Methodological Issues in Compiling Drug Dosing Histories of Ambulatory Patients

METHODOLOGICAL ISSUES IN COMPILING DRUG DOSING HISTORIES OF AMBULATORY PATIENTS

Until the later 1980s, the available methods (clinical judgment, interviews, patient diaries, counts of returned, unused dosage forms, spot checks of drug concentration in plasma) were unreliable and biased by the ease with which patients can and do easily censor evidence for omitted doses. Thus, a brief review of methods for gathering drug dosing histories in ambulatory patients is a logical part of this chapter.

CLINICAL JUDGMENT

A leading reason for the weakness of clinical judg-ment about patient adherence is that the doctor– patient relation is based on trust, which, when there is no reliable, contradictory source of information, leads most physicians to take at face value what patients tell them about their adherence to the prescribed drug dosing regimen. The result is strongly biased towards over-estimation of the patient’s adher-ence to the prescribed dosing regimen. What patients tell their physician or other health care personnel is strongly coloured by two factors: (a) recall of day-by-day drug intake is often poor, unless the patient goes to extraordinary but infrequently made efforts to keep records of what was taken and when; (b) there is a pervasive reluctance among patients to inform the prescribing physician that they have never started taking the medicine, or have started it but executed the dosing regimen poorly, or have completely discontin-ued taking the medicine much sooner than the physi-cian had prescribed.

INTERVIEWS AND PATIENT DIARIES

Reliability problems are obvious with interviews and diary entries, because patients can say or write what-ever they choose, and whenever they chose to make diary entries. Recent work with a special diary that captured and stored the time of each diary entry has shown that only 11% of diary entries bore a credible temporal relation to the event being entered into the diary (Stone, Schiffman, Schwartz, 2002).

MEASURED DRUG CONCENTRATIONS IN PLASMA

An often-misunderstood method is the direct measure-ment of drug concentration in plasma. This method, which has an aura of ultimate objectivity, runs head-long into the prevalent bias called ‘white-coat compli-ance’ (Feinstein, 1990). This phenomenon occurs in patients whose adherence is poor most of the time, but shifts suddenly to correctness during the day or two prior to a scheduled visit to the physician or other caregivers. With 1–2 days of correct dosing having preceded the sampling of drug concentration, the measured value of drug concentration will, with the vast majority of conventionally formulated drugs, reflect drug intake for only 1–2 days. What happened before that brief period of time, or what happens after-wards, is unknown. The source of these numbers is pharmacokinetic theory, which teaches that measured drug concentration in plasma at a given time will reflect drug intake during a period of prior time equal to 3–4 times the drug’s plasma half-life. It turns out that 87% of the several hundred most commonly used drugs have plasma half-lives of 12 hours or less (Benet, Oie and Schwartz, 1995). Thus, the measure-ment of drug concentration in plasma, as done in the usual way, with blood sampling done at the time of a scheduled visit, will, in most instances, reflect drug intake only during the period of white-coat compli-ance. If a measured drug concentration is zero, it signifies that no drug was taken during four prior half-lives, that is 2 days or less for the vast majority of drugs.

PILL COUNTS

With dosage form counts (‘pill counts’), many patients can and do discard untaken dosage forms before returning the drug package to the clinical staff – a prob-lem clearly identified in two studies reported about 15 years ago (Pullar et al., 1989; Rudd et al., 1989). Since then, numerous studies have compared electronically compiled dosing histories with results of pill counts, with uniform demonstration of exaggerated results with pill counts (Urquhart, 1997). Yet, to their shame, clini-cal researchers continue to perform counts of returned, untaken dosage forms, and solemnly report the results as if the method had not been thoroughly discredited, except in the infrequent instance when a patient returns all dispensed medicine untaken.

HOW PILL COUNTS ARE (MIS)INTERPRETED

The usual result of pill-count data is that somewhere in the low 90% range of trial patients were satisfac-torily compliant. ‘Satisfactorily’ usually means that the patient has returned few enough dosage forms to support the conclusion that >80% of prescribed doses were taken. This almost universally applied, ‘>80% is OK’ criterion has no roots in pharmacological science, and is supported only by uncritical repetition, having started in the 1970s as a self-evident guess (Sackett and Haynes, 1976). Yet, as already noted, such infor-mation is not only drug-specific but product-specific.

The inherently unscientific folly of universally applying the ‘80% is OK’ criterion is revealed by the well-documented fact that the pharmaceutical prod-ucts with which there is the greatest use-experience of all can fail to act even when all prescribed doses are taken, but are mistimed. The products in ques-tion are the low-dose, combined oestrogen–progestin, oral contraceptives. As the UK labelling indicates, being more than 12 hours late in taking the once-daily ‘pill’ already increases the risk of breakthrough ovula-tion and conception during the part of the monthly cycle in which ovulation is most likely (Guillebaud, 1993). Thus, a patient who routinely takes a daily ‘pill’, but wobbles in dose-taking between doing so at the usual 7 am or, exceptionally, at bed-time, creates intervals between doses that exceed 36 hours, which would appear to be the mean point at which the risk of breakthrough ovulation starts to rise. (Note that the 36-hour mean implies that half the patients can be expected to have an even shorter margin for dose-timing error than 36 hours.) Clearly, then, in the case of these most widely used products, the ‘80% is OK’ criterion, which means missing one dose in 5, and thus a series of 48-hour or longer intervals between doses, would allow many instances of breakthrough ovulation and correspondingly high likelihood of unwanted conception.

Thus, another factor having a major bearing on the question of ‘how much adherence is enough?’ is the degree of ‘forgiveness’ that each pharmaceu-tical product provides. At one extreme of forgive-ness is bendroflumethazide, a thiazide diuretic widely used in the treatment of hypertension in the United Kingdom, and which has a once-daily dosing regi-men, though it is able to maintain anti-hypertensive action for over 6 days after a missed dose (Girvin and Johnston, 2004). At the other extreme of forgiveness are, as just discussed, the low-dose, combined oestro-gen/progestin oral contraceptives, with their minis-cule average of 12 hours of safety margin beyond the recommended 24-hours interval between once-daily doses. In the latter instance, one can have product fail-ure simply from errors in dose-timing, even though 100% of prescribed doses had been taken. In the former instance, one should be able to omit several sequential daily doses and still have continuity of anti-hypertensive action.