Ultrasound supports clinical decision-making in determining the Sanders’ skeletal maturity score of the hand

The assessment of growth prognosis is crucial for the treatment of scoliosis and growth guidance measures such as temporary epiphysiodesis [1‐3]. For most physicians, the bone age determination atlas of Greulich and Pyle is still a standard method, but it is very cumbersome to implement [4]. Therefore, Sanders et al. developed a score that predicts the progression of the last growth period well and is easy to apply without the use of a detailed atlas book. However, both methods require an x-ray of the non-dominant hand [5]. Other methods like the evaluation of the growth plates or apophyses situated at the pelvis, the thumb, the proximal humerus, or the calcaneus have been described [6‐9]. Hung et al. introduced the Thumb Ossification Composite Index (TOCI) using only a radiograph of the thumb.

Although in principle this does not imply a high radiation exposure, recent high-quality studies have shown that in a growing organism even low radiation doses carry a higher risk of malignancy than in adulthood. Therefore, the ALARA principle (“As Low As Reasonably Achievable”) should always be followed [10‐12]. Especially in children, sonography has therefore proven to be an invaluable radiation-free diagnostic tool [13‐17]. These studies investigated growth plates or apophyses that were either difficult to access or poorly documented for predicting growth. Mentzel et al. and Utczas et al. created an ultrasound (U) device to assess skeletal maturity, although it is not widely available and does not utilize a standard U system [15, 17].

Therefore, we questioned whether U is suitable to identify the Sanders stages of the hand or the TOCI score of the thumb with special focus on the stages relevant for therapy. Sanders stages 3 and 4 (S3 and 4) are crucial for deciding if to start brace therapy for scoliosis, while stages 7 and 8 (S7 and S8) are significant for determining when to stop brace therapy or for deciding against epiphysiodesis in cases of limb deformities.

Growth prediction in early adolescence is of critical diagnostic importance in many musculoskeletal conditions that require monitoring until skeletal maturity and can be treated with surgical or conservative growth guidance [2, 18, 19]. Unfortunately, current methods still rely on radiation-driven diagnostics such as radiography of the hand or pelvis. Since the ALARA principle is always binding, we investigated whether the SMS and the TOCI can be performed with sufficient accuracy using U [11, 12]. In this study, we provide fundamental evidence that U is capable of detecting clinically relevant Sanders stages. While there have been previous attempts to determine bone age sonographically, we focused on comparability with the established SMS and TOCI classifications [13‐17, 20].

Regarding demographic data, the majority of our patients were between 12 and 17 years of age. This is consistent with the fact described by Sanders et al. that the significant phase of the growth spurt occurs during adolescence and that, as a result, adolescent scoliosis, leg length discrepancies, and leg axis deformities become clinically prominent [1, 21]. SMS 3, which contained the most individuals, as well as SMS 7/8, TOCI 4/5, and TOCI 8 were found to be the most common, as these stages also represent phases during growth when physicians either indicate or omit surgical or conservative treatment options when diagnosing scoliosis or limb deformities. In addition, other growth stages, such as SMS 5, represent a rapid transition between SMS 4 and 6 [21]. The low number of patients in SMS 1 and 2 also results from the fact that there is rarely a justifiable indication for radiographs of the hand before the last growth spurt, since we know without further diagnostics that otherwise healthy children will still grow significantly. For these ethical reasons, we performed radiographs of the hand only when this had the potential to lead to a therapeutic decision.

The percentages of concordance in our data compare well with previous studies, e.g., Utczas et al. [15, 17]. These authors used a quantitative ultrasound-based device measuring acoustic parameters of the wrist [15]. Although it should be remembered that the trial population studied was almost ten times larger than ours, they used the complex Greulich and Pyle method as well as a specialized, not widely available ultrasound device [17].

Comparing U and R, we found average to good probabilities of an accurate attribution to the 8 Sanders and first 8 TOCI stages. Looking at the stages separately, we found the highest discrepancy in SMS 5, 6, and 8 as well as TOCI 3, 6, and 7. The highest agreements were found in SMS 3 and 7 with 71.7% and 66.0%, respectively, as well as TOCI 8 with at least 77.7%. It is noteworthy that these are the phases with the largest patient population. This might also have resulted in a better correlation with the radiologic result due to a better statistically relevant cohort. We speculate that the greater routine with these categories enabled physicians to diagnose these categories more easily, leading to lower statistical agreement in groups with fewer patients. In addition, it is probably easier to identify wide-open or fully closed growth plates, which increases the hit probability for Sanders 3, 7, and 8 and TOCI 8. We found a very small number of patients at TOCI 9 stage and therefore could not perform a statistically accurate evaluation.

The differences in SMS and TOCI scoring are well known in the literature and are also consistent with our results [8]. Hung et al. have previously described that TOCI 5 levels are more consistent with SMS 3 and TOCI 8 are more consistent with SMS 8 when radiological analysis is performed [8]. However, according to their data, both scoring systems are reliable tools for determining skeletal maturity.

In clinical practice, it is very important to have a simple and usable tool to identify the stages that entail therapeutic consequences. Therefore, we have grouped SMS 3 and 4 (in SMS category 2) and TOCI 5 and 6 (in TOCI category 2), where a residual growth of 10% can be expected and the initiation of brace therapy in case of scoliosis or temporary epiphysiodesis in case of limb deformities might be appropriate [3, 22]. We also combined SMS 7 and 8 (in SMS category 2) and TOCI 7 and 8 (in TOCI category 3), in which no therapeutically relevant growth occurs and the above therapy methods are omitted in most cases. In particular, for SMS categories 2 and 4, we found very acceptable transferability between U and radiography, with P = 76.7% [71.8%; 81.0%] and P = 79.7% [75.0%; 83.9%], respectively, making the assessment of SMS by U a reliable, readily available, rapid, radiation-free, and cost-effective tool with low burden on children.

Our intra- and interrater reliability for the radiological SMS and TOCI were good compared with other studies validating the SMS, so the data could be used as the basis for comparison with U [8, 23]. The intrarater reliability of U was slightly lower for both SMS and TOCI, but was within the limits for clear to strong agreement (Table 2). However, the interrater reliability of U for SMS and TOCI was weaker. In particular, for SMS, the agreement using Fleiss’ Kappa was weak (k = 0.381). Looking at the different concordances when comparing R and U devided by raters, there was a wide range between P = 70.4% and P = 33.0% (see Table 1).

Differences in interrater agreement are well known in ultrasound diagnostics in different medical fields compared to other imaging modalities and often depend on the level of experience of the examiner [24‐27]. We have observed significant differences among the raters concerning concordance, but we have achieved very satisfying results in terms of intrarater reliability. This indicates that raters consistently scored at a high and reproducible level but interpreted the data systematically differently from their colleagues. Therefore, we hypothesize that providing more intensive training before conducting SMS or TOCI using US could lead to improved accuracy of the method.

Nevertheless, SMS can also be effectively determined by U by combining the clinically relevant stages SMS 3/4 and SMS 7/8. With somewhat higher intra- and interrater reliability, our data show advantage of radiographic determination of SMS compared with the use of U. However, because U is radiation-free, easy to perform, inexpensive, and rapid and involves very little psychological distress to the child, it may offer advantages in clinical practice.