Proximal humerus fractures (PHFs), AO/OTA (Ar beitsgemeinschaft für Osteosynthesefragen/Orthopaedic Trauma Association) type 11,1 are common, representing 4% to 5% of all fractures in adults.2 However, there is no consensus as to optimal management of these injuries, with some reports supporting and others rejecting the various fixation methods,3 and there are no evidence-based practice guidelines informing treatment decisions.4 Not surprisingly, orthopedic surgeons do not agree on ideal treatment for PHFs5,6 and differ by region in their rates of surgical management.2 In addition, analyses of national databases have found variation in choice of surgical treatment for PHFs between surgeons and between hospitals of different patient volumes.4 Few studies have assessed surgeon agreement on treatment decisions. Findings from these limited investigations indicate there is little agreement on treatment choices, but training may have some impact.5-7 In 3 studies,5-7 shoulder and trauma fellowship–trained surgeons differed in their management of PHFs both in terms of rates of operative treatment5,7 and specific operative management choices.5,6 No study has assessed surgeon agreement on radiographic outcomes.
We conducted a study to compare expert shoulder and trauma surgeons’ treatment decision-making and agreement on final radiographic outcomes of surgically treated PHFs. We hypothesized there would be poor agreement on treatment decisions and better agreement on radiographic outcomes, with a difference between shoulder and trauma fellowship–trained surgeons.
Materials and Methods
After receiving institutional review board approval for this study, we collected data on 100 consecutive PHFs (AO/OTA type 111) surgically treated at 2 affiliated level I trauma centers between January 2004 and July 2008. None of the cases in the series was managed by any of the surgeons participating in this study.
We created a PowerPoint (Microsoft, Redmond, Washington) survey that included radiographs (preoperative, immediate postoperative, final postoperative) and, if available, a computed tomography image. This survey was sent to 4 orthopedic surgeons: Drs. Gardner, Gerber, Lorich, and Walch. Two of these authors are fellowship-trained in shoulder surgery, the other 2 in orthopedic traumatology with specialization in treating PHFs. All are internationally renowned in PHF management. Using the survey images and a 4-point Likert scale ranging from disagree strongly to agree strongly, the examiners rated their agreement with treatment decisions (arthroplasty vs fixation). They also rated (very poor to very good) immediate postoperative reduction or arthroplasty placement, immediate postoperative fixation methods for fractures treated with open reduction and internal fixation (ORIF), and final radiographic outcomes.
Interobserver agreement was calculated using the intraclass correlation coefficient (ICC),8,9 with scores of <0.2 (poor), 0.21 to 0.4 (fair), 0.41 to 0.6 (moderate), 0.61 to 0.8 (good), and >0.8 (excellent) used to indicate agreement among observers. ICC scores were determined by treating the 4 examiners as independent entities. Subgroup analyses were also performed to determine ICC scores comparing the 2 shoulder surgeons, comparing the 2 trauma surgeons, and comparing the shoulder surgeons and trauma surgeons as 2 separate groups. ICC scores were used instead of κ coefficients to assess agreement because ICC scores treat ratings as continuous variables, allow for comparison of 2 or more raters, and allow for assessment of correlation among raters, whereas κ coefficients treat data as categorical variables and assume the ratings have no natural ordering. ICC scores were generated by SAS 9.1.3 software (SAS Institute, Cary, North Carolina).
Results
The 4 surgeons’ overall ICC scores for agreement with the rating of immediate reduction or arthroplasty placement and the rating of final radiographic outcome indicated moderate levels of agreement (Table 1). Regarding treatment decision-making and ratings of fixation, the surgeons demonstrated poor and fair levels of agreement, respectively.
The ICC scores comparing the shoulder and trauma surgeons revealed similar levels of agreement (Table 2): moderate levels of agreement for ratings of both immediate postoperative reduction or arthroplasty placement and final radiographic outcomes, but poor and fair levels of agreement regarding treatment decision-making and the rating of immediate postoperative fixation methods for fractures treated with ORIF, respectively.
Subgroup analysis revealed that the 2 shoulder surgeons had poor and fair levels of agreement for treatment decisions and rating of immediate postoperative fixation, respectively, though they moderately agreed on rating of immediate postoperative reduction or arthroplasty placement and rating of final radiographic outcome (Table 3). When the 2 trauma surgeons were compared with each other, ICC scores revealed higher levels of agreement overall (Table 4). In other words, the 2 trauma surgeons agreed with each other more than the 2 shoulder surgeons agreed with each other.
Discussion
This study had 3 major findings: (1) Surgeons do not agree on treatment decisions, including fixation methods, regarding PHFs; (2) regardless of their opinions on ideal treatment, they moderately agree on reductions and final radiographic outcomes; (3) expert trauma surgeons may agree more on treatment decisions than expert shoulder surgeons do. In other words, surgeons do not agree on the best treatment, but they radiographically recognize when a procedure has been performed technically well or poorly. These results support our hypothesis and the limited current literature.