Today’s New England Journal Hospitalist Study

Today my pals Peter Lindenauer and Andy Auerbach (and colleagues) published the largest hospitalist outcomes study to date, in the New England Journal of Medicine. It is a rigorous, important piece of work. Let me try to add a bit of context.

First, the What’s What. Using the massive database of the Premier system (which Peter has mined to tremendous advantage, such as in this study and this one), they compared the hospitalizations of nearly 80,000 adult inpatients with 7 disorders at 45 hospitals. They chose these disorders (things like pneumonia, CHF, and COPD) because they are common and are cared for by hospitalists, general internists, and family physicians.

They found that hospitalists had a length of stay 0.4 days below that of non-hospitalists, a 12 percent reduction that was highly significant. Patients cared for by hospitalists also had lower hospital costs ($268 lower than internists, $125 lower than FPs); this difference was significant for the internist comparison but not for the FP comparison. These efficiency advantages did not come at the expense of measurable quality problems – mortality and readmission rates were similar in all the groups. Nor was there evidence that hospitalists were more likely to discharge patients to SNFs (as opposed to home) than the non-hospitalists.

The paper does a nice job discussing the findings and speculating on why the differences were less than those seen in prior studies. (Of note, the paper refers to the LOS reductions as “modest” and “small” in several places. I take issue with that [and I’m guessing the authors do too, but if NEJM reviewers tell you to call something modest, you do just that]: a 12% reduction in LOS bespeaks a relatively powerful intervention, particularly when multiplied across 6,000 U.S. hospitals, many with habitually packed EDs. Even a $200 cost reduction, while well below that seen in prior studies, adds up when multiplied by a few million hospitalizations a year.) Nevertheless, the differences are smaller than those seen previously. My own thoughts on why:

  • The early hospitalist studies (which I summarized in this JAMA paper), compared just-starting hospitalists against all primary care physicians. The Lindenauer study looks at a very different world: most of the primary care docs who didn’t like or weren’t particularly good at hospital care have long since turned in their hospital parking passes, leaving a cherry-picked group of PCPs continuing to come to the hospital. This group is likely to be far more efficient than the group the hospitalists originally displaced.
  • One of the premises of the hospitalist model was that hospitalists would improve systems of care – developing checklists, protocols, IT systems, etc. Such systems, if effective, change the environment of care for everybody, not just the hospitalists. Moreover, they condition the nurses, case managers and others to expect a more efficient milieu. This new, snappier culture could improve everyone’s efficiency, and thus narrow the differences between groups.
  • The authors’ choice of seven common, relatively straightforward diseases (with ALOSs of about 3-4 days) may have biased the study toward lesser differences. Most of the earlier studies examined all-comers. It would make sense that the greatest hospitalist advantage would be in the care of the sicker patients with less common DRGs; entities that the PCPs would have minimal experience in managing and whose care would be less likely to be protocolized.

Caveats aside, the findings are there for all to see, and the socio-politico-economic fallout will be fascinating. Since most hospitalist groups get (and require) hospital support, and much of that support has been predicated on a Return-on-Investment drawn from earlier findings of 15 percent LOS and cost reductions, expect some skirmishes at budget time, with hospitals trying to tighten the screws on their hospitalist groups (“why should we raise your support – you only save us $200 per patient!”). The hospitalist group that has not convinced its CFO that the true ROI doesn’t hinge on pure cost reduction – but rather on systems improvement, QI, patient safety, and more – may be in for a bumpy ride.

In fact, as Larry McMahon’s thoughtful NEJM editorial articulates, for all its admirable exposition and scientific rigor, the new study probably doesn’t matter very much. The hospitalist genie is out of the bottle – in community hospitals, most of the PCPs have retreated to the office and aren’t returning unless somebody screws up the decimal place in next year’s CMS inpatient E&M reimbursement manual. In academia, all the researchers are safely ensconced in their labs, while the housestaff duty-hour reductions ensure that the need for hospitalists will continue to be insatiable. In other words, the demand for hospitalists is now relatively de-linked from the field’s original premise – efficiency advantages – and is now both more diversified and more robust. Larry argues, correctly I believe, that we don’t need many more hospitalists vs. PCP studies. Rather, we need better information about how to structure hospitalist programs to maximize efficiency, quality, safety, access, communication, education, and the satisfaction of all the stakeholders (patients, nurses, PCPs, and hospitalists). The anti-hospitalist traditionalists can continue to await the Next Coming, but here’s a hint: it ain’t coming.

Given the exuberant national market for hospitalists and the fact that a hospitalist (or a whole group of them) can pack up on Friday and be working across town Monday, I’m guessing that a few hospitals will negotiate their way into a highly dissatisfied group of hospitalists this year, an imploded program next year, and a very expensive hospitalist program mulligan a year later. Time will tell.


  1. Saul S. on January 15, 2008 at 8:38 pm

    Your blog is greatly appreciated!

    It is difficult to convince the administrators that hospitalists can bring tremendous value to the health care system. The current compensation structure is mainly focused on RVUs with a small subsidy . Unfortunately, this does not account for the unrealized value of hospitalists. As the above study points out, a 12% reduction in length of stay (LOS) is an accomplishment that speaks for itself. This can translate into savings which can leads to a hefty return on investment.

    I agree that there will be a bumpy ride ahead. Trying to prove our unrealized value to bean-counters can be quite challenging. A possible solution may be some form of collective bargaining agreement on a local or national level, to negotiate on the behalf of the profession.

    Have you heard of hospitalist groups forming a union? Being that most groups are now hospital owned, anti-union laws should not apply.

  2. Jackson B on February 2, 2008 at 5:51 am

    You are spot on with assessment that cost per case and los are no longer relevant in justifying a hospitalist program. Not surprised at slim cost and los benefit. A thorough severity of illness and casemix adjusted analysis in my own organization shows hospitalists cut los by 2 hrs but cost $125 more per case compared to community internists. Hospitalists aren’t the only ones that can follow a protocol. Meanwhile hospitalists lag in patient satisfaction scores. Might be the shift mentality. The genie may be out of the bottle but that doesn’t mean hospitalists should cook the goose that laid the golden egg.

Leave a Comment