Skip to content

Sickle cell and stillbirth rates

So, from here:

Of the total 63 334 deliveries, 516 (0.8%) resulted in a stillbirth (Table 2), and 2482 (3.9%) were by patients with SCT. The prevalence of stillbirth was 1.1% among patients with SCT compared with 0.8% in the general population (Table 2), and stillbirth was found to be associated with SCT after adjusting for all risk factors. In the fully adjusted model, patients with SCT were at higher risk of stillbirth relative to patients without SCT (adjusted odds ratio [aOR], 8.94; 95% CI, 1.05-75.79; P = .045) (Figure 3; eTable 2 in the Supplement).

I’ve simply not got the maths or background to be able to work this out. From here:

A study from the Centers for Disease Control and Prevention (CDC) found that Black mothers were more than twice as likely to experience stillbirth compared to Hispanic and white mothers. While the full reason for these racial disparities is not clear, differences in health problems that occur during pregnancy or underlying health conditions may be linked to these differences.

My reading of this is that SCT explains about half that difference. That about right? Or have I cocked that, as only 10% of Black Americans have SCT……

11 thoughts on “Sickle cell and stillbirth rates”

  1. The 95% CI of 1.05-75.79 tends to suggest that the figures are pretty meaningless. IMHO it’s just somebody torturing the stats to try and force a result.

  2. Are you more likely to have a still birth if you’ve already aborted a few?

    I’ve just finished watching Adam Curtis’s Russia 1985-1999: TraumaZone – highly recommended (but 7 hours long!) Like many Curtis epics, it uses live news footage (mostly shot by the BBC) to tell a story, in this case “What it felt like to live through the collapse of communism … and democracy”.

    One of the later programs (possibly episode 6, but don’t quote me) covers a woman going for an abortion, because her single room flat gives no space for a second child, and there’s also an interview with the boss of the abortion clinic, who mentions her most regular customer: aged 32, she’d had one child and fourteen abortions and was currently pregnant again.

    It seems Soviet (and probably current Russian) condoms aren’t very reliable, and birth control pills were (a) unavailable except for those with hard currency) and (b) denigrated as the product of capitalist hyenas.

    Very sad (but not the saddest part of the film series).

  3. Baron Jackfield is perceptive but what I should like to know is why the unadjusted ratio for those with SCT is approx 1.375 and the adjusted ratio is 8.94?!? What on earth among the adjustments are those environmental factors applying disproportionately to those with SCT that *reduce* the risk of stillbirths by 85%?

  4. “P = .045” It’s decades since I did any stats, but is that 0.045 to be compared with the conventional threshold of significance, to wit 0.05?

    If so it’s hard not to suspect some heroic data-polishing to allow the result just to sneak into the “statistically significant” category.

    Hm, “after adjusting for all risk factors. In the fully adjusted model, …”

    Adjusting for confounders is a game of arithmetically-assisted guesswork. You can never know that you’ve got a complete set or even a set that includes the most important variables. That, after all, is why you do a randomised controlled trial whenever you can.

    I like the good Baron’s point: if a lottery offered a single prize ‘between £1,000 and £75,000’ I don’t suppose I’d buy a ticket.

  5. Bloke in the Fourth Reich

    The important numbers are in table 2 of the publication. In the total population, 0.8% stillbirth rate, compared to 1.1% among mothers with SCT. This is not evidence of a real difference.

    If you scroll down a little and look only at the black population (which takes out all of the “confounders” that need the complicated, and never-disclosed statistical piggery-jokery, the mothers with SCT actually have a lower stillbirth rate (1.0%) than the general black population (1.1%).

    The stupid thing is, it would be totally unsurprising if there is a higher stillbirth rate among SCT mothers but the increase in risk is clearly rather small, and if it exists at all, this study has not demonstrated it. In fact, there must be close-to population level data on this, no need for statistical analysis of a small data set in one set of hospitals in one part of the country but that wouldn’t generate “more research (funding) is needed conclusions. Under the circumstances, I think I would leave it to American obstetricians and midwives to advise expectant mothers whether to pay any attention to a very small risk factor they have zero control over.

    I’m going to pick a further medical nit, and wonder what proportion of deliveries are at term. And ask a possibly more interesting scientific question as to whether the fetus’ SCT status has more influence on stillbirth risk than the mother’s.

  6. “And ask a possibly more interesting scientific question as to whether the fetus’ SCT status has more influence on stillbirth risk than the mother’s.”

    Or even SCD……

  7. @BiFR
    It *is* evidence of a real difference – just not very strong evidence of a significant real difference.

  8. @ dearieme
    Or just cherry-picking any of the results that *appear* to show “statistical significance”. In any twenty random statistical tests you should expect one to show “statistical significance” at the 5% level – so every decent statistician will want to set up an experiment (if that is ethical) or search for other independent data to test the hypothesis.
    I suspect that every Grauniad journalist/politician will seek to publish any untested cherry-picked dta that suits their political agenda

  9. Bloke in the Fourth Reich

    No, it isn’t evidence of a real difference – especially when the difference doesn’t just disappear – but reverses – when you stratify by skin colour, the thing highly correlated with SCT and also highly correlated with the various ill-defined adjustments.

    Sure, it’s a real difference in this sample, a very small real difference in the sample, but we care about the population. And the difference is not sufficient to convince us there may be such a difference in the population. It doesn’t even pass the eyeball test.

Leave a Reply

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