External validation of the BABY‐Knee Algorithm for treatment decision‐making in skeletally immature patients with anterior cruciate ligament injuries: A Dutch cohort study
Annika Babette Ito, Rob Paulus Augustinus Janssen, Stijn Thomas Henricus Croes, Robin Nicolaas Julian Voskuilen, Alberto Grassi, Maria Christine van der Steen, Martijn DietvorstAbstract
Purpose
The aim of this study was to investigate the external validity of the Best anterior cruciate ligament (ACL)‐treatment Based on the Years of the Knee (BABY‐Knee) Algorithm, developed based on an Italian population and to assess its applicability to a Dutch patient population treated in accordance with the 2018 International Olympic Committee (IOC) consensus statement.
Methods
A total of 109 skeletally immature patients with ACL injuries were included from the Dutch contribution to the European Society of Sport Traumatology, Knee Surgery and Arthroscopy (ESSKA) Paediatric Monitoring Initiative (PAMI). The BABY‐Knee Algorithm was retrospectively applied to all patients, and corresponding scores were calculated. The predictive performance of the BABY‐Knee Algorithm was assessed using the validation approach of the original validation study and an additional failure criterion based on hard clinical endpoints. Results were compared with the previously reported results.
Results
Of the 109 patients included, 78 (71.6%) were treated non‐operatively, while the remaining 31 (28.4%) underwent operative treatment. However, according to the algorithm, 34 patients (31.2%) had a score of <3, indicating non‐operative treatment, whereas 75 patients (68.8%) had a score of ≥3, indicating operative treatment. A statistically significant difference was observed between the algorithm's prediction and the performed treatment ( p = 0.003). The original internal validation study reported a positive predictive value (PPV) of 91.7% and a negative predictive value (NPV) of 87.5%. In contrast, application of the BABY‐Knee Algorithm to the present study population resulted in a PPV of 51.1% or 31.9% and an NPV of 64.5% or 67.7%, depending on the applied definition of failure.
Conclusion
The external validation of the BABY‐Knee Algorithm showed inferior predictive performance compared with the original internal validation study in an Italian population. These findings suggest that the BABY‐Knee Algorithm is not directly applicable to other populations of skeletally immature patients with ACL injuries.
Level of Evidence
Level III.