Intended for healthcare professionals

Rapid response to:

Feature Evidence Based Medicine

Why we can’t trust clinical guidelines

BMJ 2013; 346 doi: https://doi.org/10.1136/bmj.f3830 (Published 14 June 2013) Cite this as: BMJ 2013;346:f3830

Rapid Response:

Re: Why we can’t trust clinical guidelines

There have been multiple inquiries requesting the details of the estimation of the increased death toll resulting from the use of MPSS in spinal cord injury based on increased death rate from the use of MPSS in the CRASH head injury study1.

The details of this estimation are as follows:
Starting estimates:
a) 10,000 acute SCI/year occur in the US. (Considered a low estimate. The National Spinal Cord Injury Statistical Center, Birmingham, Alabama lists an incidence of 12,0002.)
b) 3.4% increase in death rate from the use of MPSS (p=0.0001) in the CRASH head injury [(25.7% with MPSS) - (22.3% without MPSS) = 3.4% increase in death rate from MPSS use. This was significant enough to cause the study to halt its planned enrollment].
c) SCI patients were at risk for 22 years (22 years since NIH notification of ERs and then the subsequent NEJM3 publication in 1990).

Thus, the group at risk would be:
(10,000 SCI patients/year) x (22 years) = 220,000 total SCI patients at risk.

If the total 3.4% increase in death rate from MPSS is transferred to the SCI patients:
(220,000 SCI patients) x 0.034 = 7,480 potential SCI patients at risk for an increase in death rate.

However, some of the SCI would be single system injuries, cervical fracture only, and not be placed on ICU ventilator treatment. Some of the spinal cord injuries would also have head injuries. I estimated that 1/2 of the SCI patients are multiple injury and on ICU ventilator treatment in their initial hospital course, and thus considered to be the truly at risk patients for the MPSS complication of increased death rate.

Thus, the increased death rate for MPSS in SCI is estimated to be 1/2 of the 7,480 figure, or 3740 extra deaths from MPSS use in SCI, as inferred from the CRASH study by the estimations and logic shown above. This is similar to previous estimates4.

To set this estimated death total of 3740 in perspective, the attacks on September 11, 2001 were used as a benchmark for an unacceptably high death toll. The 9/11 attacks actually had a lower death total of 2,996 people5, (19 hijackers and 2,977 victims). The victims included 246 on the four planes, 2,606 in New York City in the towers and on the ground, and 125 at the Pentagon.

Even if someone claimed the above estimate is high and lowered it by an order of magnitude, then even this unrealistic low-ball estimate would predict 374 extra patient deaths for a treatment with no proven clinical value and for which the primary data has never been released for a true review.

The author and BMJ requested and received these data and analysis prior to publication of the article.

References:

1. Edwards P, Arango M, Balica L, et al. Final results of MRC CRASH, a randomised placebo-controlled trial of intravenous corticosteroid in adults with head injury-outcomes at 6 months. Lancet. 2005;365(9475):1957-1959.
2. Spinal Cord Injury Facts and Figures at a Glance. 2013; https://www.nscisc.uab.edu/PublicDocuments/fact_figures_docs/Facts%20201....
3. Bracken MB, Shepard MJ, Collins WF, et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury. Results of the Second National Acute Spinal Cord Injury Study. The New England journal of medicine. May 17 1990;322(20):1405-1411.
4. Geisler FH. NASCIS publications abused the processes of peer-review, publication, regulatory approval, and clinical consensus, rapid response after, Excessively closed science hurts. BMJ. 2008-03-20 00:00:00 2008;336(7645):629-629.
5. Wikipedia. September 11 attacks. 2013; http://en.wikipedia.org/wiki/September_11_attacks.

Competing interests: No competing interests

24 June 2013
Fred H Geisler
Neurosurgeon
Chicago Back Institute at Swedish Covenant Hospital
5145 North California Ave., Chicago, IL 60625 USA