An analysis of Medicare databases showed marked regional variation in rates of operative treatment of PHFs.2 Similarly, a Nationwide Inpatient Sample analysis revealed nationwide variation in operative management of PHFs.4 Both findings are consistent with our results of poor agreement about treatment decisions and ratings of postoperative fixation of PHFs. In 2010, Petit and colleagues6 reported that surgeons do not agree on PHF management. In 2011, Foroohar and colleagues10 similarly reported low interobserver agreement for treatment recommendations made by 4 upper extremity orthopedic specialists, 4 general orthopedic surgeons, 4 senior residents, and 4 junior residents, for a series of 16 PHFs—also consistent with our findings.
The lack of agreement about PHF treatment may reflect a difference in training, particularly in light of the recent expansion of shoulder and elbow fellowships.2 Three separate studies performed at 2 affiliated level I trauma centers demonstrated significant differences in treatment decision-making between shoulder and trauma fellowship–trained surgeons.5-7 Our results are consistent with the hypothesis that training affects treatment decision-making, as we found poor agreement between shoulder and trauma fellowship–trained surgeons regarding treatment decision for PHFs. Subanalyses revealed that expert trauma surgeons agreed with each other on treatment decisions more than expert shoulder surgeons agreed with each other, further suggesting that training may affect how surgeons manage PHFs. Differences in fellowship training even within the same specialty may account for the observed lesser levels of agreement between the shoulder surgeons, even among experts in the field.
The evidence for optimal treatment historically has been poor,4,6 with few high-quality prospective, randomized controlled studies on the topic up until the past few years. The most recent Cochrane Review on optimal PHF treatment concluded that there is insufficient evidence to make an evidence-based recommendation and that the long-term benefit of surgery is unclear.11 However, at least 5 controlled trials on the topic have been published within the past 5 years.12-16 The evidence is striking and generally supports nonoperative treatment for most PHFs, including some displaced fractures—contrary to general orthopedic practice in many parts of the United States,2 which hitherto had been based mainly on individual surgeon experience and the limited literature. Without strong evidence to support one treatment option over another, surgeons are left with no objective, scientific way of coming to agreement.
Related to the poor status quo of evidence for PHF treatments is new technology (eg, locking plates, reverse total shoulder arthroplasty) that has expanded surgical indications.2,17 Although such developments have the potential to improve surgical treatments, they may also exacerbate the disagreement between surgeons regarding optimal operative treatment of PHFs. This potential consequence of new technology may be reflected in our finding of disagreement among surgeons on immediate postoperative fixation methods. Precisely because they are new, such technological innovations have limited evidence supporting their use. This leaves surgeons with little to nothing to inform their decisions to use these devices, other than familiarity with and impressions of the new technology.
Our study had several limitations. First is the small sample size, of surgeons who are leaders in the field. Our sample therefore may not be generalizable to the general population of shoulder and trauma surgeons. Second, we did not calculate intraobserver variability. Third, inherent to studies of interobserver agreement is the uncertainty of their clinical relevance. In the clinical setting, a surgeon has much more information at hand (eg, patient history, physical examination findings, colleague consultations), thus raising the possibility of underestimations of interobserver agreements.18 Fourth, our comparison of surgeons’ ratings of outcomes was purely radiographic, which may or may not represent or be indicative of clinical outcomes (eg, pain relief, function, range of motion, patient satisfaction). The conclusions we may draw are accordingly limited, as we did not directly evaluate clinical outcome parameters.
Our study had several strengths as well. First, to our knowledge this is the first study to assess interobserver variability in surgeons’ ratings of radiographic outcomes. Its findings may provide further insight into the reasons for poor agreement among orthopedic surgeons on both classification and treatment of PHFs. Second, our surveying of internationally renowned expert surgeons from 4 different institutions may have helped reduce single-institution bias, and it presents the highest level of expertise in the treatment of PHFs.
Although the surgeons in our study moderately agreed on final radiographic outcomes of PHFs, such levels of agreement may still be clinically unacceptable.19 The overall disagreement on treatment decisions highlights the need for better evidence for optimal treatment of PHFs in order to improve consensus, particularly with anticipated increases in age and comorbidities in the population in coming years.4 Subgroup analysis suggested trauma fellowships may contribute to better treatment agreement, though this idea requires further study, perhaps by surveying shoulder and trauma fellowship directors and their curricula for variability in teaching treatment decision-making. The surgeons in our study agreed more on what they consider acceptable final radiographic outcomes, which is encouraging. However, treatment consensus is the primary goal. The recent publication of prospective, randomized studies is helping with this issue, but more studies are needed. It is encouraging that several are planned or under way.20-22