The impact of fixation type for intertrochanteric femoral fracture on patient survival

Intramedullary (IM) nail fixation for intertrochanteric fractures is potentially associated with improved postoperative function but may have an increased mortality risk compared to sliding hip screw (SHS) fixation. This study investigated postoperative mortality risk between surgical fixation type for intertrochanteric fracture in patients aged 50 years and older using linked data from the Australian Hip Fracture Registry and National Death Index.


Background
In Australia, approximately 18 000 people present to hospital each year with a hip fracture representing one of the largest groups undergoing unplanned orthopaedic surgery. [1][2][3] Of these patients, the mortality rate following hip fracture within 30 days of surgery averages 8.2% with most of these deaths occurring during the initial hospitalization for the injury. 2 However, there is also an associated excess mortality that persists for up to 5 years following injury, and in the context of an ageing population, this burden is projected to increase in the foreseeable future. [3][4][5] For patients presenting with hip fracture, international guidelines provide advice on management options but the choice of type of surgical fixation is often a multifactorial decision based on fracture type, degree of fracture displacement and comminution, patient premorbid function, and surgeon experience. [6][7][8][9][10] Intertrochanteric fractures represent one of the most common types of fractures in Australia and New Zealand, and are usually treated with surgical fixation. [2][3][4][5][6] This is usually carried out via insertion of a sliding hip screw (SHS) or an intramedullary (IM) nail. [2][3][4][5][6] As discussed by Jos et al. 11 the approach to fixation appears to vary greatly due to surgeon preference and is likely related to the lack of clear consistent evidence on differences in mortality, postoperative pain scores, function, and reoperation rates. 12-14 NICE guidelines 7 recommend SHS over IM nail for intertrochanteric fracture due to the lower reoperation rate secondary to periprosthetic fracture as well as lower cost. There is also a concern that IM fixation is associated with a higher 30-day mortality when compared to SHS. 14 Whether IM fixation is associated with better functional recovery remains unresolved. [13][14][15] In a recent Cochrane review, Lewis et al. 15 reported benefits of IM nail over SHS fixation in terms of reduced risk of postoperative infection and non-union, but also recognized an increased risk in implant-related fracture. The review also highlighted the lack of quality evidence using large datasets (>1000 participants) to determine the impact of type of fixation for intertrochanteric fractures on health-related quality of life, on functional outcomes following more unstable fracture patterns and on the role of long versus short nail designs. 15 This primary aim of this study was to determine the association between 30-day mortality and choice of surgical fixation, comparing short IM nail, long IM nail and SHS fixation, for patients aged 50 years and over presenting with intertrochanteric fractures. The secondary aim was to analyse this association over the first year following surgery.
This study was granted ethical approval by the University of New South Wales Human Research Advisory Panel (HC210482).

Study design and data sources
This retrospective cohort study uses linked data from the Australian Hip Fracture Registry (AHFR) and the Australian Institute of Health and Welfare (AIHW) National Death Index (NDI). The AHFR collects individual and facility level data on hip fracture hospitalisations across Australia to facilitate improvement in clinical care and patient outcomes. 1 Eligibility criteria includes Australians aged 50 years and older admitted to an Australian hospital with a low trauma hip fracture. Representing approximately 75% of Australian public hospitals that operate of hip fractures, the AHFR contains 64 890 records from 81 hospitals. 1 Mortality data is linked annually by the AHFR with the AIHW NDI. 6

Case selection
The study cohort comprised people aged 50 years and older admitted to an Australian hospital with an intertrochanteric femoral fracture and treated with surgical fixation (short IM nail, long IM nail or SHS) between 1 January 2016 and 31 December 2020.

Data analysis
The primary outcome was mortality within 30-days of surgery and the secondary outcome was mortality within 1-year of surgery.
Covariates associated with mortality in hip fracture were selected for inclusion in the analysis; age group, sex, American Society of Anaesthesiologists (ASA) classification, usual place of residence, preadmission cognitive status and preadmission walking ability. Selection of these variables were based on clinician review of data available through the AHFR and informed by other studies. 15 Descriptive analysis was undertaken using SAS 8.3 Enterprise Guide 16 to show between group differences in mortality based on type of fixation selected at the time of the surgical procedure. Chi square tests were used for categorical variables and ANOVA tests for continuous variables. A P-value of <0.05 was considered significant. Kaplan-Meier survival analysis was used to show the unadjusted difference in mortality between groups within 30-days and 1-year from date of surgery. It should be noted that there was a difference in cohort size as the date was truncated to 1 December 2020 and 31 December 2019 for 30-day mortality and 1-year mortality analyses post-surgery, respectively to allow adequate followup time.
The primary analysis used adjusted multilevel logistic regression to test the association (expressed as an Odds Ratio and 95% confidence interval) between types of surgical fixation and 30-day mortality, adjusted for covariates and hospital-level clustering. Given the known institutional variation, an instrumental variable analysis was performed using hospital preference for type of surgical fixation as the instrument and individual 30-day mortality as the outcome. This was conducted in R Environment for Statistical Computing. 17 Instrumental variable analysis does not rely on the assumption of no unmeasured confounding and allows casual inference from observational data. To further explore the association within the 30-day period, Cox proportional hazard modelling was used (expressed as a Hazard Ratio and 95% confidence interval), adjusted for covariates.
Analyses were performed to explore whether the findings of the model continued at 1 year using Cox's proportional hazard modelling (expressed as a Hazard Ratio and 95% confidence interval) between type of surgical fixation and mortality within 1-year following surgery, adjusting for the covariates listed above.

Results
The study included 16 667 records of cases undergoing surgical fixation for intertrochanteric femoral fracture. The mean age was 83.1 years (AE9.8), 11 315 (67.9%) were female and 11 654 (70.7%) were from private residence. Twenty-five cases were excluded as the date of death preceded the date of hospital presentation. In this study, 7300 (43.8%) underwent short IM nail fixation, 4251 (25.5%) underwent long IM nail fixation, and the remaining 5126 (30.7%) cases received a SHS fixation. IM nail fixation (both short and long) was more likely in female and older patients, and less likely in those walking independently ( Table 1).
As shown in Figure 1, the Kaplan-Meier survival curve demonstrated no significant difference in 30-day mortality following surgery by type of surgical fixation (P = 0.1). As shown in Figure 2, there was no significant difference in mortality within 1-year of surgery (P = 0.8).
The primary analysis looking at 30-day mortality using adjusted multilevel logistic regression demonstrated a significant difference in mortality for the long IM nail group compared to the short IM nail group (OR 1.2, 95% CI 1.0-1.4, P < 0.05) ( Table 2). The adjusted Cox proportional hazard modelling for mortality at 30-days when comparing long IM nail to short IM nail fixation was not significant (HR 1.1 95% CI 1.0-1.3 P = 0.09). There was no significant difference in 30-day mortality between short IM nail and SHS in the adjusted multilevel logistic regression (OR 1.1, 95% CI 0.9-1.3, P = 0.5) nor in the adjusted Cox proportional hazards regression (HR 1.1, 95% CI 0.9-1.2, P = 0.4) ( Table 2).
Instrumental variable analysis revealed that type of fixation was not significantly associated with 30-day mortality when comparing short IM nail fixation to long IM nail fixation (OR 1.0, 95% CI 0.9-1.0, P = 0.4) and SHS (OR 1.0, 95% 0.9-1.0, P = 0.2). The test for weak instrument had a very low P-value (P < 0.0001) meaning that hospital preference was a strong predictor of type of fixation and therefore suitable for use as the instrument in this analysis. The adjusted multilevel logistic regression demonstrated no significant difference at 1 year when comparing short IM nail fixation to long IM nail (OR 1.1, 95% CI 0.9-1.2, P = 0.2) and SHS (OR 1.0, 95% CI 0.9-1.1, P = 0.6) fixation. Likewise, adjusted Cox regression showed no significant difference in mortality between groups at 1 year comparing short IM nail to long IM nail (HR 1.1 95% CI 0.97-1.2, P = 0.2) and SHS (HR 1.0, 95% CI 0.9-1.1, P = 0.8) fixation.

Discussion
This study demonstrated a significant increase in 30-day mortality with long IM nail fixation compared to short IM nail fixation using adjusted multilevel regression. However, this was not demonstrated by the Cox modelling. Further to this, the IV regression showed no causal association between type of fixation and 30-day mortality, suggesting the association in the primary analysis was due to  unmeasured confounding. There was no significant association in 1-year mortality between long IM nail and SHS compared to short IM nail for intertrochanteric fracture.
The main findings are consistent with several publications. 12,13,15 Parker 13 demonstrated no significant difference in postoperative mortality following IM nail versus SHS in a randomized controlled trial of 1000 participants. However, mobility outcomes were better in the IM fixation group. Similarly, Aktselis et al. 12 did not show any difference in 1 year mortality between SHS and IM nail fixation for intertrochanteric fractures (AO/OTA 31-A2.2 and A2.3) in a randomized controlled trial of 80 participants. However, they reported IM fixation was associated with a significantly shorter operative time and improved postoperative quality of life as measured by the Barthel Index, at 1 year. These postoperative mobility and quality of life outcomes were not substantiated by the recent Cochrane review by Lewis et al. 15 which suggested there was insufficient evidence to recommend one approach to fixation over another in relation to mortality, functional outcomes, and quality of life. However, it did suggest that IM nails were associated with fewer superficial infections and non-unions but a higher risk of intra-operative and later fractures when compared to extramedullary fixation.
A recently reported study (82 990 patients) by Whitehouse et al., using information from the UK National Hip Fracture database, demonstrated a significant 12.5% increase in 30-day mortality for IM fixation compared to SHS for intertrochanteric fractures. 14 The study population 14 had similar demographic features to our study except that SHS fixation use was more common (86.7% compared to 30.7% in our study). Whitehouse et al. 14 conducted sensitivity analyses to minimize the impact of seasonal mortality variation on results, however, as a causal analysis was not conducted it is not possible to know if unmeasured confounders contributed to the significant difference in mortality reported between groups.
This study selected confounders for the adjusted analysis based on the variables available within the AHFR and that of previous studies. Patient comorbidities, time to surgery from hip fracture event (rather than date of hospital presentation), duration of anaesthesia and surgery and preoperative haemoglobin were not included in this study due to lack of data availability across the study period. 18 These factors may have influenced patient mortality risk across the groups. Further to this, it is possible that fracture severity (leading to the need for a long nail) may also be related to mortality in that this is associated with a longer procedure, longer anaesthetic and more bone reaming. This effect of unknown confounders was minimized by the causal analysis using instrumental variable regression which is a strength of this study.
Over time, the ANZHFR 6 has observed a year-on-year increase in the use of the IM nail over the SHS for intertrochanteric fractures. This has not been driven by any guideline recommendations and is not consistent with the existing ANZ Guideline for Hip Table 2 Adjusted multilevel logistic regression and adjusted Cox's proportional hazards regression modelling of the association between 30-day mortality and individual characteristics for intertrochanteric fractures receiving surgical fixation in AHFR dataset in people 50 years and over (n = 13 641) fracture Care or the recently updated UK NICE guideline (ref-CG124) both of which recommend the SHS mainly because of the cost differential. 7 What is leading to this change in practice is not clear and may possibly relate to marketing from the devices industry. Of note, long IM nails were observed to be used more commonly in frail, female and older patients. These patient factors are often associated with a higher risk for advanced osteoporosis and thus the effects of this on bone density may have influenced surgical decision. There is also substantial variation in preferred device by hospital site with some sites using almost exclusively nails. 1,6 Further work to better understand the change in practice in Australia is warranted as the observed trend to IM nails will ultimately see many junior doctors completing their training without or with limited exposure to a procedure that is currently recommended care. 9,10 The results of this study are generalizable to the Australian population being a large national registry derived cohort. As with most registry data, it is subject to human error in data entry and clinical judgement including classification of fracture type. However, data integrity is maximized by undertaking annual audits of a random sample of patients from each site as well as inbuilt alerts when improbable data is entered. We had hoped to study functional outcomes at 120-days, but data were incomplete with over 70% of the data missing for mobility and place of residence at this time point. This is important given the very limited evidence to support any difference in mortality and insufficient evidence to make a recommendation based on outcomes that are important to patients-pain, mobility, living independently and with good quality of life.

Conclusion
This study demonstrated a significant difference in mortality at 30 days between short and long IM nails for intertrochanteric hip fracture in the adjusted multilevel logistic regression, but the difference is likely to be related to residual confounding given there was no significance found in the adjusted Cox model nor IV analysis. In the absence of conclusive data around differences in mortality, it is important that analysis of large datasets focus on outcomes that include cost as well as function and quality of life to better guide future practice.