VA Risk Calculator Predicts Long-Term Loss of Independence After Surgery

Redefining Postoperative Success Beyond the 30-Day Window

While surgical success is traditionally measured by 30-day mortality or the absence of immediate complications, these metrics often fail to capture the multidimensional nature of patient recovery [1]. Major surgery can trigger a cascade of long-term challenges, including postoperative sleep disturbance (a common complication occurring in 15% to 72% of patients) and perioperative neurocognitive disorder (a decline in cognitive function that adversely impacts quality of life) [2, 3]. For elderly patients, the preservation of functional independence is often hindered by sarcopenia (a muscle disease rooted in the loss of muscle strength and quality) and other physical vulnerabilities [4]. Despite the clinical importance of these long-term trajectories, point-of-care tools to predict which patients will experience a protracted recovery remain scarce [1]. A recent study addresses this gap by evaluating a predictive model designed to identify patients at risk for a permanent loss of independence months after their procedure.

Defining Protracted Recovery and Loss of Independence

Surgery often has lasting impacts on quality of life, necessitating outcome measurement well beyond the standard 30-day postoperative window. To address this clinical blind spot, researchers developed a risk calculator specifically designed to anticipate long-term functional decline before a patient enters the operating room. By utilizing Veterans Affairs Surgical Quality Improvement Program data from 2016 to 2019, the investigators sought to move past immediate surgical complications and focus on a patient's ability to return to their baseline functional state. To categorize these outcomes, the researchers divided postoperative recovery trajectories into two distinct groups based on duration and severity. The first group, classified as routine or slow recovery, included patients who returned to their baseline within a 0- to 60-day window. In contrast, the second group was defined by protracted recovery or loss of independence lasting six months or more. This six-month threshold signifies a profound shift in a patient's life trajectory and care requirements. For practicing physicians, identifying this specific subset of patients allows for a more nuanced discussion during the informed consent process. This approach shifts the clinical focus from mere survival to the preservation of functional autonomy, ensuring that both surgeons and patients have a realistic understanding of the potential for extended disability.

Leveraging Comprehensive Veteran Health Data

To build a predictive model capable of identifying long-term functional decline, the researchers conducted a retrospective cohort study using the Veterans Affairs Surgical Quality Improvement Program database. This analysis included all surgical records captured within the program from 2016 to 2019, providing a robust foundation of clinical data from a period of standardized surgical practice. The strength of the predictive model relied on the integration of disparate data points to create a holistic view of patient health. The researchers linked the primary surgical data to a variety of other Veterans Affairs data sources, allowing them to capture a broad range of candidate predictors that extend beyond the immediate surgical event. This data linkage provided a comprehensive longitudinal record for each patient, incorporating historical health trends and prior healthcare utilization patterns that are often omitted from traditional surgical risk assessments. By analyzing patients who underwent procedures in these facilities, the researchers tracked trajectories of recovery across the continuum of care. This extensive data environment enabled the team to identify that the strongest predictor of protracted recovery or loss of independence was the Care Assessment Needs score for 90-day hospitalization, followed closely by the patient's baseline functional status. For clinicians, this highlights the critical importance of pre-existing frailty and healthcare needs in determining postoperative outcomes.

Identifying the Strongest Clinical Predictors

The development of this model addresses a critical void in perioperative planning, as point-of-care risk calculators have historically focused on short-term mortality rather than long-term functional outcomes. To identify the most relevant factors from a vast array of health data, the researchers used LASSO regression (a statistical method for variable selection that narrows down a broad range of candidate predictors to those with the highest predictive value). This mathematical approach allowed the authors to estimate a predictive model that remains robust across different patient populations by penalizing less significant variables and reducing the risk of statistical noise. The analysis revealed that the strongest predictor of protracted recovery or loss of independence was the Care Assessment Needs score for 90-day hospitalization. This score is a validated metric used within the Veterans Affairs system to determine a patient's risk for acute care utilization. Following this score, the researchers identified that functional status was the second strongest predictor of protracted recovery or loss of independence, emphasizing that a patient's baseline physical autonomy is a primary determinant of their status six months after a procedure. By focusing on these specific clinical markers, the model achieved high discriminative power, maintaining a sensitivity of 83.0% to 83.6% and a specificity of 82.4% to 82.5%.

Model Performance and Machine Learning Comparison

The predictive model demonstrated high discriminative power, measured by the c-statistic (a metric indicating how well a tool distinguishes between patients who will experience an outcome and those who will not, with 1.0 being perfect prediction). Specifically, the c-statistics ranged from 0.906 to 0.908 across both the training and test samples. For clinicians, these values indicate a high degree of reliability in identifying patients at risk for long-term functional decline. The model achieved a sensitivity of 83.0% to 83.6%, meaning it correctly identified more than four out of five patients who would eventually lose their independence six months after surgery. Furthermore, the specificity was recorded between 82.4% and 82.5%, indicating a consistent ability to correctly identify patients who would maintain their independence during the recovery period. To ensure the most accurate predictive tool was selected, the researchers compared the LASSO regression model against gradient boosting (a tree-based machine learning technique that builds a model in stages to improve accuracy). While the gradient boosting model achieved a c-statistic of 0.920, the authors noted that this was not a substantial improvement over the performance of the simpler LASSO model. This finding suggests that primary clinical predictors, such as functional status and hospitalization risk, carry sufficient weight that more complex computational methods do not significantly enhance the tool's utility in a clinical setting.

Clinical Utility and Shared Decision-Making

The clinical utility of this predictive model lies in its ability to shift the focus of surgical care from immediate survival to long-term functional preservation. By identifying high-risk patients who are likely to experience a protracted recovery or a loss of independence six months post-operation, clinicians can implement targeted interventions to improve outcomes for these specific individuals. These interventions might include prehabilitation programs, which involve preoperative physical therapy and nutritional optimizations designed to increase physiological reserve, or more intensive postoperative monitoring and home health support. Having a reliable metric for long-term risk allows for the strategic allocation of resources to those most likely to benefit from enhanced perioperative care pathways. While the model provides a robust statistical foundation for predicting recovery trajectories, its integration into routine clinical practice requires further refinement. The researchers emphasize that future work is required on risk communication for shared decision-making (the collaborative process where clinicians and patients make healthcare choices together based on clinical evidence and patient values). Translating a statistical risk into a meaningful conversation about a patient's lifestyle and goals remains a challenge. Physicians must be able to explain how a high risk of protracted recovery might impact a patient's ability to live alone or perform daily activities, ensuring that the decision to proceed with surgery aligns with the patient's long-term quality-of-life expectations.

References

1. Ou-Young J, Boggett S, Ansary DE, Clarke-Errey S, Royse CF, Bowyer AJ. Identifying risk factors for poor multidimensional recovery after major surgery: A systematic review.. Acta anaesthesiologica Scandinavica. 2023. doi:10.1111/aas.14302

2. Yin H, Wang S, Huang H. Effect of perioperative esketamine administration for postoperative sleep disturbance following general anaesthesia: protocol for a planned systematic review and meta-analysis of randomised controlled trials.. BMJ open. 2026. doi:10.1136/bmjopen-2025-114079

3. Wu C, Wei X, Luo F, et al. Electroacupuncture for the prevention of perioperative neurocognitive disorder in elderly patients undergoing general anesthesia: a systematic review and meta-analysis.. Frontiers in medicine. 2026. doi:10.3389/fmed.2026.1729153

4. Cruz‐Jentoft AJ, Bahat G, Bauer JM, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age and Ageing. 2018. doi:10.1093/ageing/afy169