Introduction and Inspiration

Mental health is never too important to stress, from a personal as well as a social standpoint. Over the past 27 years, North America consistently had the highest share of population with mental health and substance use disorders (Figure 1). However, they still remain widely under-reported due to scarcer data, less attention and treatments at lower incomes. Pew Research Center reported that 2008 had seen a significant increase in the number of internet users who looked for information about depression, anxiety, or other mental health issues. The percentage rose to 28% as opposed to a relatively stable 22% over the the previous 5 years. Another report in 2013 showed that there were more progress seen on cancer, AIDS, smoking than other health problems, including mental illness (Figure 2).

Now years has past, has anything changed or has any progress been made? What are some major types of disorders and how do they vary for different groups? What are the difference between men and women, the youth and the adult? It’s important to understand the scale and status while bringing up such topics for discussion. Inspired by Figure 3, I decided to focus on the measure of estimated prevalence of mental disorders broken down by major categories and look for data that would start answering some of my questions. Through my research and data analysis, my goal was to present the estimates of mental health disorder prevalence within the United States during a 10-year period (2008-2017).

Materials & Methods

• Tableau Public 2019.2: Free software to create interactive visualizations that enables online sharing
• Global Burden of Disease (GBD) Results Tool: Online data source tool which is part of the data services provided by Global Health Data Exchange (GHDx)
• Microsoft Excel: Spreadsheet software used to explore and examine data

1.Data Selection and Collection

My data set came after exploring a number of public data sources such as Census & American Community Survey, Federal Data, Pew Research Center, United Nations, etc. On Quartz Directory, I came across a few topics regarding mental health and looked into the corresponding data sources. Global Health Data Exchange (GHDx) stood out to me because the GBD Results Tool services covers comprehensive sets of data around all GBD causes, types of measures / metrics, locations at different scales, etc. available from 1990 to 2017. There are a number of sorting options for users to query data with different combinations, and the returned dataset is free to download (Figure 4). I decided to collect data over a time frame of 10 years, which allowed me compare the data with my assumption of more recent trends.

• 

• 

• 

While this lessen the work of cleaning up data with external softwares, there was a bit of learning curve to understand what each terminology represents (e.g. Rate vs. Percent vs. Number, Figure 5 ) and how the filter works (e.g. users need to select all 51 states within the U.S. and leave out the option “United States” to retrieve data from each individual state, as shown in Figure 6; the age filter is filled with inconsistent categorizations of age groups, Figure 7), in order to get the most relevant data. The guidebook GBD 2017 Online Tools Overview was used to understand the tool and sorting options.

2.Data Visualization

• 

• 

• 

Prepared with a dataset of more than 285k rows, I began experimenting on Tableau Public. Drawing on previous examples I gathered for inspiration, I focused more on creating line charts to show trends and bar graphs to compare prevalence between genders, causes, ages and locations. Some data refinement was also done during visualization. For example, I group “Dysthymia” and “Major Depressive Disorder” under a “Depressive Disorders”, “Anorexia Nervosa” and “Bulimia Nervosa” under an “Eating Disorders”. In addition, while attempting to compare different age groups, the percentage of average prevalence rate wasn’t showing properly (Figure 8). After examining the original data (Figure 9), I created a customed calculated fields for this value in Tableau and again set the format to percentage, which then generates a number that make sense (Figure 10).

Results and Interpretation

Link to full dashboard

Notes on terms:

• Prevalence: The proportion of people in a population who are a case of a disease, injury or sequela
• Rate: measured per 100k population

1. Anxiety Disorders and Depressive Disorders are the most prevalent type of mental disorders and most distinct in gender difference. In addition to bar graph and line charts, a heatmap was also used to indicate level of severity. Figure 11 shows the prevalence of major mental disorders broken down by gender. Female with Anxiety Disorders has the highest rate, followed by its male counterpart, female with Depressive Disorders and male with Depressive Disorders. There is least visible difference between men and women in Bipolar Disorder and Schizophrenia.

2.Mental disorders are most prevalent within young adults and adults. Figure 12 shows a pattern in the overall mental disorders prevalence with relation to age and gender. The rate of prevalence peaks at females between 35 to 44 years old, followed by 25 to 34, 15 to 24 and 45 to 54, which are all female. Men have the highest rate also within the 35 to 44 age group, but still much lower than women. The rate of women having mental disorders exceeds men’s across all ages, except for those younger than 14 years old.

3. The trend remains quite stable, with slow decrease in Anxiety Disorders and slow increase in Depressive Disorders. Figure 13 visualizes the trends of each disorder over 10 years. Surprisingly, there’s no noticeable fluctuations since the average rates are almost flat. There seems to be a slight decrease in the prevalence of Anxiety Disorders, while Depressive Disorders has risen by similar rate. This chart demonstrates consistency with Figure 11, showing that these two are the most predominant type amongst all mental disorders, but it also suggests that Anxiety is almost twice as severe as Depressive Disorders.

4. Top states with most number of Depressive and Anxiety Disorders are almost identical. The top 4 states are the same, which are California, Texas and New York and Florida. Sorted using a descending order, the shape of the two graphs are also pretty close, except that Anxiety Disorders have doubled the scale (Figure 14). With this pattern, it is to be examined whether there is a correlation between Depressive Disorders and Anxiety Disorders, and if there is another metric that would better reflect the rankings in proportion to the state population.

Reflection

One of the most challenging parts of this project is finding the relevant data. Rather than searching for topics or data set that I’m interested in in an exploratory way, I had a topic in mind and came up with a rough picture of the visualizations from preliminary research. For example I knew I was looking for data available in certain demographic categories and has a timeline, but many of the data sources I initially looked at were as specific as “Percentage of adults in the United States with poor mental health as of 2017”, so it was not easy to find the exact matching data.

Using Tableau for the first time was also a trial and error experience for me. It took me a while to figure out the technical aspects of it. Working with data are not immediately straightforward. I felt like we could be easily stuck in what we have created, so I realized it is important to take a step back, think and experiment if there could be a better solution. I hope the amount of information will not overwhelm users, but at the same time informative enough for takeaways.

For future direction I plan to deepen the discussion by looking for data on some aftermath or influence caused by mental health issues (e.g. deaths, suicide rates, etc.). More information on demographics such as race/ethnicity, level of education, employment status, annual household incomes, etc. would also provide us with a broader picture about mental health issues in the U.S. Finally I will dig into details regarding some major types of disorder to present a more granular perspective, as individual characteristics could vary a lot.

After thought: After testing and critiquing during class, some changes could be made to improve my visualizations.

• Be more consistent with colors
• Make it more clear as of which graphs are combined with the filter, or make interaction unified
• Rename the headings to be more engaging (e.g. Q&A)

References

Depression, anxiety, stress or mental health issues, Pew Research Center

Americans see U.S. losing ground against mental illness, prescription drug abuse, Pew Research Center

Mental Health, Our World in Data