Cluster RCT on PlGF test to diagnose pre-eclampsia - Lancet Publication to immediate national recommendation

Based on this study (PARROT) NHS England is rolling PlGF testing nationally, to all pregnant women with suspected pre-eclampsia

What did the stepped wedge cluster trial do?

Population - women with suspected pre-eclampsia

Intervention - PlGF test results revealed. (see supplementary appendix).

If PlGF less than 12 pg/ml then clinicians are asked to admit and measure blood pressure 4 times a day, and do all the tests needed to diagnose pre-eclampsia, and ultrasound

If PlGF more than 100 pg/ml then clinicians are asked to send the patient home, and only measure blood pressure once a week.

12-200 pg/ml twice a week blood pressure, ultrasound, and blood tests for pre-eclampsia

Control: Usual care where for suspected pre-eclampsia (blinded to PlGF results) will not be checking blood pressure that often, may be twice a week or doing ultrasound for fetal growth

Outcome: Time to diagnose pre-eclampsia (which could be high blood pressure with proteinuria or abnormal fetal growth or abnormal blood tests)

So by virtue of knowing the test results, clinicians monitor intensely in test positive patients, and are more likely to diagnose pre-eclampsia early; whereas usual care group only monitors less frequently, does not do scans (a criteria for diagnosing pre-eclampsia), and hence is not likely to diagnose early.

  1. Sample size calculation:

for 50% reduction in mean time to diagnosis of pre-eclampsia.

For the actual RCT result,

Time to diagnosis of pre-eclampsia

control median 4.1 days (0.5-9.2)

intervention median 2.2 days (0.8-14.7)

The primary outcome is reported as time ratio 0·36 (95% CI 0·15–0·87; p=0·027).

There is overlapping CI in median between the two groups.

Secondary outcome of severe maternal morbidity 22 [4%] vs 24 [5%] events; aOR 0·32, 95% CI 0·11 to 0·96; p=0·043, and highlighted as benefit in abstract

Look forward to discussion on above in this forum.


Do you have any specific questions?

1 Like

We do not recommend population-based screening studies based on detection only. We do so based on meaningful improvements in patient outcomes. Similarly, evidence-based medical testing recommendations are based on (changes in) patient-relevant outcomes.

Why would the medical community reason differently about testing for pre-eclampsia?

Table 2 in the Lancet article shows a shorter time to detection, but no improvement in outcomes, only more procedures: more ultrasound, more pre-labour C-sections,


1 Like

Thank you for highlighting this article on an interesting development in monitoring for pre-eclampsia. While reading it, I was struck by the close resemblance between casually ’throwing a variable into [the right-hand side of] a regression’, and the attitude adopted in this ‘pragmatic’ trial of the ‘real-world effectiveness’ of an ‘intervention’ in prenatal care.

To the mediocre analyst, regression modeling is chiefly a matter of ‘throwing variables into’ a regression and seeing which ones ‘stick’ (say, because p < 0.05). Having no interest in the realistic detail of the world beyond the computer screen, such an analyst cares little about the functional form of the RHS of the regression, and uses whatever form (and link function) happens to be handy in off-the-shelf statistical software. This analyst scrupulously avoids any deep examination of the informational content of the variable ‘thrown in’, and makes no effort to modify the ‘default’ RHS.

Likewise, this trial seems to have ‘thrown PlGF testing’ into the existing system of care without examining closely the informational content of PlGF, and without devoting any attention to modifying the ‘functional form’ of that system. The PlGF apparently served only to intensify surveillance with the usual clinical and laboratory measures. Also notable in the report was the lack of interest in any sort of mediation or path analysis to account for causality of the reduced time to diagnosis observed in the revealed-testing group.

Somehow, terms like ‘pragmatic’ and ‘real-world’ have gained a cachet that induces us to forgive failings like these.