Association of frailty indices with post-operative outcomes in shoulder surgery: a systematic review

Abstract

Hypothesis and/or Background

Frailty is a multidimensional state of diminished physiologic reserve that increases vulnerability to surgical stress and adverse post-operative outcomes. Although frailty indices have been associated with adverse outcomes across a variety of medical and surgical specialties, their role in shoulder surgery remains unclear.

Methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines. PubMed, Embase, Scopus, Web of Science, and the Cochrane Library were searched for studies evaluating frailty in patients undergoing shoulder surgery. Eligible studies included those that applied any frailty index and reported post-operative outcomes. Study quality was assessed using the Quality in Prognosis Studies tool.

Results

Nineteen retrospective cohort studies were included. The surgical procedures evaluated were total shoulder arthroplasty, reverse shoulder arthroplasty, arthroscopic rotator cuff repair, and surgical management of proximal humerus fractures. The modified 5-item Frailty Index was the most frequently used frailty measure, followed by the Hospital Frailty Risk Score. Studies were categorized into four groups according to their original data source because several studies used overlapping databases. Frailty was consistently associated with increased post-operative complications, readmission, length of stay, non-home discharge, reoperation, and mortality. Limited evidence was available regarding the association between frailty and long-term functional outcomes. Overall, most studies were judged to have low to moderate risk of bias, although reliance on administrative databases and short-term outcomes was common.

Discussion and/or Conclusion

Frailty indices are associated with an increased risk of complications, greater health-care utilization, and mortality following shoulder surgery. However, current evidence suggests that their ability to predict shoulder-specific complications, patient-reported outcomes, and long-term functional recovery remains limited. Frailty assessment may assist peri-operative risk stratification and patient counseling, although prospective studies using standardized, multidimensional frailty measures are needed to better define its prognostic value.

Frailty is a condition defined by reduced physiologic reserve and diminished ability to adapt to internal or external stressors. Frailty is more common in older adults, but it extends beyond the physiologic changes of aging and reflects the patient's reduced ability to adequately respond to internal or external stressors.³ Frailty is also a well-known risk factor associated with adverse health outcomes and reduced well-being among older adults.¹⁴ This vulnerability becomes crucial in the surgical setting, as perioperative stress may exceed the patient's limited functional reserves.⁹

Frailty is often difficult to identify in the clinical setting, and physicians often rely on clinical judgment rather than objective, standardized diagnostic criteria.¹⁵ In recent years, there has been growing literature focused on the development and validation of frailty indices. Frailty indices were designed to move beyond simple comorbidity counts by incorporating elements of functional status and physiologic vulnerability into the assessment.⁵ However, because these indices differ in their components and methods of assessment, they may identify different at-risk patient populations and demonstrate variable prognostic performance across medical and surgical specialties.

Frailty has been associated with worse outcomes in many medical and surgical fields.^1,7,16,36 In orthopedic surgery, several studies have examined whether frailty indices can help predict post-operative complications and healthcare utilization.^18,27,33 As the number of shoulder procedures in older adults continues to increase, there has been growing interest in the application of frailty-based risk stratification in this population. Recent studies have evaluated different frailty indices in shoulder arthroplasty and reported associations between frailty and post-operative complications.²¹ However, others have questioned the strength of these associations, as substantial heterogeneity exists in both frailty measures and outcome definitions.^4,21 However, the overall prognostic value of frailty indices in patients undergoing shoulder surgery has not been evaluated.

The primary aim of this systematic review was to evaluate the association between frailty indices and post-operative outcomes following shoulder surgery and to summarize the available evidence regarding their prognostic utility. By reviewing the available literature, we sought to define the role of frailty assessment in patients undergoing shoulder surgery and its potential use as an adjunct to perioperative risk stratification.

Methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines²⁴ and the Synthesis Without Meta-analysis (SWiM) reporting guideline.² A review protocol was developed prior to study initiation to ensure a transparent process for study identification, selection, and data extraction and is provided in Appendix A. The completed Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 and SWiM reporting checklists are available in Appendix B.

Literature search strategy

The literature search was conducted on November 28, 2025, in PubMed, Embase, Scopus, Cochrane Library, and Web of Science. The search strategy was developed to identify studies evaluating frailty in patients undergoing shoulder surgery. In PubMed, we used the terms “frailty,” “frailty index,” and “frailty syndrome,” together with shoulder surgery terms such as “shoulder arthroplasty,” “shoulder,” “arthroscopy,” “rotator cuff,” “proximal humerus,” “Bankart,” “SLAP,” and “proximal humeral fracture.” Boolean operators (“OR” and “AND”) were used to combine different search terms. The search strategy was adapted for each database to account for differences in indexing and controlled vocabulary, and the complete search strategies are provided in Appendix C.

Study selection and eligibility criteria

All records were imported into a reference management software (Zotero, Corporation for Digital Scholarship, Vienna, VA, USA). Duplicate citations were identified and removed before screening the titles and abstracts. Studies were eligible for inclusion if they met the following criteria: (1) used any frailty index or frailty scoring system as a main part of their methodology; (2) included patients undergoing common shoulder surgical procedures, such as shoulder arthroplasty (anatomic or reverse), rotator cuff repair, proximal humerus fracture fixation, shoulder arthroscopy, or shoulder stabilization procedures (eg, Bankart or SLAP repair); and (3) reported perioperative or post-operative outcomes in relation to frailty index scores. Studies were excluded if they (1) were narrative reviews, systematic reviews, meta-analyses, editorials, commentaries, letters, or conference abstracts without an available full text; (2) were non-English publications; or (3) did not evaluate a frailty index or frailty scoring system. Non-English publications were excluded because the review team did not have the resources to ensure accurate translation and data extraction. Conference abstracts were excluded because they generally do not provide sufficient methodological detail or complete outcome data for quality assessment and data extraction and are often not peer reviewed.

Two reviewers independently screened the titles and abstracts of all identified records to identify potentially eligible studies. Full-text articles were then independently assessed for eligibility by the same 2 reviewers. Any disagreements during study selection were resolved through discussion and consensus. Data were independently extracted by the 2 reviewers using a standardized Microsoft Excel spreadsheet. Extracted data included study characteristics, patient demographics, type of shoulder surgery performed, frailty index used (including cutoff thresholds), outcome measures, and post-operative outcomes. Any discrepancies in data extraction were resolved by consensus.

Quality assessment

Risk of bias was independently assessed by 2 reviewers using the Quality in Prognosis Studies (QUIPS) tool. QUIPS evaluates 6 domains of potential bias: (1) study participation, (2) study attrition, (3) prognostic factor measurement, (4) outcome measurement, (5) study confounding, and (6) statistical analysis and reporting. Each domain was judged as having low, moderate, or high risk of bias. Disagreements were resolved by consensus. Consistent with the developers' recommendations, no overall numerical quality score was calculated. Because of the substantial heterogeneity among included studies and the absence of quantitative synthesis, reporting bias (eg, publication bias) and certainty assessment were not assessed.

Results

Study selection

The initial literature search identified 505 records across 5 databases, including 85 from PubMed, 101 from Embase, 92 from Scopus, 190 from Web of Science, and 37 from the Cochrane Library. After removal of 152 duplicate records, 353 records remained and were screened based on titles and abstracts. Of these, 309 records were excluded. The remaining 44 reports underwent full-text review for eligibility; 25 were excluded for failing to meet the inclusion criteria. Ultimately, 19 studies were included in the final qualitative synthesis (Fig. 1). The complete search strategies for all databases and the list of full-text articles excluded after eligibility assessment, together with the reasons for exclusion, are provided in Appendix C.

Figure 1.

PRISMA 2020 flow diagram illustrating the study selection process for this systematic review. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Study characteristics

Included studies consisted of large national administrative database analyses and single or multi-institutional cohort studies. Nine studies used the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database, 3 used the Nationwide Readmissions Database (NRD), 1 used the Premier Healthcare Database, and 1 used a German nationwide claims database. Five studies were institutional cohort studies.

We categorized the studies into 4 categories: ACS-NSQIP studies, NRD studies, other administrative database studies, and institutional cohort studies (Table I). This approach was used instead of presenting each outcome individually to minimize duplication of evidence arising from overlapping patient populations and study periods among database-based studies, particularly those using the ACS-NSQIP and NRD. By synthesizing these studies collectively, we aimed to avoid overemphasizing findings that may reflect the same underlying patient cohorts.

Table I.

Characteristics of included studies evaluating frailty indices and post-operative outcomes in shoulder surgery.

Database category	Study	Data source/setting	Follow-up	Procedure/indication	Sample size	Frailty index	Frailty definition	Comparator	Outcomes assessed
Nationwide Readmissions Database	Moore et al, 202522	Nationwide Readmissions Database (2016-2021)	180 d	Primary aTSA	184,460	HFRS; mFI-5	HFRS: <5, 5-15, >15; mFI-5 categorical	Non-frail	Any complication, adverse discharge, readmission, revision, mortality
	Silvestre et al, 202528	Nationwide Readmissions Database (2016-2020)	180 d	Primary aTSA	254,150	HFRS	≥5 vs. < 5	HFRS < 5	Any complications, readmission, LOS, revision surgery, discharge to facility, mortality
	Lakra et al, 202517	Nationwide Readmissions Database (2017-2019)	30 d	Primary aTSA and rTSA	107,774	HFRS	≥5 vs. < 5	HFRS < 5	Any complications, reoperation, readmission, LOS
Other administrative database studies	Mayfield et al, 202520	Premier Healthcare Database (2016-2022)	90 d	Primary aTSA and rTSA	86,356	Johns Hopkins ACG Frailty Index	≥1	Non-frail	Any complication, LOS, non-home discharge, readmission
	Spoden et al, 202531	German nationwide claims (2015-2021)	90 and 365 d	Proximal humerus fracture surgery	34,912	HFRS	<5, 5–15, >15	HFRS < 5	Any complication, reoperation, mortality
ACS-NSQIP	Traven et al, 201935	ACS-NSQIP (2005-2017)	30 d	Primary aTSA	18,957	mFI-5	Continuous	Lower mFI-5	Any complications, LOS, readmission, discharge to facility, mortality
	Holzgrefe et al (2019)13	ACS-NSQIP	30 d	aTSA	9,861	mFI-5	≥2 and 1 vs. 0	Non-frail	Any complications, readmission, reoperation, LOS, adverse discharge, mortality
	Zhu et al, 202540	ACS-NSQIP (2011-2022)	30 d	Primary aTSA and rTSA	39,810	mFI-5	Continuous	Lower mFI-5	Any complications, LOS, readmission, non-home discharge, mortality
	Fu et al, 20198	ACS-NSQIP (2005-2015)	30 d	Primary aTSA and rTSA	10,527	mFI-5	Continuous	Lower mFI-5	Any complication, LOS, discharge
	Seibold et al, 202526	ACS-NSQIP (2013-2019)	30 d	Revision TSA	1,393	mFI-5	≥2 vs. 0-1	Less frail	Any complications, readmission, LOS
	Ling et al, 202419	ACS-NSQIP (2015-2020)	30 d	Primary aTSA and rTSA (≥65 y)	9,228	mFI-6	≥3 vs. < 3	mFI-6 < 3	Any complication, readmission, discharge
	Yi et al, 202138	ACS-NSQIP (2005-2017)	30 d	Proximal humerus fracture surgery	3,893	mFI-5	Continuous	Lower mFI-5	Any complication, reoperation, adverse events, LOS
	Evans et al, 20216	ACS-NSQIP (2014-2017)	30 d	Proximal humerus fracture ORIF	2,004	mFI-5	Ordinal (0-3)	Lower mFI-5	Any complication, readmission, reoperation, LOS
	Traven et al, 202034	ACS-NSQIP (2006-2016)	30 d	Arthroscopic RCR	24,477	mFI-5	Continuous	Lower mFI-5	Any complication, LOS, readmission, adverse discharge, mortality
Institutional/multi-institutional cohort studies	Cox et al, 20234	Single institution (2016-2019)	90 d	Primary aTSA and rTSA	1,365	mFI-5	Continuous	Lower mFI-5	Any complications, LOS, adverse discharge, reoperation, readmission
	Smith et al, 202429	Single institution (2016-2023)	90 d	Primary aTSA and rTSA	338	mFI-5	≥2 vs. < 2	mFI-5 < 2	ED visits, readmission, post-operative surgical complications
	Smith et al, 202530	Single institution (2016-2024)	90 d	Primary aTSA and rTSA	327	mFI-5	≥2 vs. < 2	mFI-5 < 2	Discharge to facility, reoperation, readmission
	Zhang et al, 202339	Two Level-I trauma centers	Minimum 2 yr	rTSA for proximal humerus fracture	153	mFI-5	Ordinal (0-3)	Lower mFI-5	Any complications, reoperation, functional outcomes
	Moorthy et al, 202123	Single institution	24 mo	Arthroscopic RCR	340	mFI-11; CFS	Ordinal	Less frail	Functional outcomes only

ACS-NSQIP, American College of Surgeons National Surgical Quality Improvement Program; mFI-5, modified 5-item Frailty Index; mFI-6, modified 6-item Frailty Index; HFRS, Hospital Frailty risk score; CFS, Clinical Frailty scale; TSA, total shoulder arthroplasty; rTSA, reverse total shoulder arthroplasty; LOS, length of stay; RCR, rotator cuff repair; aTSA, anatomic total shoulder arthroplasty; ED, emergency department.; ORIF, open reduction and internal fixation.

The surgical procedures evaluated across the included studies were anatomic total shoulder arthroplasty (aTSA), reverse total shoulder arthroplasty (rTSA), revision total shoulder arthroplasty (revision TSA), proximal humerus fracture surgery, and arthroscopic rotator cuff repair. Study characteristics, including frailty indices, surgical indications, and outcome definitions, are summarized in Table I and complemented by additional study details provided in Appendix D.

Frailty indices

The most commonly used frailty index across all studies was the modified 5-item Frailty Index (mFI-5). However, frailty categorization and analysis were inconsistent between studies. Some authors used mFI-5 as a continuous variable, others used it as an ordinal variable, and some used a cutoff of 2. One ACS-NSQIP study evaluated frailty using the modified 6-item Frailty Index (mFI-6). All studies using the NRD evaluated frailty with the Hospital Frailty Risk Score (HFRS), typically using a cutoff of 5. One study additionally stratified frailty severity into moderate (HFRS 5-15) and severe frailty (HFRS >15).

One administrative database study introduced the Johns Hopkins Adjusted Clinical Groups (JHACG) frailty indicator as an alternative claims-based frailty index. Among institutional studies, mFI-5 remained the most commonly used frailty measure. One study used the 11-item mFI and the Clinical Frailty Scale (CFS) as their main tool for assessing frailty.²³

The mFI-5 and mFI-11 are deficit-based frailty indices derived from clinical variables, whereas the HFRS and JHACG are claims-based measures developed using administrative data. The CFS is a clinician-assessed measure of functional status and overall frailty. These instruments assess different domains of frailty and should not be considered interchangeable. Appendix E summarizes the frailty indices used across the included studies.

Nationwide Readmissions Database studies

In this group, 3 studies evaluated frailty in shoulder arthroplasty populations using the HFRS.^17,22,28 These studies had similar study periods and included overlapping patient cohorts. Accordingly, they were not treated as independent sources of evidence but were interpreted collectively within the same evidence category. Across studies, higher HFRS was consistently associated with worse short-term and intermediate-term outcomes following TSA. At 30 days, HFRS independently predicted reoperation (odds ratio [OR]: 1.61; 95% confidence interval (CI): 1.22-2.09), readmission (OR: 1.79; 95% CI: 1.63-1.97), major medical complications (OR: 2.19; 95% CI: 1.95-2.45), major surgical complications (OR: 1.51; 95% CI: 1.15-1.97), longer hospital stay (incidence rate ratio 1.46; 95% CI 1.44-1.47), and higher hospital costs (incidence rate ratio: 1.09; 95% CI: 1.09-1.10).¹⁷ At 180 days, HFRS remained associated with increased complications, readmission, mortality, and medical complications including acute renal failure, sepsis, urinary tract infection, Acute Respiratory Distress Syndrome, and pneumonia²⁸(Appendix D). Moore et al further demonstrated that HFRS had better discriminative performance than the mFI-5 for predicting any complication (area under the receiver operating characteristic curve [AUC] 0.758 vs. 0.616), medical complications (AUC 0.861 vs. 0.672), surgical complications (AUC 0.621 vs. 0.572), revision (AUC 0.629 vs. 0.593), and mortality (AUC 0.720 vs. 0.670), whereas discrimination for readmission was similar (AUC 0.562 vs. 0.568). They also compared HFRS with mFI-5, the Elixhauser Comorbidity Index, and Charlson-Deyo Comorbidity Index, and concluded that HFRS had stronger discrimination for adverse outcomes.²²

American College of Surgeons National Surgical Quality Improvement Program studies

Nine studies used ACS-NSQIP data and had large patient populations and timeline overlap.^{6,8,13,19,26,34,35,38,40} Eight studies used the mFI-5, and one study used the mFI-6. Because these studies used the same database and overlapping study periods, their study populations overlapped. Accordingly, they were not considered independent sources of evidence, and their findings were interpreted collectively to avoid overrepresenting evidence from overlapping populations.

Frailty measured using the modified Frailty Index (mFI) was consistently associated with worse 30-day post-operative outcomes following shoulder arthroplasty.^13,26,35 After adjustment for patient and procedural characteristics, higher mFI-5 scores were independently associated with any complication (ORs: 1.57-2.40), readmission (ORs: 1.45-2.80), adverse discharge (ORs: 1.45-3.21), reoperation (ORs: 1.82-2.53), prolonged length of stay (LOS), and mortality (ORs: 2.11-2.66).^{6,13,26,34,35} Similar associations were observed in revision shoulder arthroplasty, where an mFI-5 score ≥2 independently predicted readmission (OR: 2.58; 95% CI: 1.17-5.72), post-operative transfusion (OR: 3.66; 95% CI: 1.47-9.11), prolonged LOS (OR: 2.43; 95% CI: 1.37-4.34), and non-home discharge (OR: 3.22; 95% CI: 1.83-5.66).²⁶

The single study evaluating the mFI-6 likewise demonstrated that a score ≥3 independently predicted urinary tract infection (OR: 2.79; 95% CI: 1.49-5.23), blood transfusion (OR: 1.53; 95% CI: 1.01-2.32), readmission (OR: 1.58; 95% CI: 1.06-2.35), and non-home discharge (OR: 2.60; 95% CI: 2.08-3.25).¹⁹

In patients with proximal humerus fractures, increasing mFI-5 scores were associated with higher rates of adverse events and complications following operative management.³⁸ Likewise, higher mFI-5 scores were associated with increased odds of medical complications, hospital admission, prolonged LOS, adverse discharge, and mortality following arthroscopic rotator cuff repair, although no significant association was observed with surgical site infection.³⁴

Despite these consistent associations, the predictive discrimination of mFI-based models remained modest, with reported AUCs generally ranging from 0.56 to 0.63 for most post-operative outcomes.^8,38,40 Although Yi et al³⁸ reported higher discrimination for selected outcomes, including renal complications (AUC: 0.878), most models demonstrated only limited discriminative ability, suggesting that mFI alone has limited value as a standalone predictor of post-operative risk (Appendix D).

Other administrative database studies

Two other database studies investigated the association between frailty and adverse post-operative outcomes across shoulder arthroplasty and proximal humerus fracture populations.

Using German nationwide claims data, Spoden et al³¹ evaluated patients undergoing surgical management of proximal humerus fractures and demonstrated a strong relationship between increasing HFRS severity and post-operative morbidity. Patients with severe frailty (HFRS >15) had markedly increased risks of secondary surgery, general complications during index admission, and in-hospital mortality.

Mayfield et al²⁰ was the only study in our research that used the JHACG frailty index. The JHACG is a claims-based frailty measure that identifies frailty using diagnosis codes encompassing multidimensional deficits, including functional impairment, cognitive impairment, nutritional status, geriatric syndromes, and social vulnerability (Appendix E). Compared with the mFI-5, which primarily assesses comorbidity burden using 5 clinical variables, the JHACG captures a broader range of geriatric deficits. In contrast to the HFRS, which was developed to identify frailty among hospitalized older adults using International Classification of Diseases-based diagnoses, the JHACG was designed as a population-level claims-based frailty measure applicable across healthcare settings. Frail patients had significantly higher odds of prolonged LOS (aOR: 2.42; 95% CI: 2.16-2.71), skilled nursing facility discharge (aOR: 4.15; 95% CI: 3.54-4.86), readmission (aOR: 1.71; 95% CI: 1.26-2.33), and medical complications (aOR: 2.06; 95% CI: 1.58-2.68), whereas the association with surgical complications was not statistically significant (aOR: 1.30; 95% CI: 0.83-2.05).

Institutional studies

Several non-database studies further investigated the association between frailty indices and adverse post-operative outcomes following shoulder surgery.

Cox et al⁴ compared multiple comorbidity and frailty indices following primary aTSA and rTSA and showed mFI-5 had modest discrimination for post-operative complications and readmission. Smith et al²⁹ examined the association between mFI-5 and short-term outcomes after elective aTSA and rTSA. Their results showed that mFI-5 ≥ 2 was independently associated with 90-day readmission but was not associated with 90-day emergency department (ED) return, suggesting that frailty may be more relevant to unplanned readmission than to post-operative ED utilization. A separate study by Smith et al³⁰ evaluated discharge with home health services after elective aTSA rather than frailty as the primary exposure. Although patients discharged with home health services had higher baseline frailty, discharge with home health services was not independently associated with 90-day ED return or readmission. In secondary analyses, mFI-5 ≥2 was not significantly associated with 90-day readmission.

Another institutional study by Zhang et al³⁹ evaluated patients undergoing rTSA for proximal humerus fractures and found no significant association between mFI-5 score and post-operative complications, reoperation, pain, range of motion, or patient-reported outcomes.

Moorthy et al²³ was the only included study in our research to evaluate frailty in patients undergoing arthroscopic rotator cuff repair and to focus primarily on post-operative functional recovery rather than perioperative complications. Outcomes were assessed pre-operatively and at 3, 6, 12, and 24 months using the Oxford Shoulder Score, Constant Shoulder Score, University of California, Los Angeles Shoulder Score, and visual analog scale for pain. Higher mFI-11 scores were independently associated with worse functional scores and pain at several post-operative time points, particularly at 6 and 12 months. Although the mFI-11 remained a significant predictor of selected outcomes in the multivariable models through 24 months, differences between frailty groups were no longer statistically significant at the 24-month assessment, suggesting that frail patients experienced slower recovery but ultimately achieved outcomes comparable with those of less frail patients. The authors also compared post-operative improvements with published Minimal Clinically Important Difference thresholds for the Oxford, Constant, and University of California, Los Angeles Shoulder Scores and reported that these improvements were generally clinically meaningful at 12 and 24 months. However, substantial clinical benefit and patient acceptable symptom state thresholds were not assessed.²³

Quality assessment

The QUIPS risk-of-bias assessment is summarized in Figure 2. Moderate risk of bias was most commonly identified in the study confounding domain, reflecting the observational design of the included studies and the potential for residual confounding despite multivariable adjustment. Several studies also demonstrated moderate risk of bias in the statistical analysis and reporting domain because of incomplete reporting of model-building strategies, handling of missing data, or assessment of model assumptions. One study was judged to have a high risk of bias in the study attrition and study confounding domains because of substantial attrition resulting from the minimum 2-year follow-up requirement and the absence of multivariable adjustment for potential confounders.³⁹

Figure 2.

QUIPS risk-of-bias assessment of the included studies. Domains assessed included study participation (D1), study attrition (D2), prognostic factor measurement (D3), outcome measurement (D4), study confounding (D5), and statistical analysis and reporting (D6). Green denotes low risk of bias, and yellow denotes moderate risk of bias. QUIPS, Quality in Prognosis Studies.

Heterogeneity and inability to perform meta-analysis

A meta-analysis was not performed because of substantial clinical and methodological heterogeneity across the included studies, consistent with SWiM guidance. Frailty was assessed using different instruments (mFI-5, mFI-11, HFRS, CFS, JHACG, and other indices) and modeled inconsistently (continuous vs. categorical variables with varying cutoff values). The included studies evaluated heterogeneous shoulder surgery populations, including primary and rTSA, arthroscopic rotator cuff repair, and proximal humerus fracture management. Outcome definitions, follow-up periods, and effect measures (eg, odds ratios, hazard ratios, and area under the receiver-operating characteristic curve) also varied substantially across studies. Furthermore, overlap among ACS-NSQIP and NRD cohorts raised concerns regarding the independence of observations required for quantitative synthesis.

Discussion

Frailty is generally recognized as an age-related syndrome characterized by reduced physiologic reserve and diminished resilience across multiple organ systems. Despite its well-established clinical importance, no universally accepted definition or standardized method of assessment currently exists.²⁵ Consequently, numerous frailty indices have been developed, each capturing different dimensions of physiologic vulnerability and functional decline.

Within orthopedic surgery, the literature commonly relied on indices derived from large databases.¹¹ The mFI-11 was adapted from the 70-variable Canadian Study of Health and Aging deficit-accumulation Frailty index and incorporates 11 variables reflecting medical comorbidities and functional dependence.³⁷ A simplified version of mFI-11, the mFI-5, was later introduced using the ACS-NSQIP database to improve feasibility and ease of use by focusing on 5 core comorbidity and functional variables.³² The HFRS is derived exclusively from International Classification of Diseases-based diagnostic codes and emphasizes healthcare utilization patterns and comorbidity burden, without directly capturing functional status.¹⁰ These differences among frailty indices highlight that they measure different domains of frailty using different methodological approaches.

In this systematic review, frailty indices were consistently associated with adverse post-operative outcomes following shoulder surgery. Strong and relatively consistent associations were observed for medical complications, prolonged LOS, readmission, non-home discharge, mortality, and reoperation. However, when discrimination was evaluated, prognostic performance varied across frailty indices and was generally modest. In contrast, associations with surgical and shoulder-specific complications were weaker and less consistent.

HFRS generally demonstrated stronger discriminatory performance for systemic medical complications and healthcare utilization outcomes. However, all HFRS studies were derived from the same national database with partially overlapping study periods and overlapping patient populations, which limits the extent to which these findings can be considered fully independent. Studies using the ACS-NSQIP database predominantly utilized the mFI-5 and demonstrated associations between increasing frailty and higher rates of medical complications, prolonged LOS, readmission, adverse discharge, and mortality across multiple shoulder procedures. However, despite these consistent associations, reported AUC values for mFI-based models remained modest, suggesting limited discriminatory ability rather than strong predictive performance. It is also important to note that the ACS-NSQIP database tracks perioperative outcomes only within a 30-day window following surgery, further limiting the ability to generalize findings to longer-term outcomes. Therefore, these findings support the use of frailty indices as adjuncts to perioperative risk stratification rather than as standalone tools for predicting post-operative outcomes.

On the other hand, institutional studies demonstrated more variable findings, with several studies showing weaker or nonsignificant associations between frailty indices and post-operative outcomes. This may reflect that frailty indices designed for large administrative databases have limited ability to predict outcomes within independent institutional cohorts. These findings suggest that frailty indices developed and validated within one database should not be assumed to demonstrate similar performance when applied to institutional cohorts or other data sources. This may also reflect the differences in data collection methods and coding practices across databases, which can affect the performance of frailty indices. In addition, when the mFI, originally developed using the ACS-NSQIP database, was evaluated against the HFRS in NRD-based studies, the mFI only marginally outperformed models based on age and sex alone.²² Overall, it should be emphasized that each frailty index tends to perform best within the database framework for which it was originally designed, whereas its generalizability to other databases and institutional cohorts may be limited.

Collectively, these findings suggest that frailty indices may function better as broad perioperative risk stratification tools rather than standalone predictive models in shoulder surgery populations. Notably, patient-reported outcomes and functional recovery remain substantially underexplored within the shoulder frailty literature. Only a limited number of studies evaluated post-operative functional outcomes, and findings were inconsistent across procedures and frailty definitions. This represents an important gap, as frailty may influence not only perioperative complications but also rehabilitation potential, recovery trajectory, and long-term functional improvement.

This systematic review has several significant limitations. First, all included studies were retrospective, introducing inherent risks of selection bias, residual confounding, and incomplete adjustment for unmeasured confounders. These methodological limitations were reflected in the QUIPS risk-of-bias assessment, in which study confounding was the most frequently identified methodological concern. One study was judged to be at high risk of bias because of concerns related to study attrition and confounding. Second, most studies relied on large administrative or registry-based databases. Although these databases enable large-scale analyses, they provide limited clinical granularity and limited information on late readmissions, functional recovery, quality of life, implant survivorship, and shoulder-specific outcomes. In addition, reliance on administrative coding may introduce misclassification of both frailty and post-operative outcomes.¹² Third, most included studies evaluated older patients undergoing shoulder arthroplasty, limiting the generalizability of the findings to younger patients and other shoulder procedures. Furthermore, shoulder-specific complications, patient-reported outcomes, and long-term functional recovery were evaluated in only a few studies, limiting conclusions regarding these outcomes. Fourth, a review protocol was not prospectively registered, which may reduce methodological transparency. Fifth, the certainty of evidence was not assessed using GRADE because of the substantial heterogeneity across included studies. Differences in frailty indices, cutoff values, patient populations, outcomes, comparator groups, and reported effect measures limited the ability to perform meaningful outcome-level certainty assessments. Sixth, the exclusion of non-English studies and conference abstracts may have introduced language and publication bias by omitting potentially relevant evidence. Finally, patients with severe or high-grade frailty were underrepresented in most cohorts, limiting the ability to evaluate outcomes in the most vulnerable patients and potentially underestimating the impact of advanced frailty on post-operative risk.

Overall, the available evidence suggests that frailty assessment may be a useful adjunct to pre-operative risk stratification in shoulder surgery by helping identify patients who may be at increased risk of post-operative complications. Incorporation of frailty assessment into pre-operative evaluation may support surgical planning, perioperative optimization, patient counseling, and shared decision-making, particularly among patients with elevated frailty scores. However, the current evidence does not support any single frailty index as a validated tool for routine clinical use, given the variability in their performance across different populations and data sources. Future prospective studies using standardized frailty measures and patient-reported outcomes are needed to improve risk prediction and inform targeted perioperative strategies. There may also be value in developing shoulder-specific frailty indices based on shoulder surgery databases.

Conclusion

The current literature is limited by heterogeneity in frailty assessment, outcome reporting, the predominance of retrospective database studies, and overlapping study populations within national databases. Patient-reported outcomes, shoulder-specific complications, and long-term functional recovery remain underreported or inconsistently evaluated. Despite these limitations, current evidence supports an association between frailty and adverse perioperative outcomes, including complications, prolonged LOS, readmission, non-home discharge, mortality, and increased healthcare utilization. Frailty assessment may therefore serve as a useful adjunct to comprehensive pre-operative risk stratification in shoulder surgery, although standardized assessment methods and prospective studies are needed to better define its prognostic value and determine the comparative performance of individual frailty indices.

Disclaimers

Funding: No funding was disclosed by the authors.

Conflicts of interest: Jason Koh is a shareholder in Marrow Access Technologies, which manufactures a marrow access device that may be used in rotator cuff repair. Any additional authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.