FoughtFyr said:
That said,
Dr. Krueger and the other article you posted both cite this:
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=9715832
I will have to get a copy and read it but at first glance at the abstract there are SERIOUS methodological questions. I'll reserve judgement until I can read it.
- H
O.k., I've got the entire article. It does not even come close to "proving" anything. The author, who is intimately involved in the creation of a database known as the Arthritis, Rheumatism, and Aging Medical Information System (or ARAMIS).
ARAMIS, "which has been funded by the National Institutes of Health for the past 25 years, is a prospective observational data bank system that systematically collects data on individuals with chronic rheumatic disease." O.k., so where is the problem? Here is the ARAMIS data:
Gastrointestinal (GI) Complications in Osteoarthritis
(OA) and Rheumatoid Arthritis (RA)
OA Hospitalizations
Number of patients 1,283
Person-years of observation 3,234
Person-years taking NSAIDs 2,199
Number of GI events 19
GI events in NSAID users 16
Annual incidence on NSAIDs (%) 2.51
Relative risk on NSAIDs 0.73 (SE 5 0.18)*
Annual incidence of upper GI events on NSAIDs (%) 0.50
Annual incidence of lower GI events on NSAIDs (%) 0.23
RA Hospitalizations
Number of patients 2,921
Person-years of observation 12,224
Person-years taking NSAIDs 8,471
Number of GI events 134
GI events in NSAID users 124
Annual incidence on NSAIDs (%) 5.49
Relative risk on NSAIDs 1.46 (SE 5 0.13)*
Annual incidence of upper GI events on NSAIDs (%) 1.27
Annual incidence of lower GI events on NSAIDs (%) 0.19
(* P , 0.001 for RA vs OA.) {emphasis added}
From this, the author concludes: "The Arthritis Foundation conservatively estimates that at least 13 million individuals in the United States with OA or RA regularly take NSAIDs. Applying the ARAMIS data to these figures, the number of potential hospitalizations for serious GI complications is about 107,000 per year." Which sounds great, right? Except for two facts. Elsewhere in the paper, the author describes additional risk factors that seem to suggest which specific persons taking NSAIDs are at increased risk. These data are not applied to the "conservative estimate" of 13 million individuals who regularly take NSAIDs. Second, as is contained in the data above, individuals with OA are far less likely than individuals with RA to develop GI complications (RR 0.73 v. 1.46). OA is far more prevalent than RA in the general population. Again, no measure of this is made in the creation of the author's estimate of those affected by GI complications from NSAIDs. Additionally, no mention is made at all of the pre-existing (or "base rate") of GI pathology except to note that there were events in both the OA and RA populations among individuals not taking NSAIDs. There are many individuals not on NSAIDs develop GI problems, yet the author only compares his data between the OA and RA groups, not against controls. It is impossible then to draw the conclusion that the entirety of the pathology present can be "blamed" on NSAID use - especially given that the data only contain 153 events out of more than 4,000 patients.
A better look at the issue is here;
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=1834002
but their results are not as dramatic.
But, let's for a minute compare the results of the first article to the risk of vertebral artery dissection, one that the "chiro defenders" here seen to all say is minimal at best. From:
http://www.ncbi.nlm.nih.gov/entrez/...ve&db=pubmed&dopt=Abstract&list_uids=14589464
"The total number of chiropractic practice years assessed for this 10 year period was 32,885."
"A total of 43 cases were referred to CCPA during the study period. Twenty-three cases were confirmed to be a stroke that occurred at some point after cervical manipulation, and 21 of these cases resulted in claims of malpractice. The remaining 20 cases, on review of the charts, were determined by the treating physician or neurologist to be the result of factors unrelated to the treatment or the symptoms resolved without significant investigation."
Before you cry foul, remember the NSAID paper only discussed association, it did not exclude any cases for "base rate", all associated cases were "blamed" on the NSAIDs. So, we have 43 cases in 32,885 chiropractic practice years for a rate of 1 event per 764.77 years. In the OA population taking NSAIDs we have 16 events in 2,199 person years of observation for a rate of 1 event per 137.43 years. So, if a chiropractor takes NSAIDs for OA pain, he is only 5.56 times more likely to have a GI complication than he is to cause a VAD in any given year. Since many on this board like to equate VAD from chiropractic therapy to "getting hit by lightning" then something 5x more likely than getting hit by lightning isn't such a risk after all, is it. And consider this, the data for the GI article contained no deaths in the OA group. The data above measure all GI complications, including those as mild as dyspepsia...
Yes, PublicHealth, I know the last paragraph is statistically pure junk. I am just making the point that there are three types of falsehoods: lies, damn lies, and statistics. The vision skiiboy posted of "hundreds of thousands of deaths a year" (attributable to NSAIDs) isn't quite true either...
- H