Social workers did not kill Peter. They did not kill Victoria Climbié, tortured to death within a few streets of him nine years ago. Nor did they drug Shannon Matthews and hold her prisoner for 24 days. These children were abused by what we wishfully call their “families”. Baby P, now revealed as Peter, was subjected to unspeakable torture by his mother's new boyfriend, a sadist she barely knew. Shannon Matthews was abducted by her “step-uncle”. Climbié was starved and beaten to death by her great-aunt and her great-aunt's boyfriend.

Much of the outrage about these terrible events has focused on a detailed raking over of child protection procedures. This week we had frenzied government announcements of yet more targets and quangos, as well as BMA guidelines urging doctors to detain children in hospital on even the faintest suspicion of neglect. The mantra is “suspect everyone”. Few questions are being asked about root causes. We seem more squeamish about asking why these “families” have become so dangerous than we are about the appalling injuries they cause. If we really care about children (and I sometimes wonder whether we do, when so much of the response has been back-covering), we surely have to ask questions that might - God forbid - sound judgmental.

The Cinderella Effect is the name given to analysis in Canada, the US and Britain which suggests that children are at far greater risk from stepfathers and non-blood “relatives” than from natural parents. Detailed Canadian research over 20 years has put the risk of being killed by a stepparent at between 50 to 100 times greater than the risk of being killed by a parent. A 1989 study by the University of Iowa found non-biological fathers four times more likely than natural fathers to sexually abuse children in their care.

In Britain, NSPCC research has found that children living with biological parents are between 20 and 33 times safer than those living in any other type of household - despite the NSPCC being inclined to play down family breakdown.

Poverty must surely be a factor. But the American researchers Daly and Wilson, who have done the most detailed work on this subject, say that poverty pales into insignificance compared with “the presence of a step-parent, which is the best epidemiological predictor of child abuse yet discovered”.

I am not suggesting that we should start demonising stepfathers: there has been quite enough demonising already. But I do wonder why such figures are ignored by the authorities. Should we not stop the benefits system penalising marriage, as Frank Field keeps demanding? Should we not ask whether the welfare state might have created a class of serial boyfriends, who prey on single mothers on benefits?

Haringey Council moved Peter's mother from a three-bedroom council flat to a four-bedroom house soon after he was first taken to hospital. The sadist boyfriend seems to have appeared soon after, perhaps attracted by the house that was big enough to house him, his collection of neo-Nazi memorabilia and a second male lodger. The presence of those two men sealed the fate of Peter.

This is uncomfortable territory. Instead of venturing into it, the Government creates more and more bureaucracy to detect children at risk. We have Children's Trusts, Local Safeguarding Boards, multi-agency meetings and increasingly complex computer systems to share information.

The mania for information is understandable. Even in 1973, the murder of Maria Colwell by her stepfather demonstrated how deftly evil people conceal abuse. It showed that individual professionals - the policeman, the social worker, the GP - get only a partial picture. The same was true for Victoria Climbié and Peter. But the problem was not an inability to share information: it was that information needs interpreting. Social workers

dismissed a nurse's concerns about Victoria Climbié because they thought her subservience towards her great-aunt was part of “African culture”. Others believed that Peter's mother was merely inadequate.

The obsession with information-sharing has three perverse effects. First, it has become an obstacle to child protection, because it forces professionals to spend so much time entering data, and it fatally dilutes responsibility. Second, it brands all parents equally as potential abusers, which further confuses the picture. Third, it cuts away at trust between the public and the professionals.

People ask why none of Peter's neighbours alerted the local authority. They ask the same about Khyra Ishaq, the little girl who starved to death in Birmingham last year after stealing crumbs off a bird-table. It is the same reason that people no longer tell health visitors they are depressed: they do not

trust them. This week's BMA guidelines, which tell doctors to contact social services on the slightest suspicion of anything from physical to emotional harm, surely spell the death of patient confidentiality. That has profound implications.

If the statistics are right, by far the most effective check on abuse is the family. The real family, where the vast majority of real fathers see their first duty as protecting their children. Yet these are the people we have made afraid. We have institutionalised shamelessness among people with no notion of family, some of whom have killed children. Simultaneously we have sown widespread fear of decent, well-meaning professionals among the innocent, the majority of whom who do not move on from baby to baby and man to man.

Unless we address this, we will be wringing our hands in hypocritical horror again when the next Peter is discovered. And there will be one.

Times Recommends

• Copenhagen will fail – and quite right too
• They’re still wriggling to avoid the flak on Iraq
• Rule the waves? Not any more we don’t