Table 2. Estimates of Temporal Associations of Alcohol and Tobacco Consumption With Overall, Male, and Female Cancer Mortality Based on 3 Different Lag Models.

Lag Model	Cancer, Coefficient (95% CI)
	Male	Female	Total
Model with 20-y geometric lags
Alcohol	0.005 (−0.036 to 0.046)	0.031 (−0.012 to 0.074)	0.014 (−0.021 to 0.049)
Tobacco	−0.078 (−0.303 to 0.147)	−0.154 (−0.401 to 0.093)	−0.093 (−0.289 to 0.103)
Health expenditure (5-y geometric lag)	−0.105 (−0.195 to −0.015)a	−0.161 (−0.212 to −0.110)a	−0.124 (−0.200 to −0.048)a
Constant	−0.000 (−0.016 to 0.016)	−0.001 (−0.015 to 0.013)	−0.001 (−0.015 to 0.013)
Model specification	0,1,0	0,1,0	0,1,0
Box-Ljung Q (lag 10)b	7.787	8.388	6.147
P value	.65	.59	.80
R2	0.155	0.318	0.260
Model with 20-y Skog lags
Alcohol	0.061 (0.002 to 0.120)a	0.018 (−0.037 to 0.073)	0.038 (−0.007 to 0.083)
Tobacco	−0.239 (−0.502 to 0.024)	−0.114 (−0.363 to 0.135)	−0.170 (−0.368 to 0.028)
Health expenditure (5-y Skog lag)	−0.158 (−0.254 to −0.062)c	−0.105 (−0.195 to −0.015)a	−0.128 (−0.201 to −0.055)c
Constant	−0.005 (−0.023 to 0.013)	−0.002 (−0.018 to 0.014)	−0.003 (−0.017 to 0.011)
Model specification	0,1,0	0,1,0	0,1,0
Box-Ljung Q (lag 10)b	10.054	7.362	6.999
P value	.44	.69	.73
R2	0.36	0.16	0.35
Model with 20-y cross-correlation lags
Alcohol	0.043 (0.012 to 0.074)a	0.035 (0.010 to 0.060)a	0.038 (0.014 to 0.062)c
Tobacco	0.266 (0.115 to 0.417)c	0.083 (−0.066 to 0.232)	0.151 (0.078 to 0.224)d
Health expenditure (5-y cross-correlation lag)	−0.046 (−0.148 to 0.056)	−0.042 (−0.151 to 0.068)	−0.047 (−0.127 to 0.033)
Constant	0.010 (0.000 to 0.020)a	−0.002 (−0.012 to 0.008)a	0.005 (−0.003 to 0.013)
Model specification	1,1,0	0,1,1	1,1,1
Box-Ljung Q (lag 10)b	9.546	4.996	11.658
P value	.39	.84	.17
R2	0.582	0.467	0.589

^aP < .05.

^bThe Box-Ljung Q test is a diagnostic tool used to test the lack of fit of a time series model, and a P value of the Box-Ljung Q test greater than .10 indicates the test rejects the null hypothesis of lack of fit of the time series model.

^cP < .01.

^dP < .001.