The Association of Preoperative Frailty and Postoperative Delirium: A Meta-analysis

BACKGROUND: Both frailty and postoperative delirium (POD) are common in elective surgical patients 65 years of age and older. However, the association between preoperative frailty and POD remains difficult to characterize owing to the large number of frailty and POD assessment tools used in the literature, only a few of which are validated. Furthermore, some validated frailty tools fail to provide clear score cutoffs for distinguishing frail and nonfrail patients. We performed a meta-analysis to estimate the relationship between preoperative frailty and POD. METHODS: We searched several major databases for articles that investigated the relationship between preoperative frailty and POD in patients with mean age ≥65 years who were undergoing elective, nonemergent inpatient surgery. Inclusion criteria included articles published in English no earlier than 1999. Both preoperative frailty and POD must have been measured with validated tools using clear cutoff scores for frailty and delirium. Articles were selected and data extracted independently by 2 researchers. Risk of bias (ROBINS-I) and presence of confounders were summarized. Odds ratios (ORs) for POD associated with frailty relative to nonfrailty were computed with adjusted ORs when available. Original estimates were pooled by random effects analysis. Statistical significance was set at 2-sided P < .05. RESULTS: Nine studies qualified for meta-analysis. The Fried score or a modified version of it was used in 5 studies. Frailty prevalence ranged from 18.6% to 56%. Delirium was assessed with the Confusion Assessment Method (CAM) or Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) in 7 studies, Delirium Observation Scale in 1 study, and Intensive Care Delirium Screening Checklist in 1 study. The incidence of POD ranged from 7% to 56%. ROBINS-I risk of bias was low in 1 study, moderate in 4 studies, serious in 3 studies, and critical in 1 study. Random effects analysis (n = 794) of the OR for POD in frail versus nonfrail patients based on adjusted OR estimates was significant with an OR of 2.14 and a 95% confidence interval of 1.43–3.19. The I 2 value was in the low range at 5.5, suggesting small variability from random effects. Funnel-plot analysis did not definitively support either the presence or absence of publication bias. CONCLUSIONS: This meta-analysis provides evidence for a significant association between preoperative frailty and POD in elective surgical patients age 65 years or older.