Answers are grounded in this paper and cite their sources.
The classification of hypertensive patients as dippers or nondippers using 24-hour ABPM demonstrates reasonable reproducibility regardless of the specific time intervals used to define day and night.
Studies from independent centres have provided evidence that a blunted day–night blood pressure (BP) difference identifies hypertensive subjects at increased risk of cardiovascular disease [1,2]. Two main barriers, however, hindered for several years the acceptance of this approach of cardiovascular risk stratification. One barrier was the lack of consensus about the definition of time intervals for the calculation of daytime and night-time BP averages [3–5]. Another barrier was the reproducibility of the day–night BP differences, generally perceived as poor [6]. Definition of day and night Day and night have been defined in most studies using time-dependent methods with ‘large’ or ‘narrow’ fixed intervals. Time-independent methods have also been used by selecting the amount of time actually spent out of bed or in bed during 24-h BP monitoring. The large fixed-clock intervals generally include consecutive time windows for definition of day and night, whereas the narrow intervals exclude the so-called transitional hours, during which a variable proportion of subjects may be actually awake or asleep. Reproducibility studies carried out about a decade ago provided concordant results in showing a better reliability of methods based on either diaries or narrow fixed-clock intervals than on methods based on large fixed-clock intervals [4,5,7,8]. Some uncertainty, however, persists as to how better to define day and night for the calculation of daytime and night-time BP averages. Uncertainty is largely driven by the absence of event-based studies comparing different definitions. The present study An elegant study by Henskens et al.[9] published in this issue of the Journal of Hypertension, although not being an event-based study, has the merit to accurately compare some widely used definitions of day and night and to assess their reproducibility. The authors examined 150 hypertensive subjects who underwent 24-h ambulatory BP monitoring (ABPM) twice, a gap of about 1 week in the absence of antihypertensive medications. Left ventricular mass (LVM) by echocardiography and urinary albumin excretion (UAE) were also determined. ABPM recorders were set to measure BP every 15 min from 0700 to 2300 h and every 30 min thereafter. Of note, subjects with unavailability of at least 30% of expected number of measurements, BP recording period of less than 20 h, or lack of valid BP measurements in any 2-h period were excluded from data analysis. Day and night were defined as ‘wide-fixed’ intervals (day from 0700 to 2300 h and night from 2300 to 0700 h), ‘narrow-fixed-1’ time intervals (day from 0900 to 2100 h and night from 0100 to 0600 h), ‘narrow-fixed-2’ time intervals (day from 1000 to 2000 h and night from 0000 to 0600 h), ‘wide-diary’ (day and night as time spent out or in bed, respectively) and finally ‘narrow diary’ (as ‘wide-diary’, but excluding the 2-h transition period around the reported rising and retiring times). The subjects were categorized as dippers or nondippers according to the different definitions of day and night. Definitions have been compared in terms of reproducibility of the dipper/nondipper classification, and differences in LVM and UAE between dippers and nondippers. When different definitions of day and night were compared in terms of absolute BP reduction from day to night, an important finding was the slight, although significant, underestimation of the BP reduction by the wide-fixed definition as compared with the wide-diary one, in agreement with previous studies [4,5,7,8,10]. A remarkable finding of the study by Henskens et al.[9] is the high prevalence of subjects who did not change their classification as dippers or nondippers from the first to the second session. Such frequency varied from 70 [classification based on systolic BP (SBP) with wide-fixed intervals] to 88% (classification based on SBP and diastolic BP (DBP) with narrow-fixed-1 intervals). This high extent of agreement would suggest a remarkable reproducibility of the dipping status, regardless of the definition of day and night. Methodological caveats Indeed, the degree of agreement between the two sessions seemed higher for dippers (81–95%) than nondippers (37–59%). For instance, when using the widely accepted narrow-fixed-2 classification [7] (day from 1000 to 2000 h and night from 0000 to 0600 h) in association with the dipping status solely based on SBP, out of 112 dippers in session 1, 97 (87%) remained classified as dippers in session 2, whereas out of 38 nondippers in session 1, 19 (50%) remained classified as nondippers in session 2. For interpretation of reproducibility of the dipping status, the authors did not give much credit to the unadjusted degree of agreement between the two sessions, which was remarkably high, but to the kappa-statistics, a test that compares the agreement between two independent raters on a given sign whose precise quantification is inherently limited by potential rater bias [11]. The typical feature of the kappa-statistics is that it adjusts for chance (i.e., the proportion of times raters would agree by chance alone). The difference between kappa and 1 indicates the total unachieved agreement beyond chance. Examples of the use of kappa-statistics in the cardiovascular setting are the interpretation of ischemic electrocardiographic changes during exercise test [12], diagnosis of jugular venous distension [13], and so on. The authors highlighted the point that kappa values for dippers/nondippers classification ranged between 0.32 and 0.46. Such values would suggest a fair (kappa = 0.21–0.40) [11] or moderate (kappa = 0.41–0.60) [11] chance-corrected agreement between the two ABPM sessions. The criticism has, however, been moved to the kappa statistics. It has been noted that in the absence of some explicit model of rater decision-making, it is not quite clear how chance can affect the decisions of actual raters and how one might correct for it [11,14,15]. There are examples of low kappa values in the presence of accurate individual ratings, high levels of agreement between raters and also for widely used diagnostic tests. For example, interpretation of ischemic electrocardiographic changes during exercise test yielded a kappa value of only 0.25 [12] and diagnosis of jugular vein distension yielded a kappa value of 0.56 [13]. Quite low values of kappa have also been reported with other widely used binary classifications based on continuous variables. For instance, in a systematic review of studies addressing the reproducibility of impaired glucose tolerance (IGT) and impaired fasting glucose (IFG) definitions, the kappa values indicated poor to fair agreement for IGT (0.04, 0.22, 0.38, and 0.42) and moderate agreement for IFG (0.44 and 0.56) [14]. Furthermore, as kappa values are not comparable across different rating categories [15,16], the kappa test cannot be used in this study to compare different rating definitions of day and night, but only to test their session-to-session reproducibility. As a further caveat, the kappa statistics assumes total independency of raters, not achieved in this study, because the two ABPM sessions (i.e., the raters) were not fully independent, but at least time-dependent. Final point is that the applicability of kappa statistics is limited by the prevalence of the clinical category under study [17]. In our study, reproducibility of the dipping status in conditions of high proportion on nondipping may not be inferred by the results of this study, in which prevalence of nondippers ranged between 10 and 37%. Clinical implications Beyond interpretation of statistical tests, the main clinical challenge remains the correct interpretation of the day–night BP profile in individual subjects. Unfortunately, Henskens et al. [9] did not make any attempt to identify potential predictive factors for maintenance of the dipping status. For example, sleep deprivation during overnight BP monitoring may result in a flattened day–night BP profile with abolished prognostic impact of nondipping status [18]. In addition, subjects with extreme dipping or, conversely, reverse dipping, may be less likely to become nondippers or dippers, respectively. On the contrary, subjects with a day–night BP difference approaching the threshold value (i.e., close to 10%) are more likely to move from one classification to another just for the effect of spontaneous variability. Along this line of reasoning, the suggestion by the authors to repeat ABPM only in nondippers, not in dippers, should be carefully weighted against the actual values of the day–night BP drop. Event-based studies are particularly suitable to validate different classifications of the dipping status. We have recently compared different wide-fixed, narrow-fixed and diary-based definitions of day and night in 2934 initially untreated hypertensive subjects followed for an average of 7 years. The incidence of major cardiovascular events during follow-up was related to the initial dipping status, and no statistical differences were observed across the different definitions of day and night [19]. In conclusion, the study by Henskens et al[9] is an elegant methodological exercise, which provides measures of the reproducibility of the dipping status using different definitions of day and night. In our interpretation, the study substantiates the reliability and validity of the dippers/nondippers classification. After all, a 70–85% full glass is not empty or poorly filled. Acknowledgement This study has been funded in part by the Associazione Umbra Cuore e Ipertensione, Perugia, Italy.
No takes yet. Share an insight, caveat, or question.
Verdecchia et al. (2008) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: