Lies, Damned Lies, and Statistics

DianaCox

Bad Cop
Joined
Dec 30, 2013
Messages
6,343
Location
San Jose
It is unconscionable to me that surgeons would jigger the statistics in what is going to be a widely-read paper that PURPORTS to show relatively successful long term results with the VSG, in order to prop up a surgery they obviously want to do - but you have to actually READ THE PAPER to see the most egregious statistical abomination to understand how worthless their data actually is.

This is the recent publication in SOARD - here is the PubMed abstract - the paper is not accessible without a subscription, buying it, or being given access:

Surg Obes Relat Dis. 2014 January - February;10(1):177-183. doi: 10.1016/j.soard.2013.11.007. Epub 2013 Nov 21.
Review of long-term weight loss results after laparoscopic sleeve gastrectomy.
Diamantis T1, Apostolou KG2, Alexandrou A1, Griniatsos J1, Felekouras E1, Tsigris C1.
Author information

Abstract
Sleeve gastrectomy (SG) has gained enormous popularity both as a first-stage procedure in high-risk super-obese patients and as a stand-alone procedure. The objective of this study was to evaluate the long-term weight loss results after SG published in the literature and compare them with the well-documented short-term and mid-term weight loss results. A detailed search in PubMed using the keywords "sleeve gastrectomy" and "long-term results" found 16 studies fulfilling the criteria of this study. A total of 492 patients were analyzed, with a follow-up of at least 5 years after laparoscopic sleeve gastrectomy (LSG) (373 at 5 years, 72 at 6 years, 13 at 7 years, and 34 at 8 or more years). Of the total number of patients, 71.1% were women (15 studies, n = 432 patients). Mean patient age was 45.1 years (15 studies, n = 432 patients). Mean preoperative body mass index in all 16 studies was 49.2 kg/m2. The mean percentage excess weight loss (%EWL) was 62.3%, 53.8%, 43%, and 54.8% at 5, 6, 7, and 8 or more years after LSG, respectively. The overall mean %EWL (defined as the average %EWL at 5 or more years after LSG) was 59.3% (12 studies, n = 377 patients). The overall attrition rate was 31.2% (13 studies). LSG seems to maintain its well-documented weight loss outcome at 5 or more years postoperatively, with the overall mean %EWL at 5 or more years after LSG still remaining>50%. The existing data support the role of LSG in the treatment of morbid obesity.

Copyright © 2014 American Society for Bariatric Surgery. Published by Elsevier Inc. All rights reserved.

Sounds pretty good, right? But look what's missing from these stats (I have access to the paper, but can't post it - just excerpts):

"Statistical analyses were performed only on the extracted data from the selected studies, in patients who had not undergone another bariatric operation."

And:

"The long-term results after LSG are defined as the results at 5 or more years postoperatively, when LSG is performed without the addition of any other bariatric procedure."


WTF??? How do you purport to be providing data on the LONG-TERM RESULTS of VSG, when you EXCLUDE the failures who had to get a secondary procedure???

This entire paper is therefore WORTHLESS BULLSHIT without this information. They don't even give the number of patients who had another procedure for inadequate weight loss. It could be as much as 20 - 50% which would REALLY change the stats.

Ditto that they didn't report the results as stratified by starting weight or BMI, which would be REALLY useful information - I'll bet those data would show a cut off at about a BMI of 45 for acceptable results for standalone VSG.

Lying liars.
 
a few thoughts - first, the number of true "long term" patients is very low - 13 patients at 7 years, and 34 at 8 years. And no one for longer than that. Not exactly a lot of people to go on.

And I wonder what they mean by an attrition rate of 31.2%. Are these patients who were lost to follow-up? Often, the "lost to follow-up" group turns out to be people not doing so well and to embarrassed to go to follow-up, feeling that they are failures yet again.

And lastly, as we've seen before, anyone who manages to lose at least 50% of their excess weight is counted as a success. I bet a lot of those people don't feel so successful. Personally, I would have been totally dissatisfied with losing jut 54% of my excess weight. And resolution of comorbidities isn't even mentioned here. There is more to bariatric surgery than weight loss.

Is there a role for VSG as a stand alone? I believe there is. But is it adequate for most bariatric surgery patients? Not even close. I also see a role for VSG as a bridge to a larger operation for higher risk patients, but the present realities of insurance coverage make it extremely difficult for many patients to get a second operation no matter why they had the first one, and no matter who paid for the first one.
 
Ditto that they didn't report the results as stratified by starting weight or BMI, which would be REALLY useful information - I'll bet those data would show a cut off at about a BMI of 45 for acceptable results for standalone VSG.

I'm having a little trouble understanding what that would look like, Diana, but I've been awake since 2:30 am. boo.

my gut feeling (ha!) after years of reading people post is that VSG is for lightweights. less so for men, possibly.
 
I'm having a little trouble understanding what that would look like, Diana, but I've been awake since 2:30 am. boo.

my gut feeling (ha!) after years of reading people post is that VSG is for lightweights. less so for men, possibly.

It would involve assigning participants to starting weight categories and running a statistical analysis, probably an ANOVA, to compare outcomes among the groups of participants. The fact that I understand this having had six lowly undergrad credits of psych stats and research methods should tell you a lot about why Diana has a grumble in her voice as she mentions this issue: It really is a stats 101 thing. I suspect that if the authors had bothered with such an analysis (or maybe they did it and eliminated it from their article), it would have undermined their assertions about the wonderfulness of the VSG.
 
Manipulating your results to make your procedure look better... All I can say is I hope these folks learn their lesson after getting some severe criticism from their peers!
 

Latest posts

Back
Top