Teaching Psychiatric Residents Research Literacy: Use of Existing Epidemiological Databases
To the Editor: Given the need for clinicians to keep up with the pace of scientific innovation, and contribute to the formulation of new testable hypotheses, there is a current push for training programs to foster research interest and literacy among psychiatric residents.
Some of the most utilized approaches are research seminars and journal clubs, where articles are dissected and methodological issues debated. We report on the use of publicly available epidemiological datasets to foster research literacy and methodology among psychiatric residents. Given the current availability of large epidemiological databases—such as the National Comorbidity Survey and its replication—it is possible to include residents in direct exercises of data analysis and hands-on research activities. Multiple publicly available datasets are online today, such as the National Health Interview Surveys (http://www.cdc.gov/nchs/nhis.htm) and the National Health and Nutrition Examination Surveys (http://www.cdc.gov/nchs/nhanes.htm). Another resource is the large archive of computerized social science data at the Inter-University Consortium for Political and Social Research, located within the Institute for Social Research at the University of Michigan (http://www.icpsr.umich.edu).
At the Medical College of Georgia Adult Psychiatry Residency Program, we have implemented a program that has so far produced three resident-led research posters in national meetings and two articles submitted to publication, along with heightened research interest among the residents. Some of the advantages of this approach include faster IRB approval, given the public and nonidentifiable nature of these datasets, and a hands-on approach to epidemiology and research questions, leading to a direct experience with research design, methodological problems, and analytical solutions. We have structured this approach in a 12-step process:
1. Finding a resident-specific research interest: each resident is prompted to think of an area of psychiatry in which he or she has a particular interest.
2. Finding a dataset and perusing the codebook (we have thus far used the NCS-R).
3. Asking an answerable scientific question such as, “Are persons with history of seasonal allergies, compared to those without it, more likely to report having suicidal ideation and history of attempts?”
4. Describing the data and types of variables such as categorical versus continuous measures.
5. Summarizing data: presenting the demographics and other main variables.
6. Measuring associations: odds ratio and relative risk, along with discussion about confidence intervals.
7. Looking for possible confounding variables and biases: discussing the role of gender distribution and age.
8. Dealing with confounding: stratification and regression.
9. Discussing biological plausibility and possible causal paths.
10. Inference and interpretation: arguing for clinical and public health implications and future research.
11. Presenting research: creating tables and graphs, preparing posters and papers.
12. Authorship issues: measuring contribution from each participant, submitting manuscripts, and dealing with reviewers’ comments.
A requirement in adopting this program is the presence of a faculty member skilled with data analysis. However, if a faculty member with epidemiology skills is not available within the psychiatry department, liaisons may be established with departments of epidemiology and biostatistics. Statistical programs suitable for such analysis need to be available as well.
At the end of this process the residents involved reported a greater sense of comfort around research questions. The use of existing databases also leads to concrete results and presentations at meetings and publication of papers, which benefits the residents directly. As they move on to independent practice they can better appreciate the work involved in researching new therapies and be better prepared to critically assess the scientific literature. A variety of available datasets would foster this approach, and we should support NIH policy in making them publicly available. We believe this format can be added to the current tools used to prepare physician scientists in psychiatry.
At the time of submission, the author disclosed no competing interests.