Tuesday, April 20, 2010

Lean update -- Readmissions

Several of you have asked for updates on our Lean process improvement efforts. There is lots of stuff going on at the hospital in this regard, and it is not possible to put it all in this blog. But here is a summary of the work on an important set of clinical projects related to transitions of care. It is in the form of an email from Doctor Julius Yang, who is coordinating the effort.

I offer these as works in progress, with a full admission of current flaws in our processes. I know folks in other hospitals will find the subject matter familiar, as the issue of readmissions is one we all face. Perhaps our approach will provide you with useful suggestions. We also welcome yours.

Notice below the involvement of residents. This is very important in an academic setting, as they are key participants in the delivery of care. It also provides these young doctors with a chance to learn process improvement methods, something that is usually not taught in medical schools.

Notice, too, the beginning of more outreach to patients after they leave the tertiary setting.
This is likely to be a growing trend nationwide. We are certainly not the first place to do this, but it remains relatively unusual for hospitals, which have tended to focus only on the acute care setting.

Hello all,

I wanted to take this opportunity, now five weeks after the close of our Multidisciplinary Quality Improvement Retreat, to once again thank all of you for your participation and important contributions towards improving the transition from hospital to home, and then to clinic, for our patients. Our rate of readmissions to the hospital within 30 days of discharge from the Medicine service remains high, hovering at about the 20% range. We've run more detailed analyses of these cases, and continue to find a wide variety of "root causes" for these readmissions, including medical decision-making at discharge, poor coordination of care plan with extended care facilities, lack of outpatient visit prior to readmission, ambiguous contingency planning when symptoms recur, among others. Based on this analysis, we continue to pursue multiple avenues for improvement simultaneously.

In terms of the two specific project streams generated from the QI Retreat (telephone follow-up with patients within 24hrs of discharge to home, and an electronic discharge checklist), over the past month a number of workgroups have made progress towards implementation of these initiatives. A brief update:

1. Post-discharge Telephone Follow-up
With strong support from our Nursing Directors in medicine, cardiology, and our Director of Patient Safety, we have received approval to conduct a pilot intervention for a "Nurse Discharge Specialist" whose role will include post-discharge telephone follow-up with patients at high risk of readmission. The target population for the pilot will be those patients who have been admitted to the hospital with heart failure, and especially those who have had prior readmissions. The intervention will consist of a dedicated nursing role to visit with these patients prior to discharge for both education and assessment of factors that may lead to readmission. Once discharged, this nurse will then follow-up with the patient by telephone the following day, to reinforce education and to ensure continuity in the discharge plan. The telephone "scripts" that were developed in the retreat will help to inform how this encounter is designed. During the pilot, there will be one nurse each weekday to serve in this role, to be shared across Farr 3, CC7, and Farr 2, with the intention of capturing 6-8 patients per day. The goal of the pilot will be to refine the logistics of this position, assess feasibility and generalizability, and assess impact on readmission rate for those patients seen by the Nurse Discharge Specialist. We anticipate the pilot will begin in May.

2. Electronic Discharge Checklist
Although we do not yet have a "live" electronic checklist, there has been some important progress over the last month. First of all, our Case Management leadership have embraced the concept of a "discharge checklist" to ensure that all requisite factors have been addressed prior to discharge of our patients at high risk of readmission (for now, we are defining "high risk" to include patients with heart failure, or patients who have been readmitted previously). We are therefore planning a pilot intervention in which our case managers will be in charge of a "checklist" (resembling the concept explored in our retreat) that will be used to guide care planning at daily case management rounds. The goal of the pilot will be to refine the checklist tool and its use, assess feasibility and generalizability, and assess impact on readmission rates. We anticipate this pilot to begin in the next two weeks.

Secondly, two of our QI retreat alumni, Elena and Andrea, have worked with another one of our residents, John Greenland, to develop a highly-detailed mock-up of an electronic discharge checklist, viewed through the POE alternate dashboard. It looks fantastic, and appears quite feasible to execute from existing data resources. We are now planning to present this model to I.S. for their review, with the goal of collecting their assessment of feasibility and timing for further development and implementation of such a tool.

As we continue to further develop these two project streams, I see that a major challenge still lies in establishing "stability" for change and continuous improvement at the front-line, "unit staff" level. Because of competing schedule demands, it has been difficult to coordinate routine meetings for front-line staff on each unit, in order to help teach and develop "kaizen mindset". I would like to try to gather us, as QI retreat alumni, perhaps in the next two weeks to further explore the best model by which to pursue this goal, and welcome any input or thoughts you may have in the interim.

Again, thanks to all for your work last month - the improvement continues.


Jane Sherwin said...

I've long thought of "Lean" as a source of hope in health care reform, ever since writing about its use in an orthopedic clinic. It's a wonderful, if complex, linking up of two such very different fields. Thank you Paul for sharing Julius's e-mail.

Anonymous said...

Not to rain on the parade, but why not step back and ask why we (NEJM, CMS, etc) are suddenly fixated on readmissions? It's not obvious that this is a particularly important measure of performance.

Some guiding questions:

What's the alternative to readmission? [death at home? That would certainly lower the readmission rate, as would keeping patients in the hospital longer than necessary or transferring them to rehab when rehab is overkill]

What's the optimal rate of readmission? [probably not zero]

If readmission isn't the most important thing to capture our attention, what is? [think about overall well-being of the patient and efficiency of the system]

If you're going to stay on readmissions, where are the primary loci of control, and where should they be? [hint: not everything is about the hospital. As a PCP, the idea of having a hospital-based nurse call a discharged patient hurts my head. Why not get in touch with me and have my staff handle this? A Kaiser leader once told me that discharge truly occurs when the patient gets home; not before]

Just some thoughts...

Engineer on Medicare said...

I suggest that a more appropriate number than readmissions might be the sum of deaths + readmissions.

Anonymous said...

Excellent, excellent, excellent. In my mind the primary factor in process improvement is tenacity, which is being well demonstrated here. The "competing schedule demands" euphemism also demonstrates the difficulty of instilling the idea that this work is priority work.

I am guessing, without any special knowledge at all, that your primary sources of readmission are heart failure and readmission from nursing homes. Just to personalize the difficulties - my uncle was a chronic heart failure patient who, despite his and his hospital's best efforts, suffered multiple and frequent readmissions for his condition.
(I was not close enough to know whether he was a candidate for an assist device or other more definitive therapy.) One day he just decided that he was not going to the hospital any more, and the next time he had a crisis, scared but determined, he just said, "this is it", refused medical treatment, and died.


Anonymous said...

Anon 4:06:

I think one can always question the validity of any quality indicator. For instance, is the indicator of CABG operative mortality within 30 days a valid one by which to publish a public report card on cardiac surgeons? I'm sure many cardiac surgeons would argue no.

The problem is, one could spend all one’s time arguing which indicators to monitor of the million available. Is this about arguing, or action?

“What’s the alternative to readmission?” Your suggested alternatives indicate what many hospitals could do to “pass” this measurement while not really addressing the underlying patient care problem. A pretty cynical response, eh? BIDMC apparently intends to address the real problem.

“If readmission isn't the most important thing to capture our attention, what is? [think about overall well-being of the patient and efficiency of the system]”. Well, let’s see – is not the patient’s well being improved if his heart failure is treated well enough he doesn’t have to come back in within 30 days because an “efficient” hospital system treated him properly and quickly? What other specific indicators would you suggest to better illustrate your desired metrics?

As to why not let your staff handle it, you have, perhaps inadvertantly, hit the nail on the head. If that were being done consistently, one would expect readmission rates to be lower - no? Or one could ask, why doesn't your office already know your patient has been discharged and take proactive action, instead of waiting for the hospital to call you? I am sure they would love it if you would take the responsibility - cheaper for them and better continuity of care for the patient. It must not be happening right now.

As you see, this becomes an endless and ultimately unproductive debate. We no longer have time for that.

nonlocal MD

Anonymous said...


I agree that no measure is perfect, but what is the sense in rushing to improve a fundamentally construct that is devoid of inherent meaning? Getting the right measures is a worthwhile undertaking. Action without thought is dangerous.

But I do love your casual assumptions about how my office runs and what I do and don't know about my patients. So nice to see projection is alive and well!

My main point was that from the patient's perspective, discharge actually occurs at home, and therefore it makes the most sense for the ambulatory/continuity providers to make sure this goes well. The fact that transitions often don't go well is the result of systems properties that frustrate good, reliable communication across care settings. The hospital is just one part of the overall system, and to be honest, you can't blame the hospital for trying to solve the issue on its own. After all, hospitals have been deemed the units of accountability. However, the attribution of system properties (e.g., readmission rates) to a single actor within the system may inadvertently encourage a suboptimal approach that ignores other actors.

Speaking of inadvertent, you've actually made a meaningful point here:

"Well, let’s see – is not the patient’s well being improved if his heart failure is treated well enough he doesn’t have to come back in within 30 days because an “efficient” hospital system treated him properly and quickly? What other specific indicators would you suggest to better illustrate your desired metrics?"

To your first question: Ummm...yes? But that's exactly the problem with just looking at readmissions, right? You can't tell why readmission didn't occur. Death? Fear of another bad hospital experience? Elder abuse? Excellent outpatient care facilitated by good communication? Who knows?

You also can't tell whether a given readmission is undesirable. Was an adverse outcome nipped in the bud? Was outpatient management inadequate? A responsible manager will _always_ need to answer these questions in order to determine the appropriate system design changes. There may be a case for using readmissions to trigger further case investigation, but the aggregated readmission rate is just a lousy performance measure.

One other serious conceptual problem with readmission rates: the denominator is completely endogenous. Think about it. It's not like survival to discharge (and therefore the opportunity to be readmitted) is independent of the quality of inpatient care. I'll let you figure out the paradoxical results that can occur when you have denominator endogeneity.

There are ways to measure fundamentally good care. Though it's more expensive than looking at administrative data, I'm a fan of actually getting information directly from a sample of patients (or their caregivers) about their transitions. Then couple these data to fundamental health outcomes: death, disability, quality of life, etc. If it's costs we're after then, dare I say it, we should actually measure costs.

When we settle for bad measures because they're cheap, we shortchange our patients. But don't let me stop you! It's good to be enthusiastic, and despite a pretty wacky take on things, your heart seems to be in the right place.

Anonymous said...

Anon 4:06, 4:36;

I cannot speak for BIDMC, but yeah, gee I bet they picked readmissions as ONE quality indicator (which Paul chose merely as an example) because this is an indicator for CMS. Not too difficult to understand. So you can criticize its validity all you want, but it’s still going to be there.

The real question is, are they going to use it to improve patient care, or just to report a number that’s as low as possible to look good? Read the post and you have your answer. They are indeed “using readmissions to trigger further case investigations” – and along the way, they can establish that endogenous denominator you talk about, while improving care for the numerator, hmmm?

And sure, it would be desirable to evaluate and follow all discharged patients to see how they do. Are there staff and $$ resources to do so? Ask Paul; I don’t know.( Some of this may be captured in the now ubiquitous patient satisfaction surveys.) If not, how would you select a patient population to survey/follow and how long would you follow them for your metrics of morbidity/mortality/quality of life?

As for “settling for bad measures because they’re cheap” - and my unfettered but ignorant enthusiasm, you obviously have come to this blog after my 2007-era disagreements with Paul’s posts calling out physicians for nitpicking each and every quality indicator or metric. I used some of the same arguments you have. Over time, I have become a convert. Far from “rushing”, people have actually been talking about these things for a very long time. It is now time to take action.
And ps, I recommend Dr. Yang’s video to you as well as Paul’s speech to the SHM, both in adjacent posts.

nonlocal MD

76 Degrees in San Diego said...

Looking at readmissions from an outpatient perspective is really a fruitful exercise. When we had our FM residency, we faculty would engage our residents in "what could have been done to prevent this" readmission. It is analogous to "Wait, Wait, Don't Tell Me"(you start with the answer, and you try to reconstruct the question). It created a curiosity amongst the residents in thinking "how can we manage our patients better". You need to have an EMR that multiple participants can look at simultaneously, particularly for the outpatient care. It was great "exercise" for us all.

Anonymous said...

Here's an interesting recommendation from Robert Wachter to the Joint Commission when they solicited feedback after an inspection of his hospital (which is astounding in itself):

"First, they should require that hospitals deliver more than 90 percent of their discharge summaries to the follow-up provider within 24 hours of hospital discharge. We know that rates of readmissions are stunningly high; one reason is that more than two-thirds of follow-up providers lack a discharge summary when they see patients for first time after discharge. That’s crazy, and it’s unacceptable. TJC could fix it tomorrow." (Links to references embedded in this paragraph at his blog, Wachter's World).

This may be less applicable to academic centers and their clinics, but would be very important to PCP's such as anon 4:06.

nonlocal MD

Mike Caron said...

There are some things that fall naturally to lean processes. A case in point, my daughter needs to receive infusions on a monthly basis as an outpatient. She went to one setting, and in 30 minutes registered, had her infusion, and then left. She went to another hospital and in 2.5 hours performed the task. The difference in time also corresponded to a difference in price by a factor of 10. We tried calling ahead to save the 30+ minutes involved with transporting the medicine from the pharmacy but that could not happen until she was registered on site. Neither institution was your place.

Anonymous said...

Interesting Mike; I had a similar experience with a friend at none other than Johns Hopkins hospital. She has ovarian cancer and received weekly infusions. By the time we had gotten through registration, waiting for and talking to the coordinating nurse, waiting again in a tiny crowded clinic room for her port access (where the desk clerk was eating at her desk), then finally arriving at the infusion room only to be told the medicine couldn't be sent up till her WBC count arrived at the pharmacy (which should have happened long ago by then) - I was completely beside myself, as a Hopkins medical alumnus. The sad thing is, things were the same 30 years ago when I trained there; nothing changed.....
when are hospitals going to see the light?

nonlocal MD