Depression and social factors

The USA seems to be the world leader in both incidence and prevalence of major depression, and if anything, the rate is increasing.

In an Op-ED piece in today's New York Times, "Our Great Depression", Andrew Solomon argues that "We need a network of depression centers, much like the cancer centers established in the 1970s." He says:

DEPRESSION is the leading cause of disability worldwide, according to the World Health Organization. It costs more in treatment and lost productivity than anything but heart disease. Suicide is the 11th most common cause of death in the United States, claiming 30,000 lives each year...

Following this model, the National Institute of Mental Health should coordinate and subsidize a national network of depression centers, ideally based at research universities with good hospitals and departments devoted to the subject.

The University of Michigan, host to the country’s first national depression center, which opened its doors last month, has been a pioneer in this regard. More than 135 experts on depression and bipolar disorder will collaborate there, about half of them psychiatrists. The center has a large clinical treatment program and a genetic database that will house samples from tens of thousands of depressed and bipolar patients. It is sponsoring social and biological research and pressing for policy initiatives related to mental illness.

And finally adds "(Full disclosure: my father is the chief executive of a pharmaceutical company that manufactures antidepressants.)" His facts may be correct, but he is not an unbiased observer. And, the U of Michigan depression center certainly supports Cognitive Behavioral Therapy as much it supports pharmaceutical "solutions," so it is not just a thinly-disguised retail outlet for the largest company in Ann Arbor, Pfizer.

Still, while it is clear that "psychosocial factors" such as depression, isolation, and social support have a dramatic predictive value on the outcomes of "medical" disorders, such as cardiovascular disease, it is less clear to what extent depression is itself largely predicted by, or in some causal loop with these other social factors.

(See "Depression, Isolation, Social Support, and Cardiovascular Disease in Older Adults" by Heather M. Arthur, Journal of Cardiovascular Nursing, Vob 21, No. 55, pp S2-S7 for some links into the literature on the former subject.)

A different viewpoint can be found in literature off the continent, that is less supported by the pharmaceutical industry. Here's an example from the National Medical Journal of India
2006 Jul-Aug;19(4):218-20.

The cultures of depression.

Jacob, KS

Department of Psychiatry, Christian Medical College, Vellore 632002, Tamil Nadu, India. ksjacob@cmcvellore.ac.in

Diverse frameworks, models and 'cultures' of depression have been postulated and promoted by psychiatrists, the pharmaceutical industry, general practitioners, primary care psychiatrists and the general population.

Psychiatrists and the pharmaceutical industry endorse the medical model while general practitioners and the public subscribe to social and psychological frameworks. [emphasis added]

These models are partial truths and should be viewed as complementary rather than competitive, some more valid in a specific context than others. The issues that need to be resolved include: (i) reexamination of the validity of the psychiatric diagnosis of depression in the primary care context; (ii) a review of the adequacy of a single label of depression to describe the diverse human context of distress; (iii) acknowledging the problems of using a symptom checklist in diagnosing depression; (iv) recognizing the need for psychosocial diagnostic formulations which clearly state the context, personality factors, acute and chronic stress and coping; (iv) highlighting the fact that antidepressant medication should be reserved for severe forms of distress; (v) re-emphasizing the need to manage stress and alter coping strategies in the treatment of people with such presentations; (vi) de-emphasizing medicalization of all forms of personal and social distress; (vii) focusing on other underlying causes of human misery including poverty, unmet needs and lack of rights. Clinically, there is a need to look beyond symptoms and explore personality, life events, situational difficulties and coping strategies in order to comprehensively evaluate the role of vulnerability, personality factors and stress in the causation of depression.

Possibly, however, we have simply run into the largest single reason to be considering systems thinking - namely, the occurrence of feedback in models of causation.

Standard statistical techniques are fine at dealing with open-loop causality, where A "causes" B, or B causes A, and there is a clearly defined start and end point. The General Linear Model covers that reasonably nicely.

But, as soon as you close that loop, so that A causes B which in turn causes A, that model breaks down. This behavior (a feedback loop) is very common in engineering, and no big deal, but it remains not only perplexing, but almost heretical in the epidemiological community. Even the mention of "psychosocial factors" for medical disorders causes tempers to flare and voices to be raised. The battles go on between arguments such as "bullets cause death" versus "guns cause death" versus "angry people who just happen to have guns at hand cause death" versus "bad economic and political situations cause massive unemployement and unhappiness and anger, which ultimately express themselves in gunshots which cause death."

Still, it seems a reasonable hypothesis to me that social factors, such as isolation and loneliness and lack of social support, result in depression; and, then that depression results in further actions or non-actions that increase isolation and lack of support; and, etc. in a mutually reinforcing feedback loop.

This is "hard to study" in the sense that people don't have desktop software that lets them compute such things as a "p-value" to distinguish whether they are being too credible, or not credible enough when looking at this possible causal loop to explain observational data.

The lack of such software is, of course, precisely the type of gap that the R21 research request for proposals I mentioned in earlier posts is designed to address. (I'm available to work on such a project if there are others also interested in a joint proposal.)

Why does this matter? It matters because it can completely change the interventions required to address the problem. If depression is largely an internal phenomenon, caused by genetics and bad wiring in the brain, that leads to one type of intervention - drugs and CBT. If depression is largely a social phenomenon, related to the well-documented collapse in social interaction documented by Putnam and the group at Duke, then personal intervention will simply deal with symptoms, and result in an ever growing prevalence of drug-dependent victims of social dysfunction - precisely the observation we find about the USA today.

In the latter case, what we need to address is why people are losing the ability to make friends, to keep friends, and to be a friend -- because it is that low-level breakdown that is emerging on a national scale as an epidemic of "depression."

=============
The Duke study is "Social Isolation in America: Changes in core discussion networks over two decades" by Miller McPherson, Lynn Smith-Lovin and Matthew E. Brashears, American Sociological Reviews , (2006), vol 71, June (p 353-375)

Putnam's famous book is Bowling Alone: The Collapse and Revival of American Community by Robert D. Putnam (New York , Simon and Schuster, 2000).
As that site says,

In a groundbreaking book based on vast new data, Putnam shows how we have become increasingly disconnected from family, friends, neighbors, and our democratic structures-- and how we may reconnect.

Putnam warns that our stock of social capital - the very fabric of our connections with each other, has plummeted, impoverishing our lives and communities. Putnam draws on evidence including nearly 500,000 interviews over the last quarter century to show that we sign fewer petitions, belong to fewer organizations that meet, know our neighbors less, meet with friends less frequently, and even socialize with our families less often. We're even bowling alone. More Americans are bowling than ever before, but they are not bowling in leagues. Putnam shows how changes in work, family structure, age, suburban life, television, computers, women's roles and other factors have contributed to this decline.

( originally published 11/16/06 on cscwteam.blogspot.com)

Heading upstream

There's a very basic concept in Public Health known as "going upstream". The cartoon above illustrates the concept. (if you "click" on the picture it should zoom up to a bigger size.)

Imagine our hero, Tim, sees smoke coming up over the mountain, but he cannot see the source because the mountain is in the way. Say the smoke is killing the crops and Tim wants to "fix" the problem. Where should he go to start looking for the solution?

He could head towards the largest amount of smoke, to the right.
He could head towards the "center" of the problem, directly above.
He could head towards the "worst problem area" or densest smoke, to the upper left.
or
He could follow the smoke "upstream", going around the mountain or possibly over it, until he finds the "source" of the smoke.

I relate to this problem. I was in Edmonton once, visiting, and went to the top of a high rise building to catch the view. We saw all this distant smoke and asked where it was coming from. They said, "Oh, that's from a forest fire in the Rocky Mountains, about 45 miles west from here." So, we got in the car and headed west and went to fight the fire, 50 miles up a dirt logging road from Revelstoke. I'll describe our narrow midnight escape someday.

But, the point is, it is not really true that "Where there's smoke, there's fire." Many people seem to take that much too literally, and head for the densest smoke to look for the fire. Others head for the "center" of the visible problem, and others head for the largest amount of smoke.

In Public Health, we're taught to forget all that, sigh, pack a bag, and head "upstream" to locate the actual source of the problem. Often the source is not visible from where we are.

So, whether it's cancer along the Mississippi river, or developmental problems from lead paint poisoning, or gunshot wounds in the Emergency Room, we follow the Toyota Way and ask "Why?" at least five times - the same way you always got in trouble with your parents when they told you to do something.

For example - Why are so many children getting poisoned by old lead-based paint? Because the paint is peeling off and hasn't been replaced.
Why?
Because they live in terrible housing that's falling apart and neglected.
why?
Because they're poor and the poor are exploited and no one seems to care. Because despite tremendous technology, we can't make decent housing for $1000. Because despite amazing science we can't make companies and jobs that seem able to stay alive and in business. Because the people who could help don't realize there is a need, or are overwhelmed with how large the need is. Because the people who live there don't realize they could get subsidized housing in a much better place and don't know how to "sign onto our website and register for housing now!"

Why?

Now, you're getting into culture and how we distribute resources and education, and how we help or don't help each other, and how we respond to need by hiding the problem and pushing it out of our backyard into someone else's, instead of fixing what's wrong.

The Toyota Way really emphasizes that problems need to be brought to the surface, and made visible, so they don't fester and result in bad results later. Here's a view out of the window of where I'm currently writing this. Can you spot a "hiding" place and see what's happening here?


A huge pile of trash has built up just around the corner and out of sight of the main road.

In any Toyota plant, or anywhere near it, you would not find such a thing. They find they get better results if they deal with problems as they arise, instead of letting them stack up until the total pile becomes so overwhelming that no one wants to even think about it anymore.

Well, I hear a reply, that's because everyone is overwhelmed and stressed-out these days and no one has TIME to deal with "other people's problems."

Why?

This is actually a puzzling problem, related to multi-level depression or something. The "poor" in this country are poor at $10,000 a year, versus $200/year in India or China, if that. I think the figure is that something like a billion people earn less than $1 per day on this planet.

Why?

What's the most intriguing to me is that people in the US seem so fragmented and often unwilling to help each other out, or be helped, even when there are many really good-hearted people who are trying to help.

Or, even when the problem becomes desperate. A family about to lose their home because the mortgage payment just doubled on their fancy new loan would rather lose the home than try to have a second family move in and share the space and share the mortgage payment.

Why?

Because people just don't know how to get along with each other and things always turn bad.

Why?

After easily 5,000 years of written history, why is it that people haven't yet figured out how to get along with each other? If this is a big deal, here, in poverty, in Iraq, why isn't THAT what we study in school, from kindergarten through PhD level work, instead of algebra and physics?

"Because we need all this science and technology to save us from the mess we've made of things here."

Umm... Isn't the dependence or science and technology and the rejection of "learning how to get along" precisely the reason WHY we just spent $1,000,000,000,000 on the post-9/11 "homeland security" and war? That would have bought a lot of houses. Isn't the failure of management and labor to talk one of the big reasons GM lost its lead in the auto business and had to layoff hundreds of thousands of workers?

Well, for "cultural reasons" learning to get along is not a high priority.

Why?

In my book, it keeps coming back to this. We have what appear to be "technical" or "production" or "cost effectiveness" or "safety" problems, and they appear to be intractable, unsolvable by anything we can do. Then we find that "anything we can do" excludes the one thing that seems like it WOULD help, namely, putting a lot of resources into understanding how people should work together, relate, overcome conflict, and fix each other's roofs.

Why?

And that is precisely the point of the "Health, Behavior, and Society" focus on the role that "culture" and "distal factors" play on the visible immediate problems in front of us.

Don't look at the smoke. Go find the fire. Put the fire out, and the smoke will stop.

One last thought - some people argue that this kind of reasoning is no good because it doesn't involve mathematics. They've somehow deified the idea that there is such a thing as rigorous qualitative reasoning. I'm against sloppy thinking, sure.

But I've had more math than most people in this discussion. I've had 6 years of calculus, quantum mechanics, general relativity, statistical thermodynamics, etc. I taught financial modeling to MBA's.

Too often, the request for more math is an effort to avoid doing something that you already know you should be doing. We know enough now, with no more math at all, to know that a root cause of most of the mess we're in is that we don't know how to live with each other and work together. If we could solve that single problem, most of the rest would just dissolve, like a pearl necklace with the thread pulled out.

But, for those who insist on math, and are so deeply rooted in the culture of worship of "hard sciences", be of good cheer. I'll give you the math and you can be happy. You just may need to "come to the mountain" a little and learn about feedback control loops and all the rest of the non-linear, loop-based mathematics that YOU, dear you know who I mean, have been avoiding hoping that everything would fall neatly into linear causality, open paths with clear starts and finishes, and the General Linear Model and its grip on research.

And, in fact, with a little Laplace Transform wizardry, even those dreaded loops will flatten out and you can use your existing math and solve the problems with STATA -- even though the basic assumptions about unbiased estimators won't be met. Who ever checks those anyway?

So, enough. I'm off to breakfast.