Best Practices for Computerized Adaptive Tests To Address Health Disparities

In Part 2 of this video, Robert Gibbons, Blum-Riese Professor of Biostatistics; Professor, Public Health Sciences; Director, Center for Health Statistics, University of Chicago, Chicago, Illinois, discusses best practices for clinicians to use computerized adaptive tests (CAT) to address health disparities based on his recent research, which examined to concordance of CAT and traditional measures of depression among low-income Black and Latina women.  

In the previous part 1, Professor Gibbons discussed the most significant findings from his research.

Read the Transcript:

Were any of the outcomes different than you expected?

Professor Gibbons:  The question is where some of these outcomes unexpected? Were they different than what we had hypothesized? Several of the findings were unexpected. Perhaps the most unexpected is the finding that there is almost 10-fold increase in the rate of perinatal depression among Black and Latina women with a history of depression.

Women with a history of depression had a 15-fold increase in the rate of perinatal anxiety. However, a history of anxiety only conveyed a 40% increase in perinatal anxiety. A history of depression appears to drive both perinatal depression and perinatal anxiety. The magnitude of these effects of prior depression are much, much larger than expected.

The role of the history of depression on perinatal anxiety was also unexpected, and extremely important for identifying those women who are going to be at the highest risk for developing perinatal depression and anxiety.

What are best practices for clinicians to use CAT to address health disparities?

Professor Gibbons:  The question is what are the best practices for clinicians to use computerized adaptive testing in routine practice in the clinic to address health disparities, which are so important?

The take-home message of our study is that it is of critical importance for clinicians to evaluate Latina women for perinatal depression and anxiety during the first trimester overall and in particular, those Latina women with a history of depression, even more, important than a history of anxiety.

By contrast, the risk of depression and anxiety for low-income Black women is much more homogeneous over the trimester and postpartum. The CAT-MH, the computerized adaptive testing mental health, is an excellent tool for conducting these assessments.

Our study has demonstrated their importance in low-income Black and Latina women. Another advantage of these tools is that they can be administered in or out of the clinic, and do not require a trained clinician to administer or interpret results.

This makes them ideally suited for primary care and OB-GYN practices, where clinicians may not feel comfortable doing structured clinical mental health diagnostic interviews or trying to characterize the severity of these disorders. It makes them transportable to large-scale population screening assessment and measurement-based care.

Are you considering any further research?

Professor Gibbons:  The question is am I considering any further research? I'm an academic. I always consider further research. I wake up in the middle of the night thinking about further research. The answer is always. As an example, Maggie Alegria from Harvard University.

Maggie is one of our nation's leading global health disparities researchers, particularly in the area of mental health. She and I are studying additional sources of bias in mental health measurement in minority immigrant populations. Latinx and Chinese patients coming to emergency departments and taking our tests in Spanish and in Chinese.

One of our early findings is that minority immigrants are less likely to endorse extreme categories, severe impairment for a particular symptom. What this does is it leads to an underestimate of their depression and their anxiety scores.

This is true for any mental health measurement, either traditional or fixed length or adaptive. But because the CAT-MH is based on a statistical model of measurement, it is much easier to remove the effect of this bias statistically and provide unbiased estimates of the real severity of special minority immigrant populations. That's one example.

Do you have any final thoughts?

Professor Gibbons:  In terms of final thoughts, there are numerous exciting applications of this technology that we and others have already been exploring. Screening and measurement, of course, in primary care, and emergency departments is a critical application.

We are doing studies throughout the United States and throughout the world in these settings. Let me give you a few interesting examples. At UCLA, we have offered screening and measurement-based care to 85,000 students, undergrads, over the past four years using our adaptive test for depression, anxiety, and suicidality.

These tests are administered completely remotely on computers or Smartphones. On the basis of these tests, we've triaged these students who need treatment either into Internet-based cognitive behavior therapy and peer counseling, these are for those students who have mild to moderate depression or anxiety and no particular suicide risk, or clinic or emergency department services based on clinical severity and suicide risk. Yes, we are screening college students at UCLA for suicide risk remotely.

They can take these tests at three o'clock in the morning. What we've done to create a safety net is to have a direct real-time link of the results of these tests to national suicide hotlines that will then reach out to the student to make a referral of what they should be doing and conduct a further assessment.

This is possible. It is possible to screen people for depression, anxiety, and suicide risk in a safe way. The effectiveness of these treatments has been remarkable both for an Internet-based cognitive behavior therapy with peer support. We've trained over 400 college students as peer support students.

This is work with Michelle Craske and Nelson Freimer at UCLA. It has been so successful that we're now rolling this out to the 2.1 million students in the state of California Community Colleges. Again, this idea of providing mental health screening and assessment in a global way to entire populations is important.

SAMHSA and Research Triangle Institute are conducting a $30 million national prevalence study of mental health and substance use disorder.

They're using the CAT-MH at the advice of the National Academy of Sciences as a first-stage screener to determine who will receive a full-structured clinical interview that lasts an hour to two hours and requires a trained clinician.

This is to obtain incidents estimates of various mental health and substance use disorders. We are dramatically improving the efficiency of this national survey by using the CAT-MH as a first-stage screener to target which people in the US population should receive a full structure clinical interview.

Our colleagues and friends at the Veterans Administration are using our tools for the prediction of future suicidal behavior at completion, and screening and measurement of PTSD.

Lisa Brenner at the VA and her group with us have published a new approach to the assessment, both for screening and measurement of PTSD based on computerized adaptive testing and computerized adaptive diagnosis.

As of this week, all VA clinics in the United States now have access to the CAT-MH directly through their electronic health record system. Again, a wonderful example of population mental health.

Finally, I want to compliment my clinical colleagues at the University of Illinois in the University of Iowa, for pursuing this very important work in terms of looking at disparities in low-income Black and Latina populations.

I'll also want to acknowledge the over 15 years of continuous funding, which is continuing, I think my grandchildren will be funded by this line of investigation from the National Institute of Mental Health that made all of this possible.

To my clinical co-investigators, David Kupfer, Ellen Frank, David Brent, Paul Peel Conus, and Ben Leahy, and my statistical collaborator, David Weiss. This is the future of mental health measurement. Thank you.