Table 3.

Regression parameter estimates (standard errors) for the direct modeling of the crude probabilities of death from cancer and other causes, as obtained with three models using pseudo-observations with link functions: cloglog, identity, and log, and assuming an independence working covariance structure.

	cloglog		Log		Identity
	Cancer	Other causes	Cancer	Other causes	Cancer	Other causes
(Intercept)	-2.313(0.072)	-5.835(0.149)	-2.178(0.055)	-5.743(0.14)	0.099(0.01)	0.002(0.001)
t = 969 days	0.626(0.028)	0.721(0.018)	0.477(0.023)	0.702(0.017)	0.098(0.004)	0.008(0)
t = 1826 days	0.845(0.031)	1.207(0.032)	0.612(0.025)	1.163(0.029)	0.14(0.004)	0.017(0.001)
t = 2132 days	0.881(0.032)	1.339(0.036)	0.633(0.026)	1.286(0.033)	0.147(0.004)	0.021(0.001)
t = 2487 days	0.927(0.033)	1.474(0.04)	0.659(0.026)	1.41(0.036)	0.156(0.005)	0.025(0.001)
Age	0.452(0.014)	0.702(0.034)	0.33(0.009)	0.67(0.032)	0.101(0.003)	0.017(0.001)
Deprivation 2	-0.025(0.091)	0.158(0.147)	-0.016(0.067)	0.151(0.141)	-0.001(0.014)	0.002(0.002)
Deprivation 3	0.134(0.087)	0.099(0.137)	0.085(0.063)	0.094(0.131)	0.031(0.014)	0.003(0.002)
Deprivation 4	0.2(0.083)	0.135(0.126)	0.125(0.06)	0.125(0.121)	0.047(0.014)	0.005(0.002)
Deprivation 5	0.186(0.082)	0.223(0.129)	0.12(0.06)	0.203(0.124)	0.043(0.013)	0.008(0.002)

(Age at diagnosis-47 (mean age in the data set))/10