Errors in trough: peak ratio determinations induced by patient behaviour.

Abstract

BACKGROUND: Trough: peak ratio is often used to evaluate the duration of antihypertensive action. Whatever the method of measurement chosen, trough effect has to be measured 24 h after the last drug intake for a once daily regimen. Peak effect is usually measured 4-6 h after drug intake. If patients' compliance to therapeutic instructions is perfect, then the 'intrinsic' trough: peak ratio of the drug is equal to the measured trough: peak ratio. Some patients do not follow these instructions, leading to biases in the evaluation of the ratio. For trough evaluation, all patients (N) are supposed to take the last dose of the drug the day before blood pressure measurement. However, if some patients (n1) wrongly take the drug in the morning of the visit, they will be evaluated at peak (type A error). For peak evaluation, all patients (N) are supposed to take the drug a few hours before blood pressure measurement. If some patients (n2) miss their morning dose, they will be evaluated at trough (type B error). METHODS: In the MACH 1 study, the use of an electronic pill count monitor allowed us to quantify n1/N and n2/N. A total of 452 hypertensive patients were randomly assigned to two groups. Patients in group 1 received written instructions to take their last dose during the morning of the day before the visit, whereas patients of group 2 had to take their last dose on the morning of the visit. RESULTS: Electronic pill-box recording revealed that 32.9%: of patients in group 1 committed type A error, whereas 27.7% of patients in group 2 committed type B error. The resulting 'pill-box corrected' trough: peak ratio was lower (87.5% for diastolic blood pressure and 93.1% for systolic blood pressure) than the uncorrected trough: peak ratio (95.2% for diastolic blood pressure and 96.0% for systolic blood pressure) of the population. CONCLUSION: The random behaviour of patients, with respect to treatment compliance, results in a systematic overestimation of the measured trough: peak ratio. The computation of this ratio may be optimized by improving patient compliance. Alternatively, only data from a patient subpopulation that complies with the therapeutic protocol, as reported by readings from electronic pill boxes, should be taken into account for its calculation.