Abstract
Background Two recent studies have reported contradicting conclusions regarding the prognostic value of preoperative posterior tilt, following internal fixation of undisplaced fractures of the femoral neck. We therefore conducted a study based on a longitudinal cohort of 322 patients with a median follow-up of 943 days (range, 4 – 2963). Specifically, we wanted to investigate if differing methodology could explain some of the discrepancies. Method Two raters evaluated preoperative posterior tilt, and measurements were divided into two or three categories, in accordance with the previously published reports. We also simplified the assessment of posterior tilt, describing fractures as displaced or non-displaced . Both logistic and cox regression models were used to evaluate the influence of posterior tilt upon the risk of fixation failure, adjusting for the effects of age, gender, ASA-score, cognitive function and delay to surgery. Results There were a total of 44 (14%) fixation failures, of which non-union caused 31 and the remaining 13 were due to avascular necrosis (AVN). All AVNs occurred later than the first year. Based on our data, posterior tilt was associated with risk of fixation failure during the first year with odds-ratio (95% CI) of 2.8(1.1-7.0) (p=0.03). However, using cox regression for the entire follow-up period, the risk was no longer statistically significant, and the hazard ratio (HR) was 2.1(0.9-5.0) (p = 0.07). When failures were separated into non-union and AVN, posterior tilt predicted the risk of non-union with HR of 3.4(1.3 – 9.1) (p = 0.01), but not AVN, HR 1.6(0.5-4.9) (p=0.38). The simplified classification based on displacement, predicted risk of failure regardless of statistical model used. Conclusion Posterior tilt was associated with risk of fixation failure during the first year, but not for the entire follow-up period. Our data therefore suggest that the previously reported conflicting results can partially be explained by differences in methodology. We also show that a simplified classification of posterior tilt predicts risk of failure, regardless of methodology applied.
Background Two recent studies have reported contradicting conclusions regarding the prognostic value of preoperative posterior tilt, following internal fixation of undisplaced fractures of the femoral neck. We therefore conducted a study based on a longitudinal cohort of 322 patients with a median follow-up of 943 days (range, 4 – 2963). Specifically, we wanted to investigate if differing methodology could explain some of the discrepancies. Method Two raters evaluated preoperative posterior tilt, and measurements were divided into two or three categories, in accordance with the previously published reports. We also simplified the assessment of posterior tilt, describing fractures as displaced or non-displaced . Both logistic and cox regression models were used to evaluate the influence of posterior tilt upon the risk of fixation failure, adjusting for the effects of age, gender, ASA-score, cognitive function and delay to surgery. Results There were a total of 44 (14%) fixation failures, of which non-union caused 31 and the remaining 13 were due to avascular necrosis (AVN). All AVNs occurred later than the first year. Based on our data, posterior tilt was associated with risk of fixation failure during the first year with odds-ratio (95% CI) of 2.8(1.1-7.0) (p=0.03). However, using cox regression for the entire follow-up period, the risk was no longer statistically significant, and the hazard ratio (HR) was 2.1(0.9-5.0) (p = 0.07). When failures were separated into non-union and AVN, posterior tilt predicted the risk of non-union with HR of 3.4(1.3 – 9.1) (p = 0.01), but not AVN, HR 1.6(0.5-4.9) (p=0.38). The simplified classification based on displacement, predicted risk of failure regardless of statistical model used. Conclusion Posterior tilt was associated with risk of fixation failure during the first year, but not for the entire follow-up period. Our data therefore suggest that the previously reported conflicting results can partially be explained by differences in methodology. We also show that a simplified classification of posterior tilt predicts risk of failure, regardless of methodology applied.