Influenza Season: Hospital Staffing
This project analyzes historical influenza trends across all 50 U.S. states to forecast peak flu activity and identify when and where to deploy temporary medical staff. The goal is to help a staffing agency ensure hospitals are adequately staffed during seasonal surges in flu-related hospitalizations.
Audience
Medical Frontline Staff
Hospitals & Clinics
Influenza Patients
Staffing Agency Admin
Data
Influenza Deaths by Geography
Source: CDC
Population data by geography, time, age, and gender
Source: US Census Bureau
Final Integrated Data Set:
An Integrated Data set was created by combining the Influenza Deaths data with the Census Population Data using the combined key [State, Year]. The Integrated Data set shows both Population and Death counts by age group, broken down by each state, each year.
Techniques used for the project:
Microsoft Excel for data analysis
Tableau for statistical, temporal, spatial and textual visualization
Designed a data research project
Data merging, transforming, and manipulation
Statistical Hypothesis testing
Data Storytelling and Presentation
Key Business Questions:
● What is the main problem we are focusing on solving?
● Who are the people with the highest risk of death or complications due to influenza?
● What states have the highest number of fatalities due to Influenza?
● When is “Flu Season” at its peak?
● Where should we focus our additional staffing during peak “Flu Season”?
What is the problem we are trying to solve?
In the United States, there is an "Influenza Season" during which flu-related hospitalizations and deaths rise significantly, especially among vulnerable populations.
Who are the ‘Vulnerable Populations?
Hypothesis: If a person is over 65 years old, they are at higher risk of death due to Influenza.
This bar chart confirms our hypothesis:
People over 65 (particularly those over 85) are at highest risk of death from Flu.
*0 Deaths reported for the age group 0-4, so we can assume that children under 5 are not ‘vulnerable’.
Descriptive Analysis
t-Test (Hypothesis Testing)
Null Hypothesis: The death rate from Influenza in people 65+ is less than or equal to the death rate from Influenza for people younger than 65.
Alternative Hypothesis: The death rate from Influenza in people 65+ is more than the death rate of Influenza for people younger than 65.
This is a one-tailed test because we are only interested in whether individuals aged 65 and older die from influenza at a higher rate than those under 65.
The p-value is effectively zero, well below the 0.05 significance level, indicating a statistically significant difference.
We can reject the null hypothesis and conclude that older adults are more likely to die from influenza.
This insight is valuable for our medical staffing project, as it supports prioritizing additional resources in states or counties with larger elderly populations.
Where is the problem? In what states are the flu deaths happening?
Flu deaths occur more prevalently in states with higher populations, but especially in states with higher populations of 65+ adults.
When is ‘Flu Season’ at it’s peak? When do hospitals need the additional staffing the most?
The bar graph of Influenza deaths by month shows that ‘Flu Season’ peaks from December through March, with January showing the most deaths.
The bar chart of Influenza deaths by year shows that Influenza deaths are not consistently going up or down by year. They are staying relatively consistent over time. This would be a good KPI for us to come back to later to see if our staffing plan works.
Data Limitations:
Flu Death data: 54,013 of the 66,096 records in this data set were input with ‘suppressed’ as the amount of deaths for that county that year. This means that the amount of deaths was between 0-9 deaths, but the actual number could not be reported due to risk of privacy infringement.
I decided to mark all ‘suppressed’ entries as 0 deaths for two reasons:
It is likely that the majority of these were actually 0 deaths.
Reporting any number of deaths between 1-9 is insignificant when dealing with
our medical staffing project because we are focusing on sending help to states
with much larger numbers of Influenza deaths.
This means that, for instance, states like Alaska probably did have some Flu deaths, but
the number is so small that it would not have changed the outcome of our analysis.
Results and Recommendations:
There is a very strong correlation between population of 65+ adults and amount of Flu deaths. We should focus our staffing efforts to primarily help out in states and communities with larger populations of 65+ adults.
The peak ‘Influenza season’ happens from December through March (the Winter months). We should ensure that medical staffing is fully ready to deploy by December this year.
