Vulnerable Populations Affected by Influenza
Challenge
As a data analyst for a medical staffing agency, I was tasked with forecasting influenza trends to optimize staffing during flu season. With a rise in flu cases, especially among vulnerable populations, hospitals and clinics faced a shortage of staff. My goal was to analyze data on mortality rates and age correlations to provide insights on when and where additional staff would be needed, particularly focusing on patients aged 65 and older.
Context
The stakeholders, a national medical staffing agency, wanted to proactively plan for flu season by identifying staffing needs based on flu trends. The agency's priority was to allocate staff to states and regions with the highest demand, preventing staff shortages during peak periods. By understanding the correlation between flu mortality rates and patient age, they aimed to improve resource allocation and reduce the strain on healthcare systems during high-demand periods.
Project Scale
3 weeks
Data
- “Influenza deaths by geography" Accessed from the CDC https://wonder.cdc.gov/ucd-icd10.html
- “Population data by geography, time, age, and gender” Accessed from US Census Bureau
Primary Stakeholder
CareerFoundry Data Analytics Course
Skills
- Data cleaning, integration, & transformation
- Data wrangling and merging
- Visual analysis
- Hypothesis testing
- Tableau visualizatoins
Questions from the Stakeholders
Who is most at risk?
Where are majority of vulnerable populations?
When is Influenza season?
Where do we need to send most of our staff?
Hypothesis
If we examine the age or patients in relation to influenza-related death, then we will see that older patients (over 65+) have a higher rate of death caused by influenza than younger patients.
Descriptive Analysis
The Correlation Coefficient for the variables of “Influenza Deaths (65+)” and “Population (65+)” is 0.94. Since the number is closer to 1, this suggests that there is a stronger correlation between these variables.
Results and Insights:
- Null Hypothesis: The risk of dying from Influenza is equal to or higher for people under the age of 65.
- Alternative Hypothesis: The risk of dying from Influenza is higher for people over the age of 65.
The significance level (0.05) is much higher than the p-value calculated (5.782E-175) which means we can reject the null hypothesis. Meaning that the risk of dying from Influenza is higher for people over the age of 65.
The Process
- Data Collection: Gathered data on Influenza deaths, population by state and age (from 2009-2017), and calculated the percentage of Influenza deaths relative to state populations.
- Exploratory Data Analysis: Conducted correlation analysis between Influenza deaths in individuals aged 65+ and the 65+ population by state, finding a strong correlation (0.94). Tested and rejected the null hypothesis, confirming that those over 65 have a higher risk of death from Influenza.
- Trend Identification: Analyzed Influenza death rates across different states and age groups, identifying the highest death rates specifically for patients over 65.
- Prioritization for Staffing: Categorized states into low, medium, and high priority based on Influenza death rates among vulnerable populations, focusing on areas with the highest need.
Analysis and Insights
The data tells us that the vulnerable population is anyone over the age of 65.
The population of who’s at risk, also know as vulnerable populations, are patients who are more likely to develop flu symptoms that can lead to death.
65% of Influenza deaths were in the vulnerable population.
States with a higher percent of vulnerable populations will have a higher death counth due to Influenza compared to those with a lower percentage of vulnerable populations.
States with the highest vulnerable population and highest death count.
This map shows us that the states with a higher vulnerable population also have a higher death count. However it could be missleading because that doesn’t necessarily mean that these states have the highest death rate in the country.
This is what we are actually interested in!
The top 10 states that have a higher vulnerable population death rate are different than the states that have the highest vulnerable population. Meaning that theses states are the states that struggle the most with staffing issues during influenza season.
The peak months are December-March.
Influenza season starts around November/December and ends March/April.
Recommendations
- Using the Influenza death count data, we classify the data into three priority levels: low (light blue), medium (medium blue), and high (dark blue).
- High priority states, including Alaska, Wyoming, Vermont, South Dakota, North Dakota, and District of Columbia, should receive top priority for additional staff allocation when devising the staffing plan.
- Medium priority states, such as Idaho, Montana, New Hampshire, Maine, Rhode Island, Delaware, and Hawaii, should be the second priority for staff allocation when devising the staffing plan.
- Remaining staff should be distributed among the states categorized as low-priority.
- It is crucial for the staff to be prepared for deployment to these states by early November, before the onset of the Influenza season. so they ca train for the peak season.
Retrospective
What went well?
Using Tableau was one of the highlights of this project, as I found the platform intuitive and fun to work with. It allowed me to quickly visualize trends and insights, making the analysis processs smoother.
What didn’t go well?
Initially, I faced challenges getting Tableau to load my entire dataset, which delayed progress. After spending time experimenting with different options and settings, I was able to resolve the issue and continue with the analysis.
Future steps
In future analysis, vaccination rates should be examined to understand their effect on Influenza death rates. This additional data could help improve the accuracy of staffing predictions and enhance overall recommendations.
Final thoughts
This was my first experience using Tableau and my first project in data analytics, and I found the entire process exciting and rewarding. It sparked a deeper interest in the field, and I’m eager to take on more projects to continue growing my skills and knowledge.
Want to see more?
Check out my Tableau Presentation to see more recommendations and a more in depth analysis and my Github to see more information.
