Background: The decision between open reduction and internal fixation (ORIF) and arthroplasty for a displaced femoral neck fracture in a patient ≤65 years old can be challenging. Both options have potential drawbacks; if a fracture treated with ORIF fails to heal it may require a revision operation, whereas a relatively young patient who undergoes arthroplasty may need revision within his/her lifetime. The purpose of this study was to employ decision analysis modeling techniques to generate evidence-based treatment recommendations in this clinical scenario.
Methods: A Markov decision analytic model was created to simulate outcomes after ORIF, total hip arthroplasty (THA), or hemiarthroplasty in patients who had sustained a displaced femoral neck fracture between the ages of 40 and 65 years. The variables in the model were populated with values from studies with high-level evidence and from national registry data reported in the literature. The model was used to estimate the threshold age above which THA would be the superior strategy. Results were tested using sensitivity analysis and probabilistic statistical analysis.
Results: THA was found to be a cost-effective option for a displaced femoral neck fracture in an otherwise healthy patient who is >54 years old, a patient with mild comorbidity who is >47 years old, and a patient with multiple comorbidities who is >44 years old. The average clinical outcomes of THA and ORIF were similar for patients 40 to 65 years old, although ORIF had a wider variability in outcomes based on the success or failure of the initial fixation. For all ages and cases, hemiarthroplasty was associated with worse outcomes and higher costs.
Conclusions: Compared with ORIF, primary THA can be a cost-effective treatment for displaced femoral neck fractures in patients 45 to 65 years of age, with the age cutoff favoring THA decreasing as the medical comorbidity and risk of ORIF fixation failure increase. Hemiarthroplasty has worse outcomes at higher costs and is not recommended in this age group.
Level of Evidence: Economic and decision analysis Level III. See Instructions for Authors for a complete description of levels of evidence.
Displaced femoral neck fractures, resulting from ground-level falls or high-energy trauma, are common in patients of all ages1-3. They can be challenging to treat, as the femoral neck lacks periosteum and therefore relies on direct bone healing, leading to higher rates of nonunion or osteonecrosis of the femoral head4. Additionally, these injuries frequently occur in patients with medical comorbidities, making the prospect of extensive, complex operations and possibly revision surgery daunting5-7.
Arthroplasty is usually the treatment of choice for elderly patients, as it has the benefit of allowing earlier weight-bearing and does not rely on fracture-healing. Alternatively, open reduction and internal fixation (ORIF) is usually the treatment of choice for younger patients because, if the fracture unites, outcomes can be favorable8-10. However, nonunion and osteonecrosis are both concerns, and failure of the fracture to successfully heal may require repeat surgery such as salvage osteotomy in younger patients and conversion to arthroplasty in the middle-aged population11,12. Additionally, outcomes of arthroplasty after failed ORIF have been shown to be generally worse than those after successful primary total hip arthroplasty (THA)13-16.
The exact patient age at which to transition from ORIF to arthroplasty for displaced femoral neck fractures has been poorly defined. Both treatment options have potential drawbacks for “middle-aged” patients; ORIF may be unsuccessful and result in nonunion or osteonecrosis requiring a revision operation, whereas an arthroplasty in a relatively young patient may have to be revised within the patient’s lifetime. It is unclear which of these factors plays a greater role in clinical outcomes or at what patient age that balance shifts.
The purpose of this study was to employ decision analysis modeling techniques to generate evidence-based treatment recommendations to aid in the decision of whether to perform ORIF, THA, or hemiarthroplasty, as a function of both age and medical comorbidity, in a patient with a displaced femoral neck fracture.
Materials and Methods
This economic decision analysis was conducted in accordance with guidelines set by the Panel on Cost-Effectiveness in Health and Medicine17. A lifetime time horizon was used, with a discount rate of 3%, and a societal perspective was used for economic analyses. An incremental cost-effectiveness ratio (ICER) cutoff of $100,000/quality-adjusted life year (QALY) was used18,19. Analysis was performed using decision analysis modeling software (TreeAge Pro 2015, version 15.1). To add context to the results, we also conducted an analysis of the National Inpatient Sample (NIS) database to estimate the volume of these injuries and describe current practice behavior.
A Markov model was used to estimate outcomes in a theoretical cohort of patients with an acute displaced femoral neck fracture treated with ORIF, THA, or hemiarthroplasty (Fig. 1). After ORIF, patients either had successful healing or had nonunion or osteonecrosis requiring conversion to THA. In that cohort, conversion of failed ORIF to THA was modeled to behave as a revision THA, as the literature has shown that conversions of ORIF procedures to THAs have worse outcomes than primary THAs and there are often issues with deformity, loss of bone stock, and scarring from prior approaches that are closer to those encountered in a revision THA11-16.
Patients who underwent primary arthroplasty (either THA or hemiarthroplasty) were modeled to have a specified annual rate of arthroplasty failure requiring revision and, once they underwent revision, were assigned to have annual rerevision rates. Each procedure was assigned a cost and, for each year of life, patients were given QALYs based on assigned utility values for that health state.
Model Inputs (Table I)
Clinical Outcome Probabilities
The probability of revision to arthroplasty after ORIF for femoral neck fracture was estimated on the basis of a systematic literature review. On February 1, 2016, a literature search was performed through PubMed with the search terms “displaced,” “femoral neck,” “fracture,” and “operative fixation.” Studies were selected for inclusion when the subjects had undergone ORIF of a femoral neck fracture, the average duration of follow-up had been at least 1 year, and rates of revision to arthroplasty had been documented; 2,019 patients were included for final analysis (Table II). The quality of the literature included was assessed using the Methodological Index for Non-Randomized Studies (MINORS) criteria20 (see Appendix).
The revision rate of primary THA was obtained from the Australian Orthopaedic Association National Joint Replacement Registry21, using survivorship data on patients who had undergone THA for a femoral neck fracture. Rather than use a single constant annual revision rate, the failure rate varied by postoperative year according to the rates found in the registry data, as recent studies suggested that short-term complications may have a larger effect on outcomes than long-term failure rates22. Failure rates at >10 years postoperatively were not available from the registry data, so they were obtained from a literature review through PubMed on February 1, 2016, with the search terms “total hip arthroplasty,” “longterm outcomes,” and “survivorship.” The Appendix shows a full list of failure rates by year along with references.
The hemiarthroplasty failure rate was expressed as a value relative to the annual revision rate for THA, and has varied from roughly equivalent to the THA revision rate23,24 to rates 4 to 6 times higher than the THA revision rate25-28. For the base case, we used the relative revision rate in the Australian registry for patients who sustained a femoral neck fracture before the age of 75 years; the revision rate in those cases was 1.7 times higher than the revision rate for THA21.
The rerevision rate for patients who had already undergone a revision THA was also based on the Australian Orthopaedic Association National Joint Replacement Registry21 (see Appendix). On the basis of the registry data, it was assumed that, once >10 years had elapsed since a revision, the rerevision rate was 3 times the rate of revision for a primary THA.
Annual all-cause mortality rates were obtained directly from U.S. life tables as a function of age29 (see Appendix). The perioperative mortality rate for the index procedure was calculated as 5 times the baseline annual mortality rate for a patient of the same age23,30-32 and was assumed to be equivalent for all primary procedures. The perioperative mortality rate for revision surgery was calculated as 3.24 times that for an elective primary THA30, with the mortality rates for elective primary THA obtained from the National Joint Registry for England and Wales33.
The effect of medical comorbidity on mortality and life expectancy was modeled using the Charlson Comorbidity Index (CCI)34-36. Patients were divided into 3 groups: (1) relatively healthy (CCI of 0), (2) mild comorbidity (CCI of 1 or 2), and (3) multiple comorbidities (CCI of ≥3). For each category, the mortality risk ratio was estimated on the basis of values reported in the literature34,37,38 (Table I). This was modeled to affect both the annual all-cause mortality as well as the perioperative mortality. For our base case, we assumed patients were relatively healthy (CCI of 0), although that was varied in the sensitivity analysis.
All historical costs were scaled to 2015 U.S. dollars according to the Consumer Price Index for Medical Care39.
Total costs were estimated by combining the facility costs along with individual surgeon fees. Facility cost data were obtained from the Healthcare Cost and Utilization Project (HCUP) NIS database40 from 2011 (the most recent year that had full data on treatment type). We identified patients who had been treated for a femoral neck fracture (International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] diagnosis codes 820.00 through 820.19, 820.8, and 820.9) and obtained the mean hospital cost using Clinical Classifications Software (CCS) procedure codes 146 and 153 for ORIF and arthroplasty, respectively, resulting in costs of $16,312 and $17,223, respectively. Costs were obtained from the web site of the HCUP, which estimates costs using documented charges and known institutional cost-to-charge ratios41.
Surgeon fees were estimated using relative value unit (RVU) estimates from the American Academy of Orthopaedic Surgeons (AAOS) practice management data42, which showed work RVUs of 17.61 for ORIF, 21.79 for THA, and 16.64 for hemiarthroplasty. We estimated a conversion factor of $60 per RVU, which led to total overall costs of $19,252 for ORIF, $20,529 for THA, and $20,203 for hemiarthroplasty (in 2015 U.S. dollars). The total cost of revision THA was based on values reported in the literature22,23,43-45, with $34,700 used for the base case.
Health-Related Quality of Life (Health Utility States)
For patients undergoing primary and revision THA, health utility was modeled using data from studies specifically designed to assess that population46 (Table I). Utility after ORIF and hemiarthroplasty was obtained from Swedish registry data, which showed that ORIF and THA resulted in similar functional scores in patients >65 years of age but THA had superior outcomes for patients >6547 (see Appendix). Functional outcome scores after hemiarthroplasty were 28% lower than those after primary THA according to the Swedish registry data. Finally, there was a QALY loss of −0.20 (disutility) with each revision23.
All of the inputs were varied through their 95% confidence intervals (CIs) (Table I) in 1-way sensitivity analysis, and the change in the cutoff age at the extreme values was calculated. If changing the value through that interval modified the cutoff age by <5%, the decision was considered to be relatively insensitive to that variable. For the variables that were >5% sensitive, the relationship was explored with multiway sensitivity analysis.
To determine the overall effect of the simultaneous uncertainty of all of the variables on the model conclusions, probabilistic sensitivity analysis was performed. In this type of analysis, all of the variables are allowed to vary randomly within their 95% CIs (full variable distributions shown in the Appendix), and the model is iterated multiple times using Monte Carlo simulation to assess the stability of the results.
Current Practice Patterns
To evaluate national volume and practice behavior, an analysis of the NIS database was conducted. Data were evaluated for all patients who had been treated for a femoral neck fracture (ICD-9-CM diagnosis codes 820.00 through 820.19, 820.8, and 820.9) and were compared with U.S. Census data to calculate incidence. The ICD-9-CM diagnosis code does not distinguish between displaced and nondisplaced fractures, so all treatment choices including percutaneous fixation with or without reduction were returned in the initial results. We chose to directly compare the rates of the 3 treatment choices being investigated with this analysis, and the ICD-9-CM procedure code for ORIF (79.35), THA (81.51), or hemiarthroplasty (81.52) was used to estimate rates of treatment decisions.
The analysis of the NIS database showed that, in 2011, 162,000 femoral neck fractures occurred, 18,000 (11%) of which were in patients 45 to 64 years of age. Patients younger than 45 were treated with ORIF 83% of the time, whereas patients 65 or older were treated with arthroplasty 88% of the time. The transition occurred in the 45-to-64 age range, in which 27% were treated with ORIF; 22%, with THA; and 51%, with hemiarthroplasty.
Costs and Reoperation Rate
The total lifetime costs, health utility states (QALYs), and reoperation rates for ORIF, THA, and hemiarthroplasty as a function of age at the time of injury are shown in Table III. The results of the model predicted that hemiarthroplasty would generally have the highest lifetime costs. THA had higher lifetime costs than ORIF when performed on relatively young patients, but the cost of THA decreased as patient age at the time of injury increased. Once patients were older than 56, the cost of THA was lower than that of ORIF (Fig. 2).
Hemiarthroplasty also had the highest reoperation rate in most age groups. Younger patients who underwent THA had a higher average reoperation rate over their lifetime than did those treated with ORIF, although once patients were 59 years old at the time of injury THA was followed by fewer total lifetime revisions than ORIF (Fig. 3-A). The patients who underwent ORIF were subdivided into those who had successful healing after the index surgery and those in whom the fracture failed to unite and required revision surgery. This analysis showed that patients between 40 and 65 years old in whom the fracture failed to heal after ORIF required an average of 1.9 additional operations throughout their lifetime, whereas those who underwent primary THA eventually required 0.4 operation on average (Fig. 3-B).
Hemiarthroplasty had uniformly inferior health outcomes. Results were similar between ORIF and THA in patients 40 to 65 years old, whereas THA began to have superior outcomes after the age of 65 (Fig. 4-A). Again, the results were subdivided according to the success of the primary ORIF, which showed that patients between 40 and 65 with successful fracture union after ORIF had utility outcomes that were, on average, 2.6% better than those after primary THA, whereas those in whom the ORIF failed had, on average, outcomes that were 11.6% worse than those following THA (Fig. 4-B).
Cost Effectiveness and Threshold Ages
When comparing both cost and quality of life, we found that THA became cost-effective after the age of 54 years for patients with no medical comorbidities, after 47 years old for patients with mild comorbidity, and after 44 years old for patients with multiple comorbidities. Using an alternative ICER threshold of $50,000/QALY changed those threshold ages by <1 year.
The model was highly sensitive to the failure rate of fracture-healing after ORIF and to the health utility after successful ORIF or after THA. Figure 5 shows the relationship among patient age, ORIF failure rate, and medical comorbidity level. The results were moderately sensitive to the revision rate of primary THA and were relatively insensitive to the remaining variables.
Probabilistic statistical analysis conducted to evaluate the stability of the conclusions showed that hemiarthroplasty was never the preferred option under any modeled circumstances. The conclusions derived from comparing ORIF with THA were stable for patients <45 years old (ORIF was generally the preferred strategy) or >65 years old (THA was the preferred strategy). For patients 45 to 65 years old, THA and ORIF had nearly equivalent cost-effectiveness (Fig. 6, with full acceptability curves shown in the Appendix).
Displaced femoral neck fractures in patients younger than 65 years of age can be challenging to manage. Case series have shown the failure rates of ORIF to be as low as 8%9 or as high as 35%48 in this patient population, and revision surgery can be challenging12,13,15. Similarly, the 10-year survival rate of THA in a younger population has been shown to be as high as 100%49 and as low as 75%50. When uncertainty about outcomes makes decision-making difficult, quantitative analyses can help objectify risks, define areas where there is more or less ambiguity, and be used to support clinical decision-making44. By defining the critical variables that determine outcomes, we can add structure and evidence-based recommendations to the decision process.
The results of this study showed that both ORIF and THA can be cost-effective options for patients who sustain a displaced femoral neck fracture between 40 and 65 years of age, whereas hemiarthroplasty has inferior results. Although ORIF and THA have similar health outcomes on average, patients who undergo ORIF and have fracture healing have slightly better outcomes than those who undergo THA, with considerably lower costs, while those in whom the fracture fails to heal after ORIF have notably worse outcomes and a higher reoperation rate than patients treated with THA. The success rate of ORIF is a function of (1) patient factors such as injury mechanism, medical comorbidity, and fracture pattern; (2) surgeon factors, including comfort with performing complex ORIF and primary THA; and (3) system factors, including operating room resources/accessibility and availability of specialist surgeons trained in either arthroplasty or advanced fracture fixation.
Our study suggests that the transitional age at which THA should be considered is 54 years old for healthy patients, 47 years for those with mild comorbidity, and 44 years for those with multiple comorbidities. However, rather than specifying treatment decisions, it is our hope that this study can add clarity to the conversation between patient and surgeon. The surgeon can evaluate the patient’s injury characteristics and comorbidities to estimate the likelihood of ORIF resulting in successful healing and use that estimate to counsel the patient about treatment options through a process of shared decision-making.
This study is strengthened by the body of high-quality literature describing outcomes after both arthroplasty and ORIF, which improves the strength and reliability of the conclusions. Furthermore, this is a clinical scenario that is amenable to well-established modeling techniques, which facilitate testing of areas where clinical probabilities are less defined and allow quantification of uncertainty. The use of the NIS database analysis adds context to these conclusions, showing that more than 1 of 10 femoral neck fractures occur in patients in the age group evaluated in this study and that this truly is a transitional age group for which decision-making is challenging.
There are several limitations and caveats to this analysis. When interpreting these results, it is important to recognize that the conclusions are sensitive to several variables. Whenever possible, we attempted to improve confidence in the results by obtaining critical values from only the highest-quality evidence (usually from national clinical registries with data from thousands of patients), and we used literature review when registry data were not available. For example, we relied on systematic review to estimate failure of ORIF and modeled failure to occur immediately postoperatively, although in reality these failures are likely spread out throughout a 1 to 3-year postoperative period. We included data from studies with follow-up as long as 10 years. However, there are limited data on ORIF failure rates beyond that point, so it is possible that our model is overestimating the long-term success rate of ORIF. For this variable, and other areas where there was uncertainty and registry data were not available, we used multiway sensitivity analysis to quantify that relationship to allow recommendations to be adjusted for alternate clinical scenarios. Additionally, our cost estimates required the use of cost-to-charge ratios and were obtained from the HCUP database, which relies on ICD-9-CM diagnosis coding; such coding does not distinguish between displaced and nondisplaced femoral neck fractures. Although this decreases the precision of the cost estimates, the net cost estimates for the 3 primary procedures were relatively similar, and the overall conclusions were relatively insensitive to the initial procedure costs. As a result, we believe that, despite these areas of uncertainty, this model represents valid evidence-based recommendations based on the highest-quality data available.
It is also important to clarify the clinical scenario that was being evaluated. This analysis focused on patients 40 to 65 years old, for whom the operative decision is between ORIF and THA and a failure of ORIF would most likely be treated with conversion to THA. The decision between THA and hemiarthroplasty in a more elderly population was outside the scope of this analysis and has already been extensively analyzed23,25. Additionally, there is also a younger patient population for whom a salvage procedure such as valgus osteotomy would be considered following failed ORIF; that is a much more complex and patient-specific decision process.
In conclusion, the results of this analysis support the decision to perform either ORIF or primary THA in patients between the ages of 40 and 65 years who have a displaced femoral neck fracture, with the age at which to transition to primary THA found to be between 44 and 54 years old depending on medical comorbidities and the risk of fixation failure. The results of this study can be used to provide quantitative support for the shared decision-making that should occur between the patient and physician.
Tables showing the quality of the studies included in the analysis, THA implant survival rates, annual all-cause mortality rates by age, perioperative mortality rates during revision THA by age, utility ratios and failure/revision rates of ORIF by age, and distribution of values used in the probabilistic sensitivity analysis, as well as figures showing acceptability curves at different ages and medical comorbidities derived from the probabilistic statistical analysis, are available with the online version of this article as a data supplement at jbjs.org.
Investigation performed at the Department of Orthopaedic Surgery, Carolinas Medical Center, Charlotte, North Carolina
Disclosure: There was no external funding source for this study. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work and “yes” to indicate that the author had other relationships or activities that could be perceived to influence, or have the potential to influence, what was written in this work.
- Copyright © 2017 by The Journal of Bone and Joint Surgery, Incorporated