Methods: Our choice of six countries or states, Sweden, USA, Manitoba, Western Australia, England and New Zealand, is based on the availability of data, with Manitoba and Western Australia included because of the long-standing availability of high quality, linked data for these states. We also selected countries because of differences in welfare inequalities and support for parents and in policies on child maltreatment. Administrative data from both health (ICD) and social service agencies were used to develop age adjusted indicators for trend analysis from the countries. We determined time trends in the annual incidence or prevalence of child maltreatment indicators within each country fitting Poisson and negative binomial regression models. We compared continuous linear trends against change–point models with two slopes including a parameter corresponding to the time at which the change of slopes occurred. We also compared rates for each indicator for 2005 using the mean rate for three years 2004 to 2006. As absolute rates (though not trends) differed for children aged between 1 and 5 years compared with children aged between 5 and 11 years, we estimated rate ratios for each of the three age groups (<1 year, 1-<5 and 5-<11 years).
Results: While these countries have seen numerous policy initiatives over the last 30 years, the impact on indicators of child maltreatment has been limited. Trends for violent deaths in children, hospital admissions for maltreatment-related injury, and officially recognised physical abuse or neglect in children under 11 years old as primary indicators for physical abuse or neglect, showed no consistent evidence for a decline across the six countries. Rates for these indicators remained stable from the mid 1990s, but a decline in rates of violent death in a minority of age-country groups coincided with declines in maltreatment-related injury in Sweden and Manitoba. Overall, one or more agency indicators (notification, investigation, officially recognised physical abuse or neglect, or out-of-home care) increased in all countries except Manitoba, particularly in infants, possibly reflecting early intervention policies.
Mean rates between countries revealed 5 to 10 fold differences in rates of agency indicators, but less than two-fold variation in violent deaths or maltreatment-related injury, except for high rates of violent death in the USA.
Conclusions and Implications: These findings raise questions about the effectiveness of child protection policies which affect a substantial minority of children. Robust evaluation, including use of randomised controlled trials, is needed to determine whether the potential harms of the high and rising levels of agency contact and investigation and out-of-home care in some settings are outweighed by their benefits.
