This study demonstrated the ability of functional imaging to predict the return-to-sports-time of athletes after a stress fracture. For this purpose, both the severity and location of the injury need to be determined. It could be shown that patients with stress fractures of low-grade and low-risk required significantly shorter RTSTs than patients with high-grade and/or high-risk stress fractures.
An estimation of return-to-sports-time plays an important role in the management and treatment of high performance athletes [6
]. Therefore a simple and reliable method of estimating the healing time of stress fractures is desirable. Due to the lack of sensitivity of plain radiographs, accurate diagnosis relies heavily on either magnetic resonance imaging (MRI) or bone scintigraphy (BS) [1
]. Both methods have been found to deliver reliable results of equivalent accuracy, with BS appearing to provide a slightly higher sensitivity, and MRI a higher specificity for the diagnosis of a stress fracture [12
In our study, stress fractures were rated as either high- or low-grade, using MRI and BS as equally suitable imaging modalities for the classification. Despite their use of different imaging modalities, Ishibashi et. al [16
]. showed an agreement of 86.1% in a prospective study of 36 cases, while Fredericson et. al [17
]. showed an agreement of 77.8% between MRI-grading and BS-grading in stress fractures. Both studies used a 4-point scale, from low- to high-grades of fracture. This study used a simplified 2-grade grading-system for BS and MRI, combining grade 1 and 2 to low-grade stress fractures and combining grade 3 and 4 to high-grade stress fractures. A previous study of ours demonstrated the usefulness of a simplification of the BS grading-system, showing a significant difference between high- and low-grade SFX. Furthermore, a close correlation was demonstrated between the healing times for grade 1 and 2 fractures as well as for the grades 3 and 4 [12
]. Arendt et al. have also shown a significant difference in healing time between high-grade and low-grade stress injuries using their MRI grading-system [5
]. This approach provides easier grading, satisfactory accuracy, and results of statistical significance. The almost perfect interobserver agreement (>0.80) for the grading of stress fractures by both modalities emphasizes the robustness and reliability of the chosen, simplified grading systems [18
With regard to the correlation of RTST and image-based grading, published results are contradictory. Studies failing to show a significant correlation have been published by Dutton et. al [19
] retrospectively evaluating 37 BS images of tibiae with stress fractures, Yao. et. al [20
] examining 35 stress fractures using an MRI grading system, and Fredericson et. al. investigating an MRI grading system for stress fractures with a study of 18 symptomatic tibiae [17
In contrast, a significant difference in RTST between high- and low-grade stress fractures was shown by Arendt et. al. who retrospectively compared the healing time of athletes (n
61) recovering from a stress fracture with an MRI-grading system [5
]. In a previous analysis evaluating a BS grading-system, our group was able to demonstrate a significant difference in healing time between high- and low-grade lesions [12
One reason for these discrepant results might be that all the aforementioned studies graded stress fractures on imaging alone, independent of the site of injury. For example, the tibia, a frequent and well-examined stress fracture site, contains multiple high-risk sites (anterior midshaft or medial malleolus
high-risk, posteromedial aspect
]. Recent case reports and publications therefore emphasize the need to differentiate between high- and low-risk sites for treatment and estimation of RTST [10
To our knowledge, the present analysis is the first study to assess RTST and grading based on MRI or bone scintigraphy in combination with SFX risk-classification.
According to our data, only stress fractures classified as low-grade and low-risk have a healing time that is significantly shorter than all other measured categories. As soon as a stress fracture is either high-risk or high-grade, it can be assumed that the course of recovery will be prolonged.
It is reported that higher rates of nonunion and complications delay the healing time of injuries at high-risk sites [10
]. Reviews in the literature also state that complications involving prolonged healing are rare in the group of low-risk fractures [10
]. Though no complications or specific reasons could be identified with certainty in every case of this retrospective study, prolonged healing times of more than 200
days were seen in all categories but the low-risk/low-grade group. (Figure
) This again underlines the different character of the low-risk/low-grade group.
These findings show that early diagnosis is of paramount importance and progression from a low-grade SFX to a high-grade SFX has to be avoided by early intervention. For high-risk injuries a more aggressive treatment with stringent restrictions is advisable, regardless of the severity. Immobilization should possibly be prolonged, despite the absence of symptoms, but to answer this question adequately, prospective studies first need to be performed. The group of low-risk and low-grade stress fractures did not show complications under the presented treatment-plan, but in order to avoid progression of the injury, an accelerated reintegration into training cannot generally be recommended. In a previous study, a proportion of patients with low-grade stress fractures were found to recover under ongoing stress, while in others the injury progressed to a high-grade fracture [8
]. Therefore the decision has to be made individually and in consultation with the athlete.
Follow-up examinations may be useful to monitor the healing process, in which case we propose MRI, as it does not involve exposure to radiation. In the case of an inconclusive initial MRI, or a negative follow-up MRI with persisting clinical symptoms, we propose bone scintigraphy, due to its higher sensitivity and high negative predictive value.
Certain limitations arising from the retrospective setting of this study should be mentioned. Treatment was not entirely standardized and clinical suspicion influenced some decisions regarding time of immobilization, but with all the patients being competitive athletes, the aim was nevertheless, to achieve the quickest possible return to sports. Furthermore, the cases of stress fractures presented form a heterogeneous group with regard to location, which may limit the quality of the conclusion. On the other hand, a general statement on the RTST of high- and low-risk SFX can only be drawn from a study assessing a large variety of different locations. Finally, the return-to-sports-time was determined clinically and not confirmed with imaging follow-up examinations, which would be desirable.
Still, we believe that these results make a useful contribution to current sports medicine practice. To our knowledge, no prospective study evaluating RTST of stress fractures has yet been published, though such a study would be of great interest.