Do 12-step programs for addiction treatment work?
Are 12-step programs like Alcoholics Anonymous and Narcotics Anonymous effective treatments for addiction? That long-time dispute has just popped up again, prompted mostly by an Atlantic article with the click-worthy title “The Irrationality of Alcoholics Anonymous.”
Gabrielle Glaser’s central argument is that there are other addiction therapies that work, but the 12-step programs are just not based on science. It’s a reasonable point historically. For one thing, many 12-step studies have tended to ignore a crucial problem in research design: selection bias. It’s reasonable to think that people in AA, NA, etc joined because they were especially motivated to quit. Members are probably not a random sample of addicts.
But Jesse Singal’s riposte to Glaser at Science of Us argues semi-persuasively that there may be recent data showing that 12-step programs can be effective.
Glaser emphasized the equivocal nature of the science collected in the 2006 meta-analysis in a Cochrane review, which found “no experimental studies unequivocally demonstrated the effectiveness of AA or [12-step] approaches for reducing alcohol dependence or problems.” Singal seems to interpret that as saying the programs were not effective, but of course that isn’t what the Cochrane folks said (or, it appears, what Glaser was saying either.) They said effectiveness had yet to be demonstrated. The quality of previous studies was often questionable, and it was also hard to compare them.
New evidence that AA works?
Singal quotes doc John Kelly thus: “There’s quite a bit of evidence now, actually, that’s shown that AA works.” Kelly is an addiction specialist at Mass General and Harvard Med and one of the authors of a new Cochrane review to be published in August.
But the Singal post has problems too. It does not, for instance, take up drug therapies for addiction: methadone, naltrexone, etc. That leaves me in the dark about their effectiveness. I gather that 12-step programs in general don’t approve in principle of substituting one drug for another and so don’t want their participants to be using “therapeutic” drugs.
In talking about the newly revealed success rate, Kelly seems to be emphasizing research not on 12-steps themselves but on 12-step facilitation programs. Acronymed TSF, these programs sound something like prep courses for AA. Kelly claims these (plus a 12-step) have been shown to have up to a 20% advantage over one alternative treatment, cognitive behavioral therapy.
The lead author of the 2006 Cochrane Review, Stanford doc Keith Humphreys, will also be an author on the upcoming revisionist paper. He posted on his updated views about a strengthened scientific basis for AA etc last June at WonkBlog. This post, however, also was about the effectiveness of TSF, not 12-step programs by themselves with no introductory TSF boot camp.
Are 12-step facilitation programs the new AA?
But since TSF programs are not themselves regular 12-step sessions, I find these claims confusing. Is TSF a new, improved AA? Can measuring the effects of TSF plus a 12-step sojurn legitimately serve as what the clinical trials folk call “surrogate endpoints?” Is a TSF required for 12-step success, or does it simply increase the success rate? (Or, of course, neither, but both Kelly and Humphreys appear to rule that out.)
What Humphreys concludes is that TSF followed by 12-step participation is as effective as any other professional therapy. That may be good news, but it is also not an evaluation of the plain old AA, NA, etc. that millions of hopeful recoverers count on.
Also, a not unimportant point: the TSF pre-interventions require a therapist. Classic TSF is a three-month program. Humphreys also praises a mini-TSF, a “brief, structured introduction to AA,” without defining “brief.” A three-month stint run by a therapist will cost at least several hundred dollars, and even a brief one will cost something.
Contrast that with the ubiquitous–nearby and frequent–12-step sessions, which are free. Although you are supposed to contribute for the coffee. You can’t beat that for accessibility.
Glaser points out that the question of what works has taken on new urgency because Obamacare requires insurers and Medicaid to pay for substance abuse treatment. I wonder if that fact might help speed up reliable efficacy research. It would be nice to spend that money on stuff that works. Patients trying to get clean would probably prefer that too. Most of them.
Or, assuming that it turns out there is good evidence that 12-step programs can work, with or without TSF prep sessions, is it possible to do a better job of moving that information into physicians’ offices? Not to mention their brains? Along with the data on other approaches like cognitive behavior therapy and meds?
We are, we are being told, entering the age of Personalized Medicine. Surely time to lay out the smorgasbord of addiction therapies with attached price tags and data on what works.
Does breastfeeding increase adult intelligence and income?
The media loved the idea that breastfeeding improves intelligence. And, even better, that a breastfed baby earns more in adulthood. Who wouldn’t?
There have been other studies saying breastfeeding raises IQ (and some that say not). The New England Journal of Medicine‘s JWatch has made its past comments on these studies available free. But this new study, from Brazil and published in The Lancet, attracted so much attention because it went on for more than 30 years, which made it seem particularly persuasive.
Even venues where one might expect critical consideration reported with little analysis. The New York Times health blog Well, for instance. Vox’s Julia Belluz, an unusually skeptical eye, was pretty accepting of the Brazil study too. (Although a few days later she was writing about why you shouldn’t believe what you hear about exciting new medical studies, a series of cautionary tales for journos if there ever was one.)
There is much to praise in this ambitious long-term study. Spare a thought just for the amount of work involved over decades. At the Guardian’s Brain Flapping blog, Dean Burnett’s critical take begins with praise: “It’s a well-designed study which tries to take into account as many variable and confounding factors that could distort the results as possible.” Burnett’s critique leans heavily on the fact that IQ was measured with the always-controversial Wechsler Adult Intelligence Scale.
He also wonders how this widely-praised study will increase the already heavy pressure on new mothers to breastfeed. This concern he bolstered with personal experience: his wife’s breastfeeding instructor, a health care worker, vilified women who considered formula feeding.
What’s wrong with the Brazil study of breastfeeding and intelligence?
Adrienne LaFrance based her Atlantic criticism largely on details of the study–especially that so many of the study subjects enrolled in 1982–5914 of them–were, as population researchers say, “lost to follow-up.” After 30 years, only 3493 were still available. That’s perhaps understandable in such a long-term study, but it does raise questions about the data’s validity.
She points out other difficulties too. Accounts of breastfeeding practices were collected years after the births, and were based on self-reports by the mothers. Self-reports are intrinsically subject to question.
“And even if breastfed babies really do grow up to have higher IQs and make more money, how do we know it’s because they were breastfed?” LaFrance asks. Although social class differences in breastfeeding practices seem to be less important in Brazil than in many other countries, it still is the case that nursing mothers tend to be better off and more educated than mothers who don’t (or can’t) breastfeed.
There were other reported adult differences between those breastfed as babies and those who were not. The researchers compared babies who were breastfed for a month or less with those who were breastfed for a year or more. The differences were statistically significant, but not enormous. The income difference amounted to a little over $100 per month at today’s exchange rate, and as with IQ can perhaps be explained by other advantages in the breastfed. Breastfed children stayed in school longer, but less than a year longer. Also, IQ differences amounted to less than 4 points on average.
LaFrance concludes, “And though breastfeeding may play a role in how a kid turns out, 30 years of parenting surely has more to do with what kind of adult a baby becomes.”