Predicting Surgery: Accuracy of the BrAIST Endpoint Definitions at 2-Year Follow-up

Lori Dolan¹ , Stuart Weinstein¹

1) University of Iowa Department of Orthopaedics and Rehabilitation

Introduction

BrAIST defined “success” as Risser 4+ and Sanders 7 with a Cobb angle of <50 degrees, at point where future surgery would be unlikely, whereas surgery was expected in the “failed” group (>50 degrees prior to maturity). The prognostic accuracy of these endpoints has been questioned, as curves may continue to progress and surgery may still occur post-maturity. Thus, this study evaluated the predictive value of the BrAIST endpoints via a minimum 2-year radiographic/surgical follow-up.

Research Question
What is the prognostic accuracy of the BrAIST definitions of success and failure?

Methods

The 272 subjects who completed BrAIST were eligible for follow-up if surgery occurred or at 2 years-post BrAIST. The 25 BrAIST sites located and consented their own subjects. A pre-op or follow-up film was submitted to the coordinating center for measurement. Progression was calculated as the difference between the maximum Cobb angle at BrAIST exit and at follow-up. The surgery rate, and the mean length of follow-up and curve progression were calculated. Prognostic accuracy was judged via the positive predictive value (PPV, % surgery in the “failed” group) and negative predictive value (NPV,% no surgery in the “success” group).

Results

Of 272 subjects, we obtained documentation of surgery (date only or pre-op x-ray) or a new x-ray for 198(73%); 115 (85%)in the success group(SG)and 83(94%)in the failed group(FG). 15% of the SG group underwent surgery compared to 94% of the FG. X-rays were available for 108 patients in the SG; of 83 (78%) with no curve progression 8 had surgery; 25 curves progressed and 6 of these were operated on. The mean pre-op Cobb angle was 48.1(range 40-58). Surgery in the SG occurred at an average of 2.5 years (range 6 mo to 4.2 years)after skeletal maturity. In this sample, the PPV was 94% (all but 6% of the FG had surgery) and the NPV was 85% (15% of the SG had surgery).  Overall, the BrAIST definitions predicted surgery correctly in 89% of subjects. 

Conclusions
BrAIST endpoints were set with future curve behavior in mind, yet surgery decisions not solely based on Cobb angles. Curves as small as 40 degrees were operated in this sample, but one at 63 degrees was not. Despite this variation, the BrAIST definitions of “success” and “failure” were correct for 89% of this sample at a minimum of 2-year follow-up.

Discussion

These endpoints therefore seem reasonable for use in future studies. 

Disclosures (any Conflicts of Interest)
The authors have no conflicts of interest related to this work.  The work was funded by the Joan and Phill Berger Charitable Fund.