To evaluate whether in children with knee pathology there is any correlation between clinical diagnosis, magnetic resonance imaging and arthroscopy.

Between 1993 and 2001 children age 3–16 years old, who presented in the orthopaedic clinics of our institution with knee pathology were included in this study. All of them underwent MRI investigation. Their history, physical examination and clinical diagnosis were ascertained from their case notes. Some of these children underwent arthroscopic surgery of the knee and findings were also recorded. Clinical data, MRI findings and arthroscopic findings were computerised and analysed. Results were analysed and compared in the following 3 groups: a) clinical data versus MRI findings, b) clinical data versus arthroscopic findings and c) MRI report versus arthroscopic findings. Comparisons were rated in one of three categories: total agreement, partial agreement or total disagreement. Partial agreement was defined as the partial correlation of findings.

130 children (131 knees, one bilateral) were included in this study. The mean age was 8.5 years (range 3–16). 81 were male and 49 were female, ratio 1.7:1. 38 (30%) patients underwent arthroscopy. 43 (33%) of the MRI scans were reported as normal. Lesions reported on MRI included meniscal and ACL tears, osteochondritis dessicans, osteochondral fractures and discoid lateral meniscus. Overall, the results between the comparison of the 3 groups are summarised as follows:

In this study 1/3 of the knee MRI was normal and there was only 26% of total agreement between the clinical and MRI findings. Further more in 50% of cases that underwent arthroscopy, there was no correlation of arthroscopic and MRI findings. This study supports the view that knee MRI investigation in children may not provide a reliable diagnosis and guidance in children with knee pathology.