Analysis of risk factors and construction of nomogram model for arthroscopic single-row rivet repair.

BACKGROUND

The factors influencing the clinical outcome of arthroscopic rotator cuff repair are not fully understood.

PURPOSE

To explore the factors related to the postoperative outcome of arthroscopic single-row rivet rotator cuff repair in patients with rotator cuff injury and to construct the related nomogram risk prediction model.

METHODS

207 patients with rotator cuff injury who underwent arthroscopic single-row rivet rotator cuff repair were reviewed. The differences of preoperative and postoperative Visual Analogue Score (VAS) scores and University of California, Los Angeles (UCLA) scores were analyzed and compared. The postoperative UCLA score of 29 points was taken as the critical point, and the patients were divided into good recovery group and poor recovery group, and binary logstic regression analysis was performed. According to the results of multivariate logistic regression analysis, the correlation nomogram model was constructed, and the calibration chart was used, AUC, C-index. The accuracy, discrimination and clinical value of the prediction model were evaluated by decision curve analysis. Finally, internal validation is performed using self-random sampling.

RESULTS

The mean follow-up time was 29.92 ± 17.20 months. There were significant differences in VAS score and UCLA score between preoperative and final follow-up (p < 0.05); multivariate regression analysis showed: Combined frozen shoulder (OR = 3.890, 95% CI: 1.544 ∼ 9.800), massive rotator cuff tear (OR = 3.809, 95%CI: 1.218 ∼ 11.908), More rivets number (OR = 2.118, 95%CI: 1.386 ∼ 3.237), lower preoperative UCLA score (OR = 0.831, 95%CI: 0.704-0.981) were adverse factors for the postoperative effect of arthroscopic rotator cuff repair. Use these factors to build a nomogram. The nomogram showed good discriminant and predictive power, with AUC of 0.849 and C-index of 0.900 (95% CI: 0.845 ∼ 0.955), and the corrected C index was as high as 0.836 in internal validation. Decision curve analysis also showed that the nomogram could be used clinically when intervention was performed at a threshold of 2%∼91%.

CONCLUSION

Combined frozen shoulders, massive rotator cuff tears, and increased number of rivets during surgery were all factors associated with poor outcome after arthroscopic rotator cuff repair, while higher preoperative UCLA scores were factors associated with good outcome after arthroscopic rotator cuff repair. This study provides clinicians with a new and relatively accurate nomogram model.