Thursday, February 03, 2011

Surgeon General Botches Breastfeeding Call to Action: A Suboptimal Study in the National Spotlight

 Surgeon General Regina Benjamin (courtesy AP)
“I believe that we as a nation are beginning to see a shift in how we think and talk about breastfeeding,” said Dr. Benjamin. “With this ‘Call to Action,’ I am urging everyone to help make breastfeeding easier.”

A shift!  Amen! It's about time we change our talk about breastfeeding.   

Bless dear Dr. Benjamin, the Surgeon General, Her Health Excellency who declared breastfeeding a national priority in a press release last week. 

Yes, let's please talk it out, Dr. Bennie.

We can agree breast milk is the best choice for building tiny brains and bodies, even the formula makers admit it.  I breastfed all three of my babies and would do it again in that hypothetical younger-me universe where I spend a couple years in Paris eating baguettes and painting large canvases. 

However, the top doc and I would disagree over the degree of these health bennies, especially those loose links involving the health outcomes years if not decades past the actual "laying on of the lips" so to speak.

I can forgive a lapse or two here and there but Dr. Bennie highlights a particularly poor piece of what I can only describe as propaganda "research" in her press release, uses it as evidence of the high health care costs attached to not breastfeeding enough. 

In fact, it's the only study she cites in the press release (The Burden of Suboptimal Breastfeeding in the United States): 

A study published last year in the journal Pediatrics estimated that the nation would save $13 billion per year in health care and other costs if 90 percent of U.S. babies were exclusively breastfed for six months. Dr. Benjamin added that, by providing accommodations for nursing mothers, employers can reduce their company’s health care costs and lower their absenteeism and turnover rates.

Now I don't know if Dr. Bennie's read the actual journal article but if she had she'd find this:

The authors, Bartick and Reinhold, relied on a 2007 government report reviewing research on the health benefits.  We can debate the accuracy of the review - it was based on data from the 80s and 90s - but let's excuse that for now.  We'll ignore the accuracy of the findings linking 10 health conditions to (not!) breastfeeding.  While we're in a forgiving mood we'll also ignore the authors' ridiculous assumption that 90% of mothers can and want to breastfeed exclusively for 6 months.

Bartick and Reinhold then computed the health care costs for treating these conditions ranging from the  minor and fleeting like ear infections and stomach upset (the strongest links) to the more serious like childhood obesity, Type 1 diabetes even SIDS.  We'll not quibble too much there, we got a  bigger battle ahead.  Fine. We'll accept all those estimates. Whatever you say, Bart and Reiny!

Here's the rub, the ruse really.  It's the 900 dead babies.

If you remember the headlines this study captured you'll recall the dead babies.  The authors concluded over 900 baby deaths a year could be prevented if their mommas complied with the breastfeeding recommendations. 

900 dead babies!!!  Really?

Do we know breastfeeding actually prevents "premature death?" Good question.  The answer is pretty much no, no we don't know if babies actually die from "suboptimal breastfeeding".   

Nor is there much in the way of empirical numbers to help us figure that out.  But that didn't stop the researchers.  Somehow they computed how many babies die each year due to a variety of conditions, in fact, this was the first study to do so - in a peer-reviewed journal, the premiere Pediatrics journal - and reported "premature deaths" for a variety of conditions.  So it was an important study not only because it was the first large-scale analysis of the health costs associated with not breastfeeding - but also because it estimated baby deaths.

Despite setting a precedent for future research - not to mention future government Calls to Action, the authors never revealed how they calculated the annual baby deaths.  This is a very large hole.  Huge. 

This omission makes me very suspicious and very irritated.  The journal editors and reviewers should have demanded to see these calculations and it's anyone's guess why they didn't insist on them in the actual text.  Me suspects the estimates are jerry-rigged.  In other words highly speculative and thus inaccurate.  It's not only aggravating and sloppy but disingenuous.

And that's not the worst of it. 

The authors took those 900 annual deaths and estimated the cost of their premature deaths.  Basically how much those dead babies would have earned over their lifetimes, roughly $10 million each.  That's 9 billion a year.  Then Bartick and Reinhold rolled that 9 billion into their big fat estimate of how much not breastfeeding costs us each year. 

The whopping $13 billion in the headlines last spring and the Surgeon General's press release - over 70% of it isn't even a health care cost.  Most of it involves the lifetime earnings of all those dead babies Bartick and Reinhold somehow calculated die each year from not enough breast milk. 

 Forgive me for thinking we were talking about the medical costs of not breastfeeding.  When most of us hear about health care costs we imagine doctor bills, prescription medication, etc.  Not forgone earnings.

Bartick and Reinhold seriously inflated the numbers and did so in a very deceptive manner.  Not only did they not reveal how they computed the actual number of annual dead babies but they vaguely refer to the costs of premature death that I naively assumed meant the colossal end-of-life bills.  They include a line or two of economic mumbo-jumbo that I only realized meant lost lifetime earnings after I'd read the paper they cited.  Now that's pretty darned ballsy.  You claim 13 billion and hundreds of dead babies but somehow never spell out how you got that many deaths or that you've included not only true health costs but lost earnings of the deceased.  Remarkable. 

I'm not sure I've read anything this disappointing in years - at least not in a prominent journal from researchers at a prominent institution (Harvard).  

I expect a little braggadocio, the speculations in the conclusions, the exaggerations.  I know researchers have to make a good argument for their data, they gotta have a good "story" and stretch the results from time to time. We all know they sometimes poke around in the data for those significant results.  I get it, been there, done that.

But this study, "The Burden of Suboptimal Breastfeeding" is not only suboptimal but sneaky, deceptive, disingenuous and it's got the weight of Harvard and the US Surgeon General behind it. 

When I first read it a few months back I worried it would become the bench mark for health authorities.  So much so I named it one of the year's worst pieces of research in my column Naked Data at Parent Dish.  Voila, now the suboptimal study officially entered into the public health canon and will haunt me for years to come. It's there, folks, taken as matter of scientific fact.

With no questions asked.  Amazing.

Sweet Jesus, just when I thought I could finally stop writing and thinking about breast milk.  

Was Dr. Bennie aware of these lifetime wages snuck into that 13 billion? In her press release she refers to the $13 billion as "health care costs and other costs" - so if she did know, how sneaky of the Surgeon General. If she didn't, why the hell isn't our top doc doing her homework?  
So yes, Dr. Bennie, a shift is needed.  Let's start talking honestly about breastfeeding for a change.

8 comments:

Anonymous said...

I am so glad that you are posting about this! I read the report and found the "Excess Risk" tables suspect. What do you think about them? Although the paragraph above talks about the AHRQ study not "necessarily representing cauality," I think people are just going to look at the % and that's it!

Polly Palumbo, Ph.D. said...

Hey Anony! Yes those "excess risks" sound and look so scary. Those health outcomes come directly from the 2007 AHRQ report and were used in the Suboptimal Study. So for the ear infections the excess risk of "100" means exclusively formula-fed babies have 100% more infections, in other words, twice as many. The risks reflect the "odds ratios" in the studies.

The kicker, of course, is that other more recent studies that have found no differences between BFing and FFing (formula) were not included in any of these estimates. The AHRQ report doesn't include newer and thus better data and the suboptimal study drew directly from the AHRQ despite a variety of null result studies. The American Academy of Pediatrics has even doubted the link to SIDS.

While I'm still ranting, I didn't even mention the "psychosocial" benefits of breastfeeding which amount to anecdotal musings. The cogntive benefits - there is no solid evidence to suport it. The recent study showing BF-fed boys do better in school - not such a winner. Those claims are built on a house of cards.

Becca said...

I'm actually pretty comfortable with mentioning lost-earnings costs (it's not uncommon when people are writing the significance portion of grants on malaria, so I've seen it done a lot; it's not a 'healthcare cost' but it is part of the 'economic burden'), but I think they totally should have separated it and been explicit.
What bugs me more is that if you can't figure out how we are coming up with those 900 deaths, something is very, very wrong.

Polly Palumbo, Ph.D. said...

Hey Becca, thanks for the input! I hear you. How about the authors should never have included it in the first place - because it's mostly a figment of their imagination!

But yes, if we really had reason to assume the deaths were attributable to BFing then okay, separate those earning out and be honest about it - and don't include it in the headlines. In all the media reports and the Surgeon General's Call to Arms I never once heard mention of these deaths or "lifetime lost wages" costs. Most of us aren't economists so it's not what we think of as health care costs.

Melissa Bartick said...

Thanks for all your attention to the Bartick/Reinhold paper. In fact, the paper clearly shows how we arrived at the numbers of deaths, by taking the mortality rate for each relevant disease, multiplied by the number of excess cases, and crosschecked with national mortality data. We also made it clear that the mortality costs are different from the other costs. All of this is n table 3, Figure 1, and Figure 2. By definition, SIDS is a condition with a 100% mortality rate, which would be obvious to the readers of this journal, thus the death rate for SIDS was not spelled out.

The costs attributed to premature death are not from lost wages. They are a standard economic construct called the revealed preference method, which is similar to a willingness to pay model. If they were only lost wages, the figure would be much lower. The US government uses this model in determining costs and policy-- it was big news when George Bush's EPA cut its cost of death down by a $1million to about $7.6 million a few years back.

The AHRQ report used a compendium of all well-done studies done up to that point. Notably, it did not find an odds ratio in favor of breastfeeding for everything the authors studied. They used studies that both did and didn't show an effect from breastfeeding.

There are not a totality of studies published since 2007 refuting all the AHRQ findings from the 10 dieases we studied.

You may not agree with the figures and mortality rates we came up with, but the figures were certainly not zero, as was previously presumed. We were very conservative in all our estimates, and even had to leave out things known to be expensive, such cost of type 2 diabetes in children, and mortality from childhood obesity. The $13 billion is a conservative estimate by any stretch. It also does not include the maternal health and mortality costs, which are likely considerable.

-Melissa Bartick and Arnold Reinhold

The Fearless Formula Feeder said...

Bravo.

This is a perfect example of what Jules Law detailed in his oft-cited paper many years back. Breastfeeding "science" is often more like breastfeeding folklore. One overstatement gets taken as gospel and repeated, like a farked-up game of telephone... sigh.

I'm so glad you are out there, Polly, explaining these things so clearly so that the rest of us can understand how we are being manipulated by research.

Polly Palumbo, Ph.D. said...

Hey Drs. Bartick and Reinhold, thanks for weighing in. Yes, I did catch the mortality rates but I wasn't sure how you used them to calculate the number of premature deaths.

Nor was I clear what they really were. So I guess from your comments the mortality rates reflect the portion of individuals with a certain disease that die of the disease. I wasn't sure if it was as simple as that. As opposed to something like the odds ratio for the actual diseases i.e. a rate reflecting observed differences in death among the breastfed and non-breastfed groups, in other words, rates derived from studies looking at deaths among breastfed v. nonbreast-fed babies.

So, for readers here, the premature death estimates were calculated based on the mortality rates for each disease (i.e. the risk of death - regardless of breastfeeding) then multiplied by the # of excess cases (the # prevented if more women breastfeed). So there, mystery solved.

As for the costs, I think most people would be very surprised to learn the 13 billion is not primarily "health" or "medical" costs. I don't think anyone would have argued the costs of not breastfeeding were zero. The study might not have been done before, the evidence not collected and collated. So yes, we will have to agree to disagree on the extent of the costs.

Anonymous said...

Have you seen this? Thought of this exact post and your comments back and forth with Drs. Bartick and Reinhold.

http://stats.org/stories/2011/breastfeeding_risk_sids_jul11.html