Numerous studies describe the features of state EI programs, such as the characteristics of children and families referred, evaluated, and enrolled, and the kinds of services received,24,26 but very few studies examine outcomes using rigorous methodology sufficient for attributing causal impact to EI services. The vast majority of outcome studies do not have a control or comparison group to measure against the group receiving EI services, relying instead on a single group’s pre-intervention and post-intervention data, and many studies use parent self-reports through surveys to understand child and family outcomes, rather than using direct assessments of children’s progress by trained observers.14 As one researcher put it, a major “challenge to assessing impact of EI services on child outcomes is the mandate that services be available to all children who meet eligibility requirements. Conventional experimental approaches are, therefore, unobtainable and unethical” (pp. 74–75).27 The studies that do employ treatment and control groups tend to examine specific programs unique to a particular community and often have small sample sizes, limiting generalizability.28,29 Another EI researcher has cited “a critical gap in the literature linking [EI] service use and functional outcomes” (p. 2),30 and still others have acknowledged “intense professional debate surrounding claims of the effectiveness of [E]arly [I]ntervention” (p. 320).G
Despite these limitations in the research, some consistent evidence has emerged from rigorous studies of EI programs, particularly those focused on children born premature or low birthweight, demonstrating that participation in services can boost children’s developmental trajectories. More rigorous studies, with comparison groups when possible and larger sample sizes, would be valuable to broaden the evidence base and determine the impact of EI on more diverse groups of infants and toddlers.
The peer-reviewed research on EI to date does not generally focus on the impact of a specific state policy lever on child and family outcomes, although some correlational studies have examined whether state eligibility thresholds impact participation in EI, given that eligibility is one of the aspects of the program that states independently determine. The research and most recent available data on the effect of broader or narrower eligibility thresholds find mixed results in terms of whether broader thresholds lead to more eligible children actually receiving services.42,47,48,49,50,77 Using a broad eligibility threshold means that a state may serve children with less severe delays than states that report moderately inclusive or narrow criteria, but states with broader eligibility do not necessarily serve more children than states with more restrictive criteria. The percentage of children ages 0 to 3 served in a given state is tied more closely to states’ efforts to marshal resources for EI (e.g., funding and personnel) and their investments in Child Find and outreach activities.
The research discussed here meets our standards of evidence for being methodologically strong and allowing for causal inference, unless otherwise noted. Each strong causal study reviewed has been assigned a letter, and a complete list of causal studies can be found at the end of this review, along with more details about our standards of evidence and review method. The findings from each strong causal study reviewed align with one of our eight policy goals from Table 1. The Evidence of Effectiveness table (Table 2) displays the findings associated with Early Intervention (beneficial, null,v or detrimental) for each of the strong studies (A through Gvi) in the causal studies reference list, as well as our conclusions about the overall impact on each studied policy goal. The assessment of the overall impact for each studied policy goal weighs the timing of publication and relative strength of each study, as well as the size and direction of all measured indicators.
Of the seven causal studies included in this review, two studiesB,C examined how outcomes differed by race or ethnicity (beyond simply presenting summary statistics or controlling for race/ethnicity). Where available, this review presents the analyses’ causal findings for subgroups by race/ethnicity and other aspects of variation, such as family socioeconomic status. A rigorous evaluation of a policy’s effectiveness should consider whether the policy has equitable impacts and should assess the extent to which a policy reduces or exacerbates pre-existing disparities in economic and social wellbeing.
Table 2: Evidence of Effectiveness for Early Intervention Services by Policy Goal
|Policy Goal||Indicator||Beneficial Impacts||Null Impacts||Detrimental Impacts||Overall Impact on Goal|
|Parental Health and Emotional Wellbeing||Maternal Self-Confidence||B, D||Mixed|
|Maternal Role Satisfaction||D|
|Nurturing and Responsive Child-Parent Relationships||Maternal Sensitivity Toward Infants||B||Trending* Null|
|Optimal Child Health and Development||Cognitive Assessment Scores||A, B, C, D, F, G||Positive|
|Behavior Assessment Scores||C|
|Maternal Report of Infant Temperament||D|
|Receptive Language Skills||E|
|Expressive Language Skills||E|
*Trending indicates that the evidence is from fewer than two strong causal studies or multiple studies that include only one location, author, or data set.
Parental Health and Emotional Wellbeing
Although many studies examine family and caregiver outcomes after participation in EI services for their infants and toddlers, most studies employ designs that preclude causal conclusions. For example, many studies draw on surveys that lack a control group, rely solely on parent self-report, have a small sample size, and/or may be vulnerable to self-selection bias.31,44 Three strong studies measured some indicators of parent wellbeing.B,D,E
A 1988 experimental study of infants born between 1980 and 1981 examined the long-term results of an 11-session EI program that sought to offer mothers support as they adjusted to parenting a low birthweight infant over the first 3 months of life.D Mothers who received the treatment scored significantly higher on scales of maternal self-confidence when the child was 4 years old (a difference of 1.3 points in the mean score) and on maternal role satisfaction at age 6 months (a difference of 3.1 points in the mean score) than the control group.vii The study found no significant differences in maternal anxiety between the groups. A 2009viii randomized study of 138 Black mothers and their low birthweight, premature infants found that after an 8-session, 20-week EI program, mothers in the treatment group had significantly higher self-efficacy scores (the mean score was 1.2 points higher on the 40-point Maternal Self-Efficacy scale).B
A 2015 study recruited toddlers between 2009 and 2013 to participate in a 28-session EI program focused on language development.E The study focused on caregiver-implemented interventions for children with language delays, and although the study found significant impacts on child receptive language (discussed in the Optimal Child Health and Development section of this review), there were no significant findings for reducing caregiver stress.E
Nurturing and Responsive Child-Parent Relationships
The 2009 study of Black infants and their mothers, introduced in the Parental Health section, examined maternal sensitivity using the Maternal Behavioral Q-Set (MBQ) instrument.B Observers rated the sensitivity of mother-child interactions during a 2-hour period, and although the EI treatment group scored higher overall than the control group on maternal sensitivity, the difference was not statistically significant.
Optimal Child Health and Development
Cognitive, Motor, and Behavioral Outcomes
Evidence from meta-analysesA,F,G and randomized controlled trialsB,C,D,E (RCTs) suggests that EI services make a positive difference for the cognitive, motor, and behavioral skills of infants and toddlers with disabilities. For example, a 1987 meta-analysis of 31 studies found an average effect size of 0.62 for the cognitive skills of children in EI services.F The authors considered this impact to be a “moderate, positive effect” (p. 652)F and they noted some key features of programs that produced the greatest effects: They had more structured curricula, they enrolled children before 6 months old, and they involved parents to a greater degree. For example, services that involved parents and children together produced an average effect size of 0.74, compared to 0.44 for programs that involved either parents or children separately. The effect size for motor skills was somewhat smaller, at 0.43.F
A widely cited RCT begun in the 1980s, called the Infant Health and Development Program, involved 985 families with infants born low birthweight and premature at eight hospitals across the country.C The infants who were assigned to the EI treatment, which involved intensive services throughout the first 3 years of life, saw better cognitive and behavioral outcomes at age 3 (an average of 9 points higher on the Stanford-Binet Intelligence Quotient test, and an effect size of -0.2 on a behavior problems scale) than the control group, and a greater degree of participation in the program was associated with higher scores on the cognitive assessment. A 2006 follow-up to the study found positive long-term impacts at age 18 on those who had participated in the trial compared to the control group, on both academic and risk behavior assessments.H
The 1988 RCT involving infants born low birthweight, discussed in the Parental Health section of this review, found that children who received EI services from a nurse showed significantly higher cognitive scores at 36 and 48 months than similar children who did not receive the intervention (a difference of 9.5 points at 36 months and 12.9 points, or approximately 0.8 standard deviations, at 48 months on the McCarthy Scales of Children’s Abilities).D The McCarthy General Cognitive Index used in this study has a mean standard score of 100 and a standard deviation of 16 points. The study also found that infants in the experimental group scored better on a scale of infant temperament (the Carey Infant Temperament Questionnaire, in which lower scores reflect more favorable infant temperament) at 6 months old. Low birthweight infants in the EI group scored 1.3 points lower than their counterparts in the control group on the 4-point scale.D
A study using propensity score matchingix to examine the outcomes of premature and low birthweight infants found that those who received EI services showed significantly better cognitive development trajectories by age 24 and 36 months (using the Bayley Scales of Infant Development at 16 months and the Stanford-Binet scales at 24 and 36 months) than those who did not receive services.69 The study also found that among children who received EI services, those whose mothers reported higher levels of maternal support (on an index of emotional, financial, respite, and other supports offered by the mother’s family and social networks) saw better cognitive outcomes relative to the control group. This finding underscores that children’s developmental trajectories are closely tied to their caregivers’ wellbeing.
The 2009 randomized intervention discussed previously, with 138 Black infants born low birthweight and premature, found that the extremely low birthweight infants (born weighing less than 1,000 grams) who were assigned to the EI group scored 10 points higher on the Bayley Mental Development Index (a scale with a mean of 100 and standard deviation of 15) than those infants assigned to the control group.B The study did not find significant effects for infants born weighing more than 1,000 grams. In addition, the study found that socioeconomic status mediated the intervention effects; the Bayley scores for infants living below the federal poverty threshold did not show a significant improvement based on the EI treatment.B The authors hypothesized that mothers facing resource deficits may have experienced greater stress and may have had less time and less support from partners to successfully implement what they learned from the program. For example, the study found that “mothers living above poverty thresholds were significantly more sensitive during interactions with their infants than mothers living in poverty, and this finding may be a direct result of the former having more financial, material, and perhaps interpersonal resources at their disposal” (p. 160).B Research should continue to investigate the impacts of poverty on the success of EI interventions to ensure that programs work for the most socially and economically disadvantaged infants and toddlers.
A 1998 meta-analysis of studies that employed a “randomized, prospective, longitudinal design with appropriate control groups” (p. 321) found effect sizes for cognitive outcomes that ranged from 0.50 to 0.75, depending on the characteristics of the samples examined.G Finally, a 2009 meta-analysis of 25 RCTs that analyzed various EI programs for premature infants found significant differences between the cognitive scores of treatment participants as compared to control groups (a weighted mean difference of 9.7 points at 36 months for studies using the McCarthy and Stanford-Binet scales).A Differences that were identified at 36 months were no longer detected at 5 years old.A
Language and Communication Skills
Studies with rigorous methods have also shown positive impacts of EI services on infants’ and toddlers’ communication skills. In fact, the 1987 meta-analysis discussed previously found the greatest effect size for language skills compared to other developmental domains, at 1.17.F A 2015x RCT examined the effects of a caregiver-led communication intervention on toddlers facing language delays, and the authors found that the treatment produced significant positive effects on receptive, but not expressive, language skills (a 0.27 to 0.35 effect size for receptive language skills, depending on the instrument used).E
- An impact is considered statistically significant if p≤0.05. Results with p-values above this threshold are considered null or nonsignificant.
- Studies H and I are longer-term analyses for Study C and are not included in Table 2.
- Maternal self-confidence was measured using the Seashore Self-Confidence Rating Paired Comparison Questionnaire, in which a total score is measured by counting the “number of items on which the mother rates herself at least as competent as…five other potential caretakers (spouse, own mother, another experienced mother, a pediatric nurse, and a physician)” (p. 548).D Maternal role satisfaction was measured using a semistructured interview when the infant was 6 months old, and two independent raters used 4-point Likert scales to score the interviews on 10 questions.D
- Recruitment for this study took place between 2002 and 2004.
- The study did not meet our standards for causal evidence as a result of a small sample size coupled with a non-randomized design, and therefore it is not reflected in Table 2, but the findings are promising and align with the results of causal studies reviewed.
- Recruitment took place between 2009 and 2013.