I wrote a monograph a few years ago on how to improve clinical care. I will post sections on this website. Hope you find it interesting. Also considering making it into podcasts.
Chapter 1: What is Error?
When I lecture medical students in the months before they graduate, I often begin with the following statement: “Y’all need to be aware, you don’t know anything about medicine, I wouldn’t trust you taking a temperature. The purpose of residency is to turn you into doctors, medical school is more of a hurdle than a real training program. All medical school teaches you is how to pass medical school courses. So, remember, you are novices. Really, you’re worse than novices, because most novices will not be called Doctor. Just keep your eyes open and soak up new knowledge. Try to think of yourselves as paranoid sponges and you will do well”.
While this may seem harsh, I feel it is important to make the stakes of the medical game clear to our new practitioners. Medicine is unique in that we give novices the title of an expert (Doctor) and place them immediately in situations where expertise may be necessary. They are supervised, but not at the elbow, and a common problem in novices is the inability to recognize when they are over their head. In addition, medical school graduates, in general, are life-time overachievers with a high regard for their intellectual capabilities. They do not inherently believe they are incompetent.
Herein lies the problem, we don’t stress fear of the unknown sufficiently in our training, and we instill too much pride in the accomplishments of our trainees. The hard truth is that an Paramedic with a year under his or her belt is better able to keep you alive than a recent medical school graduate (unless they were also an EMT). The fact is that you don’t gain practical knowledge until you gain experience. Experience is gained by actually doing things (seeing patients and making decisions and seeing if those decisions were effective). Talking about medicine, reading about medicine, answering multiple choice questions about medicine, get you nothing in the way of practical, tactical experience. What I will try to explain in this book, using references to non-medical processes, is how our present system (and possibly any conceivable system) of medical care can not provide “perfect” care every time, or even most of the time. Error is intrinsic to our clinical care, it is how we learn, how we improve, and how we discover new ways to do things that will provide better outcomes in the long run. Revamping our systems in such a way to provide seemingly “error free” or perfect care will create a system where learning is limited, innovation is stymied, and (I predict) will provide no real long term improvement in outcomes. In fact, almost all of the recent studies of the effect of implementation of core measures and duty hour restrictions have not demonstrated significant improvements in outcomes, but have found greater costs (hence these changes will not be cheap). In reality, duty hour restrictions are being rolled back, and many of the metrics we used to measure “quality” are being reconsidered.
My hypothesis is that our current method for improving care promotes gaming of data, and institution of practices that are centrally conceived and very difficult to implement locally. This leads to a tremendous waste of effort in implementing new processes, with very little to show for it afterwards. Performance improvement is difficult. It takes time, patience, and willing participants. The impression that doctors and nurses know the right things to do, and simply do not do them out of laziness and /or ignorance is counter productive. We must change these strategies, and employ different measurement tools if our goal is truly “better” health care, and not just better statistics.
Freeing an activity from error and achieving perfection
Think of anything you do routinely in real life, and consider what would be required to make that process completely error-free (or perfect). Here is an example
Getting the Kids to School
Every morning my wife gets my kids ready for school (you might ask where I am when she does this…………let’s move on). She gets up at 6AM, takes a shower, and goes downstairs. Some of her tasks include: getting everyone fed, getting the dogs to their daily backyard play date, getting everyone dressed, out to the bus stop and off to school. She also does a double check on school assignments and homework. She’s done it for over 18 years, every weekday, for 10 months a year. She tries hard to do it well and give the kids a good start to the day. But can something as simple and repetitive as this be done perfectly and error-free?
The weather is different every day, therefore the clothes she chose the night before may need adjustment, the kids may want something for breakfast we don’t have in the house or is in the freezer. When we had a nanny, sometimes she called in sick, when my wife had an early OR case or clinic. There may be sports that evening and she needs to get the lacrosse bags together, the girls may have swimming and she needs to find a ride for them, etc., etc. (my wife is also an actively practicing Board-Certified Obstetrician and a Saint).
Lets think of two possible goals for this routine daily process. First, to do this every day safely, and second to do this perfectly (without error) along with a monitoring and accountability system for non-compliance. My wife gets this done safely every day. Her actions are carried out in such a way to provide time buffers to each of the activities (so the kids don’t miss the bus), she knows she has some food in the house so the kids will eat something, etc. This gives her the opportunity to make little changes in her routine daily as needed, and even to improve the process. She only really get angry if the kids leave their lacrosse bags in my car, or at someone’e house, etc. throwing off her processes and adding time that she does not have to the evolution.
But what if she was told she had to do this perfectly with no errors daily, or the kids would be put into protective custody? Cameras with time stamps are placed in our house, and at the bus stop. Specifications are written by a national agency for: number of calories ingested by age group for breakfast, time leaving the house prior to bus arrival, time arrived at bus stop, and whether the clothing met specifications specific to different weather conditions. The data from the cameras would be reviewed quarterly and the results published in our local newspaper. If the results had more than a 30% error rate, the children would be sent to foster care.
How would she meet these standards? How would you do it? Would it be more expensive? Would you want to have kids if this was implemented? Would you game the system?
If we apply the Lean methodology to this, we would define our background (which I have somewhat done), current condition (as above), goals (as above), gather data and define countermeasures.
For countermeasures, you would have to define exact pre-programmed processes for everything that would reduce variability to as near zero as possible. Get out of the shower at an exact time, have a wider variety of foods in the house (or manage the expectations of the kids with regard to breakfast). She would have to practice her food preparation to insure it is made according to standard, have alarms set up to indicate when time intervals had been reached, etc.
You can see how much this changes the way she would do things, make them much more difficult, and change her mindset from doing her best, to meeting the standards. It would also be far more costly, since she would have to purchase more food, more clothes (in case what was needed for that day was dirty), and more alarm clocks. If her performance was borderline after the first quarter, she may have to hire consultants to help her, or go back into the A3 process and re-examine and re-tool, two things she has no time for. She also may have to hire morning staff to help facilitate the process. This staff would have to be trained and paid.
Welcome to American Medicine in 2016
Despite the comical nature of this example, is it really that far from what we have put in place in American hospitals? Most guidelines have good intentions (certainly you want children to get sufficient food in the morning, and be properly clothed, and who would object to being measured if you were really doing a good job?) but their downstream effects are often not realized when they are implemented. A system as I’ve described would also change my wife’s disposition and attitude toward regulation.
Lets look at some other examples of the difference between low adverse event rate and a zero adverse event rate and how that relates to how you carry out your business:
The military undertakes dangerous missions as part of their daily work. Some are very risky (special operations missions into hostile territory), and some are much less risky (moving supplies through friendly territory). Let’s look at a mission of average risk, a patrol from a base into the local environment, the sort of mission that went on in Afghanistan and Iraq every day. Commanders plan these patrols to keep their soldiers reasonably safe, and their tactics will reflect that goal. The vast majority of patrols are executed with no casualties, but every once in a while, someone will take a shot at a soldier, or set off an explosive device and cause casualties. The commander arranges contingency plans for when events turn hostile (troops to respond if more help is needed, air support, and medical evacuation).
Now, think about how that commander would have to plan the mission if the goal was to NEVER suffer a casualty, analogous to setting up your medical care such that you never have an unexpected death or serious adverse event (or a decubitus ulcer or a line infection). The commander would have to significantly change the way the patrol is carried out. He would have to increases resources (more troops), provide greater protection (more weapons), plan more significant and more rapid backup in case the patrol is caught in an ambush (air support overhead at all times). The commander would also probably put more experienced people on the patrol, and remove novices, since they could not count on the novices to perform in a error free manner under stress. They would also probably change their objectives (make the radius of the patrol smaller, avoid potentially dangerous areas).
So, it would be more expensive, use more resources, get less done, have less resources available for other emergencies, and would not provide an environment for training. Most of this is common sense. Risk is inherent in any action, and trying to reduce the risk to zero will necessitate significant changes in the action. In most activities, we accept some risk in order to get things done. But naturally, we should mitigate any unnecessary risk. For instance, the commander might want to show his superior he can get more patrols done than a commander in another platoon at lower cost. If he did this by sending two soldiers out on each patrol, only allowing them to carry pistols with only 2 rounds of ammunition, and placing “shoot me” placards on their back, he would likely be reassigned to the motor pool.
How did the military improve their performance in Iraq and Afghanistan? They are diligent about after-action reviews. They try to foster an attitude in their junior officers promoting constant improvement and individual innovation. They want their people on the ground to use basic “rules” and “processes” and adapt them to the specific needs of the mission. These adaptations are critically dissected prior to and after the mission. If successful, they are added to the armamentarium of the commanders for future missions. If they do not work well, they may be completely discarded, or specific kernels of innovation may be preserved and re-tested. In this way the military slowly improves. In almost every American military conflict, the performance of commanders and soldiers becomes more effective and efficient as they gain experience. Results achieved toward the end of conflicts, would have been impossible to achieve at the beginning. Thus the military is a learning organization (when it is at its best). This is essentially the opposite of what is being implemented in hospital based medicine.
The Current State of American Medicine
Clearly American medical care has issues. A significant percentage of care is unnecessary and wasteful, complication rates may be higher than other high tech health systems, our costs are high, and we have a system that does not provide equal access. How do we improve hospital based medical care without the use of the strategies I’ve outlined above.
The answer is basic performance improvement. A major problem is what is “best” care, of the most “evidence-based care”. The undeniable fact is we do not have a great handle on that anywhere in medicine. Much of what was thought to be optimal, has through scientific investigation found not to be optimal, and in some cases detrimental. So reliance on “core measures” at best is a rough guess at optimal care, usually with no conclusive data that following such measures provides significant improvement in outcomes. An unfortunate problem with medicine is that it is very difficult to empirically know what is right and wrong. As an example we now perform CPR without breaths! If that had been brought forward 20 years ago, it would have been laughed at. But as we increase our understanding, things change. How many patients were hurt by the previous CPR protocols? We will never know.
We need to make improvement intrinsic to how we do business, but our goals should be to do better each day, not perfection.