Pharmionics in Overview

| Home | | Pharmacovigilance |

Chapter: Pharmacovigilance: Introduction to Pharmionics

This topic, if one takes a broad view, is one of many aspects of pharmacotherapeutics that was largely neglected until relatively recently.


This topic, if one takes a broad view, is one of many aspects of pharmacotherapeutics that was largely neglected until relatively recently. A major reason for neglect of patient adherence was the poor state of available methods for compiling drug dosing histories in ambulatory patients. Sometimes called ‘external drug exposure’, reliable drug dosing histories are the cornerstone of understanding how prescribed drugs are actually being used by ambulatory patients. That understanding, in turn, is the foundation for under-standing the clinical and economic consequences of observed patterns of drug usage/misusage. Thus, the qualities of methods for compiling drug dosing histo-ries of ambulatory patients are a natural topic of this chapter. So too are the methods of analysing the clini-cal and economic consequences of variable adherence to prescribed drug dosing regimens.


Three basic patterns characterize the main devia-tions from prescribed drug dosing regimens. Some patients – usually in the range of 5%–10%, but sometimes more or sometimes less – never start the prescribed course of drug dosing. This pattern is called ‘nonacceptance’. It is shown by the abrupt drop at time zero in the percentage of patients engaged with the drug dosing regimen, the line labelled ‘persis-tence’ in Figure 48.1. These are patients who never start the dosing regimen, though have enrolled in the treatment programme. They may take an initial dose or two, but most of them take none, and then disap-pear from the treatment programme. There may be a time that they come back to treatment, but it does not fall within the duration of the study or treatment plan in question.

Once the patient engages with the drug dosing regi-men, there is an ongoing question of the quality of the patient’s execution of that regimen. The main errors that patients make in execution are to delay or omit doses. Sometimes they sequentially omit multi-ple scheduled doses, which are called ‘drug holidays’ when they exceed 3 days’ duration. Occasionally, some patients take an extra dose, but missed doses generally outnumber extra doses by 4:1 or more. On any given day, within a group of patients still engaged with the dosing regimen, about 10% of prescribed doses are not taken, giving rise to the gap, seen in Figure 48.1, between the ‘persistence’ line and the lower, somewhat irregular line, labelled ‘adher-ence/compliance’ – the irregularities being due to day-to-day variations in the proportion of prescribed doses that are missed. Within that gap, of course, lie some important details, the first of which is that most of the gap arises from dose omissions made by about a third of ambulatory patients (Urquhart, 1997), and of course includes drug holidays, most of which are taken by a small minority of patients, although within 6 months about half of patients monitored in the stud-ies that comprise Figure 48.1 had had at least one holiday. The third major deviation from prescribed drug dosing regimens is early cessation of dosing, such that dosing stops, and remains stopped without resumption within the time frame of the study or clinical situation.

Figure 48.1 illustrates the foregoing points. Follow-ing the immediate drop due to non-acceptors, we are left with patients who engage with the dosing regi-men. They dwindle in numbers throughout the one-year period shown in Figure 48.1. By the end of the first year, in the 15 214-patient cohort represented by Figure 48.1, about a third had discontinued what was meant to be multi-year, if not lifetime, treat-ment. Note the large gap between the ‘persistence’ line and the ‘perfect adherence’ line. This gap, which grows with time, indicates both the loss of patients from beneficial treatment, with its implications for public health, and the loss of sales revenues for the drug developer/manufacturer/marketer. When one sees year-by-year growth in revenues from a pharma-ceutical indicated for long-term use, it signifies that the product’s marketing effort must not only recruitreplacements for the non-persisters, but also recruit additional patients. That process of intensive recruit-ment of new patients continues year after year. Some analysts refer to this costly and inherently wasteful process as ‘churn’, the high costs of which could be reduced if the gap between actual and perfect persis-tence could be narrowed.

One can expect to see variation within the above numbers, from one treatment situation to another, but the basic patterns of non-acceptance, incomplete execution and early discontinuation are pervasive in long-term ambulatory pharmacotherapy. To illustrate one end of the range of variation, Catalan and LeLo-rier studied the persistence of Canadian patients with prescribed drugs of the statin category, following the patients for 5 years after they were prescribed a statin. Each patient’s drugs were fully reimbursed, which means that economic obstacles to continuity of treat-ment were nullified. Switches between one drug and another within the ‘statin’ class were considered to represent continuity of statin treatment. Following are the percentages of patients still persisting from the first to the fifth anniversary of the original prescrip-tion: 33, 24, 17, 14, 13. This pattern shows twice the loss of patients within the first year as shown in Figure 48.1. Perhaps the reasons for this exceptionally high rate of discontinuation in the Catalan–LeLorier study lie in the fact that the patients in this study were on full social assistance, which means that they were eligible for economic support by the state, in addition to getting prescription drugs at no cost. The various problems that led these patients to qualify for full social assistance may include factors that espe-cially discourage long persistence with chronic-use medicines for asymptomatic conditions.

To illustrate the other end of the range of variation, the big confirmatory trials of several major drugs of the statin class show that over 90% of patients enrolled in the studies were continuing to attend scheduled clinic visits, and presumably were still taking the trial medication at some level of adherence/compliance (Scandinavian Simvastatin Survival Study Group, 1994; Shepherd et al., 1995). That level of adher-ence/compliance could only be crudely indicated from these trials’ reliance on returned tablet/capsule counts as estimators of patients’ exposure to the test drugs, for reasons discussed later. It remains to be seen how many patients in big clinical trials continue to keep scheduled appointments but surreptitiously discon-tinue dosing, or take too few doses to have more than de minimus clinical effects. Suffice it to say, though, that these confirmatory trials certainly demonstrate that it is possible to maintain nominal persistence with trial medications at a very high level.

It seems reasonable to infer that the administra-tive apparatus of big clinical trials – the process of securing informed consent, multiple phone calls from trial staff to patients, other reminders, all adding up to more than usual professional attention paid to patients – serve to keep the vast majority of patients engaged with the treatment process over long periods of time, with an evident > 90% persistence through year 5 – a stark contrast with the much lower persis-tence observed in studies carried out on routine medi-cal practice (Jones et al., 1995; Catalan and LeLorier, 2000; Benner et al., 2002).


The pharmionics field is just at the beginning of systematic work along these lines, with as yet few published studies, and even fewer studies of satis-factory design and analysis. The best in this cate-gory is the recently published, 392-patient, one-year study (Vrijens et al., 2005a), which has shown that community pharmacies, cluster-randomized between practice-as-usual and measurement-guided interven-tion, could use electronically compiled drug dosing histories to guide their interventional discussions with the patients, and achieve a statistically signif-icant improvement in both persistence and compli-ance with the daily dosing regimen of atorvastatin, a leading drug in the statin class. This result clearly needs to be repeated, and to benefit from knowl-edge of, and avoidance of, problems that lurked beneath the surface of this study. For example, the interventional programme was designed by commit-tee, several members of which were adamant that the provision of a credit card-sized beeper would suffice to remind patients when to take the once-daily dose; in the event, however, only 22% of patients accepted the beeper card, and half of those rapidly discontinued its use – a phenomenon well known in the consumer electronics arena as ‘beeper-fatigue’. Another limitation was that each pharmacist in the intervention group was allowed to improvise his/her interventional manoeuvres.

Despite these problems, however, the study showed clear-cut benefits of measurement-guided medication management, as improvised on intuitive grounds by community pharmacists. The results of this study are probably best seen as a starting point for learning-curve-based improvements in results, combined with simplifications in method and corre-sponding economies.


The foregoing discussion makes clear the three major categories of deviation from a prescribed drug dosing regimen in ambulatory care: acceptance, execution and discontinuation. The time between the first-taken and the last-taken dose is called ‘persistence’, expressed in units of time. The quality of regimen execution is the outcome of a comparison between the patient’s dosing history and the prescribed drug dosing regimen – the outcome of the comparison of two time-series. As there are many facets to time-series data, there is no single parameter that captures all facets, so there are a number of ways to express the data.

Many investigators have used only aggregate expressions such as the percentage of prescribed doses taken, the percentage of days on which the correct number of doses was taken, or the percentage of interdose intervals that fall within certain limits of the interval implicit in the prescription, for exam-ple 24 hours for once-daily dosing. Aggregate figures across long periods of time hide informative time-variations in dose-taking behaviour. For example, there is a marked ‘weekend effect’ frequently evident, by which substantially and significantly more doses are missed on weekends than on weekdays. Another time-dependency is the tendency for the quality of regimen execution to decline gradually over long peri-ods of time.

The choice of limits on the dosing interval should ideally relate to the pharmacometric properties of the drug in question, for example bendroflumethazide, the diuretic widely used in the United Kingdom for hyper-tension treatment, has a 3-hour plasma half-life (Jackson, 1995), but a 6.3-day duration of anti-hypertensive action after a last-taken dose; if one considers only the pharmacokinetic properties of that drug, the range would be set quite narrowly, perhaps ±1 5 hours, but given that the pharmacodynamic properties of the drug dominate, and confer a 6.3-day duration of action (Girvin and Johnston, 2004), one could reasonably accept a range of ±2 days.

In the known pharmacometric properties of bendroflumethazide, one gets a glimpse of how the search for a sound quantitative answer to the question ‘how much adherence is enough’ represents a chal-lenge to pharmacometric understanding of drugs and the dose- and time-dependencies of their actions. It also emphasizes the importance of examining not only pharmacokinetic information about the drug in ques-tion, but also pharmacodynamic information, particu-larly the duration of drug action(s) after a last-taken dose. Either can be the determining factor in judging ‘how much adherence is enough’, which of course is a crucial but neglected aspect of determining an optimal drug dosing regimen. The ‘neglect’ arises probably in large part from the prevailing delusion that achieving a once-daily dosing regimen for a product will auto-matically solve adherence problems. The case studies presented later serve to disabuse anyone of that naive notion.

Contrasting Dynamics of Acceptance, Execution, Discontinuation: why no Single Parameter can Encompass all Major Dosing Errors and Support Sound Quantitative Analysis

Acceptance and discontinuation are more or less binary occurrences, in that they are usually abrupt. Execution, in contrast, is a continuous process that can vary within days, between days, from week to week, or from month to month, and indeed does so, as noted above. It is not possible to combine binary and continuous processes in one parameter, except in a literary sense, but certainly not in the sense of having one parameter that supports sound, quantitative analysis.

‘Adherence’ is generally used as a blanket term for all aspects of how well or poorly a prescribed drug dosing regimen is followed by patients. As a literary expression, it serves a certain purpose, but it does not support sound measurement, which must distinguish between non-acceptance, poor execution and early discontinuation. As a concrete example, consider the following statement by the 6th Joint National Commission on High Blood Pressure (The Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure, 1997): ‘Poor adherence to anti-hypertensive therapy remains a major therapeu-tic challenge, contributing to the lack of adequate control in more than two-thirds of patients with hypertension.’ The problem with this statement is that it does not distinguish between non-acceptance, poor execution or early discontinuation. In all likelihood each plays some role in the overall problem. Based on Figure 48.1, which includes a considerable amount of data from studies of hypertensive patients, early discontinuation is almost certainly the biggest contributor to the distressing shortfall in the quality of anti-hypertensive drug treatment. As the Belgian atorvastatin study illus-trates, a programme of measurement-guided medi-cation management can not only prolong patients’ engagement with the drug dosing regimen, other-wise known as extended persistence, but if we are to do better than the results in that study, it prob-ably means attacking specifically each of the three major errors: non-acceptance, poor execution, early discontinuation.

The Importance of Distinguishing Early Discontinuation and Poor Execution in Analysing Drug Dosing History Data

A common manner of expressing adherence/ compliance data goes as follows:

Rates of adherence for individual patients are usually reported as the percentage of the prescribed doses of the medication actually taken by the patient over a specified period The average rates of adherence in clinical trials can be remarkably high, owing to the attention study patients receive and to selection of the patients, yet even clinical trials report average adher-ence rates of only 43 to 78 percent among patients receiving treatment for chronic conditions (Osterberg and Blaschke, 2005).

Expressing the percentage of prescribed doses taken during a fixed interval of time inevitably mixes together execution and early discontinuation. Thus, a patient will be categorized as having 50% adher-ence who doses strictly punctually but discontinues at month 6 in a 12-month study. Of course, if dura-tion of the study had been set at 24 months, then the patient who discontinues at 6 months would be cate-gorized as a 25% adherer. Also categorized as a 50% adherer will be a patient whose execution is such that he takes only half the prescribed doses, but continues to be engaged with the dosing regimen throughout the 12-month study. These two contrasting patterns of dosing, both of which are common occurrences, not exotic oddities, call for very different interventional approaches: targeted motivation in the first patient to abort his intention to quit, versus a careful review with the patient of his day-to-day dosing patterns, with assistance in finding robust routines in his daily life to which his daily dosing can be linked, as suggested by Cramer and Rosenheck (Cramer and Rosenheck, 1999). Then ongoing follow-up is needed in the latter instance to see how well specific suggestions work and to provide changes and/or motivation, as needed, to maintain high quality of execution. Ongoing obser-vation of daily dosing patterns may, if the quality of execution starts to dwindle, signal a pending episode of discontinuation.

There are two important points in the foregoing. One is that the improvement of poor quality of execu-tion is self-evidently a more difficult management problem than is the postponement of discontinuation to achieve longer persistence. The second point is that it is a fundamental mistake in the analysis of dosing history data to ignore the distinction between poor execution and short persistence. ‘Execution’ self-evidently relates to what happens while the patient is engaged with the dosing regimen; when that engage-ment halts, execution is finished.

One might argue that, from a practical point of view, taking half the prescribed doses is the same, whether it occurs by ongoing faulty execution or by early discon-tinuation of correct execution. The counter-arguments are as follows. First, since there appears to be a major difference in the complexity and cost of interven-ing to improve execution vs. intervening to prolong persistence, we only engage with intervention when we know which we are trying to fix. Second, life goes on past the end of an arbitrarily defined study period, so that the patient who has quit taking the medicine will, unless re-recruited, generate no revenues for the manufacturer/developer/marketer, whereas the faulty executor will, for as long as he persists, continue to generate revenues, albeit at a rate reduced by the extent of his ongoing underdosing. Third, the percent-age of prescribed doses taken by the short persis-ter varies with the duration of the study, as noted above; in contrast, the percentage of prescribed doses taken by a consistently poor executer is unchanged by altering the duration of the study, setting aside the tendency for the quality of execution to decline gradually with time since the start of treatment.

Figure 48.1 provides the best format for express-ing the basic findings from analysis of drug dosing histories in groups of patients. One can and should go further to characterize the occurrence of omitted doses and drug holidays on a patient-by-patient basis. The clinical correlates of substantial underdosing should be examined carefully, as they may, among other things, show whether the recommended drug dosing regimen provides either insufficient or a substantial excess of ‘forgiveness’, which is defined as the post-dose duration of the drug’s therapeutically effective action(s) minus the recommended drug dosing interval (Urquhart, 1997).

Note that Figure 48.1 is a very simple, straight-forward summary of pharmionic data. Complexity in this field arises at the level of individual pharma-ceuticals, because each has its own recommended dosing regimen and pharmacokinetic and pharmaco-dynamic properties. The clinical and economic conse-quences of early discontinuation, dose omissions, and drug holidays will depend directly on these product-specific properties. They are indeed more than drug-specific, because differences in drug formulation can not only prolong drug entry into the bloodstream, but also alter its rate in sometimes clinically impor-tant ways – a key example being how the pharmaco-dynamics of nifedipine were beneficially altered by its re-formulation in an oral, osmotic pump dosage form that releases the drug at a constant rate, versus the rapid highly time-varying release profile asso-ciated with the original dosage form (Breimer and Urquhart, 1993). Thus, the main complexities in this field arise from the fact that each of hundreds of phar-maceutical products can be expected to have different answers to the question of the clinical and economic consequences of commonly occurring dosing errors.


The usually abrupt cessation and resumption of dosing that characterizes drug holidays provide an opportu-nity to search for important clinical correlates that may contribute to the understanding of adverse reac-tions occasioned by either rebound effects, as dosing stops, or recurrent first-dose effects when post-holiday dosing resumes in patients who should be re-titrated after some period of interrupted dosing, as was done prior to the initial start of treatment. One of the missing elements of pharmacodynamic information about drugs with first-dose effects is the length of time, after dosing stops, needed to restore drug naiveté and the need for re-titration for least-hazardous resumption of dosing post-holiday. Such information would inform the answering of reasonable questions about the role of drug holidays and their potential hazards in trials of drugs like, for example, encainide and flecainide, which have hazardous or even lethal pro-arrhythmic effects that are triggered by unduly high rates of dose-escalation in the drug-naïve state. By the same token, the role of drug holidays remains unclear in the case of peripheral vasodilators that can have hazardous hypotensive episodes or reflex tachycardia when the rate of dose-escalation is too high, or full-strength dosing resumes abruptly in the drug-naïve state.

While the various patterns and extents of underdosing seen in patients’ dosing histories are, in a strict sense, observational data, their clinical correlates may send up useful ‘red flags’, tentatively identifying, for example, dosing regimens that are set too high (Cross et al., 2002; Heerdink, Urquhart and Leufkens, 2002), hazardous rebound effects (Urquhart, 1997) and recur-rent first-dose effects (Urquhart, 1997). Clinical corre-lates of a single holiday would naturally be difficult to interpret, but if holidays recur, as they do in some patients, one has the potential opportunity to see repe-tition of holidays and their associated events. Repeti-tion and consistent time-sequence greatly strengthen the inference of causality. A common problem, of course, is that most clinical events cannot be measured continuously, and are only intermittently sampled, which, via white-coat compliance (Feinstein, 1990), is likely to prevent the occurrence of holidays in temporal proximity to the sampled clinical events. In contrast, holidays can be captured by means of automatic, continuous electronic compilation of drug dosing histories.

A noteworthy technical advance is the ability of the latest generation of implanted cardiac pacemakers and defibrillators to automatically compile complete records of electrophysiological activity through-out multi-week intervals between data-downloads. That capability, combined with the prevalence of pro-arrhythmic effects among leading cardiac anti-arrhythmic drugs, provides a potentially rich area for enlightening research on the pharmacodynamics of the anti-arrhythmic drugs.

Contact Us, Privacy Policy, Terms and Compliant, DMCA Policy and Compliant

TH 2019 - 2022; Developed by Therithal info.