INTRODUCTION

Hip fractures represent a pathology with a high impact on public health, being an important cause of mortality and morbidity in the elderly population¹. Approximately 300,000 hip fractures occur in the United States each year with a projected increase due to increased life expectancy. Regarding Colombia, it has been found that between 8,000 and 10,000 hip fractures occur annually, speaking only of those that occur in women, with osteoporosis being the main risk factor in this group, older than 60 years^2,3.

The anatomical location of the proximal femur is divided into head, neck and intertrochanteric regions; fractures in this area can be classified as intracapsular and extracapsular, depending on the joint^4,5. Intertrochanteric fractures represent an important 50% of proximal femur fractures, mostly associated with low-energy traumatic mechanisms in the elderly, considered as a product of metabolic pathologies that can affect bone quality⁶.

Throughout scientific development, different treatments have been proposed for the management of hip fractures, including conservative and surgical management, the incidence of postoperative complications is high and entails a complex rehabilitation period and mortality per year corresponding to the value already mentioned. However, without surgery, mortality is 50%, the results are less promising, so surgical treatment is the standard treatment for older adult patients^7,8,9. Regarding the surgical decision, it has several supports, among them, the patient's comorbidities, her anesthetic risk, and the type of fracture line (stable or unstable)¹.

The classifications allow establishing the best intraoperative procedure according to the patient's injury, so it is necessary to have an intra and interobserver concordance when using them as a therapeutic reference^10,11. It is for this reason that the objective was to evaluate the interobserver reproducibility of the Tronzo classification for intertrochanteric fractures to establish the suitability of its clinical application and the variability according to the level of experience.

METHODS

RESULTS

56 medical records were analyzed, of which 24 corresponded to the year 2019 and 32 to 2020, it was decided to leave the same number of elements to be analyzed, for a total of 48 medical records. The radiographs obtained from each of the HCs were classified according to the Tronzo classification, by 10 observers: 6 specialists, 1 second and third year resident, and 2 final year medical students who were completing elective rotation in the service, taking into account all observers. According to the results observed inTable 1, the highest concordance was obtained when evaluating the presence of Tronzo 1 (kappa = 0.51) and Tronzo 5 (kappa = 0.31) hip fractures, which indicates that there is a greater concordance in the radiological identification of hip fractures of lesser and greater degree of severity.

When observers were divided into groups of orthopedic and trauma specialists versus trainees, we found the following results: for specialists, the overall assessment of hip fractures reflects an agreement of 0.27 but this is higher when assessing only between the hip peaks of lesser severity presenting a kappa of 0.53 and those of greater severity with a result of 0.58 for this index; results that when contrasting with the training staff (orthopedic residents and final year students with orientation in orthopedics) a total value of 0.22 is found, but there is good agreement between the identification of low severity fractures with a kappa of 0 .59 for those categorized as Tronzo 1, however, not having found a good identification for the group in fractures with a higher degree of severity (Table 2).

Finally, when all the groups observed are considered to evaluate the concordance in the radiological identification of hip fractures by Tronzo classification, the best concordance is obtained for specialist doctors when determining Tronzo 1 fractures, with a kappa of 0.53 and the Tronzo 5 with an index of 0.58, orthopedic residents show a kappa of 0.32 for all types of fracture, and the degree of concordance is higher for the Tronzo 1 (kappa = 0.55) and Tronzo 3 (kappa = 0.49). Finally, for interns with special training in orthopedics, there is only a concordance of 0.10 for all types of fracture, however, when we evaluate according to the different categories, there is a good identification among the subgroup to identify less severe fractures having a kappa of 0.48 for impact hip fractures 1 (Table 3).

DISCUSSION

The classifications work as a source for determining clinical behaviors, influencing therapeutic decision-making, so their validity becomes important. University hospitals have personnel in training between undergraduate and postgraduate, this subgroup being in most cases in the mentioned scenario, the first contact with the patient by specialty, therefore, having concordance of concepts among all the personnel involved with the patient with the same pathology, it becomes relevant. Landis and Koch proposed in 1977 a scale for interpreting the kappa value where they considered acceptable those values greater than or equal to (0.40) and excellent those greater than 0.75 once the concordance had been applied¹³.There are different studies in orthopedics where the classifications have been subjected to the evaluation of their reproducibility, an example of this is the great interobserver variability demonstrated by La Villa et al. on the Neer classification for proximal humerus fractures with concordance of (0.18)¹⁴. The authors Neto, J. B. A et al. had a concordance of (0.58) for the Fraser classification and 0.46 for Blake & McBryde, in floating knee, which was interpreted as acceptable¹⁵.

The Tronzo classification was evaluated in a study published in the Revista Brasileira de Ortopedia, Oliveira, F.A.S. et al. where they assessed its reproducibility, this being moderate (0.44). A common factor of the mentioned studies was the inclusion of personnel in training, residents of the specialty, however in our searches in the Medline and NCBI databases we did not find any study that took into account the evaluation of the concordance index of the classification of Tronzo that included, in addition to specialists and residents, undergraduate students.

As it could be observed, the degree of concordance of the classifications in Orthopedics varies, despite its relevance, there is little evidence of the assessment of the concordance that exists between the different levels of medical specialization, residents, general practitioners and doctors in training. despite its importance in training for the proper referral of patients¹⁶.

In our study, we identified that from the Tronzo classification, type I fracture had the highest concordance in all subgroups (0.48 - 0.59), which corresponds to an acceptable variability of concepts among the observed observers. This type of fracture corresponds to stable fractures, however when there is suspicion of comminution of the posteromedial wall and/or suspicion of lesser trochanter involvement, it is when the concordance becomes lower (0.02 - 0.06) as the worst results; but if the total group is looked at separately, it has an adequate concordance for the group of residents (0.49), this may be due to the challenge involved in recognizing the commitments of the femur mentioned as criteria of instability of the intertrochanteric fracture such as Tronzo II and III . However, for those cases of totally unstable IV and V fractures, it is the person with the most training who represents an adequate variability (0.58). This is explained by the influence of experience in the field of specialists and staff in training by rank (residents over doctors in training). An unexpected result of this study was the poor concordance of specialists with Tronzo II and III fractures, since in general what is expected from greater training together with experience is greater reliability. One limitation of our study is the absence of an interobserver assessment that allows evaluating the operative options in terms of the clinical and/or surgical behavior to be performed for each subtype of the classification.

CONCLUSION

The Tronzo classification presented little concordance in fractures II to IV, and in the cases of fractures I and V the reproducibility is greater. The identification for the types of high complexity fractures was better in orthopedic specialists with more experience. However, this is not the same for staff in training by rank (residents over doctors in training) because the best agreement was for the first types I, IV. Therefore, the Tronzo classification does not adequately fit the criteria for an adequate classification for clinical practice in this hospital.