Sleep training methods and the state of the evidence
Few subjects set such sharply drawn camps against each other on such thin data. Here is what the published trials actually measure, and what they stay silent on.
What the term covers
The protocols studied range from graduated extinction, where parents progressively space out their responses, to bedtime routine preparation and parent education about sleep. Public debate often lumps them under a single label although they differ greatly.
That confusion weighs on the discussion: a criticism aimed at a strict protocol is received as targeting the whole, and a result obtained with a light intervention is presented as validating the harshest forms.
They also differ in the age at which they are applied: most trials concern children past the first months, and their results say nothing about the ages they did not study.
What the trials find
Systematic reviews find a moderate reduction in the number of reported wakings and in time to fall asleep, measured over a few weeks. Effects on mothers' reported state, notably depressive symptom scores, are found in several trials.
Effect sizes are modest and intervals wide. Trials are few, often small, and blinding is impossible: parents know what they are applying, and they are the ones reporting the outcome.
The main criticism
It concerns measurement. If the outcome is the number of wakings reported by parents, an intervention that teaches a child not to call produces a fall in reporting without our knowing whether wakings have decreased. The rare comparisons with actigraphy have not settled the point.
A second criticism concerns follow-up length: most trials stop at a few weeks or months, which allows no conclusion either about a lasting effect or about its absence.
What long follow-up showed
An Australian trial followed families to the age of six and found neither lasting benefit nor lasting harm on measures of attachment, behaviour and stress assessed by salivary cortisol.
That result is regularly cited by both camps: as proof of harmlessness by some, as proof of no lasting benefit by others. Both readings are compatible with the published data.
A single trial, however well conducted, does not close a question. It concerns a specific protocol, a given population and a set starting age, and the absence of a detected effect is not proof that there is no effect, especially for measures as indirect as those of attachment at six.
Why the debate does not close
Not all the questions at stake are empirical. The discussion also bears on what is considered acceptable to ask of a child and of parents, a trade-off no trial settles.
To which is added a rarely named inequality of circumstances: whether a protocol is feasible depends on housing, working hours and the number of adults present, variables trials control poorly and recommendations ignore.
Key points
The term covers very different protocols, from the strictest to simple parent information.
Measured effects are moderate, short-term, and rest on parental reporting.
Follow-up to age six found neither lasting harm nor lasting benefit, a result invoked by both camps.
This article describes documented mechanisms and disagreements. It recommends no position, no party and no vote, and commits to no reading other than that of the work it cites.
And you, where do you stand?
The Civix test situates your convictions on six dimensions and matches you with one of the twelve profiles, without ever telling you what to think.
Start my test