TYPE II ERROR IN THE SPINE SURGICAL LITERATURE

Abstract

This study examined the occurrence of type II (Beta) error in the spine surgical literature. A literature search from 1966 until present identified twenty-nine randomized control trials, which had a two-group parallel design and reported a non-significant difference in the primary outcome measure. We determined whether these trials had sufficient sample size to detect a 25% and 50% relative change in the primary outcome. Nine studies specifically identified a primary outcome. Four studies reported a sample size calculation. Therefore, twenty-five trials were at risk of committing a type II error.

The purpose of this study was to determine the frequency of potential type II error in randomized controlled trails reported in the spine surgical literature.

A literature search was conducted of the Medline, Pubmed, and the Cochrane databases using the keywords of “spine” and “surgery”, between 1967 until present to identify randomized controlled trials involving spine surgery. Trials were included in this study if they were of a two-group design, with at least one of the groups being a surgical cohort, and that reported a non-significant difference in the primary outcome. We determined the frequency for which the primary outcome and sample size calculation was reported. The sample size was assessed to determine whether the trial had sufficient subjects to detect a 25% and 50% relative difference in the primary outcome for a power of 80%.

Twenty-nine studies satisfied the inclusion criteria. Nine studies specifically identified a primary outcome. All others reported multiple outcomes with no specified primary measure. Four studies reported a sample size calculation. Of the remaining twenty-five that did not, three had sufficient power and the rest were at significant risk of committing a type II error.

The spine surgical literature is plagued with a high potential for type II errors in the published trails with a non-significant outcome.

In the spine surgical literature, a randomized controlled trial that fails to reject its null hypothesis, requires careful scrutiny of its methodology to prevent misinterpretation of the results.