CAN THE 5-ITEM MODIFIED FRAILTY INDEX AND CHARLSON COMORBIDITY INDEX BE USED TO OPTIMIZE PATIENT SELECTION FOR TOTAL KNEE ARTHROPLASTY?

Comorbidities are one of the well-recognized predictors of postoperative prognosis and can be used for risk stratification and patient optimization. This is essential to counter the exponentially rising healthcare costs associated with postoperative complications. Charlson Comorbidity Index and 5-Item Modified Frailty Index are proposed as clinical tools to predict postoperative outcomes based on patient comorbidities in a standardized way. However, there is no literature evaluating the efficacy of these tools for Total Knee Arthroplasty (TKA) patients. We conducted a retrospective chart review on TKA patients at a tertiary care hospital from 2014 to 2020. Measures of diagnostic test accuracy (Sensitivities, specificities, Positive Predictive Values (PPV), and Negative Predictive Values (NPV)) were calculated, using postoperative outcomes as the gold standard. The Receiver Operator Characteristics (ROC) Curves were generated, adjusting for age and BMI; CCI and mFI-5 were used as continuous variables for ROC. A total of 1,061 TKA patients were included in this study. The mean CCI score was 1.32 ± 1.01, and in categorized CCI, 110 (10.4%) patients had high CCI and 951 (89.6%) had low CCI. The mean mFI-5 score was 1.04 ± 0.76; Three-hundred-twelve (29.4%) had high mFI-5 and 749 (70.6%) had low mFI-5. Both CCI and mFI-5 had low sensitivity and PPV but higher specificities and NPV (Table 1). In ROC, both CCI and mFI-5 had significant AUC for postoperative complications and 30-day mortality only (table 2). CCI and mFI-5 both had low sensitivities and positive predictive values, ranging below 30%, for postoperative complications, reoperation, 30-day readmission, and 30-day mortality. This data suggests that these indices are not appropriate as screening tools to predict prognosis after TKA. However, the specificity and NPV of both the indices range in higher 70%, making these indices very good tools to rule out the poor prognosis. Using mFI-5 and CCI as continuous variables, ROC curves indicate that CCI is a poor indicator of post-operative complications. However, both mFI-5 and CCI are acceptable tools to predict 30-day mortality, only when adjusted for the patient's age and BMI. It is important to note that our data suggest that these indicators are poor tools to predict 30-day readmissions and reoperation rates. For any figures or tables, please contact the authors directly.