Disparities in Healthcare System and Insurance Coverage in the Modern US

Authors: Patricia Cheng (Project Lead), Aashman Rastogi, Larry Lu, Aida Mohasesi, Janet Yu

Introduction

In a post-pandemic United States, understanding the intricacies of healthcare is more crucial than ever. Those who were uninsured and became hospitalized due to coronavirus find themselves struggling to pay off their medical bills. As hospitals recover from the long two-year inflow of patients, health insurance companies have started hiking up rates to match the increasing number of people seeing doctors who may have put off routine health check ups during the pandemic. It remains clearer than ever that the healthcare system is flawed and should be more equitable and affordable for all. Although it remains a vast and complicated realm of knowledge, several crucial aspects are worth taking a deeper dive into. In this article, we attempt to uncover insights and discover connections between key topics within the healthcare landscape, including finding health insurance coverage variations across racial groups in order to discover any racial disparities, looking into the impact of healthcare funding on mortality rates, determining whether prevalence of cardiovascular diseases, pneumonia, and influenza differ geographically, and predicting what factors are most indicative of a high health insurance premium using machine learning. Read on to learn what we found!

Crude Rates and deaths of Influenza and Pneumonia

Influenza and Pneumonia are significant causes of death in the United States. According to data from the Centers for Disease Control and Prevention (CDC), influenza and pneumonia consistently rank among the top 10 leading causes of death in the country. Influenza, commonly known as the flu, is a contagious respiratory illness caused by influenza viruses. It can lead to severe illness, hospitalization, and even death, particularly among vulnerable populations such as the elderly, young children, and individuals with underlying health conditions. Influenza-related complications, such as pneumonia, contribute to mortality. Pneumonia is an infection that inflames the air sacs in one or both lungs. It can be caused by various pathogens, including bacteria, viruses, and fungi and can be severe and life-threatening especially in older adults, young children, and individuals with weakened immune systems.

Tracking mortality trends associated with influenza and pneumonia provides valuable information for disease surveillance systems in order to control the spread of diseases. It helps identify regions or populations that may be experiencing higher mortality rates, allowing for targeted interventions and surveillance efforts. This information aids in early detecting outbreaks, monitoring the impact of interventions, identifying vulnerable populations that may require additional support, and developing effective strategies such as vaccination campaigns, public health messaging, and preparedness for outbreaks or pandemics.

Here we provide a general overview of how mortality trends have changed over the years 1999–2020 using the data from CDC, which looks at places with highest mortality rates and focuses on states which have been able to implement good systems in place following outbreaks.

Animation video shows changes in Death rates across the US

• Made using plotly-express — not sure how to upload it onto the web. I do have an HTML link which I can share.

The map displays the average crude rates of deaths across U.S. states from 1999 to 2020, revealing that Arkansas has the highest number of deaths. Several factors may contribute to this situation. The CDC provides an equation for how to calculate the Crude Death Rate using this formula:

Crude Rate = Count / Population * 100,000

Firstly, Arkansas’s humid, cold, and wet conditions create a favorable environment for the virus to thrive and spread. Secondly, the state’s relatively low vaccination rate leads to higher transmission rates of respiratory infections like influenza and pneumonia. These infections can exacerbate underlying health conditions and ultimately result in higher mortality rates.

Similarly, New York and West Virginia have high crude rates, owing potentially to high population density and higher rates of transmission. Alaska’s geographic location and consistently cold climate throughout the year, coupled with its low population density, also contribute to the state having lower crude rates of influenza. The extreme cold climate limits the survival and transmission of the virus, while the lower population density reduces opportunities for person-to-person spread, particularly during the winter months when social interactions are naturally reduced.

Here we introduce a video showing the changes in mortality across years.

The aforementioned graphs for New York, Florida, and California reveal intriguing patterns, particularly in terms of the high prevalence of influenza and pneumonia at the start of the twenty-first century. Notably, California experienced a significant surge in deaths within a single year, specifically from 1999 to 2000, attributable to the outbreak of the H3N2 virus. The outbreak resulted in a rapid doubling of deaths in California during that period.

Following a peak in the years 2000–2003, the mortality rates for influenza and pneumonia in California have notably declined and have since remained within the range of 6000–7000 deaths per year. Similarly, in New York, the mortality rates have stayed relatively consistent within the range of 4500–5000 deaths per year, while in Florida, they have remained in the range of 2000–3000 deaths per year.

The intriguing aspect of these graphs lies in the opportunity they present to study the response of federal and state governments in containing and controlling the disease. By analyzing the measures implemented during these specific outbreaks, valuable insights can be gained, allowing for the application of similar strategies in other states that may face or anticipate similar outbreaks during specific seasons in the future. This cross-state learning and implementation of effective measures can contribute to better preparedness and a more coordinated response to influenza and pneumonia outbreaks nationwide.

IGNORE BELOW GRAPHS ONLY NEED VIDEO

The changes might me be inherent form just the images but the video gives a good Idea

Crude Rates and deaths of Cardiovascular Diseases

Next, we try to relate the prevalence of cardiovascular disease by region. Cardiovascular disease, also called heart disease, is currently the leading cause of death for people of almost all racial and ethnic groups in the US. look at how the prevalence of cardiovascular disease is affected by region. It could be caused by an array of factors, including high blood pressure, smoking, diabetes, high blood cholesterol, and poor diet, etc. The National Center for Health Statistics provided this map showing the significance of cardiovascular mortality rate:

To learn more about the trend, we used the data from CDC (Centers of Disease Control and Intervention) in 2021 and tried to rank the states based on their mortality rate per one death. Our result was as the following:

It seems like , among the regions noted in red, Oklahoma has the highest mortality rate for cardiovascular disease in 2021, whereas Minnesota has the lowest. In fact, both the map and the bar graph shows that people in the Southeast seem to suffer from cardiovascular diseases the most, and the percent mortality rate is also much higher than that of all the other regions. In order to see whether location actually affects the mortality rate of those with cardiovascular diseases in the US, we also plotted the crude rate of cardiovascular diseases by region from 2010–2020.

Click here for Video of Transgression from 2010 to 2020

Dashboard of Crude Death Rate of Cardiovascular Diseases, 2010–2020

This video highlights the transgression of cardiovascular disease related deaths determined from their crude rate from 2010 to 2020. The major cardiovascular diseases accounted for in the crude death rate displayed in the map include, but are not limited to the following list, plus their disease code on the CDC:

Major Cardiovascular Diseases

Diseases of Heart:

Acute Rheumatic Fever & Chronic Rheumatic Heart Diseases (I00-I09)

Hypertensive Heart Disease (I11)

Hypertensive Heart & Renal Diseases (I13)

Ischemic Heart Diseases (I20-I25)

Acute Myocardial Infarction (I21-I22)

Other Acute Ischemic Heart Diseases (I24)

Other Forms of Chronic Ischemic Heart Disease (I20, I25)

Atherosclerotic Cardiovascular Disease, so described (I25.0)

All Other Forms of Chronic Ischemic Heart Disease (I20, I25.1-I25.9)

Other Heart Diseases (I26-I51)

Acute & Subacute Endocarditis (I33)

Diseases of Pericardium & Acute Myocarditis (I30-I31, I40)

Heart Failure (I50)

All Other Forms of Heart Disease (I26-I28, I34-I38, I42-I49, I51)

Analyzing the transgression of the crude death rate as displayed by the video and maps, the legend on the right shows that the lower bound of the crude death rate interval increases by ~30 deaths per 100,000 capita and the upper bound of the crude death rate interval increases by ~40 deaths per 100,000 capita. Overall, this highlights an increase in the number of deaths related to major cardiovascular diseases across the US. Visually, the dispersion of crude death rate across the US states remains largely the same, with a concentration in Eastern States not along the coast like Mississippi and Alabama while the crude death rates in the Western and Northern states seem to be slightly decreasing.

After looking further into studies that analyze mortality rates of cardiovascular diseases, we found that mortality rates have decreased overall since 2010 to present day with a slight spike during 2020 due to COVID-19. Such information correlates with the findings from the map above showing an increased crude rate, which could mean a consolidation of deaths in a certain area, but a more significant decrease in other regions. This insight brought us to the question of what factors might have contributed to people in these regions having heart diseases the most. The first factor we considered was race, and we did so by examining how insurance coverage differed by race as well as how it was related to the proportion of different racial groups residing in each of the 50 states in the US.

Racial Disparities in Insurance Status

Racial and ethnic disparities in health insurance coverage remain a persistent challenge in the United States. While inequities in health care have been reported for decades, the COVID-19 pandemic’s uneven impact on persons of color has placed increasing attention on these inequities. Health insurance coverage significantly affects people’s ability to access health care and protects people from high medical costs. However, according to 2021 data supplied by the United States Census Bureau shown below, the number of Hispanics currently uninsured drastically exceeds the national average in comparison to Whites, non-Hispanic Whites, Blacks, and Asians. The percent of uninsured Hispanics is over 2 times higher than that of Whites, and Black individuals also appear to be more likely to be uninsured than Whites and non-Hispanic Whites suggesting that people of color still disconcertingly fare worse in terms of access to health care resources.

Further racial and ethnic disparities regarding health care can be seen in terms of care quality via the percentages of racial groups covered by private versus public health insurance. Existing literature has shown that healthcare service quality, defined as staff responsiveness measured by patient satisfaction, is better in the private sector than in the public sector (Morgan et al., 2016). According to an NIH study that compared the performance of private and public health care systems in low and middle income countries, clients thought the service quality of private providers was better than that of public providers due to better hospitality, better staff availability, shorter waiting times, greater time spent with doctors, and cleaner facilities (Basu et al., 2012). Authors Bhatia and Cleland also compared public and private health care of female outpatients in south-central India, and they reported that “private practitioners are providing a better service, defined in relation to consultation time, privacy, and likelihood of receiving information about diagnosis and prognosis, than their public sector counterparts” (Bhatia & Cleland, 2004).

These comparative studies thus reflect that private health insurance provides higher quality care than public health insurance providing a basis to interpret the disparities shown in the following visualizations. Evidently, Black and Hispanic persons are less likely to be covered by private insurance than White, non-Hispanic White, and Asian persons. The number of Blacks and Hispanics currently covered by private insurance falls significantly below the national average while the number of Whites, non-Hispanic Whites, and Asians currently covered by private insurance exceed the national average. Staying consistent with the conclusions of the aforementioned comparative studies regarding the healthcare sector, Black and Hispanics seem to receive lower quality health care than Whites, non-Hispanic Whites, and Asians, on average.

Prevalence of Cardiovascular Diseases Explained

Recall that, from the Cardiovascular Disease crude rate section, we found that people in the Southeast regions seem to be more subject to cardiovascular diseases compared to those in other regions. We then decided to look at how racial composition in each of the 52 states in the US could potentially shape the prevalence of cardiovascular disease, to which we plotted the total population by race in each state.

While some of the Southeastern states, such as Alabama, Louisiana and Mississippi, do seem to have more Black people in proportion to other states, which might be linked to the lack of insurance status, the result did not seem to be very significant as there were some outlier states, not to mention that none of the southeastern states seem to have more Hispanic people compared to other states. Therefore, we decided to consider another factor that might be directly related to the health status of the general population, which is the percentage of obesity by states.

Map: Percentage of Total Obese Population by States

Obesity is a condition in which a person is overweight compared to his or her height, which is defined as having a Body Mass Index of 30.0 or above. Based on the map from CDC shown above, we can see that the people who are more likely to be obese tend to cluster around the south and east. In particular, the U.S. Census Bureau has identified some of these regions with the highest obesity rates as some of the poorest states in the US.

This is an unfortunate observation, as we often see poverty associated with the more consumption of fast food, which tends to mostly be high in cholesterol and therefore very unhealthy. In fact, not only is Mississippi the state with the most obese people in the nation, but also the poorest, with 21% of its residents living below the poverty line. These are also regions that are home to a lot of Hispanic or Black people, who we see from the insurance analysis that are more likely to be uninsured . It is possible that little access to healthy food, a more sedentary lifestyle, and an overall worse life quality might have all contributed to these people being more susceptible to diseases and conditions such as cardiovascular diseases, which might be a result of systematic racism.

Influential Factors in Raising Insurance Bill

One of the most interesting questions that arises from investigating insurance in general is what factors most affect the amount of money people are charged for their annual premiums. Anyone who’s just begun learning to drive knows that their parents hate paying for their car insurance because of how expensive it is. This is because car insurance companies know inexperienced drivers who have never been on the road before are much more likely to be involved in an accident of some sort, meaning the insurance firm has a high chance of paying out compared to an adult who has been driving for years. The same concept applies to health insurance. But what factors are the most influential when determining how much health insurance premiums should be?

To answer this question, we can build a random forest regression model using age, sex, BMI, number of children, whether or not one smokes, and region of residence as the predictors and the annual health insurance charges as the target variable we want to predict using the machine learning model. By using only 80% of the data to train the model on, we can then utilize the remaining 20% of data to test how accurate the model is at predicting health insurance premiums.

With an R2 value of 0.86, this model is doing pretty well at predicting how much people are being charged! Most of the points lie along the line indicating 100% accuracy, and many of the outliers are most likely due to other health issues like chronic illnesses that aren’t accounted for in the dataset. It’s quite plausible that health insurance companies might be using a similar model to predict how much they’ll be paying out on claims based on an individuals’ stats. But what factors are most important to an insurance company? Turns out, we can calculate an index to describe how much each feature (predictor variable) affects the overall prediction of health insurance cost.

Somewhat unsurprisingly, whether or not one smokes (0.61) is the biggest decider in how much health insurance companies charge for their premiums. This makes sense, given how hazardous smoking is to one’s health. Research has proven that active smokers suffer higher mortality rates than overweight and obese ex-smokers, which may explain why BMI is much lower in the rankings with an importance of 0.22 (Siahpush M;Singh GK;Tibbits M;Pinard CA;Shaikh RA;Yaroch A;). Finally, age, while not as much of a deciding factor as BMI, sits at 0.13. The rest of the predictors, number of children, region of residence, and sex, do not hold much bearing on the insurance costs.

From this model, we can attempt to explain why certain people are unable to afford healthcare, and what measures can be taken to improve the issue.

Government Funding and its Effect on Mortality Rate

In the pursuit of understanding the relationship between medical funding and mortality rates, we embarked on a comprehensive analysis utilizing regression models and examining correlations. Our goal was to uncover any significant trends or associations that could shed light on the impact of funding on mortality. However, our findings revealed a complex and nuanced relationship that cannot be generalized easily.

We began our exploration by examining a graph depicting mortality rates and expenditure over time. To explore the potential trend or correlation between healthcare funding and mortality rates over the years, we once again utilized data from CDC for mortality rates and the Centers for Medicare & Medicaid Services for national funding. By examining these datasets, we aimed to gain insights into any potential relationship between healthcare funding and mortality rates.

The first graph below shows the population growth and mortality rates over the years where we can see a sharp increase in death due to Covid. The second graph below shows the comparison between health expenditure and mortality rates. Interestingly, the shape of both graphs, although with different y-axis, match due to pure coincidence.

Below we can see the composition of health expenditure, which helps us understand where the majority of our government funding is coming from.

Additionally, we can analyze specific programs within the federal government by examining how their funding has evolved. The provided graph illustrates the trends in Medicaid and Medicare funding over the years, which are two of the most popular public health insurance offered by the government towards specific groups of people. By observing these funding patterns, we can gain insights into the changes and allocations within these crucial healthcare programs.

Surprisingly, we observed no discernible correlation or trend in the data. Despite the increase in medical funding, mortality rates continued to rise consistently. This initial observation left us intrigued and eager to delve deeper into the data. To gain further insights, we conducted correlation analyses between mortality rates and various streams of expenditure. However, our findings also indicated weak correlations, with values ranging from 0.01 to 0.2. These results suggested that the influence of individual expenditure streams on mortality rates was minimal.

In summary, our regression analyses suggest that National Health Expenditure (NHE) and population size are significant factors associated with mortality rates. However, the presence of multicollinearity indicates that other variables or factors might be influencing the relationship. It is crucial to recognize that medical funding and mortality rates are multifaceted, influenced by various streams of expenditure, allocation strategies, and utilization patterns.

Our findings underscore the need for a more nuanced understanding of the relationship between medical funding and mortality rates. The effects of funding are not straightforward or uniform; they vary depending on how funding is provided, allocated, and utilized across different healthcare domains. To gain comprehensive insights, future research should consider exploring the specific streams of funding and how they impact mortality rates in different healthcare contexts, and diving deeper into the intricate interplay between funding streams, resource allocation, and healthcare utilization.

Conclusion

Now that we’ve gained a better understanding of how healthcare companies think and operate, more informed decisions can be made by customers, researchers and policymakers alike. Discovering which states are most susceptible to which sorts of diseases can be invaluable in order for an increased budget to be allotted for the treatment of such diseases. However, as previously mentioned, more funding does not necessarily lead to an instant improvement in quality of life or lower mortality rate; the effect of the expenditures vary greatly based on how efficiently and wisely they are allocated across different branches. Furthermore, although affordable and equitable healthcare for all would be ideal, this is sadly not the case right now. Seeing that the uninsured rates across different races are drastically different, more of the budget can be dedicated toward providing healthcare for all while also increasing the quality of public healthcare to match private companies. Finally, by discovering what exactly insurance companies are looking for when deciding on a health insurance premium can help those with higher rates make healthier choices to lower their costs of healthcare. With these insights, we hope that America will become an overall healthier country in the future.

Sources:

Basu, S., Andrews, J., Kishore, S., Panjabi, R., & Stuckler, D. (2012). Comparative performance of private and public healthcare systems in low-and middle-income countries: a systematic review. PLoS medicine, 9(6), e1001244.

Bhatia, J., & Cleland, J. (2004). Health care of female outpatients in south-central India: comparing public and private sector provision. Health policy and planning, 19(6), 402–409.

Morgan, R., Ensor, T., & Waters, H. (2016). Performance of private sector health care: implications for universal health coverage. The Lancet, 388(10044), 606–612.

https://www.ahajournals.org/doi/10.1161/circ.146.suppl_1.9853

Manemann, S.M., Gerber, Y., Bielinski, S.J. et al. Recent trends in cardiovascular disease deaths: a state

specific perspective. BMC Public Health 21, 1031 (2021). https://doi.org/10.1186/s12889-021-11072-5. Accessed 3 June 2023.

Woodruff, Rebecca C., Tong, Xin, Jackson, Sandra L. et al. Abstract 9853: Trends in National Death

Rates From Heart Disease in the United States, 2010–2020. (2022). https://www.ahajournals.org/doi/10.1161/circ.146.suppl_1.9853. Accessed 3 June 2023.

Siahpush M, Singh GK, Tibbits M, Pinard CA, Shaikh RA, Yaroch A. It is better to be a fat ex-smoker than a thin smoker: findings from the 1997–2004 National Health Interview Survey-National Death Index linkage study. Tob Control. 2014 Sep;23(5):395–402. doi: 10.1136/tobaccocontrol-2012–050912. Epub 2013 Apr 10. PMID: 23574644.

“Heart Disease Facts.” Centers for Disease Control and Prevention, 15 May 2023, www.cdc.gov/heartdisease/facts.htm#:~:text=Heart%20Disease%20in%20the%20United%20States&text=One%20person%20dies%20every%2033,United%20States%20from%20cardiovascular%20disease.&text=About%20695%2C000%20people%20in%20the,1%20in%20every%205%20deaths.&text=Heart%20disease%20cost%20the%20United,year%20from%202018%20to%202019.

“Adult Obesity Prevalence Maps.” Centers for Disease Control and Prevention, 17 Mar. 2023, www.cdc.gov/obesity/data/prevalence-maps.html.