Table 3 Results of forward selection using a set of predictor variables based on improvement in concordance (δC).

From: Blunted rest-activity circadian rhythm increases the risk of all-cause, cardiovascular disease and cancer mortality in US adults

Variable

Cumulative concordance

δC

 

Stopping rule: δC ≥ 0.005

Age

0.8225

 

0.089 (0.079, 0.098)

RA

0.8315

0.0090

− 0.377 (− 0.507, − 0.248)

General health

0.8374

0.0059

 

(Very) good/excellent

  

− 0.799 (− 0.985, − 0.614)

Fair/poor

  

Ref.

Stopping rule: δC ≥ 0.001

Age

0.8225

 

0.082 (0.074, 0.092)

RA

0.8315

0.0090

− 0.333 (− 0.481, − 0.185)

General health

0.8374

0.0059

 

(Very) good/excellent

  

− 0.590 (− 0.814, − 0.367)

Fair/poor

  

Ref.

CVD

0.8408

0.0034

0.325 (0.044, 0.606)

Physical activity

0.8431

0.0023

 

Poor

  

Ref.

Intermediate

  

− 0.577 (− 0.917, − 0.237)

Ideal

  

− 0.425 (− 0.778, − 0.073)

Ratio of family income to poverty

0.8451

0.0020

− 0.114 (− 0.214, − 0.013)

Race

0.8480

0.0028

 

NH White

  

Ref.

NH Black

  

− 0.237 (− 0.619, 0.145)

Mexican American

  

− 0.411 ( − 0.789, − 0.025)

Other

  

− 0.520 (− 0.878, − 0.162)

  1. RA Relative amplitude; NH Non-Hispanic. Variables were listed in order of their inclusion into the model and the tenfold cross-validated concordance from a model which included all variables up to the current variable that was presented in the second column. The fourth column presented the estimated coefficient and 95% CIs obtained from the final models fit for each stopping rule. RA has been divided into quartiles.
Back to article page