Stories

Fetal heart-rate recovery

Open the research page

From a 500 mL bolus to the 30-minute checkpoint, and from step curves to a forest of relative risks.

03060120MinutesmL deliveredPercent reached0%50%100%BolusContinuousp = 0.010p < 0.001p < 0.001500 mL75 mL725 mL300 mLBolusContinuousI, II, I over timeRecovery remains counted33.3 pt40.0 pt43.3 ptA 70.0%A 50.0%B 73.3%B 43.3%30 min1.62 [1.13,2.34]33.3 [10.2,52.3]Reviewer A1.40 [0.91,2.15]20.0 [-4.6,41.5]Reviewer B1.69 [1.07,2.69]30.0 [5.1,50.3]Consensus1.71 [1.12,2.62]33.3 [8.9,52.8]60 min1.71 [1.24,2.35]40.0 [19.2,57.7]120 min1.76 [1.29,2.41]43.3 [23.8,60.8]Adjusted35.7 [14.3,57.2]0.501202404608Relative risk, bolus / continuousDifference, percentage pointsNNT 3 [1.9,9.8]HR 3.20 [1.71,6.00]Multicentre trialStata release · synthetic runRationale1 / 21
  1. Rationale

    1We prescribed 1000 mL in both groups: a 500 mL bolus at 1000 mL/h for 30 minutes, then 150 mL/h, versus continuous infusion at 150 mL/h throughout.From Manuscript draft, 2026-09

  2. 2By 30 minutes we had delivered about 500 versus 75 mL, and by 120 minutes about 725 versus 300 mL, making the early window most revealing.From Manuscript draft, 2026-09

  3. 3We randomised 60 women with Category II tracings at one centre; both groups also received left lateral positioning, oxytocin stopped where applicable, and oxygen at 10 L/min.From Manuscript draft, 2026-09 · Figure 1, trial flow

  4. 4We masked participants and the attending by wrapping the bag and drip chamber in foil, housing the pump in cardboard behind a curtain, and reading tracings from a separate room.From Manuscript draft, 2026-09

  5. 5We defined the primary endpoint as recovery to Category I within 30 minutes; once recorded, recovery remained counted even if a later reading returned to Category II.From Manuscript draft, 2026-09 · src/00_preprocess.do

  6. Progress

    6At 30 minutes we knew only who had already reached Category I: 86.7% after bolus versus 53.3% after continuous infusion, with p = 0.010.From Manuscript draft, 2026-09 · src/04_cumulative.do, Figure 3

  7. 7We found recovery of 96.7% after bolus versus 56.7% after continuous infusion at the secondary 60 minute checkpoint; Fisher's test gave p < 0.001.From Manuscript draft, 2026-09 · src/04_cumulative.do, Figure 3

  8. 8We found 100.0% recovery after bolus at exploratory 120 minutes, while continuous infusion remained at 56.7%; the comparison gave p < 0.001.From Manuscript draft, 2026-09 · src/04_cumulative.do, Figure 3

  9. 9I measured the vertical gap at each checkpoint: 33.3, 40.0 and 43.3 percentage points at 30, 60 and 120 minutes, respectively.From Manuscript draft, 2026-09 · Figure 2B, estimates

  10. 10We asked two blinded reviewers to reread the primary tracings; their agreement with our bedside call was 76.7% for reviewer A and 85.0% for reviewer B, and their own 30 minute recovery calls were 70.0% versus 50.0% for A and 73.3% versus 43.3% for B.From Manuscript draft, 2026-09 · Stata release, reviewer aggregate

  11. 11I expressed each gap as a bolus to continuous relative risk on a log axis, while keeping the same time coordinates above for the original observations.From Manuscript draft, 2026-09 · Figure 2A, forest plot

  12. Findings

    12I estimated the primary 30 minute relative risk as 1.62 (1.13 to 2.34) and linked its interval directly to the observed 30 minute gap.From Manuscript draft, 2026-09 · Figure 2A, forest plot

  13. 13We found reviewer estimates in the same direction: A 1.40 (0.91 to 2.15), B 1.69 (1.07 to 2.69), and consensus of 2 of 3 readers 1.71 (1.12 to 2.62).From Manuscript draft, 2026-09 · Figure 2A, forest plot · Figure 2B, estimates

  14. 14We found ratios of 1.71 (1.24 to 2.35) at 60 minutes and 1.76 (1.29 to 2.41) at 120 minutes, so the earlier lead remained.From Manuscript draft, 2026-09 · Figure 2A, forest plot

  15. 15I then translated each ratio to a risk difference in percentage points; the forest rows slid onto the same linear measure as the gaps above.From Manuscript draft, 2026-09 · Figure 2B, estimates

  16. 16After covariate adjustment, I estimated a difference of 35.7 (14.3 to 57.2) points; our primary number needed to treat was 3 (1.9 to 9.8).From Manuscript draft, 2026-09 · Figure 2B, estimates

  17. 17As an exploratory check, I modelled intervals from 0 to 30 and 30 to 60 minutes; the hazard ratio was 3.20 (1.71 to 6.00).From Manuscript draft, 2026-09 · src/04_cumulative.do

  18. 18We saw no progression to Category III in either group, while two secondary outcomes did not differ; I returned to the delivery schedules to interpret the recovery comparison.From Manuscript draft, 2026-09

  19. Impact

    19We assigned both groups the same 1000 mL of saline through the same volumetric pump; because only the assigned rate differed, the trial supports an effect of faster administration on early recovery to Category I.From Manuscript draft, 2026-09

  20. 20We propose a multicentre trial that measures neonatal acid base status and NICU outcomes as the next step for bolus infusion in intrauterine resuscitation.From Manuscript draft, 2026-09

  21. 21I developed the methods, analysis, visualisation, validation and software, and released the Stata code so others can run it end to end on synthetic data.From Stata release README

Open the research page