Telephone-Based Developmental Screening and Care Coordination Through 2-1-1: A Randomized Trial

This study was approved by the University of California, Los Angeles Institutional Review Board and registered at clinicaltrials.gov.

Clínica Monseñor Oscar A. Romero (CMOAR) is a federally qualified health center with 2 clinic sites in mainly Hispanic and underserved neighborhoods of Los Angeles. Pediatric primary care is provided by pediatricians, family physicians, and supervised physician assistants. Potential participants were recruited from June 2015 to January 2016 by the study’s research associate (RA) in clinic waiting rooms, by clinic staff when patients checked in for visits, and through phone calls by clinic staff to patients in the appropriate age range. We included patients between ages 12 and 42 months at baseline if they received primary care at CMOAR, were not currently receiving intervention services for a developmental delay or disability, and spoke primarily Spanish or English.

The recruitment sample and randomized assignments are described in Fig 1. After informed consent, each family was randomly assigned to the intervention or control group by using a random number generator with 1:1 allocation to the 2 groups. The RA then conducted a structured interview with the parent or caretaker at the time of enrollment. Baseline interviews included sociodemographic and health information about the child and family, questions about parental concerns related to child development or behavior, and items from the Promoting Healthy Development Survey (PHDS)²¹ to measure families’ experiences with primary care. For control families, the RA also conducted the PEDS, PEDS: Developmental Milestones, and MCHAT-R screens (the MCHAT-R is validated for children ages 16–30 months and was only administered to children in that age range at baseline). Baseline screening was conducted by the RA for the control group and by 211LA (as part of the intervention) for the intervention group. When screening results suggested that control families had developmental or behavioral concerns, the RA encouraged parents to raise those concerns with their health providers, but results were not shared directly by the RA with health providers. This was done to ethically address needs of control families without unduly influencing usual care. Both control and intervention families received care as usual from their primary care providers.

Intervention procedures are summarized in Fig 2. For families in the intervention group, the RA attempted to connect the parent to the 211LA care coordinator. Whenever possible, this connection was made by using a “warm hand-off” immediately after the baseline interview, calling the 211LA care coordinator directly and connecting the parent over the phone. If that was not immediately feasible, the RA shared the care coordinator’s phone number with the parent and sent the parent’s contact information to the care coordinator.

When the care coordinator and family were able to speak by phone, the care coordinator administered the PEDS Online screening tools, entering parent responses into the PEDS Online and 211LA data systems, sharing results with the parent, and making referrals on the basis of developmental risk. Detailed screening and care coordination protocols and procedures were developed and tested before this study.¹⁷ Typically, children with high or moderate developmental risks were offered referrals to their local EI (for children <3 years old) or ECSE (for children ≥3 years old) programs for developmental evaluations and services. When parents indicated behavioral concerns, additional referrals were made for behavioral health services. All families were also offered referrals to ECE programs and social services if parents expressed a need.

Whenever possible, the 211LA care coordinator made 3-way calls to connect the parents to local service agencies directly. When a 3-way call was not possible, the 211LA care coordinator completed referral forms and sent these securely to the service agencies or called agencies on the parent’s behalf. Parents were also given agency contact information so they could contact them directly. Results of screening and referrals were summarized in a care plan and sent to the primary care provider to be scanned into the child’s medical record.

Finally, the 211LA care coordinator made follow-up calls with parents and service agencies, typically ∼15, 30, and 60 days after their initial contact, to determine the status of referrals and address any barriers. As part of the Los Angeles County health and social services safety net, 211LA care coordinators also connected families with other services, such as transportation, housing, food, and utility assistance, as needed to address barriers and ensure timely interventions for children. Families received between 4 and 30 calls, as needed, to complete care coordination. Six months after each intervention-group family’s enrollment in the study, a follow-up document was sent to the clinic to describe final outcomes of each referral, including whether a child was evaluated, whether he or she was found eligible for services, and, if receiving services, what specific services (eg, speech therapy, physical therapy, behavioral therapy, or ECE).

Other data sources included medical record abstraction for each child at baseline and at 3 and 6 months after enrollment. The RA and medical-student volunteers conducted these abstractions, noting any documentation in the medical record about developmental screening, developmental-behavioral concerns, or referrals to intervention services in addition to general health information, such as health conditions, medications, clinic use, and vaccination status. The RA then conducted a follow-up structured interview with all study participants 6 months after enrollment, which included the previous PHDS items and follow-up questions about developmental-behavioral concerns, referrals, connection to services, and experiences with 211LA, if applicable.

A trained medical-student volunteer also conducted retrospective medical record reviews of an entirely separate cohort of children of similar age who were seen for well-child care at CMOAR during the 6 months before the intervention began to assess clinician practices before the study and to determine if the study might have altered clinician practice with respect to developmental screening and referrals among control patients.

All of the data collected were entered into the Research Electronic Data Capture system at University of California, Los Angeles.²² Analyses were conducted by using Stata software (version 14; Stata Corp, College Station, TX). Primary outcomes were dichotomous and included (1) referral to EI or ECSE and (2) receipt of intervention services. Each of these outcomes was considered a “yes” if documented in the medical record, reported by the parent in the baseline or follow-up interview, or included in the 211LA care plan. Additional outcomes, including whether developmental surveillance (asking developmental milestone questions without using a structured screening tool) was performed by the health provider, whether the primary care provider conducted developmental screening with a validated tool, and whether a developmental or behavioral concern was noted, were taken from the abstracted medical records. Because medical records typically do not document whether a child has been evaluated or found eligible for services by an EI or ECSE program, these variables were measured by parent report and 211LA data only. A comparison of the percentage of children with each of these outcomes between intervention and control groups was performed by using a 2-sided Fisher’s exact test. To measure differences in family experiences with primary care over time, we analyzed the following domains from the PHDS at baseline and at the 6-month follow-up: percentage of recommended anticipatory guidance topics discussed, percentage of family-centered care items reported as “usually” or “always,” assessment of parental smoking and substance use, and assessment of parental well-being. We built difference-in-differences models, with clustered SEs and controlling for child age, sex, and primary home language, to determine if there were statistically significant differences in change over time on the basis of group assignment.

Medical records were reviewed for all children ages 12 to 42 months who were seen in the clinic for well-child care during the 6 months before the start of the study and included 142 records. These abstractions revealed that although almost all children (99%) had documented developmental surveillance performed during well-child visits, only 4% had been screened by using a structured, validated screening tool, 11% had developmental-behavioral concerns noted in the medical record, and 8% had documentation of a referral for EI or ECSE.

Recruitment for participation in the RCT resulted in 182 children being assessed for eligibility and 152 children enrolling in the study after 12 were excluded for not meeting eligibility criteria and 18 declined to participate (Fig 1). Of the 152 children enrolled at baseline, 77 were randomly assigned to receive the intervention, and 75 were randomly assigned to receive usual care alone. Baseline sociodemographic characteristics of the children and families participating in the study are in Table 1. Average child age was 24.5 months, and child sex was evenly split between boys and girls (50%). Families predominantly self-reported as Latino or Hispanic (97%), mostly Spanish speaking (68%) or bilingual, having low annual household income, and having parental education at the high school level or lower. None of these family characteristics revealed a statistically significant difference between the intervention and control group. In terms of developmental-behavioral concerns, 38% of parents reported having a concern in the previous 6 months. This percentage did not differ between groups and is similar to previous national estimates.²³ Although there were more children in the control group scoring in the high-risk developmental category on the basis of the PEDS Online screening system (18% vs 7%), the difference was not statistically significant, and the combined number in the high- and moderate-risk categories (the criteria for EI or ECSE referral) was similar in the 2 groups. Overall, 8% of children who were tested failed the MCHAT-R, with no statistically significant difference between groups.

Descriptive statistics showing the primary study outcomes for the intervention and control groups are in Table 2. The percentage of children in the intervention group screened with a validated tool indicates the percentage that reached 211LA and received the intervention during the 6 months between enrollment and follow-up. None of the children in either group had documentation of a clinician-performed validated screening tool in their medical records. Developmental surveillance (milestone questions from the electronic medical record) conducted by the health providers in primary care, along with documentation of developmental-behavioral concerns by clinicians, was similar in the 2 groups. However, we found a large and statistically significant difference between groups in terms of referrals to EI and ECSE programs for evaluation (32% intervention versus 9% control, P = .001), completed evaluations (19% intervention versus 1% control, P < .001), eligibility for services (16% intervention versus 1% control, P = .002), and receipt of services (16% intervention versus 1% control, P = .002). These differences remained statistically significant after adjustment for child age, sex, and primary language in multivariable logistic regression models (Supplemental Table 4). Other referrals made by 211LA to families in the intervention group included child care and preschool programs, family literacy programs, and behavioral health and social services. We did not see any documentation in medical records about referral to these other programs and services and did not hear from parents in the control group that their child health providers made such referrals.

Parents’ reports about primary care experiences, based on PHDS items at baseline and at the 6-month follow-up, are in Table 3. The results for our sample are comparable to those found in other clinics and other studies.²⁴ These indicators of primary care components did not change significantly over time or show statistically significant differences between groups.

We also examined whether there were any differences in demographic characteristics between children who followed-up at 6 months and those who did not and found only that children in the group lost to follow-up were 5 months older on average, possibly because these older children were not as likely to return to the clinic for well-child care.

To our knowledge, this study was the first used to test telephone-based early childhood developmental screening and care coordination in an RCT. We found that children assigned to the intervention group had significantly higher odds of being screened with a validated tool, being referred for evaluation, being found eligible for services, and receiving services compared with those receiving usual care alone, even after adjusting for other variables. Families working with 211LA were also connected to community services beyond EI and ECSE programs, including ECE and behavioral health and family-support services. The intervention did not seem to significantly affect components of primary care received.

Our study has several limitations. First, the sample of 152 children enrolled is relatively small, comes from a single clinic system, and was fairly homogeneous in terms of race and ethnicity, primary language, and other sociodemographic variables. Therefore, it is unclear that our findings would be generalizable to other patient populations or other clinic settings. Also, the large effect of the intervention seen in this study may be related in part to the low baseline rates of screening and referral done at our partner clinic. The intervention might reveal smaller effect sizes in other settings. However, the rates of screening and enrollment in services among families in the intervention group were also substantially higher than those reported nationally in previous studies.^8,9,25,26 Finally, we were limited by available data sources, including parent report and medical record review, both of which may have omissions or inaccuracies, as well as a substantial loss to follow-up of parents in both groups over time.

Despite these limitations, this study has important implications for clinical practice and future research. Many studies have revealed that child health providers struggle to achieve recommended practices in developmental screening and connection to services for families with developmental-behavioral concerns. Families and providers face multiple barriers limiting the numbers of children enrolled in appropriate interventions, and these barriers are even more pronounced for children from low-income families or families of racial and ethnic minority, families with low levels of parental education, and families for whom English is not a primary language. 211LA has developed a care coordination program that appears to be more effective than a primary care clinic in connecting children with developmental-behavioral concerns to intervention services and has the potential to address barriers for these vulnerable families. 2-1-1 call centers around the United States have comparable resource directories and data tracking systems and have the potential to replicate this program with training of existing staff. This model could potentially be implemented at relatively low cost as a centralized utility that could be shared among multiple clinics in a region, reducing the burden of care coordination for clinic staff and more effectively linking families with services. Future studies are needed to test whether this program can be scaled up and replicated in other clinic sites, with different patient populations, and ultimately in other 2-1-1 call centers.

In addition to these demonstrated improvements in connecting children with developmental delays to intervention services, partnerships between primary care and 2-1-1 may have the potential to benefit children and families who have psychosocial risks associated with developmental-behavioral concerns and poor school readiness.^27,–30 Because 2-1-1 call centers and resource directories are designed to address basic needs and connect families to social services, they have the potential to support vulnerable families well beyond linkages to developmental services. The initial description of the 211LA Developmental Screening and Care Coordination Program demonstrated that families are willing to engage in screening even when in crisis and that the young children of families calling 2-1-1 have a higher risk of developmental-behavioral concerns.¹⁷ This current study suggests that 211LA may also be an exceptionally effective partner for primary care clinics struggling with their own developmental screening and care coordination. Future research should examine whether this model can improve long-term child health, development, and educational outcomes.

Telephone-based early childhood developmental screening and care coordination through a 2-1-1 call center has the potential to improve screening rates, referrals, and connection to services compared with usual pediatric primary care alone. This is a model that could be scaled up through a national network of 2-1-1 call centers.

211LA

2-1-1 Los Angeles County

CMOAR

Clínica Monseñor Oscar A. Romero

ECE

early care and education

ECSE

Early Childhood Special Education

EI

early intervention

MCHAT-R

Modified Checklist for Autism in Toddlers, Revised

PEDS

Parental Evaluation of Developmental Status

PHDS

Promoting Healthy Development Survey

RA

research associate

RCT

randomized controlled trial