Enhanced mortality prediction using frailty models in cardiac surgery patients: evidence from the MIMIC-IV database

Q: What are the key findings of this study?

Adding both the Frailty Index based on Laboratory Tests (FI-Lab) and Hospital Frailty Risk Score (HFRS) to a base clinical model significantly improved prediction of in-hospital mortality after cardiac surgery, increasing the AUC from 0.847 to 0.910.

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Cardiac surgery (n=4204). Intervention: Frailty assessment models (FI-Lab and HFRS) vs. Base clinical prediction model (without frailty scores). Primary outcome: Model discrimination (AUC) for in-hospital mortality (ΔAUC 0.063, p=<0.001).

Q: What does this research mean for the field?

Combining the Frailty Index based on Laboratory Tests (FI-Lab) and the Hospital Frailty Risk Score (HFRS) significantly improves postoperative mortality prediction in cardiac surgery patients compared to standard risk models alone. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Q: What question did this study set out to answer?

To evaluate the individual and combined prognostic utility of the Frailty Index based on Laboratory Tests (FI-Lab) and the Hospital Frailty Risk Score (HFRS) for predicting short- and long-term mortality following cardiac surgery.

Jiang Dai; Tong Tan; Hao Cui; Alimujiang Awaker; Yongqiang Lai; Ying Li

doi:10.1007/s00423-026-04268-z

September 21, 2026Langenbeck s Archives of SurgeryOpen Access

Enhanced mortality prediction using frailty models in cardiac surgery patients: evidence from the MIMIC-IV database

View Full Paper

Ask AI

Bookmark

Share

Key result

Adding FI-Lab and HFRS to clinical models improves post-cardiac surgery mortality prediction to ~0.91 AUC.

ΔAUC 0.063
P<0.001
n=4,204

Why the study?

The comparative and combined prognostic utility of FI-Lab and HFRS frailty measures in cardiac surgical populations remains unclear.

Does adding the FI-Lab and HFRS frailty models to standard clinical variables improve postoperative mortality prediction in adult cardiac surgery patients?

Key Points

To evaluate the individual and combined prognostic utility of the Frailty Index based on Laboratory Tests (FI-Lab) and the Hospital Frailty Risk Score (HFRS) for predicting short- and long-term mortality following cardiac surgery.
Retrospective cohort study of 4,204 adult cardiac surgery patients from the MIMIC-IV version 3.1 database.
Evaluated baseline predictive models incorporating age, gender, race, Charlson comorbidity index, and SOFA score with and without FI-Lab, HFRS, or both across in-hospital, 28-day, and 1-year mortality endpoints.
Adding FI-Lab to the base model increased AUC from 0.847 to 0.903 for in-hospital mortality, 0.835 to 0.888 for 28-day mortality, and 0.780 to 0.827 for 1-year mortality (all p < 0.001), yielding NRI improvements from 0.145 to 0.214.
The combined FI-Lab + HFRS model achieved the highest discrimination across all endpoints, with AUCs of 0.910 for in-hospital, 0.892 for 28-day, and 0.836 for 1-year mortality (all p < 0.001 vs. base model).
FI-Lab and HFRS demonstrated moderate correlation (Spearman ρ = 0.40) and fair categorical agreement (weighted κ = 0.26), with in-hospital mortality increasing from 0.3% in FI-Lab non-frail patients to 13.9% in frail patients.

Study Design

Type

Cohort (n=4,204)

Multicenter

No

Structured PICO

Does adding the FI-Lab and HFRS frailty models to standard clinical variables improve postoperative mortality prediction in adult cardiac surgery patients?

P

Population

4,204 adult patients aged 18 and older undergoing cardiac surgery with postoperative ICU admission, evaluated for frailty and followed for up to 1 year.

E

Exposure

Prognostic evaluation combining the Frailty Index based on Laboratory Tests (FI-Lab) and the Hospital Frailty Risk Score (HFRS) in mortality prediction models.

C

Comparator

Base clinical prediction model including age, sex, race, Charlson comorbidity index, and SOFA score.

O

Outcome

In-hospital mortality, 28-day all-cause mortality, and 1-year all-cause mortality.hard clinical

Main Result

Effect estimate: ΔAUC 0.063

Absolute Event Rate: 0.91% vs 0.847%

p-value: p=<0.001

Limitations

Single-center retrospective analysis limits generalizability
Residual confounding is possible
Lack of granular perioperative variables (e.g., cardiopulmonary bypass time, intraoperative events)
HFRS relies on administrative ICD codes which may introduce misclassification due to variable documentation quality
Direct comparisons with established cardiac surgical risk scores (EuroSCORE II, STS score) were not feasible
Lack of universally accepted cut-off values for FI-Lab classification
Single-center retrospective analysis limiting generalizability
Absence of granular perioperative variables (e.g., cardiopulmonary bypass time, intraoperative events)
Reliance on administrative ICD codes for HFRS, introducing potential misclassification
Inability to compare with or adjust for established cardiac surgical risk scores like EuroSCORE II and STS due to missing variables
Lack of universally accepted cut-off values for FI-Lab

Implication

Combining laboratory-based and diagnosis-based frailty indices significantly improves postoperative risk stratification and mortality prediction in cardiac surgery patients.

Abstract

The Frailty Index based on Laboratory Tests (FI-Lab) and the Hospital Frailty Risk Score (HFRS) represent two frailty measures, yet their comparative and combined prognostic utility in cardiac surgical populations remains unclear. We conducted a retrospective cohort study using the Medical Information Mart for Intensive Care IV (MIMIC-IV) version 3.1 database, including 4,204 adult patients who underwent cardiac surgery. We evaluated the association of each frailty measure with in-hospital, 28-day, and 1-year mortality and compared the predictive performance of models incorporating FI-Lab, HFRS, or both. Both FI-Lab and HFRS were strongly associated with all mortality endpoints, with FI-Lab consistently providing greater incremental discriminatory value. Adding FI-Lab to the base model increased the AUC from 0.847 to 0.903 for in-hospital mortality, from 0.835 to 0.888 for 28-day mortality, and from 0.780 to 0.827 for 1-year mortality (all p < 0.001), accompanied by significant NRI improvements (range: 0.145–0.214). The combined FI-Lab + HFRS model achieved the highest discrimination (AUC 0.910, 0.892, and 0.836, respectively). FI-Lab and HFRS demonstrated moderate correlation (Spearman ρ = 0.40) and fair categorical agreement (weighted κ = 0.26). FI-Lab and HFRS each capture distinct dimensions of frailty and independently predict postoperative mortality in cardiac surgery patients. Their combination provides superior discrimination. Incorporating both FI-Lab and HFRS into perioperative evaluation may enhance early identification of high-risk individuals and support more individualized postoperative management. Further studies are required before these measures can be recommended for routine perioperative risk assessment.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Dai et al. (2026) conducted a cohort in Cardiac surgery (n=4,204). Frailty assessment models (FI-Lab and HFRS) vs. Base clinical prediction model (without frailty scores) was evaluated on Model discrimination (AUC) for in-hospital mortality (ΔAUC 0.063, p=<0.001). Adding both the Frailty Index based on Laboratory Tests (FI-Lab) and Hospital Frailty Risk Score (HFRS) to a base clinical model significantly improved prediction of in-hospital mortality after cardiac surgery, increasing the AUC from 0.847 to 0.910.

synapsesocial.com/papers/6ad0fd403afa093ea9ad02ce https://doi.org/10.1007/s00423-026-04268-z

View Full Paper

Ask AI

Bookmark

Share

Enhanced mortality prediction using frailty models in cardiac surgery patients: evidence from the MIMIC-IV database

Key Points

Study Design

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Enhanced mortality prediction using frailty models in cardiac surgery patients: evidence from the MIMIC-IV database

Key Points

Study Design

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

More related papers

More from Langenbeck s Archives of Surgery

Keep reading

More related papers

More from Langenbeck s Archives of Surgery