Does Survival Vary for Breast Cancer Patients in the United States? A Study from Six Randomly Selected States

Background Breast cancer is the most common cancer in women. Disparities in some characteristics of breast cancer patients and their survival data for six randomly selected states in the US were examined. Materials and Methods A probability random sampling method was used to select the records of 2,000 patients from each of six randomly selected states. Demographic and disease characteristics were extracted from the Surveillance Epidemiology and End Results (SEER) database. To evaluate relationships between variables, we employed a Cox Proportional Regression to compare survival times in the different states. Results Iowa had the highest mean age of diagnosis at 64.14 years (SE = 0.324) and Georgia had the lowest at 57.97 years (SE = 0.313). New Mexico had the longest mean survival time of 189.09 months (SE = 20.414) and Hawaii the shortest at 119.01 (SE = 5.394) months, a 70.08-month difference (5.84 years). Analysis of stage of diagnosis showed that the highest survival times for Whites and American Indians/Alaska Natives were for stage I cancers. The highest survival times for Blacks varied. Stage IV cancer consistently showed the lowest survival times. Conclusions Differences in breast cancer characteristics across states highlight the need to understand differences between the states that result in variances in breast cancer survival.


Introduction
Breast cancer is a worldwide public health concern within many countries, including the United States (US) which has experienced a recent 20% increase in breast cancer diagnosis. Breast cancer is the most prevalent type of cancer (29%) in women in the US. Additionally, breast cancer is the most commonly diagnosed malignancy among women in the US, accounting for nearly one out of every three diagnosed malignancies [1,2]. Between 1975 and 1995, breast cancer incidence in women increased from 81 out of 100,000 personyears to 97 out of 100,000 person-years (Althuis, Dozier, Anderson, Devesa, and Brinton, 2005). Secondly, the US had a 29% rate of breast cancer in women, with women over the age of 50 years having the highest incidence ( [3]; Kohler et al., 2015). Furthermore, it is estimated that one in eight women living in the US will develop breast cancer in their lifetime [1]. In 2013, there were 232,340 cases of invasive breast cancer in women in the US and 39,620 associated deaths [1]. In the US, over three million women are currently living with a history of invasive breast cancer, with 40% of cases occurring in women over 65 years of age and 20% among women younger than 50 years of age [4]. Therefore, the increase in breast cancer incidence and the cancer-related complications have remained a significant public health issue within the US over the last few decades.
Prevention and Treatment. Early detection is the primary focus of breast cancer prevention efforts in most settings. In 2015, the American Cancer Society (ACS) reported that 66% of women aged 40 years and older had a mammogram within the past two years [5], an increase from 2010 where it was estimated that at least half of all women in the US between ages 40 and 74 years had received a mammogram [6].
The National Breast and Cervical Cancer Early Detection Program (NBCCEDP) is a program developed by the Centers for Disease Control and Prevention (CDC) to increase breast and cervical cancer screening rates among economically disadvantaged women. In the years 2011 to 2012, NBCCEDP screened 549,043 underserved women between the ages of 40 and 64 years. NBCCEDP screened about 10% of eligible women with screening rates ranging from 3.2 to 52.8% of the eligible population [7]. Overall, the program screened over half a million women; however, most eligible women remained unscreened for breast cancer and may require increased outreach efforts [7]. Mammogram use was lowest among American Indians and Alaska Natives at 36% [7,8].
Research shows that 43.5% of the 137,274 eligible women had at least one mammogram screening. Additionally, women 66 to 74 years old were more likely to get mammography screening compared to those 85 to 100 years or older (57.2% versus 15.2%, resp.; < 0.001) [9]. Further research shows that 50.1% of Black women and 40.8% of White women aged 65-74 years received either no or one screening mammogram from 2005 to 2008 [10].
Health Disparities. There have been major improvements in screening and treatment; however, there remain racial and ethnic differences in breast cancer screening and mortality. As of 2014, the rates of breast cancer have increased for Black Americans and have decreased for Hispanics [1]. The public health community has focused on increasing the breast cancer screening rates of ethnic and racial minorities and has had some success increasing the rates of breast cancer screening for Black Americans and some Hispanic groups, with the exception of Mexican Americans [11]. Ethnicity and race are found to be major predictors of breast cancer prognosis and incidence; social, environmental, and hereditary determinants directly affect the development of breast cancer [3,6,8,12,13].
Black women have an increased risk for more aggressive forms of breast cancer, such as estrogen receptor negative tumors, which frequently do not respond well to current therapies [6]. Black women under the age of 50 years also experience higher rates of breast cancer compared to White women of comparable ages. Additionally, Black and White women who had one mammogram annually had a lower 10year mortality than those who received screening irregularly or biennially [10]. When diagnosed at the same stage, Black women face higher rates of mortality associated with breast cancer than White women and are more likely to be diagnosed at advanced stages of the disease [14]. Mammogram use is 33% lower among immigrants who migrated to the US within the last 10 years [8]. Many of these immigrants lack health insurance and have lower education levels and limited income, which can contribute to lower screening rates, and women who experience inconsistent screening rates have a shorter 10-year survival time after breast cancer diagnosis [8,11].
Structural, organizational, and political factors further exacerbate issues of racial and ethnic disparities in cancer mortality [14]. Issues of poverty, poor access to care, poor transportation, low or no income, and lack of health insurance increase the chances of a poor prognosis and lower screening rates [14]. Additionally, some groups are diagnosed with varying types of breast cancer. For example, estrogen receptor negative breast cancers have decreased across all ethnicities, but rates of estrogen receptor positive breast cancer have increased in young White women, older Hispanic women between the ages of 60 and 69 years, and all groups of Black Americans except the eldest groups [14,15] [16]. In this paper, we provide the observed numbers of breast cancer cases and deaths in the US from 1973 to 2011 for six states randomly selected out of nine recorded states, as well as a broad summary of breast cancer incidence and survival times.

Materials and Methods
The study used data from the Surveillance Epidemiology and End Results (SEER) database . The SEER database started collecting data in 1973 for about 10% of the US population from nine states. Currently, the SEER program collects and publishes cancer incidence and survival information from cancer registries covering 28% of the US population [17]. The SEER website includes data from twenty populationbased registries across varying states and territories. For this study, only data collected from the years 1973 to 2011 will be evaluated from each of the six states: California, Connecticut, Georgia, Hawaii, Iowa, and New Mexico. The representative probability sample data were randomly selected from the six SEER registries, and the selected data were summarized for information on stage of cancer, overall survival, and the lifetime probability of developing breast cancer.
Inclusion criteria for the present study are female gender, first and primary diagnosis of stages I, II, or III breast cancer, no previous cancer(s) being registered, and age 20 years or older. A participant's contribution to the person-years at risk began from the date of breast cancer diagnosis to the date of death or loss to follow-up, whichever occurred first. Women diagnosed with breast cancer at autopsy were excluded. Since breast cancer is uncommon in males, only female cases were included in this study. The SEER-coded categories of registry ID (REG) were used to classify participants into six mutually exclusive categories. Simple random sampling was used to select 2,000 cases from each registry. Information on the methods used for random sampling can be found in previously published literature by Khan et al. [18][19][20][21]. In addition, we used subject demographic information (age at diagnosis, marital status, race, and ethnicity) and survival time from the SEER dataset for statistical analysis. Data regarding other socioeconomic factors, such as income and health insurance status, were not available.
A total of 12,000 women with breast cancer were included in the analysis (for each state's registry = 2,000, Figure 1) and individual survival time is defined by t, where t contains 12,000 survival data points and = 2,000 for each state. Survival analysis accounted for both censored (patients who survived till the end of SEER registry's cutoff date) and uncensored (any patient who died within the SEER registry's cutoff date) data. Survival time was calculated in months using the Cox proportional hazards model, adjusting for age at diagnosis, race, ethnicity, and marital status. Cox proportional hazards models generated the adjusted hazard ratio and their 95% confidence intervals (CI). Data analysis was conducted using SPSS software (IBM SPSS for Windows version 20, 2011) and SAS5 software version 9.4.

Results and Discussion
Using a probability sampling method, which is simple random sampling method, 2,000 patients were selected from six state cancer registries (California, Connecticut, Georgia, Hawaii, Iowa, and New Mexico). Tables 1 and 2 contain the descriptive statistics. Table 1 Table 1 also indicates that Iowa has the older stage at diagnosis for breast cancer patients followed by Connecticut. Hawaii and Georgia have the lowest ages of breast cancer diagnosis. The 25th and 50th quartiles of age at diagnosis ranged from 48-53 to 57-65 years of age, respectively, for all six states.
Mean survival days (in months) were also calculated and stratified by each cancer registry. New Mexico reported the longest mean survival time of 189.09 months (SE = 20.414), and Hawaii reported the shortest mean survival time of 119.01 (SE = 5.394) months, representing a 70.08-month difference (5.84 years). The 25th and 50th quartiles of survival times generally ranged from 33-41 to 83-96 months, respectively, for all six states. Table 2 shows the frequency and percentage of each race, ethnicity, and marital status for each of the six states studied. Most participants were married with widowed being the second most common relationship status, and separated women were the smallest group. Married women made up over half of the women with a breast cancer diagnosis, ranging from 52.95% in Connecticut to 59.6% in Hawaii.
We have stratified data by race and 5-year-time intervals for age at diagnosis and obtained the following summary statistics for six states: in California during 1973-1980, 61.50 years (SD = 13.68) was the mean age of diagnosis for White patients, compared to 57.52 years (SD     Table 3 shows the analysis of Cox proportional hazard ratio and its corresponding confidence intervals. New Mexico, with the highest mean survival time, was made the referent group for calculating hazard ratios, adjusting for states using the Cox Proportional Regression. Hazard ratios compare the probability of an event occurring in one group versus another group considering the time elapsed until the event occurs. Hazard ratios were calculated for five states.   respectively. New Mexico, California, and Connecticut have the longest survival times. In Table 4, Hawaii was used for the referent group. Hawaii had a significantly increased risk of death compared to California (hazard ratio: 1.177; 95% confidence limits (1.066-1.300)); Connecticut (hazard ratio: 1.261; 95% confidence limits (1.143-1.390)); Georgia (hazard ratio: 1.257; 95% confidence limits (1.134-1.394)); Iowa (hazard ratio: 1.232; 95% confidence limits (1.119-1.356)); New Mexico (hazard ratio: 1.272; 95% confidence limits (1.151-1.406)).
This study's findings showed that New Mexico reported the longest mean survival time compared to Hawaii, which had the shortest mean survival time, demonstrating a fiveyear difference. This suggests that there are potential differences across states that affect survival time. In

Study Limitations
SEER has collected cancer data for over thirty years from cancer registries throughout the United States; it is nationally recognized and considered reliable source of information on incidence, mortality, and other related variables. Although use of SEER lends this study strength, it is limited in the information it can provide. SEER lacks insight into many variables, such as social and economic factors, that affect the survival time of breast cancer patients and may explain much of the disparities experience by more disadvantaged groups.

Conclusions
This study concludes that there are several factors accounting for breast cancer survival including state, local, and individual level factors. For example, the age of diagnosis was greatly different for those living in Iowa compared to Georgia. This presented a seven-year difference in disease presentation. It is imperative to understand what differences are found in breast cancer prevention between those two states. Additionally, there are other factors that may account for the differences in age of diagnosis. Even more concerning is the stark differences in disease prognosis. At the state and governmental level, future studies may compare state policies, preventive breast cancer systems, and the current state of the health care systems and their effect on the outcome of women diagnosed with breast cancer.
However, it may also be important for future studies to address other demographic variables including income, education level, and health insurance status. To better understand patient survival, primary factors affecting patient prognosis should be addressed. These factors can be current medical treatment and prevention regimen that affect breast cancer diagnosis and treatment, and more indirect factors will also affect survival. Additionally, further investigation should be considered at the community level to determine what specific factors prevent the receipt of preventive screening or access to adequate care. This will inform the design and troubleshooting of prevention and treatment efforts.
These study findings give credence to the importance of early detection and treatment in reducing breast cancer incidence nationwide. Given the patient's current location and reported conditions, our findings suggest that geographical and local characteristics affect the survival rates of breast cancer patients. More in-depth research can help highlight factors contributing to the disparities seen across states. State health policies, access to health care, breast cancer screening and awareness programs, and cultural norms may affect screening and preventive care, which will likely affect longterm patient survival. The mean age at diagnosis ranged from 58 to 64 years, underlining the importance for following the United States Preventive Services Task Force (USPSTF) recommendations for biennial screening mammography for women aged 50 to 74 years. Breast cancer takes approximately 2 to 5 years to develop; therefore, women should be screened regularly to identify and treat precancers as well as prevent the progression of breast cancer. The findings in this study reinforce the importance of early detection and treatment in reducing breast cancer. Increasing patient survival first begins with improving screening efforts among poor and underserved women who may be at highest risk. However, it is important to recognize that some women, such as those with a family history of the disease, are at risk and may experience earlier presentations of breast cancer and should be screened earlier than the standard guidelines suggest. Public health and health care prevention efforts should target the most disadvantaged communities to help eliminate breast cancer-related health equity. For example, Black women experience more aggressive forms of breast cancer and are often diagnosed at early ages; therefore, this group may require more attention in screening and education efforts. This study showed that stage IV cancer had the lowest survival times among all groups. The highest survival times for Whites and American Indians/Alaska Natives were for stage I cancers, and the highest survival times for Blacks varied in the 5-year stratification analysis. Furthermore, economically disadvantaged patients that lack health insurance often do not receive adequate or appropriate care for their diagnosis adversely affecting their prognosis. This may include improving the treatment options for these groups and improving screening efforts to ensure earlier disease detection. Additionally, improving outreach efforts for programs such as NBCCEDP will help to target many underserved communities. Furthermore, high survival estimates in New Mexico, California, and Connecticut indicate the need to gather information about regional and environmental characteristics affecting the survival rates of breast cancer patients. The combination of current knowledge and previous research regarding predictive modeling can be used to inform policy decisions and to plan allocation of future resources and interventions [18][19][20][21]. As it relates to cancer prevention and control there are several other factors that have been associated with breast cancer risk including lifestyle practices such as alcohol and tobacco use, obesity, and physical inactivity. Medical therapies such as high duration of hormone therapy to treat menopause have also been associated with increased breast cancer risk. Consistent biennial screening and genetic testing for those who are genetically predisposed to breast cancer have also been strongly linked to reducing breast cancer risk and should be encouraged. However, many at risk groups also face issues of alcohol and tobacco use, poor diet, obesity, and physical inactivity which increase their risk of cancer. Therefore, these lifestyle changes should be addressed in more disadvantaged groups to help prevent and control the incidence of cancer. Additionally, women in these groups often experience no or inadequate health insurance coverage that provides affordable screening options. Therefore, there is more need to provide affordable screening options in minority and socioeconomically disadvantaged communities as well as providing nutrition and fitness counseling services that encourage healthy lifestyle choices that are reasonable and practical for economically disadvantaged communities that face greater barriers and poorer breast cancer outcomes.

Conflicts of Interest
The authors have declared no conflicts of interest.