Skip to main content

Table 4 Excess hazard ratios with Cox and RBS model. Results of the application on prostate cancer data

From: Correcting for misclassification and selection effects in estimating net survival in clinical trials

Variable & modalities

Cox model

RBS model

EHR [95% CI]

EHR [95% CI]

Treatment

High-dose DES

0.43 [0.30; 0.64]

0.62 [0.40; 0.96]

Low-dose DES

Age (centered)

0.97 [0.95; 1.00]

0.98 [0.95; 1.01]

Weight index a

< 80

1.45 [0.75; 2.81]

2.05 [1.04; 4.05]

80–99

1.29 [0.87; 1.92]

1.73 [1.13; 2.66]

≥100

  

Performance rating

Limited activity

1.49 [0.87; 2.52]

1.62 [0.98; 2.66]

Normal activity

  

History of cardiovascular disease

Yes

0.86 [0.58; 1.28]

1.70 [1.15; 2.52]

No

  

Haemoglobin

< 9 g/100 mL

9.13 [3.64; 22.88]

5.66 [2.12; 15.09]

9–12 g/100 mL

1.01 [0.64; 1.59]

1.32 [0.86; 2.01]

≥12 g/100 mL

  

Size of primary lesion

≥30 cm2

4.02 [2.62; 6.16]

3.25 [1.91; 5.50]

< 30 cm2

  

Gleason stage/grade category

> 10

9.58 [5.50; 16.67]

2.22 [1.31; 3.77]

≥10

  

Selection effect ( \( \widehat{\alpha} \) )

 

1.51 [0.89; 2.56]

  1. aWeight (kg) - Height (cm) + 200 - RBS Rescaled B-spline model, CI confidence interval, EHR excess hazard ratio, DES diethylstilbestrol – α: parameter of the RBS model used to rescale the all-cause mortality in participants in a clinical trial