This project utilizes publicly available data to visualize temporal trends of smoking rates, lung/bronchus cancer incidence and cigarette tax across 6 geographic regions within the US. Smoking is associated with negative health outcomes such as lung cancer and excise tax on cigarettes is a public health approach to lower smoking rates within the US.1 ,2 Immediately below, you will view a series of questions with each followed by several data visualizations. Details on data and methodology are located at the end of this blog post. All data loading, manipulation and visualization is performed using the R statistical software and the code is located at the end of this blog.
Cigarette Tax = Smoking Rates?
Figure 1: Cigarette Tax ($ Per Pack) Over Time
Figure 1 shows that cigarette taxes increase over the time period 1999 - 2012 across all regions within the US. You will notice that the pacific and northeast regions have the highest state level excise tax on cigarettes whereas the NC-KY-GA and south regions have the lowest amount of state level excise tax across this time period.
Figure 2: US Region Current Smoking % Over Time
Figure 2 shows that smoking rates across all regions within the US decrease over the 199 - 2010 time period. The NC-KY-GA and south regions have the highest current smoking rates whereas the northeast and west have the lowest current smoking rates across this time period.
Figure 3: Current Smoking (%) & Cigarette Tax in the US
Figure 3 incorporates 3 dimensions in order to visualize the relationship among current smoking percentage and cigarette tax across time. You will notice that regions (such as the northeast, west and pacific) with higher cigarette taxes tend to have lower current smoking percentages across time compared to regions with lower cigarette taxes such as NC-KY-GA and the south.
Smoking Rates = Lung/Bronchus cancer rates?
Figure 4: Regional Lung/Bronchus Cancer Incidence Over Time
Figure 4 identifies lung and bronchus cancer incidence for each of the 6 US regions across the 1999 - 2010 time period. You will notice that the NC-KY-GA and south regions have higher lung/bronchus cancer incidence rates compared to the west, pacific and northeast regions.
Figure 5: Current Smoker % and Cancer Incidence Over Time
Figure 5 incorporates the relationship between current smoking percentage and lung/bronchus cancer incidence for each of the six regions across the 1999 - 2010 time period. You will notice that regions with higher current smoker percentages such as the NC-KY-GA and south regions have higher lung/bronchus cancer incidence compared to regions with lower current smokier percentages such as the west and pacific. The northeast seems to contradict this trend by having low current smoker percentages but with a moderate lung/bronchus cancer incidence rate.
Which regions have the highest per capita average health expenditures attributable to smoking?
Figure 6: Health Expenditure Attributable to Smoking
Figure 6 identifies the per capita health expenditures attributable to smoking for each of the 6 regions. You will notice that the northeast has the highest per capita health expenditures attributable to smoking whereas the NC-KY-GA region has the lowest per capita health expenditures attributable to smoking.
Data & Methodology
Data Source: Center for Disease Control and Prevention
Behavioral Risk Factor Surveillance System
State Tobacco Activities Tracking and Evaluation System
Smoking-Attributable Mortality, Morbidity, and Economic Costs
enigma.io/publicdata/
Metrics
Smoking Rates– “Current Smokers”
Cancer – “Lung and Bronchus Cancer Incidence”
Smoking Attributable Economic Costs – “Excess health care expenditures attributable to smoking”
Midwest:ND, SD, NE, KS, MN, IA, IO, MO, WI, IL, IN, OH, MI
South: TX, OK, AR, LA, MS, AL, GA, FL, SC, NC, TN, KY, WV, VA, DC, MD, DE
Pacific: HI, AK
West: MT, WY, CO, NM, AZ, UT, ID, OR, WA, NV, CA
NC-KY-GA (High Tobacco production states)
Conclusion
Smoking rates follow a downward trend over time.
Lung and Bronchus Cancer rates follow a similar downward trend.
Cigarette Tax follows an upward trend across time.
Regions with higher smoking rates tend to also have higher lung/bronchus cancer incidence.
Regions with higher amounts of excise tax tend to have lower smoking percentages and lower lung/bronchus cancer incidence.
Next Steps
Apply rigorous statistical methods in order to identify whether any of these initial analyses are statistically significant
Incorporate additional data on other public health interventions related to state level tobacco control programs
References
S. Department of Health and Human Services.The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2014 [accessed 2015 Oct 5].
[sourcecode language="r"]
##### ##### ##### ##### ##### ##### #####
##### GRAPHS CREATED BELOW ##### #####
##### ##### ##### ##### ##### ##### #####
## Smoking All States
theme_set(theme_tufte(base_size = 20))
smoking_plot_st<- as.data.frame(filter(analytic_dataset_state, (1998< year & 2011 >year))%>% group_by(region))
smoking_plot_st<-ggplot(smoking_plot_st, aes(x=year, y=smoking_rate, color=region, size=2)) +
geom_point(position="jitter") +
ylab("Current Smoking (%)") + xlab("Year")+
labs(title="State Smoking Rates Over Time")
#1
smoking_plot_st
smoking_plot_region<- as.data.frame(filter(analytic_dataset_state, (1998< year & 2014 >year))%>% group_by(region))
smoking_plot_region<-ggplot(smoking_plot_region, aes(x=year, y=smoking_rate, color=region, group=region)) +
geom_smooth(linetype=1, size=2, se=FALSE) +
ylab("Current Smoker (%)") + xlab("Year")+
ggtitle("US Region Current Smoking % Over Time")
#2
smoking_plot_region
## (Taxes v Time)
tax_plot_region<- filter(analytic_dataset_state, (1998< year & 2015 >year))%>% group_by(region)
tax_plot_region$ProvisionValue<- as.numeric(tax_plot_region$ProvisionValue)
tax_plot_region<-ggplot(tax_plot_region, aes(x=year, y=ProvisionValue, group=region, color=region)) +
geom_smooth(linetype=1, size=2, se=FALSE) +
scale_y_continuous(breaks=seq(0, 4, 0.5)) +
ylab("Cigarette Tax ($ Per Pack)") + xlab("Year")+
ggtitle("Cigarette Tax ($ Per Pack) Over Time")
#3
tax_plot_region
## Smoking V Tax for one year - ?
##Main Visualization 1 (Smoking Rates v Taxes)
smoking_tax_plot<- filter(analytic_dataset, (1998< year& 2011 >year), region!="NULL") %>%
ggplot(data = ., aes(year, mean_smoking_rate, color=region)) +
geom_point (aes(size=mean_provision_value)) + scale_size_continuous(name="Cigarette Tax ($ per pack)", range = c(3,15)) +
ylab("Current Smoking (%)") +
xlab("Year")+
ggtitle("Current Smoking (%) & Cigarette Tax in the US") +
theme(legend.title=element_text(size=15)) +
guides(shape=guide_legend(override.aes=list(size=50)))
#4
smoking_tax_plot
## Cancer v Time
cancer_plot_region<- as.data.frame(filter(analytic_dataset_state, (1998< year & 2011 >year))%>% group_by(region))
cancer_plot_region<-ggplot(cancer_plot_region, aes(x=year, y=age_adjusted_cancer_incidence_rate, group=region, color=region)) +
geom_smooth(linetype=1, size=2, se=FALSE) +
ylab("Cancer Incidence per 100,000") + xlab("Year")+
ggtitle("Regional Lung/Bronchus Cancer Incidence Over Time")
#5
cancer_plot_region
##Main Visualization # 2 SMOKING AND CANCER RATE TRENDS
rates_time_plot<- filter(analytic_dataset, (1998< year& 2011 >year), region!="NULL") %>%
ggplot(data = ., aes(year, mean_smoking_rate, color=region)) +
geom_point (aes(size=mean_incidence_rate)) + scale_size_continuous(name="Lung/Bronchus Cancer Incidence (per 100,000)", range = c(3,15)) +
ylab("Current Smoker %") + xlab("Year")+
ggtitle("Current Smoker % and Cancer Incidence in the US")
#6
rates_time_plot
## Expenditure v time - Boxplots?
expenditure_region<- as.data.frame(filter(analytic_dataset_state, (2008< year & 2010 >year))%>% group_by(region))
expenditure_region<-ggplot(expenditure_region, aes(x=region, y=per_capita_expenditure, color=region, group=region)) +
geom_boxplot() +
ylab("Per Capita Cost ($)") + xlab("Region")+
ggtitle("Health Expenditure Attributable to Smoking")
#7
expenditure_region
[/sourcecode]
About Author
Brian Saindon
As a Health Data Scientist, Brian Saindon (MPH) leverages innovative data science tools to identify underlying patterns within healthcare systems. As a Health Data Analyst for Predilytics, he applied advanced statistics to predict disease likelihood, member disenrollment, member...