According to data from the National Institute of Health (NIH), cancer receives more research funding than any other disease, with an appropriation of over $5 billion per year. The National Cancer Institute (NCI) is the primary department that is responsible for allocating these funds to various researchers. This project is intended to be presented to an NCI committee that oversees grant reviews. In this project, we sought to analyze whether different types of cancer receive different amounts of funding, and whether these differences are due to factors such as incidence and mortality rates, years of life lost, and economic impacts. We also examined the overall cancer burden and recommended that some funding be allocated toward public health research that addresses geographical and racial disparities in cancer outcomes.
How has the overall cancer burden in the United States changed from 1999 to 2016, in terms of incidence and mortality rates?
Before investigating NCI funding data, we obtained an overview of how the cancer burden in the U.S. has changed over time, in order to gauge the progress that has been made in cancer treatment and prevention. This plot shows the age-adjusted incidence and mortality rates, as well as the mortality-incidence rate ratio (expressed as a percentage), for the leading cancers in the U.S., from 1999-2016. The mortality-incidence age-adjusted rate ratio decreased for 13 of 20 leading cancer sites, indicating that chances of survival have markedly improved. However, incidence rates only decreased in 7 of the 20 leading cancer sites. Incidence rates actually increased for several cancers, especially melanoma of the skin, whose rate increased by ~10 individuals per 100,000. Thus, while there have been significant improvements in survival rates, there is still room for progress to be made, especially by lowering incidence rates. This could potentially be achieved through public health initiatives that focus on prevention and early diagnosis. As an example, we analyzed the geographic distribution of melanoma incidence and mortality rates over time, to investigate whether this cancer might benefit from research for public health initiatives.
Have incidences of melanoma increased or decreased over time?
Yes. This animation displays melanoma incidence age-adjusted rates (per 100,000) by state over the years 1999-2016. The map is color-coded so that darker states have higher incidence rates. The rates vary from 8.337 (Nevada, 1999) to 42.693 (Utah, 2014). Several states were missing data for a given year; they have been left white in this map.
Melanoma incidence rates rose significantly over this time period. In 1999, the states reporting data averaged 15.97 incidences per 100,000 people. By 2016, that number had grown to 24.11 cases. It appears that mountainous states in the North are seeing the most cases of melanoma, with Utah, Vermont, New Hampshire, Oregon, and Washington as the top five. The public health implications of this observation are discussed below (along with plot 3).
Have melanoma mortality rates also increased over time?
No. This animation displays melanoma mortality-incidence age-adjusted rate ratios by state over the years 1999-2016. A darker state means a higher percentage of patients with melanoma died than a lighter state. The ratios vary from .052 (Vermont, 2016) to .398 (Oklahoma, 1999). Once again, states lacking data have been left white. As the animation shows, mortality-incidence ratios dropped significantly over this time period. In 1999, the M-I ratio averaged .187 across states. In 2016, that number was .098, roughly half of what it was in 1999! However, this coupled with the increase in melanoma cases means that the melanoma mortality rate has only diminished slightly over the 1999-2016 time period, from an average of 2.79 per 100,000 in 1999 to 2.28 per 100,000 in 2016.
These numbers suggest that although great strides have been made in melanoma treatment, prevention measures aren’t adequate. Although melanoma can be caused by a variety of factors (including factors out of human control like genetics), researchers have found evidence that exposure to UV rays and a history of sunburns can dramatically increase someone’s risk of melanoma. Practices like avoiding tanning beds, wearing sunscreen and protective clothing, and staying out of the sun during peak hours are all steps that can be taken to reduce the risk of melanoma. Despite the fact that researchers have discovered more about how to reduce this risk, incidence rates have continued to climb. This suggests an opportunity for public health research to come up with ways to improve the public’s use of skin protection, particularly in mountainous Northern states where the incidence of melanoma is especially high.
How do the 5-year survival rates of different cancers compare among races?
In formulating recommendations for the funding of public health research projects, we investigated whether there are any disparities in cancer outcomes. Here, we show 5-year cancer survival rates among races. For 13 of the 18 cancers for which data was available, the survival rate was higher among whites than blacks. Furthermore, the overall survival rate of all cancers combined was about 5% higher in whites and other races compared to the African American population. Disparities were especially large in breast, melanoma, and urinary bladder cancers, where survival rates in African Americans were over 10% lower than in whites. Thus, there are clear disparities in cancer outcomes among races. We recommend that some funding be allocated to research projects that investigate the biological and sociological causes of these disparities.
What is the current distribution of NCI research funding for different cancer sites?
Next, we began to analyze NCI funding patterns. This graph shows the distribution of NCI research funding for the top 21 most common cancers in the United States. Breast cancer receives the highest amount of NCI funding (over $600 million), which is more than two times the next closest site (prostate cancer). There are also five types of cancer that receive over $200 million in funding. These cancers- breast, colon/rectum, leukemia, lung, and prostate- make up 63% of all research funding for leading cancers. Thus, it is evident that there is an unequal distribution of funding for different cancers. We hypothesized that these inequities might stem from differences in incidence, mortality, years of life lost, and financial burden. In our subsequent analyses, we sought to answer the question of why the top five highest-funded cancers receive such high levels of funding.
How does the breakdown of spending compare among the five most-funded cancers? Is NCI funding too high compared to other sources of spending?
In this graph, we quantified the overall “cost” of the top five cancers by analyzing three sources of spending- NCI funding for research, Medicare Spending for senior medical care, and National Spending on medical care for non-seniors. To create the graph, Medicare Spending and NCI Funding data were converted from millions to billions of dollars spent, so that they would be comparable. Overall, NCI funding makes up less than 9% of the total “cost” of each cancer (except in leukemia, where it is ~20%). Due to the low contribution of NCI funding to the “total” cancer cost, we do not think the NCI is overfunded. Among the top five highest-funded cancers, leukemia appears to be an outlier. Despite being the second highest NCI funded cancer, it has only about 40% of the national and medicare funding compared to other highly funded cancers. This drew our attention to potential disproportionalities in funding, which we investigated further in terms of incidence, mortality, years of life lost, and economic impact.
How does the percent of NCI funding allocated to each of the top 5 cancer sites compare to their percent of mortalities and incidences?
In this graph, we explored how funding among the top 5 cancer sites was allocated, and whether it was correlated with incidence and mortality rates. We divided the amount of funding, mortalities, and incidences of each of the top 5 cancer cites by the total of the top 5 in each category. We expected that if NCI funding was being allocated proportionally to mortalities and incidences then all of the bars would be at approximately the same level. This was not the case, as evidenced by lung cancer, which accounts for over 52% of the total mortalities between the top 5 cancer cites, but only 16% of the funding. Additionally, leukemia accounts for 7% of the mortalities in the top 5 cancer cites but has 17% of the funding. Percent of funding is closer to the percent of incidences in four out of the five clusters, but based on the bars in this graph the funding allocated does not seem to directly correlate with the amount of incidences or mortalities.
Do differences in incidence and mortality rates account for inequities in NCI funding of leading cancers?
Although the previous graph had shown that mortality and incidence rates did not account for differences in funding within the 5 highest-funded cancers, we wondered if it might separate the top 5 cancers from the others. We divided the amount of NCI funding by the number of incidents (or mortalities), to calculate the funding per incident (or mortality). We expected that if cancers were being funded based on their proportion of total incidences or mortalities, the funding per incident or mortality would be approximately equal for all cancers. However, there are drastic differences in funding, even after normalizing for incidence and mortality rates. For example, bladder cancer receives only $320 per incident and $1,500 per mortality, while brain cancer receives almost $9,000 per incident and nearly $15,000 per mortality. This graph also identified cancers that are potentially underfunded or overfunded. The orange and blue dotted lines indicate the median NCI funding per mortality and NCI funding per incident, respectively. If resources were allocated based on mortality and incidence rates, cancers above these lines would be considered overfunded, and cancers below these lines would be considered underfunded. Out of the top five highest-funded cancers (breast, colon/rectum, leukemia, lung, and prostate), only two are overfunded with respect to the number of incidences, and three are overfunded with respect to the number of mortalities. Thus, differences in incidence and mortality do not explain why cancers receive varying levels of funding, but the top five highest-funded cancers are not severely overfunded based on these metrics. Since cancer funding does not seem to be dependent on incidences and mortalities, we hypothesized that differences in years of life lost and economic impact might play larger roles in resource allocation. We investigated this question in the next two plots.
Do cancers with more severe effects on lifespan tend to receive more NCI funding than other cancers?
Next, we wondered if the NCI tends to allocate more funding toward cancers with more severe impacts on lifespan, which was operationalized by average years of life lost (AYLL), and years of life lost per individual (YLLPI). Interestingly, the slope of the regression lines for both AYLL and YLLPI were nearly 0, indicating that there was no correlation between these metrics and NCI funding. Out of the top five highest-funded cancers (breast, colon/rectum, leukemia, lung, and prostate), only one (breast) ranked in the top five in AYLL, and none were in the top five in YLLPI. Thus, the “severity” of cancers does not have a noticeable impact on how much they are funded. A possible explanation for this is that although certain cancers might kill people at younger ages, these cancers are relatively rare, and have a lower economic burden on society. To test this theory, we investigated the correlation between NCI funding and the economic impact of different cancers.
Do cancers with greater economic impacts tend to receive more NCI funding than other cancers?
After finding no correlation between “severity” of cancers and NCI funding, we speculated that economic factors might have a greater impact on NCI funding. Lost productivity and national spending were moderately correlated with NCI funding, with R2 values of 0.54 and 0.53, respectively. When calculating the R2 value for lost productivity, we excluded lung cancer, since it was a drastic outlier with a lost productivity value that was nearly 4x greater than the second-highest cancer (colon/rectum). This is most likely due to the fact that lung cancer had the highest incidence rate, and tends to affect individuals who are still of working age. Additionally, out of the top five highest-funded cancers (breast, colon/rectum, leukemia, lung, and prostate), three were ranked in the top five in lost productivity, and four were in the top five in national spending. These results indicate that NCI research funding tends to be biased towards cancers with higher economic impacts. We agree that economic impacts should be considered when allocating resources for cancer research. However, we would recommend that the budget be re-formulated to correlate more strongly with incidence and mortality rates, to place more of an emphasis on the health effects of cancer.
The goal of this project was to investigate the overall cancer burden in the United States, and to analyze how factors such as incidence rates, mortality rates, years of life lost, and economic costs may contribute to NCI research funding decisions. First, we observed overall decreases in incidence-mortality rate ratios from 1999-2016, but a less significant decrease in incidence rates. This indicated that while treatments have likely improved, there is still progress to be made in terms of prevention, which may necessitate a need for improved public health research. To illustrate this point, we used the melanoma as an example of how public health initiatives might be targeted toward particular regions (eg: incidence of melanoma appears to be higher in Northern, mountainous states), and racial differences in 5-year survival rates as an example of how public health initiatives should be used to address socioeconomic disparities. Thus, we would advocate for additional funding of research projects relating to public health.
Next, we investigated the distribution of NCI funding for various cancers. We were surprised to find that NCI funding does not correlate strongly with mortality rates, incidence rates, or years of life lost. However, it does have a moderate association with economic metrics such as lost productivity and national spending. Our analysis indicates that the NCI does not have a systematic approach for ensuring that cancers receive funding levels that are proportionate to their societal impact. Currently, the NCI states that grants for funding are evaluated based on “scientific merit, potential impact, and likelihood of success,” but that it “does not make decisions about funding based on predetermined targets for a specific cancer type or research category.” In order to maximize the return on investment for research expenditures, we recommend that the NCI consider factors such as incidence and mortality rates, along with economic costs, when reviewing future grant applications.
BACK TO INDEX