How Accurate Are Fulcrum Bending Radiographs in Estimating Postoperative Outcomes in Adolescent Idiopathic Scoliosis? A Systematic Review and Meta-analysis

BACKGROUND: Fulcrum bending radiographs can be used to assess coronal flexibility in patients with adolescent idiopathic scoliosis (AIS) to estimate postoperative correction. To obtain fulcrum bending radiographs, patients are passively bent over a radiolucent fulcrum at the apex of the curve. Available studies have disagreed about the accuracy in estimating postoperative correction, although these studies differed in terms of patients' baseline characteristics as well as other methods. Moreover, factors associated with accuracy were never explored. By pooling (meta-analyzing) results from these studies, we hoped to address these gaps in knowledge. QUESTIONS/PURPOSES: In a meta-analysis, we asked: (1) Can fulcrum bending radiographs accurately estimate postoperative curve correction in patients with AIS? (2) What factors are associated with the accuracy of fulcrum bending estimation on postoperative coronal correction? (3) Is fulcrum flexibility associated with other surgical outcomes such as shoulder and coronal balance? METHODS: PubMed, Embase, Medline, Journals@Ovid, Web of Science, and Scopus were searched from their inception up to August 27, 2024. Studies that (1) included patients with AIS undergoing single-stage posterior spinal fusion surgery without anterior release, (2) used fulcrum bending radiographs, (3) assessed radiographic surgical outcomes, and (4) had a minimum follow-up of 2 years were included. Studies that did not evaluate the use of fulcrum bending radiographs, those that did not report a p value, and studies with poor methodological quality were excluded. Our initial search yielded 433 articles, of which 172 remained after duplicate articles were removed. A total of 161 articles were excluded as the studies included patients who did not have AIS (n = 14), did not undergo surgery (n = 14), or did not undergo posterior spinal fusion (n = 23) or the studies did not evaluate the use of fulcrum bending radiographs (n = 59); had an insufficient follow-up duration of < 2 years (n = 15); did not evaluate the relationship between fulcrum bending radiographs and postoperative outcomes (n = 1); were reviews, commentaries, articles, conference proceedings, or non-English studies (n = 33); were animal studies (n = 1); or had poor methodological quality (n = 1). This left 11 studies for analysis. The Newcastle-Ottawa Quality Assessment Scale was used to evaluate the quality of evidence in three domains, including participant selection, comparability, and outcome measurement. Eleven included studies were of good quality except one with poor-quality evidence that was subsequently excluded from analysis. A random-effects meta-analysis was used to pool the data because of substantial statistical heterogeneity (I 2 > 50%) in the included studies. The estimation of absolute correction was pooled using standardized mean differences, referred to as the mean difference; a value > 0 indicated overestimation and vice versa. Estimation of percentage correction was pooled using ratio of means between correction rate and fulcrum flexibility, referred to as fulcrum bending correction index (FBCI); a value > 1 indicated underestimation and vice versa. RESULTS: Fulcrum bending radiographs tended to underestimate postoperative curve correction, although the difference was not clinically important (immediate postoperative mean difference -0.6° [95% confidence interval (CI) -0.9° to -0.4°], p < 0.001; immediate postoperative FBCI 1.15 [95% CI 1.09 to 1.21], p < 0.001; 2-year follow-up mean difference -0.43° [95% CI -0.6° to -0.2°], p < 0.001; 2-year follow-up FBCI 1.10 [95% CI 1.04 to 1.16], p = 0.001). To address the high between-study heterogeneity, we adjusted for potential confounders, which found that more flexible curves (regression coefficient 0.07 [95% CI 0.01 to 0.13]; p = 0.02) and proximal thoracic (immediate postoperative main thoracic versus proximal thoracic curves mean difference -0.8° [95% CI -1.4° to -0.2°], p = 0.01; 2-year follow-up main thoracic versus proximal thoracic curves mean difference -0.7° [95% CI -1.3° to -0.1°], p = 0.03) curves were associated with less underestimation. Segmental and alternate level screw placement were associated with underestimation of curve correction by fulcrum bending radiographs, although the difference was clinically unimportant. The degree of underestimation was worse with segmental screw placement at immediate postoperative (mean difference -1.0° [95% CI -1.9° to -0.1°]; p < 0.001) and 2-year follow-up (mean difference -1.0° [95% CI -1.6° to -0.4°]; p < 0.001). However, evidence surrounding more serious underestimation in segmental compared with alternate level screw placement was uncertain as only one study used a segmental screw placement strategy. Regarding the relationship between fulcrum flexibility and other radiographic outcomes, more rigid main thoracic curves were at risk of coronal imbalance, while more flexible curves were associated with postoperative shoulder imbalance. However, the evidence was inconclusive as it was reported by two or fewer studies. CONCLUSION: Fulcrum bending radiographs offer a reliable estimate of postoperative coronal correction; the amount of underestimation that we observed on some endpoints was too small to be clinically meaningful. Although there was substantial statistical heterogeneity, the direction of effect was similar across all studies. Fulcrum bending estimation was also reliable when using alternate pedicle screw constructs. More flexible curves and proximal thoracic curves were associated with less underestimation. In more rigid curves, results of fulcrum bending estimation should be interpreted with caution, and alternate flexibility assessment methods such as traction should be considered. LEVEL OF EVIDENCE: Level III, diagnostic study.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.