Pages

Friday, October 15, 2010

An interesting exercise

An interesting exercise.

I'm going back through CDC abortion reporting data, looking at how they calculated abortion mortality, and what they concluded. We'll start with the oldest one I can find, 1971, which is in my files here at home. When an online version is available, I will provide a link (even if only to a summary or abstract).

  • 1971: No national mortality data.

  • 1972: "In 1972 CDC initiated a surveillance system to determine the number of abortion-related deaths occurring annually in the United States. Data were collected through the cooperation of personnel in charge of vital statistics in the state health departments. Reports were received from all 50 states, New York City, and the District of Columbia." They investigated and tallied 19 legal abortion deaths and 35 illegal abortion deaths, but did not explain how they distinguished between legal and illegal abortions.

  • 1973: CDC notes the addition of "state medical or hospital associations, CDC investigation, published case histories, and records from other federal agencies." They also noted, "The delay in reporting abortion deaths to CDC has ranged from 1 to 33 months." This resulted in the discovery of an additional two legal abortion deaths, and five additional illegal abortion deaths. For 1973, they noted 24 legal abortion deaths and 19 illegal abortion deaths. And now, for the first time, they indicate how they distinguish between the two: Legal abortion deaths are deaths from abortions performed by licensed physicians; all other induced abortion deaths (including those for whom the perpetrator or his license status is unavailable) are counted as illegal. Thus, if a woman keeps a legal abortion secret from her family and then dies, her death is counted as an illegal abortion death, since the CDC would be unable to verify that it was indeed performed by a licensed physician.

  • 1974: Missing

  • 1975: Missing

  • 1976: The data collection method seems unchanged since 1973; however, how each individual death is investigated is expanded upon. The distinction between legal and illegal (presumed illegal unless proven to have been performed by a licensed physician) remains. Delays of up to 37 months in reporting deaths were noted. The tally of legal abortion deaths from 1972 has risen to 24. For 1973, the number of legal deaths tallied rose by 2, while illegal deaths remained at 19. Total deaths reported so far:

    1972: L = 24, I = 39
    1973: L = 26, I = 19
    1974: L = 27, I = 6
    1975: L = 29, I = 4
    1976: L = 10, I = 3

    Why did deaths suddenly fall by about 33% in 1976? There's no exploration and no explanation. Did they fall? Or did something change in the data collection?

  • 1977: No change in surveillance methods noted. Delays of up to 6 years in reporting of deaths noted, with a median delay of 7 months. The definition of "legal" abortion is expanded to include abortions performed under the supervision of a physician.

    Here's where we see something interesting happen. For 1973 and 1974, the number of reported abortion deaths went down from what had previously been reported, with one less abortion-dead woman in 1973, two fewer in 1974, and one less in 1976. Surely these women didn't return from the dead! What happened? (At Life Dynamics, when Mona and I noticed this, we asked if, like Elvis, these women had somehow been spotted somewhere, and thus dubbed this phenomenon "Elvising" the woman, as in "For some reason the CDC Elvised a woman who'd died in 1973.") There is no explanation for why previously confirmed deaths are no longer being counted.

    1972: L = 24, I = 39
    1973: L = 25, I = 19
    1974: L = 25, I = 6
    1975: L = 29, I = 4
    1976: L = 11, I = 2
    1977: L = 15, I = 4

    1978: Missing

    1979 - 80: (Combined) "Of the subsequently confirmed abortion-related deaths for 1972 - 1980, 65% were reported through State health departments. We also obtained reports from medical and hospital associations, maternal mortality committees, NCHS, the Commission on Professional and Hospital Activities, case histories published in professional journals, and private sources." But they also indicate that beginning in 1979, ectopic pregnancy deaths would no longer be counted among abortion deaths unless the abortion attempt directly caused the woman's death. This is mightily convenient for abortion cheerleaders, since women often fail to seek health care for ectopic pregnancies immediately after abortion procedures because they assume that they can't possibly be pregnant, and abortion facilities have often been caught advising women to avoid seeking care they might have otherwise sought. So though the abortion led to the death by preventing timely intervention, the CDC doesn't count these deaths as abortion-related.

    The tally:

    1972 - 1976: No change
    1977: L = 17, I = 4
    1978: L = 7, I = 7
    1979: L = 18, I = 0
    1980: L = 8, I = 1

    It was around the time that this report was published that Willard Cates was demoted and transferred to the Sexually Transmitted Disease unit of the CDC in the wake of a shake-up by the Reagan White House. Though Cates had been aggressive in seeking out and analyzing abortion deaths, he was also a vociferous and unabashed proponent of abortion and published a lot of articles, written at taxpayer expense, calling for public funding of abortion, "educating" all pregnant women about "the availability of safe, legal abortion", excusing bad behavior by abortion doctors and instead blaming safety regulations for patient deaths, etc. His partner in many of these publications, David Grimes, left the CDC about two years later.

  • 1981: This year they looked at deaths by type of abortion performed, rather than by year.

  • 1982 & 1983: Missing

  • 1984 - 85: They no longer describe their mortality collection method. The tally:

    1972: L = 24, I = 39
    1973: L = 25, I = 19
    1974: L = 26, I = 6 (Legal up 1 from before)
    1975: L = 29, I = 4
    1976: L = 11, I = 2
    1977: L = 17, I = 4
    1978: L = 9, I = 7
    1979: L = 18, I = 0
    1980: L = 9, I = 1
    1981: L = 7, I = 1
    1982: L = 11, I = 1
    1983: L = 10, I = 1
    1984: L = 11, I = 0
    1985: L = 8, I = 1

  • 1986 - 87 & 1988: Did not examine mortality. And the fairly large team (Herschel W. Lawson, Hani Atrash, Audrey Saftlas, Lisa M. Koonin, Merrell Ramick, and Jack C. Smith) has been cut to Koonin, her assistant (Ramick), Smith, and Atrash.

  • 1989: Mortality information not available. Atrash has left the team, which is down to Koonin, Ramick, and Smith.

  • 1990 & 1991: Did not address mortality.

  • 1992: "Sources for obtaining such data include national and state vital records, maternal mortality review committees, surveys, private citizens, the media, health-care providers, and medical examiner reports. Clinical records and autopsy reports are requested and reviewed by medical epidemiologists to determine the cause of death and to verify that the death was abortion-related." Note that the aggressive search for abortion deaths is no longer mentioned. The CDC is now, more by omission than commission, admitting to their passive system described to me by Clarise Green (a research fellow at the CDC) when I was researching Lime 5. As Ms. Green indicated, the primary source of abortion deaths for the CDC was death certificates provided by the NCHS. My contacts at state vital records offices indicated that NCHS gets only a sample of death certificates from each state.

    The tally:

    1972: L = 24, I = 39
    1973: L = 25, I = 19
    1974: L = 26, I = 6
    1975: L = 29, I = 4
    1976: L = 11, I = 2
    1977: L = 17, I = 4
    1978: L = 9, I = 7
    1979: L = 22, I = 0 (Legal up from 18 at last count)
    1980: L = 9, I = 1
    1981: L = 8, I = 1 (Legal up from 7, and they still missed at least 1)
    1982: L = 11, I = 1
    1983: L = 11, I = 1 (Legal up from 10)
    1984: L = 12, I = 0 (Legal up from 11)
    1985: L = 11, I = 1 (Legal up from 8)
    1986: L = 11, I = 0 (I know they missed some.)
    1987: L = 7, I = 2 (They missed some.)
    1988: L = 16, I = 0
    1989: L = 12, I = 1
    1990: L = 5, I = 0
    1991: Not reported
    1992: Not reported

    I'm very curious as to why, in this report, they suddenly noted ten deaths that had previously eluded them. This after three years in which they expressed little interest in abortion mortality. The year they worked on this report, however, is also the year research fellow Clarise Green joined the crew.

  • 1993 - 94: Brief mention. The tally:

    1972 - 1990: no change
    1991: L = 11, I = 1 (They missed one that I've verified.)
    1992: Not reported
    1993: Not reported
    1994: Not reported

  • 1995: Mortality not reported.

  • 1996: "CDC has periodically reported information on abortion-related deaths; 1972 was the first year for which such data were available. Sources for such information included national and state vital records, maternal mortality review committees, surveys, private citizens, the media, health-care providers, medical examiners' reports, and, more recently, computerized searches of several full-text newspaper databases."

    The tally:

    1972 - 1991: No change
    1992: L = 10, I = 0
    1993 - 1996: Not reported

  • 1997: Mortality not addressed.

  • 1998: No change noted in mortality collection and analysis.

    The tally:

    1972 - 1992: No change
    1993: L = 6, I = 1
    1994: L = 10, I = 2
    1995: L = 4, I = 0
    1996: L = 9, I = 0
    1997: L = 7, I = 0
    1998: Not reported

  • 1999: No change noted in surveillance methods. No changes in death numbers, just added for 1998: L = 9, I = 0.

  • 2000: No changes noted in data collection methods, however, they identified four more legal abortion deaths for 1990 and one more legal abortion death for 1997, and note 4 confirmed legal abortion deaths for 1998.

  • 2001: No changes noted. The following years are added:

    1999: L = 4, I = 0
    2000: L = 11, I = 0

  • 2002: Mortality not addressed other than an intention to update numbers at some point.

  • 2003: They mention that they counted deaths for 2001 and 2002, 6 legal and 1 illegal for 2001, and 9 legal for 2002.

  • 2004: No change, just 10 legal deaths noted for 2003.

  • 2005: No changes, just 4 legal and 1 illegal death noted for 2004.

  • 2006: Suddenly they find it necessary to add "Since 1987, CDC has monitored abortion-related deaths through its Pregnancy Mortality Surveillance System," which they don't describe in any detail. They go into a bit more detail on how they define abortions: "An abortion is defined as "legal" if it was performed by a licensed physician or an appropriately licensed advanced practice clinician acting under the supervision of a licensed physician; an abortion is defined as "illegal" if it was performed by any other person." They add 3 legal abortion deaths for 2004, and also note 7 legal abortion deaths in 2005.

    While I'm at it, I've been doing some data analysis (or trying; I'm not a statistician so I'm learning on the fly here), and found this great article on basic data analysis. I really liked this bit:

    Finally, be aware of numbers taken out of context. Again, data that are "cherry picked" to look interesting might mean something else entirely once it is placed in a different context.

    Consider the following example from Eric Meyer, a professional reporter now working at the University of Illinois:

    My personal favorite was a habit we use to have years ago, when I was working in Milwaukee. Whenever it snowed heavily, we'd call the sheriff's office, which was responsible for patrolling the freeways, and ask how many fender-benders had been reported that day. Inevitably, we'd have a lede that said something like, "A fierce winter storm dumped 8 inches of snow on Milwaukee, snarled rush-hour traffic and caused 28 fender-benders on county freeways" -- until one day I dared to ask the sheriff's department how many fender-benders were reported on clear, sunny days. The answer -- 48 -- made me wonder whether in the future we'd run stories saying, "A fierce winter snowstorm prevented 20 fender-benders on county freeways today." There may or may not have been more accidents per mile traveled in the snow, but clearly there were fewer accidents when it snowed than when it did not.


    I've wondered this myself -- whether there are actually more or fewer accidents during bad weather (due to there being fewer people on the roads), and whether the accident rate per number of miles driven actually drops on holiday weekends because everybody's being hyper-vigilant.
  • No comments:

    Post a Comment