Patient-Surgeon Agreement in Assessment of Frailty, Physical Function, & Social Activity

The shared decision-making process between surgical providers and patients relies on a joint understanding about the risks of different treatment options based on a patient's individual health state. However, it is unclear whether a patient's perception of their own condition is congruent with the health state assigned by their surgical providers. This study was designed to compare provider assessment of frailty versus patient-reported outcome (PRO) measures of their own frailty status, physical function, and social activity level.

We prospectively assessed patients presenting to a vascular surgery clinic at an academic institution between May 2018 and June of 2019. Before clinic examination, patients completed PROs of their frailty status (Frail Non-Disabled survey), physical function (patient-reported outcome measurement information system [PROMIS] v1.2), and social activity level (PROMIS v2.0). Next, each patient's frailty status and overall health were scored by a surgical provider using the 9-point Clinical Frailty Scale, a validated frailty assessment tool that incorporates their functional status and level of activity. The correlation between the provider and PROs for frailty, physical function, and social activity was determined using the Spearman rank test, sensitivity/specificity tests, and receiver operating curves. Logistic regression models were used to predict 1-y mortality after assessment.

A total of 118 patients were evaluated in clinic (50% male with mean age of 60 y), including 35 (30%) who were categorized as being frail by the surgical provider. In comparison, the same patients were much more likely to self-report as having low physical function (73%), being frail or disabled (79%), and/or unable to engage in social activities (78%). Although there was high sensitivity (89%) between a provider's and PROs for frailty, the specificity was low (26%) resulting in a receiver operating curve area of 0.57. Overall, there was low correlation between PROs for frailty (r = 0.16), physical function (r = 0.21), and social activities (r = 0.21) when compared with a provider's assessment of patient frailty. Models using PROs for frailty had better discrimination for predicting 1-y mortality (c-statistic: 0.72) than those using the Clinical Frailty Scale (c-statistic: 0.62).

Patients are more likely to self-report being frail, having low physical function, and limited social activity than what is detected by their surgical providers. These findings suggest that low levels of patient activity and its associated risk may often be underappreciated by surgical providers. Efforts are needed to improve how PROs are incorporated into surgical decision-making and outcome assessment.