Impact of preoperative frailty on venous thromboembolism risk following total hip and knee arthroplasty: a meta-analysis — correspondence
Year: 2026
Type: article
Abstract: Lan and colleagues synthesized ten observational cohorts including more than two million patients and showed that preoperative frailty is associated with a higher risk of venous thromboembolism (VTE) after total hip and knee arthroplasty (THA/TKA)[1]. In parallel, contemporary guidance such as the TITAN guideline on Transparency in the reporting of Artificial INtelligence (AI) research has highlighted the need for transparent, clinically oriented reporting of prediction models and decision-analytic measures in surgical studies, which is directly relevant to frailty-based VTE risk prediction[2]. Their work adds to a growing body of evidence that frailty is a powerful predictor of adverse outcomes in older surgical patients and is particularly relevant as the prevalence of frailty approaches one quarter to one half in people aged over 85 years[3,4]. By using a prespecified protocol, comprehensive database searches and random-effects models, the authors have provided an important foundation for incorporating frailty into perioperative risk stratification pathways[1,3,4]. We would like to offer several comments on how these findings could be translated into frailty-informed risk prediction tools and practical thromboprophylaxis decisions in arthroplasty practice. First, the results are mainly presented as pooled odds ratios and relative risks, whereas the absolute risk of VTE across frailty strata receives less emphasis. Across the included cohorts, the overall incidence of postoperative VTE was very low, even though frail patients were at roughly double the relative risk compared with non-frail peers[1]. In clinical decision-making, surgeons and patients must weigh benefit and harm on an absolute scale: the implications of frailty raising 90-day VTE risk from 0.5% to 1.0% are very different from an increase from 3% to 6%. Reporting frailty-specific baseline risks (for example, events per 1,000 operations), absolute risk differences and numbers-needed-to-treat or harm under standard prophylaxis strategies (aspirin alone, low-molecular-weight heparin or direct oral anticoagulants) would make the results more actionable at the bedside[5]. Such estimates could also be embedded into decision aids that explicitly contrast VTE and bleeding risks in older adults who often value functional recovery and avoidance of complications as highly as survival[3,5]. Second, Lan et al. appropriately acknowledge considerable heterogeneity in frailty definitions and measurement tools, but the consequences of this heterogeneity for model transportability are not fully explored. Different instruments—including modified frailty indices, deficit-accumulation scores and claims-based frailty algorithms—capture overlapping yet distinct domains of vulnerability such as comorbidity burden, mobility limitation, cognitive impairment and nutritional status[3,4]. Pooling effect estimates across these instruments implicitly treats them as interchangeable, which may obscure which domains actually drive VTE risk after THA/TKA[1,3,4]. We suggest that future work should consider stratified analyses by frailty construct, cut-off and timing of assessment, as well as by follow-up horizon (for example 30 versus 90 days), to identify which combinations of domains and thresholds provide the greatest prognostic value and are realistic to implement in busy arthroplasty clinics[4]. Third, the meta-analysis briefly mentions the potential for risk scores but does not elaborate on how its findings could seed modern prediction pipelines. Recent large-scale arthroplasty cohorts have used logistic regression and machine-learning algorithms to predict 30-day VTE and major bleeding events, yet most models do not incorporate validated frailty measures and focus primarily on discrimination rather than clinical usefulness[6,7]. Building on Lan et al., frailty should be treated as a core candidate predictor whose incremental value is quantified through calibration, reclassification metrics and decision curve analysis, alongside traditional C-statistics[8,9]. In practice, this would allow clinicians to move from the general statement that “frail patients have higher VTE risk” to patient-specific estimates such as “for this 78-year-old frail woman undergoing primary TKA, the predicted 90-day risk of symptomatic VTE is X% with aspirin and Y% with a more potent anticoagulant, while the corresponding bleeding risks are …,” which could then be visualised through simple web-based nomograms or explainable machine-learning dashboards[6–9]. Finally, we encourage researchers to explicitly link frailty-based risk stratification with modifiable perioperative pathways[4,5]. Prehabilitation programmes, nutritional optimisation, early mobilisation protocols and tailored thromboprophylaxis have all shown promise for improving outcomes in frail or high-risk surgical populations[4,5]. Frailty-informed prediction models that jointly consider VTE and bleeding, report net clinical benefit and pre-specify risk thresholds for intensifying prophylaxis could support multidisciplinary teams in designing pragmatic, risk-stratified care bundles for THA/TKA[5–9]. In summary, Lan et al. have provided robust evidence that preoperative frailty is associated with increased VTE risk after arthroplasty. The next step is to integrate frailty into well-calibrated, decision-curve–evaluated prediction tools and prospective implementation studies, so that relative risks can be translated into concrete bedside decisions for our growing population of frail arthroplasty patients[6–9].
Source: International Journal of Surgery
Authors: Yuhuai Wu, Jianping Liu
Language: English
Subfield: Internal Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $3,450
Funder: Kunming Medical University
OpenAlex ID: W7124317551