Orthopedic research is usually performed on shoulders, elbows, hips, knees, ankles, etc., and the presentation of body parts is usually given precedence before patients when the researchers describe what they have studied.
We studied 52 hips (30 patients)…
It is, however, a fundamental assumption underlying many statistical methods that all analysed observations are independent of all other. This assumption is not fulfilled when body part is used as analysis unit, instead of patient, and when one or more patients contribute more than one body part each. Statistical methods for dealing with this problem have been developed, but they are complex and not generally known.
Two systematic reviews have studied the prevalence of articles presenting research findings based on erroneous analyses of correlated observations.
Bryant et al. (1) reviewed 143 articles involving multiple observations from single individuals published during 2003 in the American and British volumes of JBJS, Clinical Orthopaedics and Related Research, The American Journal of Sports Medicine, Journal of Shoulder and Elbow Surgery, The Journal of Arthroplasty, and the The Journal of Arthroscopic and Related Research. Of the 143 articles 76 (53%) presented statistical tests between groups, and an additional 60 (40%) included group comparisons without statistical tests. Of the 136 studies with group comparisons only 16 (12%) used any technique or methodological approach to adjust the results for the correlation between observations.
Parke et al. (2) reviewed 486 articles published in the American volume of JBJS during 2007-2008. Bilateral cases were included in 151 articles, and of these 120 (79%) possibly violated statistical independence. It was unclear whether this actually was the case in 33 articles (22%), but it was clear in 87 (58%). The authors also showed that the correlation between observations on the same individual can lead to substantial underestimation of the uncertainty of the findings, unless it is taken into account.
Both author groups conclude that orthopedic researchers need to attend to this issue when reporting results.
1. Bryant D, Havey TC, Roberts R, Guyatt G. How Many Patients? How Many Limbs? Analysis of Patients or Limbs in the Orthopaedic Literature. JBJS Am 2006;88:41-45.
2. Park MS, Kim SJ, Chung CY, Choi IH, Lee SH, Lee KM. Statistical Consideration for Bilateral Cases in Orthopaedic Research. JBJS Am 2010;92:1732-1737.