Measuring the service system impact of a novel telediagnostic service program for young children with autism spectrum disorder

Participating systems and families

Participating families were served at a rural medical facility approximately 130 miles from the tertiary center where the remote assessor was located. Families were recruited from a target region of 23 counties immediately west of the tertiary center’s home region. This region was selected based on its identification as “under resourced” by the state Department of Education as well as its number of referrals to the tertiary center. According to the most recent US Census estimates, the mean population of represented counties was 34,408; median household income, $40,541.20 (range: 29,893–55,972), and percentage of persons living in poverty was 20.25 (range: 16–42.7) (Juárez et al., 2018).

For comparison, we analyzed patterns of tertiary center referral rates from a second rural region of the state with Part C support. The comparison region (17 counties) is immediately east of the tertiary center. It was selected based on its comparable distance, history of referrals, and demographic similarities.

Identification of appropriate referrals

In contrast to a comprehensive evaluation model (i.e. Autism Diagnostic Observation Schedule-Second Edition (Lord et al., 2012), cognitive assessment, adaptive skill interview), the telemedicine service relies on a singular assessment and a caregiver interview regarding developmental history and Diagnostic and Statistical Manual of Mental Disorders 5 (DSM-5) diagnostic criteria. As such, the evaluating team worked with referring providers to identify children and families most likely to be served well by this model. Referred families were enrolled in the state Part C system, whose service coordinators identified children showing ASD characteristics based upon parental report and/or direct observation.

Because ASD presentation in young children varies and early intervention providers have inconsistent expertise, a standardized pre-screening decision tree was developed for determining the appropriateness of the telemedicine service for families, and distributed among referring service coordinators. It included the following description:

The program is designed to help rapidly and accurately identify ASD (or clearly rule-out ASD) when this is a clear and pressing issue for families. The program is not intended to replace more in-depth assessments for families where complex issues require more substantial evaluation over time.

Guidelines were provided (Figure 1) for determining appropriate referral routes. Exclusion criteria necessitating referral for comprehensive evaluation at the tertiary center included complicating medical conditions (e.g. prenatal substance exposure), psychosocial factors (e.g. history of abuse), significant sensory impairment, or limited English proficiency. All referrals were then reviewed by the in-person assessor for appropriateness.

OPEN IN VIEWER

Figure 1.

Telemedicine referral decision tree.

We established regular meetings with the Part C system as part of an iterative program evaluation procedure to maintain ongoing communication about how this decision-making process functioned. This included discussing case examples of appropriate referrals as well as referrals of families not adequately served by this abbreviated model. This formalized check-in procedure between referring providers, in-person testers, and remote assessors was critical in assuring the telemedicine service was useful and positive for these rural families, for whom the Part C system was their first and primary point of continued contact.

Overview of the telemedicine diagnostic consultation model

At the rural medical facility, visits were conducted in an exam room outfitted with Health Insurance Portability and Accountability Act (HIPAA) compliant technology. In-person testers had college degrees in child development, experience working with families in homes, and training on the Screening Tool for Autism in Toddlers & Young Children (STAT, Stone et al., 2000). The STAT, a Level 2 ASD screener for use with children ages 24–36 months, was selected because it is well-validated and easily implemented in community settings (Stone et al., 2004). The remote assessor, a licensed clinical psychologist with extensive ASD expertise, observed and interviewed caregivers. At the conclusion of the 60-minute appointment, the remote assessor discussed diagnostic impressions with the family, which included making an ASD diagnosis, ruling ASD out, or recommending further evaluation (Juárez et al., 2018).

Impact of the telemedicine diagnostic consultation on regional referrals

From 2012 to 2016, the target region referred an average of 33.2 children per year (2.76 per month) to the tertiary center. After telemedicine services started in August 2016, the average total referral rate for 2016 (tertiary center, 38 + 5 months of telemedicine service, 20) rose to 4.8 per month (see Figure 2). This represents a 56.76% total increase in referrals for evaluations from 2015 (tertiary center only, 37) to 2016 (tertiary center + 5 months of telemedicine service, total of 58), and an 81.25% increase over the average of referrals made over the previous four years, 2012–2015 (average of 32).

OPEN IN VIEWER

Figure 2.

Target region referrals and evaluations.

We next compared referral rates across the first full telemedicine year (2017) to the last full year without telemedicine (2015). As seen in Figure 2, in 2017, the Part C system made 69 referrals for any type of evaluation, representing an 89.19% increase over 2015. Fifty-six of these referrals were for telemedicine evaluations; 13 were for the tertiary center, representing a 64.86% decrease in referrals for tertiary center evaluation. Of the 12 families then completing tertiary evaluations, 7 chose to do so rather than accessing telemedicine services, an option offered to every participating family. The remaining five children were medically complex, which made telemedicine services less appropriate. Information on the child who did not complete any evaluation was not available.

Importantly, the telemedicine service sped access to diagnostic consultation relative to waits at the tertiary center. In 2017, the statewide early intervention system referred 323 children under the age of 3 years for comprehensive evaluation, resulting in a 15-month wait time by January 2018. In contrast, families waited an average of 11.01 weeks for telemedicine visits from time of intake to time of evaluation (standard deviation (SD) = 8.83). This wait time has slowly increased (August-December 2016: mean = 8.17 weeks; August-December 2017: mean = 15.03 weeks) but still remains far below that of the tertiary center to which children would otherwise have been referred. Families are also keeping more appointments. Show rates from the target region to the tertiary center were 85% from 2012–2016; show rates for the telemedicine service were 100% in 2016 and 94.64% in 2017. Of families referred to the tertiary center rather than the telemedicine service in 2017, the show rate was 92.30%.

Comparison across rural geographies with versus without telemedicine service access

To determine whether differences in referral rates were attributable to telemedicine services versus statewide changes in Part C referral patterns, we compared the number of referrals from the target region with the number of referrals from the comparison region. From 2015–2017, referrals to the tertiary center from the target region decreased by 64.86% (from 37 to 13), whereas referrals from the comparison region increased by 114.71% (from 34 to 73).