S. Subramaniam1, J. J. Aalberg1, R. P. Soriano2, C. M. Divino1 1Icahn School Of Medicine At Mount Sinai,Division Of General Surgery, Department Of Surgery,New York, NY, USA 2Icahn School Of Medicine At Mount Sinai,Department Of Geriatrics And Palliative Medicine,,New York, NY, USA
Introduction: The modified frailty index (mFI-11) is a NSQIP based 11-factor index that has been proven to adequately reflect frailty and predict mortality and morbidity. These 11 factors, made of 16 variables, map to the original 70 item Canada Study of Health and Aging Frailty Index. In the past years, certain NSQIP variables have been removed from the database; as of 2015, only 5 out of the original 11 factors remain. The predictive power and usefulness of these five factors in an index (mFI-5) have not been proven in past literature. The goal of our study was to compare the mFI-5 to the mFI-11 in terms of value and predictive ability for mortality, post-operative infection and unplanned thirty-day readmission for future research and clinical use.
Methods: The mFI-5 is made up of the following factors: functional status, diabetes, history of COPD, hypertension, and history of CHF and was calculated by dividing the number of factors present for a patient by the number of available factors for which there were no missing data. Spearman’s Rho was calculated in order to compare mFI-5 and mFI-11 value. Predictive models, using both unadjusted and adjusted logistic regressions were created for each of the three chosen outcomes using 2012 NSQIP data, the last year all mFI-11 variables existed. Adjusted models were controlled for ASA classification, wound class, age, transfer status, surgical complexity represented by RVU, inpatient status, anesthesia type, and emergency type. Both adjusted and unadjusted models using mFI-5 were run on 2015 data to validate results. All above methods were conducted for the following nine surgical subspecialties including general surgery, cardiac surgery, neurosurgery, gynecology, orthopedics, otolaryngology, plastic surgery, thoracic surgery, urology, vascular surgery and then completed for all surgical subspecialties combined
Results: Correlation between the mFI-5 and mFI-11 were above 0.9 across all surgical specialties except for cardiac and vascular surgery. Adjusted and unadjusted models showed similar C-statistics for mFI-5 and 11 and strong predictive ability for mortality and post-operative complications. Predictive value for thirty day readmission was weak for both the mFI-11 and the mFI -5.
Conclusion:The mFI-5 is an equally effective predictor as the mFI-11 in all subspecialties and is a strong predictor of mortality and post-operative complication. It has credibility for future use to study frailty within the NSQIP database. It also has potential in other databases and for clinical use.