Tuesday, August 10, 2010

Reducing Risks During Hand-offs

The hand-off of a patient from one doctor to another is an episode ripe for potential problems. Important data about the patient's condition might not be transferred, and there is also the potential for miscommunication between the caregivers. In academic medical centers, the responsibility for the hand-off is often in the hands of the interns, i.e., the first year residents. As duty hours for residents have become more restricted to avoid overtired doctors, the number of hand-offs that occur has necessarily increased -- by something like 40%.

(The trade-offs between the dangers associated with tired doctors and those associated with increased hand-offs has been well discussed elsewhere, and it is not my purpose here to argue the case. For the former, you find serious medical errors, medication errors, diagnostic errors, car crashes, depression, and burn-out. For the latter, you find longer lengths of stay, medication errors, and more adverse events, especially those associated with communication failures.)

Three years ago, the Risk Management Foundation published an edition of its Forum entirely devoted to the subject of how to reduce risks during hand-offs. It remains a good summary of the issues, and you can view it here.

Last year, one of our Senior Residents, Kelly Graham, decided to use the research phase of her residency to test out some interventions to see if they could reduce the likelihood of hand-off related errors. She compiled the following baseline assessment for BIDMC (which, as noted, was similar to a previous assessment at Brigham and Women's Hospital):

Kelly decided to focus on three aspects of hand-offs: The systems in place, the written communications, and the oral communications. Her hypothesis was that by taking a systematic approach to intervening in each component of the patient hand-off, we could improve the quality of sign-outs, patient safety, and intern satisfaction.

The "prior" that Kelly was trying to change is the age-old system: Interns learn how to do hand-offs on the floors by watching their senior residents. Process improvement folks reading this know that is a recipe for a high degree of variation in practice and for a systematic transmittal through time of bad habits and approaches that increase the likelihood of harm.

So, Kelly's aims were to provide resident physicians and patients with safe hand-off practices; to promote a “standard operating procedure” for hand-off; and to take hand-offs out of the hidden curriculum of medical training and make it part of our formal education process.

On the system side of the equation, she noted that many hand-offs actually did not occur between the doctor leaving the service and the doctor arriving. Instead, an intermediary person often took the information from the departing doctor and later relayed it to the arriving doctor. Like the old game of telephone, this increased the likelihood of flawed information transfer. (In fact, prior studies indicated a loss of 22% of the desirable information that should be passed along at the time of transfer.)

The alternative was to require direct communication between the departing intern and the arriving intern, in a standard location (the house officer lounge). Doctor-to-doctor interaction increased from 25% to 100%.

The next intervention was designed to present a common template of information to be transferred. Pull-down menus on the computer helped to ensure that standard categories would be discussed, and standard language would be used as much as possible to reduce variation in the transmittal of patient data.

And the final intervention, the one that is likely to raise eyebrows among my lay readers, is the idea of teaching how to do a sign-out in the classroom before arriving on the medical floors. Huh?

Well, the baseline assessment was that interns are not prepared for hand-offs during medical school. 91.3% of interns at BIDMC reported no hand-off training prior to residency; and 92% interns nationally report no hand-off training prior to residency. So Kelly designed and implemented a case-based, interactive, sign-out workshop during the interns' orientation.

As the year went along, she surveyed the residents and also kept track of patient data. She reached the following conclusions:

Interns are ill-prepared for transitions of care; “double hand-offs” may reduce work hours slightly, however the trade-off is that they may be unsafe for patients; involving the primary team in the hand-off process has a powerful effect of patient safety and physician satisfaction; electronic templates are reliable tools to ensure sign-outs are complete; and interns respond well to incorporating hand-off training into their education.

And, now look at the clinical efficacy of the experiment. There was a dramatic reduction in adverse events, near misses, and data omissions. In fact, the first two interventions were so powerful that it was not possible to fully evaluate the strength of the last one -- but the training did help to improve interns' job satisfaction.

The interventions are now embedded in our Department of Medicine's system of training and care. The interns who just arrived don't know enough to know that they are doing something different from the past because they never experienced the ad hoc system that was in place before. Congratulations to Kelly and her colleagues for demonstrating how an academic medical center can contribute to the improvement of clinical processes, something just as important as our contributions to basic and translational research about disease.


olaoluwatomi said...

Lovely demonstration of what happens when a process is designed in place of 'business as usual'. It reminds me of Atul Gawande's checklist and shows that medicine is moving from an era of paternalism and 'i cant be wrong' to one where steps are taken to actively reduce errors that are sure to happen. (The last time I checked doctors are humans)

e-Patient Dave said...

FABULOUS! If Kelly drinks Starbucks, I'll buy her a $50 card for the Shapiro lobby!

Hot diggety, smart process thinking applied to a real-world problem.

This is serious IHI Forum material for December. Let's go.

The other thing I was looking for was *bedside* handoffs, comparable to what I've read is effective for nursing shift changes. I'd be eager to hear if that was considered, for the same reasons.

(btw, Paul, this is a case where your automated export-prolog-to-Twitter thing did not do justice to the importance of this post. Pay attention to yer first few words!)

Anonymous said...

Is there a reason a woman initiated this research, instead of a man?! (just kidding).
Hopefully the NIH and other funders of research will come to recognize that this sort of research is equally, if not more, important than basic and translational research; and it will come to be associated with the same prestige.

(female) nonlocal MD

mdspencer said...

Groundbreaking work. It's amazing how much groundbreaking work is about basic medicine or basic science. This is a great example. I hope it spreads internationally.

76 Degrees in San Diego said...

Of course, you fumble less if you keep the ball and run with it, instead of handing it off.....

jonmcrawford said...

@76 - not if you've been running for too many hours, you don't. Shorter shifts are just common sense, I don't want my surgeon to be operating on me at the end of their shift, I want them to be at the TOP of their game.

Bravo to Kelly and all involved!

Anirudh said...

This is an amazing intervention that clearly saves patients.

I would love to see what the template on the computer looked like (ie what categories were included and what the pull down options were for each category).

I think this system should be used in more hospitals and I would love to introduce a modified version of it to our service.

jonmcrawford said...

Had a followup question Paul, did this study find that the information that was being documented in the chart was not sufficient, and the face-to-face interaction and/or the Intervention #2 e-form now gives the receiving doctor more useful info? If so, what kinds of data are not being captured in the chart, and what can we do to highlight that information (change orientation in the system so it stands out, highlight/color code info/sticky flags/etc) so that it is easier to find? (the categories on the written hand-off seem like data that should be accessible in the chart to me, except perhaps for Contingencies and Tasks, not sure how those are being used)



Kelly Graham said...

To answer your question Jon- the data in the chart is usually a progress note, which has an expiration date. The progress note usually makes it into the chart by noon, and the primary teams are leaving for the night around 6PM- so this document is already 6+hours hold by the time the cross-covering physician needs the information.

In addition, the progress note really addresses the plan for the day on each active issue, but not always the "what to do if" (ie "contingency planning") types of information that physicians exchange about worst case scenarios (ie- if Ms. X 's heart failure worsens, use 80mg of IV furosemide, not 40- she only responds to really high doses). These are the types of scenarios that lead to adverse events (even if they are rare). The "tasks" section is the to-do's for the night, (ie- please check ins/outs at midnight, if not at goal of -1L, give another dose of 20mg furosemide). Patient care doesn't stop at night (especially with our ever-shortening lengths of stay, and this portion of the written sign-out outlines what needs to be done overnight to continue providing patient care.

Anonymous said...

Great project and kudos for tackling such a thorny problem. How did you standardize direct communication between the departing intern and arriving intern? Are interns now required to stay at work until night float arrives?

jonmcrawford said...

Thanks for the response Kelly, great idea to get this in a written form that can be accessed during the gap between the notes. That type of info seems vital and is obviously making a huge impact!
(Paul, Kelly needs a raise)

76 Degrees in San Diego said...

Obviously, having a concise, standardized clinical summary is very useful. Do you have to be logged into the EHR to see it, or, is it tranferrable to portable hand held devices? Many times in practice, calls come from hospital staff when you are not logged into the EHR...

Michael Kirsch, M.D. said...

Handoffs are vulnerable periods, but not just for housestaff. We practicing physicians 'hand off' regularly, often several times each week when we are on-call. In addition, we routinely take phone calls at night from our partners' patients whom we do not know. It's not always easy navigating through conversations of patients with all varieties of abdominal pain. Is it their chronic irritable bowel or could it be appendicitis? Do we want to send them all to the ER? The practice of medicine is not seamless.

e-Patient Dave said...


It sounds like for management of any active case, a vital design goal for a medical record system would be a "handoff view" that shows the "must not miss" info - whatever that is.

I say that with joyous naivete, so feel free, all, to correct that. I just know that modern databases offer views, which can typically be configured for specific use cases, and this would seem a biggie.

Michael Kirsch, M.D. said...

Yes, Dave, anytime there is a transfer or responsbility from one physician to another, there is potential for a misaventure. Amazingly, we physicians usually get it right, but these are vulnerable points in the chain of patient custody. A particularly vulnerable point is when patients are discharged from the hospital back to their primary care physicians, who may not have seen the patient during the hospitalization. Often, important clinical data is still pending at the time of discharge. The accepting primary care physician may not even be aware that this data exists.

Kelly Graham said...

To answer more of your questions:

-You do need to be logged into the EMR to see the sign-out view. It is easily accessed from home with a login and password.

-We standardized the hand-off by asking the primary teams to stay until nightfloat arrives (5:30PM), this had the most profound impact on the patient safety data and physician satisfaction.