Seen from the perspective of the patient, going to the doctor is an event! The environment of the clinic, office, or hospital is uniquely different from every other place that the patient goes. Allostatically, a blood pressure taken during the event (being in the clinic) will reflect the patient’s adaptive response to it. As Riva-Rocci noted (see above), arterial occlusion is enough of a stimulus to initiate an increase in blood pressure, but because the taking of a blood pressure is also an entirely unique social interaction involving a physician, nurse, or other medical professional and the patient, there will also be effects related to the perceptions of the patient connected to that interaction. Even if the pressure is taken by an automatic device with no one present, that situation still requires an adaptive response from the patient. When the blood pressure response to this peculiar environment exceeds the average response to all other daily environments, the patient is said to exhibit a white coat effect, but if that effect leads to ausculted blood pressure measurements that exceed 140/90 (hypertension Rubicon) the patient is diagnosed with white coat hypertension. Whether blood pressure responds with an acute heightened response in the clinic may largely depend on prior patient experiences with the setting and prior relationships with the people within it.

That a blood pressure measurement can be profoundly influenced by the perceptions of the patient was dramatically demonstrated by Mancia and colleagues (1987) in their classic study in which blood pressure readings were continuously taken intra-arterially on one arm while a nurse or physician took an ausculted blood pressure from the other. The intra-arterial measurements showed that relative to the pressure prior to the ausculted measurement interaction with the physician, there was an increase of some 23/18 mmHg when the physician took the ausculted pressure. Further, the increase in pressure by the physician was about twice the effect seen when a nurse took the pressure.

What did the patient perceive that lead to the increase in pressure? A more recent study by Jhalani et al. (2005) provides some answers. They examined the acute effects of anxiety and expectancy on clinic measured pressures and found that when assessed as a specific office related effect, anxiety had a substantial influence on increasing pressure in the office. In their study, they measured anxiety before, during, and after blood pressure was measured. They also showed that there is an effect related to the patients’ expectations about what their blood pressure measurement will be. Their findings suggest that prior experience can trigger anxiety regarding this peculiar environment and the relationships within it, so that the blood pressure response is elevated. These psychological factors will lead to a diagnosis of hypertension if the ausculted numbers exceed 140/90.

Masked hypertension is defined by the precise opposite effect seen with white coat hypertension. Specifically in these patients, adaptation to the peculiar clinical environment requires less of a response than an average of the responses to all other events outside the clinic. Rather than being made anxious, they may be calmed by the setting and interpersonal interactions. Interestingly, masked hypertension is seen not so much a relaxed adaptation as it is an absence of high risk behaviors which elevate pressure outside the clinic such as alcohol consumption, smoking, or contraceptive use (see Longo et al. 2005 for example).

Studies have been done which have evaluated the morbidity and mortality risk associated with the diagnosed conditions of white coat and masked hypertension which are defined from the average blood pressure in the clinic and the average blood pressure response to all other conditions during the day (e.g. everything not in the clinic). Pierdomenico and Cuccurullo (2011) did a metaanalysis comparing the risk for cardiac and cerebral events among patients who were diagnosed as normotensive, white coat hypertensive, masked hypertensive and essential hypertensive based on the out of clinic-inside clinic blood pressure difference and found that white coat and normotensive patients had similar risk as did the masked hypertensives and essential hypertensives. This kind of finding suggests that inside clinic-outside clinic variation may not be important to cardiovascular health, and that in fact, the determination of who really has hypertension should be made from average pressure experienced across many different situations and not from the peculiar setting of the clinic or office.

Given that blood pressure is a response to ambient conditions, it would stand to reason that an evaluation of the relationship between its circadian variation and morbidity or mortality would necessarily involve assessing the appropriateness of the pressure responses to the various external and internal conditions that drive the continuous changes (see Zanstra and Johnston 2011 for example). However, virtually every study that examines blood pressure variation as a risk factor for cardiac or cerebral events ignores the dynamic interplay between blood pressure and the specific environmental demands an individual confronts during daily life. Instead, studies of blood pressure variation and vascular risk focus on the event predictability of some measure of the statistical dispersion or cumulative differences of the sample of blood pressures taken with a non-invasive ambulatory blood pressure monitor over the course of one 24-h period (a day) or the average waking-sleep blood pressure transitions (either “dipping”-the difference between average waking pressure and average sleep pressure, or the “morning surge”- the difference between various pressures prior to and just after morning awakening), (see for example Asayama et al. 2015; Hansen et al. 2010; Palatini et al. 2014; Parati et al. 2015; Taylor et al. 2015). These measures are examined only with regard to a possible linear relationship; that is, the studies only address the question of whether risk is related to being too low or too high on the various parameter scales. The inconsistent results from these studies, where some suggest variability is an important risk factor and others find little or no effect has spurred a controversy as to whether blood pressure variation should be a target for treatment (e.g. Asayama et al. 2015). Before this type of issue can be addressed, it is useful to examine what each indicator of the variability, or variation in these risk related studies is measuring. Are the indexes and parameters that are employed in these studies suitable and meaningful indicators of blood pressure variability?

Standard deviations (SD) or coefficients of variation (CV) are measures of the dispersion around a mean of a variable that is normally distributed. These are calculated from presumably random samples of a population of measurements. However, if the distribution of the overall population is not normal and the sampling is unrepresentative and small, these measures will be biased, inaccurate, and uninformative (Cochran 1977). Given that 100,000 or more systolic and diastolic pressures are generated over a 24-h period, and non-invasive ambulatory monitors sample perhaps 50 of those (5/100ths of 1 % of all those generated) which vary with time and conditions in a systematic way (pressures change to adapt the person to continuously changing circumstances) what is the value of the SD or CV of that sample in predicting risk? Parati et al. (1992) some 25 years ago noted that these kinds of measures don’t tell you anything about how single values, as collected, are distributed around the mean. Do the pressures spread out or is there perhaps a bimodal shape? Many odd distributions could provide the same calculated SD or CV. These measures do not provide any information about the pattern and extent of individual pressure responses, and because as noted above, what needs to be evaluated in an assessment of how variability affects pathology is the appropriateness of the variation, they really are unsuitable variability indicators for examining morbidity and mortality risk.

Furthermore, the SD and CV as indicators of 24-h blood pressure variation are poorly reproducible over 24-h (see for example, James et al. 1990a; and the review by Asayama et al. 2015). In our study, we compared 24-h variability in normotensive and hypertensive patients over 2 weeks. Figure 3 shows the timing and spacing of each measurement on each day for both groups of subjects. Note how different the days are. This disparity is actually typical when comparing daily non-invasive ambulatory monitoring data. What we found is that people did different things on different days, and while there were enough pressures to provide a reasonably stable average over time, the varying mix of conditions and times when pressures were taken, were poorly matched day to day. This mismatch profoundly affected the distribution of the pressures around the mean, rendering the distributionally tied measures of variation (SD, CV) irreproducible (James et al. 1990a).

The “average real variability” (ARV24) has also been used as an indicator of blood pressure variability and is defined as the mean of the absolute differences of consecutive non-invasive ambulatory measurements. The effects of this parameter on the predictability of events are small or inconclusive (e.g. Asayama et al. 2015; Hansen et al. 2010). Bearing in mind that each of the sequential blood pressures taken by non-invasive ambulatory monitors are a response to the ambient condition in which they are taken, the ARV24 can go up or down depending upon what the person confronts and is doing during the day. What this quantity really represents is a summary score of the differences between peak blood pressure responses to some indeterminate number of sequential unknown stressors. Ultimately, the magnitude of this parameter depends solely upon the variability of the environments experienced and the behavior/emotional responses of the patient (James 1991, 2013). Thus, a patient who is monitored on a day where they are inactive, remain at home and are emotionally stable will have low ARV24, whereas one that performs multiple varying tasks, transitions through many daily microenvironments (goes to work, out to dinner, etc.) and experiences an array of emotions will have a high ARV24. Since the blood pressure changes are adaptive and are a normative response to the tribulations of everyday life, it is not clear from the studies that have used this parameter why high (or low) values of ARV24 would be indicative of pathology or health.

The other variation measures used in risk studies are “dipping” and the “morning surge.” These are measures of blood pressure change between the state of waking and the state of sleep. Conceptually, dipping refers to the blood pressure transition from waking to sleep, whereas the morning surge refers to the transition from sleep to waking. Operationally, there is no consistent definition for either measure across studies, although with dipping, a Rubicon of 10 % decline, particularly for systolic pressure seems to be the popular demarcation line for normalcy and pathology, although there is no definitive reason why this value is the clinically relevant cut-point (Asayama et al. 2015; Flores 2013; Taylor et al. 2015). Again, seeing blood pressure as an adaptive response, the waking average that is used to determine dipping is based on a mean of values that are tied to the conditions experienced on the day of study. So depending upon whether a person had a difficult day or an easy day, the waking average could be higher or lower. There are ample data showing that excessive psychological stress during the day can also carry over and increase sleep pressure (see James et al. 1989 for example), so non-dipping may occur on a given night simply because it was a stressful waking day. Another problem with the concept of dipping is that it assumes that all people experience just the two distinctive periods (waking and sleep) over the day, so that “waking” and “sleep” happen during the day and night. This presumption is demonstrably false as there are plentiful data showing that waking-sleep patterns can change with age and that this affects the circadian patterns of adaptive blood pressure responses in ways that confound the determination of dippers and non-dippers (see Ice et al. 2003). Likewise, whatever pressure(s) chosen to define the post- awakening point and the low pre-awakening point in defining the morning surge are also adaptive responses to the conditions when they are measured, so that its relative magnitude may be related to any number of factors affecting both sets of measurements. And, as previously noted, there are also other issues with “sleep” pressures taken by a cuff occlusion method that have to do with the position of the cuff relative to the heart that will influence the level and variability of “sleep” blood pressures (James et al. 2015).

It is not surprising that waking-sleep transition measures are often found to have poor reproducibility as well as differential effects in different populations (Asayama et al. 2015; Taylor et al. 2015). Patterns of behavior, stress, and sleep quality vary from day to day, and all these are factors that may be influenced by the cultural background and occupation of the patient (James 2007). While there may be theoretical reasons to believe that the variability in blood pressure associated with wakefulness and sleep ought to have health implications, the operationalization of the concepts using non-invasive ambulatory measurements are inadequate because they don’t embrace the adaptive nature of blood pressure which makes it impossible to define what normative transitions ought to be. Without a clear definition of normalcy, there is no way to coherently use these measures for treatment purposes (Flores 2013).

So, after evaluating the nature of the parameters that have been employed to assess the morbidity and mortality risk of blood pressure variability in large international and community based populations, it appears that none of them are meaningful indicators of what is or is not appropriate variability, and therefore can’t really address the question of whether blood pressure variability ought to be treated.

To understand why blood pressure varies during the day, you need to have information regarding the ambient conditions when measurements are made. If blood pressure is responding to these conditions during everyday life, you need to be able to show that as they change, so does blood pressure.

Several means have been used to classify the conditions of ambulatory blood pressure measurements. While direct observation of subjects wearing the monitor has been used (e.g. Ice et al. 2003), for most studies of blood pressure variation, subjects have self-reported the ambient conditions of each blood pressure measurement in a diary, which have taken on a variety of forms, from pencil and paper to hand held computers as has been discussed (see James 2007, 2013). Most behavioral studies of blood pressure variation have not been conducted with a focus toward allostasis, or even understanding cardiovascular adaptation. Rather, studies have simply defined the sources of diurnal blood pressure variation, or evaluated whether people with specific characteristics differ in their responses to similar lifestyle related stimuli (Gerin and James 2010; James 2013).

Studies designed to evaluate what affects blood pressure variation and by how much have generally taken two forms. As has been noted (James 2007, 2013), the first approach is one where each blood pressure measurement is assessed with regard to simultaneously recorded circumstances reported in a diary (often called ecological momentary data) using inferential statistical models (see for example Brondolo et al. 1999; Gump et al. 2001; James et al. 1986; Kamarck et al. 2002; Kamarck et al. 1998; Schwartz et al. 1994). In this analysis, the sources of blood pressure variation are separated based on the reported diary entries (such as the posture of the subject, the location of the subject, etc.). The proportion of variation associated with each is quantified, as is the number of mmHg the alternative levels of each (such as posture-standing, sitting, reclining) contribute to either increasing or decreasing the values of individual blood pressure measurements. In evaluating blood pressure variation this way, the choice of diary reporting alternatives is critical. The potential sources of variation chosen to have reported in the diary and how they get recorded will dictate how the variation in blood pressure gets analyzed (James 2007, 2013). Analysis of ecological momentary blood pressure data has been undertaken using raw (e.g. Brondolo et al. 1999; Kamarck et al. 2003; Schwartz et al. 1994) and standardized e.g. (Brown et al. 1998; Ice et al. 2003; James et al. 1986) data. The estimated effect sizes from different studies using these approaches vary considerably, due in part to the fact that there is no consensus as to what ought to be the standard value against which sources of variation should be measured, but also because of the demographic and cultural diversity of the groups studied (James 2007, 2013).

The second form employs what might be termed a “natural experiment” which has been discussed at length elsewhere (see James 1991, 2007, 2013). However in brief, natural experiments are studies in which there are a priori design elements that define predictable dynamically changing behaviors or situations that occur during a typical day (James 2013). This kind of study is done by anthropologists and human population biologists, and it is an approach that has its roots in psychological and psychophysiological paradigms in which blood pressure reactivity to various stressful tasks are evaluated in the laboratory, (see for example, Pickering and Gerin 1990; Linden et al. 2003; Kamarck et al. 2003). In these laboratory experiments, a baseline condition is established and then the subject undertakes a series of predefined tasks that will elicit a response. The difference between the baseline measurements and those during the tasks define the magnitude of blood pressure reactivity (James 2013). Because they are conducted in a laboratory, there are controls in the experiment such that specific effects can be isolated, measurements can be taken in a systematic way, and all the participants experience the same protocol. Control groups can also be included in the experiment. Moving this experimental paradigm to a “natural” setting (e.g. into real life and outside the laboratory) requires modification because no true baseline can be established. But, a “natural experiment” can be designed where blood pressure changes can be evaluated as people move from situation to situation (such as their work and home situations) during the course of their everyday lives. For example, a person who lives in a suburb and commutes to an urban workplace every day likely has a structured, urban work environment where economic related activities occur, where social interactions take place with non-relative co-workers, and where a specific occupational hierarchy dictates the nature of social relationships (James 2013). The characteristics of this situation diverge sharply with that of the suburban home, where domestic tasks and leisure activity happen in a social context where interactions are with relatives and neighbors (James 2013). The variation in blood pressure required to adapt to these relatively predictable situations can be assessed by comparing the average blood pressure while in them with that during overnight sleep, or more specifically, while the person is quietly recumbent in a dark room acting as a pseudo-baseline.