The Art and Science of Inspiring Quality in Clinical Medicine
Friday, September 26, 2008
| Jeff Gruen, MD, MBA
About the author - Jeff Gruen, MD, MBA is a partner and co-head of Healthcare at PRTM, a global consulting firm focused on helping healthcare and other organizations implement operational excellence. He formerly held a teaching position at Harvard Medical School and was on the staff in Pediatrics at Massachusetts General Hospital. He has a distinguished 25 year career helping to found and lead healthcare organizations including Revolution Health Group the country’s leading consumer health web portal and has a passion for health, healing and business strategy. He can be reached at jgruen@prtm.com or 202-729-1641.
It is axiomatic that clinical medicine is both an art and a science. But how do we measure the quality of the art component, and how do we encourage the more fruitful interplay between the art side and the science side? I would like to suggest that there are rich undiscovered answers to this question in the untapped vein of the single case clinical study.
To understand how and why, let us examine the analogy of quality in a well known service industry. There is much talk lately in healthcare of the business case for quality, but 10 years ago Starbucks Coffee already definitively proved the business case for quality. At over 6,000 stores worldwide the company demonstrates every day that it’s possible to dominate a commodity market by achieving an elevated level of product quality, especially when it is matched with a consistently more delightful service experience.
To make the model economic, Starbucks needed to put in place systems that delivered this quality. To make the business case, Starbucks needed to demonstrate that the quality could be achieved at an acceptable cost (since the critical question for service businesses is “what level of quality can we afford to produce consistently”?)
Prior to Starbucks, a cup of coffee was not a destination, was rarely differentiated and was always unbranded. To learn from the Starbucks example, the brutally honest strategic service question for healthcare organizations is this: for the next new customer walking in the door, does my product level and service experience meet the Starbucks threshold of differentiation, or is it more similar to the average coffee shop experience of the 1970’s?
To be fair, it is at least an order of magnitude more difficult for healthcare organizations to clearly differentiate based on quality. I would like to suggest there are three core reasons why:
- We have limited understanding of the nature of the quality we are trying to produce and of how to measure it.
Since healthcare creates both a service (the consumer experience during service events) and a product (health outcomes from a sequence of service events), the challenge is doubly hard
- We are not generally clear on the cost of producing a given level of quality (recall the key question above that
Starbucks needed to answer). Without knowing the true cost, we do not know where to set the bar, leading to constantly varying levels of service quality
- We have limited understanding about specifically how to raise the level of quality
To date, major advances in improving the quality of clinical medicine have leveraged the brilliant insights of Deming, Jurand and others which arose largely from the study of industrial manufacturing processes. These insights derive power from the statistical control model and study of the human factors necessary to affect statistical control. An underlying assumption of this work is that the processes and their causal link to the outcomes we wish to achieve are clearly known.
Statistical control methods have been and will continue to be an extraordinarily powerful and needed approach for reducing unwarranted practice variation, and for thereby improving clinical quality. This is especially true for those processes for which there is incontrovertible evidence of benefit, e.g. the use of ACE inhibitors in certain clinical situations following myocardial infarction. One might call this approach the statistical control model for improving the science of medicine. It has been championed by the pay for performance movement and by the well known work of Jack Wennberg and others on practice pattern variation.
Our firm (PRTM) and others have implemented the statistical processes control approach and helped drive innovations in its use across industries with considerable effect. For example, one of the projects we helped launch is a company called WellDoc (www.welldoc-communications.com). By ingeniously combining wireless telecom technology with artificial intelligence engines, WellDoc is able to provide the evidence-based skills of the world’s expert Diabetologists from places like the Joslin Clinic constantly at the fingertips of diabetics throughout the day for just pennies of delivery cost. This is because the service allows diabetics to interact via cell phone with an expert system engine to allow for virtual “tight control” of their diabetes and early controlled trial data seems to indicate the potential ability to lower HgA1C values by up to 2 points within a few months.
But the delivery of healthcare is confoundingly complex and refractory to simple, linear solutions for improvement. In contrast to say a pilot’s pre-flight checklist, a physician has hundreds if not thousands of potential routines to rehearse in each clinical encounter. Systems with point of service messaging can help eliminate variation from those mechanical tasks for which we known the correct routine.
But as we increasingly move from a focus on improving individual processes, toward a focus on producing the more broadly defined outcome of “value” in the patient’s terms, two key questions arise. How far are we down the road of identifying the processes that are most essential to producing the most value for the patient? And what do we understand about how to effectively encourage healthcare workers to follow these processes?
To understand the question better, return for a moment to the Starbucks analogy. Consider the statistical control mothod to have effectively identified and managed the need for tight regulation of water temperature. Starbucks does this well (ever notice the timers and thermometers on the coffee urns?), and it is part of their secret for keeping fresh unburned coffee in your cup.
But imagine what the experience of walking into a Starbucks would be if they had not identified the processes needed to be controlled at the Barista counter. Without the “call and check” specified dialog between cashier and barista, flow at peak demand hours would slow to the crawl you experience in a long airport security line, and the rate of mistakes would be so high that Starbucks would be forced to contract their coffee options down to a few favorites from the thousands of combinations currently offered. And without the ambience of the unique Starbucks culture at the cashier level, the generally pleasant, rapid transactions could devolve into painful and annoying experiences.
Starbucks does all of this well because they have defined the key processes necessary to deliver the experience. It is interesting to note by the way that Starbucks does all of this with only a day or so off-site training --- most of the processes are learned in an apprentice on-the job training mode, something I’d like to come back to in a future blog.
In healthcare, I’d like to suggest, we know a fair amount about the processes that we are trying to control with statistical means (by analogy the water temperature and heating timing at Starbucks). We know much less about the processes to produce the real health outcome (analogous to the barista’s methods for quickly moving through product). And we know very little about creating environments that support consistently superlative service (analogous roughly to the Starbucks cashier experience, which, if not always superlative, is at least consistently acceptable). Our problem is much harder because healthcare services are considerably more complex than a coffee bar. They are the processes, in addition to following known protocols, of effective intuitive diagnosis, of empathic listening, of powerful communication and mobilization across multidisciplinary teams, and of healing.
In short, we do not have an effective method for discovering and improving the processes of the art of clinical medicine.
This at heart is a fundamental source of disgruntlement and disillusionment on the part of physicians and other healthcare workers with current quality efforts, including pay for performance. It is partly why these programs often have serious limitations.
I would like to suggest that the reason physicians have for the most part not exactly embraced statistical control programs runs deeper than their maverick nature (although this is a factor, perhaps as a credit to them, perhaps as a detriment). Physicians and healthcare workers inherently know that there are many subtleties to producing the right result. They do not culturally resonate with programs that do not recognize these subtleties in a sophisticated and meaningful way. I would further suggest, that unless these subtleties are effectively addressed, the credibility of quality efforts will always be limited.
Fortunately, there may be a parallel, less travelled path that could speak loudly and viscerally to healthcare workers, and that could open up fruitful lines of research and innovative service design. Instead of focusing solely on deviations from the mean for standardized processes with a goal of measuring populations that are as large as possible, this approach focuses on an n of exactly one. The method is the lowly and yet majestic single case study.
The case studies that may be of use are single examples of something that went very right in clinical medicine, or even of something that consistently seems to go generally right. The examples of things that go wrong are already used to great effect in Deming and lean processes to identify “defects” and to drive toward zero defects. We at PRTM have consistently implemented quality circles focused on zero defects, which when well managed, can produce great results.
I’d like to focus on less tilled soil: case studies of things that go right, specifically of things that go very right. I’d like to suggest that each case study of what goes very right is a gold mine of insights about valuable processes, ways of thinking, and cultural directions. By studying these examples, we can gain insights that can lead to new modes of leadership and eventually innovations in systems that support the right processes. For as we all know, it is less individuals who fail in healthcare, it is the systems that fail to support the individual in consistently performing the right actions.
There is a long and beautiful legacy to the use of case examples to point toward directions in clinical medicine. This narrative legacy has been less emphasized as we have moved further toward the pursuit of the improbable dream that we can develop incontrovertible evidence with statistically significant samples for all of the possible diagnostic and treatment variations that occur in clinical practice.
To learn how to use a single case study however, we most go to the modern master, the celebrated neurologist and author Oliver Sacks. When Sacks first began his work documenting unusual case reports, he was following in a tradition which many neurologists (perhaps the most scientific of clinicians) take for granted. It is neurologists after all who have long shared the darkly humorous yet revealing clinical saw “one learns the functioning of the brain stroke by stroke”. So it is fitting that a great neurologist in this tradition would seek to press the limit of our understanding of functional and anatomical neurology case by case. Were Dr. Sacks simply to have revived the art of the clinical anecdote, he would not be nearly as memorable and innovative.
Rather, Oliver Sacks’ central Copernican insight came when he first read Aleksandr Luria's The Mind of a Mnemonist. In the book Luria described a man with a photographic memory so strong that he could recall endless stanzas from The Divine Comedy in Italian (a language he did not speak) which had been shown to him only once briefly 25 years before. Luria’s interest though was not just in understanding the neuroanatomy responsible for this bizarre capability. Luria was interested in the effect of this bizarre capability on the patient’s identity. To quote from a wonderful article in Wired on the subject:
“Luria's work suggested that the act of recovering one's own story was itself healing. He called the sort of writing he had done in The Mind of a Mnemonist and The Man With a Shattered World "romantic science." The two books had a profound impact on Sacks. They suggested a new form of writing that combined the clinical precision of 20th-century neurology with both the humane observations of the great Victorian physicians and the explorations of the psyche that Freud undertook in his own case histories….
The "great crisis" in neuropsychology, as Sacks' Russian mentor saw it, was reconciling two modes of scientific observation. One reduces complex phenomena to their constituent parts - the way neurology had narrowed its focus from observation of behavior to specific areas in the brain and then to individual neurons - which Luria paralleled with the evolution of chemistry, from the study of gross matter to the study of compounds, to the study of individual atoms and elements. The other mode relies on the description of phenomena and intuition to comprehend the interactivity of whole systems. Either one, he thought, was inadequate without the other.”
Could we perhaps derive a similar double order of benefit from using case examples of great service delivery, extraordinary clinical acumen, valor in the performance of everyday duties? The case examples could teach us what these momentary masters employ (the “functional anatomy” as it were of the process). But they could perhaps also teach us about the effect on the “identity of the individual”, (by way of analogy, on the identity of the institution). Because ultimately delivering the best is an individual and institutional question of identity – it is a function of what individuals and organizations come to expect of themselves. Case studies may in this way lead us toward a more “wholistic” view of encouraging quality in the highly complex service environment called healthcare.
As we move hopefully toward a value-driven healthcare system (and away from an emphasis just on single processes or transactions), we will need better measures of value from the perspective of the ultimate user (the patient), and better ways of creating this value. Perhaps the individual case study of great results can help to light the way on this journey.
To this end, I hope to spark with this entry a dialog on this approach, and as importantly perhaps inspire others to share examples they have witnessed or lived of extraordinary clinical medicine. Allow me then to kick off this exploration with my own memorable moment, one that occurred in January 1981 when I was an intern on a freezing cold weekend day in New Haven.
I was on call in a busy Pediatric emergency room together with a more experienced resident for a 12 hour shift. As if the bitter New Haven weather and the long hours weren’t bad enough, it turns out we were in the middle of a virtual epidemic of gastroenteritis – stomach flu. Barely taking 5 minutes for lunch, we worked from 8 in the morning until 8 at night, going from room to room. The routine was almost monotonous, grab a chart, head into a room where yet another mother from the inner city sat with crying child. We would each listen to the story, examine the child, determine if the dehydration were solvable in the emergency room, provide instructions and carefully explain the instructions.
I think between the resident and I we set a near record that day, as I recall (and I’m sure my memory is colored over the years) we saw something like 178 children between the two of us over 12 hours. The diagnosis on 90 % of them was the same: stomach flu, gastroenteritis.
Somewhere around 7 pm, my colleague went in to examine his near 100th child of the day, an 8 month old with stomach cramps. By routine of course he went through the motions, asked the same questions, examined the same child with vomiting, some diarrhea, a little fever just like all of the rest. As everyone in clinical medicine knows, these are symptoms of gastroenteritis, and certainly a diagnosis of something like appendicitis would be very low on the list. Appendicitis rarely occurs in an 8 month old. And when you hear hoofbeats as they say (another case of gastroenteritis), don’t think of zebras.
My colleague came out of the room and said to me: I think we have a case of appendicitis in this 8 month old. I looked at him incredulously as he drew the blood tests himself and ordered a surgical consult. The child looked little different from the others --- who would even think of this diagnosis in these circumstances.
And I recall vividly the admiration I felt when the surgical resident, after spending almost 20 minutes in the room, emerged and confirmed the suspicion. And my awe grew when we got the surgical report back later that night that this child’s appendix had been near rupturing, despite a rather benign clinical presentation. In effect, my colleague, by maintaining his composure and focus, had perhaps saved the life of this child.
I am left wondering at how this kind of clinical composure could be recognized, encouraged, institutionalized. These are large and open questions, but I do believe that recognizing the importance of the excellence I witnessed that day, and the difficulty of producing it, is a first step.
These are the kinds of subtleties that the study of case examples may add to our understanding of quality and value, and of how to credible inspire healthcare workers to lift the level of service in the increasingly difficult environments in which we operate. When combined with advanced methods to implement statistical control of definable processes, we may yet achieve a promised land of a step-level improvement in quality.
I would be gratified to hear your comments, stories, case examples and insights to the collective enterprise of this blog.