Assessing Impact of Orthopedic Subspecialty Training on Outcomes of Non-Arthroplasty Fixation of Geriatric Hip Fractures

Abstract

Purpose

In a single health system (NYU Langone Health) with established management protocols, our study examined differences in postoperative outcomes among patients treated with non-arthroplasty hip fracture fixation by orthopedic trauma (OT), adult reconstruction (AR), and sports medicine (SM) subspecialists.

Materials and Methods

We performed a retrospective review of 1,588 geriatric patients who underwent internal fixation for a hip fracture using a standard perioperative treatment protocol. Patients were assessed for in-hospital outcomes as well as readmissions and complications.

Results

Race/ethnicity, American Society of Anesthesiologists classification, ambulatory status, and energy of injury differed among subspecialty cohorts. The 30- and 90-day readmission rates were higher in the AR group than in the OT group (P=0.001, P=0.028). However, these differences were no longer significant after adjusting for baseline health, fracture type, and surgical construct. Mortality during admission, fracture-related infections, and revision surgery did not differ significantly across the 3 groups. Although the difference in the length of stay (LOS) (<1 day) across the groups was statistically significant (P=0.004), it was not clinically meaningful in adjusted analyses.

Conclusion

Within standardized care pathways, short-term outcomes did not meaningfully differ by fellowship background after adjustment. Surgical experience and consistency in care may be more influential, and future studies should account for surgeon case volume.

INTRODUCTION

With the growing prevalence of hip fractures, the worldwide incidence is projected to reach 4.5 million by 2050¹⁾. It is increasingly important to optimize health and surgical costs because of the serious morbidity and mortality risks associated with hip fracture injury as well as the cost-intensive acute care pathways currently in practice^2-5). Optimizing care factors, such as type of surgical intervention, type of anesthesia, and time to surgery, among others, have been thoroughly investigated^5-8). However, there is no clear consensus on the importance of training fellowships of orthopedic surgeons on hip fracture surgery outcomes.

Surgical fixation is a common intervention for all patients with intertrochanteric (IT) fractures or subtrochanteric (ST) fractures and relatively healthy patients with femoral neck fractures (FNF)^9,10). Although research has delved into the impact of optimum surgical fixation constructs and perioperative factors, there is a dearth of information on the impact of fellowship training of surgeons on surgical outcomes^9-11).

More specifically, no study to date has compared the impact of fellowship training of orthopedic surgeons on short-term outcomes of surgical fixation in IT, ST, and FNF fracture types of hip surgery¹¹⁾. The purpose of our study is to identify baseline-controlled short-term outcome differences based on the subspecialty fellowship training of the treating surgeon. We specifically compared patients treated by orthopedic trauma (OT), adult reconstruction (AR), and sports medicine (SM)-trained orthopedic surgeons.

MATERIALS AND METHODS

This retrospective study was approved by the Institutional Review Board of NYU Langone Health (approval No. i06-581 CR21), and the requirement for informed consent was waived. We reviewed data from patients treated for hip fractures between October 2014 and October 2024 at a single urban, university-based orthopedic department (NYU Langone Health).

We searched the database for all patients aged 65 years or older who sustained an Orthopedic Trauma Association type 31A, 31B, or 32A-C fracture treated with either closed reduction percutaneous pinning (CRPP), sliding hip screw (SHS), intramedullary nail (IMN), or other open reduction and internal fixation constructs, such as locked plating, femoral neck systems, or other less common screw configurations^12-16).

We applied a standardized, algorithmic approach to care for all the patients¹⁷⁾. The baseline health demographics included age, sex, race or ethnicity, body mass index, Glasgow coma scale, age-unadjusted Charlson comorbidity index, pre-injury ambulatory status, American Society of Anesthesiologists (ASA) score, the STTGMA (Score for Trauma Triage in the Geriatric and Middle-Aged) score as well as a frailty measure¹⁸⁾. Patients were excluded if they were lost to follow-up (minimum six months). We grouped the patients into three cohorts based on the fellowship training of the surgeon who treated them: OT-trained surgeons, AR-trained surgeons, and SM-trained surgeons.

Our analysis included a total of 1,588 patients who underwent surgical repair for hip fractures. Nineteen AR surgeons, 11 OT surgeons, and 10 SM surgeons operated on the patients in this cohort. The experience of the surgeon (years post fellowship) was extracted from departmental credential records and compared across subspecialties using ANOVA. Most patients came through the emergency department and were treated by an attending on-call doctor at the time. In rare cases, patients were transferred to a specific attending doctor or presented to the outpatient office of a specific surgeon. Patients were assigned to specific faculty on the basis of the “on-call” schedule or a pre-existing relationship (patient-doctor). An important fact to note is that the complexity of the fracture pattern or the comorbidities of a patient did not dictate how a surgeon was assigned to a specific case. Under the direction of the attending surgeon, the surgical intervention was performed in collaboration with the resident training staff.

We conducted an initial analysis of baseline health status of the three groups and performed an in-depth comparison of surgical outcomes. The in-hospital variables analyzed included length of stay (LOS) (days), in-hospital death, minor and major in-hospital complications. Minor in-hospital complications included deep vein thrombosis, urinary tract infection, anemia, and decubitus ulcer, among other less frequent complications. Major complications included sepsis, myocardial infarction, pulmonary embolism, pneumonia, stroke, major wound dehiscence, major blood loss, and multisystem organ failure. We also analyzed long-term outcomes, including healing rate, fracture-related infection (FRI), 30- and 90-day readmission, and 30-day mortality. Recognizing the potential for selection bias, we analyzed exploratory radiographic healing in patients with ≥6-month postoperative hip radiographs (n=616; 38.8% of the cohort). Nonunion, fixation failure, and necrosis were some of the reasons patients were considered unhealed at follow-up.

ANOVA was used to assess continuous variables and reported as mean±standard deviation. Chi-square analysis was used to analyze categorical variables. Following the initial analyses of baseline health characteristics and outcome variables, we performed logistic regression for all outcome variables that demonstrated statistical significance, while adjusting for baseline health, fracture, and surgery variables that were also found to be significant. Specifically, regression analyses had covariates of fracture classification, surgery type, ASA, ambulatory status, race or ethnicity, and energy. We also analyzed the LOS using an ANCOVA with the same control covariates. The significance level (alpha) was set at 0.05. IBM SPSS Statistics software (ver. 28; IBM Corp.) was used for all statistical analyses.

RESULTS

Univariate analyses of baseline health factors demonstrated that several variables differed between the subspecialty groups. In terms of race or ethnicity, the OT cohort had a higher distribution of White–non-Hispanic people than those in the SM cohort (83.9% vs. 77.0%, P<0.001). The OT group had a smaller distribution of Asian patients than those in the AR cohort (5.9% vs. 10.2%, P<0.001) (Table 1). The OT cohort also had a significantly smaller distribution of ‘unknown or other’ race classifications than those in both the AR and SM cohorts (3.4% vs. 7.7% vs. 8.9%, P<0.001) (Table 1). When scrutinizing baseline health metrics, the OT group had a significantly lower ASA score average than those of both AR and SM (2.73 vs. 2.99 vs. 2.97, P<0.001) (Table 1). Finally, the OT group had a larger distribution of baseline community ambulators than those of both the other cohorts (73.2% vs. 63.6% vs. 63.3%), a smaller distribution of household ambulators than those of the SM group (24.9% vs. 34.3%), and a smaller distribution of non-ambulators than those of the AR group (1.9% vs. 5.2%, P<0.001) (Table 1).

Table 1.

Demographic and Baseline Health Comparisons among Subspecialty Cohorts

Variable	Orthopedic trauma (n=977)	Adult reconstruction (n=363)	Sports medicine (n=248)	P-value
Age (yr)	82.75±8.51	83.11±8.10	84.15±8.43	0.063
Sex				0.528
Female	703 (72.0)	271 (74.7)	176 (71.0)
Male	274 (28.0)	92 (25.3)	72 (29.0)
Race/ethnicity				<0.001*
White: non-Hispanic	820 (83.9)*	285 (78.5)	191 (77.0)*
Black	29 (3.0)	8 (2.2)	4 (1.6)
White: Hispanic	37 (3.8)*	5 (1.4)*	6 (2.4)
Asian	58 (5.9)*	37 (10.2)*	25 (10.1)
Unknown/other	33 (3.4)*	28 (7.7)	22 (8.9)
Body mass index (kg/m2)	24.04±4.83	24.46±4.95	24.37±4.93	0.322
Glasgow coma score	14.91±0.36	14.85±0.65	14.86±0.55	0.056
Charlson comorbidity index	1.32±1.53	1.23±1.29	1.26±1.53	0.517
ASA score	2.73±0.68*	2.99±0.66	2.97±0.61	<0.001*
STTGMA	0.0106±0.0265	0.0126±0.0306	0.0132±0.0396	0.301
Ambulatory status				<0.001*
Community	715 (73.2)*	231 (63.6)	157 (63.3)
Household	243 (24.9)*	113 (31.1)	85 (34.3)*
Non-ambulatory	19 (1.9)*	19 (5.2)*	6 (2.4)
Low-energy	961 (98.4)	351 (96.7)*	247 (99.6)*	0.024*

Values are presented as mean±standard deviation or number (%).

ASA: American Society of Anesthesiologists, STTGMA: Score for Trauma Triage in the Geriatric and Middle-Aged.

*P<0.05.

Regarding fracture characteristics and surgical procedures, the AR group treated fewer low-energy mechanisms of injury than those treated by the SM group (96.7% vs. 99.6%, P=0.024) (Table 1).

The AR group also had more IT fracture patterns (31A) and fewer FNFs (31B) than those of the OT group (IT: 81.3% vs. 73.6%, FNF: 16.5% vs. 23.7%; P=0.045). We did not find any difference in the proportions of ST fractures between the two groups. Additionally, we did not find any difference in the time of arrival at the hospital to the start of surgery between the three groups (OT: 1.13 days, AR: 1.22 days, SM: 1.28 days; P=0.288). When comparing types of surgical fixation, there were more CRPPs in the OT group than in both AR and SM groups (CRPP: 14.6% vs. 9.4% vs. 8.9%; P<0.001) and fewer IMNs in the OT group than in the AR group (IMN: 75.8% vs. 83.5%; P<0.001). Significantly more SHS was performed in the OT group than in the AR group (4.6% vs. 1.1%; P<0.001). We did not observe a significant difference in post-fellowship surgeon experience among the cohorts, with 13.9±11.5 years for AR, 14.1±10.3 years for OT, and 12.2±11.6 years for SM (P=0.903).

When we compared short-term outcomes following surgery, we observed that patients treated by AR surgeons had a higher frequency of 30- and 90-day readmissions than those treated by the OT group (P=0.001, P=0.028) (Table 2). We performed logistic regression analysis of both the short-term outcomes with covariates of fracture classification, surgery type, ASA, ambulatory status, race or ethnicity, and low energy mechanisms of injury. There was no statistical significance for both the outcomes in the logistic regression analysis (30-day readmission: odds ratio [OR]=0.62, 95% confidence interval [CI] 0.38-1.02, P=0.058; 90-day readmission: OR=1.21, 95% CI 0.83-1.76, P=0.32). Finally, univariate analyses of the subspecialty groups revealed that they did not differ significantly in mortality during admission (P=0.974), FRIs (P=0.389), and length of hospital stay (P=0.471) (Table 2). Although the rates of revision surgery differed numerically across the three groups (OT: 10, AR: 1, SM: 0), the analysis was likely underpowered to demonstrate significance (P=0.122). The subspecialty groups demonstrated a small yet significant difference of adjusted mean LOS (P=0.004) in ANCOVA analysis that adjusts for covariates of fracture classification, surgery type, ASA, ambulatory status, race or ethnicity, and energy. Adjusted mean LOS was 6.4 days for OT, 5.6 days for AR, and 5.7 days for SM. However, the differences among groups were less than one day and unlikely to be clinically significant. There were no significant differences in the exploratory healing rates, though these data were only available in 38.8% of the cohort (P=0.650) (Table 3).

Table 2.

Short-Term Outcomes of Hip Fixation Surgery among the Three Subspecialty Cohorts

	Orthopedic trauma (n=977)	Adult reconstruction (n=363)	Sports medicine (n=248)	P-value
Length of stay (day)	6.21±4.90	5.87±3.76	6.22±4.61	0.471
In-hospital death	12 (1.2)	5 (1.4)	3 (1.2)	0.974
ICU admission	136 (13.9)	64 (17.6)	45 (18.1)	0.108
Minor complication incidence	352 (36.0)	127 (35.0)	98 (39.5)	0.494
Major complication incidence	93 (9.5)	31 (8.5)	21 (8.5)	0.794
30-Day readmission	60 (6.1)*	43 (11.8)*	25 (10.1)	0.001*
90-Day readmission	98 (10.0)*	54 (14.9)*	34 (13.7)	0.028*
Infection	7 (0.7)	1 (0.3)	3 (1.2)	0.389
Revision surgery	10 (1.0)	1 (0.3)	0 (0.0)	0.122
30-Day mortality	27 (2.8)	12 (3.3)	3 (1.2)	0.265

Values are presented as mean±standard deviation or number (%).

Means analyzed with ANOVA; frequencies analyzed with chi-squared test.

ICU: intensive care unit.

*P<0.05.

Table 3.

Exploratory Healing Rates for the Three Cohorts

	Orthopedic trauma (n=414)	Adult reconstruction (n=136)	Sports medicine (n=66)	P-value
Healed	399 (96.4)	132 (97.1)	65 (98.5)	0.650

Values are presented as number (%).

Subjects were excluded if there was not an X-ray at least six months postoperatively (38.8% of cohort included).

DISCUSSION

Our study found no significant clinical differences in the outcomes of the treatment of patients with hip fracture by the three orthopedic subspecialties—OT, AR, and SM. Although readmissions were significant in univariate analyses, logistic regression demonstrated that these differences were more likely due to patient cohort differences rather than surgeon subspecialty. Although the LOS differed statistically, the mean difference was less than a day across the subspecialties. This suggested there was minimal clinical significance to this observation. Univariate analyses revealed no significance in all the other outcomes. After adjusting for baseline health, the results of our study suggest no detectable short-term differences by subspecialty within a standardized care pathway. The hip fracture surgical fixation is a core component of all orthopedic training, which potentially limits the additional value of fellowship specialization for these procedures due to comparable outcomes¹⁹⁾. Furthermore, many procedures are executed in academic environments where residents rotate across subspecialties²⁰⁾. These trainees may incorporate techniques learned from trauma specialists, when assisting in cases staffed by SM or arthroplasty faculty. This could promote consistency in care, regardless of the surgical subspecialty. In tandem with standardized treatment protocols, the consistency in care likely reduces strategic variation among surgeons of different subspecialties²¹⁾. At approximately 90%, orthopedics has one of the highest fellowship participation rates across all specialties²²⁾. Hence, it is important to examine the utility of this extra training for common procedures²³⁾.

The current data are in line with previous research that found no difference in surgical fixation outcomes with OT surgeons versus non-OT surgeons¹¹⁾. Compared to the previous study, the current study included a large cohort with more operative fixation types. This supports the evidence that fellowship training does not seem to influence patient outcomes following hip fracture operative fixation. In another analysis that compared OT, AR, and ‘other’ orthopedic surgeons, there was no difference in surgical complication risk among the cohorts²¹⁾. The authors found that increased case volume rather than subspecialty was associated with more favorable surgical outcomes²¹⁾.

This emphasizes the importance of practical experience over formal fellowship training. Similarly, both Forte et al.²⁴⁾ and Ames et al.²⁵⁾ demonstrated that experience of surgeons is likely more important for hip arthroplasty outcomes. These conclusions have been replicated for other common orthopedic procedures, including knee arthroplasty and scoliosis^26,27). Surgeons with high case volume may develop the ability to take rapid decisions and demonstrate intraoperative efficiency, both of which could lead to better outcomes. The findings of our study also add to growing evidence that fellowship training is not a prerequisite for surgical personnel to achieve the best outcomes in their patients with hip fracture^28,29). Although the study analyses adjusted for numerous patient and surgical variables, we did not measure surgeon-specific case volume. Similar work experience and case volume rather than a lack of subspecialty effect may explain comparable outcomes across subspecialties.

It appears that orthopedic residents receive adequate training to achieve optimal outcomes in surgical fixation procedures. This widespread proficiency across subspecialties may enable more expedient surgery following admission, aligning with one of the prime tenets of hip fracture care–reducing the time to surgery as much as possible³⁰⁾. Given the growing need to manage hip fractures in an aging population, the findings of our study may have implications for workforce planning and surgical training. More specifically, these results may support increased flexibility of staffing to schedule trauma calls in order to prioritize timing over fellowship training.

Our study has several limitations. We did not assess long-term outcomes, such as functional recovery. Since healing outcomes were only available for 38.8% of the cohort, these results have to be interpreted with caution due to the possibility of selection bias that could limit the generalizability of this specific outcome. Although radiographs of some patients revealed they had healed within six months, they were excluded if a six-month follow-up radiograph was unavailable. Given the elderly population in our cohort, a possible reason for incomplete follow-ups is that some stop returning once they recover functionality, are pain-free, or find the logistics to return become burdensome. Thus, the subset with documented fracture healing outcomes may not fully represent the cohort. Therefore, our estimates are best considered exploratory rather than representative of the overall sample. Furthermore, we did not analyze the duration of surgery among the cohorts. Although the experience of surgeons was compared, the specific case volume of each individual surgeon was not scrutinized. Since case volumes of surgeons are associated with outcomes, its omission may confound subspecialty comparisons and possible differences. This lack of granularity could limit the precision of the presented analyses. Although multivariate tests accounted for broad fracture classification, we did not utilize more specific fracture classifications that accounted for factors such as displacement. Granular fracture morphology is known to influence complications and healing. Apparent similarities may be exaggerated if these details are not captured. This raises the possibility of overgeneralized analyses. The absence of bone mineral density comparisons among the groups is another limitation; however, the study did control for a broad range of health-related variables, which may help mitigate this drawback. Finally, the retrospective nature of the study limits the ability to establish causation.

Future studies should incorporate more prospective data from a large population to allow analyses to be stratified by fracture and surgery type. This will help facilitate the comparison of long-term outcomes such as functionality.

CONCLUSION

Adjusted analyses eliminated subspecialty differences in readmission rates. There were no unadjusted differences in other short-term outcomes among the three groups. Surgical outcomes within standardized care pathways may depend more on surgical experience and consistency in care rather than on fellowship background. From our observations, we emphasize the need for future research that accounts for case volume of surgeons.