For Best Results, Trust in God and Use Large Sample Sizes

Earlier this week, Crommunist pointed me to this article in The Atlantic, People Who Believe in God are More Responsive to Treatment of Depression. And ohhh boy do I have feelings and crinkly skeptic eyebrows.

[As a side note, I’d like to thank The Atlantic for putting a link to the original research in their footnotes. Nice touch, actually including the science you’re talking about. Everyone else, take heed.]

While I wasn’t particularly impressed with the article in the Atlantic, there were some good things to be said about the original research (found here if you have access. At the very least you should see the abstract and summary).

First, a word of caution. This is a prospective study. In other words, a study that followed people over time, in real time. That means that there’s a small sample size (159 participants), because it’s hard to find funding for an following a huge group of people. Retrospective studies can offer the ability to have a large group of participants, but also mean that you can quite easily miss important variables. As a result, prospective studies are considered to yield data a cut above that of retrospective. The tradeoff is, the smaller sample size can cause the occasional news site to get super excited about what are pretty conditional results.

What did they do?

Okay, so what actually happened? Researches went to McLean Hospital (an excellent place to get psychiatric care, I hear), and talked to patients in their day program. These are clients who aren’t living in the hospital, but are coming to receive services all day and returning to their homes at night. (This sort of thing is also called partial hospitalization).

All participants were given a battery of statistically sound tests to measure congregational support, religious affiliation, depression, self-harm, and belief in God, as well as psychological well-being.

Of Note: No God was specified, the participants were simply asked “Do you believe in God?” and told to circle a number from 1 to 5, where 1 is defined as “not at all” and 5 is defined as “a strong sense of belief”.

No part of treatment was changed–the researchers were just interested in who seemed to respond best to treatment.

Who did they look at?

The methods section is a little fuzzy on whether they approached 159 patients and two refused from that point, or whether 159 was the total number of participants. Regardless, the sample size wasn’t terrible large.

On one hand, this is a group of people in a very well controlled environment–a hospital–one of the few ways you can have control over environmental factors during the study. On the other, anywhere between many and most–those are the scientific terms–of people with depression and anxiety won’t be hospitalized. So you’ve got a slice of the population with very serious manifestations of these disorders, possibly with presentations that aren’t responsive to traditional coping mechanisms or therapy. [Keep that last part in mind, it comes up again.]

On the other, you have a  very particular subset of people with these disorders. Partial hospitalization programs take your entire day–that means you’r either taking time off from your job or not working. While insurance usually covers partial hospitalization, you have to actually have insurance. That means that what we’re looking at here is a subset of people with access to this kind of care.

And what does the methods section say about the demographics?

 Most participants were Caucasian (83.6%) and single (61.4%), and there were a high number of college graduates (45.3%). Impairment in the sample was high in that 56% of participants were unemployed, and all patients presented with global assessment of functioning (GAF) scores of <45, representing serious symptoms/impairment.

…Sounds a lot like what we’d expect.

What was found?

Even after controlling for age and gender (Women and those who are older are more likely to believe in God), belief in god was related to having better outcomes in the program. Not related to outcome: religious affiliation. You didn’t have to believe a certain kind of God, you just had to believe there was a god out there.

Belief in God was also related to having greater support from the religious community, (I’m shocked, I tell you, SHOCKED), but not related to having a greater ability to regulate emotions.

So should you convert to vaguely-unspecific-god-belief? Probably not. Researchers actually concluded something entirely different than the title of the article would lead you to believe. It appears that there was an important mediating variable: belief in the success of the treatment process. People who believed in God were significantly more likely to place their trust in the ability of the program to help them.

Belief in treatment credibility and expectancy of success across levels of belief in God.
Belief in treatment credibility and expectancy of success across levels of belief in God.

And that, not belief in God, strikes me as the more important link. People who believe in the the ability of a type of therapy to help them are far more likely to see results of the therapy than those who are skeptical. And it appears that those who have faith in a deity are also more likely to believe in the authority of their psychologists and psychiatrists–perhaps an expected result?

Some words on race as it influences this research:

Eighty-three percent of these participants were Caucasian. In the United States, even when controlling for income, mental health spending for outpatient care (aka partial hospitalization) for Latino and black consumers is only 60-75% of that for whites. Add to that the long history psychiatry and medcine has of unacceptable medical experiements on minorities. Testing psychotropic drugs on black and Hispanic children, overseas pharmacological experiments, and on and on. Is it so surprising that we find significant differences in trust of physicians in ethnic minorities?

What does this mean with respect to this study? Research has shown that African Americans are more likely to avoid seeking early treatment, like outpatient care for their depression. (Research is contradictory on this phenomenon in Latino populations.*)

Black and hispanic rates of admission to inpatient hospitalization care are, as a result, much higher. Without the ability to trust in or access preventative health care, many more are going to need emergency services, including involuntary commitment, which can be an unpleasant process; first responders are often untrained in compassionate care of psychiatric patients. And so the cycle of distrust repeats. Which means fewer minority participants in studies like these, which means care tailored to non-minority clients.

General conclusions:

This is…an interesting study.

It’s not badly done by any means. But it is a small sample size, and not very generalize-able to the population of people with depression. I’d like to see more research that examines a cross-section of people in inpatient, outpatient, and therapeutic settings, with a careful eye to the influence of trust in psychiatry. Until then, I’m willing to reach a cautionary conclusion that for white participants who can and need access outpatient programs, belief in god is linked to belief in the promised results of treatment, which leads to better outcomes, with the caveat that the populations of People Who Can Access Outpatient Care and People Who Trust Outpatient Care In The First Place overlap heavily.

 * I know I’m actually just talking about two specific minorities in this section. Unfortunately, there’s basically no other research available. If someone say, had extra money to throw at research into other minorities and psychiatric care, they should do that posthaste. 

For Best Results, Trust in God and Use Large Sample Sizes

6 thoughts on “For Best Results, Trust in God and Use Large Sample Sizes

  1. K

    This jibes with my experience, but not for the implied reason.

    Several years ago I was institutionalized after a suicide attempt. One of the first questions asked was about my religious beliefs. When the doctor assigned to me learned that I was agnostic, he was quite clear that he was worried that my (lack of faith) would lead me to harm myself again, and in the end that was one of the reasons given for me being committed (although strangely it didn’t appear on his written assessment.) I ended up having to file an appeal (which took two weeks.) The day before my appeal was to be heard, he declared that I wasn’t a danger to myself anymore and I was released.

    So yeah – this comes as no surprise to me – if the doctors know you don’t believe, you get treated differently.

    1. 1.1

      Yes! That’s a side I left out to–if your lack of religion is something your mental health professionals view negatively, it can impact treatment.

      Something else you didn’t mention, but that also is worth saying is that therapeutic alliance matters: if you don’t connect well to your therapist or counselor, you are less likely to have a positive result. Religious people may find a professional with religious symbols or who mentions god to be more like them or likeable than a non-religious person. For instance, if my therapist doesn’t mention God, I don’t know if it’s because they’re non-religious or because they’re just not mentioning it. But if Sally Catholic sees a religious saying on Amy Therapist’s wall, she has an immediate touchstone.

  2. 2

    It really would be both interesting and informative for this sort of study to be performed again specifically in non-white populations to parse out belief in deity vs. belief in treatment. Many non-white populations still have a higher % of religiosity than white populations, but for the reasons you mentioned they may be less likely to trust the medical establishment. This would provide a great opportunity to parse out where the benefits are coming from when you have participants where these two factors (deity belief vs. doctor belief) do not correlate well. Unfortunately, finding the patients may not be easy, since (as you mentioned) they are less likely to be found in this sort of clinical setting, and also those that would sign up for the study may be more trusting of the medical establishment (giving you a biased sample).

  3. pdk

    159 is a small sample size? Your claims of statistical deficiency are embarrassingly qualitative. Please make the effort to cash out such accusations with power analysis.

  4. 4

    This post is really interesting. I heard about something along these lines in an intro psych class a couple months ago. I wish I had asked for a reference, it couldn’t have been the same study, but I wonder if there is another study out there showing something similar.

    Looking back at various notes/texts all I can find is a reference to “Psychotherapy and Religious Values” by Allen Bergin, which is less of a study and more of an op-ed about how clinical psychology should be centred on religion.

Leave a Reply

Your email address will not be published. Required fields are marked *