1 Introduction

Pilon fractures account for 3–10 % of all tibial fractures and are serious injuries that often result in significant morbidity and complications.¹ These high energy injuries involve complex fracture patters associated with significant soft tissue damage and a high risk of complications.² Despite advances in surgical techniques, pilon fracture patients have long recovery periods, high risk of soft-tissue complications, and notable risk of deep tissue infections.³ A thorough understanding of the risk factors as well as the development of predictive models is needed to help improve postoperative outcomes in patients with pilon fractures. Being able to identify these factors would significantly improve patient care and surgical strategies.

Frailty is defined as the decrease in physiological reserve which has emerged as an important consideration in surgical outcomes. It has been linked to an increase in postoperative complications, extended hospital stays and non-home discharge (NHD) across different types of surgical specialties.⁴ Specifically, through the lens of orthopedics, frailty has been linked to a significantly higher risk of fractures in adults.⁵ It has been well documented in the literature that quantifying frailty via validated models such as 5-factor modified frailty index (mFI-5) and the Risk Analysis Index-revised (RAI-rev) has been shown to be a predictor for surgical outcomes.^6,7 The mFI-5 is a widely used tool for assessing frailty in various surgical populations. The frailty index incorporates five variables including diabetes, congestive heart failure, functional status, chronic obstructive pulmonary disease (COPD), and hypertension. The mFI-5 has been validated in many different surgical contexts with a strong predictive ability for length of stay (LOS), NHD, and postoperative complications.⁸ Introduced by Hall and colleagues,⁹ RAI-rev is another frailty tool developed to assess frailty with a focus on preoperative risk prediction. The tool incorporates 11 variables including cognitive decline, residence other than independent living, functional status, unintentional weight loss, poor appetite, renal failure, congestive heart failure, shortness of breath, cancer, age, and sex.^9,10 Studies have found that RAI-rev has superior predictive ability for outcomes such as 30-day postoperative morbidity and mortality.^11–15 Thus, it has been used increasingly to help improve both preoperative management and optimization of patient care.

Despite the advances, there remains a paucity in the literature that investigates the link between frailty and pilon fractures. The complexity of these fractures, often caused by high-energy trauma, usually involves significant soft tissue damage making them difficult to manage while also exposing the role frailty can play in these scenarios. With the importance of preoperative planning and the high complication rates surrounding this type of injury, our study aims to investigate the role frailty, established by RAI-rev and mFI-5, has in predicting outcomes following surgical fixation of pilon fractures. The ability to reliably quantify the risk of postoperative complications would aid in preoperative planning and be of great benefit to surgeons, and most importantly, patients.

2 Methods

3 Results

The total patient cohort consisted of 2935 individuals who met the inclusion criteria for this study (Table 1) which was patients ages ≥18 years diagnosed with pilon fracture treated with external or internal fixation or arthrodesis. The median age of the cohort was 46 years (range 33–59), with the majority being White (64.1 %) and males comprising 53.2 % of the population. Among the cohort, 27.8 % were current smokers and 26.5 % had hypertension requiring medication. When examining frailty cohorts, 43.2 % (1296/2935) of patients were in the frail cohort of RAI-rev and 11.2 % (330/2935) in the severely frail cohort. For the mFI-5, 30.1 % (883/2935) were frail, while 1.5 % (43/2935) were categorized as severely frail.

A multivariate analysis was conducted to assess the predictive values of the mFI-5 and RAI-rev for poor postoperative outcomes (Table 3). The analysis was performed while controlling for age, sex, ethnicity/race, BMI, ambulation status, nonelective-surgical status, and total operation time for each cohort. The mFI-5 was a significant predictor for major complications, readmission, and CDIV especially in the severely frail cohort compared to the non-frail reference group with OR 4.30 (95 %CI 1.94–9.50; p < 0.001), 5.18 (95 %CI 2.16–12.43; p < 0.001), and 9.82 (95 %CI 2.47–39.07; p = 0.001), respectively. The mFI-5 was also a significant predictor for eLOS and NHD for both frail and severely frail cohorts compared to the non-frail reference group: eLOS frail: 1.54 (95 %CI 1.22–1.89; p < 0.001), severely frail: 5.80 (95 %CI 2.74–12.29; p < 0.001), NHD frail: 2.49 (95 %CI 1.86–3.33; p < 0.001), and severely frail: 10.23 (95 %CI 4.61–22.70). The RAI-rev was not a significant predictor for any outcome for the frail or severely frail cohort.

To determine the discriminatory accuracy of mFI-5 and RAI-rev, AUC analysis (Table 4 & Fig. 1) was performed for eLOS, major complications, NHD, readmission, and CDIV. AUC analysis revealed that RAI-rev significantly outperformed mFI-5 for eLOS (RAI-rev 0.658 vs mFI-5 0.621; p = 0.0012), major complications (RAI-rev 0.671 vs mFI-5 0.623; p = 0.0254), NHD (RAI-rev 0.798 vs mFI-5 0.733; p < 0.001), and readmission (RAI 0.656 vs mFI-5 0.597; p = 0.0220).

Close

Fig. 1 ROC curves comparing the discriminative performance of the RAI-rev (red) and the mFI-5 (blue) for predicting 30-day outcomes after operative treatment of pilon fractures. Outcomes assessed are eLOS, NHD, major complications, 30-day readmission, and CDIV complications. Curves plot sensitivity over 1-specificity. Diagonal dashed line represents no discrimination (AUC = 0.50). Across outcomes, RAI-rev demonstrates equal or greater discrimination than mFI-5. Exact AUCs are reported in Table 4.

Export to PPT

4 Discussion

This longitudinal study of 2935 patients extracted from a multicenter database was used to assess the predictive ability and discriminative accuracy of RAI-rev compared to mFI-5 for postoperative patient morbidity in surgical pilon fractures. Only mFI-5 was found to be a significant positive predictor, indicating that higher frailty scores were associated with increased odds for major complications, eLOS, NHD, readmission, and CDIV complications. However, RAI-rev demonstrated significantly increased discriminative accuracy in predicting eLOS, major complications, readmissions, and most markedly NHD compared to mFI-5, evidenced by higher AUC values in ROC analysis indicating its superior ability to distinguish between patients with and without these outcomes.

Evaluating short-term outcomes, such as non-home discharge, provides valuable clinical insights in the acute trauma setting. In the context of pilon fractures, which often occur due to high-energy mechanisms, accurate short-term risk stratification is crucial for early discharge planning, resource allocation, and multidisciplinary coordination. The improved AUC for RAI-rev in predicting NHD underscores its practical utility in identifying acute physiologic vulnerability, particularly when applied preoperatively to identify patients at risk for early complications and disposition challenges.

The findings from this study underscore the importance of integrating frailty assessments into preoperative risk stratification conversations for complex orthopedic injuries. This is the first study to apply frailty models to patients with pilon fractures and was done using a large, multi-institutional cohort.

The mFI-5 is an abbreviated version of the original mFI-11, developed to function as a comorbidity index that can often independently predict adverse postoperative outcomes.²⁰ The mFI-5 is an older frailty model compared to RAI-rev and in regards to orthopedic surgery, it has been validated as a significant predictor of postoperative morbidity and mortality in elective total joint, shoulder arthroscopy and spine surgeries based on ACS-NSQIP data.^20–24 In contrast to the mFI-5, the RAI-rev was designed to be mathematically and conceptually superior to mFI-5 and mFI-11, accounting for and correctly scoring advanced patient age and impairment in activities of daily living.^9,10 While there has not been any published literature comparing mFI-5 to RAI-rev in lower extremity orthopedic injuries, RAI-rev has had superior discriminative accuracy compared to mFI-5 based on ROC analysis in neurosurgery and orthopedic spine surgery patients.^11,12 Based on a similarly structured retrospective database analysis, the RAI-rev was superior in predicting CDIV, NHD, and reoperation rate in neurosurgical patients.¹¹ Similar to this current study, Conlon et al. showed that the RAI-rev is more accurate than mFI-5 for eLOS, readmission, and NHD in patients undergoing surgical fixation for traumatic spine injuries.¹² While the literature supports the utility of RAI-rev for elective procedures, this study adds to its superior accuracy in traumatic, non-elective procedures.^11,12

While frailty has been closely associated with aging,²³ the cohort in this current study had a median age over a decade younger than similar studies investigating RAI-rev.^11,12 Studies that have investigated orthopedic trauma and its relationship to frailty include Rege et al., which by using ACS-NSQIP found that the mFI-5 was a significant predictor of mortality and CDIV complications in their young cohort (<60 years old) who underwent surgical fixation for pelvis, acetabular and lower extremity trauma.²⁴ Similarly, out of a single level 1 trauma center, Phen et al. showed that mFI-5 was a significant predictor of any postoperative complication and mortality in patients less than 65 who underwent surgical fixation for lower extremity trauma.²⁵ Similar to Rege et al. and Phen et al., in this study, mFI-5 was a significant predictor of eLOS, and NHD for all frail cohorts, and of major complications, readmissions and CDIV complications for severely frail patients with pilon fractures.^24,25 However, among the studies who investigated frailty in young trauma patients using mFI-5, none compared it to any other frailty model, nor was significance for discriminative accuracy using AUC analysis achieved.^24,25 In contrast, this current study compared mFI-5 to the RAI-rev for predictive ability and discriminative accuracy. We displayed that while mFI-5 was a significant positive predictor for postoperative complications, RAI-rev had superior discriminative accuracy in a young, non-elective cohort. It is worth noting that while RAI-rev had superior accuracy for postoperative morbidities, the overall predictive value was low with only NHD demonstrating acceptable discriminative ability, consistent with Conlon et al.¹²

Other risk factors that have been linked to higher postoperative complication rates in surgical pilon fracture patients include open fractures secondary to high energy trauma,^26,27 preoperative comorbidities such as diabetes and hypertension,^27–29 active smoking,³ and psychiatric conditions.^30,31 When comparing these risk factors to the cohort in this study, 27.8 % of our cohort includes current smokers, and 26.5 % have diagnosed hypertension requiring medication. Interestingly, psychiatric conditions, and diabetes have all been linked to frailty,^32–34 with diabetes and hypertension being accounted for by mFI-5.

It is also important to address the clinical utility and applicability of RAI. It has already been effectively integrated into electronic health record systems such as Epic and Cerner, making the ability to collect a preoperative frailty assessment accessible to a wide range of clinicians across a variety of different settings.¹⁹ Additionally, while not as concise as the mFI-5, the RAI is practical and can be calculated in as little as 30 s without disrupting workflow, which is especially crucial for large, high volume healthcare systems.¹⁹

There are several limitations to this study. Regarding discriminative accuracy, while RAI-rev was superior, the overall AUC predictive values were low with c-statistic falling below 0.80 which has been associated with predictive capacity. C-statistics of 0.70 is the lower limit of what is considered an informative model, and this was only accomplished for NHD, which for RAI-rev was just shy of 0.80 at 0.798. Another limitation is inherent to using a database and involves the potential presence of data inconsistencies and inability to clarify certain details such as differentiating between severity or mechanism of injury. Due to the small cohort size, there was an unequal distribution between mFI-5 and RAI-rev severely frail groups and overall between severely frail and frail. This may have artificially inflated the mFI-5 frail cohort, potentially indicating that frailty has less of a role than perceived in this type of injury and influence the risk for postoperative complications with its relationship with frailty. Additionally, postoperative complications were limited to within 30 days of the operation. Further research should incorporate a database such as the Trauma Quality Improvement Program (TQIP) that allows for differentiation between injury severity. Such a study may hold potential in validating either mFI-5, or in the development of a novel model for predicting postoperative complications in pilon fracture patients.

5 Conclusion

This study is the first to directly investigate the relationship between frailty and postoperative outcomes in pilon fracture patients. Our study demonstrated that while mFI-5 was a significant predictor of most complications, RAI-rev had superior and more acceptable accuracy compared to mFI-5 for predicting NHD in pilon fracture patients. However, due to the inability to distinguish if injuries were due to high or low energy mechanisms, further research using a trauma specific database is required to establish the role of frailty in surgical fixation of traumatic injuries. A simple assessment for comorbidities such as those measured in the mFI-5 could be convenient in these patients, but ultimately, RAI-rev demonstrated a stronger ability in assessing frailty for predicting postoperative complications. This study contributes to the existing literature investigating the role of frailty in younger patients. In the context of pilon fractures, a cohort underrepresented in frailty research, this study builds on existing evidence by demonstrating the complementary strengths of frailty models in a high-risk, complex population.

Consent statement

-This study was performed under a Health Insurance Portability and Accountability Act-compliant Participant Use Data File which is considered exempt from institutional review board approval. Patient/guardian consent not required.

Ethics statement

-This study was performed under a Health Insurance Portability and Accountability Act-compliant Participant Use Data File which is considered exempt from institutional review board approval. Patient/guardian consent not required.

CrediT author statement

Gabriel DeOliveira: Conceptualization, Writing, visualization. Amber Park: Software, formal analysis, writing. Arsalaan Sayyed: Writing. Aruni Areti: Writing. Nithin Gupta: Conceptualization, formal analysis, supervision. Taylor Manes: Conceptualization, supervision. Morgan Turnow: Supervision. Benjamin Taylor: Supervision. Jack W. Weick: supervision.

Funding

There was no funding available to or provided for this project.