SPECIAL ARTICLE ARTICLE Annual Report to the Nation on the Status of Cancer, 1975 – 2002, Featuring Population-Based Trends in Cancer Treatment Brenda K. Edwards , Martin L. Brown , Phyllis A. Wingo , Holly L. Howe , Elizabeth Ward , Lynn A. G. Ries , Deborah Schrag , Patricia M. Jamison , Ahmedin Jemal , Xiao Cheng Wu , Carol Friedman , Linda Harlan , Joan Warren , Robert N. Anderson , Linda W. Pickle D o w clinical advances to cancer prevention, detection, and uni- n lo Background: The American Cancer Society (ACS), the Cen- formly high quality of care in all areas and populations of the a d ters for Disease Control and Prevention (CDC), the National United States. [J Natl Cancer Inst 2005;97:1407 – 27] ed Cancer Institute (NCI), and the North American Association fro m of Central Cancer Registries (NAACCR) collaborate annu- h ally to provide information on cancer rates and trends in the The American Cancer Society (ACS), the Centers for Disease ttp s United States. This year’s report updates statistics on the 15 Control and Prevention (CDC), the National Cancer Institute ://a (NCI), and the North American Association of Central Cancer c most common cancers in the fi ve major racial/ethnic popula- a d Registries (NAACCR) collaborate each year to produce a report e tions in the United States for 1992 – 2002 and features p opulation- m to the nation on the current status of cancer in the United States. ic based trends in cancer treatment. M ethods: The NCI, the The initial report, in 1998, documented the fi rst sustained decline .ou CDC, and the NAACCR provided information on cancer p in cancer death rates since national record keeping was instituted .c cases, and the CDC provided information on cancer deaths. in the 1930s ( 1 ) . Subsequent reports have generally confi rmed om Reported incidence and death rates were age-adjusted to the this fi nding and provided updated information ( 2 – 7 ) . /jnc 2ra0t0e0s fUo.rS .fi xsteadn idnatredrv aplosp wulaast ieosnt,i maantneuda bl yp elirnceeanrt rcehgarnesgsei oinn, the T inhcei d2e0n0c4e r aenpdo rdt e( a 7t )h f roactuess efdo ro tnh ec a1n5c emr ossutr cvoivmaml torenn cdasn acnedrs oinn i/article and annual percent change in trends was estimated with join- each of the fi ve major racial/ethnic populations in the United -a b point regression analysis. Population-based treatment data States — white, black, Asian/Pacifi c Islander (API), American stra were derived from the Surveillance, Epidemiology, and End Indian/Alaska Native (AI/AN), and Hispanic/Latino. In this ct/9 Results (SEER) Program registries, SEER-Medicare linked report, we update the cancer incidence and death rates and 7 /1 databases, and NCI Patterns of Care/Quality of Care studies. 9 /1 Results: Among men, the incidence rates for all cancer sites 4 0 7 combined were stable from 1995 through 2002. Among women, /2 5 the incidence rates increased by 0.3% annually from 1987 Affi liations of authors: Division of Cancer Control and Population Sciences, 2 1 through 2002. Death rates in men and women combined National Cancer Institute, Bethesda, MD (BKE, MLB, LAGR, LH, JW, LWP); 38 4 decreased by 1.1% annually from 1993 through 2002 for all Division of Cancer Prevention and Control, National Center for Chronic Disease b cancer sites combined and also for many of the 15 most com- PArtelavnetnat,i oGnA a n(dP AHWea, lPthM PJr,o CmFo)t;i oNno, rCthe nAtemrse rfiocra nD Aisessaosec iCatoionntr oolf aCnedn Ptrraelv Cenatniocenr, y gue mon cancers. Among women, lung cancer death rates in- Registries, Springfi eld, IL (HLH); Epidemiology and Surveillance Research st o creased from 1995 through 2002, but lung cancer incidence Department, American Cancer Society, Atlanta, GA (EW, AJ); Memorial Sloan n 0 rates stabilized from 1998 through 2002. Although results of Kettering Cancer Center, New York, NY (DS); Louisiana State University Health 8 Science Center, Louisiana State University, New Orleans, LA, and North American A cancer treatment studies suggest that much of contemporary p cancer treatment for selected cancers is consistent with evidence- Association of Central Cancer Registries, Springfi eld, IL (XCW); Division of ril 2 Vital Statistics, National Center for Health Statistics, Centers for Disease Control 0 based guidelines, they also point to geographic, racial, and Prevention, Hyattsville, MD (RNA). 19 economic, and age-related disparities in cancer treatment. Correspondence to: Brenda K. Edwards, PhD, Division of Cancer Control and Conclusions: Cancer death rates for all cancer sites combined Population Sciences, National Cancer Institute, 6116 Executive Blvd., Suite 504, Bethesda, MD 20892 – 8315 (e-mail: [email protected] ). and for many common cancers have declined at the same See “ Notes ” following “ References. ” time as the dissemination of guideline-based treatment into DOI: 10.1093/jnci/dji289 the community has increased, although this progress is not © The Author 2005. Published by Oxford University Press. All rights reserved. shared equally across all racial and ethnic populations. Data The online version of this article has been published under an Open Access model. from population-based cancer registries, supplemented by Users are entitled to use, reproduce, disseminate, or display the Open Access linkage with administrative databases, are an important re- version of this article for non-commercial purposes provided that: the original source for monitoring the quality of cancer treatment. Use of authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation this cancer surveillance system, along with new developments details given; if an article is subsequently reproduced or disseminated not in its in medical informatics and electronic medical records, may entirety but only in part or as a derivative work this must be clearly indicated. facilitate monitoring of the translation of basic science and For commercial re-use, please contact: [email protected]. Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 SPECIAL ARTICLE 1407 identify trends in these rates in the United States for men and for most common cancers among all racial and ethnic populations women, separately. combined, we used SEER incidence data from nine registries There has been a growing concern that not all cancer patients ( 23 ) , which cover approximately 10% of the U.S. population, and in the United States receive the full benefi t of cancer treatments 100% of the U.S. population death data from the CDC. For the that have been shown to be effective and appropriate on the basis more recent (i.e., 1992 through 2002) trend analyses of incidence of the accumulated evidence from controlled clinical trials ( 8 ) . and death rates for the 15 most common cancers in each major In response to this concern, there has been an increased effort in racial and ethnic population (i.e., white, black, API, AI/AN, and recent years to monitor trends in cancer treatment, to understand Hispanic/Latino populations), we used incidence data from 13 patterns of cancer treatment across different population groups, SEER cancer registries (SEER13), which cover approximately and to elucidate the factors that determine these patterns. For 14% of the U.S. population, including 12% of whites, 12% of example, a 2000 Institute of Medicine report ( 9 ) on data systems blacks, 36% of APIs, 21% of AI/ANs, and 22% of Hispanics/ to improve the quality of cancer care stated, “ complete ascertain- Latinos ( 13 ) and total U.S. death data from the CDC. We examined ment of incident cancer cases by cancer registries is a prerequisite incidence data for 23 cancer sites or types and death data for 22 for national quality assessments, allowing case selection for cancer sites or types to ensure that we would cover the 15 most studies whose results can be generalized to the total population, common cancers in each racial and ethnic population. In this D o w as well as assessments of quality for important subgroups, for report, we considered Kaposi sarcoma and mesothelioma cases n lo example, individuals of low socioeconomic status, and individu- separately from other cancer sites or types, in contrast to previous a d als enrolled in certain types of health plans or delivery systems. ” reports. Kaposi sarcoma and mesothelioma were reported as ed This report also includes a special section in which we review causes of death beginning in 1999; therefore, we did not report fro m and update population-based studies of trends in cancer treatment them separately for mortality. Although cancer registries col- h and determinants of cancer treatment patterns of care, using can- lected information on cancer incidence and death among specifi c ttp s cer registry data to select cases and to identify demographic and API and Hispanic/Latino populations, incidence and death rates ://a clinical characteristics. for these populations could not be calculated because of the lack ca d of intercensal county population estimates from the U.S. Bureau e m S UBJECTS AND ME THODS of the Census. Estimates of cancer incidence and death rates ic.o aggregated from data across different geographic regions can u p Cancer Cases and Deaths differ, refl ecting regional and racial/ethnic variations ( 17 , 24 ) . .co m /jn Sta Itnesfo wrmasa btiaosne do no nn deawtaly c odlilaegcnteods ebdy ccaanncceerr rceagsiesst riines tphaer tUicnipitaetd- Cancer Treatment ci/artic ing in the NCI’s Surveillance, Epidemiology, and End Results Data regarding patterns and trends of cancer treatment in the le (SEER) Program and/or the CDC’s National Program of Cancer United States were obtained from published and unpublished -ab s Registries (NPCR) ( 10 – 13 ) . All cancer registries are members of SEER and SEER-Medicare databases and from NCI Patterns of tra the NAACCR ( 14 ) . For all cancers except bladder cancer, inci- Care/Quality of Care (POC/QOC) studies that were based on ct/9 dence data refer to invasive but not in situ cancers (data on the samples of SEER cases. The NCI has been conducting POC/QOC 7 /1 incidence of bladder cancer refers to both invasive and in situ studies since 1987 ( 25 ) and has linked SEER data to Medicare 9/1 cancers). For incident cases diagnosed in 2001 or later, all infor- data since 1986 to create a database for health services research 40 7 mation concerning the primary cancer site and histology was ( 26 – 28 ) . Information about surgery for early-stage breast cancer /2 5 coded according to the International Classifi cation of Diseases was derived from cases diagnosed from 1991 through 2002 2 1 3 for Oncology, third edition (ICD-O-3; 1 5 ); cases diagnosed and reported in SEER11 registries [i.e., Connecticut, Hawaii, 8 4 before 2001 were coded according to the ICD-O, second edition Iowa, Utah, and New Mexico, and the metropolitan areas of b y (ICD-O-2; 16 ) and then converted to ICD-O-3 codes. For analy- San Francisco, Detroit, Atlanta, Seattle-Puget Sound, San Jose- g u sis, all cases were categorized according to SEER site groups Monterey, and Los Angeles County ( 29 ) ]; SEER11 has population es ( 17 ) . To maximize comparability between ICD-O-2 and ICD- coverage similar to that for SEER13 but does not include Alaska t o n O-3, borderline tumors of the ovary, refractory anemias, and Natives or residents of rural Georgia. Information about adjuvant 08 other myelodysplastic syndromes were excluded, and pilocytic therapy for early-stage breast cancer was from a population- A p astrocytomas (which were excluded from ICD-O-3 as malignant based POC/QOC study of a sample of cases that were diagnosed ril 2 tumors) were included in analyses for this report. in 1987, 1990, and 1995 ( 30 ) and was updated with data from 01 9 Cancer deaths in the United States occurring from 1975 cases diagnosed in 2000 ( 30 ) . Treatment data for stages II and III through 2002 that are reported to state vital statistics offi ces and colorectal cancer were from a POC/QOC study for a sample of consolidated into a database by CDC through the National Vital cases diagnosed in 1987, 1991, and 1995 ( 31 ) and from SEER- Statistics System ( 18 ) were coded according to the version of the Medicare databases for cases diagnosed from 1997 through 1999 ICD in use in the United States at the time they occurred ( 19 – 21 ) . ( 32 ) . Analyses of treatment for non – small-cell lung cancer Beginning with 1999 mortality data, ICD-10 was used to code (NSCLC) were based on data from a POC/QOC study ( 33 ) the cause of death. Cancer was slightly more likely to be selected and a SEER-Medicare database ( 34 ) . Information about the as the underlying cause of death under ICD-10 rules than under receipt of guideline therapy for ovarian cancer was derived previous ICD rules ( 22 ) . Cancer sites were grouped by the SEER from a POC/QOC study conducted on sample cases diagnosed in Program to allow for maximum comparability between versions 1991 and 1996 ( 35 ) . The two studies of trends in treatment for of ICD codes ( 17 ) . early-stage prostate cancer patients diagnosed in 1986 through For the long-term (i.e., 1975 through 2002) trend analyses of 1993 and 1991 through 1999 were based on SEER – Medicare incidence and death rates for all cancers combined and for the 15 data ( 36 , 37 ) . 1408 SPECIAL ARTICLE Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 Statistical Analysis the trend is statistically signifi cantly different from zero (two-sided P <.05); otherwise, we use the terms “ stable ” or “ level. ” Absolute Cancer incidence and death rates, expressed per 100 000 per- changes in cancer incidence and death rates were calculated as the sons, were age-adjusted by 19 age groups, typically in 5-year age difference between the 2002 and the 1992 age-adjusted rates. categories (i.e., younger than 1 year, 1 – 4 years, 5 – 9 years, 10 – 14 More detailed information and methodology related to this years, … , 75 – 79 years, 80 – 84 years, and 85 years or older) to the report are available at the NCI Web sites www.seer.cancer.gov/ 2000 U.S. standard population ( 17 ) . Weights were calculated report_to_nation/1975_2002/ and www.cancer.gov . Additional using population projections from the U.S. Bureau of the Census data on cancer incidence and mortality are available from the ( 38 , 39 ) for the year 2000 and methodology developed from pub- following sites: www.cancer.org (ACS); www.cdc.gov/cancer/ lished guidelines ( 40 ) . Rates calculated using the new 19 age- npcr/index.htm and www.cdc.gov/nchs/about/major/dvs/mortdata. group 2000 U.S. standard population weights ( 41) were virtually htm (CDC); and www.naaccr.org/CINAP/index.htm (NAACCR). identical to those calculated using the 19 age-group U.S. 2000 standard million ( 41 ) weights developed for earlier reports. The R ESULTS CDC’s National Center for Health Statistics publishes age- adjusted mortality rates, obtained using the established method- D Update on Long-Term Incidence Trends for All Cancer o ology, that are based on 11 age groups, typically in 10-year age w Sites Combined and for the 15 Most Common Cancer n categories (i.e., younger than 1 year, 1 – 4 years, 5 – 14 years, Sites for All Races, 1975 – 2002 load 15 – 24 years, … , 65 – 74 years, 75 – 84 years, and 85 years or older) e d and weights from the year 2000 population projections ( 38 , 42 ) . For all populations combined ( Table 1 ), age-adjusted cancer fro The number of age groups used for age adjustment may affect incidence rates for all sites combined were stable since 1992. m h estimated rates slightly, whereas the effects of weights within an Among men, incidence rates for all cancer sites combined were ttp s age group are negligible when an adequate number of signifi cant stable from 1995 through 2002. Among women, incidence rates ://a digits is retained ( 41 ) . More detailed information on the change increased by 0.3% per year from 1987 through 2002. ca d to the U.S. standard population and comparison to the U.S. stan- Among men, cancer incidence rates increased during the most e m dard million population can be found at http://seer.cancer.gov/ recent segment (from the last joinpoint until 2002) for melanoma ic stdpopulations/single_age.html . The process of age adjustment of the skin (melanoma) and cancers of the prostate, kidney and .ou p allows comparisons of rates among multiple groups, such as renal pelvis (kidney), and esophagus, but decreased for cancers .c o tihnogs teh ed eefif fneecdt obfy dsiefxfe, rreanccee, sa nind tyheea ar goef sdtiraugcntuorseiss, owf hthilee gerliomupinsa. t- oanf dt hpeh laurnygn xa n(odr ablr ocnavchituys) , (slutonmg)a,c cho, laonnd alnadry rnexc.t uInmc,i doernacl ec arvatietys m/jnc val Es swtiemrea tegse noef rraatteeds , usstainngd aSrdE EerRro*rSst,a at nvde 9rs5i%on c5o.n0fi. 1d7e nscoef tiwntaerre- fino r itnhcei dreemncaein riantge s1 5f ocra npcreors staittees cwaenrcee srt aflb ulec.t uTahtee dlo gnrge-atetlrym, tarse nwdes i/article ( 23 ) ; standard error estimates [based on the gamma method ( 43 ) ] previously reported ( 3 – 7 ) . -ab s and 95% confi dence intervals results are available at http:// Among women, cancer incidence rates increased during the tra jncicancerspectrum.oupjournals.org/jnci/content/vol97/issue19 . most recent segment (from the last joinpoint until 2002) for ct/9 The long-term trends (i.e., from 1975 through 2002) in cancer leukemia, non-Hodgkin lymphoma, melanoma, and cancers of the 7/1 incidence and mortality among all races/ethnicities combined were breast, t hyroid, urinary bladder (bladder), and kidney. Female breast 9/1 described by joinpoint regression analysis, which involves fi tting a cancer incidence rates increased by 0.4% per year from 1987 40 7 series of joined straight lines on a log scale to the age-adjusted through 2002, a slower rate of increase than in the previous time /2 5 rates ( 44 ) . Line segments are joined at points called joinpoints. period (i.e., increase of 3.7% per year from 1980 through 1987). 21 Each line segment is described by an annual percent change, which The cancer incidence rates decreased for cancers of the colon and 38 4 is based on the slope of the line segment, and each joinpoint rectum, ovary, cervix uteri, oral cavity, and stomach. Incidence rates b y denotes a statistically signifi cant change in trend. The overall for the remaining cancer sites were stable. It is important to note that g u statistical signifi cance was set to P <.05, and a maximum of three the stabilization of lung cancer incidence rates among women since es joinpoints and four line segments was allowed for each model. We 1998 occurred after they had increased for many decades, i.e., by t on present incidence trends that are based on observed data and on 5.5% per year from 1975 through 1982, by 3.5% per year from 1982 08 data that were adjusted for delays in reporting ( 45 ) . This delay through 1990, and by 1.0% per year from 1990 through 1998. Ap adjustment facilitates the interpretation of incidence trends, espe- ril 2 cially because recent diagnosis years are most affected by report- Update on the Long-Term Mortality Trends for All 01 9 ing delays. We used statistical models to adjust the current cancer Cancer Sites Combined and for the 15 Most Common counts for anticipated future improvements to the data on the basis Cancer Sites for All Races, 1975 – 2002 of long-term observed reporting patterns in SEER registries ( 45 ) . Descriptions of long-term cancer incidence trends are based on Overall cancer death rates for all racial and ethnic populations the delay-adjusted rates, except where specifi cally noted. For each combined decreased by 1.1% per year from 1993 through 2002; racial and ethnic population, the annual percent changes for a more the decline was more pronounced among men (1.5% per year from recent fi xed time period (i.e., 1992 through 2002) were estimated 1993 through 2002) than among women (0.8% per year from 1992 by fi tting a linear regression line to the natural logarithms of the through 2002) ( Table 2 ). Mortality trends for the 15 most common rates, using calendar year as the independent variable ( 3 ) . The an- cancers differed slightly between men and women. Death rates nual percent change for the fi xed time period is the best measure decreased for 12 of the 15 most common cancers in men (i.e., lung, for comparison among groups because the beginning and ending prostate, colon and rectum, pancreas, non-H odgkin lymphoma, years are the same. In describing trends, we use the terms leukemia, bladder, stomach, and brain and other nervous system “ increase(d) ” or “ decrease(d) ” only when the slope (coeffi cient) of [brain], myeloma, oral cavity, and melanoma) and for nine of the Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 SPECIAL ARTICLE 1409 Table 1. S EER cancer incidence rate trends with joinpoint analyses for 1975 through 2002 for the 15 most common cancers, by sex, for all races * Joinpoint analyses (1975– 2002) † Trend 1 Trend 2 Trend 3 Trend 4 Cancer site or type Years APC ‡ Years APC ‡ Years APC ‡ Years APC ‡ All sites § Both sexes 1975 – 1983 0.9 || 1983 – 1992 1.8 || 1992 – 1995 − 1.6 1995 – 2002 0.0 (Delay-adjusted) 1975 – 1983 0.9 || 1983 – 1992 1.8 || 1992 – 1995 − 1.7 1995 – 2002 0.3 Male 1975 – 1989 1.3 || 1989 – 1992 5.1 || 1992 – 1995 − 4.6 || 1995 – 2002 − 0.2 (Delay-adjusted) 1975 – 1989 1.3 || 1989 – 1992 5.2 1992 – 1995 − 4.7 || 1995 – 2002 0.2 Female 1975 – 1979 − 0.2 1979 – 1987 1.5 || 1987 – 1999 0.3 || 1999 – 2002 − 0.8 (Delay-adjusted) 1975 – 1979 − 0.2 1979 – 1987 1.5 || 1987 – 2002 0.3 || 15 most common cancers for males Prostate 1975 – 1988 2.6 || 1988 – 1992 16.4 || 1992 – 1995 − 11.2 || 1995 – 2002 1.3 || (Delay-adjusted) 1975 – 1988 2.6 || 1988 – 1992 16.5 || 1992 – 1995 − 11.2 || 1995 – 2002 1.7 || D Lung and bronchus 1975 – 1981 1.7 || 1981 – 1991 − 0.3 1991 – 2002 − 2.0 || o (Delay-adjusted) 1975 – 1982 1.5 || 1982 – 1991 − 0.4 1991 – 2002 − 1.8 || w n Colon and rectum 1975 – 1986 1.1 || 1986 – 1995 − 2.1 || 1995 – 1998 1.0 1998 – 2002 − 2.9 || lo a (Delay-adjusted) 1975 – 1986 1.1 || 1986 – 1995 − 2.1 || 1995 – 1998 1.0 1998 – 2002 − 2.5 || d e Urinary bladder 1975 – 1987 1.0 || 1987 – 1996 − 0.5 1996 – 2000 1.5 2000 – 2002 − 3.5 d (Delay-adjusted) 1975 – 1987 1.0 || 1987 – 1996 − 0.5 1996 – 2000 1.6 2000 – 2002 − 2.6 fro Non-Hodgkin lymphoma 1975 – 1991 4.3 || 1991 – 2002 − 0.1 m (Delay-adjusted) 1975 – 1991 4.3 || 1991 – 2002 0.2 h Melanoma of the skin¶ 1975 – 1985 5.7 || 1985 – 2000 3.6 || 2000 – 2002 − 1.2 ttp s (Delay-adjusted) ¶ 1975 – 1985 5.8 || 1985 – 2002 3.8 || ://a Leukemia 1975 – 2002 − 0.3 || c a (Delay-adjusted) 1975 – 2002 0.1 d e Oral cavity and pharynx 1975 – 1983 − 0.1 1983 – 2002 − 1.5 || m (Delay-adjusted) 1975 – 2002 − 1.2 || ic.o Kidney and renal pelvis 1975 – 2002 1.7 || u p (Delay-adjusted) 1975 – 2002 1.8 || .c Stomach 1975 – 1988 − 1.2 || 1988 – 2002 − 2.1 || om (Delay-adjusted) 1975 – 1988 − 1.2 || 1988 – 2002 − 2.0 || /jn Pancreas 1975 – 1981 − 1.8 || 1981 – 1985 1.2 1985 – 1989 − 2.4 1989 – 2002 − 0.1 c Liv (Dere alanyd- aindtjruashteedp)a tic 1 1997755 – – 1 1998814 − 1 1..87 | | 1 1998814 – – 1 1998959 1 4..15 || 1 1998959 – – 1 2909002 −− 2 2..11 1990 – 2002 0. 1 i/artic bile duct le-a (Delay-adjusted) 1975 – 1984 1.7 1984 – 1999 4.5 || 1999 – 2002 − 0.7 b s Brain and other nervous 1975 – 1989 1.2 || 1989 – 2002 − 0.5 tra system c (Delay-adjusted) 1975 – 1989 1.2 || 1989 – 2002 − 0.3 t/9 7 Esophagus 1975 – 2002 0.7 || /1 9 (Delay-adjusted) 1975 – 2002 0.8 || /1 Larynx 1975 – 1988 − 0.2 1988 – 2002 − 2.8 || 4 0 (Delay-adjusted) 1975 – 1988 − 0.3 1988 – 2002 − 2.8 || 7/2 15 most common cancers 5 2 for females 1 3 Breast 1975 – 1980 − 0.5 1980 – 1987 3.8 || 1987 – 2002 0.3 || 8 4 (Delay-adjusted) 1975 – 1980 − 0.4 1980 – 1987 3.7 || 1987 – 2002 0.4 || b Lung and bronchus 1975 – 1982 5.5 || 1982 – 1990 3.5 || 1990 – 1998 1.0 || 1998 – 2002 − 1.1 y g (Delay-adjusted) 1975 – 1982 5.5 || 1982 – 1990 3.5 || 1990 – 1998 1.0 || 1998 – 2002 − 0.5 u e Colon and rectum 1975 – 1985 0.3 || 1985 – 1995 − 1.8 || 1995 – 1998 1.5 1998 – 2002 − 1.9 || s (Delay-adjusted) 1975 – 1985 0.3 || 1985 – 1995 − 1.8 || 1995 – 1998 1.5 1998 – 2002 − 1.5 || t o n Corpus and uterus, NOS 1975 – 1979 − 6.0 || 1979 – 1988 − 1.7 || 1988 – 1998 0.6 || 1998 – 2002 − 1.1 0 8 (Delay-adjusted) 1975 – 1979 − 6.0 1979 – 1988 − 1.7 || 1988 – 1997 0.7 || 1997 – 2002 − 0.6 A Non-Hodgkin lymphoma 1975 – 1990 2.9 || 1990 – 2002 0.9 || p (Delay-adjusted) 1975 – 1990 2.9 || 1990 – 2002 1.2 || ril 2 Ovary # 1975 – 1987 0.1 1987 – 2002 − 0.9 || 0 1 (Delay-adjusted) 1975 – 1985 0.2 1985 – 2002 − 0.7 || 9 Melanoma of the skin¶ 1975 – 1980 6.1 || 1980 – 2002 2.6 || (Delay-adjusted)¶ 1975 – 1981 6.1 || 1981 – 1993 2.2 || 1993 – 2002 4.1 || Pancreas 1975 – 1984 1.3 || 1984 – 2002 − 0.3 || (Delay-adjusted) 1975 – 1984 1.2 || 1984 – 2002 − 0.2 Thyroid 1975 – 1977 6.4 1977 – 1980 − 4.9 1980 – 1993 2.1 || 1993 – 2002 5.0 || (Delay-adjusted) 1975 – 1981 − 1.2 1981 – 1993 2.0 || 1993 – 2002 5.3 || Cervix uteri 1975 – 1981 − 4.6 || 1981 – 1997 − 1.1 || 1997 – 2002 − 4.8 || (Delay-adjusted) 1975 – 1981 − 4.6 || 1981 – 1997 − 1.1 || 1997 – 2002 − 4.5 || Leukemia 1975 – 2002 − 0.1 (Delay-adjusted) 1975 – 2002 0.2 || Urinary bladder 1975 – 2002 0.2 || (Delay-adjusted) 1975 – 2002 0.2 || (Table continues) 1410 SPECIAL ARTICLE Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 Table 1 (continued). Joinpoint analyses (1975– 2002) † Trend 1 Trend 2 Trend 3 Trend 4 Cancer site or type Years APC ‡ Years APC ‡ Years APC ‡ Years APC ‡ Kidney and renal pelvis 1975 – 1991 2.7 || 1991 –2 002 1.3 || (Delay-adjusted) 1975 – 1990 2.8 || 1990 – 2002 1.6 || Oral cavity and pharynx 1975 – 1980 2.6 || 1980 –2 002 − 1.0 || (Delay-adjusted) 1975 – 1980 2.5 1980 –2 002 − 0.9 || Stomach 1975 – 2002 − 1.7 || (Delay-adjusted) 1975 – 2002 − 1.7 || * Sources of data are the Surveillance, Epidemiology, and End Results (SEER) Program registries that include Connecticut, Hawaii, Iowa, Utah, and New Mexico, and the metropolitan areas of San Francisco, Detroit, Atlanta, and Seattle-Puget Sound (i.e., SEER9). Joinpoint analysis was performed with the use of the Joinpoint (JP) Regression Program, version 3.0, April 2005, National Cancer Institute. The 15 most common cancers were selected on the basis of the sex-specifi c age-adjusted incidence rates for 1992 – 2002 for all races combined. A spreadsheet that contains the standard errors and 95% confi dence intervals for the APCs is available at D http://jncicancerspectrum.oupjournals.org/jnci/content/vol97/issue19 . APC = annual percent change; NOS = not otherwise specifi ed. ow † Joinpoint analyses with up to three joinpoints are based on rates per 100 000 persons and are age-adjusted to the 2000 U.S. standard population (using 19 age nlo groups, with data provided from Census Current Population Reports series, P25-1130 ( 39 ) . ad ‡ APC is based on rates that were age-adjusted to the 2000 U.S. standard population (using 19 age groups, with data provided from Census Current Population ed Reports series, P25-1130 (3 9 ) using joinpoint regression analysis. fro § Excludes myelodysplastic syndromes and borderline tumors. m h || APC is statistically signifi cantly different from zero (two-sided P <.05). ttp ¶ Age-adjusted rates for melanoma of the skin are calculated using white patients only. s # Excludes borderline tumors. ://a c a d e m ic .o 15 most common cancers in women (i.e., breast, colon and rectum, 1992 through 2002 ( Table 3 ) masked a change in the direction of u p non-Hodgkin lymphoma, leukemia, brain, stomach, myeloma, the incidence trend that began in 1995, as identifi ed by the join- .c o cervix uteri, and bladder). For melanoma, death rates decreased in point analysis ( Table 1 ) for all race/ethnic groups combined. Pros- m /jn men (by 1.6% per year from 1998 through 2002); for multiple tate cancer incidence rates increased from 1995 through 2002. c myeloma, death rates decreased in both men (by 0.9% per year Among women, breast cancer incidence rates increased among i/artic from 1994 through 2002) and women (by 0.5% per year from 1993 API women, decreased among AI/AN women, and were stable for le through 2002). Among men, death rates increased for esophageal other women; lung cancer rates decreased in AI/AN and Hispanic/ -ab s cancer (by 0.5% per year from 1994 through 2002) and for liver Latina women and were stable for the other populations; colorectal tra cancer (by 1.6% per year from 1995 through 2002). Among cancer incidence rates decreased only for white women. ct/9 women, death rates increased for lung cancer (by 0.3% per year Trends in incidence rates for other cancers also varied among 7 /1 from 1995 through 2002). Death rates were stable for kidney can- different racial and ethnic populations and by sex, although most 9/1 cer in men and women and for fi ve of the 15 most common cancers trends among AI/ANs could not be evaluated because of the small 40 7 in women (i.e., kidney, pancreas, ovary, corpus uteri, and liver and number of cases ( Table 3 ). Incidence rates for liver cancer in- /2 5 intra hepatic bile duct [liver]). Further joinpoint analyses of age- creased among men in all groups except APIs and AI/ANs and 2 1 3 specifi c lung cancer death rates among women by 10-year age in- among white and Hispanic/Latina women. We observed declines 8 4 tervals showed that lung cancer death rates increased for women in the incidence rates for cancers of the stomach and larynx among b y aged 40 – 49 years and 70 years or older (data not shown). men in all populations except AI/ANs. Incidence rates for cancers g u of the oral cavity also decreased among men in all populations es Cancer Incidence and Death Rates for the 15 Most except APIs, and incidence rates for Kaposi sarcoma decreased t on Common Cancer Sites by Race and Ethnicity, 1992 – 2002 among white, black, and Hispanic/Latino men. Cervical cancer 08 rates decreased among women in all racial and ethnic populations. A p We ranked the 15 most frequently occurring cancers diag- Incidence rates for thyroid and kidney cancer increased among ril 2 nosed from 1992 through 2002 in terms of age-adjusted incidence women in all racial and ethnic populations except AI/ANs. 01 9 rates ( Table 3 ) and age-adjusted death rates ( Table 4 ) for all races An analysis of recent (i.e., from 1992 through 2002) mortality combined and for each major racial and ethnic population by sex. trends revealed declines in death rates for lung, prostate, and The highest cancer incidence and death rates for each racial and colon and rectal cancers among men in most racial and ethnic ethnic population continued to be for cancers of the prostate, populations; the exceptions were death from lung cancer among lung, and colon and rectum among men and for cancers of the AI/AN men and from colon and rectal cancer among AI/AN and breast, lung, and colon and rectum among women. Hispanic/Latino men ( Table 4 ). Death rates for colorectal cancer Examination of age-adjusted incidence trends by race and eth- declined among white, black, and API women, and death rates nicity from 1992 through 2002 revealed that the incidence rates for for breast cancer declined among white, black, and Hispanic/ lung and prostate cancers declined among men in all racial/ethnic Latina women. Although the lung cancer death rates continued to populations; colorectal cancer incidence rates decreased only for increase among white women and black women, these annual white men ( Table 3 ). Although our analysis of trends for this increases are substantially lower than increases reported for all period permits valid comparisons of trends by race, sex, and cancer women before 1992 ( Table 1 ) and are consistent with long-term site, the declines in prostate cancer incidence rates observed from trends of slowing rates of increase over time. Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 SPECIAL ARTICLE 1411 Table 2. U .S. death rate trends with joinpoint analyses for 1975 through 2002 for the 15 most common cancers for all races* Joinpoint analyses (1975– 2002) † Trend 1 Trend 2 Trend 3 Trend 4 Cancer site or type Years APC Years APC ‡ Years APC ‡ Years APC ‡ All sites Both sexes 1975 – 1990 0.5 § 1990 – 1993 − 0.3 1993 – 2002 − 1.1 § Male 1975 – 1979 1.0 § 1979 – 1990 0.3 § 1990 – 1993 − 0.4 1993 – 2002 − 1.5 § Female 1975 – 1992 0.5 § 1992 – 2002 − 0.8 § 15 most common cancers for males Lung and bronchus 1 975 – 1978 2.4 § 1978 – 1984 1.2 § 1984 – 1991 0.3 § 1991 – 2002 − 1.9 § Prostate 1975 – 1987 0.9 § 1987 – 1991 3.1 § 1991 – 1994 − 0.6 1994 – 2002 − 4.0 § Colon and rectum 1975 – 1978 0.8 1978 – 1984 − 0.4 1984 – 1990 − 1.3 § 1990 – 2002 − 2.0 § Pancreas 1975 – 1986 − 0.8 § 1986 – 2002 − 0.3 § Non-Hodgkin lymphoma 1975 – 1981 1.8 § 1981 – 1990 3.0 § 1990 – 1997 1.6 § 1997 – 2002 − 2.8 § D Leukemia 1975 – 1995 − 0.2 § 1995 – 2002 − 0.7 § o Urinary bladder 1975 – 1983 − 1.4 § 1983 – 1987 − 2.7 § 1987 – 1993 0.1 1993 – 2002 − 0.6 § w n Esophagus 1975 – 1985 0.7 § 1985 – 1994 1.2 § 1994 – 2002 0.5 § lo a Stomach 1975 – 1987 − 2.3 § 1987 – 1991 − 0.9 1991 – 2002 − 3.5 § d e Liver and intrahepatic 1975 – 1986 1.7 § 1986 – 1995 3.9 § 1995 – 2002 1.6 § d bile duct fro Kidney and renal pelvis 1975 – 1991 1.1 § 1991 – 2002 − 0.1 m Brain and other nervous 1975 – 1977 4.4 1977 – 1982 − 0.4 1982 – 1990 1.5 § 1990 – 2002 − 0.7 § h system ttp s Myeloma 1975 – 1994 1.5 § 1994 – 2002 − 0.9 § ://a Oral cavity and pharynx 1975 – 1991 − 1.8 § 1991 – 2002 − 2.6 § c a Melanoma of the skin 1975 – 1987 2.4 § 1987 – 1998 0.8 § 1 998 – 2002 − 1.6 § d e 15 most common cancers m for females ic.o Lung and bronchus 1 975 – 1982 6.0 § 1982 – 1990 4.2 § 1990 – 1995 1.7 § 1995 – 2002 0.3 § u p Breast 1975 – 1990 0.4 § 1990 – 2002 − 2.3 § .c Colon and rectum 1975 – 1984 − 1.0 § 1984 – 2002 − 1.8 § om Pancreas 1975 – 1984 0.8 § 1984 – 2002 0.1 /jn Ovary 1975 – 1982 − 1.2 § 1982 – 1992 0.3 § 1992 – 1998 − 1.2 § 1998 – 2002 0.8 c N Leounk-Hemodiag kin lymphoma 1 1997755 – – 1 1999840 2 0..28 § 1 1999840 – – 1 2909072 − 0 1..40 § 1997 – 2002 − 3. 2 § i/artic Corpus and uterus, NOS 1975 – 1989 − 1.6 § 1989 – 1997 − 0.7 § 1997 – 2002 0.5 le-a Brain and other nervous 1975 – 1992 0.9 § 1992 – 2002 − 1.0 § b s system tra Stomach 1975 – 1987 − 2.8 § 1987 – 1990 − 0.5 1990 – 2002 − 2.6 § c Myeloma 1975 – 1993 1.5 § 1993 – 2002 − 0.5 § t/9 7 Cervix uteri 1975 – 1982 − 4.4 § 1982 – 1996 − 1.6 § 1996 – 2002 − 3.8 § /1 9 Liver and intrahepatic 1975 – 1978 − 1.5 1978 – 1988 1.4 § 1988 – 1995 3.9 § 1995 – 2002 0.4 /1 bile duct 4 0 Kidney and renal pelvis 1975 – 1992 1.3 § 1992 – 2002 − 0.5 7/2 Urinary bladder 1975 – 1986 − 1.7 § 1986 – 2002 − 0.3 § 5 2 1 3 * Sources of data are the Surveillance, Epidemiology, and End Results (SEER) Program registries that include Connecticut, Hawaii, Iowa, Utah, and New Mexico 8 4 and the metropolitan areas of San Francisco, Detroit, Atlanta, and Seattle-Puget Sound (i.e., SEER9). Joinpoint analysis was performed with the use of the Joinpoint b y (JP) Regression Program, version 3.0, April 2005, National Cancer Institute. The 15 most common cancers were selected on the basis of the sex-specifi c age-adjusted g u incidence rates for 1992 – 2002 for all races combined. A spreadsheet that contains the standard errors and 95% confi dence intervals for the APCs is available at e s http://jncicancerspectrum.oupjournals.org/jnci/content/vol97/issue19 . APC = annual percent change; NOS = not otherwise specifi ed. t o † Joinpoint analyses with up to three joinpoints are based on rates per 100 000 persons and are age-adjusted to the 2000 U.S. standard population (using 19 age n 0 groups, with data provided from Census Current Population Reports series, P25-1130 ( 39 ) . 8 A ‡ APC is based on rates that were age-adjusted to the 2000 U.S. standard population (using 19 age groups, with data provided from Census Current Population p Reports series, P25-1130 ( 39 ) using joinpoint regression analysis. ril 2 § APC is statistically signifi cantly different from zero (two-sided P <.05). 01 9 Mortality trends for cancers other than the three most com- Hispanic/Latina women, and cervical cancer death rates declined mon cancers also varied by racial and ethnic group and by sex among women in all populations. ( Table 4 ). From 1992 through 2002, the death rates for liver can- cer increased among white, black, and Hispanic/Latino men and among white and Hispanic/Latina women. Stomach cancer death S PECIAL S ECTION : MO NITORING C ANCER TR EATMENT rates declined for men and women of all racial and ethnic popula- TR ENDS A ND D ETERMINANTS U SING D ATA FROM tions except for AI/AN men and women. Similarly, declines in PO PULATION -B ASED C ANCER R EGISTRIES death rates for oral cavity cancers were observed among men and women in most populations, except for AI/AN men and women, One strategy for reducing the number of cancer deaths and im- API women, and Hispanic/Latina women. Finally, death rates proving survival among those diagnosed with cancer is to ensure for cancers of the gallbladder declined among white, API, and that evidence-based cancer treatment services are available and 1412 SPECIAL ARTICLE Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 accessible. Here we discuss the results of several studies that have inverse relationship between age at diagnosis and receipt of ra- examined trends in the delivery and determinants of cancer treat- diation therapy among women aged 65 years or older, even after ment and present results of some new analyses. Although we focus adjusting for comorbidity (adjusted odds ratios for women aged on the treatment of breast cancer, we also present abbreviated fi nd- 65– 6 9 years, 70– 7 4 years, 75 – 79 years, and 80 years or older ings for the treatment of colorectal cancer, NSCLC, and cancers of were 1.00, 0.70, 0.44, and 0.12, respectively). Riley et al. (5 2) the ovary and prostate. We also briefl y summarize fi ndings on the found that patients with early-stage breast cancer (stage I and II) relationship between provider procedure volume and outcomes, on who were enrolled in health maintenance organizations (HMOs) care during the last year of life, and on cancer treatment cost. were, on average, as likely to undergo breast-conserving surgery as women who were enrolled in fee-for-service (FFS) health Evaluations of Cancer Care Delivery plans in the same geographic area and that patients in HMOs were more likely than women in FFS health plans to receive Several studies ( 30 , 31 ) have examined trends in the dissemina- radiation therapy. These investigators also found that results tion of appropriate cancer treatment as defi ned by statements varied across the nine SEER areas studied (e.g., in three of the issued by the National Institutes of Health (NIH) Consensus areas, FFS patients were more likely to receive radiation treat- Devel op ment Program ( 46 ) or by specifi c NCI clinical alerts that ment than were HMO patients). D o w refl ect emerging evidence of treatment effi cacy from controlled Adjuvant chemotherapy and hormonal therapy for early-stage n lo clinical trials. In other studies ( 47 , 48 ) , treatment patterns have been breast cancer. In recent decades, the treatments recommended for a d compared with clinical guidelines issued by professional organiza- women with early-stage breast cancer by NIH Consensus Develop- ed tions concerned with specialty cancer care. In all of these studies, ment Conference statements and clinical alerts ( 49 , 53 ) have fro m patterns of care were evaluated relative to the most recently pub- changed ( Fig. 2 ). In 1985, recommendations for lymph node – h lished guidelines before the year of diagnosis under examination. positive breast cancer patients were established. Multiagent che- ttp s Trends in Early-Stage Breast Cancer Treatment. Data from motherapy was recommended for premenopausal women and for ://a SEER registries and NCI POC/QOC studies have been used postmenopausal women with estrogen receptor– negative tumors. ca d to study two treatment regimens for early-stage breast cancer: For women with estrogen receptor – positive tumors, tamoxifen was e m 1) breast-conserving surgery and radiation, and 2) adjuvant recommended. In May 1988, on the basis of new clinical trial re- ic .o chemotherapy and hormonal therapy. sults, the NCI issued a clinical alert advising the use of multiagent u p Breast-conserving surgery and radiation. Clinical trials have chemotherapy for lymph node – negative patients with tumors larger .c o demonstrated that women with early-stage breast cancer who than 3 cm and for lymph node – negative patients with estrogen m /jn receive breast-conserving surgery followed by radiation have receptor – negative tumors of 3 cm or smaller ( 53 ) . c survival outcomes similar to those of women who receive a Harlan et al. ( 30 ) conducted an NCI POC/QOC study to de- i/artic mastectomy ( 49 ) . A 1990 NIH Consensus Development Panel scribe therapies that were being used in community practice to le concluded that “ breast conservation treatment (breast-conserving treat women who resided in SEER areas and were newly diag- -ab s surgery followed by radiation therapy) is an appropriate method nosed with breast cancer. They used comprehensive medical tra of primary therapy for the majority of women with stage I and II record reviews and physician contact to collect information about ct/9 breast cancer and is preferable because it provides survival treatment. Data from this study of women diagnosed with early- 7 /1 equivalent to total mastectomy and axillary dissection while pre- stage breast cancer in 1987, 1990, and 1995 ( 30 ) , which we have 9/1 serving the breast ” ( 49 ) . Breast-conserving surgery followed by updated using unpublished data for a sample of cases diagnosed 40 7 radiation therapy is associated with a lower rate of local recur- in 2000 ( Table 5 ), indicate that, by 1987, a substantial proportion /2 5 rence than breast-conserving surgery alone ( 49 , 50 ) . of node – positive women with stages I – IIIA breast cancer were 2 1 3 Data from SEER11 on trends in the treatment of early-stage being treated with adjuvant therapy. Between 1987 and 2000, the 8 4 breast cancer ( Fig. 1 ) show that the proportion of women diag- use of concurrent chemotherapy and hormone (i.e., tamoxifen) b y nosed with stage I and II breast cancer who received breast- therapy increased for node – positive women being treated for g u conserving surgery and radiation treatment increased substantially early-stage breast cancer, although its use remained relatively es during the 1990s and that the proportion of women who received low among women aged 65 years or older, who were more likely t o n breast-conserving surgery only also increased modestly. The pro- to receive tamoxifen alone. 08 portion of women 65 years or older at diagnosis who received To translate these observed patterns of treatment into esti- A p breast-conserving surgery and radiation treatment was lower than mates of the proportion of women who received care according ril 2 the proportion of women younger than 65 years at diagnosis who to guideline recommendations, Harlan et al. ( 30 ) also took into 01 9 received this treatment ( Fig. 1 ). account the year of diagnosis (and the corresponding guidelines Two studies examined factors associated with breast- available during that year) and the estrogen receptor status of conserving surgery and radiation among women aged 65 or older each patient. On the basis of this analysis, the authors ( 30 ) esti- at diag nosis. Ballard-Barbash et al. ( 51 ) , using SEER-Medicare mated that 70% of the patients with node - positive early-stage data, found that, from 1985 through 1989, 55% of women en- breast cancer treated between 1987 and 1989 received therapy as rolled in Medicare who were diagnosed with early-stage breast specifi ed by guidelines, 75% of those treated in 1990 received cancer (stage I and II) and received breast-conserving surgery guideline therapy, and 73% of those treated in 1995 received also received radiation treatment. Women with preexisting guideline therapy, after adjusting for age, race/ ethnicity, registry, c omorbid conditions were less likely to receive radiation ther- estrogen receptor status, tumor grade, tumor size, and the number apy than were women with no comorbid conditions (adjusted of positive nodes. odd ratios for women with no comorbid conditions, with one The proportion of women with early-stage node-negative dis- comorbid condition, and with two or more comorbid conditions ease treated with adjuvant therapy increased from 1987 through were 1.00, 0.52, and 0.33, respectively). There was also a strong 1995 but remained lower than the proportion of women with Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 SPECIAL ARTICLE 1413 Table 3. S EER incidence rates and trends for the 15 most common cancers by sex and race/ethnicity for 1992 through 2002 * All races Whites Sex/cancer site or type Rank Rate ‡ APC § AC || Rank Rate ‡ APC § AC || Male All sites ¶ 570.9 − 1.3 # − 111.5 572.7 − 1.3 # − 114.3 Prostate 1 180.1 2.0 # − 62.5 1 175.5 − 2.1 # − 67.9 Lung and bronchus 2 82.7 − 2.2 # − 19.9 2 81.5 − 2.1 # − 18.5 Colon and rectum 3 64.0 − 1.2 # − 10.4 3 63.8 − 1.3 # − 12.0 Urinary bladder 4 36.1 − 0.2 − 2.2 4 39.7 − 0.1 − 2.5 Non-Hodgkin lymphoma 5 23.7 − 0.5 − 0.5 5 24.9 − 0.4 − 0.6 Melanoma of the skin 6 20.4 2.5 # 3.8 6 24.0 2.9 # 5.3 Leukemia 7 16.4 − 1.1 # − 2.6 7 17.4 − 1.1 # − 2.6 Oral cavity and pharynx 8 16.4 − 1.8 # − 2.7 8 16.2 − 1.5 # − 2.3 Kidney and renal pelvis 9 15.5 1.4 # 1.9 9 15.9 1.6 # 2.0 Stomach 10 13.1 − 2.1 # − 2.5 11 11.3 − 2.1 # − 2.1 Pancreas 11 12.7 − 0.4 # − 0.6 10 12.5 0.1 − 0.1 D Liver and intrahepatic bile duct 12 8.6 3.0 # 2.4 15 6.8 2.9 # 1.8 ow Brain and other nervous system 1 3 7.7 − 0.7 − 0.9 12 8.5 − 0.4 − 0.9 n Esophagus 14 7.6 0.3 − 0.3 13 7.4 1.5 # 0.5 loa Larynx 15 7.2 − 3.3 # − 2.7 14 7.1 − 3.3 # − 2.6 de MKaypeolosim saa rcoma 1 186 4 7..41 − − 2 00..54 # − − 0 8..35 1 168 6 4..85 − − 2 02..34 # − − 0 9..34 d fro F e m TGAhalalyll lesrb iotlieadsd ¶d er 1 390 412 3 0...179 − 1 0 3... 511 # −− 0 5 1...101 3 129 42 05 3...879 − 0 0 3... 924 # − 0 1 1...093 m https://a Breast 1 132.4 0.4 0.0 1 138.3 0.5 1.3 ca Lung and bronchus 2 49.2 − 0.2 − 1.1 2 51.3 − 0.1 − 0.6 de CCoolropnu sa nandd r euctteurmus , NOS 4 3 2 446..44 −− 0 0..62 # − − 3 0..98 3 4 4 255..99 − − 0 0..73 # − − 4 1..38 mic.o Non-Hodgkin lymphoma 5 15.5 0.8 # 1.2 5 16.3 0.8 # 1 .1 up Ovary † † 6 14.2 − 0.9 # − 1.6 7 15.1 − 0.8 # − 1.8 .co Melanoma of the skin 7 13.2 2.3 # 2.3 6 15.9 3.0 # 3.5 m Pancreas 8 9.9 − 0.3 0.0 11 9.6 − 0.3 − 0.1 /jn Thyroid 9 9.8 4.8 # 4.6 8 10.2 5.2 # 5.3 ci/a CLeeurvkiexm uitae ri 1 101 9 9..76 −− 2 0..88 # # − − 2 0..88 12 9 1 90..31 − − 2 0..25 # − − 2 0..46 rticle Urinary bladder 12 9.2 − 0.4 # − 0.4 10 9.9 − 0.2 − 0.2 -a Kidney and renal pelvis 13 7.6 1.4 # 0.9 13 7.9 1.5 # 0.8 bs Oral cavity and pharynx 14 6.7 − 1.1 # − 0.4 14 6.7 − 1.2 # − 0.4 tra SBtroamina acnhd other nervous system 1 1 65 5 6..42 − − 0 0..75 − − 0 0..63 1 165 5 6..10 − − 1 0..01 − − 0 0..72 ct/97 Myeloma 17 4.6 − 0.8 − 0.6 17 4.3 − 0.8 − 0.7 /19 Liver and intrahepatic bile duct 18 3.3 3.3 # 1.0 18 2.7 3.7 # 0.9 /1 Gallbladder 23 1.6 − 1.5 # − 0.4 23 1.6 − 1.4 − 0.3 40 7 /2 (Table continues) 5 2 1 38 e arly-stage node-positive tumors treated with adjuvant therapy sus Development Conference recommended treatment with adju- 4 b ( 30 ) . The 1990 NIH Consensus Development Statement on the vant chemotherapy for patients with stage III colon cancer ( 54 ) . y g u ptrre eaftemreenncte o fs heoaurllyd- sdteatgeer mbrienaes tt hcea nccheori c( 4e 9o ) fi ntrdeiactamteedn tth faot rp natoidene-t Ttrheaet ienvgi dsetnagcee fIoI rc tohleo ne fcfeacntcievre nweasss olefs sa ddjeufiv nanitti vceh.e Am ostuhbesreaqpuye nint est on n egative breast cancer, given that cure rates after surgery alone meta-analysis of data from multiple trials for 1016 stage II colon 08 are relatively favorable and that chemotherapy results in only a cancer patients indicated that patients who were treated with A p m odest improvement in these rates. After controlling for age, adjuvant 5-fl uorouracil plus leucovorin did not have statistically ril 2 rac e/ethnicity, registry, estrogen receptor status, tumor grade, and signifi cantly better 5-year disease-free or overall survival than 01 9 tumor size, Harlan et al. ( 30 ) estimated that the percentage of untreated control patients ( 55 ) . women with early-stage node-negative breast cancer who re- The 1990 NIH Consensus Conference ( 54 ) also recommended ce ived adjuvant therapy increased from 34% in 1987 – 1989 to combined adjuvant chemotherapy and high-dose external-beam 51% in 1990 and to 53% in 1995, following the 1988 publication radiotherapy to treat patients with stage II or III rectal cancer. of a clinical alert ( 53 ) that recommended adjuvant therapy for The Consensus Conference noted that, although radiation therapy node – negative tumors. There were relatively few changes in did not appear to affect disease-specifi c or overall survival, it general treatment patterns between 1995 and 2000, except for a substantially decreased local recurrence— an outcome associated continued increase in the percentage of women younger than with substantial morbidity in rectal cancer — and should therefore 51 years who were treated concurrently with chemotherapy and be considered an indicator of high-quality care. tamoxifen and a continued decrease in use of chemotherapy for Table 6 presents treatment patterns for colon and rectal can- tumors smaller than 1 cm. cers that were documented in an NCI POC/QOC study ( 31 ) of Trends in Colorectal Cancer Treatment. On the basis of the data from patients diagnosed in 1987, 1991, and 1995. By ana- accumulated evidence from clinical trials, a 1990 NIH Consen- lyzing these data, Potosky et al. ( 31 ) found that the percentage of 1414 SPECIAL ARTICLE Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 Table 3 (continued). Blacks API AI/AN Hispanics/Latinos † Rank Rate ‡ APC § AC || Rank Rate ‡ APC § AC || Rank Rate ‡ APC § AC || Rank Rate ‡ APC § AC || 715.6 − 1.7 # − 127.1 394.4 − 1.2 # − 55.6 285.5 − 3.8 # − 101.3 429.6 − 0.9 # − 44.5 1 283.8 − 1.9 # − 55.0 1 104.6 − 1.7 # − 27.5 1 63.4 − 6.8 # − 44.9 1 143.1 − 0.7 − 13.3 2 122.8 − 2.5 # − 35.7 2 61.2 − 1.4 # − 14.0 2 49.8 − 5.4 # − 20.3 3 47.2 − 2.0 # − 7.2 3 72.9 − 0.5 − 4.6 3 56.9 − 0.4 − 0.6 3 40.8 − 2.7 − 7.3 2 48.1 0.0 − 5.0 5 20.2 − 0.2 1.5 6 16.8 1.1 2.8 9 8.3 ** ** 5 19.0 − 0.2 0.0 7 18.8 − 1.2 − 1.2 7 16.7 0.2 0.0 7 9.8 2.2 3.8 4 19.3 − 0.9 − 2.6 23 1.3 ** ** 20 1.7 2.9 1.4 20 2.2 ** ** 17 4.1 3.2 # 1.6 11 12.9 − 1.6 − 3.6 10 9.8 − 0.3 − 0.2 12 5.5 ** ** 9 11.8 − 0.6 0.1 4 20.7 − 3.1 # − 6.0 8 12.6 − 1.5 − 0.9 6 11.3 − 9.2 # − 11.8 11 10.1 − 2.8 # − 2.3 8 18.5 2.2 # 4.3 11 8.7 0.1 − 0.1 4 15.6 − 4.0 # − 4.2 7 14.7 2.0 # 2.5 6 19.5 − 2.8 # − 5.1 4 23.1 − 3.3 # − 7.7 5 14.6 ** 4.8 6 18.3 − 2.5 # − 3.1 9 17.5 − 2.5 # − 5.4 9 10.7 − 2.8 # − 1.3 10 7.8 ** ** 10 10.8 − 0.9 − 1.0 D 14 10.8 4.5 # 4.3 5 20.9 1.0 2.0 8 9.0 ** ** 8 13.4 2.2 # 3.7 ow 16 4.7 0.0 0.3 14 4.1 − 1.6 − 1.0 14 3.3 ** ** 14 5.7 0.1 − 0.1 n lo 13 12.4 − 5.7 # − 7.4 12 5.1 − 2.0 − 0.5 11 6.1 ** ** 13 5.9 − 1.2 − 1.2 a d 12 12.7 − 3.2 # − 5.8 16 3.5 − 2.5 # − 0.8 18 2.3 ** ** 15 5.6 − 1.6 # − 1.2 e d 1 105 13 6..54 − − 1 02..77 # − 0 7..85 1 234 4 1..22 − 1 *.*4 − − 0 1..15 1 139 4 2..12 * *** * *** 1 126 6 4..59 − − 1 19..15 # − 0 8..58 fro m 20 2.1 1.0 − 0.3 15 3.9 1.6 0.7 21 2.1 ** ** 20 2.9 3.2 0.2 h 32 0.8 ** ** 21 1.4 − 5.0 − 1.0 15 2.9 ** ** 25 1.3 ** ** ttp 401.6 − 0 .2 1.1 301.2 0 .3 7.2 229.8 − 1 .7 # − 58.9 309.0 0.1 − 1.0 s://a c 1 120.2 − 0.2 − 1.6 1 92.8 1.5 # 8.1 1 60.7 − 3.5 # − 22.9 1 88.2 0.6 1.9 a d 3 53.6 0.5 2.8 3 28.4 0.0 − 1.2 3 25.8 − 2.8 # − 10.2 3 24.1 − 1.5 # − 3.4 e m 2 55.9 − 0.1 − 2.3 2 39.3 0.0 − 0.1 2 32.3 − 1.0 − 9.5 2 32.3 − 0.2 − 1.3 ic 4 18.4 1.8 # 6.0 4 17.0 1.6 # 3.3 4 10.1 ** − 1.6 5 16.6 0.7 1.7 .o u 7 10.9 2.5 # 2.1 7 11.2 1.2 2.4 9 6.9 ** † † 6 13.3 0.4 0.7 p 8 10.3 − 1.6 # − 0.9 9 10.4 0.3 1.4 5 8.9 ** − 2.6 7 11.9 0.2 0.5 .co 29 0.8 ** ** 21 1.4 4.4 0.6 19 1.9 ** ** 17 4.1 2.8 0.7 m 5 14.7 − 1.6 # − 0.9 10 8.3 2.0 # 1.5 8 7.3 ** ** 10 9.4 0.2 1.0 /jn c 15 5.5 4.5 # 2.4 6 11.7 2.0 # 1.5 11 6.1 2.3 0.7 9 9.7 2.9 # 3.0 i/a 6 12.6 − 3.8 # − 4.4 8 10.5 − 4.9 # − 4.3 10 6.6 − 6.9 # − 6.3 4 17.3 − 3.3 # − 6.3 rtic 12 8.0 − 1.2 − 1.1 12 6.2 − 2.2 # − 0.8 13 4.5 ** ** 12 7.7 − 0.6 − 0.4 le 13 7.5 0.9 1.0 14 4.2 − 1.3 − 1.7 18 2.0 ** ** 14 5.2 0.3 0.5 -a b 11 9.1 2.5 # 3.3 15 4.1 2.8 # 1.3 7 7.9 ** − 3.4 11 7.8 2.7 # 2.7 s 14 6.5 − 1.7 # − 0.3 13 5.8 − 0.1 − 0.7 14 4.0 ** ** 18 4.0 − 1.3 0.0 tra c 10 9.7 0.0 0.4 5 12.9 − 2.9 # − 4.0 6 7.9 ** ** 8 10.2 − 0.6 − 1.2 t/9 18 3.5 − 1.3 − 0.4 16 3.0 − 1.6 0.4 20 1.6 ** ** 15 4.6 0.6 0.4 7 /1 9 10.1 − 1.8 # − 1.4 17 2.8 1.5 0.1 16 3.5 ** ** 16 4.4 − 0.1 − 0.1 9 17 3.6 1.4 0.1 11 7.9 0.2 0.4 12 5.6 ** ** 13 5.4 5.0 # 2.6 /1 4 24 1.6 0.2 0.3 20 1.7 − 4.0 − 1.2 15 3.9 ** ** 19 4.0 − 3.0 − 0.5 0 7 /2 5 * Sources of data are the Surveillance, Epidemiology, and End Results (SEER) Program registries that include Connecticut, Hawaii, Iowa, Utah, and New Mexico; 2 1 the metropolitan areas of San Francisco, Detroit, Atlanta, Seattle-Puget Sound, San Jose-Monterey, and Los Angeles; rural Georgia and Alaska Natives in Alaska (i.e., 38 SEER13). Cancers are sorted in descending order according to sex-specifi c rates for all races. More than 15 cancers may appear for males and females to include the 4 b 15 most common cancers in every racial and ethnic group. APC = annual percent change; AC = absolute change; API = Asian/Pacifi c Islander; AI/AN = American y g Indian/Alaska Native; NOS = not otherwise specifi ed. u e † Data for Hispanics/Latinos excludes cases diagnosed in Detroit, Hawaii, Alaska Natives, and rural Georgia. st o ‡ Rates are per 100 000 persons and are age-adjusted to the 2000 U.S. Standard Population (using 19 age groups, with data provided from U.S. Bureau of the Census, n Current Population Reports, Series P25-1130. 08 § APC is based on rates that were age-adjusted to the 2000 U.S. Standard Population (using 19 age groups, with data provided from U.S. Bureau of the Census, A p Current Population Reports, Series P25-1130) ( 39) . ril 2 || AC was calculated as the difference in the age-adjusted rate for 2002 minus age-adjusted rate for 1992. 0 1 ¶ All sites excludes myelodysplastic syndromes and borderline tumors; ovary excludes borderline tumors. 9 # APC is statistically signifi cantly different from zero (two-sided P <.05). ** Statistic could not be calculated. APC based on fewer than 10 cases for at least 1 year during the time interval. † † Excludes borderline tumors. patients who received adjuvant therapy for colon and rectal can- clinical trials, treatment rates were lower for patients with stage cer increased rapidly between 1987 and 1992, following publica- II disease than for patients with stage III disease. Treatment pat- tion of results of relevant clinical trials in 1989 and 1990 ( 56 , 57 ) terns for colon and rectal cancers have also been examined using and the 1990 NIH Consensus Conference ( 54 ) . However, dis- SEER-Medicare data, and the fi ndings of the POC/QOC studies semination of adjuvant therapy varied with patient age, with have largely been corroborated ( 32 , 58 – 64 ) . much lower rates of treatment among older patients. Not surpris- A SEER-Medicare analysis examined the association between ingly, given the lack of positive evidence from randomized adjuvant therapy and patient characteristics. For Medicare Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005 SPECIAL ARTICLE 1415 Table 4. U .S. death rates and trends for the 15 most common cancers by sex and race/ethnicity for 1992 through 2002 * All races Whites Sex/cancer site or type Rank Rate ‡ APC § AC || Rank Rate ‡ APC § AC || Male All sites 258.6 − 1.5 ¶ − 36.7 252.5 − 1.4 ¶ − 32.9 Lung and bronchus 1 80.8 − 1.9 ¶ − 14.5 1 79.3 − 1.7 ¶ − 13.3 Prostate 2 33.9 − 3.6 ¶ − 11.1 2 31.2 − 3.7 ¶ − 10.5 Colon and rectum 3 26.3 − 2.0 ¶ − 5.5 3 25.8 − 2.2 ¶ − 5.8 Pancreas 4 12.3 − 0.3 ¶ − 0.5 4 12.0 0.0 − 0.2 Non − Hodgkin lymphoma 5 10.4 − 0.7 − 0.7 5 10.8 − 0.7 − 0.7 Leukemia 6 10.4 − 0.7 ¶ − 0.6 6 10.6 − 0.6 ¶ − 0.6 Urinary bladder 7 7.7 − 0.6 ¶ − 0.4 7 8.0 − 0.5 ¶ − 0.3 Esophagus 8 7.6 0.6 ¶ 0.5 8 7.2 1.6 ¶ 1.1 Stomach 9 7.1 − 3.4 ¶ − 2.2 9 6.3 − 3.6 ¶ − 2.1 Liver and intrahepatic bile duct 10 6.5 2.1 ¶ 1.4 12 5.9 2.1 ¶ 1.3 Kidney and renal pelvis 11 6.2 − 0.1 − 0.1 10 6.2 − 0.1 − 0.1 D Brain and other nervous system 1 2 5.7 − 0.7 ¶ − 0.4 11 6.1 − 0.7 ¶ − 0.4 ow Myeloma 13 4.8 − 0.5 ¶ 0.0 13 4.5 − 0.3 0.2 n Oral cavity and pharynx 14 4.5 − 2.7 ¶ − 1.1 15 4.2 − 2.3 ¶ − 0.8 loa Melanoma of the skin 15 3.9 − 0.1 − 0.1 14 4.4 0.0 0.0 de LSoarfyt ntixss ue including heart 1 167 2 1..76 −− 2 1..53 ¶ ¶ −− 0 0..72 1 167 2 1..46 − − 2 1..32 ¶ ¶ −− 0 0..62 d fro Fe m ale m h ALBurllen asgsi ttae nsd bronchus 1 2 1 4 26089...352 −− 0 2 0...746 ¶ ¶ ¶ − − 1 61 2...168 2 1 1 26 4871...091 − − 2 0 0...577 ¶ ¶ ¶ − − 1 60 3...574 ttps://a Colon and rectum 3 18.3 − 1.8 ¶ − 3.3 3 17.8 − 1.9 ¶ − 3.4 ca Pancreas 4 9.2 − 0.1 − 0.1 5 8.9 0.0 0.0 de ONvoanr −y H odgkin lymphoma 5 6 9 6..08 −− 0 0..59 ¶ − − 0 0..45 4 6 9 7..31 − − 0 0..49 − − 0 0..36 mic.o Leukemia 7 5.9 − 0.6 ¶ − 0.4 7 6.0 − 0.5 ¶ − 0.4 up Corpus and uterus, NOS 8 4.1 − 0.1 − 0.1 9 3.9 − 0.2 − 0.1 .co Brain and other nervous system 9 3.8 − 1.1 ¶ − 0.4 8 4.1 − 1.0 ¶ − 0.4 m Stomach 10 3.4 − 2.6 ¶ − 0.9 10 3.0 − 2.9 ¶ − 0.9 /jn Myeloma 11 3.2 − 0.4 − 0.1 11 2.9 − 0.4 ¶ 0.0 ci/a CLievrevri xa nudte irni trahepatic bile duct 1 123 3 2..09 − 3 1..12 ¶ ¶ − 1 0..04 1 134 2 2..77 − 2 1..71 ¶ ¶ − 0 0..74 rticle Kidney and renal pelvis 14 2.8 − 0.4 − 0.1 12 2.9 − 0.4 − 0.1 -a Urinary bladder 15 2.3 − 0.4 − 0.1 15 2.3 − 0.3 0.0 bs Esophagus 17 1.8 − 0.2 − 0.1 18 1.6 0.7 ¶ 0.1 tra c Oral cavity and pharynx 18 1.7 − 2.4 ¶ − 0.4 17 1.7 − 2.3 ¶ − 0.4 t/9 Gallbladder 20 0.9 − 2.5 ¶ − 0.3 20 0.9 − 2.7 ¶ − 0.3 7/19 (Table continues) /1 4 0 7 /2 5 patients aged 65 years or older who were diagnosed with stage III diagnosed with stage III colon cancer or with stage II or III rectal 2 1 3 c olon cancer from 1997 through 1999, SEER-Medicare data are cancer from 1996 through 1997. They found that 88% of patients 8 4 av ailable on patients’ receipt of adjuvant therapy by patient age younger than 55 years of age received adjuvant treatment, com- b y a t diagnosis and the number of comorbid conditions ( Table 7 ) pared with only 11% of patients older than 85 years. Physician g u ( 3 2 ) . Although receipt of postoperative adjuvant chemotherapy interviews revealed that patient refusal (30%), comorbid illness es w as inversely associated with the number of preexisting concur- (22%), and lack of perceived clinical benefi t (22%) were the t o n rcerne at sceodn wdiittiho nasg,e t, heev elink ewlihheono cdo omfo rrebciediivtyin wg aasd tjaukveann ti ntthoe aracpcyo udnet-. mo Tstr cenomdsm ino nL ruenagso Cnas nthceart aTdrjeuavtamnet ntht.e Sraepvye rwala sst undoite ds ehlaivveer beede. n 08 A p These data also show that among elderly patients the rates of conducted on patterns of care for patients with stage IV NSCLC. ril 2 hospitalization for complications from chemotherapy increased Although treatment for advanced-stage NSCLC is not without 01 9 modestly with increasing age. controversy ( 34 ) , published guidelines ( 65 , 66 ) indicate that che- The specifi c reasons for the failure to receive treatment are not motherapy may be benefi cial for patients whose cardiopulmo- directly documented in SEER-Medicare data. However, because nary status is adequate to allow them to undergo the treatment. Medicare claims permit ascertainment of referral patterns, Schrag Three studies examined the use of chemotherapy in patients et al. ( 32 ) found that most colon cancer patients who did not re- aged 65 or older at diagnosis with stage IV NSCLC. By using ceive chemotherapy never had a consultation with a medical on- instrumental variable analysis and propensity score techniques to cologist. This observation suggests that more research is needed analyze SEER-Medicare data, Earle et al. ( 67 ) found that the es- to determine how referral and access to specialty physicians may timated survival benefi ts for stage IV NSCLC patients aged 65 infl uence whether patients receive appropriate treatment. years or older who were treated with chemotherapy were similar Provider and patient perceptions and preferences also infl u- to the survival benefi ts found in randomized trials of younger ence treatment. To supplement SEER registry data from northern patients. In another study using SEER-Medicare data, Earle et al. California, Ayanian et al. ( 64 ) performed physician interviews ( 34 ) found that 22% of stage IV NSCLC patients diagnosed and extensive medical record reviews for patients who were from 1991 through 1993 received chemotherapy. Using NCI 1416 SPECIAL ARTICLE Journal of the National Cancer Institute, Vol. 97, No. 19, October 5, 2005