Monday, May 12, 2008

Triggers happy

A year ago or so, I wrote about the introduction of our Triggers Program, a rapid response team approach to patients on medical floors who might soon decompensate or have other serious changes in their condition. The program has been incredibly successful in reducing mortality and morbidity. In fact the number of "codes" on our floors has gone down so dramatically that residents now need to practice emergency resuscitation mainly in the simulation center because so few actual patients need it.

I recently asked a couple of our folks who were deeply engaged in the design and
implementation of this program -- Dr. Michael Howell and Patricia Folcarelli, RN, Ph.D. -- to tell me what lessons have come out of the last year's experience with Triggers. Here is what they sent me. I offer it in the spirit of sharing information with people in other hospitals.

In the year after implementation of the Triggers program, one of the major focuses of our reviews was on patients who had major adverse events happen in spite of the Triggers program. When these adverse events occurred, we tried to understand the factors that contributed to them even being possible in our organization. A few months after Triggers began, we began to notice some patterns. Here are some examples of the things we learned.

Oxygen is not a utility

Patients in the hospital sometimes need extra oxygen. Low oxygen levels in the blood can be due to pneumonia, heart failure, or a number of other problems. Surprisingly, extra oxygen usually does not help with the feeling of shortness of breath, but rather prevents further problems from not getting enough oxygen to vital organs. We found that, in many cases, providers often treated oxygen as a utility -- like the water that comes out of the sink – rather than as a drug used to support a feeling organ system. (The members of our Triggers Steering Committee had worked in about twenty other hospitals total, and we all felt it was the same in every hospital in which we’d ever worked.)

We saw a pattern in which providers would repeatedly increase the amount of extra oxygen that was being provided to patients. We often monitor the oxygen level in the blood through a noninvasive device -- as his number was normal, providers felt reassured – not taking into account the fact that the patient was needing higher and higher levels of artificial support to keep this number at the “right” level.

In fact, interns would sometimes round in the morning and would find their patients on oxygen with no explanation, and the patient had been breathing room air the night before. Sometimes, neither the nurse nor the intern knew why the patient got put on oxygen; it had happened overnight and was viewed as an unimportant event.

As a result of this we conducted a Failure Mode Effects and Criticality Analysis, a tool used in the military and industry to understand points at which complex systems are likely to fail, and implemented substantial changes in the ways that we order oxygen, in a way that patients are monitored from a respiratory standpoint. We also introduced physician, nurse, and patient care technician education on this matter.

Aspiration risk

We also learned that aspiration was a bigger threat to patient safety than was usually appreciated. When physicians and nurses talk about "aspiration" they are talking about when a patient swallows something the wrong way. This can be the person's own saliva and secretions or, more commonly, can occur when they try to eat or drink something. Since the mouth is usually full of bacteria, this can lead to pneumonia; sometimes, the person actually swallows his or her stomach acid in the lungs, which can lead to very severe chemical injury to the lungs. In some cases, aspiration leads to death. For this reason, when we think that someone is at high risk for aspiration, we put them on “aspiration precautions." This means that nurses, patient care technicians, and physicians are all alerted to the increased risk of this problem. In addition, we put a sign up on the patients at the patient's bedside to warn visitors and those providers who may be seeing the patient before seeing the chart.

As we dug a little deeper into some of these cases, we learned that patients sometimes aspirated food that their families brought in. Family members obviously did this out of love, but it sometimes led to very severe consequences for their loved one. When we tried to figure out why this happened, we found that our warning signs depended heavily on written English, rather than on easily interpretable symbols. This meant that if family members came to visit and English was not their first language, or if they had trouble reading English, we might not convey the right information to them. In coordination with a provider education campaign about the risks of aspiration, we therefore redesigned our signage to overcome these barriers – by using multiple languages and universal symbols (think Mr. Yuck!) that were likely to be interpretable even if the family member was unable to read the sign -- see above.

Who does what?

As inpatient medical care has become more complex, more people are needed to provide it. For example, our nurses do a number of safety checks as they're preparing various medications because these medications have inherent risks. There is also substantial amount of documentation that nurses have to do for safety, compliance, and legal reasons. This means that nurses need extra manpower to get work done. Most hospitals, therefore, have a group of providers who are variously known as nursing assistants, nurse’s aides, or patient care technicians. These providers are trained by the hospital, and sometimes by external schools, but are not licensed in the same way that nurses and physicians are. Patient care technicians may check vital signs, help with turning patients, assist with toileting, etc. In our hospital, for example, many of the routine vital signs are taken by patient care technicians. The Triggers program taught us a few things about patient care technicians and their relationships with our other existing systems of care. In particular, when we did our initial education for the Triggers roll out, we forgot to include patient care technicians in the educational campaign. This was a huge oversight, which we quickly learned when we would see patients who did not Trigger even though they had abnormal vital signs. Why didn't they Trigger? They didn't Trigger because we forgot to provide education to this very important a set of providers in our institution. Once we had included them in the educational campaign, this mechanism of Trigger failure essentially vanished.

We also learned that what patient care technicians do on any given floor is extremely variable. We therefore began a program to help standardize the scope of practice for patient care technicians at BIDMC.

Unintended consequences of improving patient satisfaction

A few years ago, as we tried to improve patient satisfaction, we changed the way that patients order their hospital food. The program was called “At Your Request" and let patients call up to order their meals from a menu of options – at essentially anytime they wanted to eat. (From a practical standpoint, this works a lot like room service: you call and order your meal, and it shows up half an hour later.)

However, this turned out to be another way that patients who were at high risk for aspiration (see above) could get food that was unsafe for them to eat. A patient on aspiration precautions, for example, could literally call and order a hamburger, which would generally be delivered, warm and tasty, a half hour later. When we saw events related to this, we redesigned the process by which food was delivered, creating an electronic Diet Dashboard and directing the delivery of all food for patients on aspiration precautions to the nursing station. (Sometimes, patients at high risk for aspiration just need help eating food safely, which we can now provide.)

If the nurse is worried, you should be worried too.

This is an example where our analysis confirmed something we already believed to be true.

The Triggers Program has various specific criteria mandating a response from providers. For example, if the pulse rate is acutely greater than 130 beats per minute, a Trigger is called and the team responds. However, we have one criterion which is much more subjective: "marked nursing concern." When we implemented the Triggers program, many physicians were very nervous about giving this criterion. They were afraid that they might be called in the middle of the night for things that weren't really important, and that nurses might use this as a weapon if they did not like the physician or if they disagreed with the plan of care.

Well, it turns out that nurses use this Trigger quite judiciously – only 15% of our Triggers are called only for nursing concern. (In another 27% of cases, nurses express “marked concern” but the patient also meets other criteria simultaneously.) It also turns out that if nurse has “marked nursing concern,” it means you’re really sick. The in-hospital mortality rate for a patient who has a Trigger called for “marked nursing concern” is 10.7%.

This is roughly twice as bad as showing up to the Emergency Department with a heart attack. Literally.


Ron Newman said...

Can you show us a picture of the "do not bring outside food to this patient" signs that you put up? Thanks.

Anonymous said...

Excellent review and a good example of the evaluation phase of performance improvement. I, too, was skeptical of the "nursing concern" indicator - but here is good evidence that it works. Kudos to the nurses. Now go ahead and call out those SPIRIT problems too - a year from now you'll be vindicated again!

nonlocal MD

Anonymous said...


Done, as requested. Thanks for the suggestion.