Story-3: Obesity and Healthcare Spending

Instructions:

Assignment: Investigating the Relationship Between Healthcare Spending and Adult Obesity

Public health outcomes often reflect differences in policy priorities and resource allocation. This assignment uses state-level adult obesity prevalence data to explore whether differences in healthcare spending correlate with obesity rates.

Research Question:

Do differences in per capita healthcare spending correlate with adult obesity prevalence?

Your task is to analyze the provided obesity data and source additional data on healthcare spending to address this question.

Assignment Requirements:

Data Collection:

1. Obesity Data:

The state-level adult obesity prevalence dataset is provided in the attached file.

Obesity 2024

2. Per Capita Healthcare Spending Data:

Find and download state-level per capita healthcare spending data from a reliable source, such as:
- Kaiser Family Foundation (KFF): State Health Expenditures.
- US Census Bureau: Health spending data.
You will assign states to quintiles based on healthcare spending levels (lowest 20%, second lowest 20%, middle 20%, second highest 20%, highest 20%).

url <- "https://raw.githubusercontent.com/Aconrard/DATA608/refs/heads/main/adult_obesity.csv"
df_1 <- read.csv(url)

url2 <- "https://raw.githubusercontent.com/Aconrard/DATA608/refs/heads/main/raw_data.csv"
df_2 <- read.csv(url2)

url3 <- 'https://raw.githubusercontent.com/Aconrard/DATA608/refs/heads/main/scraped_table.csv'
df_3 <- read.csv(url3)

Data Preparation:

1. Merge the Datasets:

Combine the obesity prevalence data with the per capita healthcare spending data.
Assign each state to its respective spending quintile.

df_2$State <- rownames(df_2)
rownames(df_2) <- 1:nrow(df_2)
df_2 <- df_2[, c("State", setdiff(names(df_2), "State"))]

df_2 <- df_2[-(1:3), ]

colnames(df_2)[2] <- "Expenditures_Millions"

df_2 <- df_2[1:(nrow(df_2)-7), ]

colnames(df_3)[1] <- "State"

combined_df_1 <- merge(df_1, df_2, by = "State", all = TRUE)

combined_df <- merge(combined_df_1, df_3, by= "State", all = TRUE)

combined_df$Expenditures_Millions <- gsub("[$,]", "", combined_df$Expenditures_Millions)
combined_df$Population_Estimate.2024 <- gsub(",","", combined_df$Population_Estimate.2024)
combined_df$Expenditures_Millions <- as.numeric(combined_df$Expenditures_Millions)
combined_df$Population_Estimate.2024 <- as.numeric(combined_df$Population_Estimate.2024)

combined_df <- subset(combined_df, State !="Puerto Rico")

combined_df <- combined_df |>
  mutate(per_capita_exp = round((Expenditures_Millions*1000000)/Population_Estimate.2024))

q1 <- quantile(combined_df$per_capita_exp, 0.2)
q2 <- quantile(combined_df$per_capita_exp, 0.4)
q3 <- quantile(combined_df$per_capita_exp, 0.6)
q4 <- quantile(combined_df$per_capita_exp, 0.8)

combined_df$Quintile <- ifelse(combined_df$per_capita_exp <= q1, "Lowest 20%",
                ifelse(combined_df$per_capita_exp <= q2, "Second Lowest 20%",
                ifelse(combined_df$per_capita_exp <= q3, "Middle 20%",
                ifelse(combined_df$per_capita_exp <= q4, "Second Highest 20%",
                "Highest 20%"))))

combined_df$Quintile <- factor(combined_df$Quintile,
                              levels = c("Lowest 20%",
                                       "Second Lowest 20%",
                                       "Middle 20%",
                                       "Second Highest 20%",
                                       "Highest 20%"))

2. Focus Metric:

Use obesity prevalence as the health outcome metric to analyze differences across spending quintiles.

## [1] "Mean Obesity Prevalence by Spending Quintile:"

##             Quintile Obesity..
## 1         Lowest 20%     33.10
## 2  Second Lowest 20%     35.70
## 3         Middle 20%     34.34
## 4 Second Highest 20%     32.86
## 5        Highest 20%     32.24

Visualization:

1. Create a Single Visualization in R (ggplot2):

Develop a visualization that directly addresses the research question. Examples:
- A box plot comparing obesity rates across spending quintiles.
- A scatter plot with spending on the x-axis and obesity prevalence on the y-axis, grouped by quintile.
Your visualization must be clear, professional, and directly relevant to the research question.

2. Accessibility Requirements:

Use an accessible color palette (e.g., viridis) to ensure clarity for all viewers.

ggplot(combined_df, aes(x = Quintile, y = Obesity.., fill = Quintile)) +
  geom_boxplot() +
  scale_fill_brewer(palette = "Set2") +
  labs(x = "Per Capita Healthcare Spending Quintiles",
       y = "Obesity Prevalence (%)",
       title = "Obesity Rates by Per Capita Spending Quintiles") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1),
        legend.position = "none", plot.title = element_text(hjust = 0.5))

Interpretation:

Write a single paragraph (3–4 sentences) summarizing your findings:
- Does the data suggest a correlation between healthcare spending and adult obesity prevalence?
- What elements of your graph directly answer the research question?
Highlight any anomalies or trends observed.

Analysis of obesity rates across healthcare spending quintiles reveals a complex relationship. While the second lowest spending quintile shows the highest median obesity rate at approximately 36%, the highest and second-highest spending quintiles demonstrate lower obesity rates, suggesting that increased healthcare spending might have some protective effect. However, this relationship is not straightforward, as evidenced by Hawaii’s position as an extreme outlier in the middle quintile (26.1%). When Hawaii is excluded from the analysis, the middle quintile actually emerges with the highest obesity prevalence. The substantial overlap in obesity rates across all spending quintiles indicates that higher healthcare spending does not necessarily translate to lower obesity rates, suggesting that states, particularly those in the second lowest spending quintile, may need to examine their healthcare strategies beyond simply increasing expenditure to effectively address obesity issues.