In the past year, my research on this project was concentrated in three areas: (1) analyses of long-term linear trends in cancer incidence and mortality, (2) analyses of cancer incidence and mortality trends that allow for curvature, and (3) comparisons of statewide (Pennsylvania) and national (US) data on the incidence of non-Hodgkin's lymphoma. Research in these areas involved age-period-cohort and joinpoint analyses. This work is collaborative with researchers at the University of Pittsburgh. The three areas of research are described in more detail below.? ? Area 1:? ? We examined cancer incidence and mortality trends in the United States using population-based data from the SEER program. The incidence data were from 1975-2004 and the mortality data were from 1970-2004. We used APC models to investigate the effects of age group, time period, and birth cohort on the rates of cancer in three categories: cancers related to tobacco use, cancers detectable by screening, and all other cancers. By design, time trends observed in the residual category cannot be attributed to temporal changes in cigarette smoking or screening practices.? ? Incidence and mortality rates were analyzed with respect to long-term trends, which can reflect changes in cancer risk factors. Studying trends in light of known risk factors may indicate unexplained cancer patterns. Specifically, we focused on linear trends in the log-transformed rates, summarized by average annual percentage changes and generational risks. The latter concept assesses relative cancer rates between one point in time and another 25 years (i.e., one """"""""generation"""""""") earlier.? ? Among whites over a 25-year span, cancer incidence in the residual category increased 34% in men and 23% in women, while mortality decreased 14% in men and 18% in women, with similar results among blacks. Changes in tobacco use and screening practices do not completely explain observed changes in cancer trends over the last three decades. Hopefully a focus on the cancers in the residual category will provide clues about the causes of these unexplained increases in cancer incidence.? ? A manuscript was submitted to Environmental Health Sciences.? ? Area 2:? ? We extended our APC analysis to account for non-linear trends. Specifically, we assumed a constant curvature model for the log-transformed incidence and mortality rates, of which linearity is a special case. Again we focused on a residual cancer category, but in addition to tobacco-related and screen-detectable cancers, we also excluded cancers associated with HIV infection.? ? Incidence rates increased in every race-sex group, and factors related to both time period and birth cohort membership appeared to accelerate these increases in women. Mortality rates decreased in black and white men and women, with the declines decelerating in white women but accelerating in the other race-sex groups. Increasing incidence of cancers not related to tobacco, screening, or HIV may signify increasing cancer risks, changing diagnostic practices, or better case ascertainment. Declining mortality may signify improvements in cancer care.? ? A manuscript was submitted to Cancer.? ? Area 3:? ? We investigated the incidence of non-Hodgkin's lymphoma (NHL) in greater detail, studying the effects of both temporal and demographic factors. We used data from the SEER program and the Pennsylvania Cancer Registry (PCR), which is not part of the SEER program. Data from the PCR were available from 1985 to 2004, so we restricted our attention to the same 20 years for the SEER data. Results based on the national (SEER) and statewide (PCR) data were compared and contrasted.? ? The incidence of NHL rose between 1985 and 2004 among black and white men and women, both nationally and in Pennsylvania. The magnitudes of the increases were similar, ranging from 1.5% to 3.2% across the four race-sex groups, except among white men, where the SEER increases were smaller than the PCR increases. Diffuse lymphoma appeared to be the most important component of this increase. In Pennsylvania, NHL incidence was higher in counties with a greater percentage of urban residents.? ? A manuscript is in preparation and should be ready to submit to a scientific journal soon.
