Report to the Nation, 19752004/Espey et al.
FIGURE 1. States and Contract Health Service Delivery Area (CHSDA) counties used in cancer incidence analyses for the American Indian/Alaska Native popu- lation, by Indian Health Service region.
adults aged 18 years. Questions also were asked regarding participation in mammography screening among women aged 40 years, colorectal cancer screening among persons aged 50 years, and pros- tate-specific antigen (PSA) testing among men aged
50 years who reported no history of prostate can-
cer. Adults aged 20 years were asked about self- reported obesity prevalence.
The distribution of stage of disease at diagnosis for cancers of the breast, prostate, colon and rectum, and cervix uteri (cervix) was examined for non-His- panic white (NHW) and AI/AN populations region- ally and in all regions combined. Cancer incident cases were staged using the 2000 SEER Summary Stage. To eliminate the effect of different staging sys- tems both before 2001 and after 2003, only indivi- duals who were diagnosed from 2001 through 2003 were included for cancer stage statistics.38
Statistical Analysis For all populations, cancer incidence and death rates were expressed per 100,000 persons and were age- adjusted by 19 age groups (aged < 1 year, ages 1–4 years, ages 5–9 years, . . ., ages 80–84 years, aged 85 years) to the 2000 U.S. standard population. Age adjustment summarizes age-specific rates and elimi- nates the effect of differences in age composition among the populations being compared.39 The AI/
AN population is younger on average than the NHW
Rates, standard errors, and 95% confi-
dence intervals (95% CIs) were generated using SEER*Stat software, version 6.3 (available at URL: http://www.seer.cancer.gov/seerstat/).20 All incidence and death rates that were calculated with < 16 observations were suppressed. Similarly, the annual percent change (APC) statistic was suppressed if it was based on < 10 cases for at least 1 year within the time interval.
Long-term cancer incidence and death trends (1975 through 2004) were described using joinpoint analysis, including the APC for each interval, for all races/ethnicities combined.41 Statistical significance was set at P < .05. The observed incidence trends and those adjusted for reporting delay (mostly affecting recent years) were presented using models based on long-term reporting patterns in SEER.42 Nondelay-adjusted and delay-adjusted APCs are pre- sented; however, only delay-adjusted results are referred to in the text. APC analysis was used to describe fixed interval trends (1995 through 2004); the incidence data for the fixed interval APCs were not adjusted for reporting delay. For both methods of trend analysis, the terms ‘‘increase’’ or ‘‘decrease’’
were used when the slope of the trend
cally significant ‘‘stable’’ or ‘‘level’’
(P < .05); were used.
was statisti- the terms