Parameter | Relative excess hazard ratio (95% CI) |
---|---|
Intercept | 1.561 (1.246–1.955) |
Year of diagnosis | |
1990–1994 | Reference |
1995–1999 | 0.755 (0.622–0.917) |
2000–2004 | 0.724 (0.603–0.870) |
2005–2009 | 0.732 (0.615–0.871) |
Follow-up after diagnosis, yr | |
1 | Reference |
2 | 0.601 (0.552–0.654) |
3 | 0.519 (0.461–0.584) |
4 | 0.551 (0.477–0.637) |
5 | 0.359 (0.286–0.450) |
Age at diagnosis, yr | |
< 60 | Reference |
60–69 | 0.967 (0.750–1.246) |
70–79 | 1.347 (1.032–1.757) |
≥ 80 | 1.698 (1.165–2.475) |
Sex | |
Female | Reference |
Male | 1.123 (0.965–1.307) |
Charlson–Deyo comorbidity index | |
0 | Reference |
1 | 0.788 (0.722–0.859) |
2 | 0.547 (0.492–0.607) |
≥ 3 | 0.496 (0.442–0.558) |
No hospitalization record | 0.520 (0.478–0.566) |
HCC treatment | |
Curative (yes v. no) | 0.223 (0.199–0.250) |
Non-curative (yes v. no) | 0.651 (0.590–0.717) |
Palliative care (yes v. no) | 1.696 (1.470–1.957) |
No treatment (yes v. no) | 1.403 (1.223–1.608) |
Interactions | |
Age at diagnosis × sex (male), yr | |
Female/< 60 | Reference |
60–69 | 1.071 (0.851–1.347) |
70–79 | 1.139 (0.889–1.458) |
≥ 80 | 1.923 (1.335–2.772) |
Age at diagnosis × year of diagnosis, yr | |
1990–1994/< 60 | Reference |
1995–1999/60–69 | 0.827 (0.644–1.062) |
2000–2004/60–69 | 0.870 (0.680–1.114) |
2005–2009/60–69 | 0.887 (0.694–1.133) |
1995–1999/70–79 | 1.014 (0.791–1.299) |
2000–2004/70–79 | 0.825 (0.650–1.046) |
2005–2009/70–79 | 1.018 (0.801–1.295) |
1995–1999/≥ 80 | 1.142 (0.799–1.632) |
2000–2004/≥ 80 | 0.784 (0.568–1.084) |
2005–2009/≥ 80 | 0.750 (0.550–1.023) |
Age at diagnosis × palliative care, yr | |
No palliative care/< 60 | Reference |
Palliative care, yr | |
60–69 | 1.146 (0.863–1.523) |
70–79 | 1.674 (1.236–2.267) |
≥ 80 | 2.275 (1.480–3.499) |
Note: CI = confidence interval, model deviance = 3844.21, df = 2836. Interactions between age at diagnosis and sex (p = 0.009), year of diagnosis (p = 0.003) and palliative care (p < 0.001). Interactions between age at diagnosis and follow-up after diagnosis, Charlson–Deyo comorbidity index, curative treatment, noncurative treatment and no treatment were not significant. *Generalized linear model using Poisson error structure. |