How did Arizona safely reduce its investigation backlog?

In April 2015, Arizona’s Department of Child Safety (DCS) faced an enormous investigation backlog, with 33,245 open reports of child maltreatment. In a state that averages about 4,000 reports per month, the number of open reports was equivalent to eight months of work. The backlog put children at risk, left families in limbo, and increased pressure on agency staff and leaders. A multitude of factors contributed to the backlog, and an intensive, multi-pronged strategy was required to address it. Over the course of two years, DCS strategically worked through its backlog and by August 2017, the number of open reports had fallen to about 5,500.1

The local context

DCS’ backlog had been growing for many years. State leaders attributed it to inefficient processes, staff turnover, burdensome documentation requirements, and lack of consistent practices. Legislation passed in 2014 created the Department of Child Safety pulling the child welfare agency out of the Department of Economic Security, which meant that child welfare leaders and staff faced the added challenge of getting a new department up and running.

In addition, Arizona’s population had grown dramatically during this period. As the overall child population increased, so did the number of maltreatment reports. Between 2010 and 2014, DCS experienced a 44% increase in the number of maltreatment reports, particularly those involving allegations of neglect. This, coupled with a decrease in the number of caseworkers and supervisors due to the economic downturn, contributed to the growth of the backlog.

Preparation to address the backlog

As part of the 2014 legislation establishing DCS,2 independent consultant Chapin Hall Center for Children conducted a review of Arizona’s child safety system. Findings from the review helped to identify the root causes of the backlog, including operational ambiguities in the hotline decision-making process. To address the identified issues and drive the change necessary to reduce the backlog, new management processes were developed beginning in 2015. A workgroup was established to oversee implementation of the new processes, which were partially informed by learning from the experiences of other child welfare agencies and included five subcommittees to address specific issues:

  1. Increase the accuracy of referral categorization at the hotline;
  2. Implement targeted prevention strategies to reduce the need for DCS intervention;
  3. Create manageable workloads;
  4. Increase accuracy of safety and risk assessments during investigations;
  5. Reduce the number of inactive cases.

Each task developed by the workgroup and its subcommittees was assigned an executive leader and a deadline, and the status of all five tasks was reviewed at weekly status meetings. Recognizing that each region struggled with a different constellation of issues contributing to the backlog, each region developed a specific plan within the context of the overall statewide plan to reduce overdue investigations, and, as appropriate, shared successful.

One of the primary reasons we’re able to sustain our performance is the energy and effort we put in at the hotline to ensure that our screen-in rate is appropriate, and we have a secondary quality control for anything that’s not screened in.

– Michael Faust, Arizona Department of Child Safety

Intensive backlog reduction strategies

To reduce its backlog, DCS implemented several strategies and continues to employ many of them today.

Revise hotline screening processes

Arizona began by significantly revamping hotline screening processes. Chapin Hall’s 2015 review found that decision-making protocols related to reporting and investigation were not being used regularly and that the hotline’s safety priority matrix was unclear. Hotline staff had a great deal of discretion on cases of neglect, and they tended to err on the side of caution, setting a higher priority level than the cases required. As a result, caseworkers were being sent out to investigate families unnecessarily.

Arizona developed a clear hotline decision tool that includes specific examples of maltreatment and summaries of relevant legal statutes. In addition, the tool collects information on vulnerability, such as young age or mental/cognitive disability, and guides staff through the determination of priority level. Hotline staff also received coaching to increase interrater reliability. As a result, the screen-in rate for investigations decreased from 70% to about 55%.3 Calls that are screened out undergo a secondary quality control process to ensure that cases that should be investigated are not incorrectly being screened out.

Assess and address children’s safety immediately

In cases where a report is accepted for investigation, Arizona requires investigators to make a safety determination and document the finding within 48 hours of seeing the child. If the child is determined to be unsafe, the investigator documents the safety plan, such as moving the alleged perpetrator outside the home or moving another caregiver into the home. The investigation may still be in process, but child safety is assessed and addressed immediately.

Triage overdue investigations

When Arizona was beginning its backlog reduction, reports over 60 days old were triaged based on vulnerability and date of report receipt. Children age 5 and younger, substance exposed infants, families experiencing domestic violence or substance use issues, and children with diminished physical or mental capacity received the highest priority. In addition, given that more recent reports were considered to present the greatest risk of continuing maltreatment, reports that were two to six months old received the highest priority, followed by those that were six to 12 months old, 12 to 18 months old, or more than 18 months old.

Use data to target resources

When Arizona started its backlog reduction, timely data were not available to staff on a regular basis to inform decision-making.

To address this challenge, Arizona developed a set of statewide and regional reports that include weekly and monthly metrics (target and actual) to monitor ongoing hotline work. In addition to metrics on staff vacancies, overtime, the number of children in care, and hotline service (such as average number of seconds to answer the phone call, call duration, and call abandonment rate), the dashboards include metrics specifically related to hotline calls and investigations.

Dashboard metrics related to hotline reports:

  • Total number of hotline reports received
  • Total number of open DCS reports
  • Number of inactive cases
  • Percent of calls screened in
  • Percent of calls with on-time response (lagging 30-day metric)
  • Percent of reports with all victim children assessed within response time
  • Percent of open reports with a safety decision
  • Percent of reports with timely documented safety decisions
  • Number of reports without a safety decision (at the end of each month)
  • Percent of screened-out calls changed to investigation
  • Inter-rater reliability: screen-in, response time, and criminal conduct
  • Response time (lagging 30-day metric)

Deploy resources where they are most needed

In examining the data, Arizona found that 87% of all reports came from three of its five regions. Rather than try to reduce the backlog in all 55 field offices simultaneously, an “overdue reduction team” was deployed to a small cluster of field offices at a time, beginning with those that had the biggest backlog. Focused solely on reducing overdue reports, these teams completed two to three times more reports per week than the field office would typically receive. Once the field offices with the largest backlogs were sufficiently caught up, the team’s efforts were redirected to a different set of field offices.

The vast majority of the backlog reduction was due to intensive work by existing staff, using overtime and stipends. The overdue reduction team consisted of central office staff and staff in various field offices with investigative backgrounds, including DCS specialists, supervisors, reviewers, field responders / runners (who moved reports to “pending closure” status for reviewers), and assistant program managers. All received a stipend to work ten extra hours per week on backlog reduction, in addition to their regular positions. Field office staff received overtime pay. Arizona created a protocol and workflow for overdue reduction teams, including specifying expectations for all team members.

In addition to the intensive work by existing personnel, Arizona hired clerical staff who primarily assisted with communications and typing up notes and contract staff. Contract staff were permitted to conduct administrative tasks such as acquiring police records and interview certain individuals, though they were not permitted to interview children, alleged adult perpetrators, or parents when there was suspicion of maltreatment or domestic violence.

You need to be able to look at and manage the number of total open and the percentage that are overdue at the state level, at the regional level, at the section level or field office level, and then at the individual unit level down to the individual investigator level.

– Michael Faust, Arizona Department of Child Safety

Support field offices in developing and implementing local plans

Although the overall framework for the backlog reduction effort was organized statewide at the central office, field offices (with support from the central office) developed region-specific plans to address their local backlogs, and local leaders had to “own” the improvement. This permitted field office staff to tailor plans to their own contexts and to have a sense of ownership in the process. Strategies developed by individual field offices that were effective in reducing the backlog were shared with other field offices so they could be replicated.

Aim for “net zero”

When Arizona embarked upon its backlog reduction plan, staff were completing 3,000 cases per month, but receiving 4,000 reports per month. One of the agency’s first steps was to implement a “net zero” policy, which set the expectation that staff would complete as many cases as they were receiving (that is, the backlog was not to be increasing). Once select field offices began to make progress in reducing the backlog, collective energy towards the effort increased; other field offices began to take notice of the positive momentum, and feel encouraged by the positive peer pressure to take steps to reduce their own backlogs.

Host regular meetings focused on the backlog

Arizona created a weekly schedule of phone calls that continues to this day, with deliverables to track backlog reduction progress and make changes in resource allocation as necessary. After reviewing and discussing data locally, regional leaders also began to hold accountability calls with the deputy director of field operations to review trends, discuss performance issues, and remove barriers as needed. At the end of the week, regional offices send updates the central office.

When we could actually go to the workforce and tell them, ‘You’re keeping up, things aren’t getting worse,’ right then and there, we saw a huge shift in their work output and their demeanor towards the situation… You could see people breathe a sigh of relief.

– Michael Faust, Arizona Department of Child Safety

Ongoing strategies to prevent backlog

In addition to continuing many of the strategies described above, Arizona has maintained its backlog reduction for the last 3 years by strategically assigning new cases, providing ongoing support to staff, and continually monitoring for backlog.

Strategically assign cases

Calls to Arizona’s statewide hotline are dispositioned to field offices based on the ZIP code of the report, at which point the local supervisor assigns each case to a specific investigator based on workload, experience, and the nature of the case. For example, a particularly complex case wouldn’t be assigned to an investigator who was new or had just completed a similar difficult case. At least once per year, the ZIP codes are evaluated and, if necessary, re-mapped to field offices based on the number of reports and/or changes in the population.

To support backlog prevention, Arizona also created groups of three neighboring offices called “geo groups” to help each other with reports. For example, if one office has a high level of vacancies or a high caseload, it will shift some of its reports to one or both of the other offices in its geo group. The group also communicates throughout the week via a shared electronic document.

Provide ongoing support to staff

After Arizona’s leaders traveled to Tennessee to learn about safety science, they changed their culture from one that was compliance-driven to one that recognized challenges and offered concrete supports. The state has since created protocols for onboarding of new staff, supporting existing staff, and developing transition plans when staff leave.

In addition, Arizona employs 20 coaches whose full-time job is to help the state’s 240 supervisors improve their administrative and clinical supervision skills. Prior to starting to coach, these staff undergo nearly a year of training and development.

Continually monitor for backlog

Finally, Arizona continually monitors its performance metrics and has weekly accountability calls with regional leaders and the deputy director of operations. If the percentage of open reports in overdue status exceeds the target performance goal (17%), a series of formal agency procedures is triggered to address and problem-solve the source of the overdue reports.4 This includes identification of the unit(s) that have the highest number of overdue reports, identification of reasons that cases are overdue, commitments to resolve and close reports, and discussion of the reports during the weekly accountability call.

1 Based on conversations with Zeinab Chahine, Casey Family Programs, September 20, 2019; and Michael Faust, Arizona Department of Child Safety, October 21, 2019 and November 8, 2019.
2 Laws 2014, 2nd S.S., Ch. 1, §159.
3 Arizona’s research nationally indicates that a normative screen-in rate is about 55% – 60%.
4 The number of hotline reports ebbs and flows around when school is in or out of session because of the change in children’s contact with mandated reporters. Arizona plans for the cyclical nature of reporting and adjusts its resources and targets accordingly.