Time magazine gives some context to the recent Lancet study on the incidence of abuse, and the difficulty of getting accurate statistics...

Based on a review of research conducted on child abuse between 2000 and June of this year, researchers estimate that 4% to 16% of children are physically abused each year in high-income nations, including the United States, United Kingdom, Australia and Canada. As many as 15% are neglected, and up to 10% of girls and 5% of boys suffer severe sexual abuse; many more are victims of other sexual injury. Yet researchers say that as few as 1 in 10 of those instances of abuse are actually confirmed by social-service agencies — and that measuring the exact scope of the problem is nearly impossible.

The issue lies in the delicate nature of the crimes — and the consequences of intervention. Many cases of abuse are rife with potential for long-term harm of the child, whether or not the assault is reported. The decision to report is rarely clear-cut, says Theresa Costello, director of the National Resource Center for Child Protective Services, who was not involved with the new research. "Professionals want to advocate for their clients, but they also know the reality of the public child-welfare system," she says. "There is a natural professional dilemma when you see a kid and you think, 'I should make a report,' but you're not sure you want to subject that child to the system."

Indeed, the second study in the Lancet analysis, citing previous research, reveals that physicians reported only 6% of children's injury cases to protective services, even though they suspected the injury was a result of abuse 10% of the time. Further, researchers say that many more cases of maltreatment — particularly of sexual abuse — are never even suspected, and the victimized children never come forward to report the assaults.

"The official statistics agencies produce are conservative estimates of probably the lowest level of child maltreatment," says Dr. Cathy Spatz Widom, a psychology professor at John Jay College of Criminal Justice, who specializes in the long-term effects of child abuse and is a lead author on one of the Lancet studies.

Unfortunately, the article only mentions prevention in passing:

The ultimate goal is to prevent abuse in the first place, says Widom, and to protect the well-being of children who have been victimized. "It would be wrong to assume that all maltreated children are going to turn out to have all of these problems," she says.

The Lancet has a series of articles on incidence and treatment of child abuse. There is one on injury prevention, and the abstract for this survey of prevention techniques sounds interesting:

The Lancet, Early Online Publication, 3 December 2008
doi:10.1016/S0140-6736(08)61708-0
Interventions to prevent child maltreatment and associated impairment

Prof Harriet L MacMillan MD a , C Nadine Wathen PhD b, Prof Jane Barlow DPhil c, Prof David M Fergusson PhD d, Prof John M Leventhal MD e, Heather N Taussig PhD f
Summary

Although a broad range of programmes for prevention of child maltreatment exist, the effectiveness of most of the programmes is unknown.
Two specific home-visiting programmes—the Nurse—Family Partnership (best evidence) and Early Start—have been shown to prevent child maltreatment and associated outcomes such as injuries. One population-level parenting programme has shown benefits, but requires further assessment and replication. Additional in-hospital and clinic strategies show promise in preventing physical abuse and neglect.
However, whether school-based educational programmes prevent child sexual abuse is unknown, and there are currently no known approaches to prevent emotional abuse or exposure to intimate-partner violence. A specific parent-training programme has shown benefits in preventing recurrence of physical abuse; no intervention has yet been shown to be effective in preventing recurrence of neglect. A few interventions for neglected children and mother—child therapy for families with intimate-partner violence show promise in improving behavioural outcomes.
Cognitive-behavioural therapy for sexually abused children with symptoms of post-traumatic stress shows the best evidence for reduction in mental-health conditions. For maltreated children, foster care placement can lead to benefits compared with young people who remain at home or those who reunify from foster care; enhanced foster care shows benefits for children.
Future research should ensure that interventions are assessed in controlled trials, using actual outcomes of maltreatment and associated health measures.