Appendix 2: Pilot Studies Using the SIGN Database

Appendix 2. Pilot studies using the SIGN database

The general information that can be extracted from the SIGN database has been reported elsewhere [2]. However, this report did not attempt to use the SIGN database for a study of the distribution of fracture patterns. It is not clear whether the SIGN database contains information of sufficient quality to allow such scientific studies of that type. Accordingly a pilot investigation was undertaken to discover whether a valid study on the classification of fracture patterns could be undertaken using the database.

One hundred eleven database entries were randomly selected for study. Four were excluded (4%). Two had been deleted from the database. One case did not have information about the size of the implant. One case was excluded because radiographs of a femur were associated with a tibia case. The remaining 107 cases each had at least a minimum set of information: year of surgery, gender, bone fractured and surgical approach, Gustilo classification, open or closed reduction and implant size.

For these cases, a single observer (JFMC) examined the radiographic images on the database to determine whether they were of sufficient quality to allow classification of the fracture using the OTA system [1]. Cases were excluded if the radiographic images were inadequate or if the images were taken more than 6 weeks from the time of injury. Thirty cases (28%) were excluded. Sixteen excluded cases had no x-ray in the database. In three cases the images were of such poor quality that they could not be used for fracture classification. The remaining 11 excluded cases had delayed presentation with evidence of previous surgery, nonunion or delayed union. One of these cases presented for ankle fusion. Seventy-seven cases (two humeri, 47 femurs, and 28 tibias) were fresh fractures with images of sufficient quality to classify the injury under the OTA Classification system.

The distribution of femur fractures classified by this system (Table 1) shows that there were 11 simple fractures (32-A), 25 fractures with minor comminution (32-B), 10 segmental or multiply-comminuted fractures (32-C) and one simple distal femur fracture (33-A). One pattern, 32-B2.2, accounts for 11 of 47 (23%) femur fractures in this sample. Overall there are 23 diaphyseal fracture patterns reported in this sample of 46. There is only limited evidence of clustering into specific fracture patterns in this small sample.

A similar pattern is seen in the distribution of 28 diaphyseal fractures of the tibia (Table 2). As with the femur there is a very flat distribution across a wide variety of fracture patterns. The 28 fractures are classified into 21 different groups.

To explore the potential contribution of the SIGN database to bone morphometry we studied radiographic images of tibial fractures from two different nations. Each postoperative image has a reference object (the IM nail) of known dimensions so that scaling of the image is possible. A random selection of 30 cases of male tibia fracture patients from Kenya and Cambodia was made. Three cases were excluded; two were recorded as tibias but have radiographs of fractured femurs and one case could not be found. Unfortunately, 36 of 57 cases had radiographs which did not show both the proximal and distal joint. Proportionately more cases were excluded from the Kenyan cohort. Radiographs of longer tibias might be expected to fail to show both joints more often. Thus there was both a high proportion of inadequate data and a suspicion that there might be selection bias. We conclude that the records in the SIGN database could not be used for bone length measurements. Bone width measurements could be made but are of limited clinical significance. Bone thickness and cortical width measurements could also be made; however a variable amount of reaming has been done on cases with an intramedullary nail in situ, so these measurements would be suspect. It is unfortunate that the data in the SIGN database cannot be used for bone morphometry because of the large amount of material. Exhortations to improve the quality of the data by requesting full radiographs of the injuries are not likely to change that situation.

References

1. Marsh JL, Slongo TF, Agel J, Broderick JS, Creevey W, DeCoster TA, Prokuski L, Sirkin MS, Ziran B, Henley B, Audigé L. Fracture and dislocation classification compendium - 2007: Orthopaedic Trauma Association classification, database and outcomes committee. J Orthop Trauma. 2007;21(10 Suppl):S1-133.
2. Shearer D, Cunningham B, Zirkle LG Jr. Population characteristics and clinical outcomes from the SIGN online surgical database. Tech Orthop. 2009;24:273-276.