A baby is born with severe neurologic injury.
Months or years later, lawyers, physicians, nurses, experts, and jurors gather around strips of paper from the labor room.
The fetal heart rate tracing is reconstructed minute by minute.
A deceleration at 1:42 PM.
Reduced variability at 2:07.
Another deceleration at 2:31.
By 3:15, someone argues, the baby should have been delivered.
Looking backward, the story can appear remarkably clear.
But there is a problem.
The clinicians caring for the patient at 2:07 PM did not know the outcome.
We do.
That difference changes everything.
The Monitor Is Not a Crystal Ball
Electronic fetal monitoring is one of the most ubiquitous technologies in modern obstetrics.
Its basic physiologic premise is reasonable.
Changes in fetal oxygenation can alter fetal heart rate patterns. Monitoring those patterns can therefore provide information about fetal status during labor.
But somewhere along the way, a physiologic surveillance tool acquired a much larger cultural role.
It began to be treated as though it could reliably predict which fetus would suffer permanent neurologic injury.
The evidence does not support that degree of certainty.
In one of the landmark studies, Nelson and colleagues examined fetal monitoring among children with moderate or severe cerebral palsy and controls.
Certain fetal heart rate abnormalities were associated with increased risk.
But association is not prediction.
That distinction is fundamental.
A finding can be associated with an outcome and still perform poorly as a clinical predictor when the outcome itself is uncommon.
What Randomized Trials Tell Us
The Cochrane review of continuous cardiotocography included 13 trials involving more than 37,000 women.
