point out that the time of discontinuation of dosing, as measured by prescription refill intervals, is highly uncertain, because it takes 2 sequentially missed refills to support the conclusion that the patient has stopped taking the medicine. Because refill intervals are typically 2 to 3 months, this means 4 to 6 months of uncertainty as to when the patient quit taking the medicine.The electronic monitoring of drug adherence provides a unique description of the temporal patterns of drug intake and behavior in the population, including in the hypertensive population. This point is illustrated in a recent analysis of 4783 patients who participated in 1 of 21 different phase IV studies involving 43 different once-daily antihypertensive drugs, mostly of relatively recent origin, including recipients of angiotensin II receptor blockers (n=2088), calcium channel blockers (n=937), angiotensin-converting enzyme inhibitors (n=665), β-adrenergic receptor inhibitors (n=195), and diuretics (n=155).37 The study encompassed 478 630 days of dosing history data. These data were gathered from a series of organized phase IV clinical studies, not from routine care.Three types of deviation from prescribed dosing instructions are shown in Figure 2, which extended the results of the above-mentioned analysis in hypertension to 36 907 patients prescribed oral medications for one of a variety of medical conditions in 95 studies16: (1) noninitiation; (2) short persistence; or (3) nonexecution of the dosing regimen.Download figureDownload PowerPointFigure 2. Kaplan–Meier plots of the time course of adherence parameters of 36 907 patients prescribed oral medications for one of a variety of medical conditions in 95 studies during the first year of electronic compilation of the patients' dosing histories. The horizontal dashed line at the top illustrates how perfect adherence of all patients would be depicted. The dark blue line shows the percentage of patients still engaged with their dosing regimen as time passed after the start of treatment. The abrupt drop in the dark blue line at zero time reflects noninitiation of treatment by ≈4% of the patients. The subsequent decline of the dark blue line arises from patients' permanent discontinuation of their dosing, as they ended their persistence with the prescribed once-daily dosing regimen. The irregular red line shows the percentage of patients who were (1) still engaged with the dosing regimen and (2) who dosed correctly on each day of the observation. The slight wobble in the red line arises from day-to-day variation in the proportion of patients who dosed correctly on each day. The pink area between the dark blue line and the irregular red line indicates the shortfall in drug intake that arose from intermittently omitted doses. The light blue area between the horizontal dashed line and the dark blue line indicates the shortfall in drug intake arising from the combined influences of (1) noninitiation of dosing by ≈4% of the treated patients and (2) early, complete discontinuation of dosing, also known as short persistence, by ≈35% of the patients. Data from the iAdherence database (www.iadherence.org). Adapted with permission from Blaschke et al.16Noninitiation is indicated by an abrupt decline of ≈4% in the dark blue line; it is small in this set of clinical circumstances, but there are other clinical situations in which noninitiation can be much larger. In the context of this review, the main point is that noninitiation is a situation-specific part of the overall adherence story. Noninitiation needs to be measured and included in the overall analysis of adherence within every study involving ambulatory patients who have the responsibility for medication-taking.Short persistence is indicated by the gradual but inexorable downward movement of the dark blue line, falling more rapidly during the first 90 days than later, reaching the 50% point at 330 days after the start of treatment. These patients ceased their engagement with the dosing regimen on their own initiative—an act that is inherently willful, not arising from forgetfulness.Lapses in implementation (or execution) of the once-daily dosing regimen gave rise to the irregular red dashed line, which reflects the fact that, on any given day, ≈8% to 10% of patients omitted that particular day's scheduled once-daily dose, although the patients could be seen to have remained still engaged with the dosing regimen, in that they returned to dosing after ≥1 days of omitted doses. Lapses in implementation (or execution) are typically a consequence of forgetfulness or negligence: most of such errors involve a single day's dose, but some are part of a multiday sequence of omitted doses, thus giving rise to exceptionally long intervals between sequential doses, so-called drug holidays, which are an important aspect of patient nonadherence. The frequency of occurrence of lapses in dosing has been analyzed recently.37 From this analysis, it seems that longer lapses occur less frequently than shorter lapses.Figure 3 shows the similarity of 4 studies of patients' persistence with antihypertensive drugs in routine practice.38–41 There is remarkable similarity between the results of these 4 pharmaco-epidemiological studies of in-practice persistence with antihypertensive drugs, but there is also remarkable similarity of these pharmaco-epidemiological results to the aggregated results of the 21 clinical trials from Figure 2. There is no evidence here for a substantive difference between in-practice and in-trial adherence.Download figureDownload PowerPointFigure 3. The red line in this figure is the same as the dark blue line in Figure 2, showing the time course of early discontinuation of (short persistence with) the prescribed once-daily dosing regimen for each of the 43 once-daily antihypertensive drugs tested in the 21 phase IV studies summarized in Figure 2 and based on data generated by continuous electronic monitoring. The individual points in, respectively, yellow,33 blue,34 green,35 and black36 show persistence reported by 4 different trials in which adherence to once-daily antihypertensive drugs was studied by means of prescription refill records. The ubiquity of short persistence is indicated by both methods. PKC indicates pharmacokinetics.Drug Adherence and BP ControlAs shown above, approximately half of the patients who start treatment for hypertension abandon treatment during the first year of treatment. Other studies show that nonpersistence continues to reduce the number of patients still engaged with antihypertensive drug dosing regimens out to ≥5 years after the onset of treatment, by which time only 10% to 15% of the originally treated patients are still engaged with the regimen. These strikingly high rates of abandonment of drug treatment help explain why a substantial majority of patients with resistant hypertension turn out, after careful evaluation, to be nonadherers, with pharmacological nonresponders being only a minority of patients. Some propose to use the term pseudoresistance for the nonadherers, but, given that nonadherers constitute the vast majority of patients deemed resistant, they deserve a straightforwardly descriptive term instead of a euphemism. Thus, the identification of resistant hypertension is a unique opportunity to reconsider drug adherence and to investigate this issue adequately.Of note, a relationship between the percentage of drug adherence and the level of BP control has been difficult to demonstrate in hypertension.42 Several reasons can explain why it has been so difficult to establish such a relationship. First, as soon as the attention is focused on drug adherence, the latter increases substantially in all groups, including controls. Hence the difference between the intervention and the control is reduced, leading to the absence of statistically significant differences in BP, as differences between treated and control patients narrow and the variances in the 2 sets of data increase. It is also possible that patients who accept to participate in adherence studies have a higher drug adherence than those who refuse to be enrolled (selection bias). And finally, any measure of adherence per se increases the patient's adherence. In fact, the latter observation has been used clinically to determine the true efficacy of a prescribed drug regimen.26,43 We observed that the monitoring of drug adherence during a period of 2 months, without changing the prescribed treatment, enables to assess how much BP decreases when the therapy is taken adequately. This approach provides an opportunity to interfere with the real problem, that is, with drug adherence if the patient does not take his/her medications or with the drug prescription if adherence is good but BP remains uncontrolled.26 In our study, about a third of patients with resistant hyperten
Referência(s)