Skip to main content

Table 6 Multivariate linear regression analysis (biopsy group vs. no-biopsy group) in the cohort before and after propensity score matching

From: Biopsy in emergency gastroscopy does not increase the risk of rebleeding in patients with Forrest I acute nonvariceal upper gastrointestinal bleeding combined with suspected malignant gastric ulcer: a multicenter retrospective cohort study

 

Before matching

Hospitalization costs

After matching

Hospitalization costs

B

95% CI

P value

B

95% CI

P value

Unadjusted

0.447

− 0.178 to 1.072

0.161

− 0.817

− 1.616 to − 0.017

0.045

Adjusted for all covariatesa

− 1.476

− 1.629 to − 1.324

 < 0.001

− 1.424

− 1.584 to − 1.263

 < 0.001

Multiple imputationb

− 1.481

− 1.602 to − 1.361

 < 0.001

   
  1. aAll variables in Table 2 included as covariates. For model with Hospitalization costs
  2. bMultivariate linear regression was performed for all the patients before matching after multiple imputation for missing data using multivariate normal distribution