Chronobiology predicts actual and proxy outcomes when dipping fails.

To the Editor: As a gauge of variability, dipping, based on day-night ratios of blood pressure, is much discussed,1 with 2194 hits in a search of the literature on the Internet. But, as compared with dipping, a classification based on chronobiological end points (such as the circadian amplitude and phase) interpreted in the light of reference values specified by gender and age offers superior discrimination in our data. Abnormality in the normal range can occur as (1) a (circadian) blood pressure overswing or circadian hyperamplitude-tension (CHAT) gauged by a circadian amplitude exceeding the upper prediction limit of presumably clinically healthy peers of the same gender, age group, and ethnicity; (2) an excessive pulse pressure gauged by a persisting excessive difference between systolic pressure, when the heart contracts, and diastolic pressure, when the heart relaxes, measured around-theclock; (3) circadian ecphasia, an odd timing of the daily blood pressure swing in the absence of an oddly timed daily heart rate pattern to rule out effects of work and sleep schedule shifts that may affect the timing of both blood pressure and heart rate rhythms; or (4) too little heart rate jitter, gauged by a reduced around-the-clock standard deviation of heart rate. In a 6-year prospective study of 297 patients with no initial history of morbid cardiovascular event undergoing 48-hour ambulatory blood pressure monitoring,2 a circadian amplitude above the upper 95% prediction limit of clinically healthy peers matched by gender and age had a relative risk of 4.27 (95% CI: 2.43, 7.51; P 0.001), whereas nondipping was not discriminatory (RR 1.37; 95% CI: 0.75, 2.51; P 0.05). Analyses of 1179 untreated patients3 indicate that the concomitantly assessed left ventricular mass index (LVMI) of patients with an abnormal circadian pattern of diastolic blood pressure (DBP) is greatly elevated (Fisher statistic from 1-way ANOVA: F 15.959, P 0.001), contrasting with the LVMI of reverse dippers, nondippers, dippers, or extreme dippers (F 1.605, P 0.186) (Figure). Much larger LVMI values, considered as a surrogate outcome measure available from all 1179 patients, are observed in the presence of abnormal circadian patterns of DBP, whether the phase occurs at an odd time (ecphasia) or whether the amplitude is excessive (CHAT). Comparable elevations in LVMI are not seen for patients with an abnormal day-night ratio. A similar comparison based on systolic blood pressure (not shown) also favors a classification based on cosinor-derived circadian characteristics, whether considering all patients (All) or only women (F) or men (M), as does broader evidence in the Table. In populations of presumably healthy subjects and untreated or treated hypertensive patients, usually with no prior cardiovascular morbidity, a classification in terms of dipping based on the day-night ratio, routinely assessed in our analyses, has not contributed risk information beyond prediction achieved by means of chronobiological end points. To detect variability disorders, we need to replace the single office measurement of blood pressure by a 7-day profile of 3-hourly (self-) or denser (eg, half-hourly) automatic measurements analyzed chronobiologically, because there can be large day-to-day variability in circadian characteristics.4,5 We may find a midline estimating statistic of rhythm (MESOR)hypertensive patient, whose overall blood pressure is seemingly well treated by drugs, with acceptable measurements during office hours, but who may still have much too high or much too