Sunday, December 28, 2008

Burnout and Suicidal Ideation among U.S. Medical Students

I came across this article in the Annals a few months ago. I have been tapped to lead an electronic journal club in January, and this is the article that I will be using. If anyone has any input that they want included in the discussion, lets hear it.
Linky
I know the authors from medical school and have a great deal of respect for them. I exciteded to give it a critical read and will post my thoughts here soon. I will say though, that I know these are huge issues for some medical students.

1 comment:

DrJaymez said...

I stumbled across this article last fall in the Annals. I think it is relevant for us, at minimum, because we have the most face time with students who may have more on their minds than the finer points of distinguishing between anemia of chronic disease and iron deficiency anemia on morning rounds.

It is a study of prevalence.

Funding came from an education grant from the Mayo Clinic, presumably a non-biased non-profit. We're not testing some drug or marketable test here.

This is a web based survey based study of medical students at all levels of training at 7 different medical schools (Mayo, U of Washington, U of Chicago, U of Minnesota, and U of Alabama in 2006). UCSD and Uniformed Services University of the Health Sciences also participated in 2007. Students who responded in both years were included in longitudinal analysis. They used established instruments to measure burnout (Maslach Burnout Inventory) and quality of life (Medical Outcomes Study Short Form-8). The article gives normal scores for the general population and medical professionals. Depression was evaluated with the validated 2-item Primary Care Evaluation of Mental disorders. Suicidality was evaluated by simply asking direct questions, listed in the article.

The authors hypothesized that burnout was related to suicidal ideation.

Statistical analysis was a linear regression evaluating for independent associations of the independent variables with suicidal ideation.

I can't easily think of a way to apply our handy 2x2 table in this situation because there isn't really a gold standard evaluation for these conditions, and this is more a study of prevalence than of diagnosis.

From my perspective, this was as well conducted as a survey trial for mental health data can be. The response rate was 52%, which strikes me as not bad for medical students.

There are a number of results reported. I will summarize those that I think are interesting.
Ever considered suicide: 25%
Ever made an attempt: 2%
Considered suicide in the last year: 11%
47% screened positive for depressive symptoms on an instrument that would predict that 50% of these have major depression (using what I think is Bayesian logic).
Emotional Exhaustion, Depersonalization, Personal Accomplishment, Mental Quality of Life, and Positive Depression Screen are all associated with suicidal ideation. For the burnout scale elements, this relationship, interestingly, is incremental, in that the worse your burnout scores are, the more likely you are to have suicidal ideation.
When all this was run through the multivariate regression controlling for the variables associated with suicidal ideation on univariate analysis, only the burnout domains and mental quality of life fell out as associated with suicidal ideation.
The really interesting data, in my mind, is the longitudinal data. Those students that were no longer burned out were less likely to report suicidal ideation, which makes sense. Furthermore, those that developed burnout now had an increased suicidal ideation. All of these numbers were statistically significant.

Discussion Points

The authors have actually demonstrated both a dose-response relationship of burnout to Suicidal Ideation as well as demonstrating that if burnout goes away, so does suicidality. The implication here is causality, which the authors imply but won't confirm. The further implication is that if we could identify and treat burnout in medical students that suicidal ideation would therefore decrease.

I feel that this is as strong as a study of this nature can be. This is a confidential, diverse, multi-institutional, longitudinal study.

There are some limitations of the study. Many are pointed out by the authors. It is not clear what type of responder bias a study like this gets. Do people respond because their burnout and mental health are so poor that they want to make sure that somebody knows about it, or do they choose not to respond because they are concerned about anonymity, or because they are so “burned out” that they don't do the survey. The authors make the point that even if none of the non-responders were suicidal the suicidality would still be 5.8%. But the more involved calculations related to burnout and its relationship to suicidality are more complicated.

I have more trouble interpreting the somewhat advanced age of the average responders, which was a bit high, 54.9% between 25-30. I'm not sure if there is more maturity associated with likelihood to fill out a survey of this nature. I'm not sure in what way it would skew the results.

I don't know what “we” do at OSU to identify at-risk students. While further investigations are reasonable, as the authors indicate, I certainly think this data is good enough to act on.