Data collection and study population
This nationwide population-based observational study was conducted using the Swedish Fracture Register (SFR). The SFR is a national quality register compiling data on patient characteristics, injury mechanism, fracture classification and treatment. The treating physician is responsible for registration including classification of the fracture. The register was established in 2011 and since May 2015 the register has expanded to include also paediatric fractures (13). Since 2021 all orthopaedic departments in Sweden report to the SFR (100% coverage). Comparing the registered paediatric humeral fractures in the SFR with the National Patient Register, the completeness was >70% in half of the affiliated hospitals and varying between 10 and 70% in the remaining hospitals. Most of the latter only registered surgically treated fractures (internal audit, data not published).
All patients <16 years of age at the time of injury, with a proximal or diaphyseal humeral fracture registered between 1 January 2015 and 31 December 2019, were included. In total 8052 humeral fractures were registered between 2015 - 2019. Supracondylar humeral fractures (6056 fractures) were excluded. Patients with two fractures (127 patients) registered within the time of study (either patients with multiple fractures and/or refractures) remained in the study. Thus, the final study population consisted of 1996 registered fractures with 1696 proximal humeral fractures and 300 diaphyseal humeral fractures.
Variables
Variables of interest included: patient’s age, sex, injury date and injury mechanism. Furthermore, anatomical fracture classification and treatment method were assessed.
The mechanism of injury was divided into eight groups: simple falls, falls from height, unspecified falls, transportation accidents, stress/pathological/spontaneous fractures, non-accidental, other accidents and no mechanism registered (missing data). ”Simple fall” included all falls in the same level (trips, in snow/ice, ice-skating/skiing, nudges) and “fall from height” all falls from another level (from furniture, playground facilities, stairs, trees, buildings). “Unspecified falls” were injuries registered as falls without information about which level they occurred in. “Transportation accidents” were divided into several subgroups; pedestrian, bicycle/motorcycle (including other small vehicles), car (including truck), horse-riding and unspecified transportation accidents. “Non-accidental injuries” were all injuries due to fights, abuse and self-destructive acts. “Other accidents” were injuries due to mechanical and living forces.
The fractures were categorized according to the paediatric AO/OTA classification system (Fig. 1 and 2). The fractures were also divided into groups of adult patients with closed growth plates or paediatric patients with open growth plates. In proximal fractures, fractures categorized as Salter-Harris III, IV, and intraarticular fragments were all categorized as “intraarticular fractures”.
Treatment was divided into two main categories; non-surgical and surgical treatment. The surgical treatment category also included those patients which had a change of their treatment regime from initial non-surgical treatment to surgical within 2 weeks of the injury. The non-surgical group was further divided into a non-surgical or non-surgical with closed reduction group, and the surgical group to; a) closed reduction under general anaesthesia, b) k-wire and cerclage, c) intramedullary nailing (including both rigid and flexible nails), d) plate fixation, e) screw fixation, f) combined osteosynthesis and g) external fixation groups. The chosen method of treatment was then surveyed in four different age groups; infancy and toddlerhood (0-3 years), preschool (4-6 years), school-age (7-12 years) and adolescence (13-15 years).
Statistics
The statistical analyses were performed using Excel (Microsoft Excel for Mac version 16.31, Microsoft Corporation, Redmond, WA) and R version 3.6.3 (February 29, 2020). Descriptive statistics (counts, median with interquartile range and percentage) were used to describe the collected data. Logistic regression was performed to estimate the odds ratio (OR). Statistical significance was defined as p < 0.05.