TITLE: Clinical methodologies and incidence of appropriate statistical testing in orthopaedic spine literature.
Are statistics misleading?
Vrbos LA; Lorenz MA; Peabody EH; McGregor
MAUTHOR AFFILIATION: Loyola University
Medical Center, Maywood, Illinois.
SOURCE: Spine 1993 Jun
15;18(8):1021-9CITATION IDS: PMID: 8367770 UI: 93377123
ABSTRACT: An analysis of
300 randomly drawn orthopaedic spine
articles, published between 1970 and 1990,
was performed to assess the quality of
biostatistical testing and research design
reported in the literature.
Of the 300 articles, 269 dealt with topics of an experimental nature, while 31 documented descriptive studies.
Statistical deficiencies were identified in 54.0% of the total articles.
Conclusions drawn as the result of misleading significance values occurred in 124 experimental studies (46%) while 96 failed to document the form of analysis chosen (35.7%).
Statistical testing was not documented in 34 studies (12.6%), while 20 (7.4%) employed analyses considered inappropriate for the specific design structure.