A Lean Machine

Virginia Mason Medical Center in Seattle has become famous as the hospital in America that has most dramatically endorsed the Toyota Lean Production System. The senior administrative and medical team, led by CEO Dr. Gary Kaplan, started the process with a visit to Japan and then designed a hospital-wide program to bring greater efficiency to many aspects of the institution's operations. They entitled their program the Virginia Mason Production System and made significant improvements in many aspects of health care delivery.

At BIDMC, we were not prepared to go quite so far as VM, but we did create a small office to test out the heart of the Lean process, Rapid Process Improvement Workshops. In these short-term intense exercises, a team of people from a variety of jobs categories in a given service or production area get together to map out every step in a customer service or production process. Each step is labeled as "value added" on "non-value added", i.e., with regard to accomplishing the objectives of the area, and then the group decides on strategies to eliminate NVA steps. Then, they actually put them into practice to test their efficacy towards meeting goals of service quality and/or efficiency. Other, longer terms plans and objectives are also decided upon and put in place.

One target area for us was our orthopaedic clinic. Like most clinics, there would be check-in, delays waiting to see a doctor, delays waiting for an X-ray, delays waiting to see the doctor again after the X-ray, and so on. In sum, the average time for a clinic visit was about three hours. Is there any doubt as to why there were disgruntled patients, cranky front-desk staff, frustrated X-ray technicians, and angry doctors?

With great support from our Chief of Orthopaedics, a team was assembled, and they went to work, aided by our Lean coordinator and other helpers. The top chart above shows how many NVA steps (the ones with red dots) were in the "before" process, i.e, the "current state". (Observers often find that over 90% of steps in any service or production process are NVA.) The "final state" chart underneath shows the change in relative NVA and VA steps after the Lean review.

The third chart shows the overall improvement in the amount of time a patient has to spend getting that X-ray and physician consult: Down from three hours to about an hour! Let's repeat that. Previous time for a visit -- 187 minutes. Hoped for target by the Lean team after its analysis -- 84 minutes. Actual results -- 6o minutes or less.

Of course, patients were happy. The staff was very pleased, too. Fewer cranky patients at the front desk complaining about long waits. Efficient use of X-ray equipment and Rad Techs' time. And, doctors being able to stay on schedule all day long. And then being able to add additional appointment slots because they knew they could stay on schedule.

The biggest problem: Patients finished their appointments so quickly that their spouses were nowhere to be found. They were still downstairs at the cafe having a cup of coffee, without enough time to read the whole newspaper!

Forbidden pleasures

I lied. This is really about forbidden abbreviations. But I will do anything for more readers. (No, not really anything....)

One safety improvement I have learned about during the past several of years was the effort to prohibit certain medical abbreviations. This is a Joint Commission requirement, and it is a really, really good one.

For those of you non-medical folks out there, let me provide some examples relating to medication dosages. Now, you have to imagine these things being written in a doctor's handwriting to get the whole point. (Admittedly, this is corrected with computerized order entry, but that still does not exist everywhere.)

Trailing zeros. If you write a dosage that is supposed to be in whole units like this -- 1.0 mg -- it will often be read as "10 mg". That is a big difference, an entire order of magnitude. So the rule is "don't use terminal zeros for doses expressed in whole units." So it is properly done like this -- 1 mg.

Missing leading zeros. In contrast, when a dosage is a fractional amount, it is unacceptable to leave off the zero, like this -- .5 mg -- because it is easy for the decimal not to be seen and the dosage read as "5 mg", another order of magnitude problem. So the rule is to always use a preceding zero when the dose is less than a whole unit -- 0.5 mg.

You have to use some imagination for these next two, but remember, some people have really bad handwriting! An international unit was often called an IU. With bad handwriting, this could be read as "IV" (intravenous) or "10" (ten). The solution: Write out "international unit." Ditto for unit, or U, which could be mistaken for 0 (zero); 4 (four); or cc (cubic centimeter). So, it is now written out as "unit".

For frequency of medication, the old QOD, every other day, is gone. The "O" can be confused with a period, as in Q.D (once per day). What's the new terminology? -- "every other day."

When these requirements were first introduced, there was substantial resistance from some in the profession. That is pretty much past now, as people have become retrained, but there is occasional backsliding. I hope the medical schools are teaching the new rules so we don't have to retrain all those interns!

What Works -- Part 6 -- Triggers

If you are a patient in an academic medical center, who is watching over you in the middle of the night? Chances are it is a young nurse or intern, among the least experienced people in the hospital. Do these folks have the judgment and experience to respond to potential patient instability?

So imagine, on a regular late night visit to the room, a nurse notices that a patient has developed a fast respiratory rate, a drop in blood pressure, a drop in blood oxygen saturation, or a drop in urine production. S/he needs to make a decision about whether to call the intern. When the doctor returns the call, s/he needs to decide whether or not to actually come to see the patient -- or to just make treatment recommendations over the phone. Then, the intern needs to decide whether or not to wake up a more senior resident or the attending doctor in charge of the patient -- or to make the decisions around changing the plan of care autonomously.

The next morning, after rounds, the attending or another senior physician arrives and decides that there is a need to change the patient's treatment regime or even move the patient to the ICU because of a severely deteriorated condition. Hours of proper attention have been delayed, and the treatment plan now has to make up for lost time.

Or worse, before the attending arrives, the patient suffers cardiac arrest and a "code blue" is called to resuscitate the person.

Extreme example, maybe. Exaggeration of the usual mode of care, no.

When a patient on a medical or surgical unit becomes unstable, early intervention can be very important. Knowing this, the Institute for Healthcare Improvement has recommended that hospitals deploy a response team at the first signs of a patient's decline. But, how do you make sure this happens in the middle of the night? Standardizing a response makes a lot of sense when you think about the complex communication systems that exist in most hospitals.

In 2004, a series of events led us to recognize the need to change. First, a journal entitled Critical Care Medicine published an article in April on rapid response teams that caught the attention of our ICU doctors. Next, we had two very serious adverse events in which well-meaning, very involved junior providers did not recognize the tempo of patients' deteriorating condition. Our folks analyzed those cases and concluded that the care patterns for "acutely decompensating inpatients were complicated, sometimes disorganized, and had multiple single-point failure modes." Among other things, we conducted a survey of our house staff (i.e, the interns and residents) and found that they would contact attending physicians for many acute patient events only about 25 percent of the time. This suggested an area for major improvement.

(But first, an aside. Why wouldn't an intern or resident call the attending physician? Part of the mentality of medical training is an overstated belief that you don't really learn unless you do it yourself. Young doctors often believe that it will be viewed as a sign of weakness to call for help. Their senior residents reinforce that belief, based on their own training.

Another factor is the outright fear of calling an attending physician at 2:30 in the morning and getting the following response: "You woke me up for THAT?! What did they teach you in medical school anyway???")

Here is how things are today at BIDMC. The nurse notices that the patient has developed a certain condition, based on a standardized set of criteria (Triggers). The nurse is required to call the doctor, the senior nurse in charge and the respiratory therapist -- and they all come to see the patient. They collaborate on a plan of care for the patient going forward. Regardless of the time of day or night, the intern/resident then calls the attending doctor in charge of the patient to let him/her know that the patient has "triggered".

The standard set of Triggers we use are based on changes in heart rate, blood pressure, oxygen saturation, urine output, an acute change in the patient's conscious state, or marked nursing concern. Wait, what is this one called "marked nursing concern"? This means that if the nurse has any concern whatsoever about the patient, based on observation or instinct, s/he is authorized to call a Trigger. It turns out that this last criterion is just as valid an indicator of patient distress as the more quantitative measures. (This will come as no surprise to nurses or to those of us who highly respect their judgment!)

So how much of a difference has this made? Over the course of the past year we have have observed significant reductions in "code blue" cardiac arrest events and a significant reduction in relative risk (a 47% decrease) of non-ICU death for our patients. We are also learning a lot about teamwork, communication and systems of care as a result of closely reviewing our responses to Triggers that are called.

One side effect of this improved coordination of care and the decrease in frequency of "code blue" events is that we now need to use our simulation center to train more of our interns, residents and nurses so they can get enough experience resuscitating patients! What a lovely problem to have . . . .

Thank you, social workers

I sometimes say that our doctors and nurses cure the patients, but the social workers heal the families. The Social Work Department has a pervasive presence in the hospital by assisting patients and families with a huge range of problems. Social workers are the unsung heroes in many regards, helping patients and families through extremely stressful situations.

But beyond patient consultations, the department also is involved in many special programs that are highly regarded: Celebration of Life; DriveWise; more than fifteen support groups for patients and families; the Room Away from Home program, annually providing housing for hundreds of needy out-of-town patients and families; Cancer Navigator, a program designed to assist patients of color to access and utilize cancer services; the patient special needs fund; and the Patient-to-Patient, Heart-to-Heart volunteer program for cancer patients.

Several years ago, our hospital was in financial meltdown, and the social work department was a target because so many of its services were not reimbursed by insurance companies. In fiscal year 2000, the department had about 43 FTEs. By 2002, it had been cut to 28, a reduction of 15 FTEs or 35%.

Even though we were in dire financial straits, I decided that we needed to enhance our social work staff and program. We needed more people, and we needed to put more money into professional development programs to maintain the expertise of our staff.

In our 2003 Strategic Plan, I challenged our Board to expand this program and other programs that had shrunk over the years but that were important to help maintain our reputation for warmth, caring, and compassion, and to prove our commitment to the neighborhoods of Boston. As I look back at it now, I made what was a rather pushy request:

The cost of [expanding these programs] rises from approximately $500,000 in fiscal year 2004 to $1.3 million in fiscal year 2007. Our strategic recommendation is that the Board of Directors commit itself and the other governing bodies to cover these costs on an annual basis from unrestricted donations. . . . Just as the Board of Directors is expecting every doctor, nurse, and administrator to make a full-fledged commitment to the success of BIDMC, so too should it be willing to make as strong a pledge of behalf of the lay members of the community who are given governance privileges for this institution.

The response from our lay leaders was immediate and unequivocal. I had asked them to increase their annual gifts by 15% in the first year to cover the new program costs. Instead, they responded with a fifty percent increase in donations!

This was a moving and incredible statement of support from our lay leaders. I believe their response emanated from a desire to express their appreciation to the staff and to say in part, "Thank you, social workers."

Data, data, everywhere

A good and thoughtful follow-up on the Jayco issue this morning by the Globe editorial staff. The final line:

Patients need a reliable source to tell them which institutions do the most to minimize errors and correct those that occur, and have the best outcomes at treating disease.

Who can disagree? But, here is the upshot. There are already tons of public reports about hospital performance on a variety of metrics. Two problems: (1) They are based on administrative (i.e., claims data) rather than more accurate clinical data; and (2) they are way out of date. These numbers might as well not be published at all -- in terms of the useful information they provide to referring doctors and to patients. Think about it, are you going to make a decision about treatment based on numbers that are two years old? The numbers remind me of the accurate, but useless, answer given by the apocryphal student in the exam question displayed above.

The writer says that my proposal for self-reporting is "is no substitute for comparative data from an unbiased source". It is not a substitute, but it is available in real time and it is based on exactly the same data each hospital uses to make decisions about clinical improvement.

I have now heard a lot about the lack of comparability across hospitals. For example, I am told that each place measures central line infections differently. A simple solution: Normalize the results. Let's set June 2006 as the base period for all hospitals, with a value of "1". Each month, show whether the chosen metric has risen or fallen relative to "1". That way, the public can see whether things are getting better or worse in that hospital.

You know, what is striking about this "debate" is the lack of criticism I have received on this blog about this kind of proposal. Indeed, most comments have been quite favorable. If you folks out there who think it is stupid, useless, or otherwise bad would just submit comments (even anonymously), all of our readers would have a chance to judge for themselves. Are you that reluctant to engage here on the issue for fear of giving credibility to this medium (i.e, a blog)? Please: I won't be offended by disagreement. We work in academic medical centers, where open discourse is to be encouraged and treasured. (See the marvelous comments below on an even more controversial subject to understand how vibrant the conversation can be.)

All right, call the MSPCA: I am beating this horse past death!

And a note to the interns, mentioned below: Please read through the various postings and comments on this topic. It will be the major issue facing hospitals during your residency and after. Perhaps your arrival in our institutions will help us all come up with better solutions!

There is no mystery in Mystery Shoppers

Nobody wants to be thought of like Ernestine, Lili Tomlin's rude telephone operator! We have been trying to improve customer service in our hospital. Academic medical centers are often not great at helping patients navigate their way through their clinics, and we are hoping to set a higher standard at our place.

We have borrowed the concept of "mystery shoppers" from other service industries. We train people to pretend they are patients or patients' family members, and then we send them into a clinic to make observations and take notes. We also do this with our call centers, so we can see how our people serve the public on the telephone.

We then share the results with the chiefs, the clinic managers, and, of course, the front-line staff. As in the case of clinical improvements, we do not engage in the "blame game", but rather we use the shoppers' reports to offer helpful suggestions to people. Often, too, the problem is not with the front-line person, but there is some systemic problem behind the scenes that needs to be fixed.

Curious? OK, here are samples from two of our clinics. In the first, the Emergency Department after a very busy day, you can see areas for improvement. In the second, the more sedate Infectious Disease clinic, things look pretty good. (Excuse the stream-of-consciousness feel of the reports. We ask our shoppers to maintain a running commentary of what they see and hear.)

Remember, these take place in the waiting rooms -- not the patient care areas. These particular surveys are designed to review service quality, not the quality of the medical care offered in the exam rooms.

The Emergency Dept wait area is comfortable, well designed area with corner views of the Medical Center area. It is well lit with natural Department and fluorescent lighting. Area is modern but messy and dirty. On today’s visit, which occurred late in the day after the ED had been on diversion, there was litter strewn about, empty soda bottles on side tables and food crumbs all over the floor. Entryway is free of obstruction to passage. The first group of seating is reserved for patient triage. Security greeted me as soon as I entered and told me where to wait. The guard would not allow me close to the desk area unless I required nursing assistance. I wasn't sure how patients were actually checked-in, since the security guard primarily interacted with the patients who arrived and he asked them to have a seat. There is a large C shaped desk that seemed to be shared by security and nursing that was part of the wait area. Once I took a seat further back in the area I was not acknowledged again. Co-payment and referral signage were not posted. Patient Rights and Health Care Proxy information were available in various languages on a nice turnstile rack that needed restocking. There was no PRC information. Infection control information was posted throughout the area. One box of tissue was on a side table; no Calstat bottles. There is a vending machine in the back along with restrooms. Restrooms needed cleaning. General appearance of the waiting area was messy. At the check-in desk there were soda bottles and a coffee cup. Area was lacking in entertainment reading; there were a couple of old, torn magazines scattered around. Addresses were visible and there were no instructions for coping. Several informative brochures were neatly arranged on a long side table. Area had a plasma screen television which was on, tuned to nightly news. RN triage area is further along the large C shaped desk area and provides for privacy. Ten patients arrived within fifteen minutes of one another and overall the staff managed the patients efficiently through the triage area. Information regarding waits was not provided once in rear wait area. Staff was pleasant and courteous. Name tags were visible. Nursing personnel wore lab coats. At times personal conversations occurred between security and nursing but were not overheard. The security guard was chewing gum. I asked security (my only option) for directions to the cafeteria. He instructed me on how to proceed to the Farr Building then said -- "you’ll find it". My assessment of customer satisfaction in this clinic would be 3.0 -- good patient flow system but lacking in friendliness.


The Infectious Disease Clinic shares a good size waiting area with the Travel Clinic. Area is up to date, warming, inviting and comfortable. When first entering the French doors you see the water cooler and excess bottles. It does not pose an obstruction to passage. The reception assistants made eye contact with me immediately. I chose a seat in the corner as she inquired how they could be of assistance. There was no posted signage on the walls or framed on desks. Kleenex was on all side/reception tables. One Calstat dispenser was on the wall by the entrance area; no others visible in the area. Appearance of the waiting area was more than satisfactory -- clean, neat and orderly. Plenty of updated reading material for entertainment and infectious disease related. None of the magazines had stickers with copying instructions. Names were blacked out. Most of the magazines were addressed to MDs in the Lowry Building. There was no television in the waiting area, more than adequate seating and no clutter on the reception desk. A small bouquet of fresh flowers was in the room along with several healthy appearing plants. One of the staff members was caring for the plants and making pleasant small talk with those waiting. I would rate the area a strong 5 in the customer satisfaction area. There were 3-4 attendants behind the desk at all times. They spoke quietly, discreetly and professionally. None wore lab coats but were dressed in professional work attire. All had their name tags visible but I was unable to obtain names without further calling attention to myself. Twice 2 different attendants approached me in the corner and asked how they could be of assistance. Therefore, I was addressed/approached a total of 3 times in 22 minutes. One gentleman was seated several chairs next to me the entire time who was never approached. But he seemed content. One patient was identified by their first name but it was clear they had an established relationship. Physicians in the area wore lab coats, walked in the wait area to greet their patient, shook hands and lead them to the exam area. A physician approached one gentleman to offer an explanation of the wait after I was there approximately 15 minutes. As I was exiting the area I asked directions to the bathroom -- interrupting a work related phone conversation. The attendant was extremely pleasant, giving specific directions. As I was exiting the area a physician who heard my request offered additional instructions.

Better grades on VAP

Good progress on implementing steps to reduce ventilator-associated pneumonia. Recall my story in January on this topic.

Our goal is to make sure we are carrying out the five-part Institute for Healthcare Improvement "bundle" to reduce the incidence of this disease. As I noted back then, if you want to reduce VAP, you institute this bundle of steps. But, like your toughest sixth grade teacher would say, "There is no partial credit!" Unless you carry out all five steps, you do not get a perfect score. Here are the monthly stats:

April 06: 83%
May 06: 74%
June 06: 82%
July 06: 80%
August 06: 76%
September 06: 86%
October 06: 92%
November 06: 85%
December 06: 87%
January 07: 93%
February 07: 98%
March 07: 100% (to date)

In addition, our folks are working on instituting improved oral hygiene for these patients, another aspect of reducing the VAP problem. There we have gone from 58% (in April 06) to 91% (in March 07).

But, I want to make clear that these are process metrics. This differs from the central line infection story I have mentioned elsewhere, where we measure the actual number of cases. My folks tell me that actual VAP rates themselves may not be accurate because it is the most subjective of all nosocomial infections and the most difficult to track. We are currently engaged in refining our methodology for actual VAP surveillance.

Finally, I want to point out that this effort requires a variety of specialties to work together -- several types of doctors, nurses, respiratory therapists, and pharmacists. In this case, 14 people serve in the VAP Working Group. They develop protocols, data monitoring, and training programs and then have to persuade dozens of other people to engage in the program on a sustained basis. As I have noted below, the unique environment of academic medical centers creates interesting challenges in process improvement. These folks deserve a lot of credit for what they have accomplished, but the real test will be to see if they can keep it going. Knowing the people involved, I am optimistic.

These things happen

I was reminded of this by our Chief of Medicine. In the movie, It's a Mad, Mad, Mad, Mad World, Ethel Merman, playing Mrs. Marcus, says:

Now what kind of an attitude is that, 'these things happen?' They only happen because this whole country is just full of people who, when these things happen, they just say 'these things happen,' and that's why they happen! We gotta have control of what happens to us."

I am struck by the relevance of this to running a hospital.

Several years ago, we had that attitude in our hospital with regard to certain types of medical outcomes. For example, we were content with our level of central line infections because we were below the national average. After all, these things happen. Then our chiefs of medicine and surgery said, "No, they don't have to happen. When they happen, people die. We are going to insist that we achieve zero central line infections." And then they got to work. As I have noted below, it is not an easy problem to solve, but it is worth the effort, and you can improve.

One way to encourage organizational improvement is to publicize the results of your program. I have done that below for our hospital, and I have made the suggestion that others in the city could do the same. As I noted, I did not make the suggestion for competitive purposes -- after all, I don't know if our numbers are better or worse than those of other hospitals -- but because public exposure of all our efforts will drive all of us to do better. Also, it will build, rather than erode, public confidence in the academic medical centers in our city.

The response, as you have seen from the press reports, ranges from simple recalcitrance to technically sophistic arguments about comparability of data. Please, does anyone argue that the goal should not be zero? If it is zero, it does not matter whether the data is measured in cases per thousand patient-days, cases per thousand catheter-days, or just the raw number of cases.

We all keep track of these numbers in some form or another. We could easily post them in real time voluntarily on a website maintained by the state or an insurance company, along with our own explanations of how and what we measure. (And perhaps, over time, we will agree on what single metric is most useful.)

People can and will understand this. They already spend hours on the Internet reading medical websites. Why do we give them so little credit? It will demonstrate to the public that we care about this problem, and will show our individual progress towards our ultimate goal.

Finally, it will enhance the reputation and credibility of all of the academic medical centers, two aspects of our character that will be more and more under siege because of the broader problems of the health care system.

Addendum (November 2010): Here's the video clip from the movie: