Thurs., 10/13/11 Polytrauma, Paper #31, 3:31 pm OTA-2011
Loss of Follow-Up in Orthopaedic Outcome Studies: Is 80% Follow-Up Still Acceptable?
Boris A. Zelle, MD1; Mohit Bhandari, MD, MSc2; Alvaro I. Sanchez, MD, MS1;
Christian Probst, MD3; Hans-Christoph Pape, MD4;
1University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, USA;
2McMaster University, Hamilton, Ontario, Canada;
3University of Witten/Herdecke, Witten, Germany;
4University of Aachen, Aachen, Germany
Purpose: Loss of follow-up is a potential source of bias in orthopaedic outcome studies. Previously suggested guidelines propose 20% loss of follow-up as acceptable. However, these guidelines for acceptable loss of follow-up were arbitrarily derived and have not been established through scientific investigations. The goal of this study was to evaluate how loss of follow-up influences the statistical significance in an orthopaedic database.
Methods: In this simulation study, we used a database of 637 polytrauma patients who had been reexamined at an average follow-up of 17.5 years (range, 10-28 years) after their injury. The functional outcome as measured by the Hannover Score for Polytrauma Outcome (HASPOC) of workers’ compensation patients versus non–workers’ compensation patients was compared using a logistic regression model with adjustments for age, gender, ISS, and head injuries. As documented in a previous publication, a significant difference between the 2 groups was found (P <0.05). In the current simulation study, we simulated a gradually increasing loss of follow-up by randomly deleting an increasing number of patients from 2%, 5%, 10%, and then increasing in increments of 5% until the significance changed. This process of simulating an increasing loss of follow-up was repeated 50 times. A different electronic random generator was used in each of these 50 simulation series. In each simulation series, we documented the simulated loss of follow-up at which the results turned from statistically significant to nonsignificant at the level of P = 0.05.
Results: Among the 50 simulation series, the turning point from significant to nonsignificant varied between 15% loss of follow-up and 75% loss of follow-up. The average turning point from significant to nonsignificant was at 40% ± 18.3% simulated loss of follow-up. In 14 of the 50 simulation series (28%), the results changed from significant to nonsignificant with a simulated loss of follow-up of 20% or less.
Conclusions: A loss of follow-up of 20% or less may frequently change the study results. In the planning phase of clinical studies, researchers should carefully establish protocols to minimize the loss of follow-up. The actual loss of follow-up should be clearly stated in manuscript publications.
Alphabetical Disclosure Listing (628K PDF)
• The FDA has not cleared this drug and/or medical device for the use described in this presentation (i.e., the drug or medical device is being discussed for an “off label” use). ◆FDA information not available at time of printing. Δ OTA Grant.