Methods for time trend analysis of cancer incidence rates
To introduce statistical methods of time trend analysis on cancer rates. Cancer incidence data collected by the Shanghai Cancer Registry during 1991 to 1999 was used in the analysis to calculate the crude and age-adjusted rates, percent changes (PCs) and annual percent changes (APCs). APCs were estimated by a linear regression of the logarithm on the incidence rates during the nine years. It also introduced a method for partitioning a linear trend in age-adjusted rates into site-specific contributions to the overall floating trend. 95% confidence intervals for the APCs and contributions were described in the paper. A decreasing rates were observed for cancers of stomach and esophagus among both men and women in urban Shanghai from 1991 to 1999. The increasing rates among men would include cancers of colon, rectum, gall bladder, pancreas, prostate, urinary bladder, kidney and leukemia. The rates of cancers among women increased for colon, rectum, lung, breast, gall bladder, endometrium, ovary, urinary bladder and kidney. The changes of above cancers over time were statistically significant (P < 0.05 or P < 0.01), but rates for other cancer sites changed little. The APCs (weighted method) and contributions for the cancers of stomach, esophagus, colon, rectum and prostate were -2.99% and -65.72%, -2.90% and -17.07%, 12.30% and 21.46%, 2.94% and 18.62%, and 3.11% and 15.09% among men, and -6.05% and -39.55%, -1.08% and -35.19%, 2.81% and 28.64%, and 3.69% and 15.70% for the cancers of stomach, esophagus, breast and colon in women, respectively. APC, and related statistics could be used to describe and analyze the time trend of cancer rates rather than PC or/and graphical method alone.