Reducing Ventilator-Associated Pneumonia

Some patients go to the hospital because they have pneumonia. Other people go to the hospital for other reasons (e.g, stroke), are put on ventilators, and get pneumonia. We call this ventilator-associated pneumonia, or VAP. It is a big problem:

• It is common, with 10 to 20% of patients ventilated for two or more days;
• It is lethal, roughly doubling the risk of death;
• It is expensive, adding $20,000 to $40,000 in extra costs per case.

The good news is that it is often preventable, and we could be pretty good at preventing it if we took the right steps all the time. Our good friends at the Institute for Healthcare Improvement suggest the following "bundle" of steps to help avoid VAP:

1. Elevation of the head of the bed;
2. Daily "sedation vacation", i.e., some removal of sedation medication;
3. Daily assessment of readiness to extubate, i.e, don't keep the breathing tube in longer than necessary;
4. Stress ulcer disease prophylaxis, to reduce the risk of upper GI bleeding;
5. Deep venous thrombosis prophylaxis, to prevent formation of embolisms.

So, if you want to reduce VAP, you institute this bundle of 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.
We started working hard on this problem last year at BIDMC. Why? Because we looked at our rate of this disease, and we were not pleased. Here are our compliance scores on the IHI bundle, after lots of analysis, training, and follow-up:

FY06 Q3: 79%
FY06 Q4: 81%
FY07 Q1: 88%
FY07 Q2: 92% (only includes January, through today)

This looks pretty good, right? It appears that we are making constant improvement. Not so. Unfortunately, the quarterly figures mask monthly variations:

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: 92%

Still, the trend is good, but the difficulty of carrying out the full bundle for all patients is real. For example, we have virtually 100% compliance with stress ulcer disease prophylaxis; but we do not always carry out a daily assessment of the readiness to extubate. On that metric, we have ranged from 88% to 98%. Sometimes, even when you know what you would like to do, the patient's condition or other exigencies make it impossible. Sometimes, even when you know what you should do, it doesn't get done -- for a variety of reasons: training, follow-up, schedules, competing demands of other patients.

Sometimes, there are unexpected reasons. At one point, we could not elevate some beds properly because other patient-related equipment was in the way! (We fixed that. And, yes, we bought contractors' protractors, the same ones used in construction to measure the angle of a pipe bend. How else will you know if the bed angle is correct?)

IHI has published stories of places with great success in this arena. Congratulations to those hospitals. We hope to be in one of those stories some day.

Why can't we do that?

A recent note from our chief operating officer to several of our clinical chiefs:

I came across the NEJM from 12/28/06 and the article and editorial on catheter-associated blood stream infections done in Michigan (my medical school alma mater!)

In that study of 108 ICU’s in Michigan, the institution of a set of evidence-based interventions reduced the median rates of infection per 1000 catheter days from 2.7 to 0 and the mean rates from 7.7 to 1.4. These impressive gains were held through 18 months of follow-up. If more than half of these ICU’s can virtually eliminate these infections, it seems that we should be able to do so as well. We had 2 months in the last year when we achieved this goal, but the last two months showed a sizable bump. Are we doing everything possible to eliminate these on a sustainable basis? What will we do differently going forward to hold the ‘zero rate’ for every month?

For the record, here are the numbers for BIDMC, updated from my posting of December 17, below.

Month ----- Infection Rate
Oct 05 ----- 1.67
Nov 05 ----- 1.28
Dec 05 ----- 2.43
Jan 06 ----- 3.07
Feb 06 ----- 1.40
Mar 06 ----- 1.07
Apr 06 ----- 0.00
May 06 ----- 0.59
Jun 06 ----- 1.15
Jul 06 ----- 0.57
Aug 06 ----- 3.03
Sep 06 ----- 2.50
Oct 06 ----- 0.00
Nov 06 ----- 2.38
Dec 06 ----- 1.87

We are serious about this and, like Michigan, we will share any lessons learned with others in the medical community. Recall that we engaged in an intensive training and improvement program in this arena, and the result we want is tantalizingly within reach. Nonetheless, as noted by our COO, we are not yet "best in class," which is our goal.

Here is what we are doing for now: Every single infection is now viewed as a sentinel event, for which we conduct a root cause analysis and from which we learn how to do better. Stay tuned to see how we progress over the coming months.

What Works -- Part 5 -- Team Training

We have all heard stories of cockpit behavior in an airplane that causes a crash. The navigator or first mate says to the pilot, "Watch out for that mountain." The pilot ignores the advice, and the aircraft ends up in flames. This kind of thing can happen in a hierarchical environment.

The same kind of thing can happen in an operating room, where the surgeon is the "pilot" and the nurses and anaesthesiologists are the support crew. Doctors, like pilots, are trained to be in charge and to make split-second decisions.

A few years ago, a series of errors and bad communication in our obstetrics department resulted in the loss of a baby and almost resulted in the death of the mother. For a department that had always prided itself on providing extraordinary care, the event was a shock and caused an intense self-evaluation.

Using the experience of the military -- indeed from those cockpit situations --the Department proceeded on a full-fledged series of courses in team training.

This was not a simple seminar or two. It was a process that took many months. After all, it had to break down barriers and behaviors that had taken years to develop. Nurses had to feel comfortable offering suggestions to doctors, and doctors had to learn how to hear the nurses' comments.

Here's an article that describes the whole thing. I urge you to take the time to read it. Our OB staff would tell you that it has changed their view of practicing medicine. They would also tell you that it has created unusual bonds of collaboration and friendship in their department, even for a group that had always had a strong group ethic. Most important, the program has actually had a measurable difference in clinical results. Our folks now participate in programs across the country to spread the word.

I wish I could tell you that we have taken this experience and have infused it throughout our own hospital. We have not, at least to the extent I would like. Not that we are not trying, but it turns out that the culture of each department and each division is a bit different, even within the same hospital. So it takes longer than you might expect. What might work in OB needs to be modified to work in surgery or orthopaedics. Even within surgery, what might work for the pancreatic surgery group -- see the November 27 discussion below on Whipple procedures (What Works -- Part 2) -- might not be quite right for the transplantation group. Like other medical centers, we are still feeling our way through this issue of the diffusion of practice improvements.

First, kill as few patients as possible

That was the name of a humorous book by Oscar London, but there is a serious side to the concept:

For years, Don Berwick and his colleagues at the Institute for Healthcare Improvement have been proselytizing and working to improve care in the nations' hospitals. They conduct important research and offer training programs for all types of hospitals, medical staff, and administrators. Recently, they have offered a metric that is the grandaddy of all metrics, the
hospital standardized mortality ratio. This is a disease and procedure based, risk-adjusted single number that tells you how you are doing in term of deaths compared to the average and compared to other hospitals. According to IHI, "the HSMR, appropriately adjusted for multiple variables such as population characteristics and diagnoses, provides an essential starting point in improving care and reducing mortality. "

In shorthand, for us lay people, the metric gives a sense of your likelihood to die at a specific hospital compared to other hospitals. If your hospital has a value of 1.0, it is average. If you have below 1.0, it is better than average. If you have above 1.0, it is worse than average. [Note: See correction to this statement in my comment below.] As with all metrics, you can quibble with the components and argue with the calculations, but it is as powerful a tool as I have seen. It is rapidly becoming the touchstone for many hospitals as they review their safety and quality programs.

IHI offers this tool to help people do better. It is not meant for advertising purposes or punitive purposes. As they note: "Many hospital deaths could be prevented if all the factors that contribute to them were better understood. Each hospital death provides an opportunity for learning -- by understanding and addressing local conditions that contribute to mortality."

We recently asked a group of outside experts from places with the strongest national programs to review BIDMC's progress in patient safety and quality. We received a good grade, but we also received a number of thoughtful and helpful suggestions for improvement. We have high aspirations. Our goal is to set audacious targets for improvement in overuse, underuse, misuse, and waste in the care of patients -- to set plans and milestones for doing so -- and to manage towards those targets.

Academic medical centers have a special responsibility in this regard, to create within the safety and quality program an academically rigorous examination of what works and what does not in various health care settings. I have given you a few examples in the postings below, entitled "What Works". But no single hospital has a monopoly on ideas when it comes to this field, and the first step is for all of us to disclose publicly how we are doing.

This HSMR number is not published anywhere unmasked by name, but if you contact IHI they will give you your own data, which is what we did. To relieve your suspense, 0urs is 0.71, which just puts us in the top ten percent in the nation. (Frankly, if a Harvard-affiliated academic medical center were not better than average, everyone would have a reason to wonder why.)

I wonder if my academic medical center colleagues in Boston and around the country would similarly be willing to post their HSMR number publicly on their own and to authorize IHI to maintain a publicly available list on their website. With a national debate swirling about the cost of care and value of academic medical centers, what would be more powerful than a grand display of openness about our progress in trying to kill fewer people?

What Works -- Part 4 -- Central Line Infections

Central line-related bloodstream infections are a serious problem in hospitals. A central line is a port installed directly into a major blood vessel to permit a catheter to be used for the quick delivery of medication for patients in ICUs and in other settings. Because of the direct connection to major blood flow, an infection associated with the installation will flow quickly into the blood stream and to major organs. This article from the Centers for Disease Control attributes a mortality rate of 12 to 25 percent (!) for each infection -- not to mention increasing costs by about $25,000.

The Institute for Healthcare Improvement likewise notes that "up to 4,000 catheterized ICU patients die each year in the US from avoidable infections and organ failure (sepsis) related to central venous catheters (CVCs). . . . Forty-eight percent of ICU patients in the US have central venous catheters, accounting for 15 million central-venous-catheter-days per year in ICUs. . . . Within this population, studies indicate an estimated 4% to 20% (500-4,000) of patients will die from catheter-related bloodstream infections."

Like others in the country, the medical leadership at BIDMC decided that our current rate of central-line infections was too high and set about to change it. When we started, our average rate of central line infections per thousand patient days in the ICUs was about 3. This was better than what we often see nationally, but our doctors were impatient to improve it. After all, each case has a high potential for serious patient injury or death. So the goal is to get to zero.

This turned out to be a multi-faceted problem. Central lines are often inserted by residents who have been trained how to do the insertion by other residents. (Dr. Atul Gawande provides a vivid description of this learning process in his book Complications: A Surgeon's Notes on an Imperfect Science.) Beyond the insertion process, decisions must be made about how long the line should stay in, and how often it should be maintained. Very often, there are only informal rules of thumb in a hospital for these determinations -- and there is often wide variation even within a single hospital.

Our folks set about to make this process more rigorous, analytical, and controlled. Sessions were held among surgeons, medical doctors, anaesthesiologists, nurses, and residents to reach a consensus on the proper method for inserting a central line. A specific kit was designed, so that anyone inserting a line had the full complement of supplies at hand. Detailed rules were established for the protocols surrounding maintenance of the line and its withdrawal. And, a system was set up so that every single infection that occurred would be analyzed to determine its cause -- so corrective measures would be taken going forward.

Here are the month-to-month results for the first year of the program:

Month ----- Infection Rate
Oct 05 ----- 1.67
Nov 05 ----- 1.28
Dec 05 ----- 2.43
Jan 06 ----- 3.07
Feb 06 ----- 1.40
Mar 06 ----- 1.07
Apr 06 ----- 0.00
May 06 ----- 0.59
Jun 06 ----- 1.15
Jul 06 ----- 0.57
Aug 06 ----- 3.03
Sep 06 ----- 2.50
Oct 06 ----- 0.00

As you can see, the figure goes up and down, although progress is good. The key thing is that every single case of infection is analyzed thoroughly, with the results shared across the broad range of hospital staff in the ICUs. What goes wrong? As many things as there are people. For example, one day, our chief of medicine happened to go by as another member of the staff was not following the protocol. When he pointed it out -- and none too gently! -- the person was embarrassed and really had no excuse for doing it wrong. So human nature often comes to play. Sometimes more technical factors arise. Regardless of the cause, each case is used to reinforce the program.

With about 1600 ICU patient days per month at BIDMC, the difference between an infection rate of 0.0 and one of, say, 2.5 is 4 actual people. Over the course of a year, that same difference amounts to 48 people who either get or do not get an infection. Applying the CDC's cited mortality rate of 12 to 25 percent, the difference amounts to saving the lives of 5 to 12 people -- just at our hospital.

When you look at numbers like those, you can see why our medical staff -- and people around the country -- are rabid about making this improvement real and permanent. Doctors and nurses devote their lives to alleviating human suffering caused by disease. They are heartbroken by the thought that their own well-intentioned actions might lead to death, and they are driven to get better and better at what they do.

What Works -- Part 2 -- Clinical Pathways

The Whipple procedure is a complicated and difficult surgical procedure. It is also called a pancreaticoduodenectomy, and it generally encompasses the removal of the gallbladder, common bile duct, part of the duodenum, and the head of the pancreas.This operation was first described by Dr. Alan O. Whipple of New York Memorial Hospital (now Memorial Sloan-Kettering).

The surgeons at BIDMC have developed a "clinical pathway" to guide themselves, related physician specialists, nurses, case managers and other involved in performing Whipples and taking care of patients before, during, and after this surgery. The clinical pathway is intended to assist physicians in clinical decision making by describing a range of generally acceptable interventions and outcomes. In other fields, it might be called a "decision tree." The guidelines attempt to define practices that meet the needs of most patients under most circumstances. While the physician must remain alert to deviations from the expected, the use of the clinical pathway can bring greater predictability to the entire treatment process in many cases.

The introduction of the clinical pathway for Whipple procedures at BIDMC has had very positive results. Here is a summary of the data pre- and post-clinical pathway.

Pre-clinical pathway period: October 2001-January 2004 -- 64 patients (42% male)
Post-clinical pathway period: February 2002-October 2006 -- 121 patients (53% male)

Age distribution: Mean for both periods = 64 (comparable range of ages)

ASA class of patients (degree of disease/difficulty)
Pre-clinical pathway: I (1.6%); II (51.6%); III (46.8%); IV (0%)
Post-clinical pathway: I(0.8%); II (39.7%); III (54.5%); IV (5.0%)

Pre-clinical pathway mortality = 1 death (1.6%)
Post-clinical pathway mortality = 2 deaths (1.7%)

Pre-clinical pathway ICU admissions = 8 patients (12.5%)
Post-clinical pathway ICU admissions = 16 patients (13.2%)

Pre-clinical pathway of stay/cost = 10.8 days/$23,536
Post-clinical pathway length of stay/cost = 9.8 days/$19,999

Pre-clinical pathway readmission/reoperation = 4 (6.3%)/4 (6.3%)
Post-clinical pathway readmission/reoperation = 10 (8.3%)/7 (5.8%)
(Both readmission and reoperation are measured within 30 days.)

To put all this into English, after the introduction of the clinical pathway, nothwithstanding a greater percentage of sicker patients, cost and length of stay decreased without negatively affecting mortality, readmission, or reoperation rates.
Patients and families also like the new pathway because they receive a roadmap of what to expect and when to expect it, and they can literally follow their own progress by looking at a chart on the wall of their room.

I don't mean to suggest that we are the only place to have clinical pathways, but they remain less prevalent than you might think. This is but one example to show how effective they can be when well designed and implemented by the entire medical team.

What Works -- Part 1 -- PatientSite

The first in a series of innovations that work (!) and make a difference in patients' lives:

A recent story on MSNBC called "Tired of waiting for the doctor?" made reference to PatientSite, our user-friendly, personalized electronic communications link between patients and their doctors, using a secure website:

Most patients have experienced playing phone-tag to get test results. That’s at least partly because of the traditional paper-based method of relaying information. Test results are transcribed onto paper, then given to the doctor, who then phones or snail-mails them to patients.

At Boston’s Beth Israel Deaconess Medical Center, patients can get test results electronically the same time as the doctor through a private online account called PatientSite. “There is no waiting for paper printouts to arrive by mail,” said Dr. John Halamka. All test results show up on the site, except those involving diagnosing cancer or HIV, “assuming that this news should be delivered in person,” he said.

PatientSite has been up and running for many years, well ahead of most of the industry. Patients like the fact that they can use it for a variety of routine functions -- from requesting prescription renewals, to making appointments and getting referrals, to viewing their own electronic medical records, including medications, allergies and test results, radiology reports and electrocardiograms. Doctors like it, too, because it frees them up to spend person-to-person time with patients on more important matters.

The long-term success of PatientSite also means that it can be used to study other new ways to improve care. Here is an example funded by the Robert Wood Johnson Foundation.

Try out the demo!