Friday, April 29, 2011

DNR Patients and Outcomes of Surgery: Two Papers, Same Data, Different Results

Just the other day, a paper from Archives of Surgery reported that patients who have existing DNR orders have a higher postoperative mortality rate than patients who do not. This, of course was not surprising since people with DNR status are usually old and sick. The paper did receive some media attention with routine “churnalistic”* coverage. [HealthDay News, Medscape, MSN Health]

Having read the paper with mild interest, I didn’t think too much about it until yesterday when the May issue of the journal Critical Care Medicine [CCM] arrived. In it was a paper which is very similar to the Archives article. Let’s compare the two papers. The Archives paper was entitled “High mortality in surgical patients with do-not-resuscitate orders: analysis of 8256 patients,” and the CCM paper “Pre-existing do-not-resuscitate orders are not associated with increased postoperative morbidity at 30 days in surgical patients.”

Both papers used data from the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) Database for the years 200-2008. The Archives paper found some 4128 DNR patients while the CCM paper found only 2199. Both studies matched the DNR patients with non-DNR patients who had similar operations and demographics.

Both papers found that DNR patients were a little more than twice as likely to die within 30 days than non-DNR patients, but the mortality rates were markedly different. The Archives mortality rates were 23.1% for the DNR and 8.4% for the non-DNR versus the CCM cohorts which were 12.5% and 5.2% respectively.

The Archives paper noted a significantly higher rate of major complications for the DNR patients with no difference in minor complications whereas the CCM paper reported that the DNR patients actually had significantly fewer wound infections and no difference in the major complication rate.

I could go on but I think you get the idea. Both sets of authors did mention that their studies could be confounded by the limitations of retrospective studies and the use of an administrative database, such as the NSQIP.

What is not clear and certainly beyond my ability to explain, are the marked discrepancies in the number of patients found with DNR orders when the time period of both studies was identical and the substantial differences in mortality and complication rates.

Bottom line. Although they are at a higher risk of death, DNR patients may undergo necessary surgical procedures with a reasonable expectation of success.

*Churnalism is a news article that is published as journalism, but is essentially a press release without much added


Anonymous said...

Not surprising. Just read both method sections... one retrospective, one prospective. very different inclusion/exclusion (CCM excluded all pts on vent duirng 48 hours prior, sepsis, coma>24 hours.

Once they got their data (quite differently), they analyzed them very differently.

Basically, I would have to say, garbage in... garbage out...squared.

Skeptical Scalpel said...

I appreciate your comments. I try to keep my posts under 500 words and did not go into detail about the methods.

The data for both papers were collected prospectively but both analyzed the data retrospectively. In other words, the database was "mined." That's how the NSQIP works.

I find it hard to believe that there were nearly 2000 patients excluded due to mechanical ventilation, sepsis and coma. Also, the statistics that were used by the two sets of authors were different.

I am glad to see that you agree with my main point. You just said it more bluntly.

Anonymous said...

This is why EBM is so hard to actually apply (its painful to learn and much easier to just swallow what is fed to us!). Only a small few actually like the statistics and most practicing physicians don't want to have to read the methods.. just the abstracts. I now only skim abstracts to see if the topic interests me and then go straight to the methods. cuts down on the number of articles I actually read!

The amount of actual clinically usable literature is exceptionally small. There is a decided lack of critical thinking skills in the world, and particularly in medicine.

Throw in the NYT science section and every other media outlet, and its just horrifying. :) I am reminded of reading some things about "spermgate" and a media person saying: the lifesaving greenfield filter. Still looking for that article as I am pretty sure the literature shows no change in mortality. *sigh* /rant.

Skeptical Scalpel said...

Very few people understand statistics well. Yesterday I was involved in a Twitter debate with two others, one of whom tweets as DietsandScience, who insisted that vegetarians have a higher overall mortality rate than meat-eaters. This was based on an article that said the all-cause death rate ratio was 1.03 (95% CI: 0.90, 1.16). The CI includes the number 1, thus it is not a significant difference.

Almost nothing that we do is really evidence-based. Sometimes that isn't so bad. For example, did you know that the parachute, which has been used for years to escape a crippled aircraft, has never been the subject of a randomized clinical trial?

Post a Comment