When hospitalists stay longer, patients leave sooner


An extended stretch of days on service can be grueling for a hospitalist, but a new study has found that the effort may pay returns in reduced length of stay for patients.

The study, published in the July/August 2010 Journal of Hospital Medicine, analyzed the hospitalizations of more than 10,000 pneumonia and heart failure patients treated by IPC The Hospitalist Company practices. Each patient's fragmentation of care was calculated, based on the percentage of care that was handled by a hospitalist other than the hospitalist who provided the majority of care.

Photo courtesy of Kenneth R Epstein
Photo courtesy of Kenneth R. Epstein.

Researchers, led by Kenneth R. Epstein, FACP, now chief medical officer for Hospitalist Consultants, Inc. (HCI), found that as fragmentation increased, length of stay did, too. The association was significant, even when researchers controlled for a number of factors, including admission day of the week. Dr. Epstein recently spoke with ACP Hospitalist about the study's results, problems with handoffs, and pros and cons of block schedules.

Q: What motivated you to research fragmentation of care?

A: We thought it was a very important area that there was no hospitalist research on: What's the impact of different hospitalist group schedules? We always say it's better to try to schedule a block of days that is at least three to four days long, but some hospitalist groups have one or two days on and then switch back and forth. We were interested to see if there was an impact of the continuity of care on length of stay or any other quality parameters.

Q: How did the results compare to your expectations?

A: I thought we might find a little bit of a relationship. We found a very dramatic relationship—for every 10% increase in fragmentation, the length of stay went up significantly. So when fragmentation went up by 30% or 40%, then length-of-stay increased by more than a full day.

We did not find a significant effect of day of the week of admissions on length of stay. We thought maybe fragmentation was a proxy for people who come in on weekends. But it was the fragmentation that caused length of stay, not the day of the week of admission. That was an interesting finding.

Q: Could that indicate that the causation goes the other way—that the greater fragmentation of care on weekends causes longer lengths of stay for patients admitted then?

A: Yes, I think weekends prolong length of stay for two reasons. If a physician comes in for the weekend and covers, there is the risk that they will not discharge patients as much. They may see their role as keeping them clinically stable over the weekend and may be less motivated to do discharges if they're not a regular part of the group. Secondly…it's adding another doctor to the mix.

Q: What do you think causes the overall association between fragmentation and length of stay?

A: There's always a dropoff in information, despite efforts to improve handoffs. If I have a plan and I know I want to do A, B and C prior to discharge, and another physician comes along tomorrow, they may lose a half a day or a day while that physician comes up with their own plan.

It has to do with when the physician feels comfortable sending a patient home. If I've been seeing a patient for three or four days, I know what's going on and…when certain clinical problems are getting better, I feel comfortable sending the patient home. Whereas, every time there is a transfer of care, a new physician may come on and need to assess the situation and decide whether the patient is ready to go. [The incoming physician] may review everything today and say, ‘It looks pretty good, I'll try to get the patient home tomorrow,’ rather than getting them home today. It doesn't happen in 100% of cases—otherwise we would have found an even higher percentage—but I think it happens often enough that it created those results.

Q: What are the best potential solutions to this problem?

A: One is to try to minimize the fragmentation by having more block schedules, having physicians come in for at least three- to four- to five-day blocks. Seven on/seven off is probably a good system. When people come in for a day or two, it definitely increases fragmentation.

The second thing is improving handoffs in order to increase the quality of information. The third thing, which is related to the second, is to really work out a system with your hospitalist group so that you all feel comfortable discharging patients that the other physicians took care of. The group members need to determine ‘What information do I need to feel comfortable discharging the patients?’ If I'm coming off service and I know someone's going home tomorrow, I try to prepare everything ahead of time. I may then write an order stating that ‘May discharge tomorrow if clinically stable after seen by Dr. So-and-so.’ It's clear to the other physicians that the person's doing well, and I've done a lot of the prep work.

Q: What obstacles cause some hospitalist programs not to use block schedules?

A: I think it's just a lack of awareness of the advantages. Hospitalist groups that don't have block scheduling base the schedule more on physician requests: I'd like to work just Monday and Tuesday and be off Wednesday. In terms of lifestyle, it's easier to work a couple days, take a day off, rather than committing to five or seven days of working. [It's] putting physician desires above quality of care. [Also] if a group is short-staffed and is filling in various spots, they are going to have more discontinuity.

Q: Have the results of this study affected the way you approach scheduling?

A: I wear a couple hats. I'm chief medical officer for HCI. In terms of putting together hospitalist programs for HCI, I'm very aware of this data and we really push block scheduling as a way to promote quality of care. The other hat I wear is to still do hospitalist management consulting. Certainly part of my recommendations, when I go into a program, is to help them create more continuity of care in their hospitalist scheduling.

Q: Most prior research in this area has shown that hospitalists reduce length of stay. Does this study point out a limitation of the hospitalist model?

A: No, I think hospitalists do reduce length of stay. There's just as much fragmentation in nonhospitalist systems, with group members from internal medicine or family practice groups covering the hospital. I think it's just a way of improving it even further. Hospitalist programs that have improved handoffs as well as longer blocks of working are going to have an easier time reducing length of stay than those that don't do those two things.