The Impact of Specific Fracture Characteristics of Low-energy Fractures of the Pelvis in the Elderly on Mortality


 Background:Pelvic fractures in the elderly are associated with relevant morbidity and mortality. Both might be determined by fracture morphology and/or patient characteristics. The aim of this project is to investigate the prognostic value of specific fracture characteristics with respect to overall survival and to compare it with an established classification system. Methods:Retrospective analysis of patients ≥60 years, treated conservatively for a CT-scan verified, low-energy pelvic ring fracture between August 2006 and December 2018. Survival data was available from patients’ charts and cantonal or national registries. The prognostic value of fracture characteristic describing the anterior and posterior involvement of the pelvic ring was investigated. This analysis was repeated after patients were stratified into a high-risk vs a low-risk group according to patient characteristic (age, gender, comorbidities, mobility, living situation). This allowed to assess the impact of the different fracture morphologies on mortality in fit vs. frail senior patients separately Results:Overall, 428 patients (83.4% female) with a mean age of 83.7 years were included. Two thirds of patients were still living in their home and mobile without walking aid at baseline. In-hospital mortality was 0.7%, overall, one-year mortality 16.9%. An independent and significant association of age, gender and comorbidities to overall survival was found. Further, the occurrence of a horizontal sacral fracture as well as a ventral comminution or dislocation was associated with an increased mortality. The effect of a horizontal sacral fracture was more accentuated in low-risk patients while the ventral fracture components showed a larger effect on survival in high-risk patients.Conclusion:Specific fracture characteristics may indicate a higher risk of mortality in conservatively treated patients with a low-energy pelvic ring fracture. Hence, they should be taken into account in future treatment algorithms and decisions on patient management.

So far, age has been found to be a prognostic factor for mortality in patients with low-energy pelvic ring fractures [3]. Further, the AO-classi cation of pelvic fractures has been shown to correlate with mortality [19]. A correlation of the "fragility fractures of the pelvis" (FFP) classi cation to mortality was not found [20].
Here, we focus on single fracture characteristics of low-energy pelvic ring fractures which are also re ected in the FFP classi cation like the extent of the dorsal and ventral aspect of the pelvic ring fracture (absent, unilateral, bilateral), the presence/absence of a horizontal sacral fracture and the presence/absence of comminution and/or dislocation of the ventral fracture.
The aim of the current analysis is to evaluate these fracture characteristics considering their association with mortality and to assess prognostic factors in elderly patients with a low-energy pelvic ring fractures regarding outcome.

Material And Methods
For the present analysis, all patients ≥60 years old treated for a CT scan veri ed low-energy pelvic ring fracture at our hospital between August 2006 and December 2018 were included unless a dissent for the use of routine data was documented. The study was approved by the competent ethical committee "Ethikkommission Nordwest-und Zentralschweiz; EKNZ" (Ref. 2017-01859, ClinicalTrials.gov Identi er: NCT03476824). All methods were performed in accordance with the relevant guidelines and regulation. It followed applicable law as well as good clinical practice (GCP) and the Declaration of Helsinki.
Patients were identi ed by screening the electronic patients' charts for the following keywords: pelvic ring fracture, pelvic fracture, ala fracture, sacrum fracture and ramus pubis fracture. We also screened for acetabular fractures in order not to miss fractures of the anterior wall as extension of a pubic ramus fracture. The injury mechanism from anamnestic information allowed a discrimination between patients with low-or high-energy injuries according to a predetermined list of possible trauma mechanisms.
Patients with high-energy trauma like tra c accidents, bicycle accidents, falls on stairways or falls from considerable height other than standing position were excluded.
Additionally, the following six patient characteristics were collected from patients' charts: age, gender, comorbidities (assessed by the Elixhauser Comorbidity Index (ECI) [21]), mobility in daily activities (assessed by the Parker Mobility score (PMS) [22], living situation, use of walking aids. Information on treatment (surgery vs. conservative) was obtained from the medical records. For the purpose of this study, only conservatively treated patients were considered.
Each CT scan was classi ed according to the traditional FFP classi cation [23] and the fracture additionally broken down into the description of ve speci c fracture characteristic: extent of the dorsal aspect of the pelvic ring fracture (absent, unilateral, bilateral) presence/absence of a horizontal sacral fracture extent of the ventral aspect of the pelvic ring fracture (absent, unilateral, bilateral) presence/absence of comminution of the ventral fracture presence/absence of a dislocation of the ventral fracture Information on survival was obtained from patient charts as well as by contacting the registration o ce of the canton and the national pension scheme.
Survival times are de ned as the time from rst presentation and fracture diagnosis until death. If information on survival was obtained from the registration o ce of the canton, relocation to another canton as well as the end of the observation period (January 2020) were regarded as censoring events. If survival information was obtained from the national pension scheme, only the end of the observation period was regarded as censoring event.
Strati ed Kaplan-Maier-curves are used to visualize the association of overall survival with single variables. Age, the PMS and the ECI were categorized for these analyses. The log rank test was used to assess the statistical signi cance of the differences between the strata shown.
The Cox proportional hazard model was used for multivariate analyses. Hazard ratios together with 95%con dence intervals and p-values are reported. The extent of dorsal and ventral fractures was handled as continuous covariates.
For the development of a prognostic index, age, PMS and comorbidity enter the models as quadratic functions. Living status was handled as a categorical covariate. Missing values in the PMS, living status or use of walking aids were handled by the missing indicator approach, as these missing values are likely to be informative and carry prognostic information.
The independent prognostic value of the ve fracture characteristics was assessed using a multivariable Cox model with all ve fracture characteristics.
The analyses on the correlation of speci c fracture characteristics to mortality are then repeated separately in low-risk and high-risk patients using patient characteristics (age, gender, comorbidity, mobility, living situation and use of walking aid) for risk strati cation ( Figure 1). Considering the prognostic value of the above-described patient characteristics, a strati cation into six risk classes of equal size was made, whereby class 1-3 (high risk, frail patients) and class 4-6 (low risk, t patients) were then grouped and used for further analysis.
The statistical signi cance level was set to 0.05. All computations were performed using STATA 16.1 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC).

Results
Overall, 548 patients treated for a CT scan con rmed FFP within the de ned time period could be identi ed in the clinic information system. Eighty-six patients were treated operatively and thus excluded from this analysis. Among the remaining 462 patients, survival data could not be obtained from 34 patients. Hence, 428 patients were considered for this study. The median follow-up time was 59 months, and one year mortality was 18.5%. In total, 225 patients died in the course of follow-up.
Patient characteristics are depicted in Table 1. The mean age was 83.7 years (range 61-104 years), the majority were women (83.4%) and 2/3 still living in their home. Before the accident, two third were mobile without walking aid. Half of the patients had 3 or more relevant comorbidities. Most patients were discharged either to a rehabilitation centre or nursing home. In-hospital mortality was 0.7%. The fracture classi cation according to the modi ed FFP classi cation with its speci c fracture characteristics and the traditional FFP classi cation respectively are depicted in Table 2.

Patient characteristics and mortality
The relation of the six analysed baseline patient characteristics to survival is shown in Figure 2. We observe clear associations: male gender, higher age, living in a nursing home, use of walking aids, any de cit in mobility indicated by a PMS below the optimal value of 9, and two or more comorbidities are associated with poor survival in these patients with a conservatively treated low-energy pelvic ring fracture. A multivariate analysis con rmed an independent prognostic value of gender (p=0.001), age (p<0.001) and comorbidities measured by ECI (p<0.001) ( Table 3). The signi cant effects of age, gender and comorbidity allowed us to conclude that these three variables carry independent, prognostic value. The other three variables did not show a signi cant effect, but they showed effects in the expected directions (worse survival if living at a nursing home or alone or if using walking aids, decreased survival with lower PMS). Fracture characteristics and mortality The correlation of the individual fracture characteristics to survival is shown in Figure 4. We observe a signi cantly decreased survival in the presence of a horizontal sacral fracture and a comminuted ventral fracture. Survival was also decreased in patients with a dislocated ventral fracture, but this was not signi cant. For the extent of the fracture in the dorsal or ventral pelvic ring no association could be observed. A multivariate analysis con rmed an independent prognostic value of horizontal sacral and comminuted ventral fractures with hazard ratios (HR

Fracture characteristics of frail and t elderly patients and mortality
Further assessment of the relation of speci c fracture characteristics separated for low-risk and high-risk patients was performed. Patients were grouped into 6 classes of equal size according to their overall mortality risk considering the following variables: age, gender, comorbidities, Parker mobility score, use of walking aids and living situation.
Based on these ndings, a risk index based on all six factors was constructed, allowing to differentiate clearly among low-risk and high-risk subjects (Figure 1 and Figure 3. In the following the terms "high-risk, frail patients" and "low-risk, t patients" refer to the groups of patients in classes 1-3 and 4-6. We observe a more pronounced association for the horizontal sacral fracture component in low-risk patients than in high-risk patients and a more pronounced association for the comminuted and dislocated ventral fractures in high-risk patients ( Figure 5). Separate multivariate analyses in high-risk and low-risk patients could con rm an independent prognostic value of horizontal sacral fractures in low-

FFP and mortality
No association of the FFP classi cation with survival was observed in these cohort of conservatively treated patients (Figure 6). Neither the full classi cation with its 11 subcategories nor the concentration on the four main categories of the FFP classi cation showed a signi cant association with mortality.

Discussion
In this study of a cohort of conservatively treated patients with a low-energy pelvic ring fracture, we could demonstrate a relation of three speci c fracture characteristics, namely horizontal sacral fractures, comminuted ventral fractures and dislocated ventral fracture, with overall survival. In contrast, the extent of the fracture (absent, unilateral, bilateral) in the dorsal or ventral ring was not associated with overall survival.
Previously a signi cant difference in survival of elderly patients with AO type A compared to AO type B pelvic fractures after low-energy trauma with 80% of conservative treatment has been reported [19]. In contrast, the categories of the FFP classi cation have neither in our population nor in a previous study been reported to be associated with survival or mortality [20]. Also, the alphanumeric classi cation by Krappinger et al has not yet been reported to correlate to outcome measures [24].
Our results suggest that speci c aspects of the dorsal as well as ventral fracture seem to have an impact on survival and it might therefore be favourable to take these into account in future decision-making regarding treatment options for these patients. These three speci c aspects of pelvic fractures in the elderly seem to have a higher impact than the extent of dorsal or ventral fractures. Horizontal sacral fractures are most frequently associated with an H-type fracture pattern that often triggers surgical stabilisation, so the unfavourable prognosis associated with horizontal sacral fractures may by already taken into account. In contrast, the impact of comminuted and dislocated anterior fractures on mortality might re ect a so far underestimated functional impact of fractures in the ventral pelvic ring.
A strati cation according to patients' risk reveals further insight. Horizontal sacral fractures were found to be particularly harmful in low-risk patients, whereas comminuted and dislocated ventral fractures proved more harmful in high-risk patients with a signi cant increase in mortality in these patients. This differential impact of the dorsal vs anterior fracture characteristic in distinctly different populations of elderly patients is somewhat surprising. In general, high-risk patients are characterized in our population by a substantial short-term mortality (25% mortality within the rst six months), whereas low risk patients are characterized by a slow long-term decline. As visible in Figure 5, in frail, high risk patients, comminuted or dislocated ventral fractures seem to increase the short-term mortality, whereas horizontal sacral fracture do not show this effect. In contrast, in t, low risk patients, horizontal sacral fractures seem to have an unfavourable effect on the long-term decline. These ndings suggest that interpretation of fracture characteristics with respect to decision making may have to take into account the degree of frailty of the patients.
In line with this interpretation the impact of fracture characteristics on mortality seems to be smaller compared to patient characteristics. Age, existing decreased mobility, and comorbidities are directly associated with an increased mortality risk. The same tendency was observed for patient characteristics that did not directly and statistically relevantly correlate with mortality, ie. living situation, use of walking aid and PMS. The lack of signi cance can be explained by the fact that these three all tend to measure the same concept, namely the degree of mobility and as consequence independence in daily living.
Indeed, if added as single variables to a base model with age, gender and comorbidity as covariates, they always resulted in a statistically signi cant effect. These ndings underline the frailty of this speci c patient group and the vital importance of mobility [25]. One-year mortality after surgical fracture stabilisation has been reported between 10-28% [20,29].
We reported only on conservatively treated patients. The restriction of this analysis to conservatively treated patients avoids an undesirable effect of surgery as a confounding factor. Surgery can unpredictably impact outcome and survival in both directions with a potential improvement, but also with a potential deterioration due to postoperative complications and surgery-related morbidity. Operatively treated patients with a pelvic fragility fracture might represent a selected subpopulation. However, even if the decision for surgery depends directly on the fracture characteristics, this does not bias the estimation of the effect of the prognostic value of these factors [30].
In addition, the analysis of conservatively treated patients allows to discuss possible adaptations of current practice and speci c analyses of potentially underserved subpopulations.
The majority of our population was independently mobile and living prior to fracture diagnosis. Almost a quarter of this conservatively treated population further presented with a FFP type III or IV. Furthermore, treatment strategies generally aim at restoring baseline mobility and independence, thus rst evaluating the success of conservative treatment.
We acknowledge the limitations of our study. First, it is a retrospective analysis. As follow up information was restricted to survival, data could nevertheless reliably be obtained for almost all patients using o cial records. Nevertheless, using overall survival as measure probably allows only to have a look at the tip of the iceberg with the true triggers and points of possible interventions hidden. Rigorous follow-up assessments with respect to mobility and quality of life may produce better insights into the association with patient relevant outcomes.

Conclusion
Speci c fracture characteristics such as horizontal sacral fracture, comminuted and dislocated ventral fractures may indicate a higher mortality risk of patients with a pelvic fragility fracture. Hence, they should be taken into account in future treatment algorithms and decisions on patient management.

Declarations
Ethics approval and consent to participate: