• How we tolerate risk

    I’ve been thinking a lot about how we tolerate risk here in the United States, after we literally shut down a major city last week to catch one man. I am in no way saying that it wasn’t an important endeavor. I just think that many people I’ve talked to can only look at the positive outcome of catching him. They ignore the massive implications (socially, politically, economically) of what it took to do so. It’s a debate worth having, and I’m not seeing it as much as I’d like.

    Ironically, at the same time, a friend pointed me to a new study in the BJOG, “Light drinking versus abstinence in pregnancy – behavioural and cognitive outcomes in 7-year-old children: a longitudinal cohort study

    OBJECTIVE: To assess whether light drinking in pregnancy is linked to unfavourable developmental outcomes in children.

    DESIGN: Prospective population-based cohort.


    POPULATION:Ten thousand five hundred and thirty-four 7-year-olds.

    METHODS:Quasi-experimental using propensity score matching (PSM) to compare children born to light (up to 2 units per week) and non-drinkers.

    MAIN OUTCOME MEASURES:Behavioural difficulties rated by parents and teachers; cognitive test scores for reading, maths and spatial skills.

    RESULTS:Ordinary least squares (OLS) regression and PSM analyses are presented. For behavioural difficulties, unadjusted estimates for percentage standard deviation (SD) score differences ranged from 2 to 14%. On adjustment for potential confounders, differences were attenuated, with a loss of statistical significance, except for teacher-rated boys’ difficulties. For boys, parent-rated behavioural difficulties: unadjusted, -11.5; OLS, -4.3; PSM, -6.8; teacher-rated behavioural difficulties: unadjusted, -13.9; OLS, -9.6; PSM, -10.8. For girls, parent-rated behavioural difficulties: unadjusted, -9.6; OLS, -2.9; PSM, -4.5; teacher-rated behavioural difficulties: unadjusted, -2.4; OLS, 4.9; PSM, 3.9. For cognitive test scores, unadjusted estimates for differences ranged between 12 and 21% of an SD score for reading, maths and spatial skills. After adjustment for potential confounders, estimates were reduced, but remained statistically significantly different for reading and for spatial skills in boys. For boys, reading: unadjusted, 20.9; OLS, 8.3; PSM, 7.3; maths: unadjusted, 14.7; OLS, 5.0; PSM, 6.5; spatial skills: unadjusted, 16.2; OLS, 7.6; PSM, 8.1. For girls, reading: unadjusted, 11.6; OLS, -0.3; PSM, -0.5; maths: unadjusted, 12.9; OLS, 4.3; PSM, 3.9; spatial skills: unadjusted, 16.2; OLS, 7.7; PSM, 6.4.

    CONCLUSION:The findings suggest that light drinking during pregnancy is not linked to developmental problems in mid-childhood. These findings support current UK Department of Health guidelines on drinking during pregnancy.

    Basically, this study looked at a cohort of more than 10,000 seven year olds to see if there was an association between light drinking in pregnancy and later developmental issues in children. They didn’t find any.

    Let’s take a pause here and let me state clearly that I am not advocating that women drink during pregnancy. I’m just asking us to think about risk, and how we instinctively respond to it. OK?

    When I was a resident, one of the world’s experts in fetal alcohol syndrome told me that if a fetus is exposed to a certain amount of alcohol at a certain point in development, there appears to be a significantly increased risk of developing problems. But, he explained, we don’t know for sure what that dose of alcohol is, or at what stage of development the danger zone exists. We do know some things. It’s not very late in pregnancy. We used to use alcohol in pretty hefty doses to prevent preterm labor. Not so much anymore, but its use periodically late in the third trimester did not result in increased cases of fetal alcohol syndrome.

    So what do we tell women in light of this vague, but real, danger? In the United States, we tell them no alcohol in pregnancy. Ever. In any amounts. Here, from the NLM and NIH:

    Pregnant women are strongly urged not to drink alcohol during pregnancy.

    Drinking alcohol while you are pregnant has been shown to cause harm to a baby inside the womb and may lead to long-term medical problems in the child after birth…

    Women who are pregnant or who are trying to get pregnant should avoid drinking any amount of alcohol. The only way to prevent fetal alcohol syndrome is to not drink alcohol during pregnancy.

    If you did not know you were pregnant and drank alcohol, stop drinking as soon as you find out. While it is unlikely that the occasional drink you took before finding out you were pregnant will harm your baby, the sooner you stop drinking alcohol, the healthier your baby will be.

    Try replacing alcoholic drinks with their nonalcoholic counterparts: for example, you might opt for a nonalcoholic pina colada instead of the real thing.

    If you cannot control your drinking, avoid eating or drinking around people who are drinking alcohol.

    Here is the NIAAA:

    Drinking Can Hurt Your Baby

    When you are pregnant, your baby grows inside you. Everything you eat and drink while you are pregnant affects your baby. If you drink alcohol, it can hurt your baby’s growth. Your baby may have physical and behavioral problems that can last for the rest of his or her life. Children born with the most serious problems caused by alcohol have fetal alcohol syndrome.

    Children with fetal alcohol syndrome may:

    • Be born small.
    • Have problems eating and sleeping.
    • Have problems seeing and hearing.
    • Have trouble following directions and learning how to do simple things.
    • Have trouble paying attention and learning in school.
    • Need special teachers and schools.
    • Have trouble getting along with others and controlling their behavior.
    • Need medical care all their lives.

    Here are some questions you may have about alcohol and drinking while you are pregnant.

    1. Can I drink alcohol if I am pregnant?

    No. Do not drink alcohol when you are pregnant. Why? Because when you drink alcohol, so does your baby. Think about it. Everything you drink, your baby also drinks.

    2. Is any kind of alcohol safe to drink during pregnancy?

    No. Drinking any kind of alcohol when you are pregnant can hurt your baby. Alcoholic drinks are beer, wine, wine coolers, liquor, or mixed drinks. A glass of wine, a can of beer, and a mixed drink all have about the same amount of alcohol.

    Mind you, none of this is wrong. It’s abstinence at its finest. Drink no alcohol – none at all – and there is no risk of fetal alcohol syndrome. But go back to the British study I began this piece with. Look at the conclusion. They state that the idea that “light drinking is not harmful” is consistent with UK Department of Health guidelines. So I looked those up:

    The UK Chief Medical Officers’ advice to women is:

    ‘Women who are pregnant or trying to conceive should avoid alcohol altogether.  However, if they do choose to drink, to minimise the risk to the baby, we recommend they should not drink more than 1-2 units once or twice a week and should not get drunk.’

    Again, I’m not advocating that pregnant women go out and drink. There is no doubt that no alcohol is the safest way to avoid issues. But sometimes it feels like this is the way we approach risk in the US, even in health care. We think in absolutes, in terms of avoiding all harms and dismissing potential benefits instead of debating the relative contributions of both. It’s easy to dismiss alcohol as being all harms, as it’s certainly not providing any benefit to the fetus. But some studies have shown that small amounts of alcohol can be beneficial to health, and I know many, many people who would argue that it contributes positively to life in general. Moreover, this is how many issues are approached. “Never eat red meat”. “Never let kids watch TV or play video games”. “Never eat soft cheese while pregnant”. “Never fail to screen for disease”. And so on.

    Is this a United States thing? Or am I just wired differently? I’m just not sure that the way we react to potential harms is the best approach. This includes, by the way, our response to terrorism.


    • 1) The US does not seem to do nuance well. We appear to prefer simplistic bright lines. No alcohol is simpler than 1-2 drinks per week.

      2) For some reason, a few people killed by a planted bomb is worse than 15-20 killed by a plant explosion is worse than close to 100/day killed by gun violence (although about 2/3 of that is suicides).

      Looking at numbers killed and resources devoted in various categories over the past decades should be enough to show that cost-benefit analysis is another thing we don’t do well.

      • I live in the Boston area, so I’ve been thinking about these things over the past few weeks a lot too (though not as well or cleverly as Aaron Carroll).

        Foosian, I think you make absolutely true points, but I think there’s more there to be said:

        1) I don’t think it makes sense to say the US doesn’t “do nuance well”. That seems to me like saying, “man, Canadians aren’t good at metaphors” because their National Building Code doesn’t have enough flowery language. A limit of 2 drinks per week is no more complicated than a limit of 0 drinks per week. If someone is able to stick to a limit of 0 drinks per week, but they are unable to stick to a limit of 1 drink per week during pregnancy, my guess is the issue has more to do with the relationship to alcohol, not the understanding of 0 and 1. My completely inexpert interpretation is that the bodies that issue these guidelines (for drinking during pregnancy, screening for illness, etc.) are for some reason institutionally very risk averse. We, as consumers of their guidelines, pick them up and run with them ….. which I guess brings us right back to your point that we don’t do nuance well ….. well played. I guess the question then is why are these institutions a bit more risk averse than others?

        2) I think everyone agrees that deaths from a bombing, an explosion, a car accident, or whatever, are no better or worse than one another. They are all tragic, and each of those people were priceless, irreplaceable, and loved. My question is, would any of us really want to live in a world where we addressed issues purely based upon their relative risk? In such a world, none of us would ever drive faster than 15mph, eat a piece of cake ever again, etc. But we would take a look at the past 15 years in the US, and in the aftermath of this bombing say, “no big deal, your odds of being killed or injured are infinitesimal … we’ll work the case and catch them eventually … odds are only a few more people will die so don’t worry. The majority of our resources continue to be applied to traffic and dietary enforcement.” That might be the rational approach, but is it the human one? Is it the one you would want?

        I’m not saying we can’t be WAYYY smarter in our approach to risk, but it seems like a lot of people have been (legitimately) questioning the approach to capturing the second Boston bombing suspect as a sign that they somehow “won” or that it wasn’t an appropriate response given the level of risk, to which I say a few things:

        – There’s no such thing as winning and losing in these scenarios. I’m not even sure why the concept is so pervasive. If we all went about our business on Friday, did we “win”? If we won, would that prevent additional violence? Who’s keeping score? I personally don’t think the whole paradigm of terrorism/response as a competition even makes any sense whatsoever. These things happen, they are reprehensible, we address them as we address them, well or poorly. There are many and varied measures of defining “well” or “poorly”. Nobody wins; not us, and certainly not the perpetrators.

        – I’ve seen a lot of comparisons with Israel or Afghanistan over the past few days as an example to follow in how to keep going in the face of terrorism and violence. While their toughness and perseverance is phenomenal, is it desirable? Do we want to live in a state of affairs where attacks are so frequent that we become toughened to their practical and emotional impact? As has been pointed out, we don’t react this way to most gun violence, but instead of hardening ourselves against acts of terror in the same way, shouldn’t we instead ask how we can use some of that risk prioritization discussed above to reduce acts of gun (or any) violence to the point where they too become appalling in their rarity? If this seems a little “pie in the sky” … I agree, it is.

        – In practical terms, if everything goes according to plan on your drive to work, you don’t get in a car accident. For a criminal attempting to blow things up, if everything goes according to plan, things get blown up. I think that makes at least some difference in our response, and at least to some level I think that our more vigorous response to the latter is justified (the devil is in the details of course; what level is justified?). Frankly, I get a little angrier at a person intent upon committing terrorist acts near my neighborhood than I do at my car when its tire blows out. My car may be riskier technically, but it isn’t trying to hurt me or the events and places I hold dear. I think a bit of anger in such a situation is normal. Of course (to contradict my own point from above), the only time a terrorist REALLY wins is when they get us to turn our anger at a single perpetrator into hatred of a place, a people, or a way of live.

        – Finally, can we all just agree to absolve ourselves and blame the media? They hooked me from Thursday night on, and by the time they were done I was sick with guilt over being sucked into the spectacle of it all, and terrified to open my own closets.

      • I wouldn’t say this is anything remotely unique to the U.S. As evidence I present the entire political history of Europe from about 1900 to 1950.

    • I wonder (statistically) how many lives and/or limbs were saved by the Boston lockdown due to fewer traffic accidents (if any).

    • The debate on the extent of the shutdown isn’t happening partly because it would be seen as libertarian, and hence lots of eye-rolling on the right and left. The successful outcome reinforces this.

      Interesting link between the two though. Excessive intrusion and excessive caution are different things, though they overlap a lot. I wouldn’t say that Americans are any worse at this kind of thing in terms of belief, and nor do you (I’m Cdn) have a more egregious history of institutional follow-through on these biases. Invoking a control that is likely to fail (soda ban) is very different from an actually enforced control though.

      A social norm of martial law is a bad thing. A social norm of over-reacting to protect fetal development, and which has almost zero negative impact on ETOH sales, productivity, etc. is on balance not so bad. If one’s preoccupation is primarily about widespread mild irrationality then maybe they are equivalent, but that is a frustrating and unpragmatic place to be.

    • I do not think that Americans are worse than others. Consider the European’s attitude toward GM food and pesticides. The German move to end nuclear power after Fukashima seems a bit rash.

    • As a native Bostonian (30 year resident) and MIT grad who (30 years later) still has friends in Boston, I don’t think the reaction was over the top at all. It seemed exactly right. The Boston Marathon is an important part of Boston culture and we really needed to capture the perpetrators and determine the extent of the problem, which wasn’t know at the time of the lockdown. Boston does just fine with everything coming to a halt, as in the 1978 blizzard and the more recent ones. Things were much nastier in Tokyo, when all commuter trains stopped on a Friday afternoon (March 11, 2011) and there were (literally) millions of people without a place to sleep.