Racial Disparities in Surgical Treatment Choice for Displaced Femoral Neck Fractures

Abstract

Background:

Femoral neck fractures (AO-OTA 31B) (FNFs) are prevalent among the elderly, and the choice between hemiarthroplasty (HA) and total hip arthroplasty (THA) remains debated. Given previous evidence of treatment disparities in FNFs, we investigated racial differences in THA utilization.

Methods:

Using the National Inpatient Sample (NIS) database, a representative sample of inpatient admissions, we identified patients aged ≥65 years who underwent HA or THA for FNFs from 2000 to 2020. Records were extracted using ICD9/ICD10 codes for intracapsular FNFs, excluding basicervical and peritrochanteric fractures, and procedure codes for THA and HA were used to create the cohorts. The primary outcome was treatment with either HA or THA. Covariates include hospital characteristics, demographics, socioeconomic factors, and medical comorbidities. Logistic regression analysis evaluated the relationship between race and treatment with THA or HA.

Results:

Bivariate analyses showed similar rates of HA across races and no statistically significant differences (Hispanic 89.6%, Asian 89.4%, Black 89.4%, White 88.7%; p = 0.056). Patient factors, including age, sex, and insurance, influenced treatment choice (all p < 0.0001), as did hospital-related factors such as location, region, and bed size (all p < 0.0001). Statistically significant comorbidities included acquired immune deficiency syndrome (AIDS), solid tumor without metastasis, alcohol abuse, metastatic cancer, depression, diabetes, hypertension, chronic obstructive pulmonary disease (COPD), obesity, hypothyroidism, and peripheral vascular disease (all p < 0.05).When controlling for patient- and hospital-related variables and significant comorbidities in our multivariable logistic regression model predicting THA, Black (OR 0.83, 95% CI 0.75-0.93; p = 0.001) and Hispanic (OR 0.76, 95% CI 0.67-0.86; p < 0.001) patients had significantly lower odds of undergoing THA compared with patients, whereas differences for Asian/Pacific Islander patients were not statistically significant.

Conclusion:

Variations in patient age, insurance status, and hospital-related variables influenced HA versus THA treatment. Racial disparities persisted after controlling these factors, underscoring the need to closely evaluate potential inequities to ensure equitable access to appropriate care.

Level of Evidence:

Level III, prognostic. See Instructions for Authors for a complete description of levels of evidence.

Introduction

Femoral neck fractures (FNFs) are common in older adults, and the optimal treatment for displaced fractures remains controversial^1,2. While arthroplasty is generally accepted, the choice between total hip arthroplasty (THA) and hemiarthroplasty (HA) remains unclear. Recent studies, including the HEALTH trial and the study by Tate et al., found no major outcome differences between THA and HA although some data suggest that THA may offer better functional outcomes at the cost of increased dislocation risk^2-7. Despite this uncertainty, THA utilization rate has significantly increased since 1991. Indications remain subjective, often based on the physician’s assessment of functional status, health, and life expectancy, introducing potential for bias^1,8,9.

Although racial disparities in treating FNFs have been reported, few studies have specifically examined disparities in arthroplasty type and most have relied on controlled clinical trials or limited-year registries^10,11. These studies may not reflect national trends or evolving practice patterns. To address this, we analyzed 20 years of data from the National Inpatient Sample (NIS), a nationally representative discharge database, to evaluate racial disparities in the surgical management of FNFs. We also assessed how these disparities, if any, changed over time.

Methods

The National Inpatient Sample (NIS), sponsored by the Agency for Healthcare Research and Quality as part of the Healthcare Cost and Utilization Project (HCUP), was used for this analysis. The NIS is the largest publicly available all-payer inpatient database in the United States and contains discharge-level data from a nationally representative sample of US community hospitals. HCUP-recommended sampling weights were applied to generate national estimates¹². The cohort included patients 65 years and older who underwent surgery for FNFs between 2000 and 2020. We used ICD-9 and ICD-10 codes to identify intracapsular FNFs. Basicervical and peritrochanteric fractures were excluded. Procedure codes for THA and HA defined patient cohorts. Because ICD-9-CM does not specify displacement, all FNFs treated with arthroplasty were included during ICD-9-CM years (Supplemental Table I). Institutional review board approval was not required because patient data were deidentified.

The primary outcome was the surgical treatment type: THA or HA. Race/ethnicity (White, Black, Hispanic, Asian) was extracted from hospital-reported discharge data. Per HCUP guidelines, Hispanic ethnicity takes precedence over race. Covariates included patient (age, sex, insurance status), hospital characteristics (type, bed size, location), and comorbidities based on the Elixhauser Comorbidity Index (ECI)^13,14. HCUP’s Elixhauser Comorbidity Software (ECS), a proprietary tool developed by AHRQ, was used to identify chronic conditions from ICD-coded administrative data. Comorbidity data were unavailable for years 2000, 2001, and 2016 to 2018 because of system-level limitations in those years, and these cases were excluded from comorbidity-adjusted analyses.

Descriptive statistics summarized demographics. Bivariate analyses compared racial groups using chi-squared tests and one-way ANOVAs or Kruskal–Wallis tests for non-normally distributed variables. We evaluated normality using the Shapiro–Wilk test. Weighted multivariable logistic regression models were used to evaluate the association between race and treatment selection. A main effect model and an interaction model predicting the outcome of receiving THA, including race and year, were conducted using data from the years in which comorbidities were reported (2002-2010, 2012-2015, 2019, 2020). Year of admission was treated as categorical to allow for nonlinear changes over time. A global race × year interaction was evaluated using a likelihood ratio test comparing nested weighted multivariable logistic regression models with and without the interaction term. To focus on racial disparities in treatment selection, models predicting THA were adjusted for statistically significant-, hospital-, and comorbidity-related variables identified in bivariate analyses. Fig. 1 presents the STROBE flow diagram. All analyses were conducted using R (R Core Team, Vienna, Austria) and accounted for the NIS sampling design¹⁵. HCUP-recommended discharge weights were applied to generate nationally representative estimates¹⁴. This study was not funded.

Figure 1.

STROBE flow diagram outlining the study design for investigating racial disparities in the surgical treatment (THA vs. HA) of displaced FNFs among elderly patients. The diagram shows the inclusion criteria, data sources (NIS database from 2000 to 2020), and covariates (patient and hospital-related factors) and identifies years without complete comorbidity data. HA = hemiarthroplasty, NIS = National Inpatient Sample, THA = total hip arthroplasty.

Results

Over the 20-year study period, THA utilization increased from 8.1% in 2000 to 17.0% in 2020 within our population. THA rates rose, and HA declined across all racial groups, with the greatest increase among Asian/Pacific Islander patients followed by White, Black, and Hispanic patients. Year-to-year fluctuations were more pronounced in smaller subgroups although differences generally remained within 3 to 5% points. THA utilization increased within our population from 8.1% to 17% during this period (Supplemental Table II).

Bivariate analysis found similar rates of HA across all races and no significant differences (Hispanic 89.6%, Asian 89.4%, Black 89.4%, White 88.7%; p = 0.056; Supplementary Table III). However, significant differences were observed across all patient-related factors (age, sex, insurance type; all p < 0.001) and hospital-related factors (location, region, bed size; all p < 0.001). In analyses limited to years with available comorbidity data (excluding 2000, 2001, and 2016-2018), several comorbidities, including substance abuse disorder, metastatic cancer, solid tumor without metastasis, depression, diabetes without chronic complications, hypertension, COPD, obesity, peripheral vascular disease, and hypothyroidism, were significantly associated with the choice of procedure type (p < 0.05; Table I; Supplemental IV).

TABLE I.

Patient and Hospital Characteristics by Treatment with Total Hip Arthroplasty Versus Hemiarthroplasty, 2000–2020

	HA (n = 935,917)	THA (n = 118, 305)	p
Patient Information
Age, y (mean)	82.39	78.36	<0.0001
Sex (%)
Male	28%	30%
Female	72%	70%	<0.001
Race			0.056
White	90%	91%
Black	4%	4%
Hispanic	4%	4%
Asian	1%	1%
Expected primary payer			<0.0001
Medicare	92%	90%
Medicaid	1%	1%
Private insurance	5%	7%
Self-pay	0%	1%
No charge	0%	0%
Other	1%	2%
Hospital information
Location			<0.0001
Rural	15%	13%
Urban, no teaching	41%	36%
Urban, teaching	45%	52%
Region			<0.0001
Northeast	21%	19%
Mid-west	20%	19%
South	41%	44%
West	18%	17%
Bed size			<0.001
Small	17%	19%
Medium	29%	29%
Large	54%	52%

HA = hemiarthroplasty and THA = total hip arthroplasty.

Values are presented as mean (standard deviation) or number (%), unless otherwise indicated. p values were calculated using the chi-square or Fisher's exact test for categorical variables and the independent-sample t test for continuous variables. p < 0.05 is considered statistically significant.

In the multivariable logistic regression model, Black (OR 0.83, CI 95% 0.75-0.93; p = 0.001) and Hispanic (OR 0.76, CI 95% 0.67-0.86; p < 0.001) patients had significantly lower odds of undergoing THA compared with White patients. No significant difference was found for Asian/Pacific Islander patients (OR 0.87, 95% CI 0.74-1.02; p = 0.093). Complete multivariable regression results are reported in Supplemental Tables V, VI, and VIII. These findings suggest that while THA utilization increased across all racial groups over time (Table II), Black and Hispanic patients remained significantly less likely to receive THA after adjusting for patient characteristics, hospital factors, and comorbidities.

TABLE II.

THA Utilization by Race in 2000 and 2020

Race	THA Rate 2000 (%)	THA Rate 2020 (%)	Absolute Change (%)
White	8.0	17.5	+9.5
Black	8.7	11.7	+3.0
Hispanic	11.5	12.3	+0.8
Asian/Pacific Islander	3.9	17.0	+13.1
Overall	8.1	17.0	+8.9

THA = total hip arthroplasty.

Values are presented as percentages. Absolute change represents the difference in THA utilization between 2000 and 2020.

The global race × year interaction was statistically significant (likelihood ratio test = 86.9, p = 0.002), indicating that temporal trends in THA utilization differed across racial groups. When including a race-by-year interaction, disparities in THA use persisted. Over time, THA likelihood increased more rapidly for White and Asian patients (Fig. 2). By later years, THA rates were highest in White, with Asian groups being the second highest. As shown in Fig. 3, THA utilization increased across all racial groups from 2000 to 2020, with White patients demonstrating the largest absolute rise, followed by Asian, Black, and Hispanic patients.

Fig. 2.

The figure shows predicted probabilities based on the multivariable logistic regression model, with the x-axis representing the year and the y-axis representing the predicted probability of receiving THA. The model was adjusted for relevant patient- and hospital-related covariates and includes years 2000 to 2020. Lines represent different racial groups: White (purple), Black (Green), Hispanic (Blue), and Asian/Pacific Islander (red). Shaded areas around the lines indicate the 95% confidence intervals for each racial group. THA = total hip arthroplasty.

Fig. 3.

Rates of THA among patients >65 years old from the NIS database between the years 2000 and 2020 stratified by race. THA utilization increased among all races, with White patients displaying the largest absolute increase. NIS = National Inpatient Sample and THA = total hip arthroplasty.

Discussion

After adjusting for medical, hospital, and socioeconomic factors, we found that Black and Hispanic patients had significantly lower odds of undergoing THA compared with White patients. In our analysis, THA utilization rose steadily across all racial groups over time. Notably, Asian patients began with lowest probability of receiving THA, but demonstrated faster increases over time, converging with White patients. These differences, though small, remained statistically significant, with uncertain clinical significance. In addition, racial minority subgroups compromised a smaller portion of the study population, potentially limiting estimate precision and increasing variability. Finally, our findings align with the previous literature showing increasing THA utilization for FNFs over the past 2 decades, alongside persistent racial disparities in access and treatment patterns⁹.

Historically, HA has been preferred for FNFs in the elderly because of high reoperation rates with internal fixation. However, recent trends show that greater use of primary THA—once reserved for younger, more active patients—has increased because of concerns about acetabular erosion with HA. Comparative studies and meta-analyses have shown that THA offers better functional outcomes and lower revision rates in select populations. However, randomized trials have shown a marginal benefit in functional scores^16-19.

Consistent with the previous literature, we found that even after adjustment for covariates, White and Asian patients were more likely to undergo THA than Black and Hispanic patients^10,20,21.

The choice between THA and HA for treating FNFs is multifaceted and involves several key considerations, including patient age, functional status, and life expectancy. THA is often favored for younger, more active patients and is believed to provide superior long-term functional outcomes and a lower risk of revision²². By contrast, HA may be preferred for elderly patients or those with significant comorbidities as the shorter operative time and reduced intraoperative blood loss make it a safer option for higher-risk individuals. HA is associated with a lower risk of dislocation compared with THA, which can be an important consideration in certain patient populations. However, HA has been associated with a higher rate of acetabular erosion over time, which may necessitate future interventions²³. Ultimately, the decision between HA and THA balances the patient's current health status, the expected postoperative quality of life, and the likelihood of future revision surgeries. By understanding these factors, surgeons can tailor their treatment approach to minimize the risks of complications and ensure optimal outcomes for each patient.

Insurance status also significantly influenced the treatment choice. Our bivariate analysis demonstrated a strong association between the insurance type and the likelihood of receiving HA, with HA being more common among patients with Medicaid or no insurance²⁴. Patients with private insurance or Medicare might have access to a broader range of hospitals and doctors, which may be more likely to have the resources and personnel to perform an urgent THA, as THA often requires more experience or even subspecialty training. Miller et al. found that surgeons’ expertise and the available resources at healthcare facilities can affect whether a more complex procedure like THA is feasible²⁵.

When we controlled patient and hospital factors, we found small but significant differences in treatment choice related to race. The relationship between race and medical decision-making is multifaceted, influenced by social and cultural factors that are difficult to capture in administrative data. We were only able to quantify some of those reasons in the current study design. While socioeconomic status (SES) and hospital characteristics were controlled for, unmeasured variables—such as variability in patient-doctor communication, health literacy, and cultural preferences—are challenging to quantify and likely have a significant impact. Furthermore, cultural factors and communication barriers can affect the physician-patient relationship. Previous studies have demonstrated that poor communication can negatively affect the patient's understanding of discussions regarding their healthcare, preventing them from actively participating in their care²⁶. In addition, cultural differences in expectations of patient autonomy and lack of qualified medical translators can profoundly affect communication, trust, decision-making, and the quality of care^27,28.

Implicit bias is another possible contributor. A systemic review by Maina et al. found that healthcare provider bias was linked to disparities in treatment recommendations, pain management, and empathy in simulated settings although evidence was mixed in real-world clinical care^29,30. Surveys show that orthopaedic surgeons exhibit levels of implicit bias similar to or greater than the general population, but such bias has not been shown to affect fracture care decisions³¹. However, evidence from nonorthopaedic fields suggests that implicit bias may influence care quality, with minority patients and women less likely to receive advanced treatments in conditions like cancer, heart failure, and myocardial infarction³².

While implicit bias is beyond this study’s scope, racial disparities have been reported across orthopaedic subspecialties. For instance, Schoenfeld et al. found that Black patients with lumbar degenerative disease were less likely to undergo surgery than White patients³³. Wick et al. reported that Black patients with cervical radiculopathy were less likely to receive disk replacements³⁴. Dangelmajer et al., in 2017, also found that older age, being Asian or Pacific Islander, and Medicaid payer status were associated with lower odds of receiving THA¹⁰. Treatment at teaching hospitals was linked to decreased likelihood of THA, whereas high-volume hospitals increased the likelihood, highlighting institutional variation in care. However, Dangelmajer’s analysis was limited to 2009 to 2010 and relied solely on ICD-9 codes, whereas our study spans 20 years and includes both ICD-9 and ICD-10 data^20,21,35.

Our study has several limitations. Although data from the years 2000, 2001, and 2016 to 2018 were included in the overall analysis, these years lacked full comorbidity files, which limited our ability to assess specific comorbidities. The transition to ICD-10 codes in 2015 to 2016 expanded the number of fracture-related diagnostic codes, which might have increased the number of cases reported in the later years of the data set. In addition, the NIS sampling redesign in 2012 changed weighting and hospital sampling methods, which may influence comparisons across years. In addition, the NIS does not include specific clinical details that may be crucial to surgical decision-making, such as acetabular osteoarthrosis, undiagnosed osteoporosis, hip dysplasia, or other radiography-related findings. These unmeasured factors may partly explain the observed differences, limiting our ability to attribute our findings solely to race. Because our cohort includes only patients who underwent arthroplasty, we cannot determine whether racial differences exist in the overall incidence of FNFs or in the likelihood of receiving arthroplasty versus alternative treatments. Furthermore, as this is an administrative database, we were limited to identifiable existing codes. Critical social factors such as the presence of language barrier or surgical decision making were not available, and the data set’s cross-sectional nature prevents conclusions about causality.

Conclusion

We identified variations in patient demographics, including age, socioeconomic factors such as insurance status, and hospital-related variables, between those treated with HA versus THA for FNFs. Despite controlling these patient- and hospital-level factors, we observed a small but statistically significant racial difference in the selection of THA for FNFs. However, unmeasured clinical, social, and provider-level factors not captured in administrative databases may also contribute to treatment selection and should be considered when interpreting these findings. These findings underscore the need to explore further disparities in treating FNFs to ensure equitable access to high-quality care.

Appendix

Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (https://links.lww.com/JBJSOA/B345). This content was not copyedited or verified by JBJS.