Skip to main content

Table 4 Association of %TTL (independent variable) with FBgap (dependent variable)

From: Association of continuous renal replacement therapy downtime with fluid balance gap and clinical outcomes: a retrospective cohort analysis utilizing EHR and machine data

 

β coefficient

95%CI

p value

Interaction between %TTL and total alarms

Model 1

− 0.49

− 0.80 to − 0.19

0.002*

Significant (p < 0.001)

Model 2

− 0.31

− 0.53 to − 0.09

0.006

–

 Total alarms: 0–2

− 0.41

− 0.94 to 0.13

0.134

–

 Total alarms: 3–5

− 0.06

− 0.35 to 0.22

0.665

–

 Total alarms: 6 + 

0.72

0.34 to 1.09

 < 0.001*

–

Model 3

− 0.48

− 0.78 to − 0.18

0.002*

Significant (p < 0.001)

Model 4

− 0.26

− 0.49 to − 0.04

0.021

–

 Total alarms: 0–2

− 0.37

− 0.88 to 0.14

0.157

–

 Total alarms: 3–5

− 0.11

− 0.41 to 0.19

0.480

–

 Total alarms: 6 + 

0.87

0.48 to 1.26

 < 0.001*

–

  1. The β coefficient represents the change in FBgap for each 1% increase in %TTL
  2. Model 1: A linear regression with FBgap as the dependent variable and %TTL as the independent variable, as well as the interaction term between %TTL and total alarms
  3. Model 2: A linear regression with FBgap as the dependent variable and %TTL as the independent variable, without including the interaction term. In addition, a subgroup analysis was performed for each category of total alarms (0–2, 3–5, 6 +)
  4. Model 3: A linear regression with FBgap as the dependent variable and %TTL as the main independent variable, including the interaction term between %TTL and total alarms. Other covariates included age, sex, race, CCI, SOFA score at CRRT initiation, use of MV at CRRT initiation, use of circulatory support devices (any of IABP, VAD, ECMO) at CRRT initiation, and %fluid overload as independent variables
  5. Model 4: A linear regression with FBgap as the dependent variable and %TTL as the main independent variable, without including the interaction term. The covariates included the same as in Model 3. In addition, a subgroup analysis was performed for each category of total alarms (0–2, 3–5, 6 +)
  6. CI, confidence interval; FBgap, fluid balance gap; %TTL, percent treatment time loss; CCI, Charlson comorbidity index; SOFA, sequential organ failure assessment; CRRT, continuous renal replacement therapy; MV, mechanical ventilation; IABP, intra-aortic balloon pump; VAD, ventricular assist device; ECMO, extracorporeal membrane oxygenation
  7. *After applying the Bonferroni correction, the significance threshold was set to p < 0.005, as 10 models were used
Back to article page

Journal of Intensive Care

ISSN: 2052-0492