Developing a COMputerised Bone Age Tool

Overview

The Greulich and Pyle technique (G\&P) is that most commonly used to assess bone age in children, particularly in the context of suspected skeletal dysplasia. However it has been shown not to be applicable to children of Asian and African ethnicity and may not be applicable to United Kingdom (UK) Caucasian children, who are now relatively more mature and larger than children of the mid 1930's (when G\&P was developed). It is generally agreed that updated standards are required for rapid and reliable bone age assessment. The development of such standards requires irradiation of normal children. To do this ethically, radiation exposure must be kept as low as possible. Two recent studies suggest that bone age assessment can be reliably achieved from dual-energy x-ray absorptiometry (DXA) scans. As far as the investigators are aware, the Tanner \& Whitehouse (TW3) method, which is preferred by some (e.g. endocrinologists and nutritionists), has not been assessed from dual-energy x-ray absorptiometry (DXA) scans - although not the main focus of this study, because of its popularity amongst certain specialties, the investigators shall be assessing the TW3 method also. Modern techniques should not only be available as textbooks, but should also be available in digital (on-line) format and where possible integrated with hospital PACS systems. The investigators believe that their team has the required experience and expertise to successfully carry out such a project. Before conducting the required large cross-sectional study, the investigators must first confirm that DXA can in fact replace radiographs for bone age assessment in children; which is the objective of this current study.

This is a small prospective feasibility study to be performed over a 12-month period. Children booked for a left hand radiograph for bone age assessment (either in isolation or as part of a dysplasia skeletal survey) will be recruited. This will require a purposive sample of 2 males and 2 females from each of the following 5 age groups (\<5; 5 to 7; 8 to 10; 11 to 13; 14 to 16 years). There will therefore be a total of 20 children (10 males). Because the aim of the study is to compare results of DXA with those of radiographs in corresponding children, ethnicity of individual children will not be an issue; in other words the investigators are assessing the reliability of DXA not the reliability of G\&P or TW3. Relevant clinicians will identify patients from appropriate clinics at Sheffield Children's Hospital. Only patients having a left hand radiograph for bone age assessment on the basis of clinical need will be approached. Age appropriate information sheets will be given to them, and the study briefly discussed. When they arrive at the Radiology Department for their left hand radiograph, an appropriately certified radiographer will seek informed consent and provided the left hand radiograph is actually performed, they will then also have DXA of the left hand and those over 6 years old will be asked to complete a brief questionnaire. Researcher 1 will anonymise hand radiographs and DXA scans, such that corresponding patient images are not identifiable. Researcher 1 will also document patient age and sex, ethnicity and body weight and height (no other clinical information is required for the purposes of this feasibility study). Researchers 2, 3 and 4 will independently assess all images using both G\&P and TW3. All observers will interpret a random selection of half the images (again independently) on a second occasion at least 2 months after the first; on both occasions the Images will be interpreted in random and varied order. Images will be read in two batches of 10 radiographs and 10 (non-corresponding) DXA scans for the first read and 5 radiographs and 5 (non-corresponding) DXA scans for the second read. Statistical analysis is largely irrelevant for this small feasibility study. The investigators' main aim is to assess whether or not G\&P and TW3 can be reliably assessed from hand DXA images. The concordance between 1) independent readings of the three observers for radiographs and for DXA (i.e. comparing modalities) and 2) individual observers for both occasions (i.e. comparing observer reliability for each modality) will be evaluated using interclass and intraclass correlation coefficients. The investigators will calculate the mean and standard deviation of the differences between DXA and radiographs and between observers. Statistical analyses will be performed using the Statistical Package for the Social Sciences, version 21 for Mac (SPSS, Chicago, IL).