The JAMA article about the large increase in urinary Bisphenol A (BPA) levels after eating canned soup caught my eye because my husband and I eat a lot of canned soup.
[JAMA. 2011 Nov 23;306(20):2218-20. Canned soup consumption and urinary bisphenol A: a randomized crossover trial. Carwile JL, Ye X, Zhou X, Calafat AM, Michels KB. Source: Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts, USA.]
Naturally I was concerned, but my epidemiology background makes me a hearty skeptic when it comes to chemical/toxin “scares” from food.
My motto is “Show Me the Data”. It is SO DURN HARD to perform a really good study of the isolated effects of isolated chemicals from isolated foods ON HUMANS NOT RATS that I rarely alter my normal food intake to avoid specific compounds unless a strong body of evidence suggests there’s really something to it.
I also know just how easy it is to get a really crappy epidemiologic study published (ahem, Andrew Wakefield?).
So, here’s my project: look into this BPA thing. People take teams of researchers and spend months doing this. This is time I do not have. So, I’ll do this in parts. Part I is literature search.
SO PUBMED, SHOW ME THE DATA:
1. Search string #1: BPA w/o limits = Results 2807
2807 isn’t a manageable #. I knew it wouldn’t be, but I was curious. So, I need to refine my search.
2. Search string #2: BPA or Bisphenol A, limits: humans, english, search title only = Results 537
This still isn’t manageable for 1 person, unfortunately. I.E. it would take about 9 hours to screen all these abstracts. But, if I refine the search further, my search becomes more specific and less sensitive. I might miss an important article. So I’ll limit to full text only next, then screen the titles.
3. Search string #3: Search string #2 but with full text limit = Results 498
That isn’t too helpful. Limiting to free full text would be very helpful (i.e. 154), but then I know I’ll miss some important things. Ok, so let’s see how long it will take me to screen titles.
4. Screening titles, using timer to measure time to screen one screen of titles = Results Page 1 = 2 min, Page 2 = 1.30 Page 3 = 1.20 Page 4 = 1 min
So, I can guesstimate that it would take me about 34 minutes at 80 seconds a page to screen the titles. Out of the first 80 I screened, I chose 30 titles to review. If that is a steady ratio, I’ll pick 187 abstracts.
BUT then I have to read the abstracts to screen them (time estimate = 2 min/abstract = 6 hours), and THEN I’d have to read the actual articles. Next would be the joy of getting the articles (some I’ll have access to through my job but I am anticipating quite a few I would have to inter-library-loan, ILL). I don’t think that an estimate of 30%-50% needed through ILL is unreasonable based on my experience with the library so far. Each ILL costs ~ $15-20.
- 6 hours to read abstracts.
- If I go on to choose 50% of those abstracts for review = 93
- Cost to get full text articles if 30% require ILL = $558
- Time to read 93 articles, [no clue really, big shot in the dark here] ? 10 min? = 15 hours
There’s NO WAY I can take the time or money to do this.
Luckily as I am scrolling, I can pass over the articles on how to measure BPA and the non-human studies that didn’t get weeded out by checking the “humans only” option.
But, perhaps more efficiently I can pick up other recent review articles about the human effects of BPA (and actually read the full JAMA article, which I have not done yet). The danger in this is knowing whether the people who wrote those review articles have enough knowledge of epidemiology to accurately evaluate the quality of the studies. Also here’s a WARNING about reading only the discussion section of journal articles: these ARE NOT SYSTEMATIC REVIEWS. So, if you are “summarizing” the data, it is only natural to bias your summary to support your findings. Since humans are humans, I don’t want to trust the JAMA article discussion without looking around myself.
See how difficult this can be? It is SO much easier to glance through someone else’s review article and take their word for it that they evaluated this data objectively.
In the “good old days”, one was supposed to be able to trust the “name” or “reputation” of a journal and the authors. Well, we know now that even the bastions of medical scientific writing (names like JAMA or NEJM) can let some data slip through that it shouldn’t. [There was an article on screening for lung cancer using CT scans that used historical controls to make comparisons, published in either JAMA or NEJM a few years ago. While I was in grad school at UNC, one of my epidemiology journal clubs evaluated this article and we had serious issues with the strength and validity of their findings. Sadly, I cannot find the citation within a a reasonable period of time to write this blog, so I'll have to get back to you.]
Who do you trust? I don’t know. I can only tell you what I do, and what I do changes depending on the time I have in which to do it. Since I don’t intend to write up this exercise for anyone but myself and the readers of this blog, I’ll probably be less rigorous than I would like to be.
5. Search string #4″ Search string #3 with limits to free full text and limited to meta-analysis and review articles only, Results = 19!! Yay!!, kinda (see qualifications about about limiting it to this tight a search).
You know what? Two of these articles are EXACTLY THE KIND OF STORY I was looking for. I was amazed the first article ended up Freely Disseminated in Pub Med Central because it is written in 1st person by a mother and a scientist.
PLoS Biol. 2007 Jul;5(7):e200. Epub 2007 Jul 17. Babies, bottles, and bisphenol A: the story of a scientist-mother. Quitmeyer A, Roberts R. S. Department of Biology, Ursinus College, Collegeville, Pennsylvania, United States of America.
In this article, an undergraduate biology student and her professor (who is also a scientist and mother) discuss the history of BPA, some of the studies of BPA effects, the legislation surrounding BPA (up to the time of publication, 2007).
“Many animal studies focus on the effect of BPA exposure during fetal development . . . A similar correlation to human development is plausible. Indeed, BPA has been found in the bloodstream, placenta, cord blood, and fetal blood of humans at levels that are within the range studied in many of the animal models .”
Be very clear you understand their argument and rationale for their decisions:
- Studies of animal exposure indicate adverse effects for animal offspring.
- The same levels that are shown to be bad in rats (and other animals) have been found in humans.
- Therefore, if I believe that what is bad for a rat is bad for a human, then I believe that BPA is bad for human offspring.
They also discuss the few human studies that exist:
“Of the few human epidemiological studies, one revealed a relationship between BPA exposure and repeated miscarriage . Additionally, BPA causes a human breast cancer cell line to proliferate, indicating that estrogen-sensitive tissues and cells in the body may react similarly “
They decide to apply the effects of animal studies to humans, incorporate the human studies they reviewed, and thus they believe that current human levels of BPA exposure are too high and can be harmful.
I think it is important not to arbitrarily apply rat studies to humans. I think we should use animal studies to warn us of potential harmful human effects, then try to see if those effects do indeed show up in humans. Epidemiology allows us to address these questions in humans without subjecting us to trials where we ingest a potentially toxic substance. The caveat is that good epidemiology studies of chemicals are hard to come by–there is the potential for SO MUCH BIAS that one must be very careful when designing this type of study. You can get completely wacko results and needlessly scare a lot of people (ahem, Andrew Wakefield?).
Since the above article is now 4 years old, I want to find some more HUMAN reviews to help me form my own opinion.
Next I will turn to an article published in Germany :
Leibniz Research Centre for Working Environment and Human Factors (IfADo), University of Dortmund, Dortmund, Germany, Institute of Environmental Toxicology, University of Halle, Halle/Saale, Germany, Federal Institute for Occupational Safety and Health, Dortmund, German, Research and Development, Merck Serono, Darmstadt, Germany, Dr. Lilienblum Consulting Toxicology LiCoTox, Hemmingen/Han, Germany, Global Early Development, Bayer Schering Pharma AG, Berlin, Germany, Bavarian Health and Food Safety Authority, Munich, Germany, Lower Saxony Governmental Institute of Public Health, Hannover, Germany, Federal Institute for Risk Assessment (BfR), Berlin, Germany
I read the abstract for this, but I haven’t read the paper yet. Basically this was a panel in Germany that was tasked with deciding if BPA is good or bad for you, and if bad, how much is OK. I’m going to hope that since they were a panel, and that they took dedicated time to address the same question I’m addressing, that their conclusions will be valid. However, as pointed out in article #1, I will have to decide if politics, lobbying, and business interests play at all into this panel’s decision.
Since I have spent close to two hours on this exercise today and will now have to go back to finishing my clinic notes, working on my presentation for research conference this Friday, re-writing a paper for publication and working on my next research project, it is time to stop my BPA exploration until next time.
Stay tuned . . .test Filed under Uncategorized | Comments (2)