Estimation of return-to-sports-time for athletes with stress fracture – an approach combining risk level of fracture site with severity based on imaging

In this retrospective study 52 competitive athletes (male, n = 22; female, n = 30; mean age, 22.8 years (y), quartile(Q) (Q1, 16.0 y; Q2, 18.0 y; Q3, 24.8 y) with stress fractures were included (track, n = 18; long distance running, n = 16; handball, n = 13; soccer, n = 1; swimming n = 1; triathlon, n = 1; canoeing, n = 1; basketball, n = 1). All patients were either at a residential sports college or attending an Olympic training center or belonged to a professional sports team.

Treatment of all included patients took place at the Department of Sports Medicine, and was performed by the same team of doctors and physical therapists. Athletes presenting with symptoms suggestive of a stress injury were advised to immediately rest or minimize exercise of the affected site. Radiographs, MRI, and BS were used for initial diagnostic imaging. Only in cases of suspected aggravation or a delayed healing process, were follow-up examinations performed.

A confirmed stress injury was initially treated with an orthotic or a cast while the patient pursued a non-weight-bearing regime. The length of immobilization was dependent on the level of subjective and objective symptoms such as pain, swelling, restricted range of motion, and duration of symptoms prior to consultation. Of all 52 stress fractures, only two were treated surgically, one in the talar bone, the another in the navicular bone. Concomitantly, manual lymphatic drainage, two-cell baths and microcurrent therapy were used to reduce edema of the bone and the surrounding soft tissue. The main objective in the early stages of treatment was to prevent any pain. On that condition, alternative sports were allowed in order to maintain the level of fitness (e.g. swimming, biking, or monitored circle training in the gym) as well as proprioceptive training. In the later stages of treatment, sport-specific exercises were added to the training program. Athletes specialized in acceleration were helped to exercise explosive movements by physiotherapists, beginning with minimal weight bearing for very short periods. Athletes relying on quick and secure sideways movements practiced the likely motion processes of the respective discipline under a weight-reducing suspension. The weight and duration of the exercises were individually coordinated by the supervising doctors and physiotherapists. After progress in individual training, athletes were gradually reintegrated into their regular training program, starting with short sessions, which were prolonged according to the symptoms developing. Throughout the process, kinesiatrics, physical therapy, and occupational therapy were available to assist the athlete’s recovery.

Magnetic resonance imaging (MRI) examinations were performed at 1.5 Tesla MRI (Intera 1.5 T MRI, Philips HealthCare, Best, The Netherlands). Depending on the location, axial, coronal and/or sagittal T1-weighted spin echo sequence (TR/TE: 400–600 ms/15–30 ms) as well as T2-weighted fast spin-echo sequence (TR/TE: 3,000 ms/44 ms, echo train 8) were acquired. In addition, a fat suppressed T2-weighted fast spin-echo sequence (TR/TE: 2,000–4,000 ms/60–80 ms, echo train 8) using short tau inversion recovery technique (STIR) or frequency selective chemical presaturation pulse was conducted. MRI examinations were performed with a field of view (FOV) of 160 × 160 mm – 240 × 240 mm, a matrix of 256 × 192 or 256 × 256, a slice thickness of between 3 mm and 5 mm, an interslice gap of between 0.4 mm and 3 mm, and with a number of excitations (NEX) of 1–2.

Three phases of bone scintigraphy images were obtained using a double head gamma-camera (e.cam, Siemens Healthcare, Erlangen, Germany) equipped with low-energy, high-resolution collimators. After the intravenous bolus-injection of 350–630 MBq (9.5-17.0 mCi) Technetium-99 m 3,3- diphosphono-1,2-propane dicarboxylic acid (Tc-99 m DPD) (Teceos; CIS bio international, GIF-sur-Yvette, France), planar data of the region of injury were recorded in a 64 × 64 matrix at a 1 second frame rate for the first minute. Static blood-pool images were obtained over the following 4 minutes. For the mineralization phase (3 h after injection), planar images were acquired over 5 minutes in the same view. Anterior and posterior scans were performed in all cases, a mediolateral for fibula and tibia and lateral images for feet were also acquired. In addition, anterior and posterior whole-body scans were performed at a table speed of 10 cm/min using a 256 × 1024 matrix.

For the purpose of this study, the following information was obtained on each patient from the records of the Department of Sports Medicine and the Department of Radiology and Nuclear Medicine: age, sex, type of sport, injury localization, beginning of symptoms, and the date of diagnostic imaging as bone scintigraphy (BS) and magnetic resonance imaging (MRI).

A reference standard was created by presenting each case to an adjudication panel consisting of experts in the fields of sports medicine, nuclear medicine, and radiology. All available data (clinical records and follow-up data, radionuclide imaging, MRI, and radiographs) were taken into consideration for the classification of a high- or low-risk injury and the determination of the exact time of return-to-sports. (Table 1) Return-to-sports was defined as the point in time, when the athlete was able to return to sports without restrictions and without any clinical or subjective symptoms.

Every patient underwent at least one examination with BS or MRI. BS images were rated in a separate blinded read setting by three independent specialists in nuclear medicine according to a simplified grading system for stress fractures derived from the work of Chisin et al. [8]. (Table 2). Similarly, MRI examinations were classified by three radiologists into low- and high-grade fractures following a simplified grading system proposed by Arendt et al. [5] (Table 2, Figure 1). Where a disagreement arose between the readers, a consensus was reached.

From every examination a dataset was created, containing the risk classification, the severity classification, and the return-to-sports-time measured from the date of the respective examination to the date of painfree return to full training. This included a total of 52 primary and 31 follow-up examinations at an interval of at least 6 weeks. Main outcome measurements were the estimation of return-to-sports-time depending on risk-classification (stress fracture at a low- or high risk site) and image-based grading (low- or high grade). Due to the small sample sizes a non-parametric distribution of the data was assumed and the healing times of all groups were compared using Wilcoxon’s rank sum test and Kuskal Wallis test. The reliability of the image-based grading was examined by calculating Fleiss' kappa.

R software, version 2.11.1 (The R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical calculations.

All tests were two-sided and a P value lower than 0.05 was referred to as significant.