Which Frailty Measure Best Predicts Outcomes Following Urogynecologic Surgery?

Objectives: Multiple claims-based frailty measures have been developed and implemented in administrative database analysis for surgical outcomes research. However, the relative performance of different claims-based frailty measures in predicting adverse outcomes remains poorly described. The objective was to estimate the incremental prognostic value of three validated claims-based frailty measures: the claims-based frailty index (CFI), electronic frailty index (EFI) and hospital frailty risk score (HFRS); and one comorbidity measure, the combined comorbidity index (CCI), for predicting readmission, emergency department encounters, and non-home discharge following female pelvic reconstructive surgery.

Methods: A retrospective analysis was performed using the 2016-2019 Centers for Medicare and Medicaid Services 5% Limited Dataset. Individuals undergoing a reconstructive pelvic surgery procedure, inclusive of apical suspension, isolated compartment repair and obliterative repair, during the analysis period were assessed. The primary outcome was hospital readmission within 90 days of surgery, as this reflects the global coverage period for all included procedures. Secondary outcomes were emergency department encounter within 90 days of surgery and non-home discharge following surgery. For each outcome, a logistic regression model was fit with and without the frailty measure. The fraction of new prognostic information attributable to each frailty measure within the model was quantified using likelihood ratio chi-square statistic, Akaike information criterion, variance in model-estimated risk and pseudo-R2.

Results: 6896 female pelvic reconstructive surgery cases were identified based on procedural coding, of which 6790 (98.5%) had adequate records for evaluation. Rates of frailty varied based on the measure used, with 65.8 and 68.7% of the population being designated as frail or elevated risk by the EFI and HFRS respectively, while the CFI classified only 38.8% as prefrail or frail (Figure 1). Hospital admission occurred among 6.1% (95% CI: 5.5-6.7), emergency department encounter among 13.8% (95% CI: 12.9-14.6) and non-home discharge among 1.2% (95% CI: 1.0-1.5). The fraction of new prognostic information predicting hospital readmission was: CCI (48-57%), CFI (49-55%), EFI (49-54%) and HFRS (51-61%). This can be visualized as increased variance before and after inclusion of the measure (Figure 2). The fraction of new information for predicting an emergency department encounter and non-home discharge ranged from 37-73%. While performance was overall similar, the HFRS appeared to contribute the greatest fraction of new prognostic information among the three frailty measures.

Conclusions: The CCI, CFI, EFI and HFRS all provided substantial additional prognostic information for the prediction of hospital readmission, emergency department encounter and non-home discharge following pelvic reconstructive surgery. Among those frailty measures analyzed in this study, the HFRS appears to have the greatest amount of incremental prognostic information and may be favored over the other assessed measures.

Figure 1: Highlighted cells represent concordant frailty designations and non-highlighted discordant. Data presented as n(%).