Creating a System to Keep Children Safer

At the New Hanover County Department of Social Services, social workers juggle 17 to 18 cases at a time. Challenged by swelling workloads, they struggle to divide their attention among families.

This year, the department is unveiling a new tool that uses predictive analytics to help them do their jobs — and if the system proves to be successful, it could be a model for the rest of the state. An $800,000 grant from The Duke Endowment is allowing New Hanover County to test it as a North Carolina pilot.

“For this county to be on the cutting edge of child welfare is awesome,” says Wanda Marino, the department’s assistant director. ​“But much more exciting is saving one more child from being further abused or neglected.”

Fighting an Epidemic

Early on a Tuesday morning, two dozen people wait on plastic chairs in the New Hanover County DSS lobby. Marino, on staff since 1999, believes the department has never been busier.

Each month, DSS handles 300 reports of abuse or neglect. Last year, 275 children came into custody, compared with 142 in 2012. In 2016, the department received reports of 13 fatalities.

New Hanover County has been hard hit by prescription opioid abuse — one well-publicized study placed its biggest city, Wilmington, at the top of the nation for the problem – and officials say the epidemic has taken a heavy toll on the county and its families.

“Our social workers are very committed people who want to do all they can to keep children safe,” says Brian Bocnuk, children’s services chief at DSS. ​“But we’ve been seeing an alarming increase in serious child abuse and neglect, and more fatalities. We need help to do our jobs better.”

In the spring of 2016, New Hanover teamed with SAS, a global data and analytics company based in Cary, N.C., to test how predictive analytics could play a role in social services. The business world has used it for years to increase profit and competitive advantage.

“The question was, how could we take what we do in business and apply that technology to doing good for children?” says Jennifer Conner, a senior account executive at SAS. ​“New Hanover was looking for ways to be innovative, and had buy-in from IT, the county manager and the county board. All the key stakeholders wanted to be part of the solution.”

Putting Data to Work

The new system mines up-to-date data from public records and creates statistical algorithms to give caseworkers vital information to do their jobs. An alerts engine will trigger a red flag when changes in a child’s life signify a potential problem. Social workers can respond with appropriate follow-up.

Let’s say, for example, that a social worker has an open case involving a mother of two young children. The mother has abused prescription drugs, but she’s working through a treatment program.

An alert is triggered after someone at the address is arrested for impaired driving. The social worker responds, and discovers that a new boyfriend has moved in and has a history of drug abuse.

“If we can more rapidly assess risk and identify potentially elevated risk situations, then we can better protect children and prevent incidents that may end in serious abuse or even death,” Marino says.

The goal is to reduce child fatalities, decrease the number of children in out-of-home care, and increase the number of children in permanent homes. The system is also expected to save the department more than $1 million annually and decrease caseworker turnover.

Once the basic data structure is built, experts say expanding to additional counties across North Carolina will be less expensive.

Early Stages

Mixing predictive analytics with child welfare isn’t without controversy. The Chronicle of Social Change, an online news source, has published more than a dozen articles and blogs on the topic. The central tension, explains publisher Daniel Heimpel, is the concern that using ​“big data” could infringe on civil liberties and lead to racial profiling.

“The field of child welfare is still in the very early stages of experimenting with predictive analytics,” Heimpel writes, ​“so we will have to wait to see how good or bad this application of technology will be.”

Tamika Williams, associate director of the Child Care program area at The Duke Endowment, says she recognizes the sensitive nature of gathering data. In funding the work with SAS, ​“the Endowment has limited the scope of the project to already-available public data sources and using it only for open child welfare cases,” she says. ​“By participating in this way, this effort seeks to protect privacy while enhancing child safety.”

With the New Hanover County project launching this summer, it’s too early to measure its success. But SAS says that jurisdictions using predictive analytics on open cases have seen more than 50 percent reduction in child fatalities and more than 20 percent reduction in child abuse.

Brian Bocnuk believes this is the child welfare system of the future.

“This is the birth of a new opportunity,” he says. ​“It doesn’t take the place of the social worker, but it gives us the tools to make better decisions.”