Child Torture as a Form of Child Abuse

28 children

Children evaluated at five US medical centres with evidence of physical abuse, neglect and psychological maltreatment including terrorising and isolation; a nonconsecutive case series selected by child-abuse pediatricians as representative of torture

A clinical case series with no comparison group, no denominator, and explicitly nonconsecutive selection — the authors chose cases they judged representative of torture, so the sample is defined by the outcome. The paper's purpose was to propose clinical criteria for identifying abuse by torture, not to estimate risk by schooling arrangement, and it makes no such claim. It is recorded here because the two percentages below are the single most-cited empirical datum in the argument that home education enables severe abuse, and because that citation is a misuse of the design: a proportion within a set of victims says nothing about risk without the base rate of homeschooling among non-victims, and the temporal ordering the authors themselves describe — withdrawal typically FOLLOWING the closure of a CPS case, used to isolate the child further — is reverse causation, not an effect of home education. Both facts have to travel together or the number is misleading.

Of 28 severe child-torture cases, 47% of the children had been withdrawn from formal school to be homeschooled and a further 29% had never been enrolled — roughly three-quarters outside institutional schooling at the time of the abuse. 36% of the children died as a result; the abuse continued for up to eight years. The authors concluded that homeschooling typically occurred after a previously opened CPS case was closed and functioned to isolate the child. The defensible inference is about DETECTION and about withdrawal-as-a-red-flag, which is how Luck et al. 2025 and the Connecticut Office of the Child Advocate both use it; the indefensible inference is a homeschooling risk ratio, which this design cannot produce and the paper never asserts.

Not applicable. The confound that matters is reverse causation — abuse preceding and motivating the withdrawal — which the authors describe directly.

The pattern — withdrawal from school following a child-protection contact — recurs in the Connecticut Office of the Child Advocate's 2018 review of 380 withdrawn students and in investigative work from Iowa, Kentucky, Indiana and Illinois catalogued by CRHE. That is convergent case-level evidence about a warning sign. None of it is a replication of a risk estimate, because none of these sources has a denominator.