Before he met Mrs. B for the first time, Dr. Good glanced at the pattern of her blood pressure readings over the previous several years. He could see from her chart the fitful climb of her pressures in spite of his partner’s additions of first one medication and then a second, each with several dose increases. The picture of uncontrolled high blood pressure despite medication jumped out at him.

Dr. Good also glanced at the patterns of her glucose levels, her cholesterol, and her body mass index, all of which were climbing in the wrong direction. It took him just a few glances through her labs to adjust his hunches, and with one hand on the door knob, the focus of his attention went to her central nervous system. What did he know about the state of her central nervous system? He had some guesses, but he knew better than to search for lab values. He knew he wouldn’t find any.

We don’t measure nerves or nervousness or the central nervous system in the clinic. We don’t measure stress or the stress response system. We talk about it, but we don’t capture the essence of that talk in numbers, not usually. And our talk about stress is general, impressionistic, passionate, and often hard to translate into action.

The aim of this chapter is to describe the range of approaches to measuring stress that is available to researchers, to doctors, and to their patients in order to appreciate the kinds of choices we must make when trying to measure stress in the clinic. An appreciation of the dilemmas of stress measurement sets the stage for understanding in this chapter what the essential components of a clinically useful stress profile are.

Making a Model Work

The concept of stress has been around a long time, and building a model for how stress leads to illness is, relatively speaking, the easy part. The harder part is putting the model to work in places where it counts for patients and their docs, such as the primary care clinic, where the mission is treatment planning.

The first hurdle we face in making this model useful is measurement. Scientists generally ignore what can’t be measured. And busy primary care doctors have to treat the numbers. What numbers for stress do they treat? Very few. And if we’re hoping to motivate distressed people to change their high-risk health behaviors, we know that measurement can be a powerful motivator.

Why Not Measure?

It’s not only stress that we clinicians avoid measuring. The history of medical science includes numerous examples of our reluctance to admit measurement into the art of medicine. An article by Tony Rousmaniere¹ reminds us that in the mid-1800s, 250 years after its invention, many physicians were reluctant to use the thermometer to manage fever. In 1868 Carl Wunderlich “proposed the radical idea of tracking a febrile [fever] illness by reading the patient’s temperature at regular intervals.”

Good doctors at the time used touch and observation and intuition to track fevers. Thermometers were cumbersome and often not reliable. As technology improved along with the science of fevers, reluctance yielded eventually to including thermometer data for fever management. The switch to temperature monitoring required several decades. Now every parent knows how to watch a child’s fever of 99 degrees F, and knows to rush a fever of 104 degrees F to the emergency room. And during the worst days of COVID, we were all taking temperatures at home and at entrances to a business or health clinic.

Until the late 1990s in the US, physical pain, though common, remained unmeasured, mostly underappreciated, and often inefficiently treated. Only with the adoption in the late 1990s of the rating of pain on a 0 to 10 scale, initially in VA medical centers and later around the country, did pain become “visible” and consistently measured with a simple question.²

We doctors are often reluctant to measure a patient’s mood, which has clouded the practice of psychotherapy and psychiatry, perpetuating our confusion about what works to relieve mental suffering.

And as the technology and science of measuring stress and other physical and psychological processes improve, our reluctance to measure is gradually yielding to the practice of measuring the workings of the brain and the mind. In fact, the healthcare reform movement in the US is embracing measurement-based care as a key principle for improving the quality of healthcare. Has the time arrived for effectively measuring some aspects of the stress response system when you see your doctor?

The Sophisticated Lab Approach

To appreciate what it takes to look deeper into the workings of the stress response system, beneath our conscious awareness, I followed Nancy Murrah, RN, BSN, a nurse, deep into the basement of the Emory Hospital Clinic Building in Atlanta one Tuesday morning. Murrah has been coordinating clinical research studies at Emory for thirty years, and she still does it with enthusiasm and affection for her patients and colleagues.

At the time she was coordinating five studies run by Viola Vaccarino, MD, PhD, the chair of the department of epidemiology at Emory’s Rollins School of Public Health. Dr. Vaccarino is a cardiovascular epidemiologist married to a research psychiatrist, Doug Bremner, MD, and this marriage is part of what has spurred her to focus her career on psychosocial risk factors for heart disease.

The study I was about to observe is called the PTSD and Ischemic Heart Disease Progression: A Longitudinal Twin Study. All the twin pairs in this study are Vietnam veterans, and the question the study aimed to answer was whether and how PTSD might contribute to heart disease over twelve years. The twin design would allow the investigators to tease out genetic and environmental influences on heart disease.

Murrah led me to a suite of rooms off a main corridor in the basement where a team of three research assistants in white coats was already setting up Mr. F for the morning protocol. She handed me a white coat, we glanced in at the waiting room where Mr. F’s heavy-set twin brother was dozing, and then entered the recording room.

Mr. F sat still in a full-backed recliner chair with six EKG wires stuck on his chest trailing out from under his green checked shirt tails, four wires clipped to carotid sensors on each side of his neck, a clear IV tube for blood drawing inserted in his right arm, a blood pressure cuff on his upper left arm, a respiratory belt around his chest, and sensors clipped on three fingers of each hand.

He was an average-sized man who looked pretty fit for sixty-four, and he smiled calmly at me when I was introduced, offering me his wired hand for a shake, amused and accommodating and apparently expecting me. He had spent the previous day doing similar procedures for this study, and twelve years before, he and his twin brother had completed the baseline version of this study. So Mr. F was a study veteran as well as a military veteran. This was familiar duty for him.

Steve, the head tech, sat to his right by the BIOPAC console and screen, chatting with Mr. F, arranging the wires, and tinkering with the leads, while Belal, the assistant on Mr. F’s left, circulated to help with the setup. From where I sat on a stool behind Steve and facing Mr. F, I could watch the readouts on the screen and him. Light banter among the three of them kept the mood friendly.

I learned from the banter that Mr. F and his brother, who live in Wisconsin about a five-hour drive from each other, had flown to Atlanta Sunday on separate planes, one from Madison and the other from Milwaukee. Mr. F told me his brother got lost in the Atlanta airport for four hours and couldn’t figure out how to call him.

“Something’s not right with him,” Mr. F said in his matter-of-fact understated way. “I hadn’t seen him for the better part of a year until we got here Sunday. I don’t know what’s going on, but he’s changed.”

After Steve ran the baseline tracings for heart rate and heart rate variability, Belal took the blood draw for cortisol, epinephrine, and norepinephrine, and the oral swab for sampling the microbiome of Mr. F’s GI tract.

Then a third research assistant, Stacey, came in to run “the scripts.” She explained to him that she would play the scripts through the headphones while they would record various measures with all the wires and bands adorning his body. All he had to do was attend to the scripts.

She asked first for Mr. F to rate his emotions and his level of distress. Then she put headphones on him and played the series of four scripts, two neutral ones followed by two “traumatic events,” which he had chosen from his life and described the day before.

Stacey’s team had edited his description of these events into two scripts of a few minutes each. After she played each script, she asked him about his specific feelings (nervous, anxious, fearful, high, angry) and asked him to rate his level of distress from 0 to 100.

The video camera and I watched him during these scripts. He looked calm and untroubled by all these wires and recording devices and strangers attending to him. For the first two neutral scripts, he reported no emotions and no distress. For the first trauma script he reported mild anger and mild anxiety, but rated his distress level 0. Here is what he was listening to [KIAs are service members killed in action]:

For the second trauma script, Mr. F reported mild anxiety and rated his distress at 20 out of 100. I wondered what it would take to rattle this man and raise his distress rating over 50! See for yourself:

I could see on the monitor that Mr. F’s blood pressure readings were up in the range of hypertension: at baseline 167/85, up around 153/90 during the first trauma script and the second, and back down to 136/78 after Steve asked him to relax at the end of the procedure. His heart rate rose from a baseline of 59 to the high 70s and back down to the low 60s.

“My blood pressure is usually pretty low,” he volunteered, unaware of what his readings had been. He then talked again about concern for his brother’s confusion, and Steve noted his blood pressure rise when he talked about his brother and then return to normal during the relaxation phase at the end. Here we could see the cardiovascular reaction in his stress response system as he experienced what he called mild anger, anxiety, and distress.

Mr. F had spent the previous night in the Emory sleep lab, and after he finished this morning’s recording session, he would spend the afternoon in a PET scan of his heart to study blood flow during rest and during a drug-induced stress test. That evening he and his brother would fly back to Wisconsin, having twice donated their time, confidences, and body fluids to one of the state-of-the-art studies of stress across a crucial section of their lifespans.

Like Ted Daley who was wearing all the monitors and filling out logs for the SHINE study, Mr. F is among our unsung heroes. As are our friends and neighbors who agree to be in phase 1, 2, or 3 studies of treatments for cancer or who rolled up their sleeves to try out the COVID-19 vaccines.

People who sign up for and qualify to be in research studies that benefit the advancement of science and eventually lead to advances in treatment for each one of us are never thanked enough. We researchers learn so much from their unselfish gifts.

Dr. Vaccarino’s team plans to mine gold from these 283 twin pairs of veterans with data on their genetic, endocrine, cardiovascular, immune, metabolic, gastrointestinal, and autonomic and central nervous systems. That gold mine will produce one of the most comprehensive pictures of the stress response system in action, the whole system and its parts. Yet this study is just one of several like it that Dr. Vaccarino’s team has completed, exploring the effects of stress on heart disease.

The silent strains that I could see on the blood pressure and heart rate monitors during Mr. F’s recording session are just the tip of the iceberg of his stress response system that Dr. Vaccarino’s study measures in detail. And what will his brother’s system look like, the one who has been struggling with PTSD for decades and now can’t find his way out of an airport? Dr. Vaccarino hopes to show us in detail how those two facts of Mr. F’s twin brother’s life – a lifetime of stress and a declining brain and heart – are related.

Dr. Vaccarino and her colleagues are raising the limits of sophistication for measuring the stress response system in the laboratory. They do this by measuring multiple organ system responses over long periods of time (twelve years) under a range of conditions and stressors in a large sample, at least half of whom have a stress-related chronic condition.

So measuring stress is possible, and it is difficult to do well, and it is costly. It can also be done outside the laboratory. Studies that collect data on stressors and stress responses during everyday life, like the SHINE study using momentary assessment methods recording a week or two of sample days and nights, like Ted was doing, achieve a similar level of sophistication and comprehensiveness of measurement.

The Problem with Slices of Time

Before we dig into the technical difficulties of measuring the stress response system in clinical settings where it can immediately guide a patient’s treatment, we should remember that measurement in medicine is still young and insufficient. Most of the lab data we use to guide our treatments represent the equivalent of a few frame shots from which experts aim to guess at the whole movie of a current life.

That is, what we really care about for health and illness are the full range of daily variations over many months in our blood pressure or glucose or heart rate or mood, and we care about how all of these vary together. How well does the whole stress response system and each of its subsystems adjust over time? How much do these demands strain the capacities of this person?

Those two high blood pressure readings in Dr. Good’s office that triggered the prescription for Mrs. B’s third blood pressure medication were just a couple of frame shots in the long movie of her blood pressure before and after the bleed into her eye. These readings offer the best guess any doctor can do for now, but what we really care about is that movie of how her blood pressure has fluctuated day and night during heavy traffic, nightmares, and an argument with her sister.

If that sounds like too much data, you’re right. Among the current barriers to that ideal level of monitoring are the challenges of collecting and analyzing that much “big data” in ways that are useful for the regular patient we doctors see every day in our exam rooms. However, those kinds of big data advances are developing rapidly enough that such a monitoring scenario will surely soon become a reality for selected healthcare systems.

Wearables such as Fitbits and Apple Watches are only the beginning. The key point to keep in mind is how our current approach to monitoring the functioning of the stress response system captures only the tiniest fraction of the relevant data. Our current definitions of clinically meaningful hypertension or diabetes or depression rely on insufficient slices of data, and will surely be revised when we can watch the whole movie.

The Problem with Stressful Events

A few exceptionally well-funded research scientists measure stress in most of its dimensions for short intervals, but the rest of the clinical world waits, or looks the other way. Why is it so difficult to measure stress?

The difficulties are rooted in our many definitions of stress. Since the 1990s Scott Monroe, PhD, a professor of clinical psychology at Notre Dame, has been studying how we measure stress. Early in his career Dr. Monroe wondered how we do on what is perhaps the easiest aspect of stress to measure, stressful life events. He discovered that our recall of stressful life events is not as good as we pretend. In fact, it’s often embarrassingly poor.

To appreciate the extent of this problem, think about the last stressful life events questionnaire you saw. Most of the initial efforts to formally assess stress focused on self-report questionnaires about events (reporting them), as opposed to your responses (how you reacted) to those events.³ They asked what kinds of stressful events you have been exposed to and, for each exposure, how severely affected you were: death of a family member, job loss, promotion, wedding, car accident, for example.

Dr. Monroe and colleagues interviewed samples of people who had responded to these questionnaires and learned some sobering facts about their answers.

First, respondents commonly endorsed items that later proved not to match the questionnaire’s threshold for a stressful event. These false positives resulted from the questionnaires not specifying thresholds of severity and from wide variations in the ways we appraise what is stressful to each of us.

Is a fender bender a stressful event, or must it be a car accident that leads to car damage or injury or results in a driving phobia? And false negatives are also a problem. Checklists often fail to capture what later prove in interviews to be valid stressful events. Multiple studies have found that even the best self-report measures of stressful life events capture less than half of the valid events identified by interview.

We forget, we deny, we misinterpret the questions, we distort our answers to please our imagined raters, and we change our responses with time and prompting.

These early shortcomings in the measurement of stress are forgivable if we understand that the measurement of stress focused only on self-reports of stressful events, because that was the easy and cheap way to collect stress data on large samples. The reports on stress were unreliable and hard to interpret, sloppy science – the mismeasure of stress exposures. And they ignored the rest of the stress experience, as we will see next.

One barrier to measuring stress is that the cheap and easy way, the self-report questionnaires about stressful events, is inherently unreliable if used alone. Interviews help to obtain useful measures of exposures, but extensive interviews are not practical for large samples of people or hurried primary care practitioners.

The Problem with Stress Responses

Measuring exposures to stressful life events is only one part of the stress measurement task. If the event is the stimulus, what is the response? Four main types of responses are important to distinguish, because they don’t always work in concert: cognitive, emotional, behavioral, and physiologic responses.

In the 1970s and 1980s Richard Lazarus and Susan Folkman, both PhDs at the University of California at Berkeley, made the case for expanding the concept and measurement of stress to include how we respond to the stressful event.⁴ They argued that at least as important as what we are exposed to is how we perceive the exposure, because that’s what determines how we respond in every way.

Consistently, pessimists and optimists tend to look at the same events differently. And you might even react to the same event quite differently when feverish or frightened than when you are healthy and well rested and feeling safe.

One of the key distinctions we make when we appraise a demanding event is the difference between a challenge and a threat. To help us draw the line between stress responses that make us stronger and stress responses that harm us, Drs. Lazarus and Folkman examined how we look at the demands of a task in relationship to how we judge our resources.

When we think our resources exceed the demand (fit enough for volleyball, quick enough for trivia), we tend to view a task as challenging and feel uplifted by meeting that challenge. However, when we see that our resources don’t meet the demand – we lack the money to pay the tax collector, lack the knowledge to pass the test, lack the brute strength to lift the car off the pinned child – we see the task as a threat.

The subjective experience of stress responses is harder to measure than stress exposures, and, relatively speaking, there has been less research on how we perceive or evaluate stress than on the more measurable exposures to life events.

The most widely used, best-studied, and easiest to administer measure of stress appraisal is the Perceived Stress Scale (PSS), first published in 1983.⁵ It’s a ten-item self-report form that asks you to rate on a five-point scale your perceptions over the past month of how upset you have been, how often you felt unable to control the important things in your life, and how confident you felt about your ability to handle personal problems. It takes no more than a few minutes to complete.

The authors of the PSS have shown that mental and emotional appraisals of stress have a different impact on illness than exposures to life events. Both aspects of the stress experience are important and influential, but distinguishable in their effects on illness. This becomes useful when thinking about how the stress response system influences illness, and ultimately which treatments to try.

For example, Mrs. B may ruminate more about the stress in her life and score higher on the PSS than her sister. Mrs. B has always viewed their world as a more threatening place than her sister has, though they grew up exposed to roughly the same events and currently live in the same neighborhood. Mrs. B spends more time on high alert and more energy coping with threats each day and each night than her sister does.

Cognitive therapy could help Mrs. B reduce and manage her daily perceptions of threats and may reduce her vulnerability to illness more than changing the number of stressful events she is exposed to, though both approaches would help.

Related to and overlapping with appraisals of stress are how we look inward and how aware we are of our emotions and thoughts (are we feeling anxious or depressed, for example), which determines what we report as stressful. This, too, is a psychological skill that some people do better than others.

Stress and Resilience

As we saw earlier, the risk for illness rises as the stress burden rises, and it declines as resilience increases. People facing high levels of stressors may do well if they also can sustain high levels of resilience. Over the past half-century the measurement of resilience has lagged behind the measurement of stress, so we have fewer good measures of resilience, and we know less about their strengths and weaknesses.

Further complicating the measurement of stress responses and resilience is the fact that, in addition to life events and our perceptions of our responses to these events, we must consider the body’s responses that all too often escape our awareness.

Just as today’s temperature doesn’t tell us about global warming, today’s blood pressure or cortisol level does not tell us about our cumulative risk for illness. We can’t hook everyone up to Dr. Vaccarino’s lab wires and tubes. Even the three monitoring devices plus saliva, blood, and urine collections can be endured only for a few days by Ted Daley and the most willing and forgiving volunteers. The complexity of the stress response system – its social, psychological, and biological dimensions – and its variations over a lifetime make for a thorny measurement dilemma.

If toxic stress matters in chronic illness, then the measurement of toxic stress matters. The difficulty of any task is not a good reason to avoid trying, and the efforts of stress researchers in this area have contributed to substantial recent progress in measuring stress. The road to understanding other complex systems, such as the global climate, the human genome, stellar space, and the human experience of pain has always run through incremental advances in methods of measurement.

Making Stress Visible

When Ted Daley completed his final visit in the SHINE study, he met with one of the investigators, and she briefly reviewed his experience and handed him a sheet that summarized some of his data.

Several months later when I asked him what he had gained from participating in this study, he said one of the best parts was the half hour he spent by himself later that day looking over his data. It included the ranges of his heart rate and his blood pressure over the four days of recording. He felt reassured to notice that his heart rate varied between 58 and 96 beats per minute and stayed mostly in the 70s. He didn’t see any alarming spikes in his heart rate or blood pressure and concluded that his recorded week had been uneventful.

When Ted and I reviewed his data summary sheet, he started to notice a few patterns. His blood pressure graph on workdays was notably higher than on weekends. And the summary of his blood pressure showed an average blood pressure when he was awake of 167/97 and asleep of 148/91. That’s 79 blood pressure readings, taken about every hour over four days, and it included his relatively lower blood pressures during the weekend. By any definition, these readings add up to high blood pressure (hypertension), at least during his workdays and nights.⁶ And yet at the time of the study no doctor or nurse had ever recorded a high blood pressure reading on Ted Daley, as far as he knew. Six months after the study, however, he had had several high blood pressure readings at his doctor’s office.

He thinks about his brother who died of a heart attack at thirty-eight and his sister who is overweight, maybe prediabetic. His own BMI is 35, though he doesn’t think of himself as obese. According to his doctor, he had not yet needed any treatment for his blood pressure, but he thought he might share this data summary with his doctor next time.

Is work stress driving Ted Daley to hypertension? Measurement at this level of detail can begin to answer such a question. This simple graph and numerical summary of his blood pressure made visible to Ted both his pattern of work stress and the possibility of his developing hypertension.

Does Measurement Help?

One final point in favor of bothering to measure stress, before we look at the best measures. After years in the trenches of clinical care, I am so convinced of the therapeutic value of monitoring that I often spend a significant chunk of my early sessions with a new patient exploring ways to track our progress or the lack of it, like with Teresa or like Dr. Good with Mrs. B. This approach also makes it clear that it’s okay to discuss whether what we’re doing is effective or not. Transparency about effectiveness is usually beneficial. And it implies that there are alternatives if we’re not making progress.

Perhaps the most dramatic demonstration of the therapeutic value of measurement in clinical settings comes to us from the UK National Health Service’s program called Improving Access to Psychological Therapies.⁷ In 2008, when this project launched its first pilot, depression and anxiety were widespread (15 percent of all visits) in the UK but only 10 percent of those afflicted received therapy, and half of those lucky few got substandard or ineffective treatment.

To address this problem of insufficient supply of psychological services, a massive training and distribution of therapists to primary care health centers throughout the UK took place, raising the access to psychological therapies to 16 percent, and aiming for 25 percent by 2021.

But David Clark, PhD, the project’s director, first had to tackle the problem of how to prove that all this training and services would be worth the money invested. In his first two pilot sites, one site collected outcome data on 54 percent of the people seen, the other site on just 6 percent. Missing data were traced to those patients who left treatment before their expected completion date.

So as a next step, to make sure they collected a final measure on everyone, Dr. Clark rigged the system so that at every visit the patient would complete a symptom measure for depression, anxiety, and functioning. It took about five minutes in the waiting room each time, and, surprisingly, only a few patients objected. The data were automatically entered into the NHS chart and could be displayed graphically for review with the doctor.

By addressing his need for transparent outcome data to fund his program, Dr. Clark inadvertently also contributed to a dramatic rise in demonstrable remission rates. In 2008, when the program started collecting these data, remission rates for depression hovered around 20 percent. That is, one in five people treated for depression in this program achieved full recovery. Within two years that rate doubled, and it has gradually risen to around 50 percent in the subsequent years. That’s as good as anywhere in the world, though few other countries can boast such a data set to prove it.

On the whole, around two-thirds of all depression patients in this UK program showed significant improvement. Measurement has been good not only for the patients but for the therapists, who now have numbers to back their confidence and raise their value. This step toward transparent measurement has proven therapeutic in several ways.

In the process of identifying people for whom toxic stress may be contributing to stress-related illnesses, we begin with the simple approaches, single measures that screen for toxic stress, and we build as needed to more complex measures. In the next chapter we will look at some specific measures and what it takes to create a stress profile.