Most over-rated paper ever published

[injuryupdate]

I would like to nominate the following paper as the most over-rated paper I have ever seen in sports medicine:

The Effect of Neuromuscular Training on the Incidence of Knee Injury in Female Athletes
A Prospective Study
Timothy E. Hewett, PhD*, Thomas N. Lindenfeld, MD, Jennifer V. Riccobene and Frank R. Noyes, MD
Cincinnati Sportsmedicine Research and Education Foundation and Deaconess Hospital, Cincinnati, Ohio

To prospectively evaluate the effect of neuromuscular training on the incidence of knee injury in female athletes, we monitored two groups of female athletes, one trained before sports participation and the other not trained, and a group of untrained male athletes throughout the high school soccer, volleyball, and basketball seasons. Weekly reports included the number of practice and competition exposures and mechanism of injury. There were 14 serious knee injuries in the 1263 athletes tracked through the study. Ten of 463 untrained female athletes sustained serious knee injuries (8 noncontact), 2 of 366 trained female athletes sustained serious knee injuries (0 noncontact), and 2 of 434 male athletes sustained serious knee injuries (1 noncontact). The knee injury incidence per 1000 athlete-exposures was 0.43 in untrained female athletes, 0.12 in trained female athletes, and 0.09 in male athletes (P = 0.02, chi-square analysis). Untrained female athletes had a 3.6 times higher incidence of knee injury than trained female athletes (P = 0.05) and 4.8 times higher than male athletes (P = 0.03). The incidence of knee injury in trained female athletes was not significantly different from that in untrained male athletes (P = 0.86). The difference in the incidence of noncontact injuries between the female groups was also significant (P = 0.01). This prospective study demonstrated a decreased incidence of knee injury in female athletes after a specific plyometric training program.

[injuryupdate]

The reason for this is that this paper has already been cited on many occasions, and will continue to be cited, as it reports in an area of great interest (prevention of ACL injuries by balance training) and was published in a high profile journal.

Those fans of citation index will see this paper as a classic. However, when you read the paper, it is a classic example of "how to lie with statistics".

The subjects in this paper played different sports and could volunteer to be part of the training program (therefore the methodology was neither randomised nor blinded). Far more volleyball players chose to be part of the training program than soccer and basketball players, meaning that this paper probably proves that you are less likely to tear an ACL playing volleyball than soccer or basketball (wow!). None of the volleyball players were injured at all, so they should have just been left out of the study. In addition, the "knee" injuries included MCL injuries as well as ACL injuries, apparently just to buff up the numbers of injuries in the "untrained" group (as there were five MCLs in the untrained group and none in the trained group - but how balance training can prevent an MCL is beyond me).

In the paper itself, these major flaws are revealed. The published abstract is a disgrace and an embarassment to the journal and the authors. Without revealing how rigged the study was, the abstract truimphantly reports that this intervention study successfully lowered injury rates.

It IS actually likely that training programs such as the one described may actually work, but it is not forgiveable to draw this conclusion from such a terribly designed study. Because the flaws are not revealed in the abstract, this study will get the credit (and citations) that it is NOT due in the future.

[injuryupdate]

see attachment Letter to the Editor