Last week, another study was published (this one in the Archives of Internal Medicine ) documenting a hospitalist efficiency advantage. Coming on the heels of more than 20 studies with similar results (see, for example, this and this ), one might ask how much this study adds to our understanding of hospitalist care.
The answer: more than you might think.
The study was another quasi-randomized hospitalist vs. nonhospitalist experiment at an academic medical center (Montefiore Medical Center in the Bronx). The overall findings are consistent with many others: mean length of stay (LOS) was about 15% lower for hospitalists than nonhospitalists (5.01 vs. 5.87 days, p<0.2), with no difference in readmission or in-hospital or 30-day mortality rates.
The study adds to our understanding in three ways.
First, it found a statistically significant negative association between length of stay in days and the number of ward months per year, with each extra month associated with a 0.19 day reduction in LOS. Although prior studies ( this and this ) have shown an experience curve for hospitalists graduating from the first to the second year of practice (improved efficiency and maybe even improved mortality rates), this is the first to show a practice-makes-perfect curve in terms of hospitalist months per year. In my own program, many hospitalists now only do 2-3 months per year on the wards, since they are busy doing other clinical (e.g., palliative care, co-management) and nonclinical (e.g., research, program leadership) activities. When we last looked a couple of years ago, we found that our hospitalists were far more efficient than our nonhospitalists, but that the 2-3 month-per-year hospitalist was no less efficient than the 6-8 monther. This study motivates me to look at this issue again. Though I’m not sure I’d change things – since there are so many advantages of people having strong “diastoles” in terms of overall programmatic contribution, finances, and prevention of burnout – it’ll be good to have the information to make rational decisions.
The second interesting finding was that the LOS advantage for hospitalists was not uniform across diseases. Rather, it was exaggerated in patients with very high acuity. For example, the difference in the mean LOS for stroke and sepsis was nearly four days (for both diagnoses, approximately 8 days for hospitalists vs. 12 days for nonhospitalists). On the other hand, hospitalists didn’t do much for less acute, more straightforward admissions, particularly those for cardiac issues like chest pain (0.09 less) and arrhythmia (0.24 days more).
The authors speculate that “these findings may reflect hospitalists’ ability to closely monitor patients with dynamic conditions… the close monitoring and continuous presence offered by hospitalists may allow for earlier discharge because hospitalists are more likely to detect clinical improvement in real time and to make appropriate adjustments in treatment regimens.” I agree.
The third interesting finding was that the hospitalist effect was much more pronounced in patients in whom complex discharge planning was required. The difference in average LOS was only 0.4 days when patients went home (3.40 vs. 3.80 days) but ballooned to a full two days when patients were discharged to a skilled nursing facility (8.21 vs. 10.22 days). Here, the authors posit that this difference “reflects hospitalists’ skills in working with ancillary staff, such as social workers or discharge planners.” Again, seems right to me.
As always, one can fret that shorter LOSs reflect premature discharges, but the findings of comparable 30-day readmission and mortality rates are reassuring.
This study provides yet more evidence of impressive improvements in efficiency, and adds to our understanding of the pathophysiology of the hospitalist advantage. Although – given the national shortage of hospitalists (don’t believe me?: just check the “help wanted” ads in The Hospitalist ) – it would be tempting to use these results to focus hospitalist care on only certain types of patients (those with acute and dynamic conditions, rather than the less ill patient likely to return home), in most cases there won’t be a sensible way to enact such a strategy. But, in academic hospitals with separate cardiology services, this study might argue that placing a hospitalist in this mix on these services (unless needed for dealing with housestaff duty hours issues) might be a low yield intervention.