Trends in Utilization of Total Hip Arthroplasty for Femoral Neck Fractures in the United States
Dr. Summers is a resident in the Department of Orthopedic Surgery, University of Miami, Miami, Florida. Dr. Grau is a resident in the Department of Orthopedic Surgery, University of Miami, Miami, Florida. Mr. Massel is a medical student in the Department of Orthopedic Surgery, University of Miami, Miami, Florida. Dr. Ong is Assistant Professor in the Rothman Institute Department of Orthopedics, Thomas Jefferson University, Philadelphia, Pennsylvania. Dr. Orozco is Assistant Professor in the Rothman Institute Department of Orthopedics, Thomas Jefferson University, Philadelphia, Pennsylvania. Dr. Rosas is a research fellow in the Department of Orthopedic Surgery, University of Miami, Miami, Florida. Dr. Hernandez is Assistant Professor in the Department of Orthopedic Surgery, University of Miami, Miami, Florida.
Authors’ Disclosure Statement: The authors report no actual or potential conflict of interest in relation to this article.
Address correspondence to: Victor Hugo Hernandez, MD, MS, Department of Orthopaedics and Rehabilitation, University of Miami, 1400 NW 12th Ave, Miami, FL 33136 (tel, 305-243-3000; fax: 305-689-4979; email, email@example.com).
Am J Orthop. 2018;47(12). Copyright Frontline Medical Communications Inc. 2018. All rights reserved.
The ideal mode of fixation for patients with femoral neck fractures is not well defined in the current literature. This study describes the recent trends in surgical management of femoral neck fractures with an analysis on perioperative outcomes.
The National Hospital Discharge Survey was used to identify femoral neck fractures in the United States between 1990 and 2007 (n = 1,155,960) treated with open reduction and internal fixation (ORIF), total hip arthroplasty (THA), or hemiarthroplasty (HA). Trends were examined over the following 3 time periods: 1990 to 1995 (group 1), 1996 to 2001 (group 2), and 2002 to 2007 (group 3). Elixhauser Comorbidity Index and perioperative complications were calculated.
Use of HA increased (74.4% to 84.6%), whereas that of THA (7.3% to 4.9%) and ORIF (18.3% to 10.6%) decreased, from group 1 to group 3 in the age group of >80 years. The use of ORIF increased (63.9% to 81.4%), whereas the use of both HA and THA decreased, from group 1 to group 3 in the age group of <50 years. The rate of adverse events increased across all fixation types but was greatest among THA (32.2% to 48.3%).
The femoral neck patient population is now older and has more medical comorbidities. We observed a trend toward performing HA in older patients and ORIF in younger patients. Despite superior functional outcomes reported in THA, this study found a decreased utilization of THA in all age groups along with an increase in adverse events and nonroutine discharges for patients with femoral neck fractures treated with THA.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Femoral neck fractures are a common occurrence in the United States. A recent study estimated an incidence of >63 per 100,000 population.1-8 Although the incidence appears to have decreased over recent decades, there is a projected exponential increase in the incidence of hip fractures over the next 30 years in the baby boomer population.8,9 Given that these fractures have a significant impact on patient morbidity, mortality, and quality of life, research efforts have been directed toward optimizing the treatment of affected patients and improving the outcomes.4,9-24
The treatment of choice for femoral neck fractures and the use of total hip arthroplasty (THA)11 have been a topic of debate.4,9,10,15-17,22,25 Total hip arthroplasty has been advocated for younger, more active patients, whereas hemiarthroplasty (HA) has been reserved for patients who are older and less active. Although several studies have demonstrated that arthroplasty outperforms open reduction and internal fixation (ORIF) in the elderly population with displaced femoral neck fractures, ORIF is still commonly performed in the United States for nondisplaced fractures and in patients aged <50 years.26-29
In an attempt to quantify the use of THA in the treatment of femoral neck fractures and demonstrate the national trends, Miller and colleagues5 pooled the American Board of Orthopaedic Surgery (ABOS) database and analyzed the treatment trends of surgeons taking part II of the ABOS examination from 1999 to 2011. The authors found an increased utilization of THA by recently graduated orthopedic surgeons. In contrast, Jain and colleagues30 found different national trends when they analyzed data from the National Inpatient Sample containing data between 1990 and 2001 and further found decreased utilization of THA procedures by orthopedic surgeons of all levels of training nationwide. However, neither of these studies reported about the trends in demographics, comorbidities, risk factors, or outcomes in this patient population following surgery.
The purpose of this study was to help clarify the findings of these authors using the largest dataset to date and also report on the perioperative complications associated with each mode of fixation in patients who undergo operative treatment for femoral neck fractures in the United States. Our hypotheses were that the femoral neck fracture patient population has become older and has more medical comorbidities. We also hypothesized that there has been a trend toward performing fewer THA procedures in the United States and that THA is associated with increased perioperative complications compared to those with HA and ORIF.
MATERIALS AND METHODS
We conducted a retrospective epidemiological study using the National Hospital Discharge Survey (NHDS) on surgical trends in the management of femoral neck fractures. The NHDS is a publicly available survey that is conducted annually to provide data of nonfederal, short-stay hospitals to the public. The sample data are weighted to provide nationwide estimates of annual inpatient care. The NHDS includes up to 7 medical diagnoses and 4 procedural codes per case, which are categorized using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes, that were collected along with patient demographic information, length of stay (LOS), and discharge disposition. The diagnostic and procedural codes used for this study are presented in the Appendix. The year 2007 was chosen as the endpoint of this study due to the fact that the relative standard error of the NHDS doubled in 2008 as a result of a decrease in its survey size. As this is a publicly available database, our study was exempt from institutional review board approval.
All patients admitted with a primary diagnosis of closed transcervical fracture of the femoral neck (ICD-9-CM 820.0x) were selected. This resulted in 1,674,160 fractures. All patients with fractures with a concurrent primary procedural code of ORIF (79.35), HA (81.52), or THA (81.51) were identified, resulting in a total sample size of 1,155,960 surgical fractures. Analysis of the fractures based on additional specificity,ie subcapital versus midcervical versus basicervical, was not carried out because >90% of femoral neck fractures in the database were coded as “unspecified” or “other” (ICD9 CM 820.00 and 820.09, respectively).
Comorbidity burden was quantified using Elixhauser coding algorithms as previously described.31 The Elixhauser comorbidity measure is a model consisting of 31 conditions and has recently been identified as a better predictor of mortality in patients undergoing orthopedic procedures when compared with the Charlson Comorbidity Index.31 Dichotomous variables for each Elixhauser comorbidity were created, and χ2 tests were utilized to assess the association between each comorbidity and mortality. The weighted Elixhauser score for each statistically significant comorbidity was calculated as described by van Walraven and colleagues.32 The Elixhauser comorbidity score was then calculated for each patient by summing the individual weights of all comorbidities. Postoperative adverse events were determined using the complication-screening-package as previously described.33
All adverse events were categorized into 3 categories, including general medical complications, mechanical complications, and surgical complications. All adverse events recorded in the NHDS database are events that occurred during a single hospitalization. Therefore, it does not take into account adverse events that occurred after discharge, and, for example, mortality refers to postoperative mortality that occurs prior to discharge. The study period comprised data captured from 1990 to 2007, and 3 groups were generated from this time period to better characterize patients throughout the large study time frame. Group 1 comprised patients who underwent surgical management of femoral neck fractures from 1990 to 1995, group 2 consisted of patients treated from 1996 to 2001, and group 3 included patients treated from 2002 to 2007.
Categorical data were analyzed using the χ2 test, and continuous data were analyzed by the independent-samples t test and ANOVA. Multivariable binary logistic regression analyses were performed to assess the contributions of individual comorbidities to mortality, adverse events, and nonroutine discharge. Elixhauser comorbidities with a P value of < .10 in the bivariate analysis and presenting in at least 0.2% of the population were included in the logistic regression.31 Odds ratios and confidence intervals were calculated to assess the association between comorbidities and our dichotomous variables. A P value of < .001 defined statistical significance.33 Statistical analysis was conducted using SPSS version 21 (IBM).
Our query demonstrated a total of 1,155,960 patients who underwent surgical fixation of femoral neck fractures (Table 1). The most commonly used treatment modality was HA (75%), followed by ORIF (18%) and later by THA (7%). The majority of patients were females in each treatment group. Patients’ age varied according to treatment group, with patients undergoing HA having a mean age of 81.0 ± 9.0 years, patients undergoing ORIF having a mean age of 75.0 ± 17.0 years, and those undergoing THA having a mean age of 79.0 ± 10.0 years (P < .001). The majority of patients were ≥80 years in all treatment groups, but the ORIF group had the greatest proportion of patients <65 years (P < .001). Among patients undergoing HA, 62.4% were ≥80 years, while the ORIF and HA groups consisted of 48.6% and 51.5% of patients in that same age group, respectively.
There was a significant change in the distributions of the procedures performed according to age group over time. Patients >80 years continued to undergo primarily HA, with an increase from 74.4% during 1990 to 1995 up to 84.6% during the 2002 to 2007 period and a concomitant decrease in ORIF from 18.3% to 10.6% during the same time period in this age group. Surgical trends in patients 65 to 79 years demonstrated a significant decrease in management with ORIF from 19.1% in 1990 to 1995 to 16.8% in the 2002 to 2007 cohort (P < .001 for all, Table 2). There was an increase in the use of HA from 71.9% during the 1990 to 1995 period to 75.5% during the final study period (Table 2, Figure 1). The use of THA for all age groups decreased between 1990 and 2007, except for the 50- to 64-year-old group where THA utilization remained constant.
Management patterns in patients 50 to 64 years varied throughout the analysis and demonstrated the following trend: treatment with HA remained the most common technique used but varied slightly from 59.7% during 1990 to 1995 to 60.3% during 2002 to2007 (P < .001, Table 2). The second most common treatment used was ORIF, which decreased from 32.2% to 31.5% (P < .001, Table 2). The use of THA varied significantly from 8.2% among those managed during 1990 to 1995 to 11.7% during 1996 to 2001 but later declined to the initial 8.2% (P < .001, Table 2).
Analysis of patients ≤49 years demonstrated that ORIF was the preferred technique, which experienced a growth from 63.9% during 1990 to 1995 to 81.4% during the 2002 to 2007 period (P < .001, Table 2). A decreased use in THA was observed from 2.0% in the initial period to 0.6% in the final period (P < .001, Table 2). Use of HA decreased from 34.0% in 1990 to 1995 to 18.0% in 2002 to 2007 (P < .001, Table 2).
Length of Stay
Mean number of in-hospital days decreased throughout the study period for all treatment techniques. During the 1990 to 1995 study period, patients who underwent ORIF had a mean LOS of 8 ± 7 days, which decreased (P < .001, Table 2) to 6 ± 3 days in 1996 to 2001 and remained constant during 2002 to 2007 (mean 6 ± 4 days). This decrease in LOS was also observed in patients who underwent THA (P < .001, Table 2), who initially had a mean LOS of 11 ± 7 days during 1990 to 1995, which later decreased to 7 ± 5 days for the remainder of the study. The LOS for patients who underwent HA also decreased (P < .001, Table 2), which initially was reported to be 11 ± 11 days during 1990 to 1995, decreasing to 7 ±7 days in 1996–2001 and later to 6 ± 4 days in 2002 to 2007.
The Elixhauser Comorbidity Index varied significantly among groups over time (P < .001, Table 2). Overall mean Elixhauser Comorbidity Index score per procedure type is provided in Table 1, with HA patients having the highest score (-0.15 ± 13.09, p<.001).
Analysis of the preoperative comorbidities demonstrated significant differences among each surgical treatment group (P < .001 for all, Table 3). The most common comorbidities in patients who underwent HA were uncomplicated hypertension (33.2%), fluid/electrolyte disorders (17.4%), chronic pulmonary disease (14.9%), and congestive heart failure (13.7%). The most common comorbidities in the ORIF group were uncomplicated hypertension (30.8%), fluid/electrolyte disorders (14.5%), chronic pulmonary disease (14.0%), and uncomplicated diabetes (10.9%). Patients treated with THA had most commonly uncomplicated hypertension (30.1%), fluid/electrolyte disorders (17.2%), uncomplicated diabetes (15.5%), and chronic pulmonary disease (14.4%). The prevalence of comorbidities is displayed in Table 3.
Mortality varied significantly, being lowest in those who underwent ORIF (0.8%), followed those who underwent THA (1.8%), and HA (2.6%) (P < .001, Table 1).
The majority of patients in each group were discharged to long-term rehabilitation facilities, including 53.0% of those treated with HA, 40.4% of those treated with ORIF, and 44.3% of patients treated with THA. The second most common discharge location was home, which included 14.8% of patients who underwent HA, 32.2% of patients treated with ORIF, and 20.8% of those who underwent THA. Table 3 demonstrates the details of the discharge settings.
Mortality analysis over time demonstrated a significant decrease in each treatment group (P < .001). Mortality in the ORIF group decreased from 1.2% during 1990 to 1995 to 0.8% in 2002 to 2007. Mortality in the THA group also decreased significantly from 0.8% during 1990 to 1995 to 0.5% during the 2002 to 2007 time period. Patients who underwent HA also exhibited a decrease in mortality rate from 3.3% during 1990 to 1995 to 2.2% during 2002 to 2007 (P < .001, Table 4, Figure 2).
General Adverse Events
There was a significant difference (P < .001) in the percentage of adverse events experienced, the maximum being observed in the THA group (41.0%), followed by the HA group (37.9%) and trailed by the ORIF group (20.3%, (P < .001, Table 1). The prevalence of adverse events is detailed in Table 5.
Patients who underwent THA had the highest rate of any adverse event, LOS, and transfusion rate (Table 1 and Table 5).
The prevalence of postoperative pneumonia was highest in the HA group (3.4%), followed by the ORIF group (2.9%), and the THA group (2.6%) (P < .001, Table 5). There was also a significant difference in rates of intubation, pulmonary insufficiency, acute renal failure, pulmonary embolism, acute myocardial infarction, induced mental disorder, and deep venous thrombosis (P < .001 for all, Table 5).
Surgery-related Adverse Events
Surgery-related outcomes over the entire study period were significantly different according to the type of procedure performed (P < .001, Table 5). Patients who underwent HA had the highest rate of acute postoperative anemia (20.2%), followed by those who underwent THA (19.7%), and ORIF (10.2%). Postoperative bleeding rates also varied significantly, with 1.2% in the HA group, followed by 1.0% in the ORIF group and 0.4% in the THA group (P < .001, Table 5). Acute postoperative infection rates also varied significantly, with the highest rate being observed in the HA group (0.6%) compared to that in the THA and ORIF groups (both 0.3%) (P < .001, Table 5).
Table 6, Table 7, and Table 8 detail the results of regression analyses in patients with femoral neck fractures for individual risk factors associated with mortality, any adverse event, and nonroutine discharge to a short- or long-term rehabilitation facility, respectively. Increasing age (50–64 years, OR: 0.238; 65–79 years, OR: 1.762; and ≥80 years, OR: 2.700), THA (OR: 1.743), and HA (OR: 2.574) were found to be independent risk factors for mortality in the perioperative period (P < .001 for each, Table 6). Increasing age (50–64 years, OR: 1.888; 65–79 years, OR: 2.983; and ≥80 years, OR: 3.722), THA (OR: 2.489), and HA (OR: 2.098) were also found to be independent risk factors for any adverse event in the perioperative period (P < .001, Table 7). Age (50–64 years, OR: 1.662; 65–79 years, OR: 4.320; and ≥80 years, OR: 7.102) was the best predictor for nonroutine discharge to a short- or long-term rehabilitation facility (P < .001, Table 8).
Femoral neck fractures in the elderly population present a significant financial burden to the healthcare system.1-3,24,25 Consistent with previous epidemiological studies, our results show that the femoral neck fracture population has become older and has more medical comorbidities over the last 3 decades.27,28. Similarly, we also found that the rate of medical, surgical, and mechanical perioperative complications has increased in the same time period. Interestingly, the mortality rate has remained relatively similar.
Although patients undergoing HA for femoral neck fractures are older and have more medical comorbidities, we found that the rate of adverse events in the perioperative period for patients undergoing THA was higher than that in the HA group. Consistent with prior studies, patients who underwent THA had higher rates of blood transfusion, pulmonary embolism, and induced mental disorders.34 Multivariable regression analysis demonstrated that after controlling for age, medical comorbidity, and type of surgery performed, THA emerged as an independent risk factor for any adverse event in the perioperative period. Increased anesthesia time, reaming of the acetabulum, and increased complexity of surgery probably account for these changes.
Our study results are consistent with those of Jain and colleagues,30 which showed a decrease in utilization of THA for femoral neck fractures between 1990 and 2001. Since THA is generally indicated for younger, more active patients in relatively good health, this would explain why changes in baseline health in this cohort over the last 20 years would lead to fewer THA procedures being performed. Surgeons in the US may be finding there are fewer patients who are candidates for THA. Miller and colleagues5 reported conflicting results and showed an increase in THA utilization in this patient population. However, their study evaluated treatment trends based on data from the ABOS part II of recently graduated orthopedic surgeons and may not be an accurate representation of national practice trends in the US. The trend toward increased subspecialization may explain their findings. As the authors noted, although they found an increase in the use of THA for femoral neck fractures by new adult reconstruction surgeons, the percentage of new surgeons treating femoral neck fractures has declined.5
Our analysis showed very concrete trends in treatment management at the extremes of the age ranges. There were substantial increases in the use of ORIF for patients <50 years (from 63.9% in 1990–1995 to 81.4% in 2002–2007, P < .001) and in the use of HA for patients >80 years (from 74.4% in 1990–1995 to 84.6% in 2002–2007, P < .001). This trend parallels recent studies that purport better outcomes for young patients undergoing ORIF and elderly patients undergoing HA.30 Our analysis did not demonstrate a large shift in surgeon preference for treatment of patients between 50 and 80 years, although there was a statistically significant decrease in ORIF and THA usage and a reflective increase in HA usage in this population as well. The fact that there has not been as substantial a shift in treatment trends for this large age group is potentially due to the wide variations in comorbid conditions and the functionality that abounds in this age group.1
The limitations of the current study are those inherent with a retrospective database analysis. The reliance on accurate coding brings up a potential for error; however, it is unlikely that comorbidities and outcomes are undercoded as hospitals are incentivized to input values that increase the acuity and thus reimbursement for each hospital stay.35 The database also relies on the ICD-9 procedural and diagnostic codes, which are not as specific as the currently adopted ICD-10 codes; hence, we are unable to distinguish between different forms of internal fixation, for example intramedullary nailing versus dynamic hip screw. This also precludes us from including other critical data such as degree of fracture displacement, cemented versus uncemented implantation, surgical approach for arthroplasty, and functional outcomes of individual patients. Moreover, the database used, although the largest inpatient sample available for analysis, represents only approximately 20% of hospitals nationwide. In addition, as patients cannot be tracked over time within the database, we are limited to outcomes in the perioperative period captured in a single hospital stay and cannot identify readmissions. Finally, our analysis is limited to the years 1990 to 2007 because of an increase in the relative standard error of the database in more recent years. Although this results in data that are not the most current, we believe that this study provides valuable insight regarding the trends in surgical treatment and acute postoperative outcomes of these injuries that have hitherto not been reported. To limit the inherent biases and the limitations within this study, prospective, randomized studies with long-term follow-up comparing outcomes across modes of treatment are needed to definitively determine the optimum form of treatment for this fracture type.
This is the largest study to date reporting on national trends in the surgical treatment and outcomes of the femoral neck fracture population. Orthopedic surgeons performing THA should be aware that the femoral neck fracture population is changing and at higher risk for perioperative complications. The advent of bisphosphonate therapy has been suggested as a possible reason for the decrease in fragility fractures and why a larger proportion of the femoral neck fracture population is now >80 years.36,37 With an aging population at a higher risk for perioperative complications, clinicians must take special care in choosing the appropriate surgical intervention that will give their patients the best functional outcome while minimizing the risk of surgical complications. Orthopedic surgeons should weigh the added risk associated with THA in this population.