A data-driven exploration of healthcare infrastructure and its potential impact on mortality rates in the Philippines
Despite advancements made in technology and medicine, access to proper healthcare remains limited for many Filipinos, possibly contributing to the country's high mortality rate up to this date.
As such, we intend to use this study to examine the factors influencing healthcare accessibility and to determine its correlation, if significant, with Filipino mortality rates, shedding light on the disparities in healthcare availability across different regions.
Throughout the years, accessibility to healthcare and basic health services continue to be scarce and uneven in the Philippines.
In April of 2024, the World Health Organization stated that almost 800 million people in the Western Pacific region, including the Philippines, have insufficient access to basic healthcare services. This is further seen in the continuous increase of household expenditures from ₱686.49 billion in 2023 to ₱754.54 billion in 2024. Although the amount of facilities have increased, totaling to 40,108 health facilities in the country as of January 2024, this remains disproportionate among regions with CALABARZON having 4,415 facilities while other regions such as NCR and CAR having 2,784 and 1,278 facilities respectively.
The inaccessibility of proper healthcare for many Filipinos has been exacerbated by the lacking amount of facilities and professionals in the industry as well as the increasing pricing of healthcare itself.
Our solution is to make use of data science methods and tools to analyze these factors that affect the accessibility of healthcare among Filipinos as a way to gain insights of the current situation of our healthcare system, giving way for feasible solutions to be implemented.
We aim to answer the following questions:
and came up with:
There is little to no correlation between the amount of operational health facilities in a region and the region's corresponding mortality rates.
There is a significant correlation between the amount of operational health facilities in a region and the region's corresponding mortality rates.
Where we got our data:
The 2023 yearly report on deaths in the Philippines. The dataset has a total of 698,421 observations.
The master list of all active health facilities and health-related facilities. The dataset has a total of 118 observations.
The report on the national facility-based routine health information system used by the Department of Health to collect and manage data on health service delivery and public health programs. However, we only used data on the population in each province and region. The dataset has a total of 122 observations.
How did we explore our data?
We began our exploration by taking the master list of active health facilities, cleaning and normalizing each "Region" label, then computing a new "Total Facilities" column (summing hospitals, RHUs, barangay stations, birthing homes, etc.). Next, we grouped that table by region and plotted a simple vertical bar chart of total facilities per region. That visualization let us immediately spot which regions (e.g. Region IVA, Region V) had the largest networks of clinics and which (e.g. BARMM, Region IX) had the fewest, guiding our subsequent per-capita and mortality-rate analyses.
Some regions have significantly more health facilities than others, highlighting infrastructure disparities.
A scatter plot with each region having a dedicated data point. Each point's x-value is the region's health facility density ([amt. of health facilities / total population of the region] * 10000) and its y-value is its respective mortality rate ([amt. of deaths / total population of the region] * 10000). A best fit line created using least squares regression shows a positive relationship between a region's health facility density and its mortality rate.
Contrary to initial expectations, regions with higher health facility density sometimes show higher mortality rates, suggesting complex socioeconomic factors at play.
Pearson's Correlation Coefficient:
Indicating a positive but statistically insignificant correlation
We employed the Pearson's r correlation coefficient test to test for the correlation of the independent variable of our study, the health facility density of a region (per 10,000 people), with the dependent variable of our study, which is the mortality rate of a region (per 10,000 people). We executed this test using Python, as well as a CSV created from our dataset with the data for Region Name, Deaths, Population, Mortality Rate (MR), and Health Facility Density (HFD) -- which were used in creating the graph for Research Question 3. With this test, we found a correlation coefficient of 0.169, suggesting that there is indeed a correlation between HFD and MR. With that in mind, however, it is worth noting that the p-value computed by the correlation test is greater than the standard 0.05, at a value of 0.518. This means that 0.169 finding is statistically insignificant – or that the observed results could occur randomly, and not prove any relationship at all between the two variables. We have then failed to reject the null hypothesis.
Due to the limited metrics available in our current dataset, our sample size of 18 regions was not "sufficiently large" for the data to be treated as normally distributed. Thus, we failed to reject the null hypothesis set in this study: that there is little to no correlation between healthcare facility density and mortality rate in a region.
Be that as it may, we believe our exploration and analysis of the dataset remain relevant to the problems we set out to investigate. The scatter plot shows that, even though the correlation between the two variables was statistically insignificant, there may still be a positive relationship: increasing the number of healthcare facilities in densely populated regions appears to coincide with higher mortality rates in those same regions.
One possible interpretation is that an increase in healthcare facilities does not directly raise mortality rates, but instead leads to more accurate and complete recording of deaths. In other words, improved healthcare infrastructure might enhance logistical capabilities, such as data collection and record-keeping, rather than affect mortality rates directly.
What conclusions or assumptions can be made from this? We believe that due to the lack of data on other possible factors influencing mortality rates—such as poverty levels, unemployment, geographical conditions, and more—we were unable to establish meaningful connections to explain this relationship.
While our current findings do not support a statistically significant correlation, they underscore the importance of a multidimensional approach in health systems research in a country as complex as the Philippines. The lack of strong correlation does not necessarily imply the absence of a relationship, but rather points to the limitations of analyzing health outcomes without considering other crucial variables.
These findings should serve as a foundation for more comprehensive and inclusive studies in the future. Conducting future research can better identify the root causes of regional disparities in health outcomes and contribute to evidence-based solutions. In this way, data science can become a powerful tool not just for understanding the state of public health, but for enacting substantial, lasting change across the country.
"Our results suggest that improved healthcare infrastructure may lead to better data reporting rather than higher mortality, raising questions that require deeper investigation."
II - BS Computer Science | CS 132 WFV
III - BS Computer Science | CS 132 WFV
III - BS Computer Science | CS 132 WFV