More information about text formats
We read with great interest “Plate of ribs: single institution’s matched comparison of patients managed operatively and non-operatively for rib fractures” by Griffard et al (1). As chest injuries are one of the most prevalent injuries encountered in trauma care, the importance of evidence for or against treatments cannot be overlooked. The authors review their institution’s recent experience with surgical stabilization of rib fractures (SSRF) and draw comparisons to other patients with chest injuries cared for by their group. We found several methodological concerns (including propensity matching, injury phenotype heterogeneity, and selection bias) that arise from their design that may significantly affect their analysis and conclusions.
To start, the authors reference three prior studies that use patient-matching to study operative vs. non-operative treatments of fractures. The authors indicate that they made a similar matched comparison; however, there was significant heterogeneity in the methods of the referenced studies. The first study used a 1:4 match of age, GCS, other surgeries, mechanical ventilation, pressors and transfusion requirement, but introduced selection bias by excluding tens of thousands of patients cared for at hospitals that did not use SSRF (2). The next study used a 1:1 match, excluding significant traumatic brain injury (TBI), spinal, and pelvic injuries or other injuries AIS =5, to match age, sex and thoracic AIS (3). This resulted i...
To start, the authors reference three prior studies that use patient-matching to study operative vs. non-operative treatments of fractures. The authors indicate that they made a similar matched comparison; however, there was significant heterogeneity in the methods of the referenced studies. The first study used a 1:4 match of age, GCS, other surgeries, mechanical ventilation, pressors and transfusion requirement, but introduced selection bias by excluding tens of thousands of patients cared for at hospitals that did not use SSRF (2). The next study used a 1:1 match, excluding significant traumatic brain injury (TBI), spinal, and pelvic injuries or other injuries AIS =5, to match age, sex and thoracic AIS (3). This resulted in a very small sample size of only 10 patients matched to the SSRF group. The final study used a very robust 1:4 propensity score match for 19 characteristics, some of which had 13 variables, and many of which were continuous variables (4). Unsurprisingly, their match was so stringent that at least 5 characteristics in the “matched” group were still significantly different to a p < 0.01 level. By contrast in the authors study, only age and number of rib fractures (in groupings) were matched, albeit at a 1:4 ratio. Matching helps to create a pseudo-randomized population, helping to control for confounding pre-intervention variables that would affect down-stream outcomes. Presumably this indicates that the authors hoped the quantity of additional patients “matched” would outweigh potential confounding information. As propensity score matching is not described in the methods, a more rigorous match may give markedly different results.
Second, after matching patients for the above criteria, the operative group was found to have significantly more flail segments, displaced rib fractures, and fewer bilateral rib fractures. Essentially, the two groups have different injury phenotypes. As many prior analyses have confirmed, segmental fractures, degree of displacement, and bilaterality play a significant role in respiratory physiology, analgesia needs, and translate into worse outcomes. A standardized definition of rib displacement has only recently been clarified in updated taxonomy (5), which suggests that the radiographic definition for displacement employed over this study timeline did not standardize rib fracture cortical overlap, and is therefore heterogenous.
Third, the authors report that all patients were screened for SSRF according to current guidelines. This suggests that the 4150 patients who were treated non-operatively (including the 148 matched patients) were deemed not operative candidates. It is hard to draw meaningful conclusions from comparing patients presumably with contraindications to SSRF (no fracture instability, no pulmonary derangements, other higher priority injuries, significant TBI, infection, hemodynamic instability, etc.) to those undergoing operative management. Could the groups have been matched by other chest injury specific means, such as Rib Score? Could the non-operative group be reviewed to determine if they met any criteria for fixation?
When taken in this context, the author’s findings of “no differences” and conclusions that “SSRF may not be as beneficial as current literature suggests” is potentially misleading. At best, the authors present a case-series of SSRF use in severe chest wall injuries with outcomes congruent with prior literature. At worst, the methodological concerns and selection bias creates an apples-to-oranges comparison that does not support the conclusions drawn. We appreciate the hard work and thoughtful process evaluation by the authors and look forward to further experience with SSRF to help guide and improve patient care.
Christopher Janowak, MD, FACS
Michael Goodman, MD, FACS
Division of Trauma
Section of General Surgery, Department of Surgery
University of Cincinnati, Cincinnati, OH, USA
1. Griffard J, Daley B, Campbell M, et al. Plate of ribs : single institution ’ s matched comparison of patients managed operatively and non- operatively for rib fractures. Trauma Surg Acute Care Open. 2020;5:1-5. doi:10.1136/tsaco-2020-000519
2. Wada T, Yasunaga H, Inokuchi R, et al. Effectiveness of surgical rib fixation on prolonged mechanical ventilation in patients with traumatic rib fractures: A propensity score-matched analysis. J Crit Care. 2015;30(6):1227-1231. doi:10.1016/j.jcrc.2015.07.027
3. Uchida K, Nishimura T, Takesada H, et al. Evaluation of efficacy and indications of surgical fixation for multiple rib fractures: a propensity-score matched analysis. Eur J Trauma Emerg Surg. 2017;43(4):541-547. doi:10.1007/s00068-016-0687-0
4. Shibahashi K, Sugiyama K, Okura Y, Hamabe Y. Effect of surgical rib fixation for rib fracture on mortality. J Trauma Acute Care Surg. 2019;87(3):599-605. doi:10.1097/ta.0000000000002358
5. Edwards JG, Clarke P, Pieracci FM, et al. Taxonomy of multiple rib fractures: Results of the chest wall injury society international consensus survey. J Trauma Acute Care Surg. 2020;88(2):E40-E45. doi:10.1097/TA.0000000000002282