I began thinking about – and yes, advocating for – the concept of hospitalists in the mid-1990s, when I became convinced that having separate inpatient and outpatient physicians would improve the quality, safety, and efficiency of healthcare. A study in today’s Annals of Internal Medicine reports that, while hospitalists did cut hospital lengths of stay and costs (consistent with the prior literature), these savings were eaten up by increased 30-day post-discharge costs, driven largely by higher rates of readmission.
While the study has its limitations, overall I though it was well done, credible, and a bit disheartening. But my core belief about the value of hospitalists endures. If hospitalists are squeezing balloons when it comes to the cost of care, I think we need bigger balloons, not fewer hospitalists.
Under Medicare’s Diagnosis-Related Groups (DRG) reimbursement system, under which hospitals receive a fixed hospital payment for a given diagnosis, hospitals are highly motivated to shorten lengths of stay and decrease inpatient costs. Increasingly, they are also under pressure (via public reporting, accreditation standards, and some early payment changes) to improve quality, safety, the patient’s experience, and (in teaching hospitals), the training of housestaff and students. In all of these areas, hospitalists have made powerful contributions, which is why hospitals have been so enthusiastic about welcoming them and willing to support them politically and economically.
Moreover, hospitalists have stepped up to fill a variety of other needs that go well beyond value improvement: caring for patients who can no longer be handled by residents because of duty-hour limits, managing patients who lack primary care physicians, co-managing surgical and other non-medical patients, and leading quality, safety, and IT initiatives. In addition, after the early pushback (which I have the scars to remind me of), many primary care doctors have embraced the hospitalist idea and are unlikely to return to the chaotic, underpaid, and overstressed world in which they were responsible for their patients across the inpatient-outpatient divide.
Today’s Annals study involved nearly 60,000 Medicare fee-for-service patients who were admitted to 454 U.S. hospitals between 2001 and 2006. It found that while hospitalist care was associated with significantly shorter hospital lengths of stay and modestly lower hospital charges, the savings were consumed by higher 30-day post-discharge costs. Compared with patients cared for in the hospital by their own primary physicians, those cared for by hospitalists were less likely to see their PCPs in follow-up (but more likely to see other physicians), more likely to be discharged to a nursing home, and more likely to return to the ED and be readmitted over the next month. The authors, Yong-Fang Kuo and James Goodwin of the University of Texas’s Galveston branch, performed a variety of sophisticated statistical analyses that adjusted for all the relevant variables and tested whether small changes in assumptions would influence the results. They found that the increased readmission rate was not a very robust finding (it was no longer statistically higher under certain new assumptions), but the rest of their results didn’t waver.
Like the Annals editorialists, Lena Chen and Sanjay Saint of the University of Michigan, I find myself unable to dispute the main findings. The sample size is huge, the definitions and assumptions are reasonable, and the analysis is strong. The limitations are there and acknowledged: the study covered a period 5-10 years ago, the patients are limited to those in the Medicare fee-for-service system (it would be great to see a similar study coming out of a large integrated and capitated system like Kaiser Permanente, which would take it on the chin if all its hospitalists were doing was cost-shifting but hasn’t wavered from its wholesale embrace of the hospitalist model), the definition of hospitalists was based on inpatient volume alone, and there is no information about how programs were organized or about subgroups that might have had better (or worse) results. But these are mostly quibbles – I am willing to take the results at face value.
Explaining them is harder. Were patients simply discharged “sicker and quicker,” and thus more likely to return to the hospital? I doubt it. As Chen and Saint point out, despite an intuitive link between short lengths of stay and higher readmission rates, several studies that have looked for such a connection have come up empty.
More likely, the findings represent the cumulative effects of influences on all the players. Hospitalists – highly motivated to cut hospital days – were more likely to send patients to skilled nursing facilities when they were ready to leave and less able to hook the patients back up with their primary care doctors at the time of discharge. Primary care docs who were uninvolved in the hospitalization may have been less comfortable that they understood the ins-and-outs of the hospital stay and more likely to favor readmission for the post-discharge patient who wasn’t doing well. Patients may have believed that, since their PCP didn’t see them in the hospital, the best thing for them to do if they were wobbly was to return to the ED or the hospital.
Each of these explanations shares a common thread: while the hospitalists did what they were supposed to do in the hospital, and PCPs probably did the same once the patients resurfaced in their office, nobody – neither of these physicians nor the systems of care in which they practiced – stepped forward to fill the black hole after hospital discharge. This should not be surprising, since during the study years there was precisely no incentive for anyone to do so.
The last patient in the study was discharged in 2006. The first serious mention of bundling (a fixed payment for an episode of illnesses, including hospital and post-discharge care) in policy circles came in 2008, a year before the influential NEJM study by Jencks, Williams and Coleman reported that one in five Medicare patients was readmitted in a month, thus catalyzing efforts to penalize hospitals for excess readmissions. For a Medicare fee-for-service patient hospitalized in 2001-06, the “care continuum” was a nice phrase, and just that.
Today’s study tells us that hospitalists have done their jobs well, but the job has been defined too narrowly. As hospitals’ lenses widen to include a larger slice of the care continuum (probably not full capitation but at least a substantial period of time after discharge, coupled with readmission penalties and other maneuvers to put skin in the post-discharge game), their willingness to help support their hospitalist programs will be predicated on the latter’s prove ability to improve quality, safety, patient experience, and efficiency over that entire period, not just the hospital stay. Hospitalists have already begun to dive into this work, leading efforts to create electronic discharge summaries, post-discharge clinics, protocols to follow up on test results, and a variety of other strategies to prevent rehospitalizations and to improve handoffs.
But ensuring safe and efficient post-hospital care is not one-handed clapping. Outpatient systems need to change so that primary care physicians can make themselves available for rapid post-discharge appointments, create after-hours capacity to see patients during the dicey post-discharge period, and share in hospitals’ – and hospitalists’ – accountability for excess readmissions.
The hospitalist field is here to stay, and today’s study doesn’t change that fact. I remain confident that good hospitalist programs, working with their partners in the outpatient setting, can successfully reduce overall costs of care, while improving quality and safety. I’m betting that a repeat of today’s study performed in few years will demonstrate that hospitalists add value across the continuum, not just within the walls of the hospital. These new findings increase the pressure on the hospitalist field to make it demonstrably so, which is as it should be.