Column 1

Title: Outcomes from a simulated Ebola outbreak

Subtitle: by Hospital and Age

Author: Nisha Udupa

Data source: ebola_simulated.rds

Background: The simulated ebola dataset examines 5888 cases of ebola in West Africa, between the years 2014-2015. There are several pieces of data collected in this dataset including information about treatment hospital, patient outcome, as well as basic health information. One of our goals in this project is to examine factors that may affect the two outcomes: death and recovery. We examined the relationship between treatment hospital and outcome, and secondly, the relationship between age and outcome. Our goal in looking into treatment hospitals is to better understand where patients are being treated, how we can determine hospitals that need additional support, and identify hospitals with poorer performance. We also look into age to understand if there is a relationship between age and ebola survival.

Results: To examine the outcomes based on treatment hospital, we constructed two separate bar charts. The first stacked bar chart showed the total number of patients per hospital, separated into outcome. From this, we were able to see that Port Hospital had a significantly higher death toll. To more clearly see the distribution of outcome we constructed another bar chart using proportion of cases by hospital instead. This chart indicated that although Port Hospital may have had a higher number of deaths, it is in fact St. Mark’s Maternity Hospital (SMMH) that had a higher proportion of deaths relative to recoveries or NAs. This indicates that additional resources should be directed towards Port Hospital since they are taking in the majority of patients, and that additional analyses need to be done to understand why SMMH is having poorer outcomes relative to the other hospitals.

The increased proportion of deaths at SMMH led us to examine the relationship between outcome and age, since a maternity hospital would also treat a higher number of infants. The box plots constructed show the distribution of age across all three outcomes, and we expected to see that the average age for the outcome of death would be younger. From this, we are able to see that the median age is 13 years for both outcomes and for those without an outcome recorded. Beyond this, the distribution from Q1 to Q3 is the same for both those who recover and those who die. Based on this box plot, we cannot conclude that there is a relationship between age and outcome. Our results highlight where resources should be allocated as well as highlight future areas for research to better understand why we are seeing these results.

Count of Outcomes by Hospital

Distribution of Outcomes by Hospital