Back to 2024 Posters

---

The modified Frailty Index does not significantly help with Esophagectomy Risk Prediction Modelling
Hayley Reddington , *Samih Shafique MS, *Kevin Dickson , *Allison Crawford MS, Mark W. Maxfield MD, Karl F. Uy MD, *Isabel Cristina M. Emmerick , Feiran Lou MD
Surgery, University of Massachusetts, Worcester, MA

Background:
Frailty is defined as a progressive loss of vigor with or without coexisting conditions or diseases, along with physical and cognitive decline. The modified frailty index (mFI-5) is a comorbidity-based index containing five components: diabetes (DM), hypertension requiring medication (HTN), congestive heart failure (CHF), functional status (partially or totally dependent), and chronic obstructive pulmonary disease (COPD). This study investigates the mFI-5 score or the included 4 comorbid conditions correlated to post-esophagectomy outcomes.
Study design:
Patients who underwent esophagectomy for malignancy were identified using the National Surgical Quality Improvement Program (NSQIP) targeted database from 2016-2022. Primary outcome was major complication (presence of at least one post-operatively: re-intubation, prolonged ventilation, pneumonia, esophageal leak requiring treatment, renal failure, re-operation) or 30-day mortality (M/M). Descriptive analysis, univariable and multivariable logistic regression models correlated mFI-5 and other pre-operative characteristics with the outcomes of interest.
Results:
A total of 5657 patients met inclusion criteria. Only a small percentage of these patients had CHF (1.1%) or partially/totally dependent functional status (0.6%). Most patients had a mFI-5 <2 (81%). In the univariable regression the mFI-5 score was significantly associated with major morbidity and mortality (p<0.0001). In the multi-variable model with the mFI-5 score, a score of >2 was again strongly correlated with postoperative major complications and mortality (p=0.0002). The relationship was maintained whether mFI-5 score >2 or each of the individual components were used in the models(c-statistic 0.616 and 0.621, respectively). When the individual components of the mFI-5 score were represented separately in the model, all were statistically significant except for dependent status.
Conclusion(s):
In this large cohort of esophagectomy patients, an association between mFI-5 and adverse outcomes was identified. However, when each component of mFI-5 were assessed, all except dependent status were predictive of M/M. In the case of esophagectomies, the association of mFI-5 is mainly driven by comorbid conditions like HTN, CHF, COPD, and DM, . Future studies on frailty should focus on assessing the impact of frailty phenotype as the mFI-5 may be a surrogate for comorbid conditions.
Multivariable logistic regressions for predicting post-operative major morbidity or 30-day mortality in NSQIP 2016 to 2022, n=5266

Model fit with all mFI-5 components as predictors
Variable	% N	OR	95% CI		p-value
Hx COPD	6.9	1.55	1.24	1.94	0.0001
Hx CHF	1.1	1.74	1.02	2.98	0.042
Partially or totally dependent	0.6	1.37	0.68	2.79	0.38
HTN requiring medication	49.7	1.23	1.08	1.40	0.002
Hx Diabetes	19.5	1.17	1.003	1.37	0.046
Age, per 10-year increase		1.12	1.05	1.20	0.001
Preop creatinine, per 0.1 unit increase		1.03	1.01	1.05	0.006
Preop WBC, per 1 unit increase		1.05	1.02	1.08	0.0006
BMI (versus 18.5 to <25, n = 1641/5266)					0.005
Under 18.5	3.4	1.65	1.19	2.29	0.003
25 to <30	36.3	1.01	0.87	1.17	0.93
30 to <35	19.0	1.20	0.999	1.43	0.051
35 to <40	6.7	1.30	1.003	1.67	0.047
40 or higher	3.6	1.34	0.96	1.88	0.09
Thoracic anastomosis location (versus Ivor-Lewis, n=3319/5266)					<0.0001
Transhiatal	15.0	1.07	0.90	1.27	0.44
McKeown	15.9	1.43	1.22	1.68	<0.0001
Thoraco-abdominal	6.0	0.80	0.61	1.05	0.11
Neoadjuvant therapy	72.8	0.81	0.71	0.93	0.002
Positive margins	7.7	1.29	1.03	1.60	0.024
Current smoker	23.3	1.52	1.32	1.76	<0.0001
Bleeding disorder	3.6	1.52	1.12	2.07	0.007

C-statistic = 0.621

---

Back to 2024 Posters