By Professor Sharon Dawe from Griffith Health Institute and Paul Harnett from University of Queensland.

Imagine you’re a child protection worker who has received a notification from a teacher voicing concerns about a child in her class. The case involves a five-year-old boy named Toby.

Toby’s mum has had a long history of alcohol abuse but has managed this well for the past few years. But she has recently been seen in an intoxicated state when collecting Toby from school. Toby has had a great first year at school but has failed to attend school several days each week over the past month.

The starting point for child protection workers to respond to a report of suspected maltreatment is to estimate an overall level of risk. Many jurisdictions across the world use computerisedstructured decision-making toolsthat estimate risk based on the presence or absence of specificrisk factorsandprotective factors.

Decision-making is most straightforward when risk factors clearly outweigh protective factors or vice versa. One such example may be a case where a child shows signs of neglect, rarely attends school and lives with substance-misusing parents who are in a domestically violent relationship.

Decisions are much more difficult and prone to error when risk and protective factors appear to be fairly equal: a child lives with a mother who has a significant history of substance abuse who is believed to have relapsed, but is attending school regularly and achieving well.

Importantly, decisions about child protection do not only depend on the presence or absence of risk and protective factors. It seems to matter which state or territory the family is living in, and whether the child is Indigenous. In 2012-13,135,000 childrenwere involved with the child protection system, as either the subject of investigation, having a current care or protection order, and/or living in out-of-home care.

Victoria and Western Australia had the lowest rates of children on care and protections orders (8.6 and 8.7 per 1000 children respectively). Northern Territory and Tasmania had the highest (17.6 and 13 per 1000 children respectively).

The figures were similar for children living in out-of-home care: Victoria and Western Australia had the equal-lowest and Northern Territory had the highest (7.5 and 16.4 per 1000 children respectively). Aboriginal and Torres Strait Islander children wereeight timesmore likely to be involved with child protection services.

What we don’t know is whether jurisdictions that have lower rates of children being removed from their families are failing to identify children who are at risk of harm (known as false negative errors) or whether jurisdictions that have higher rates are incorrectly identifying children to be at risk when they aren’t (false positive errors). Or it might be that the figures reflect real differences in child protection concerns.

What we do know is that the threshold for determining that a child is at risk is related to a range of state-based factors such as the legal definition of “harm” and procedures for assessing risk. Where it is relatively difficult to argue a child is at risk of “harm”, false negatives will be more likely. And if procedures such as structured decision-making tools are risk-aversive, as suggested in the recentinquiry into the Queensland Child Protection system, false positives will be more likely.

So, what can be done to ensure the right decision is made for Toby, irrespective of his ethnicity or where he lives?

Errors can be reduced byobtaining more informationabout the family situation. Finding out more about Toby’s social and emotional development and his mother’s current well-being will increase our understanding of the family.

Perhaps his mother experienced a traumatic event and started drinking again. But will her drinking escalate to a point where she is not able to care for Toby? Or will she engage with services and prevent a brief lapse becoming a full relapse to alcohol abuse? Knowing for certain that the mother did not relapse to alcohol abuse would be significant in this case.

An assessment must be more than the simple identification of risk and protective factors; it also requires a prospective assessment. It must identify the critical problems in the family that can be translated into goals for change. Parents should be clear about the goals and feel supported in their attempt to reach them.

Keeping families involved with the child protection system simply to “keep a check on them” is counterproductive for families and governments. Parents will be empowered when the goalposts are clear and don’t shift, and when they believe that achieving a goal will be genuinely acknowledged as a meaningful achievement by the child protection system.

This raises the question of how long an assessment of parental capacity to change should continue. In the United Kingdom, the Children Act 1989 wasrecently amendedto specify a 26-week limit for finalising applications for care or supervision orders.

Is this long enough? Arguably, yes, but only if there has been an active assessment process that offers the parents the opportunity to make change with the assistance of appropriate levels of family support. This should not mean that the parents have necessarily reached a point where child protection can withdraw, but they have shown capacity to change that justifies ongoing support.

The comprehensive and prospective assessment of a parent’s capacity to provide a safe and nurturing family environment should lead to increased consistency in decision-making across jurisdictions.

Finally, comprehensive assessments require adequate resourcing. Arecent reportfrom the New South Wales Ombudsman found that the capacity of community services to respond to children at risk of significant harm was inadequate. We may have a clear idea what “best practice” involves, but ultimately we need skilled people in jobs to do the work.

This article first appeared on The Conversation: http://theconversation.com/risky-business-how-protection-workers-decide-to-remove-children-from-their-parents-32679