Cancer statistics, 2016

Incidence and Mortality Data

Mortality data from 1930 to 2012 were provided by the National Center for Health Statistics (NCHS).1, 2 Forty-seven states and the District of Columbia met data quality requirements for reporting to the national vital statistics system in 1930. Texas, Alaska, and Hawaii began reporting mortality data in 1933, 1959, and 1960, respectively. The methods for abstraction and age adjustment of mortality data are described elsewhere.2, 3

Population-based cancer incidence data in the United States have been collected by the National Cancer Institute's (NCI's) Surveillance, Epidemiology, and End Results (SEER) Program since 1973 and by the Centers for Disease Control and Prevention's National Program of Cancer Registries (NPCR) since 1995. The SEER program is the only source for long-term population-based incidence data. Long-term incidence and survival trends (1975-2012) were based on data from the 9 oldest SEER areas (Connecticut, Hawaii, Iowa, New Mexico, Utah, and the metropolitan areas of Atlanta, Detroit, San Francisco-Oakland, and Seattle-Puget Sound), representing approximately 9% of the US population.4 As of 1992, SEER data have been available for 4 additional SEER registries (Alaska Natives, Los Angeles county, San Jose-Monterey, and rural Georgia) that increase coverage of minority groups, allowing for stratification by race and ethnicity. Delay-adjusted data from these (SEER 13) registries, which represent 14% of the US population, were the source for the annual percent change in incidence from 1992 to 2012.5 The SEER program added 5 additional catchment areas beginning with cases diagnosed in 2000 (greater California, greater Georgia, Kentucky, Louisiana, and New Jersey), achieving 28% population coverage. Data from all 18 SEER areas were the source for cancer stage distribution, stage-specific survival, and the lifetime probability of developing cancer.6 The probability of developing cancer was calculated using NCI's DevCan software (version 6.7.3).7 Much of the statistical information presented herein was adapted from data previously published in the SEER Cancer Statistics Review 1975-2012.8

The North American Association of Central Cancer Registries (NAACCR) compiles and reports incidence data from 1995 onward for cancer registries that participate in the SEER program and/or the NPCR. These data approach 100% coverage of the US population in the most recent time period and were the source for the projected new cancer cases in 2016 and incidence rates by state and race/ethnicity.9, 10 Some of the data presented herein were previously published in volumes 1 and 2 of Cancer in North America: 2008-2012.11, 12

All cancer cases were classified according to the International Classification of Diseases for Oncology except childhood and adolescent cancers, which were classified according to the International Classification of Childhood Cancer (ICCC).13 Causes of death were classified according to the International Classification of Diseases.14 All incidence and death rates were age-standardized to the 2000 US standard population and expressed per 100,000 population, as calculated by NCI's SEER*Stat software (version 8.2.1).15 The annual percent change in rates was quantified using NCI's Joinpoint Regression Program (version 4.2.0.2).16

Whenever possible, cancer incidence rates presented in this report were adjusted for delays in reporting, which occur because of a lag in case capture or data corrections. Delay adjustment has the largest effect on the most recent years of data for cancers that are frequently diagnosed in outpatient settings (eg, melanoma, leukemia, and prostate cancer) and provides a more accurate portrayal of the cancer burden in the most recent time period.17 For example, the leukemia incidence rate for 2012 is 16% higher after adjusting for reporting delays.6, 18

Projected Cancer Cases and Deaths in 2016

The most recent year for which incidence and mortality data are available lags 2 to 4 years behind the current year due to the time required for data collection, compilation, quality control, and dissemination. Therefore, we projected the numbers of new cancer cases and deaths in the United States in 2016 to provide an estimate of the contemporary cancer burden. The number of invasive cancer cases was estimated using a 3-step spatio-temporal model based on high-quality incidence data from 49 states and the District of Columbia representing approximately 94% population coverage (data were lacking for all years for Minnesota and for some years for other states). First, complete incidence counts were estimated for each county from 1998 through 2012 using geographic variations in sociodemographic and lifestyle factors, medical settings, and cancer screening behaviors as predictors of incidence.19 Then these counts were adjusted for delays in cancer reporting and aggregated to obtain national- and state-level estimates. Finally, a temporal projection method (the vector autoregressive model) was applied to the last 15 years of data to estimate counts for 2016. This method cannot estimate numbers of basal cell or squamous cell skin cancers because data on the occurrence of these cancers are not required to be reported to cancer registries. For complete details of the case projection methodology, please refer to Zhu et al.20

New cases of female breast carcinoma in situ and melanoma in situ diagnosed in 2016 were calculated by first approximating the number of cases occurring annually from 2003 through 2012 based on age-specific NAACCR incidence rates (data from 44 states and the District of Columbia with high-quality data every year) and US population estimates provided in SEER*Stat. The average annual percent change in case counts from 2003 through 2012 generated by the joinpoint regression model was then used to project cases to 2016. In contrast to previous years, the estimate for breast carcinoma in situ was not adjusted for reporting delays because delay-adjustment factors were not available.

The number of cancer deaths expected to occur in 2016 was estimated based on the annual percent change in reported numbers of cancer deaths from 1998 through 2012 at the state and national levels as reported to the NCHS. For the complete details of this methodology, please refer to Chen et al.21

Other Statistics

The number of cancer deaths averted in men and women due to the reduction in overall cancer death rates was estimated by subtracting the number of recorded deaths from the number that would have been expected if cancer death rates had remained at their peak. The expected number of deaths was estimated by applying the 5-year age-specific cancer death rates in the peak year for age-standardized cancer death rates (1990 in men and 1991 in women) to the corresponding age-specific populations in subsequent years through 2012. The difference between the number of expected and recorded cancer deaths in each age group and calendar year was then summed.

Expected Numbers of New Cancer Cases

Table 1 presents the estimated numbers of new cases of invasive cancer expected in the United States in 2016 by sex. The overall estimate of 1,685,210 cases is the equivalent of more than 4,600 new cancer diagnoses each day. In addition, about 61,000 cases of female breast carcinoma in situ and 68,480 cases of melanoma in situ are expected to be diagnosed in 2016. The estimated numbers of new cases by state for selected cancer sites are shown in Table 2.

Figure 1 indicates the most common cancers expected to occur in men and women in 2016. Prostate, lung and bronchus, and colorectal cancers account for 44% of all cases in men, with prostate cancer alone accounting for 1 in 5 new diagnoses. For women, the 3 most commonly diagnosed cancers are breast, lung and bronchus, and colorectum, representing one-half of all cases; breast cancer alone is expected to account for 29% all new cancer diagnoses in women.

Expected Numbers of Cancer Deaths

Table 1 also shows the expected numbers of cancer deaths in 2016. It is estimated that 595,690 Americans will die from cancer this year, corresponding to about 1,600 deaths per day. The most common causes of cancer death are cancers of the lung and bronchus, prostate, and colorectum in men and lung and bronchus, breast, and colorectum in women. These 4 cancers account for 46% of all cancer deaths (Fig. 1), with more than one-quarter (27%) due to lung cancer. Table 3 provides the estimated numbers of cancer deaths in 2016 by state for selected cancer sites.

Lifetime Probability of Developing Cancer

The lifetime probability of being diagnosed with an invasive cancer is higher for men (42%) than for women (38%) (Table 4). Reasons for increased susceptibility in men are not well understood, but to some extent reflect differences in environmental exposures, endogenous hormones, and probably complex interactions between these influences. Adult height, which is determined by genetics and childhood nutrition, is positively associated with cancer incidence and death in both men and women,22 and has been estimated to account for one-third of the sex differences in cancer risk.23 For adults aged younger than 50 years, however, cancer risk is higher for women (5.4%) than for men (3.4%) because of the relatively high burden of breast, genital, and thyroid cancers in young women. The estimated probability of developing cancer is based on the average experience of the general population and may over- or underestimate individual risk because of differences in exposure (eg, smoking history), medical history, and/or genetic susceptibility.

Trends in Cancer Incidence

Figures 2 and 3 illustrate long-term trends in cancer incidence rates for all cancers combined and for selected cancer sites by sex. Cancer incidence patterns in the United States reflect trends in behaviors associated with cancer risk, improvements in cancer prevention and control, and changes in medical practice. Trends in overall incidence are driven by the 4 major cancers (lung, breast, prostate, and colorectal). For example, the spike in incidence rates in men in the late 1980s and early 1990s (Fig. 2) is due to the surge in detection of asymptomatic prostate cancer (Fig. 3) as a result of widespread prostate-specific antigen (PSA) testing.24 The increase in incidence in women during the 1980s reflects increases in both lung cancer, as a result of the tobacco epidemic, and breast cancer, because of changes in female reproductive patterns and the increased detection of asymptomatic disease during the rapid uptake of mammography screening.25

Table 5 presents the annual percent change in delay-adjusted incidence rates in the SEER 13 registries from 1992 through 2012 along with the average annual percent change over the past 5 and 10 years of data based on joinpoint regression analysis. Joinpoint is a tool used to describe and quantify trends by fitting observed rates to lines connected at “joinpoints” where trends change in direction or magnitude.8, 26 The overall incidence rate in women has remained stable since 1998, but has declined in men by 3.1% per year since 2009. The recent rapid decline in prostate cancer diagnoses accounts for about one-half of the total decline in men. Routine screening with the PSA test is no longer recommended because of growing concerns about high rates of overdiagnosis, estimated at 23% to 42% for screen-detected cancers.27, 28 Despite declines in the prevalence of PSA screening from 2010 to 2013, one-third of men aged 65 years and older with limited life expectancy were screened in 2013.29

Lung cancer incidence rates began declining in the mid-1980s in men and in the mid-2000s in women (Fig. 3) as a result of reductions in smoking prevalence that began decades earlier.30 Contemporary differences in lung cancer incidence patterns between men and women reflect historical differences in tobacco use. Women took up smoking in large numbers later than men, first initiated smoking at older ages, and were slower to quit, including recent upturns in smoking prevalence in some birth cohorts.31, 32 Declines in lung cancer incidence and death rates continue to be larger in men than in women (Table 5).

The long-term declines in colorectal cancer incidence rates since the mid-1980s have been attributed to both changes in risk factors and the introduction of screening.33 However, the recent rapid declines are likely driven by the increased uptake of screening with colonoscopy, which can prevent cancer by allowing for the removal of precancerous lesions.34, 35 Among adults aged 50 to 75 years, colonoscopy use increased from 19% in 2000 to 55% in 2013.36 Colorectal cancer incidence and death rates declined by about 3% per year in both men and women from 2003 through 2012, with momentum gaining in the most recent years (Table 5). However, rates increased by 1.8% per year from 1992 through 2012 in men and women aged younger than 50 years, among whom screening is not recommended for those at average risk.5

In contrast to stable or declining trends for most cancers, incidence rates increased from 2003 to 2012 (SEER registries combined) among both men and women for some leukemia subtypes and for cancers of the tongue, tonsil, small intestine, liver, pancreas, kidney, renal pelvis, and thyroid.8 In addition, incidence rates increased in men for melanoma; myeloma; and cancers of the breast, testis, and oropharynx. Recent declines in incidence for melanoma and liver cancer among young adults may portend a reduction in the burden of these cancers in future generations (Fig. 4). Among women, incidence rates increased for cancers of the anus, vulva, and uterine corpus. Uterine corpus cancer incidence rates have been increasing since the early 1990s in black women, but only since the mid-2000s in white women, perhaps due to steeper temporal increases in obesity among black women.37 Excess weight increases endometrial cancer risk by 50% for every 5 body mass index (BMI) units; although the highest risk occurs among the most obese women, some elevation in risk is apparent even within the normal BMI range.38 It is important to note that rates of uterine corpus cancer typically do not account for hysterectomy prevalence, thus substantially underestimating true disease risk. A recent study found that incidence rates corrected for hysterectomy prevalence were 73% higher among white women and 90% higher among black women compared with uncorrected rates.39

Thyroid cancer continues to be the most rapidly increasing cancer (>5% per year in both men and women), partially due to overdiagnosis because of increased use of advanced imaging techniques. A recent study estimated that over the past 2 decades, about one-half of all papillary thyroid cancers diagnosed in women, and 40% of those in men aged ≥50 years, were clinically irrelevant.40 However, increases across tumor size and stage, as well as for follicular carcinoma (a more aggressive subtype), suggest that some of the rise may be due to changes in environmental risk factors, such as obesity.41-43

Trends in Cancer Survival

Over the past 3 decades, the 5-year relative survival rate for all cancers combined has increased 20 percentage points among whites and 23 percentage points among blacks (Table 6). Progress has been most rapid for hematopoietic and lymphoid malignancies due to improvements in treatment protocols, including the discovery of targeted therapies. For example, the 5-year survival for acute lymphocytic leukemia increased from 41% during the mid-1970s to 70% during 2005 to 2011. The use of BCR-ABL tyrosine kinase inhibitors (eg, imatinib) doubled survival for patients with chronic myeloid leukemia in less than 2 decades,44 from 31% in the early 1990s to 63% during 2005 to 2011. A recent study found that improvements in survival since 1990 for the most common cancers have been much more pronounced among patients aged 50 to 64 years than among those aged older than 65 years.45 This disparity may reflect differential care and/or lower efficacy or use of new therapies in the elderly population.

In contrast to the steady increase in survival for most cancers, advances have been slow for lung and pancreatic cancers, for which the 5-year relative survival is currently 18% and 8%, respectively (Table 6). These low rates are partly because more than one-half of cases are diagnosed at a distant stage (Fig. 5), for which 5-year survival is 4% and 2%, respectively (Fig. 6). There is promise for improving lung cancer survival rates because of earlier detection through screening with spiral computed tomography.46 However, it is important to realize that screening, as well as other changes in detection practices, introduces lead time bias in survival rates, thereby reducing their usefulness in measuring progress against cancer.47 For example, the jump in 5-year relative survival rates for prostate cancer from 83% in the late 1980s to 93% in the early 1990s to 99% since 2000 predominantly reflects lead time and overdetection. Thus, advances against cancer are best measured using age-standardized death rates.48

Trends in Cancer Mortality

The overall cancer death rate rose during most of the 20th century, largely driven by rapid increases in lung cancer deaths among men as a consequence of the tobacco epidemic. Steady reductions in smoking, as well as advances in cancer prevention, early detection, and treatment, have resulted in a 23% drop in the cancer death rate, from a peak of 215.1 (per 100,000 population) in 1991 to 166.4 in 2012. The decline, which is larger in men (28% since 1990) than in women (19% since 1991), translates into the avoidance of approximately 1,711,300 cancer deaths (1,199,200 in men and 512,100 in women) that would have occurred if peak rates had persisted (Fig. 7).

Figure 8 depicts trends in cancer death rates since 1930 among men and women overall and for selected cancer sites by sex. In contrast to male cancer death rates, which rose continuously prior to 1990, female cancer death rates fell from the late 1940s to the mid-1970s (Fig. 8A). It is interesting to note that prior to 1941, death rates were higher in women than in men due to the high death rate for uterine cancer (uterine corpus and uterine cervix combined), which was the leading cause of cancer death among women in the early 20th century. Uterine cancer death rates declined by more than 80% between 1930 and 2012 (Fig. 8C), largely due to the widespread uptake of the Papanicolaou test for the prevention and early detection of cervical cancer. However, in contrast to continuing declines for cancers of the uterine cervix, death rates for uterine corpus cancer began to increase around 2000 in the wake of rising incidence rates (Fig. 8C, inset) (Table 5).8

A similarly dramatic decline occurred for stomach cancer, which accounted for 30% and 20% of male and female cancer deaths, respectively, in the 1930s, but only about 2% for each in 2012. Although reasons for the decrease in stomach cancer occurrence in the United States and most other parts of the world are complex and not completely understood, contributors are thought to include a lower prevalence of Helicobacter pylori infection because of improved hygiene and lower salt intake and a higher consumption of fresh fruits and vegetables because of advances in food preservation techniques (eg, refrigeration).49 However, studies indicate that incidence rates for certain subtypes of stomach cancer are increasing for some subsets of the US population for reasons that remain unknown.50, 51

The decline in cancer death rates over the past 2 decades is driven by continued decreases in death rates for the 4 major cancer sites (lung, breast, prostate, and colorectum). Death rates for female breast cancer are down 36% from peak rates, and those for prostate and colorectal cancers are each down about 50% as a result of improvements in early detection and treatment.8, 33, 52, 53 Lung cancer death rates declined 38% between 1990 and 2012 among males and 13% between 2002 and 2012 among females8 due to reduced tobacco use as a result of increased awareness of the health hazards of smoking and the implementation of comprehensive tobacco control.54 Researchers recently estimated that tobacco control efforts adopted in the wake of the first Surgeon General's report on smoking and health in 1964 have resulted in 8 million fewer premature smoking-related deaths, one-third of which are due to cancer.55, 56 Despite this progress, 80% of deaths from lung cancer and one-half of all deaths from cancers of the oral cavity, esophagus, and urinary bladder are caused by smoking.57

In contrast to declining trends for the major cancers, joinpoint analysis indicates that from 2003 to 2012, death rates rose in both sexes for cancers of the anus, liver, and pancreas.8 Death rates also increased in men for melanoma (slightly) and for cancers of the tonsil, oropharynx, and soft tissue (including the heart) and in women for uterine and vulvar cancers. Thyroid cancer death rates also increased slightly in men, from 0.43 (per 100,000 population) in 2003 to 0.51 in 2012.

Recorded Number of Deaths in 2012

A total of 2,543,279 deaths were recorded in the United States in 2012, of which 582,623 (23%) were from cancer. Overall, cancer is the second leading cause of death following heart disease, which accounted for 24% of total deaths. However, cancer is the leading cause of death among adults aged 40 to 79 years (Table 7). It is also the leading cause of death in 21 states (Alaska, Arizona, Colorado, Delaware, Florida, Georgia, Idaho, Kansas, Maine, Massachusetts, Minnesota, Montana, Nebraska, New Hampshire, New Mexico, North Carolina, Oregon, South Carolina, Vermont, Virginia, and Washington), primarily due to exceptional gains made in the progress against heart disease. In Minnesota, for example, the death rate for heart disease is 30% below the national average (118 vs 170 per 100,000 population) compared with a 6% lower death rate for cancer. In addition, cancer is the leading cause of death among both Hispanics and Asian/Pacific Islanders (APIs), who combined comprise one-quarter of the US population.58

Table 8 presents the number of deaths from all cancers combined and from the 5 most common sites for each 20-year age group by sex. More cancer deaths occur in men than in women except for those aged 20 to 39 years and 80 years or older. Breast cancer is the leading cause of cancer death in women aged 20 to 59 years, but is replaced by lung cancer in women aged 60 years or older. Among men, lung cancer is the leading cause of cancer death for those aged 40 years or older.

Among children and adolescents (aged birth to 19 years), brain cancer has surpassed leukemia as the leading cause of cancer death (Table 8). Although treatment options have improved for both cancers based on collaborative efforts and outcomes from randomized clinical trials,59 therapeutic advances for leukemia have been particularly dramatic. From 1970 to 2012, the death rate in this age group declined by 76% for leukemia compared with 31% for tumors of the brain and other nervous system (Fig. 9A). This progress has occurred despite a slow increase in the incidence of both cancer types over this time period (in the 9 oldest SEER areas) (Fig. 9C). Since the mid-1970s, the 5-year relative survival rate has increased from 50% to 87% for leukemia and from 57% to 74% for brain and other nervous system tumors.8

Regional Variations in Cancer Rates

Tables 10 and 11 depict current cancer incidence and death rates for selected cancers by state. Geographic patterns in cancer occurrence reflect differences in risk factors, such as smoking and obesity, as well as disparities in the national distribution of poverty and access to health care, which have increased over time.72, 73 The largest geographic variation in cancer occurrence by far is for lung cancer, reflecting the large historical and continuing differences in smoking prevalence among states.54 For example, lung cancer incidence rates in Kentucky, which has historically had the highest smoking prevalence, are 3.5 times higher than those in Utah, which has the lowest smoking prevalence. There is a 2-fold difference for prostate cancer incidence rates, which range from 89.8 (per 100,000 population) in Arizona to 184.1 in the District of Columbia, likely reflecting both state differences in PSA testing prevalence and population demographics.24 In contrast, state variations for other cancer types are smaller in both absolute and relative terms. For example, breast cancer incidence rates range from 107.7 (per 100,000 population) in Arkansas to 141.7 in the District of Columbia, a relative difference of just 24%. Some of this variation is attributable to differences in mammography prevalence.74