For my final blog update, I wrote a report summarizing my lab and my work on the project this spring and summer. Please keep in mind that my data analysis is not complete and it is a project I am continuing with the lab in the fall semester. I did include citations in this report to back up any claims I made.
To have a clear understanding of my independent project, one must be familiar with the intern health study and the work it has been involved in. The intern health study is an IRB approved longitudinal cohort study that focuses on the relationship between stress, depression, mood, health, and other factors such as suicide ideation in first year medical interns, or residents (Intern Health Study). The study currently works with a long list of medical residency programs in the United States and China. Another component of the study accesses the genetic factors that may have a connection to depression, specifically interactions on the gene level that may factor into stress occurrence. Now, my independent scope of research is aimed at the possible connection between daily mood variation and the prevalence of depression and depressive symptoms such as suicide ideation in first year medical interns. An extensive preliminary literature review revealed some data on the topic, however it has not been extensively explored.
Some of the greatest support I found for the relationship between mood changes and depression comes from a study done at the University of California Irvine Long Beach in 1984. This study looked at quantifying emotional changes experienced by medical interns and how it relates to depression during the internship year, especially in the first few months (Uliana et al). There was deep insight on the variability and intensity of mood changes in medical interns which vouched for support systems that were aimed to reduce daily stresses in interns that could possibly lead to more severe depressive symptoms. Another study done in 1982 at the Oregon Health Sciences University explored the correlation between changes in emotions and mood and how that led to the onset of depression systems in medical interns ( Girard et al). This was done via self-report methods of daily mood and periodical emotions, similar to the methods of The Intern Health Project. Finally, it is important to summarize why exploring daily (short-term) mood variation and its connection with depression is important. A 2003 study by Gregory E. Simon discussed the negative effects of mood disorder such as lost work productivity and the increased use of health services (Simon). These burdens and more are great example for the support of more short-term (day to day) treatments for mood disorders than can lead to worse symptoms like suicide ideation. I hope the product of my final research project will extend the support for these treatment programs across residency locations in the United States and beyond.
To start, I generated a couple hypotheses based on current literature and my personal scientific predictions before exploring the data made available to me through The Intern Health Study. The first is: We predict that depressive symptoms will increase in first year medical residents that show a greater variation in daily mood. The second states: We predict that suicide ideation will increase in first year medical residents that show a greater variation in daily mood. Presence of depression was measured with the well-known PHQ-9 scale with a value greater than 10 indicating the onset of at least moderate depressive symptoms. The term variation in daily mood refers to how greatly the self-reported mood score of each medical intern in question varies from their unique average mood score over the entire study duration of twelve months. The entire twelve months were utilized to derive this average as it was thought to be more representative of each subjects average daily mood. My data analysis plan could be summarized in a few steps. Before, It is important to identify that I used data from the 2017-2018 cohort of the study that was completed in June of 2018. First, I must analyze each subject’s PHQ-9 score which corresponds to a total of two weeks worth of mood data as the scale is designed. Each subject that was considered valid for my data analysis should have at a minimum of one PHQ-9 score and at a maximum of four PHQ-9 scores, as the score is collected via quarterly surveys sent out by the study. Second, I must look at the mood scores that corresponds to each “PHQ-9 depression score” and compare each daily score to each subject’s average that was mentioned above. This will give me an “average daily variation from the mean” value that basically tells me how much their mood fluctuates from day to day. Finally, I will compute some statistical tests that will be discussed in more detail soon to determine if mood changes and depression/suicide variation are significantly correlated with each other. Due to the extremely large amount of data that must be analyzed for this project’s completion, at least six weeks are required to fully complete the data analysis and determine if my findings are statistically significant , however I will provide an example of the analysis I am completing.
For example, let us take into account subject “A” who began providing data to the intern health study on June 2nd 2017. They began providing daily mood scores on June 2nd 2017 and continued doing so throughout the next twelve months, non- continuously. Their PHQ-9 score for the first quarter of data they provided was a value of seven which corresponds with mild depressive symptoms although it is not an official diagnosis of any type. Keep in mind that this score correlates to two weeks of data prior to the completion of their first quarterly survey that was sent out in September. In this case, valid mood values that are related to the PHQ-9 score of seven would fall between August 25th 2017 and September 8th 2017. The subject recorded a total of six mood scores in this two week period. Their average reported mood score throughout the entire twelve month period was a 7.29 out of 10 so their individual mean mood score will be that value. The next step would be to graphically illustrate how the six reported mood scores vary from the mean mood score for this subject and compute an average variance. That average variance will be compared to the “depression score” of seven that was given earlier. With the thousands of data points from all of the valid subjects in the study, I will have enough evidence to run tests such as a p-value test which will tell me if there is a significant correlation between daily mood variation and depression in first year medical interns. The example given above was just one of thousands as each valid subject can provide up to four valid data points, one for each quarterly survey throughout their medical internship.
The P-value test I will use to prove if my findings are statistically significant or not will allow me to compute how much valid evidence I have against the null hypothesis, which in theory would “disprove” or provide evidence against my hypothesis (P Values). I must be aware of the logical fallacy of “correlation proves causation” when running this test as it is not always the case that when two values are statistically significant to each other that they are caused by each other. I plan to compute my P-value test for my project using the statistical program of R which I considered myself advanced in. Another statistical test that I can use to look for significance in my data is called an ANOVA test. An ANOVA test is basically and analysis of variance which tests the differences between two mean values as I am doing in my analysis. This test will again either build support for my hypothesis or in turn support my null hypothesis.
Girard, Donald E. et al. “The Internship—A Prospective Investigation of Emotions and Attitudes.” Western Journal of Medicine 144.1 (1986): 93–98. Print.
“INTERN HEALTH STUDY.” The Sen Lab – Intern Health Study – University of Michigan, www.srijan-sen-lab.com/intern-health-study.
“P Values.” Prospective, Retrospective, Case-Control, Cohort Studies – StatsDirect, www.statsdirect.com/help/basics/p_values.htm.
Simon, Gregory E. “Social and Economic Burden of Mood Disorders.” Biological Psychiatry, vol. 54, no. 3, 2003, pp. 208–215., doi:10.1016/s0006-3223(03)00420-7.
Uliana, R L, et al. “Mood Changes during the Internship.” Academic Medicine, vol. 59, no. 2, 1984, pp. 118–23., doi:10.1097/00001888-198402000-00008./