SEARCH

SEARCH BY CITATION

Keywords:

  • neoplasms;
  • health status disparities;
  • minority health;
  • Georgia;
  • geographic factors

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

BACKGROUND:

The objective of this study was to evaluate racial cancer disparities in Georgia by calculating and comparing mortality-to-incidence ratios (MIRs) by health district and in relation to geographic factors.

METHODS:

Data sources included cancer incidence (Georgia Comprehensive Cancer Registry), cancer mortality (Georgia Vital Records), and health factor (County Health Rankings) data. Age-adjusted incidence and mortality rates were calculated by cancer site (all sites combined, lung, colorectal, prostate, breast, oral, and cervical) for 2003-2007. MIRs and 95% confidence intervals were calculated overall and by district for each cancer site, race, and sex. MIRs were mapped by district and compared with geographic health factors.

RESULTS:

In total, 186,419 incident cases and 71,533 deaths were identified. Blacks had higher MIRs than whites for every cancer site evaluated, and especially large differentials were observed for prostate, cervical, and oral cancer in men. Large geographic disparities were detected, with larger MIRs, chiefly among blacks, in Georgia compared with national data. The highest MIRs were detected in west and east central Georgia, and the lowest MIRs were detected in and around Atlanta. Districts with better health behavior, clinical care, and social/economic factors had lower MIRs, especially among whites.

CONCLUSIONS:

More fatal cancers, particularly prostate, cervical, and oral cancer in men were detected among blacks, especially in central Georgia, where health behavior and social/economic factors were worse. MIRs are an efficient indicator of survival and provide insight into racial cancer disparities. Additional examination of geographic determinants of cancer fatality in Georgia as indicated by MIRs is warranted. Cancer 2012. © 2012 American Cancer Society


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Racial disparities in cancer incidence and mortality are often large and may result from complex causes. Mortality rates from many common cancers are higher among blacks compared with any other racial/ethnic group.1, 2 For all cancer sites combined in the United States, the mortality rate among blacks is 25% higher than among whites.3 Incidence also is higher among blacks, although not for all cancer sites; and racial differences in incidence tend to be smaller than mortality differentials. Blacks are diagnosed with more aggressive tumors and tend to present at younger ages; findings that help explain why the racial gap in mortality from cancer is so large. Racial cancer disparities are especially prominent in the southeast4, 5; however, explanations remain unclear.

Calculation of the mortality-to-incidence ratio (MIR) provides a unique quantification of racial cancer mortality disparities. Given complete and reliable incidence and mortality data, the MIR is a valid6 indicator of the fatality ratio.6-8 The MIR approximates a standard, population-based measure of fatality (1/survival), given incidence.4 Most investigators have used MIRs to quantitatively evaluate the completeness and validity of cancer registration,7-9 to evaluate the effectiveness of secondary prevention methods,10 or to estimate incidence or mortality when direct data are unavailable.11-13 Few studies have used the MIR to compare cancer rates.4, 14-16

Recently, Hébert et al used MIRs to compare cancer fatality by race in South Carolina.4 They were the first to map MIRs, which allowed them to highlight geographies where mortality disparities may be particularly pronounced.4 Georgia, like South Carolina, has large racial cancer disparities. Despite the presence of a Surveillance, Epidemiology, and End Results (SEER)-affiliated cancer registry with valid and complete data and a large proportion of black residents that enables statistically stable comparisons across racial subgroups, there is a paucity of published research describing cancer throughout Georgia.

Deriving inspiration from the work of Hébert et al, the purpose of this study was to evaluate racial cancer disparities in Georgia by calculating and comparing MIRs for a variety of cancer sites. We calculated MIRs for all cancer sites combined and for lung and bronchus, colorectal, prostate, female breast, oral cavity, and cervical cancers by race, sex, and Georgia public health district. In this study, we also assessed the geographic pattern of MIRs overall and in relation to spatial variables, including environmental characteristics, socioeconomic factors, and health care accessibility.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Data Sources

Cancer incidence data were obtained from the Georgia Comprehensive Cancer Registry (GCCR) for 2003 to 2007. The GCCR is a statewide, population-based cancer registry that collects information on all cancer cases diagnosed among Georgia residents. The GCCR is a participant in the National Program for Cancer Registries and the North American Association of Central Cancer Registries (NAACCR). The GCCR meets national standards for cancer registration and is gold certified for data quality and representativeness. Each cancer case was associated with a county-level geographic identifier, demographic, and tumor characteristics. Cancer sites were selected using the SEER Site Groups Primary Site variable based on International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) coding.17

Cancer mortality data from Georgia Vital Records in the Georgia Department of Public Health's standardized health data repository were obtained through the GCCR. Each cancer death was associated with a county-level geographic identifier and demographic characteristics. Cancer deaths for specific cancer site subgroups were identified by the International Classification of Diseases 10th Revision (ICD-10) code for underlying cause of death.

Incidence and mortality data were obtained for all cancer sites combined and for lung and bronchus, colorectal, prostate, female breast, oral cavity, and cervical cancers and were obtained by race (white, black) and sex, when applicable. Incident cases and deaths were excluded if they were of unknown or “other” race or sex. Only malignant tumors were included. Ethnicity was not used, because the mortality data did not contain reliable ethnicity information. To avoid unstable statistical calculations, incidence and mortality data were aggregated to the public health district level (N = 18). All analyses were performed for the state of Georgia or by health district.

Age-specific Georgia population data for 2003 to 2007 were obtained from National Center for Health Statistics bridged population estimates. Health factor data were obtained from the County Health Rankings 2011 data portal,18 which synthesizes health data from national data sources, including the Behavioral Risk Factor Surveillance System, the National Center for Health Statistics, and the Centers for Disease Control Environmental Protection Agency Collaboration, among others.

Statistical Analysis

Age-adjusted mortality and incidence rates were calculated using direct standardization for Georgia, public health district, and race/sex subgroups.19 The 2000 US standard million population was used. Rates were not calculated for cancer sites with fewer than 20 cases per health district. The MIR, a measure of mortality that adjusts for underlying differences in incidence, was calculated as the age-adjusted mortality rate divided by the age-adjusted incidence rate. MIRs were calculated overall and by health district for each cancer site and for race/sex subgroups. Rates and MIRs were generated using Microsoft Excel (2007; Microsoft Corporation, Redmond, Wash) and SAS version 9.1.3 (SAS Institute, Cary, NC). We modeled the statistical analysis approach used by Hébert et al4 in our neighboring state, South Carolina, so that direct comparisons with their findings could be made.

Special methods previously used for MIRs were implemented to calculate 95% confidence intervals (CIs) using F intervals.20 These F intervals, which approximate directly standardized rate intervals, are useful for scenarios in which some data cells are sparse (eg, districts with few cases) whereas others are large (eg, districts with many cases). The resulting CIs guarantee nominal coverage yet provide conservative estimates.20 CIs were generated using R software version 2.12.1 (The R Foundation for Statistical Computing, Vienna, Austria).

A geographic information system (ArcMAP software, version 10.0; ESRI, Redlands, Calif) was used to map MIRs for each cancer site by Georgia health district. For comparison with national rates, display cutoff points introduced by Hébert et al were used. Four categories were defined for each cancer site as follows: Category 1, the mean MIR for whites nationally; Category 2, the upper bound 10% higher than the upper bound of Category 1; Category 3, the upper bound 20% higher than the upper bound of Category 1; and Category 4, the upper bound >20% higher than the upper bound of Category 1.4 The data for the computation of Category 1 were obtained from the North American Association of Central Cancer Registries' report Cancer in North America: 2003-2007.17

To explore potential geographic risk factors of cancer fatality, MIRs were compared with health factor groupings by health district. Health factors were represented by data metrics, which were chosen and weighted based on their relative importance while considering data reliability and availability. Specifically, we used data for 4 health factors including health behaviors, a composite measure of tobacco use (adult smoking rate, 10%), diet and exercise (adult obesity rate, 10%), alcohol use (excessive drinking, 2.5%; motor vehicle crash death rate, 2.5%), and unsafe sex (sexually transmitted infection rate, 2.5%; teen birth rate, 2.5%); clinical care, a composite measure of access to care (adult uninsured rate, 5%; primary care providers, 5%) and quality of care (hospitalization rate for ambulatory-sensitive conditions, 5%; diabetic screening rate, 2.5%; mammography screening rate, 2.5%); socioeconomic factors, a composite measure of education (high school graduation rate, 5%; adults with college degrees, 5%), employment (unemployment rate, 10%), income (children in poverty, 10%), family and social support (social and emotional support, 2.5%; single-parent households, 2.5%), and community safety (violent crime or homicide rate, 5%); and physical environment, a composite measure of environmental quality (unhealthy air quality because of particulate matter, 2.5%; unhealthy air quality because of ozone, 2.5%) and built environment (access to healthy foods, 2.5%; access to recreational facilities, 2.5%). The County Health Rankings, which were created around the model of the United Health Foundation's Annual State Health Rankings, originally were developed to use publicly available data to describe variation in population health measures for community health assessment in Wisconsin.21 The Mobilizing Action Toward Community Health (MATCH) project was initiated with funding from the Robert Wood Johnson Foundation and expanded Wisconsin's County Health Rankings in 2010 to the entire country.22 These rankings provide a concrete picture of overall community health and a simple method for communities to describe and track progress.23 Health determinants, as part of the County Health Rankings metric, are described using measures that are chosen based on certain criteria, including the ability to measure, either directly or by proxy, the important aspects of public health; public availability; periodic updates at the county level; consistently collected across counties; and sufficient in quantity to produce moderately stable county-level estimates.24

County-level Z-scores (ie, standardized measures of each health measure relative to other Georgia counties) for each health factor were averaged to obtain an overall Z-score for each health district. A positive average Z-score indicated an average value higher (ie, “greater risk” for worse health outcomes) than the average of all Georgia counties; a negative average Z-score indicated an average value lower (ie, “lower risk” for worse health outcomes) than the average of all Georgia counties. For each health factor, correlations of Z-scores to MIRs at the health district level were calculated. Pearson correlation coefficients and 1-sided P values were calculated for each cancer site by race and sex. Maps of the health factor groupings by health district were generated.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

In total, 186,419 incident cancers (all sites) were identified. Incident cases were excluded if they were of unknown or “other” race/sex (N = 2987; 2%). Only tumors with malignant behavior were included in the analysis (thus, 3300 cases [2%] with in situ behavior were excluded). In total, 71,533 cancer deaths from all cancer sites were identified. Deaths were excluded if they were of unknown or “other” race (N = 602; 1%).

Table 1 displays the MIRs and 95% CIs for all cancer sites combined and for female breast, cervix, colorectal, lung and bronchus, oral cavity, and prostate cancers by race and sex. When considering all cancer sites combined, as expected, white women had the lowest MIRs (0.37), and black men had the highest MIRs (0.47) (Table 1). MIRs were greater (ie, cancers were more fatal) among blacks for each and for every cancer site and sex combination compared with the MIRs among whites. These differences were statistically significant for all groups except for lung and bronchus cancer in both sexes and oral cancer among women (Table 1). Like what was observed in South Carolina,4 the largest racial differences were for oral cancer among men (the MIR was 1.81 times greater for blacks than for whites), prostate cancer (the MIR was 1.55 times greater for blacks than for whites), and cervical cancer (the MIR was 1.52 times greater for blacks than for whites). MIRs were greater for men than for women. Among all the cancer sites, lung cancer was the most fatal (MIR range, 0.72-0.83), and prostate cancer was the least fatal (MIR range, 0.16-0.24) (Table 1).

Table 1. Georgia Mortality-to-Incidence Ratios and 95% Confidence Intervals for All Cancer Sites Combined and For Specified Cancer Sites by Race and Sex, 2003-2007
Cancer Site: Sex SubgroupBlacksWhitesBlack:White Ratioa
MIR95% CIMIR95% CI
  • Abbreviations: CI, confidence interval; MIR, mortality-to-incidence ratio.

  • a

    The black:white ratio is the MIR for blacks divided by the MIR for whites.

  • b

    The difference was statistically significant for MIRs between blacks and whites based on nonoverlapping 95% CIs.

All sites combined     
 Overallb0.4500.442, 0.4590.4010.396, 0.4051.122
 Womenb0.4430.432, 0.4550.3730.368, 0.3791.188
 Menb0.4710.458, 0.4830.4320.426, 0.4391.090
  Ratio of men to women1.0631.158
Female breast     
 Womenb0.2550.242, 0.2690.1810.175, 0.1881.408
Cervix     
 Womenb0.4230.362, 0.4920.2790.247, 0.3151.516
Colon and rectum     
 Overallb0.4070.386, 0.4300.3430.331, 0.3551.187
 Womenb0.3970.369, 0.4270.3390.322, 0.3561.171
 Menb0.4260.391, 0.4630.3520.335, 0.3701.210
  Ratio of men to women1.0731.038
Lung and bronchus     
 Overall0.7930.761, 0.8260.7700.754, 0.7851.029
 Women0.7490.701, 0.7990.7210.699, 0.7441.038
 Men0.8270.783, 0.8730.8120.790, 0.8341.018
  Ratio of men to women1.1041.126
Oral cavity     
 Overallb0.3330.291, 0.3800.2130.197, 0.2301.563
 Women0.2250.172, 0.2920.2150.187, 0.2481.047
 Menb0.3980.337, 0.4690.2200.199, 0.2431.809
  Ratio of men to women1.7691.023
Prostate     
 Menb0.2420.228, 0.2560.1560.149, 0.1631.551

We compared Georgia MIR calculations with those computed for South Carolina.4 South Carolina had higher MIRs among blacks for all cancer sites except cervical cancer, for which the MIR in Georgia was equivalent (0.42 vs 0.41). Among whites, MIRs were strikingly similar for all cancer sites combined and for breast, cervical, and colorectal cancers for South Carolina and Georgia. MIRs for lung, oral, and prostate cancers among whites were lower in Georgia than in South Carolina.

Figures 1 through 7 display MIRs according to Georgia health district and race using display cutoff points that compare MIRs against national values for whites. MIRs were higher in Georgia, chiefly among blacks, compared with national data. For all cancer sites combined among men (Fig. 1A), the highest MIRs were observed in southeast Georgia, which has more Category 4 districts (ie, >20% higher than the mean MIR for whites nationally), among black men (6 vs 4). For all cancer sites combined among women (Fig. 1B), the MIRs between races differed remarkably, with a 1-category difference in every district and a 2-category difference in most districts for blacks throughout the state. Large differences in MIRs also were observed by race for breast cancer (Fig. 2). Five districts had a 3-category difference from Category 1 (comparable to the mean MIR for whites nationally) among whites to Category 4 among blacks. Many of the districts have suppressed data for cervical cancer; however, it should be noted that all of the districts highlighted among black women were in the 2 highest MIR categories (Fig. 3).

thumbnail image

Figure 1. (A) Mortality-to-incidence (MIR) ratios are illustrated by Georgia public health district for all cancer sites combined in men. (B) MIRs are illustrated by Georgia public health district for all cancer sites combined in women.

Download figure to PowerPoint

thumbnail image

Figure 2. Mortality-to-incidence ratios are illustrated by Georgia public health district for female breast cancer.

Download figure to PowerPoint

thumbnail image

Figure 3. Mortality-to-incidence ratios are illustrated by Georgia public health district for cervical cancer.

Download figure to PowerPoint

For colorectal cancer, high MIRs were detected in Midwest Georgia with all but 3 districts having higher MIRs for blacks and 4 districts with a 3-category difference for blacks compared with whites (Fig. 4). Racial differences in MIRs were less drastic for lung cancer (Fig. 5), and 8 districts had MIRs in a higher category for blacks versus whites. For oral cancer, all of the available districts were in MIR Category 4 (Fig. 6). Finally, for prostate cancer, the majority of Georgia was in Category 4 for blacks with a 3-category difference in 8 districts for blacks versus whites (Fig. 7). Across all maps, the districts with the most Category 4 MIRs were the West Central Health District 7-0 (9 appearances in Category 4; 6 black, 3 white) and the East Central Health District 6-0 (8 appearances in Category 4; 5 black, 3 white). Conversely, the districts with the most Category 1 MIRs were the Cobb Douglas Health District 3-1 (7 appearances in Category 1; 7 white), the Fulton Health District 3-2 (7 appearances in Category 1; 2 black, 5 white), and the Dekalb Health District 3-5 (7 appearances in Category 1; 7 white).

thumbnail image

Figure 4. Mortality-to-incidence ratios are illustrated by Georgia public health district for colorectal cancer.

Download figure to PowerPoint

thumbnail image

Figure 5. Mortality-to-incidence ratios are illustrated by Georgia public health district for lung cancer.

Download figure to PowerPoint

thumbnail image

Figure 6. Mortality-to-incidence ratios are illustrated by Georgia public health district for oral cancer.

Download figure to PowerPoint

thumbnail image

Figure 7. Mortality-to-incidence ratios are illustrated by Georgia public health district for prostate cancer.

Download figure to PowerPoint

Table 2 displays the correlations between health district MIRs and health factor groupings. Expectations were that higher average Z-scores (ie, health districts at a “higher risk” for worse health outcomes) would be correlated positively with higher average MIRs (ie, health districts with more fatal cancers). Correlations with moderate (ρ, 0.30-0.70) to strong (ρ, 0.70-1.00) coefficients and statistically significant (P < .05) P values were detected between increasing Z-scores for health behavior (5 cancer/race/sex subgroup combinations with strong correlations [ρ, 0.70-1.00; P < .05], and 3 with moderate correlations [ρ, 0.30-0.70; P < .05]), clinical care (2 strong correlations; 6 moderate correlations), social/economic factors (11 moderate correlations), and increasing MIRs, especially among whites for all cancer sites combined and for lung cancer. Overall, the most positive (11 of 30 cancer/race/sex subgroups), statistically significant correlations were observed between social/economic factor Z-scores and MIRs (Table 2). Z-scores for physical environment had the weakest correlation with increases in MIRs. The health districts that were highlighted for health behavior and social/economic factors were almost identical, and the physical environment groupings were the most dissimilar (Fig. 8). Five health districts, all in Southwest and South Central Georgia (Districts 07-0, 05-2, 05-1, 08-1, and 09-2) had poor rankings for 3 of 4 health factors (Fig. 8). The West Central Health District 7-0, as noted above, also had the most Category 4 MIRs for both races. No health district was ranked poorly for all health factors.

Table 2. Correlation Coefficients Between Health District Health Factor Groupings and Average Health District Mortality-to-Incidence Ratios for All Cancer Sites Combined and for Specified Cancer Sites by Race and Sex, 2003-2007
Cancer Site: Sex SubgroupBlacksWhites
No.ρPaNo.ρP
  • Abbreviations: No. the number of health districts (of 18) included in analysis; ρ, Pearson correlation coefficient.

  • a

    P values are 1-sided.

Health behavior      
 All sites combined      
  Overall180.4630.03180.860<0.0001
  Women180.2870.13180.713<.0001
  Men180.5480.01180.864<.0001
 Female breast      
  Women160.0100.49180.1510.28
 Cervix      
  Women50.4430.237−0.0580.55
 Colon and rectum      
  Overall17−0.2240.81180.3110.11
  Women15−0.3980.93180.0820.38
  Men140.0600.42180.3110.11
 Lung and bronchus      
  Overall170.3940.06180.708<.0001
  Women16−0.2820.86180.2590.15
  Men17−0.1560.73180.805<.0001
 Oral cavity      
  Overall70.4230.17160.3260.11
  Women050.0740.46
  Men40.7090.15140.1980.25
 Prostate      
  Men160.4770.03180.2510.16
Clinical care      
 All sites combined      
  Overall180.3550.08180.715<.0001
  Women180.1720.25180.5800.005
  Men180.4790.02180.718<.0001
 Female breast      
  Women160.2840.15180.1030.35
 Cervix      
  Women50.1580.407−0.2360.70
 Colon and rectum      
  Overall17−0.3200.9018−0.0930.65
  Women15−0.4580.9618−0.3030.89
  Men140.2420.21180.1090.34
 Lung and bronchus      
  Overall170.1210.32180.4610.03
  Women16−0.5510.99180.1440.29
  Men170.0320.45180.4890.02
 Oral cavity      
  Overall70.5520.10160.4810.03
  Women050.1800.39
  Men40.7340.14140.1690.28
 Prostate      
  Men160.5790.01180.2390.17
Social and economic factors      
 All sites combined      
  Overall180.5810.005180.6470.002
  Women180.4250.04180.4980.02
  Men180.6350.003180.6760.001
 Female breast      
  Women160.2680.16180.3530.08
 Cervix      
  Women50.3500.2870.6820.05
 Colon and rectum      
  Overall17−0.3010.88180.2920.12
  Women15−0.4920.97180.1230.32
  Men14−0.1430.69180.2300.18
 Lung and bronchus      
  Overall170.4670.03180.6720.001
  Women160.1070.35180.3970.05
  Men17−0.2310.82180.6180.005
 Oral cavity      
  Overall70.1470.38160.3500.09
  Women05−0.4920.80
  Men40.5160.24140.3920.09
 Prostate      
  Men160.5530.02180.4670.03
Physical environment      
 All sites combined      
  Overall18−0.2650.8618−0.4450.97
  Women180.0180.4718−0.4220.96
  Men18−0.4490.9718−0.4140.96
 Female breast      
  Women160.1840.25180.2460.16
 Cervix      
  Women50.1830.397−0.2900.74
 Colon and rectum      
  Overall170.3660.0818−0.1690.75
  Women150.1260.33180.0190.47
  Men140.2720.1818−0.1720.75
 Lung and bronchus      
  Overall170.1590.2718−0.2180.81
  Women160.3790.0818−0.0720.61
  Men17−0.2760.8618−0.1830.77
 Oral cavity      
  Overall7−0.7100.9716−0.2350.81
  Women05−0.0900.56
  Men4−0.9830.99140.0310.46
 Prostate      
  Men16−0.6940.9918−0.2310.82
thumbnail image

Figure 8. Average health factor scores are illustrated by Georgia public health district.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

This study describes racial cancer mortality disparities and their potential geographic determinants by calculating, comparing, and mapping MIRs throughout Georgia. Blacks in Georgia had more fatal cancers than whites for all evaluated cancer sites. Although this observation has been made for the nation2 and for Georgia,5 the use of MIRs in this study allowed an efficient description of these disparities in cancer mortality using pre-existing data while standardizing for incidence. This standardization is important for racial cancer disparities, as Blacks are diagnosed more frequently and with more aggressive tumors than Whites, and especially in those instances where incidence rates are lower among Blacks (e.g., breast cancer), but mortality is higher.

MIRs provide an efficient and accurate indication of cancer fatality that has previously been unreported for Georgia. This is meaningful when describing racial disparities in cancer outcomes, because cancer originating at different anatomic sites may present differently in terms of aggressiveness. In contrast to simply mapping mortality (or incidence) rates, calculating and mapping MIRs, especially by race, allows a broader and perhaps more complete depiction of the survival experience.

The MIR differential is an indicator of a difference in mortality in excess of what would be predicted based on incidence alone. For prostate cancer, for example, the MIR for blacks in this study was 1.6 times the MIR for whites. This is the amount of excess mortality among blacks controlling for the finding that blacks have a higher incidence than whites. This observation demonstrates that, even controlling for the finding that blacks get more cancer than whites, they still have a disproportionately large mortality burden. Using MIRs to assist in quantifying this difference is an essential first step and highlights where efforts need to be focused, an important task given limited resources.

Higher MIRs were observed among blacks in South Carolina compared with blacks in Georgia. MIRs among whites were comparable. The main exceptions were for cervical cancer, for which Georgia had MIRs similar to those of South Carolina among blacks. Compared with South Carolina, Georgia had lower MIRs for lung, oral, and prostate cancers among whites. A large difference was observed for cervical cancer in Georgia, for which the MIR among blacks was 1.52 times the MIR among whites. This difference is similar to that observed for South Carolina (the MIR was 1.50 times higher for blacks than for whites).4 This mortality difference is alarming, because black women are more likely to receive Papanicolaou (Pap) tests than white women. Racial differences in cervical cancer screening follow-up or lack of culturally and linguistically specific educational messages about cervical cancer and the meaning of abnormal Pap test results may be contributing to this disparity25, 26; and interventions, educational materials, and/or programs should be targeted in Georgia based on these findings. In addition, Hispanic women in Georgia have higher incidence rates of cervical cancer (14.1/100,000 population) compared with national rates (12.5/100,000 population).27 This may be 1 reason that higher MIRs were detected for cervical cancer in Georgia, but we were unable to examine the issue directly, because ethnicity data were not evaluable. However, based on the incidence data, 0.8% of black women and 2.5% of white women were classified as Hispanic. The inclusion of this small percentage is unlikely to meaningfully alter our results; and, in fact, the larger percentage of potential misclassification was among white women, and not black women, in whom the higher MIR was detected. This suggests that Georgia is performing slightly better in terms of cancer outcomes compared with South Carolina, except for cervical cancer, where efforts should be focused.

Compared with national data, Georgia had higher MIRs for each and every cancer site examined. Overall, the worst outcomes were detected in the West and East Central health districts. Identification of districts with large disparities provides support for the development of area-specific and population-specific health programs and/or focused research studies with the aim of better describing the etiologic mechanisms behind these disparities. The best outcomes were detected in the Atlanta area in the Cobb-Douglas, Dekalb, and Fulton health districts. Compared with health factors, which rank counties based on health behaviors, clinical care, and social/economic factors,18 there appeared to be a relation between the overall health of the county and MIRs. We chose to further examine the geographic relation of MIRs at the health district level by comparing MIRs by health factors from this same data repository. We observed that districts whose population had worse health behavior, worse clinical care choices and access, and worse social/economic factors tended to have higher MIRs compared with those health districts whose populations ranked better in these factors. For all cancer sites combined, higher MIRs (ie, more fatal cancers) were strongly associated with worse health behavior and worse clinical care (overall and among men) and were moderately associated with a worse social/economic environment and worse clinical care among women. Similar relations were detected for lung cancer and, to a lesser extent, prostate cancer. Although this does not suggest causality, these relations may warrant further investigation or may suggest opportunities for health behavior interventions, such as tobacco cessation for lung cancer, and/or screening programs, such as community-based education about screening practices or treatment decisions28 among blacks for prostate cancer. In general, this trend was more apparent among whites, suggesting that other factors are at work among blacks in terms of cancer outcomes. A more detailed examination of geographic risk factors related to MIR disparities certainly is warranted, but the current analysis serves as a starting point for developing more specific and testable hypotheses to further explain and eventually mitigate these disparities.

The County Health Rankings system used to describe geographic health risk factors in relation to MIRs is not without limitation. Because the County Health Rankings were not designed to be a statistically robust measure of community health24 and individual metrics may have questionable reliability because of sparse data,18 these rankings should not be used as the statistical basis for assessing policy change or priority settings, nor should small rank-order changes be over interpreted.23 However, a thorough process supported the choice of measures and weights, including literature reviews of the impact of factors and health outcomes, the ability for factors to be modified through community action, a review of America's Health Rankings methodology, the availability and reliability of indicators at the county level, and feedback from an expert panel.18 When considered in summary, the County Health Ranking system provides a high-quality picture of overall community health18 and can be used to generate hypotheses about possible drivers of racial disparities, as we did in the current study.

Because direct age adjustment was performed and Georgia has a large number of counties (N = 159), public health districts (N = 18) were used to describe the geographic pattern of MIRs throughout Georgia while ensuring a sufficient sample size for the calculation of stable rates. The GCCR suppresses rate calculations for which cell counts are <20. Therefore, rates for more rare cancer sites (eg, cervical, oral) were considered unstable and were suppressed based on this criterion. Although this may have limited our ability to distinguish patterns for these cancer sites throughout the entire state, this conservative approach ensured that the calculated rates were stable and reliable.

The MIR is an efficient indicator of fatality (or survival). Despite its utility, most studies have generated the MIR for other purposes, particularly to validate cancer registration,7, 8, 29 and only a few4, 14-16 have used the MIR to compare cancer rates across populations. Given appropriate mortality and incidence data, an MIR can be calculated easily and avoids the limitations of traditional survival studies, which require active follow-up and potential financial and logistical constraints.6 Cancer registry data in Georgia is thorough, and its completeness meets the NAACCR gold standard (95% completion) overall and by racial subgroup (completeness for whites, 106%; completeness for blacks, 99%). Since completeness is computed as the ratio of observed to estimated expected counts, where the estimate of expected counts is based on SEER rates applied to state population characteristics (in addition to other factors), the estimated number of expected counts may actually be lower than the number of observed counts, resulting in a percent greater than 100. Not only is the MIR an efficient way to calculate and compare survival, a recent study confirmed that the measure provides a valid approximation of the 5-year survival rate for most tumor sites.6

In summary, for this study, we used MIRs to describe racial cancer mortality disparities in Georgia. Larger MIRs were detected for blacks compared with whites for all cancer sites and sex subgroups, and the largest racial differentials were detected for prostate cancer, cervical cancer, and oral cancer in men. More fatal cancers were detected in West and East Central Georgia, and less fatal cancers were detected in and around Atlanta. Compared with national MIRs, geographic differentials were detected: Georgia had drastically larger MIRs for each and every cancer site examined. Better health behavior, clinical care characteristics, and social/economic factors were correlated with lower MIRs, especially among whites. MIRs are an efficient indicator of survival and are useful in describing racial cancer disparities. Additional examination of the geographic determinants of cancer fatality in Georgia indicated by these MIRs is warranted.

The consistency of our findings with the sex and racial disparities in South Carolina suggest that there are many fundamental issues contributing to differentials in survival that need to be investigated further and addressed. Our current analysis highlights regional patterns in racial disparities in cancer mortality that encourage additional research collaborations and evaluation of potential regional-specific determinants. Targeted, race-specific, and geographic-specific prevention, treatment, and follow-up practices are needed to address the observed differences in MIRs.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES

Dr. Wagner is funded by a grant from the Georgia Cancer Coalition (Proposal 038505). Dr. Hébert is supported by an Established Investigator Award in Cancer Prevention and Control from the Cancer Training Branch of the National Cancer Institute (K05 CA136975).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SOURCES
  8. REFERENCES
  • 1
    Kohler BA, Ward E, McCarthy BJ, et al. Report to the nation on the status of cancer, 1975-2007, featuring tumors of the brain and other nervous system. J Natl Cancer Inst. 2011; 103: 714-736.
  • 2
    Ward E, Jemal A, Cokkinides V, et al. Cancer disparities by race/ethnicity and socioeconomic status. CA Cancer J Clin. 2004; 54: 78-93.
  • 3
    National Cancer Institute. Cancer Health Disparities. Bethesda, MD: National Cancer Institute; 2011. Available at: http://www. cancer.gov/cancertopics/factsheet/disparities/cancer-health-disparities. Accessed July 19, 2011.
  • 4
    Hebert JR, Daguise VG, Hurley DM, et al. Mapping cancer mortality-to-incidence ratios to illustrate racial and sex disparities in a high-risk population. Cancer. 2009; 115: 2539-2552.
  • 5
    Singh S, Bayakly AR, McNamara C, Redding K, Thompson SK, Wall K. Georgia Cancer Data Report, 2005. Publication no. DPH06/007HW. Atlanta, GA: Georgia Department of Human Resources, Divisions of Public Health, Chronic Disease, Injury, and Environmental Epidemiology Section, and the American Cancer Society, Southeast Division; 2006.
  • 6
    Vostakolaei FA, Karim-Kos HE, Janssen-Heijnen MLG, Visser O, Verbeek AL, Kiemeney LALM. The validity of the mortality to incidence ratio as a proxy for site-specific cancer survival. Eur J Public Health. 2011; 21: 573-577.
  • 7
    Parkin DM, Bray F. Evaluation of data quality in the cancer registry: principles and methods. Part II. Completeness. Eur J Cancer. 2009; 45: 756-764.
  • 8
    Gajalakshmi CK, Shanta V, Rama R. Registration of cancer mortality data in a developing area: Chennai (Madras, India) experience. Cancer Causes Control. 1998; 9: 131-136.
  • 9
    Chen WQ, Zhang SW, Zou XN, Zhao P. Cancer incidence and mortality in China, 2006. Chinese J Cancer Res. 2006; 23: 3-9.
  • 10
    Taylor RJ, Morrell SL, Mamoon HA, Wain GV. Effects of screening on cervical cancer incidence and mortality in New South Wales implied by influences of period of diagnosis and birth cohort. J Epidemiol Commun Health. 2001; 55: 782-788.
  • 11
    Ferlay J, Autier P, Boniol M, Heanue M, Colombet M, Boyle P. Estimates of the cancer incidence and mortality in Europe in 2006. Ann Oncol. 2007; 18: 581-592.
  • 12
    Woo PP, Thach TQ, Choy ST, McGhee SM, Leung GM. Modelling the impact of population-based cytologic screening on cervical cancer incidence and mortality in Hong Kong: an age-period-cohort approach. Br J Cancer. 2005; 93: 1077-1083.
  • 13
    Pisani P, Parkin DM, Ferlay J. Estimates of the worldwide mortality from eighteen major cancers in 1985. Implications for prevention and projections of future burden. Int J Cancer. 1993; 55: 891-903.
  • 14
    Hsing AW, Tsao L, Devesa SS. International trends and patterns of prostate cancer incidence and mortality. Int J Cancer. 2000; 85: 60-67.
  • 15
    Sugerman PB, Savage NW. Oral cancer in Australia: 1983-1996. Aust Dent J. 2002; 47: 45-56.
  • 16
    Kamangar F, Dores GM, Anderson WF. Patterns of cancer incidence, mortality, and prevalence across 5 continents: defining priorities to reduce cancer disparities in different geographic regions of the world. J Clin Oncol. 2006; 24: 2137-2150.
  • 17
    Copeland G, Lake A, Firth R, et al. eds. Cancer in North America: 2003-2007. Springfield, IL: North American Association of Central Cancer Registries, Inc.; 2011.
  • 18
    University of Wisconsin Population Health Institute. Georgia County Health Rankings. Available at: http://www.countyhealthrankings.org/georgia. Accessed July 20, 2011.
  • 19
    Rothman KJ, Greenland S. Modern Epidemiology. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 1998.
  • 20
    Fay MP. Approximate confidence intervals for rate ratios from directly standardized rates with sparse data. Communication Stat Theory Methods. 1999; 28: 2141-2160.
  • 21
    Peppard PE, Kindig D, Jovaag A, Dranger E, Remington PL. An initial attempt at ranking population health outcomes and determinants. WMJ. 2004; 103: 52-56.
  • 22
    Kindig DA, Booske BC, Remington PL. Mobilizing Action Toward Community Health (MATCH): metrics, incentives, and partnerships for population health [serial online]. Prev Chronic Dis. 2010; 7: A68.
  • 23
    Rohan AM, Booske BC, Remington PL. Using the Wisconsin County Health Rankings to catalyze community health improvement. J Public Health Manage Pract. 2009; 15: 24-32.
  • 24
    Peppard PE, Kindig DA, Dranger E, Jovaag A, Remington PL. Ranking community health status to stimulate discussion of local public health issues: the Wisconsin County Health Rankings. Am J Public Health. 2008; 98: 209-212.
  • 25
    Brandt HM, Glover S. Introduction to the special issue cervical cancer in South Carolina: problems and solutions through research, practice, and innovation. J S C Med Assoc. 2009; 105: 263-263.
  • 26
    Savoy J, Hurley DM, Brandt HM, Bolick-Aldrich S, Ehlers M. Cervical cancer in South Carolina: epidemiologic profile. J S C Med Assoc. 2009; 105: 227-230.
  • 27
    Berzen AK, McNamara C, Bayakly AR, Redding KC, Slade J. Cervical Cancer in Georgia, 2003-2007. Atlanta, GA: Georgia Department of Community Health, Division of Public Health, Chronic Disease, Healthy Behaviors, and Injury Epidemiology Section; 2010.
  • 28
    Wagner SE, Drake BF, Elder K, Hebert J. Social and clinical predictors of prostate cancer treatment decisions among men in South Carolina. Cancer Causes Control. 2011; 22: 1597-1606.
  • 29
    Chen JG, Zhu J, Parkin DM, et al. Trends in the incidence of cancer in Qidong, China, 1978-2002. Int J Cancer. 2006; 119: 1447-1454.