CHICAGO — The largest-ever study of the influence of hospital-based physicians on outcomes and costs has failed to show significant benefits, David Meltzer, M.D., reported at the annual meeting of the Society of Hospital Medicine.
“There was a slight trend toward lowering hospital mortality. Otherwise, we found no difference in outcomes between hospitalists and nonhospitalists,” said Dr. Meltzer of the University of Chicago.
“I was somewhat surprised. We began the study expecting we'd see a larger difference,” he told this newspaper.
The study, which involved 31,013 admissions at six academic centers over a 2-year period, compared costs and outcomes of hospitalized general medical patients treated by hospitalists or by nonhospitalist physicians. The researchers used administrative data, patient surveys (inpatient interviews and a 1-month follow-up survey), a chart review looking at process of care variables, data from the National Death Index, and surveys of attending physicians, staff, and primary care physicians.
The investigators concluded that hospitalists did not affect the average length of stay, costs, or outcomes of care across all sites. “Length of stay and cost fell with increasing disease-specific experience, but hospitalist experience may have been offset by higher initial resource use,” Dr. Meltzer said during a plenary presentation of the study, which was also presented in a poster session.
Hospitalist care was associated with significant reductions in mean length of stay at two of the six sites, Dr. Meltzer said.
Earlier, single-center studies of the effects of hospitalists have produced some mixed results. A 2000 review led by Robert M. Wachter, M.D., at the University of California, San Francisco, concluded, “Empirical research supports the premise that hospitalists improve inpatient efficiency without harmful effects on quality or patient satisfaction” (JAMA 2002;287:487–94).
Dr. Meltzer's own earlier study of over 6,500 patients at the University of Chicago showed that “hospitalist care was associated with lower costs and short-term mortality in the second but not the first year of hospitalists' experience.”
During a later session at the SHM meeting, Dr. Wachter said that the latest study by Dr. Meltzer is not totally relevant to nonacademic hospitals. “It's a different kind of environment. The evidence for improvement resulting from the use of hospitalists remains robust with more than 20 published studies showing average cost and length-of-stay reductions of about 15%.”
Dr. Meltzer cited several caveats that may take some sting out of the findings. One study limitation is the “spillover effect,” he explained, which may help to raise the quality of the nonhospitalist comparison group and lead to underestimation of the value of hospitalists.
“Interns and residents work with hospitalists and learn new ways of doing things that may be more efficient and lead to better outcomes, and they remember [these new ways] at the end of the month and then go work with and teach other attendings. So we're used to thinking that teaching is from the attending to the resident to the intern, when in fact there's teaching within those levels and even up the levels,” Dr. Meltzer told this newspaper.
Another equalizer is “a sort of selective attrition effect where, because the hospitalists are taking up more ward months, the department or section can be more selective in whom they put on the wards, so you get only the best attendings on the wards and, not surprisingly, they do a little better than the group as a whole would have done if you had not been able to sort of weed out those who might not do such a good job,” he said.
Dr. Meltzer's third caveat is that, as earlier studies show, hospitalists improve over time. “I think our data are consistent with the hypothesis that hospitalists have real effects, but that those effects don't appear so immediately in the data that we see for all these reasons.”
Finally, Dr. Meltzer was impressed by the finding that the average hospitalist in the study cared for 134 patients, compared with a 46-patient case volume for the average nonhospitalist.
“What's even more striking,” he said, “is that when we go to disease-specific experience, the average hospitalist cared for two-and-a-half patients with that same diagnosis, and the average nonhospitalist cared for less than one (0.93). We found that every doubling of disease-specific experience decreases length of stay and cost by about 3%.”
The next step, Dr. Meltzer added, is for someone to conduct a similar study in community hospitals. And “further work is needed to assess physician factors, site factors, and spillover effects that could influence comparisons between hospitalists and nonhospitalists.”