Department of Cancer Epidemiology and Genetics, H. Lee Moffitt Cancer Center, Tampa, Florida
Department of Cancer Epidemiology and Genetics, Program Leader, Risk Assessment, Detection, and Intervention Program, H. Lee Moffitt Cancer Center and Research Institute, 12902 Magnolia Drive, MRC-CANCONT, Tampa, FL 33612-9416===
Although cervical cancer incidence has declined in the past decade, considerable racial and ethnic differences remain. The objective of this study was to examine differences in incidence by histology and cancer stage in Florida stratified further by race, ethnicity, and 5-year time intervals.
Women who were diagnosed with invasive cervical cancer in Florida between January 1985 and December 2004 were included in the analysis. Age-adjusted incidence rates by race and ethnicity were estimated for different histologic types and stages of cancer. The annual percentage of change in incidence also was calculated for each histologic type. Rate ratios were estimated by race and ethnicity using whites and non-Hispanics as the reference group.
Overall, the incidence in Florida of cervical squamous cell carcinoma and transitional cell carcinoma declined significantly from 9.1 per 100,000 women in 1985 to 5.6 per 100,000 women in 2004 (P < .05), whereas the incidence of cervical adenocarcinoma remained stable (P > .05). The incidence of invasive cervical cancer was 9.6 per 100,000 women among whites and 13.13 per 100,000 women among African Americans from 2000 to 2004. African-American women were nearly 2 times more likely to be diagnosed at regional and distant cancer stages than white women for all periods examined. Furthermore, among African-American women aged >40 years, the age-specific incidence of invasive cervical cancer increased considerably, whereas the rates among other racial groups decreased.
The incidence of invasive cervical cancer (ICC) has decreased over the past several decades in the United States from 14.8 per 100,000 women in 1975, to 10.2 per 100,000 women in 1985, and to 8.9 per 100,000 women in 1995.1-4 This decline in ICC incidence persisted through 2004 (7.9 per 100,000 women) according to the US Cancer Statistics report.5 More than 11,000 cervical cancer cases are expected to be diagnosed in the United States in 2008 accompanied by an estimated 3870 deaths from the disease.6 Several studies have described a significant decrease in the incidence of squamous cell carcinoma (SCC), the most common histologic type of ICC.2-4 This decline in SCC incidence is most likely because of increased screening with the Papanicolaou (Pap) test.4 In contrast, the incidence rate of cervical adenocarcinoma has increased or remained relatively stable.3, 4 The declining rates of SCC relative to adenocarcinoma suggest that the Pap test may be less effective at detecting precancerous adenocarcinoma lesions.3
Although the incidence of cervical cancer has declined in recent decades, considerable racial and ethnic disparities exist. ICC incidence was highest among Hispanics (14.8 per 100,000 women) and African Americans (AAs) (13.5 per 100,000 women) during 1998 through 2002, based on data from the National Program of Cancer Registries (NPCR) and the Surveillance Epidemiology and End Results (SEER) Program.2 This pattern persisted when 2004 rates were examined by the NPCR (12.2 per 100,000 women among Hispanics and 10.8 per 100,000 women among AAs).5 Unfortunately, rates were not presented separately for Hispanic women living in Florida, a population comprised of immigrant populations from South and Central America and the Caribbean that are under-studied and may differ on several different factors related to cervical cancer risk compared with Mexican Americans.
The objective of this study was to assess cervical cancer incidence among the diverse racial and ethnic populations of Florida and to examine changes in incidence by race/ethnicity over the extended period from 1985 through 2004. For this analysis, we used data collected before the US Food and Drug Administration (FDA) licensure of the quadrivalent human papillomavirus (HPV) vaccine (Gardasil), which protects against cervical cancer caused by HPV types 16 and 18 and genital warts caused by HPV types 6 and 11. Another promising bivalent vaccine, Cervarix, is being tested by GlaxoSmithKline but has not yet been approved by the FDA. Thus, the current analysis was designed to identify high-risk, vulnerable populations for ICC and reports the baseline disease burden in the state of Florida before vaccine intervention.
MATERIALS AND METHODS
Data on the incidence of ICC were obtained from the Florida Cancer Data System (FCDS), a population-based cancer registry supported by the state of Florida Department of Health and the NPCR of the Centers for Disease Control and Prevention and housed at the Sylvester Comprehensive Cancer Center at the University of Miami, Miller School of Medicine. The FCDS has been collecting cancer incidence data since 1981 and meets the standards for quality set forth by the North American Association of Central Cancer Registries (NAACCR).
Women who were diagnosed with primary, malignant ICC between January 1, 1985 and December 31, 2004 who were living in Florida at the time of diagnosis were selected for analysis. ICC histology was classified using the International Classification of Disease for Oncology (third edition) 4-digit codes as follows: squamous cell carcinoma/transitional cell carcinoma (SCC/TCC), codes 8050 through 8084 and 8120 through 8131; adenocarcinoma, codes 8140 through 8149, 8160 through 8162, 8190 through 8221, 8260 through 8337, 8350 through 8551, and 8570 through 8576; and other cancers of the cervix, codes 8000 through 8049, 8085 through 8119, 8132 through 8139, 8150 through 8159, 8163 through 8189, 8222 through 8259, 8338 through 8349, 8552 through 8569, and 8577 through 9110). Most cases (97.8%) were confirmed microscopically. Patterns in ICC incidence rates were investigated by race, age at diagnosis, year of diagnosis, ethnicity, stage, region of the state of Florida, and histology type. Race was categorized as white and AA. Asians/Pacific Islanders and American Indians/Alaskan Natives were not analyzed separately because of the small number of cases when stratified by histology and stage (n < 25), although they were included in the “all races” category along with other and unknown races. Age at diagnosis was grouped as neonate to age 19 years and ages 20 to 29 years, 30 to 39 years, 40 to 49 years, 50 to 59 years, 60 to 69 years, 70 to 79 years, and ≥80 years. Calendar year of diagnosis was grouped into 4 5-year periods of diagnosis: from 1985 through 1989, from 1990 through 1994, from 1995 through 1999, and from 2000 through 2004. Ethnicity was classified as either Hispanic or non-Hispanic and was analyzed independent of race. Both race and ethnicity data were extracted from the medical records and coded according to NAACCR standards. The NAACCR Hispanic Identification Algorithm (NHIA) was applied to the FCDS data. However, because maiden names were missing from many FCDS records, the NHIA did not perform well on Florida cases. Therefore, the FCDS elected not to classify Hispanics in Florida based on the NHIA. Stage at diagnosis was categorized into localized, regional, distant, and unknown categories. The counties in the state of Florida were categorized into 3 regions based on their geographic location: north, central, and south.
SEER*Stat version 6.3.6 was used to analyze incidence data. Incidence rates were estimated per 100,000 women and were age-adjusted to the 2000 US Standard Population (19 age groups; Census P25-11307) with 95% confidence intervals (95% CIs) calculated based on the Tiwari modification for the Gamma method.8 The annual percentage change (APC) in total ICC incidence was calculated for the entire study period (1985 to 2004) using the weighted least-squares approach. Differences in ICC age-adjusted rates by histology, stage, race, and ethnicity were evaluated using rate ratios (RRs) and their 95% CIs. The statistical significance of the RR was assessed using a P value <.05. Age-specific incidence rates per 100,000 women also were calculated by race and ethnicity. Ethnicity data were not available for the period from 1985 to 1989; therefore, ethnicity analyses were restricted to the period from 1990 through 2004. SAS version 9.1.3 (SAS Institute, Cary, NC) was used to compute descriptive statistics on demographic variables of interest.
In total, 18,636 cases of ICC were diagnosed in Florida between 1985 and 2004 (Table 1), and the overall age-adjusted incidence rate was 11.90 per 100,000 women. Although white women comprised the majority of patients with ICC (80%), AA women had the highest rate of ICC (18.83 per 100,000) of any of the racial/ethnic groups examined. The ICC incidence rates were comparable for Hispanics (14.38 per 100,000 women) and non-Hispanics (14.90 per 100,000 women). Forty-five percent of all patients were diagnosed at localized stage, and 38% were diagnosed at regional or distant stages. The median age of patients who were diagnosed at the local stage was 45 years, and the ages increased to 53 years and 57 years for patients with regional and distant stages, respectively. SCC/TCC accounted for approximately 69% of the cases and had the highest incidence rate (8.33 per 100,000 women), whereas adenocarcinoma accounted for 16% of the cases and had an incidence rate of 1.84 per 100,000 women. The overall incidence of ICC decreased in Florida from 13.23 per 100,000 women during 1985 through 1989 to 9.90 per 100,000 women during 2000 through 2004. The incidence of ICC was highest in the southern region of Florida compared with the northern and central regions of Florida. There was no difference in ICC incidence between metropolitan and nonmetropolitan counties of Florida (data not shown).
Table 1. Demographic Characteristics of Invasive Cervical Cancer Cases, Florida Cancer Data System, 1985-2004
Incidence rates are average annual rates per 100,000 and are age-adjusted to the 2000 US Standard Population (19 age groups; US Census publication P25-1130; U.S. Census Bureau 19967).
All invasive cervical cancers
11.90 (11.73, 12.08)
SCC and TCC
Other cancers of the cervix
Year of diagnosis
Overall, the incidence rate for all cancers of the cervix decreased from 1985 to 2004 in Florida (APC, −2.045; P < .05) (Fig. 1). Similarly, there was a downward trend for SCC/TCC from 1985 to 2004 (APC, −2.34; P < .05). There was no change in the incidence of adenocarcinoma from 1985 to 2004.
Incidence RRs are presented for AA women compared with white women in Table 2 stratified by histology and by 5-year period of diagnosis. AA women in Florida were at increased risk for all cervical cancers, SCC/TCC, and other cancers of the cervix for all 4 periods studied. Comparable rates of adenocarcinoma were observed among AA and white women during 1985 through 1999, whereas AA women experienced a statistically significant lower risk of adenocarcinoma compared with white women during 2000 through 2004 (RR, 0.73; 95% CI, 0.57-0.93). Incidence rates for SCC/TCC decreased over time in both white women and AA women. Hispanic women in Florida tended to be at increased risk of ICC compared with non-Hispanic women, although this difference reached statistical significance only for the period from 2000 to 2004 (RR, 1.1; 95% CI, 1.01-1.19).
Table 2. Age-Adjusted Average Annual Incidence Rates and Rate Ratios of Invasive Cervical Cancer by 5-Year Period of Diagnosis, Histology, and Race/Ethnicity: Florida Cancer Data System, 1985-2004
In Florida, the percentage of AA women diagnosed at regional and distant stages was greater than the percentage of white women with similar diagnoses for all periods studied (regional-stage disease: 1985-1989, 30% for AA women vs 26% for white women; 1990-1994, 32% vs 26%, respectively; 1995-1999, 34% vs 29%, respectively; 2000-2004, 37% vs 33%, respectively; distant-stage disease: 1985-1989, 12% for AA women vs 8% for white women; 1990-1994, 10% vs 9%, respectively; 1995-1999, 12% vs 9%, respectively; 2000-2004, 12% vs 9%, respectively). In contrast, there was no consistent pattern in stage at diagnosis between Hispanic and non-Hispanic women. AA women had a statistically significantly increased risk of ICC compared with white women (P < .05) for all stages examined (Table 3). This racial difference tended to decrease over time, and no difference was observed for ICC diagnosed at localized stages during 2000 through 2004 (RR, 1.01; 95% CI, 0.89-1.16). There were no ethnic differences in stage-specific ICC risk, except for regional disease, for which Hispanics were at slightly elevated risk compared with non-Hispanics.
Table 3. Age-Adjusted Average Annual Incidence Rates and Rate Ratios of Invasive Cervical Cancer by 5-Year Period of Diagnosis, Stage, and Race/Ethnicity: Florida Cancer Data System, 1985-2004
For all races combined, ICC incidence in Florida was highest among women who were diagnosed between ages 40 years and 49 years (incidence rate, 21.21 per 100,000 women) (Fig. 2). The rates for white women and AA women remained similar until age 40 years, after which the rate continued to increase with age among AA women, reaching its peak of 45.61 per 100,000 women ages 70 years to 79 years. In contrast, ICC rates for white women peaked to 20.47 per 100,000 women ages 40 years to 49 years and then declined to 14.36 per 100,000 women aged ≥80 years at diagnosis. The patterns in age-specific incidence rates for Florida were comparable for Hispanic and non-Hispanic women.
Incidence rates decreased over time in northern and central Florida, from 13.9 and 13.5 per 100,000 women, respectively, during 1985 through 1989 to 8.7 and 9.8 per 100,000 women, respectively, during 2000 through 2004 (Fig. 3). In contrast, the incidence rate per 100,000 women increased in southern Florida from 12.9 during 1985 through 1989 to 14.8 during 1990 through 1994 (P < .05), after which incidence declined to 10.6 per 100,000 women during 2000 through 2004. Trend analysis from 1985 to 2004 indicated that there was a decrease in the incidence of ICC for all regions studied (north, APC, −3.03; central, APC, −2.12; south, APC, −1.68; data not shown). Furthermore, AA women in the south of Florida were 1.28 times more likely to be diagnosed with ICC than the AA women in the north (Table 4) (P < .05), whereas Hispanic women in the south were 1.71 times more likely to be diagnosed of ICC than Hispanic women in the north (P < .05).
Table 4. Age-Adjusted Average Annual Incidence Rates and Rate Ratios of Invasive Cervical Cancer by Region and Race/Ethnicity
AA women in Florida were at a significantly increased risk of ICC, consistent with previous observations from the entire US population,2, 4, 9, 10 and were approximately 2 times more likely to be diagnosed at regional and distant stages than white women. Moreover, AA women had the highest incidence rate of SCC/TCC among all racial groups in Florida, consistent with previous national studies.2, 4 Racial differences in Pap test screening and follow-up may account for observed racial differences in ICC incidence. According to the Behavioral Risk Factor Surveillance System (BRFSS) report in 2002, 83% of white women in Florida reported having had a Pap test in the last 3 years compared with 88% of AA women.11 Incomplete rates of diagnostic follow-up and subsequent late disease stage at diagnosis consistently have been higher among AA women compared with white women,12 which may explain why AA women are diagnosed with cervical cancer at later stages in Florida. Factors that may contribute to lower rates of follow-up after an abnormal Pap test among AA women13-15 include residence in medically underserved neighborhoods and socioeconomic status (SES), both of which are associated with late stage of cancer diagnosis16-18 and represent a multitude of factors that may have a negative impact on ICC rates. In the AA community, these include not having a usual source of care and lower levels of educational attainment.19 In addition, AA women often function in a multicaregiver role, which has the potential to conflict with compliant health-promoting behaviors and may result in AA women neglecting their personal health to attend to the needs of loved ones.20 Future research focused on the influence and potential confluence of these factors is needed so that prevention intervention strategies can be designed and implemented appropriately to decrease the disparate rates of ICC among older AA women in Florida. Also, the need for improved diagnostic and follow-up care after an abnormal Pap test in AA women is warranted.
The majority of the increased risk of cervical cancer in AA women compared with white women occurred among AA women aged ≥40 years, and the between-racial group gap widened with increasing age. This finding is consistent with results from a recent national study, which reported high incidence rates among AA women aged ≥50 years and a widening gap of ICC incidence between AAs and other racial/ethnic groups, particularly whites, as age increased.2 Therefore, age-specific Pap test screening recommendations should be revisited to better elucidate the behavioral factors underlying the low rates of Pap test screening with increasing age, especially for AA women. The US Preventive Services Task Force recommends screening within 3 years of sexual onset or from age 21 years up to age 65 years only if women had a previous negative test and are not high risk of cervical cancer.21 The American Cancer Society recommends cessation of Pap test screening at age 70 years, but only if there have been at least 3 recent tests and no abnormal results for 10 years, whereas the American College of Obstetricians and Gynecologists states that there is inconclusive evidence to establish upper age limits. Our results suggest that continuing cervical cancer screening after age 65 years should be considered for AA women in Florida. Furthermore, increased efforts to improve follow-up of abnormal Pap tests for AA women of all ages should be prioritized. ICC in AA women aged ≥65 years probably has been progressing for many years and could be prevented or diagnosed early in younger women.
In contrast to previous reports, the incidence of SCC, adenocarcinoma, and cervical cancer overall were comparable for Hispanics and non-Hispanic women in Florida.2, 16 This inconsistency in findings may be caused by differences in the country of origin for Hispanics in Florida versus the entire United States or by misclassification of ethnicity. The US 2000 Census reported that, in general, Hispanics in the United States are comprised predominantly of Mexican-Americans, particularly in the western states (74%) compared with the southern states (57%).22, 23 Specifically in Florida, Mexican-Americans represented only 14% of the Hispanic population, whereas Cuban-Americans and Puerto Ricans represented 31% and 18%, respectively.22 Moreover, approximately 66% of all Cuban-Americans in the United States live in Florida, and many have lived in Florida for multiple generations.24 Cuban-Americans tend to have a higher SES compared with other US Hispanic populations.24 For example, the percentage of the population below the Federal Poverty Level for Florida Hispanics (18%) is lower than that for US Hispanics (23%),25 a factor that has been associated significantly with higher overall cancer incidence and later stage at diagnosis of cancers among different racial/ethnic groups.16-18, 26, 27 However, the percentages of Hispanics with healthcare coverage were approximately equal for those among Florida and US Hispanics, suggesting that access to care does not explain all of the ethnic differences in incidence rates.25
The overall incidence of ICC declined in Florida during the period studied (1985-2004), whereas the rates for adenocarcinoma remained relatively stable, consistent with recent publications.2-4 The decreasing rates of SCC suggests that the Pap test is effective in reducing the burden of SCC incidence and mortality, whereas stable rates for adenocarcinoma suggest the need for an improved screening tool to detect lesions that are precursors to invasive adenocarcinoma.27 Liquid-based cervical cytology was a promising detection method for high-grade intraepithelial neoplasia, but systematic reviews have indicated that this new method does not perform better than the conventional Pap smear for the detection of high-grade lesions leading to SCC/adenocarcinoma.26, 28 Two studies also have demonstrated that the Pap test is not as effective for detecting adenocarcinoma lesions as it is for detecting SCC lesions.27, 29
In this study, we observed significantly higher rates of ICC among Hispanic women in the central and south regions of Florida compared with the north region (Table 4). BRFSS data from 9 Florida counties that were surveyed in 2002 indicate that 78.4% of Hispanic women reported having had a Pap smear in the past 3 years, a rate that was lower than the 90% target set by Healthy People 2010. Differences in the rates of ICC by Florida region may be related to recency of immigration and differences in screening by years of residence in the United States. Although the overall ICC rate for Hispanics in Florida was comparable to the rate for non-Hispanics, attention should be paid to increasing screening and appropriate treatment follow-up in recently immigrated populations, like those that reside in the central and southern regions of Florida.
There are several important limitations to the current analysis. The prevalence of HPV infection by race/ethnicity and age in Florida is unknown. HPV prevalence and information on other risk factors, such as smoking and SES, may help to explain the racial/ethnic discrepancies in ICC rates. Pap test data from the National Health Interview Survey and BRFSS are self-reported; therefore, screening rates may not be reliable, especially for women of lower SES. Several studies have demonstrated that women with lower SES greatly overestimate the prevalence of Pap smear screening.30-32 Paskett et al reported that low-income minority women thought they had received Pap tests more recently than they had by an average of 23 months.31 Calculation of ICC incidence rates using US census data resulted in the inclusion of women who underwent a complete hysterectomy in the denominator, which may have underestimated ICC incidence rates in Florida, because the rates of hysterectomy are highest in the southern region of the United States,33-35 and women who have undergone hysterectomy are not at risk of developing cervical cancer.
In conclusion, the rates of cervical cancer decreased in Florida during the period from 1985 through 2004. However, significant racial/ethnic disparities exist, particularly by histologic type, age at diagnosis, and stage of disease. The increasing rate of invasive cervical cancer among AA women aged >40 years, coupled with their diagnosis at a later stage of cancer, is of great concern. The observations from these analyses highlight the need to focus prevention, screening, and treatment efforts on AA women, and particularly on women of postreproductive age.
Conflict of Interest Disclosures
Supported by Merck and Company, Inc. (proposal 32088).