The graphs tell the story. Now I want to fit this story into a bigger framework so it all makes sense again.

Paul Alper points us to these graphs:

Pretty stunning. I mean, really stunning. Why are we just hearing about this now, given that the pattern is a decade old?

And what’s this: “Data for the U.S. ends in 2007”? Huh?

Also, it’s surprising how high the rates were for Japan, Italy, and Germany in the 1970s. Whassup with that?

The whole thing is tough to understand; I just don’t know how to think about it.

One lesson from all of this, I think, is that our public space (newspapers, TV, etc.) doesn’t have enough experts on demography and public health; thus these sorts of basic statistics come as a surprise to us. Consider: compared to most people, I’m an expert on demography and public health, but these graphs came as a complete surprise to me.

In all seriousness, I think our public discussion space needs fewer doctors and fewer economists, and more demographers and public health experts. I have no problem with doctors and economists, considered individually as experts; there just seems to be an imbalance in aggregate exposure, comparing these professions to others with relevant expertise in the same questions.

29 thoughts on “The graphs tell the story. Now I want to fit this story into a bigger framework so it all makes sense again.

  1. Ah, but economics imperialism assures that economist-demographers and economist-public health experts are all you need. See Deaton and Case, Emily Oster, etc. etc.

    Please note that I’m kidding. This economist agrees with Andrew.

  2. The graphs tell the story.

    Really, I think you know better than this? Where is this data coming from? What problems may it have? Looking for the source I see a note at the bottom that says:

    Some scholars believe that the recent U.S. increase is exaggerated by past undercounting.
    Source: Organization for Economic Cooperation and Development

    I would start there in looking for the problems. And why is that citation so poor? It is almost as if the author does not want us to look too much into the background because it would conflict with his narrative… Even look at the title: “If Americans Love Moms, Why Do We Let Them Die?”

    I think you fell for clickbait.

    • Yes the trend is exaggerated by a change in reporting, but that also means that maternal mortality has probably been understated in the past in the US, and it still looks like US rates are higher than its peers.

      • I don’t know anything about this data in particular (eg whether the trend is only exaggerated or not) but would assume the data from other countries has similar issues. NB: I assumed there were problems exactly like this with the US data before seeing the footnote on the chart, because there pretty much always are such problems that need to be dealt with in some questionable way.

        I’m just surprised Andrew relaxed his skepticism for long enough to write “the graphs tell the story”. They certainly tell a story, but not necessarily the correct one.

  3. I agree with you on the role of public health. The field is underappreciated.

    Here’s a statement that reporting of maternal deaths changed in about 2004

    The 2003 revision of the U.S. Standard Certificate of Death introduced a standard question format with categories designed to utilize additional codes available in ICD–10 for deaths associated with pregnancy, childbirth, and the puerperium. As states revise their certificates, most states are expected to introduce the standard item or replace pre-existing questions (which can vary in wording and format) with the standard item, so that there will be wider adoption of a pregnancy status item across the country and greater standardization of the particular item used.

    A separate pregnancy status item on the death certificate results in the identification of more maternal deaths. Refer to the “Technical Notes” of the forthcoming full report for details (1).

    See https://www.cdc.gov/nchs/data/hestat/finaldeaths04/finaldeaths04.htm.

    See also the discussion at https://www.scientificamerican.com/article/has-maternal-mortality-really-doubled-in-the-u-s/.

    Bob

    • Yes, I was surprised to see that. However both Canada and the USA record infants under 500gm as live births and apparently other OECD courtries do not so this may account for some of the higher rates, but I do not see the two countries are tracking so closely. Common drug use patterns?

  4. The data from these graphs are from the OECD Health Statistics. According to the metadata (available here: http://www.oecd.org/els/health-systems/Table-of-Content-Metadata-OECD-Health-Statistics-2017.pdf), there are breaks in the series for many countries, including the US in 1999 and 2003. I have no specific knowledge about this, but the documentation confirms that no data are available for the US after 2007 with the following text:

    “Maternal mortality data from 2008 onwards are not currently available. The 2003 revision of the U.S. Standard Certificate of Death introduced a checkbox question format with categories to take advantage of additional codes available in ICD-10 for deaths with a connection to pregnancy, childbirth, and the puerperium. As the states of the United States revise their certificates, most states are adopting the checkbox format, resulting in wider adoption of a pregnancy status question nationwide, and greater standardisation of the particular question used. Recent data would be available as soon the majority of states adopt the new protocol and a national estimate would be reliable and statistically conceivable to be calculated.”

    • A couple of papers document this lack of recent reliable data on maternal mortality in the US. This one seems to be particularly relevant and goes beyond by providing estimates for recent years:

      – MacDorman, M. F., Declercq, E., Cabral, H., & Morton, C. (2016). Is the United States Maternal Mortality Rate Increasing? Disentangling trends from measurement issues Short title: U.S. Maternal Mortality Trends. Obstetrics and Gynecology, 128(3), 447–455. http://doi.org/10.1097/AOG.0000000000001556

      Abstract
      Background

      A pregnancy question was added to the U.S. standard death certificate in 2003 to improve ascertainment of maternal deaths. The delayed adoption of this question among states led to data incompatibilities, and impeded accurate trend analysis. Our objectives were to develop methods for trend analysis, and to provide an overview of U.S. maternal mortality trends from 2000–2014.

      Methods

      This observational study analyzed vital statistics maternal mortality data from all U.S. states in relation to the format and year-of-adoption of the pregnancy question. Correction factors were developed to adjust data from before the standard pregnancy question was adopted, to promote accurate trend analysis. Joinpoint regression was used to analyze trends for groups of states with similar pregnancy questions.

      Results

      The estimated maternal mortality rate (per 100,000 live births) for 48 states and Washington D.C. (excluding California and Texas, analyzed separately) increased by 26.6%, from 18.8 in 2000 to 23.8 in 2014. California showed a declining trend, while Texas had a sudden increase in 2011–2012. Analysis of the measurement change suggests that U.S. rates in the early 2000s were higher than previously reported.

      Discussion

      Despite the United Nations Millennium Development Goal for a 75% reduction in maternal mortality by 2015, the estimated maternal mortality rate for 48 states and Washington D.C. increased from 2000–2014, while the international trend was in the opposite direction. There is a need to redouble efforts to prevent maternal deaths and improve maternity care for the 4 million U.S. women giving birth each year.

  5. Those knee-buckling graphs come from within a Nicholas Kristof article, “If American Love Moms, Why Do We Let Them Die?” The graphs themselves have the ironic title, “American Exceptionalism.” Here is an excerpt from Kristof:

    “we’ve structured health care so that motherhood is far more deadly in the United States than in other advanced countries. An American woman is about five times as likely to die in pregnancy or childbirth as a British woman — partly because Britain makes a determined effort to save mothers’ lives, and we don’t.

    Here in Texas, women die from pregnancy at a rate almost unrivaled in the industrialized world. A woman in Texas is about 10 times as likely to die from pregnancy as one in Spain or Sweden, and by all accounts, the health care plans proposed so far by Republicans would make maternal mortality even worse in Texas and across America.”

    To see why he makes these claims, read his article.

  6. This is all quite shocking, but none of it surprising.

    (1) Maternal mortality rates rising

    Not surprising considering our healthcare system disincentivizes people from seeking medical help, and people often have to make horrifying tradeoffs to save money.

    (2) Data ends in 2007

    Not surprising considering that the ruling class only cares about holding onto money and power, fomenting chaos and desperation to maintain control, and thus are going to suppress evidence of this alarming trend. Other dummy reasons may be given but its BS theres no logical or financial reason not to collect such a fundamental data point.

    (3) People aren’t talking about it.

    See (2)

    The ruling class in this country is shockingly cynical and callous.

  7. “And what’s this: “Data for the U.S. ends in 2007”? Huh?”

    Yeah, this is weird. Detailed cause of death data are available with a ~2 year delay (2016 data currently available on CDC Wonder), and birth data will likely be about the same (checks.. okay.. 2015 then). I would think maternal deaths and births are outcomes that are counted pretty well. Maybe there was a definitional change that didn’t work in the context of the analysis?

  8. Kristoff –> likely to be tendentious cherry picking of flimsy statistics. Note the vicious headline to the story: “If Americans Love Moms, Why Do We Let Them Die?” Note the slap at Texas in the third paragraph of the article.

    Health statistics like this are suspect because methods of recording cause of death change over time and they can be very different across countries. For a long time, people thought France had a lower death rate from heart attacks. It turned out to be just a difference in the way the French recorded the cause of death.

    Bingo. 15 seconds of googling turns up a Scientific American article making this point. Kristoff couldn’t be bothered. He had a hate-Texas article to write. (Anti-Texas bigotry appears to be mandatory at the NYT.)

    … the apparent increase in maternal mortality in the U.S. “is almost certainly not a real increase. It’s better detection from the new certificates,” says Robert Anderson, chief of the Mortality Statistics Branch with the CDC’s National Center for Health Statistics. “The numbers are going up but it’s most likely not because women are more likely to die,” he contends.

    https://www.scientificamerican.com/article/has-maternal-mortality-really-doubled-in-the-u-s/

    When trying to understand a comparison of this type between the US and other developed countries, it is often helpful to break down the statistic by race. The higher US rate is often due to a higher rate among blacks.

    Bingo. The same Scientific American article also mentions this. Again, Kristoff couldn’t be bothered. It complicated his anti-US-healthcare narrative.

    One aspect of maternal mortality that has not changed over the years is the extent to which it varies by race. The risk of maternal mortality has remained about three to four times higher among black women than white women during the past six decades.

    • Uh, Terry, I think you missed the point of those articles you cite. The _trend_ may be wrong, but it’s not because Texas doesn’t have extremely high maternal mortality now, it’s because it _also_ had extremely high maternal mortality back when it was thought to be much lower. This suggests that even _more_ mothers have died in the past 15 years than Kristoff said, not fewer!

      • Good point. I should back off.

        My criticism, then, reduces to noting that some, and perhaps all of the apparent increase is due to changes in reporting, and that death-rate comparisons across countries have been wrong in the past.

        Since previous commenters made these points, I haven’t added much.

  9. My impression is that reporting of causes of death in the US varies a lot between different localities.

    https://www.npr.org/2011/02/03/131242432/graphics-how-is-death-investigated-in-your-state gives a little discussion of some differences (although the graph shading and legend don’t seem consistent with each other).

    In Texas, not all counties have medical examiners, so cause of death may be filled out on forms by someone with no medical training. I have heard that this is believed to have made reports of deaths caused by opiate overuse under-reported; it plausibly could also affect reporting of childbirth-related deaths.

    • It looks like Texas from 2010-2011 is a good place to look to better understand these reporting quirks. There was a dramatic and permanent jump of more than 60% in reported maternal deaths in Texas from 2010 to 2011. Sounds very suspicious. It also suggests that reporting issues can make a big difference.

      The rate of maternal mortality in Texas spiked from 18.6 deaths per 100,000 live births in 2010 to more than 30 per 100,000 in 2011 and remained over 30 per 100,000 through 2014, according to a recent study in the medical journal Obstetrics and Gynecology.

      https://www.usatoday.com/story/news/health/2016/09/10/texas-maternal-mortality-rate/90115960/

  10. One additional point: the title of the graph mentions “maternal deaths during childbirth”, but the data is actually about *all* maternal deaths (ICD-10 codes O00 to O99), i.e. it includes deaths related to the pregnancy that occur at different stages of the pregnancy or up to 1 year after birth. You can have a look here to see what is included: http://apps.who.int/classifications/icd10/browse/2016/en#/XV [“The codes included in this chapter are to be used for conditions related to or aggravated by the pregnancy, childbirth or by the puerperium (maternal causes or obstetric causes)”].

  11. Update.

    It appears that U.S. data for maternal deaths is quite bad. It seems to be particularly bad for Texas. And the problems appear to go beyond the addition of the “checkbox” in 2006.

    Odd that Kristoff didn’t mention this.

    Here is an article about a study that looked into the huge jump in Texas maternal death rates in 2011.

    “I guess I was surprised how bad the data was,” said Marian MacDorman, a co-author of the study and a research professor at the University of Maryland’s population research center. For instance, Texas vital records indicate that the mortality rate for expectant and new mothers over the age of 40 rose to 27 times that for younger women, a disparity that MacDorman termed “biologically implausible.”

    Such anomalies in the Texas numbers epitomize the woeful recordkeeping on maternal deaths nationwide. In the United States, where 700 to 900 women die annually from causes tied to pregnancy, maternal mortality rates have been rising for the last two decades as rates in other affluent nations have dropped. At the same time, U.S. data tracking such deaths is so shaky that health authorities haven’t released an official annual count of fatalities, or an official maternal mortality rate, since 2007.

    https://www.texastribune.org/2018/01/04/maternal-deaths-are-increasing-texas-probably-not-much-officials-thoug/

  12. How can Ireland report between zero and 10 deaths per 100,000 live births in any given year?

    Ireland has 10,000 live births per year, so a rate of 10 deaths per 100,000 live births implies one death per year. So what does it mean when Ireland reports 4 deaths per 100,000 live births? Were there 0.4 maternal deaths that year? Perhaps there was one death that year and multiple causes of death were reported, so they allocated 40% of the death to maternity?

    What am I not seeing here?

  13. I did my PhD in a school of public health, with two demographers and one public health specialist on my committee. It’s indeed striking how much higher status economists and physicians are relative to demographers and public health specialists. In fact, the public health specialist on my committee usually refers to himself as a health economist, presumably reaching for status.
    One could argue that economists are on average deserving of higher status due to their larger numbers, better organized labor market, and thus higher quality cream that rises to the top (putting aside the obvious nepotism in Cambridge-centric and Berkeley-centric networks). A less charitable perspective would be that their proximity to power (e.g. President’s Council of Economic Advisers) gives them political capital that they are not shy about applying outside of their areas of expertise.
    For physicians the charitable argument is that their authority derives from face-to-face interactions with a broader swath of society. The uncharitable take is that they rely heavily on tradition and mystique for their authority. Most of them have terrifyingly little training in population-level thinking.
    How can we raise the status of demographers and public health specialists? Call the former data scientists and the latter social entrepreneurs. Fin.

Leave a Reply

Your email address will not be published. Required fields are marked *