Tuesday, August 31, 2010

Does preoperative medical consultation improve outcomes? Might it even cause harm?

A new cohort study from a large administrative database published in Archives of Internal Medicine found:

Results Of 269 866 patients in the cohort, 38.8% (n = 104 695) underwent consultation. Within the matched cohort (n = 191 852), consultationwas associated with increased 30-day mortality (relative risk [RR], 1.16; 95% confidence interval [CI], 1.07-1.25; number needed to harm, 516), 1-year mortality (1.08; 1.04-1.12; number needed to harm, 227), mean hospital stay (difference, 0.67 days; 0.59-0.76), preoperative testing, and preoperative pharmacologic interventions...

Conclusions Medical consultation before major elective noncardiac surgery is associated with increased mortality and hospital stay, as well as increases in preoperative pharmacologic interventions and testing. These findings highlight the need to better understand mechanisms by which consultation influences outcomes and to identify efficacious interventions to decrease perioperative risk.

So---the differences, while statistically significant, were small but everything---everything related to outcomes and utilization---was in the wrong direction.

There are many caveats. First, this was a Canadian study. Applicability to US systems of care cannot be taken for granted. Second, applicability to hospitalists is limited. Although hospitalists do a lot of comanagement (and it looks as though a lot of comanagement was done in this study, though impossible to say how much) they do hardly any preoperative evaluations, because the vast majority of patients are not admitted to the hospital in advance of their surgery. Third, since the study period, 1994-2004, we have worked out a lot of the nuances of perioperative medicine, particularly in areas relating to beta blockers, evaluation, management of patients with coronary stents and perioperative anticoagulation management. Finally, the study doesn't adequately address the value of postoperative consultation, the more likely scenario today.

Although we now have research quality data (and while it needs to be reproduced it's probably the highest quality we're likely to get) saying consultation may cause harm it's easy to imagine individual circumstances in which it's likely to be helpful. Who's going to make sure the right medications are held and the right ones are continued as seamlessly as possible (e.g. statins and beta blockers)? Will somebody pay attention to the patient's diabetes and keep their chemistries from getting messed up? Will the surgeons do these things? Some will and some may not (which is why I say individual circumstances).

What might have led to the harm? The consultation group received more beta blockers, which, in the manner they are likely to have been used from 1994 to 2004, may have been harmful. Increased testing in the consult patients may have delayed some surgeries, which could have led to harm.

As the authors pointed out, these results do not argue decisively against the use of consultation. They do suggest that we shouldn't do consultation and comanagement just because we can and that more research is needed do define which processes involved in consultation and comanagement are truly useful and in what circumstances.


#1 Dinosaur said...

Two words: Puh. Leeze!

Ya think the surgeons were just requesting consultations on patients more likely to be at risk for poor outcomes? That would define the appropriate usage of consultation. The fact that there was no difference in post-operative wound infections (as expected) supports this possible interpretation. Talk about screamingly obvious.

Anonymous said...

I have a slightly different take on this paper http://tiny.cc/7bo0a.

R. W. Donnell said...

If I understand your comment correctly, surgeons selectively requesting consults on higher risk patients does not easily explain the study findings. The authors used several methods to adjust for severity. The finding of no difference in infections, in the authors' minds, supported their interpretation that the difference in outcomes was some way attributable to consultation rather than differences in baseline risk. Granted, severity adjustment is difficult in this type of design.

#1 Dinosaur said...

So how would you do a legitimate study? Take all patients for whom surgeons did NOT feel the need to have a medical consult and randomize half of them to get one anyway. Seems to me that would address the question most directly (and would be far more ethical than potentially withholding medical consults from patients whose surgeons thought they needed them).