1 Introduction

Elective total knee arthroplasty (TKA) and total hip arthroplasty (THA) are among the most commonly performed orthopaedic procedures in the United States, with their utilization projected to increase by 129 % and 182 %, respectively, by 2030.^1,2 Although modern total joint arthroplasty (TJA) techniques are highly successful, orthopaedic surgeons must accommodate changing clinical and demographic considerations in their patient populations.^3–5 One of the most significant challenges in the United States is the rising prevalence of obesity, a major modifiable risk factor for degenerative osteoarthritis (OA).^6,7 There is a growing body of evidence demonstrating increased morbidity following TJA procedures in obese patients, especially due to wound infections.^8,9 Strikingly, obese patients may face double the risk of mortality within one year compared to patients with a normal body mass index (BMI).¹⁰ As the shift toward outpatient and ambulatory TJA continues, implementing risk assessment tools such as frailty measures may aid in surgical candidate selection in older patients with higher BMI.¹¹

Frailty is defined as a decreased physiologic reserve that increases with age, predisposing patients to increased vulnerability following stressors.¹² Frailty is a well-established predictor of adverse outcomes in multiple surgical settings.^11–13 In orthopaedic surgery, the 5-item Modified Frailty Index (mFI-5) is the most commonly used tool to predict postoperative outcomes.¹⁴ This index is conceptually based on the accumulation of deficits model developed by Rockwood et al..¹⁵ However, while the mFI-5 has been widely utilized, it has been criticized for primarily measuring multimorbidity rather than true frailty. In contrast, the more recently described Risk Analysis Index (RAI) is a weighted model that incorporates functional, social, cognitive, and nutritional domains in addition to comorbidities, making it a more comprehensive measure of frailty.^11–13,16 The RAI has been shown to have superior predictive value and discriminatory accuracy compared to the mFI-5 in various surgical procedures such as spinal fusion, lower extremity reconstruction, and brain tumor resection.^17–19

The aging population of the United States coupled with high rates of obesity present a unique obstacle for orthopaedic surgeons as the rates of elective TJA continue to rise.^20,21 Given the increased risk of mortality in this population, frailty assessment may serve as a valuable tool for surgical candidate selection and risk stratification. Therefore, this study aims to evaluate the utility of frailty as a predictor of 30-day mortality in obese older adults undergoing elective TJA.

2 Methods

2.1

2.1 Data source and patient selection criteria

Data was collected from the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) from the years 2015–2019. The NSQIP is a national database with risk-adjusted surgical outcomes data in the 30-day postoperative period.²¹ Institutional Review Board approval was not required as the NSQIP is a publicly available quality improvement database containing deidentified patient data. Patients included in this study were identified using the Current Procedural Terminology (CPT) codes 27130 and 27447 for primary THA and primary TKA respectively. The 10th revision of The International Classification of Diseases (ICD10) codes for OA of the hip (M16, M160–M165, and M169) and OA of the knee (M170–M175, and M179) were applied to only include patients undergoing elective TJA. Finally, only patients ≥65 years of age with a BMI of ≥30 kg/m² were included for the final analysis (Fig. 1).

Close

Fig. 1 CONSORT flow diagram depicting the inclusion process for study cohort.

Export to PPT

3 Results

3.2

3.2 Multivariate regression and ROC analysis

Multivariate regression was performed controlling for patient age, length of hospital stay, operative time, BMI, and THA or TKA to understand the predictive value of the RAI and mFI-5 for primary and secondary outcomes. Increasing frailty as measured by the mFI-5 (prefrail: OR 1.2, 95 % CI (0.8–1.8); frail: OR 2.1, 95 % CI (1.4–3.3); severely frail: OR 7.0, 95 % CI (3.7–13.2), p < 0.001) and RAI (prefrail: OR 2.2, 95 % CI (1.2–4.3); frail: OR 5.5, 95 % CI (2.3–13.4); severely frail: OR 6.1, 95 % CI (1.7–21.6), p < 0.001) was significantly predictive of mortality within 30-days following TJA, with greater odds for RAI prefrail and frail patients (Table 3). As quantified by ROC analysis, the RAI demonstrated superior discriminatory accuracy compared to the mFI-5 for predicting 30-day mortality (C-statistic: 0.70 vs 0.61, p < 0.001) (Fig. 2).

Table 3 Results of multivariate regression analysis controlling for age, length of hospital stay, operative time, BMI and type of total joint replacement, for prediction of 30-day mortality by the RAI or mFI-5. ¹P-value = 0.015, ²P-value = 0.005, ³P-value = 0.497, ⁴P-value = 0.001.

	Odds (95 % Confidence Interval)
	RAI				mFI-5
	Prefrail	Frail	Severely Frail	P-value	Prefrail	Frail	Severely Frail	P-value
Mortality	2.2 (1.2–4.3)1	5.5 (2.3–13.4)	6.1 (1.7–21.1)2	<0.001	1.2 (0.8–1.8)3	2.1 (1.4–3.3)4	7.0 (3.7–13.2)	<0.001
Age	1.1 (1.0–1.1)			<0.001	1.1 (1.1–1.1)			<0.001
LOS	1.1 (1.0–1.1)			<0.001	1.1 (1.0–1.1)			<0.001
Op Time	1.0 (1.0–1.0)			0.003	1.0 (1.0–1.0)			0.002
TJA Type	1.4 (1.0–1.8)			0.044	1.4 (1.0–1.9)			0.024
BMI	1.1 (1.0–1.1)			0.002	1.0 (1.0–1.1)			0.015

Close

Fig. 2 AUROC analysis demonstrating superior discrimination of the RAI compared to the mFI-5 in obese frail patients for mortality within 30-days of elective TJA (RAI C-statistic: 0.70 (0.696–0.700) vs. mFI-5 C-statistic: 0.61 (0.608–0.612); Delong's: p < 0.001).

Export to PPT

4 Discussion

As TJA becomes more frequently indicated in an aging and obese population, tools such as frailty assessment may aid orthopaedic surgeons in optimizing selection of surgical candidates. Our analysis of 169,065 patients using a national database evaluated the accuracy of the RAI and the mFI-5 in predicting 30-day mortality among obese patients aged 65 and older undergoing TJA. We found that frailty, as measured by both the RAI and mFI-5, is significantly associated with increased 30-day mortality, with the RAI demonstrating superior predictive value and discriminatory accuracy compared to the mFI-5.

Primary TJA is generally well tolerated, with mortality rates well below 1 % for both THA and TKA, consistent with our findings.^23,24 However, these low mortality rates are likely influenced by surgeons’ preoperative selection criteria.²⁵ Given that TJA is an elective procedure, extensive research has been conducted to identify risk factors for adverse outcomes, many of which are incorporated into both the RAI and mFI-5.^26–28 Notably, increasing age–the key component of frailty–has been reported to be an independent predictor of perioperative mortality for TJA, with rates reaching 1.5 % in octogenarians and 2.9 % in nonagenarians.^29,30 Additionally, obesity, which affects up to 40 % of the U.S. population, is a well-established risk factor for major complications such as infection and revision surgery.^31–33 Thus, older obese patients represent a particularly vulnerable, yet increasingly common patient population for TJA procedures.

Frailty has been widely studied in the orthopaedic literature as a predictor of increased morbidity and mortality following TJA.^{14,20,27,34–38} In a study of 39,806 patients, Shin et al. reported increasing mortality rates with greater frailty status, reaching 4.2 % and 1.95 % in the highest frailty tiers for THA and TKA, respectively.³⁹ In our study, we observed mortality rates of <1 % across all frailty tiers, which increased with the mFI-5 but not with higher RAI tiers. Although these are lower than previously reported mortality rates, this could potentially be a result of improving surgical technique and/or better optimized candidate selection.³³

Compared to frailty, the literature regarding the role of obesity in TJA for increasing mortality is more varied. While higher BMI is widely recognized as increasing morbidity and complications following TJA, there is recent data to support no difference or even a decreased risk of mortality in patients with high BMI.^40–42 Studies have sought to provide explanation for this “Obesity Paradox”, including protective factors such as increased metabolic reserve in adipose tissue, more intensive preoperative optimization, and confounding comorbidities.^43–47 Although this study did not specifically compare frailty's role in obese versus non-obese older patients, our multivariate analysis controlling for BMI suggests that frailty independently predicts mortality, regardless of BMI. These findings align with those of Owodunni et al. who studied cohorts of obese, older spine surgery patients and found that frailty remained a stronger predictor of mortality across increasing frailty tiers, irrespective of BMI.¹¹ Further, they found higher odds of mortality and superior predictive accuracy utilizing the RAI compared to the mFI-5 in spine patients. These results support our findings of more accurate risk stratification and 2.5-fold greater odds of mortality demonstrated in the RAI frail cohort.

There is a growing body of evidence supporting the predictive value of various proposed frailty measures, including the Frailty Deficit Index, Electronic Frailty Index, and Hospital Frailty Risk Score (HFRS), for not only increased mortality but also complications and readmission.^36,37 However, previously utilized frailty scales, such as the mFI (including its derivatives) and HFRS, have been criticized as measures of multimorbidity rather than true frailty. For example, the HFRS was specifically developed for use in hospitalized patients and might not fully consider patient-specific factors like functional ability and living conditions, which are particularly relevant in orthopaedic patients.^15,38 While our multivariate analysis supports the predictive utility of both the mFI-5 and RAI for our primary and secondary outcomes, our ROC analysis demonstrated that the RAI (AUC: 0.70) had greater discriminatory accuracy for 30-day mortality than the mFI-5 (AUC: 0.61). Further, both frailty measures exhibited poor discriminatory accuracy in identifying patients at risk of NHD, major complications and unplanned readmission. These findings are crucial as it supports the notion that although the mFI-5 is similar in predictive value to the RAI, it does not provide accurate risk stratification for mortality following TJA. Additionally, as the RAI is a more comprehensive frailty measure, the poor discrimination of both the RAI and mFI-5 for secondary outcomes suggests that frailty alone may not play a role in accurately predicting NHD, major complications or unplanned readmission. Interestingly, there is a paucity of literature examining the accuracy of frailty scales in TJA, with many studies simply reporting increased odds of mortality through multivariate regression analysis. Our findings are supported by previous studies in the neurosurgical literature which often demonstrate significant predictive value of the mFI-5, yet poor discriminatory accuracy compared to the RAI following ROC analysis.^48,49 Not only does the ROC curve analysis show statistical significance for RAI's superior discriminatory accuracy, but the RAI has also shown clinical significance that can warrant a change in healthcare practice.

In addressing the clinical utility and applicability of RAI, it is important to note the RAI has been effectively implemented in clinical settings for the purpose of preoperative frailty assessment.⁵⁰ This is highlighted by RAI's integration into electronic health record systems, such as Epic and Cerner, making it accessible to a broad range of clinicians across different environments.⁵⁰ Major United States hospital centers have already adopted its use into their clinical practice.⁵⁰ One of the other key advantages of the RAI is its practicality; it can be rapidly calculated in as little as 30 seconds during routine clinic visits without disrupting workflow.⁵⁰ This efficient bedside screening is crucial, especially in large healthcare systems, where it is essential to process large volumes of patients efficiently. The RAI has also been integrated into various surgical quality registries, such as the Vascular Quality Initiative (VQI), allowing RAI to be retrospectively applied.⁵⁰ Furthermore, the RAI has been validated in administrative billing data (RAI-ICD), allowing for retrospective frailty assessments using ICD-10 codes, which further enhances its utility in large-scale research and quality improvement efforts.⁵⁰ This broad applicability and ease of use make the RAI a valuable tool in clinical practice, offering real-time frailty assessments that can drive better patient outcomes through targeted interventions and more accurate risk stratification.

5 Limitations

This study has several limitations that should be considered when interpreting the findings. The retrospective nature of the analysis limits the ability to control for all potential confounding variables such as patient adherence to postoperative care and differences in surgical techniques across institutions, which may have influenced the outcomes. Similarly, the reliance on a national database may limit the generalizability of our study, especially in regard to ethnicity as 73 % of the cohort was White. Further, as there is a shift towards performing TJA in ambulatory surgical centers, the use of the ACS-NSQIP would not capture this patient population. Finally, as TJA is a relatively well tolerated procedure, the results of our multivariate analysis in regard to the predictive ability of frailty for mortality demonstrated wide confidence intervals due to the low mortality rate. However, given the significance of these results after controlling for multiple variables, these results maintain statistical and clinical significance. Taken together, future studies should aim to validate these findings in diverse populations and settings to ensure broader applicability.

6 Conclusion

The findings of this study support the utility of frailty as a predictor of 30-day mortality in older obese patients following elective TJA. Given the complex relationship of increased BMI and mortality, our results underscore the clinical relevance of frailty in identifying patients at higher postoperative risk of mortality. Furthermore, by showing the RAI is superior to the mFI-5 for discriminatory accuracy, we highlight its potential as a more comprehensive and reliable tool for preoperative risk stratification in this population. As obesity rates continue to rise and the demand for TJA increases with an aging population, incorporating robust frailty assessments like the RAI into preoperative evaluations and patient counseling may enhance surgical decision-making and improve patient outcomes by optimizing candidate selection and identifying those at the highest risk of mortality.

CRediT authorship contribution statement

Nithin Gupta: Conceptualization, Methodology, Software, Validation, Investigation, Resources, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration. Jared Sasaki: Conceptualization, Methodology, Software, Validation, Investigation, Resources, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration. Victor Koltenyuk: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration. Amber Park: Conceptualization, Methodology, Software, Validation, Investigation, Writing – original draft, Writing – review & editing, Visualization, Project administration. Hikmat R. Chmait: Conceptualization, Methodology, Software, Validation, Investigation, Writing – original draft, Writing – review & editing, Visualization, Project administration. Anthony Perugini: Conceptualization, Methodology, Software, Validation, Investigation, Writing – original draft, Writing – review & editing, Visualization, Supervision. Andrew B. Campbell: Conceptualization, Methodology, Software, Validation, Investigation, Writing – original draft, Writing – review & editing, Visualization, Supervision.

Guardian patient consent

Guardian/Patient consent was not needed for this study.

Ethical statement

Institutional Ethical Committee Approval was not needed for this study. However, we upheld publishing ethics as our duty as authors.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.