Skip to main content

Table 5 Two-Stage Least Squares (2SLS) Estimates by Benefit and Detriment by Ex-Post Reference Classes for Third-level CART Splits

From: Assessing the ability of an instrumental variable causal forest algorithm to personalize treatment evidence using observational data: the case of early surgery for shoulder fracture

  1. R = early surgery rate; (*,*) = the inter-quintile range in early surgery rates across patients in the node; F = F-statistic of the effect of the instrument on early surgery choice in the first stage; IVE = the 2SLS estimated absolute effect of early surgery on the probability of the respective outcome; N = the number of patients in the node; D_IVE the study population wide 2SLS absolute effect of early surgery on detriment; and B_IVE the study population wide 2SLS absolute effect of early surgery on benefit. CCI: Charlson Comorbidity Index score based on Medicare claims in the year prior to index proximal humerus fracture. FRI: Frailty Risk Index score based on Medicare claims in the year prior to index proximal humerus fracture
  2. *, **, *** p < .1, p < .05, p < .01 respectively
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us