Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jul 1.
Published in final edited form as: J Pediatr. 2021 Mar 10;234:227–235. doi: 10.1016/j.jpeds.2021.03.009

Early and Repeated Screening Detects Autism Spectrum Disorder

Andrea Trubanova Wieckowski 1, Taralee Hamner 1, Sarah Nanovic 1, Katelynn S Porto 2, Kirsty L Coulter 2, Sherief Y Eldeeb 1, Chi-Ming A Chen 2, Deborah A Fein 2, Marianne L Barton 2, Lauren B Adamson 3, Diana L Robins 1
PMCID: PMC8238814  NIHMSID: NIHMS1683036  PMID: 33711288

Abstract

Objective:

To evaluate timing and accuracy of early and repeated screening for autism spectrum disorder (ASD) during well-child visits.

Study design:

Using a longitudinal study design, toddlers (n=5784) were initially screened at 12 (n=1504), 15 (n=1228), or 18 (n=3052) months during well-child visits, and rescreened at 18, 24, and 36 months. Of those screened, 368 toddlers attended an ASD evaluation after a positive screen and/or a provider concern for ASD at any visit.

Results:

Screens initiated at 12 months yielded an ASD diagnosis significantly earlier than at 15 months (P = .003, d = 0.99) and 18 months (P < .001, d = 0.97). Cross-group overall sensitivity of the initial screen was .715 and specificity was .959. Repeat screening improves sensitivity (82.1%), without notably decreasing specificity (all > 93.5%). Screening at 18 months resulted in significantly higher positive predictive value (PPV) than at 12 months (X2 (1, n=221) = 9.87, P = .002, OR = 2.60) and 15 months (X2 (1, n=208) = 14.57, P < .001, OR = 3.67). With repeat screening, PPV increased for all screen groups but the increase was not significant.

Conclusion:

Screening as early as 12 months effectively identifies many children at risk for ASD. Children screened at 12 months receive a diagnosis of ASD significantly earlier than peers who are first screened at later ages, facilitating earlier intervention. However, as the sensitivity is lower for a single screen, screening needs to be repeated.

Keywords: Autism spectrum disorder, Screening, Toddlers


There is a critical need to evaluate timing and accuracy of universal screening to promote early identification of autism spectrum disorder (ASD), as the average age of diagnosis in the United States is currently over four years.1 Early detection is crucial for accessing early intervention services, which ultimately influence behavioral, social, and functional outcomes.2 Although broadband screening tools are routinely used within the first years of life to monitor a child’s development toward milestones, ASD is not well captured by these inventories. The American Academy of Pediatrics (AAP) recommends ASD-specific screening at 18 and 24 months of age,3, 4 facilitating early diagnosis.5, 6 Given that symptoms often emerge by the first birthday,7 and many diagnoses are stable by 14 months of age,6, 8 it is important to consider screening younger than 18 months.

Several promising ASD screening tools have been developed for children as young as 12 months of age;3 however, evidence to support universal screening at this age is lacking. Further, whereas many children are identifiable before the second birthday, some, especially those with greater cognitive abilities, may not demonstrate detectable symptoms until later.9, 10 Screening at younger ages may increase both false positives and missed cases. Heterogeneity in age of symptom emergence may require screening at multiple timepoints to maximize sensitivity. In this study, under research conditions, we aim to examine the accuracy of early and repeated screening for ASD during well child visits beginning at 12, 15, or 18 months.

Methods

Participants

A total of 5784 toddlers were screened at 12 (n = 1504), 15 (n = 1228), or 18 (n = 3052) months at well-child visits in Atlanta (n = 6 practices), Connecticut (n = 11 practices), and Pennsylvania (n = 10 practices; Table I). Of those screened, 368 participants (244 males, 124 females) attended an ASD evaluation after a positive screen and/or a provider concern for ASD. To be recruited into the study, children had to attend a pediatric practice visit at one of the participating practices, be within the screening age range, and have an English or Spanish speaking caregiver. Toddlers were excluded if their parent enrolled them in the study but never completed screening (n = 42), or their first screen was outside of the screening age range (n = 135).

Table 1.

Sample Demographic Characteristics

12m Screen 15m Screen 18m Screen
Total
n=1504
At Risk1
n=314
Evaluated2
n=113
Total
n=1228
At Risk1
n=305
Evaluated
n=94
Total
n=3052
At Risk1
n=378
Evaluated
n=161
Sex (N)
 Male 803 200 75 629 190 62 1505 223 107
 Female 690 112 38 586 115 32 1485 150 54
 Not reported 11 2 0 13 0 0 62 5 0
Race (N)
 White/Caucasian 812 119 46 916 203 62 1650 142 62
 Black/African American 312 92 27 74 27 9 616 111 46
 Asian 57 14 7 52 21 5 142 23 12
 Native Hawaiian/Other Pacific Islander 2 1 0 1 0 0 3 1 1
 American Indian/Alaska Native 11 1 1 2 1 1 6 4 4
 Bi- or multiracial 126 22 8 96 19 10 241 33 13
 Other 37 10 3 25 10 0 88 23 13
 Unknown 147 55 21 62 24 7 306 41 10
Ethnicity (N)
 Hispanic 293 95 32 145 53 16 470 96 44
 Non-Hispanic 897 158 60 659 143 50 2045 223 96
 Unknown 314 61 21 424 109 28 537 59 21
Maternal Education (N)
 Less than high school or GED 77 36 14 18 11 5 147 36 14
 High school/GED 204 63 19 86 29 6 474 87 44
 Technical or trade school 40 12 5 47 19 4 114 18 5
 Some college 185 53 15 168 53 20 405 69 29
 College degree 439 72 24 453 113 31 946 82 34
 Advanced degree 498 58 28 408 65 25 646 49 23
 Unknown 61 20 8 48 15 3 320 37 12
1

At Risk includes children who screened positive on any screener and/or for who providers indicated an ASD concern, and therefore were referred for an evaluation.

2

One child in group was evaluated in error. Child was not at risk, but was evaluated, and received an ASD diagnosis.

Measures

Parent-Report Screeners

The Infant-Toddler Checklist (ITC)11 is a 24-item screener for social and communication delays in children 6–24 months. The ITC was originally intended to be administered in combination with the Systematic Observation of Red Flags (SORF).12 The ITC was used as a standalone questionnaire in a previous screening study to detect ASD in a general pediatric sample of children aged 10 to 16 months.13 Total score below the 10th percentile indicates risk. ITC was found to have a high degree of internal consistency (α = .86–.92), high sensitivity for ASD (.933)12, and high sensitivity and specificity for ASD and communication delays (both .889), because it is designed to detect broader communication delays.11 The ITC was used at 12 months.

The First Year Inventory – Lite Version 3.1b (FYI-L)14 is a short form of the FYI15 to screen for ASD at 12 months. Scores at or above the 95th percentile indicate risk. Although there is no published study using FYI-L at 15 months, communications with the first author (Grace Baranek) noted utility. The FYI-L shows low sensitivity (.363 at 12 m, .370 at 15 m) but high specificity (.953 at 12 months, .944 at 15 months).16 The FYI-L was used at 12 and 15 months.

The Modified-Checklist for Autism in Toddlers, Revised, with Follow-Up (M-CHAT-R/F)5, 17 screens for ASD in children 16 children to 30 months; all children complete the initial 20 items, and those in the moderate risk range complete the Follow-Up. The M-CHAT-R/F shows high sensitivity (.854) and specificity (.993), with adequate internal consistency (Cronbach α = 0.79) in a large scale study with concurrent, research-reliable case confirmation.5 Lower ranges of sensitivity and specificity estimates have been reported in large community studies that identified ASD outcomes via prospective record review.18 The M-CHAT-R/F was used at all screens from 15 months onward.

Provider Concern

Providers were asked to indicate concern about ASD at every screening, using clinical judgment. Across all initial evaluations, providers indicated whether or not they had a concern 71.79% of the time (12m: 66.95%, 15m: 83.31%, 18m: 69.54%). Forms left blank were classified as “no concern.”

Evaluation

Diagnostic classification incorporated data from the Autism Diagnostic Observation Schedule, 2nd Edition,19 an observational assessment of social communication and repetitive behavior, and a caregiver ASD interview: the Toddler Autism Symptom Inventory (TASI)20 for children younger than 36 months or the Autism Diagnostic Interview, Revised21 for those 36 months and older. The TASI demonstrates excellent discrimination (Area Under Curve = 0.915, 0.888) between ASD and non-ASD groups (https://mchatscreen.com/tasi/). Additional measures included the Mullen Scales of Early Learning22 to assess receptive and expressive language, visual reception, and fine motor skills; the Vineland Adaptive Behavior Scales, 2nd Edition23 interview to assess communication, daily living, socialization, and motor skills; and a Medical, Developmental, and Family History form.

Procedure

Pediatric practices were randomized to begin screening toddlers at 12 (11.5 – 13.99), 15 (14.0 – 16.99), or 18 (17.0 – 21.99) months, stratified by university site, practice size, and percent of patients on Medicaid. Screenings occurred during well-child visits using either electronic or paper forms (and phone for M-CHAT-R Follow-Up). Pediatric providers were asked to invite all patients in the designated age range to participate. Reduced compliance screening children younger than 18 months, also reported in prior studies,24 resulted in a larger sample size for the 18m group compared with the 12 and 15m groups. Screens were offered in both English and Spanish at Connecticut and Philadelphia sites, whereas Atlanta was an English-only site. Parents reviewed a consent script and waived documentation of consent at screening; they provided informed consent at evaluations. Following initial screens, toddlers in the 12- and 15-month groups were rescreened at 18 (17.0 – 21.99), 24 (22.0 – 29.99), and 36 (34.0 – 38.99) months. Toddlers in the 18-month group were rescreened at 24 and 36 months. Recruitment and initial screening occurred between April 2015 and February 2018, with rescreening occurring between August 2015 and October 2019.

Study participants who were screen positive and/or whose provider had an ASD concern at any screen visit were invited for a no-cost diagnostic evaluation. Two different screeners were given at the 12 and 15m visits (order was counterbalanced across participants); an at-risk score on either screener was considered to be a positive screen. Evaluations took place at the university clinic or pediatric office between May 2015 and September 2019, by a team supervised by a licensed psychologist, certified school psychologist, or developmental pediatrician. Clinical best-estimate diagnosis was based on International Classification of Diseases, tenth edition (ICD-10)25 criteria. Diagnoses of Atypical Autism, Childhood Autism, and Asperger Syndrome were grouped into an ASD classification. When ASD was ruled out, other developmental disabilities (DD) were considered or the child was determined to have no diagnosis. Caregivers received oral and written feedback about diagnoses and recommendations for intervention. Participants who did not receive an ASD diagnosis at their initial evaluation but later screened positive were invited for re-evaluation. Families who initially declined evaluations were asked to complete future rescreens; caregivers could attend evaluation until the child turned 52 months old. Overall, only 22 children received more than one diagnostic evaluation; therefore, initial diagnosis was utilized for the analyses.

Analytic Approach

Descriptive information indicates the number of toddlers diagnosed with ASD who were identified by an initial positive screen, a positive result during a rescreen, or a provider concern at the initial or rescreen check-up. Chi-square analyses were run to compare attendance rates between the three initial screen ages, providing odds ratio (OR) for 2×2 analyses and effect size (ø) for 3×2 analyses, where OR cannot be calculated. For toddlers who attended an evaluation, chi-square analyses were run to evaluate the proportion of children who received an ASD diagnosis. To evaluate differences in age of ASD diagnosis, Analyses of variance (ANOVA) were run to explore age of ASD diagnosis between the three initial screen ages, including effect size (ηp2 or d). These analyses were followed up by t-tests to explore differences between age groups. Missing data were excluded from corresponding analyses.

Psychometric properties for each group were evaluated for initial visits, and for all visits (initial + repeat). Sensitivity (ie, detecting ASD when truly present) was calculated by dividing number of true positive cases (TP; i.e., positive screen and received ASD diagnosis) by total number of children who received an ASD diagnosis. Specificity (i.e., detecting nonASD cases) was calculated by dividing true negative cases (TN; i.e., screened negative and did not receive ASD diagnosis) by total number of children who did not receive an ASD diagnosis. Positive predictive value for ASD (PPVASD; likelihood that positive result is a true ASD case) was calculated by dividing TP for ASD by all screen positives, whereas PPV for any developmental disability (PPVDD; likelihood that positive result indicates ASD or another DD) was calculated by dividing TP for ASD or DD diagnosis by all screen positives. Negative predictive value (NPV; i.e., likelihood that a negative result is a true nonASD case) was calculated by dividing TN for ASD by all screen negatives. NPV was only calculated for ASD as TN can only be presumed for ASD cases but not for all DD cases in this study. TP, TN, and false positive (FP) cases were determined by screener results; false negative (FN) cases were detected when the provider indicated concern or, for initial visit psychometrics only, when a positive rescreen led to a diagnosis. PPV was calculated based on the subsample of screen positive children who were seen for evaluation. Chi-square analyses were run to compare components of sensitivity (TP to FN), specificity (TN to FP), PPVASD (TP for ASD to FP) and PPVDD (TP for ASD or DD to FP) for initial screen group ages as well as for comparing initial to repeat screening accuracy. For these calculations, children who were not seen for an evaluation were presumed to be nonASD.

Results

Screening Results

The majority of toddlers diagnosed with ASD were identified by a positive screen at the initial enrollment check-up, for all groups (Figure). However, some toddlers with ASD in all groups were identified by a later rescreen and/or provider concern.

Figure 1.

Figure 1.

Flowchart indicating screening results, identification for an evaluation, and final diagnoses for participants with initial screen age at 12 months (A), 15 months (B), and 18 months of age (C). ASD = Autism Spectrum Disorder, DD = Developmental Disability, ND = No Diagnosis.

12 month Screening:

Of the 1504 toddlers in the 12m group, 31 were diagnosed with ASD; 20 (64.52%) were identified by positive screen at 12 months, and 5 toddlers (16.13%) were missed at 12 months, but identified by a positive rescreen. Five toddlers with ASD (19.35%) screened negative, but a provider noted concern (three at initial screen, two at rescreen). One additional child was evaluated in error and thus was excluded from age of diagnosis analyses.

15 month Screening:

For toddlers in the 15m group (n = 1228), 18 were diagnosed with ASD; 13 (72.22%) screened positive at 15 months, two (11.11%) were identified by positive rescreen, and three (16.67%) were identified by provider concern at later visits.

18 month Screening:

For the 18m group (n = 3052), 74 toddlers were diagnosed with ASD; 55 (74.32%) had a positive screen at 18 months, five children (6.76%) were identified on rescreen, and 14 toddlers (18.92%) did not screen positive but were identified by provider concern for ASD (four at initial screen, and ten at later visits).

Table 2 describes evaluation attendance and diagnostic outcomes. Across all three initial screening ages, many toddlers who screened positive and/or a provider indicated concern for ASD did not attend the diagnostic evaluation. The three groups differed significantly on attendance rates (X2 (2, n=892) = 19.15, P < .001, ø = .15) with the toddlers in the 18m group more likely to attend the evaluation (50.00%) compared with the 12m (40.07%; X2 (1, n=604) = 5.98, P = .015, OR = 1.50) or 15m groups (32.64%; X2 (1, n=610) = 18.83, P < .001, OR = 2.06); no significant difference was found between the 12m and 15m groups (P = .065, OR = 1.38). Of those who attended a diagnostic evaluation, most attended after being identified by screen and/or provider concern at a single visit (i.e., prompt evaluation). As seen in Table 2, among toddlers who attended an evaluation, the proportion diagnosed with ASD differed by group (X2 (2, n=368) = 21.81, P < .001, ø = .24), with more toddlers in the 18m group receiving an ASD diagnosis (45.96%) compared with the 12m (27.43%; X2 (1, n=274) = 9.64, P = .002, OR = 2.25) or 15m groups (19.15%; X2 (1, n=255) = 18.50, P < .001, OR = 3.59), with no significant difference between the two younger age groups (P = .163, OR = 1.60).

Table 2.

Evaluation Attendance and Diagnosis Based on Positive Screen and/or Provider Concern

Initiated Based on First Visit Initiated Based on Subsequent Visit
Initial Positive Screen & No Concern1 Initial Negative Screen & Concern Initial Positive Screen & Concern Initial Negative Screen/No Concern & Later Positive Screen Initial Negative Screen/No Concern & Later Concern Initial Negative Screen/No Concern & Later Positive Screen & Concern
12m Screens 240 15 13 24 12 10
 Prompt Evaluation2 ASD 14 (5.8%) 3 (20.0%) 3 (23.1%) 0 1 (8.3%) 5 (50.0%)
Non-ASD 60 (25.0%) 5 (33.3%) 6 (46.2%) 2 (8.3%) 2 (16.7%) 2 (20.0%)
 Delayed Evaluation3 ASD 3 (1.3%) 0 0 0 1 (8.3%) 0
Non-ASD 5 (2.1%) 0 0 0 0 0
 Declined Evaluation --- 158 (65.8%) 7 (46.7%) 4 (30.8%) 22 (91.7%) 8 (66.7%) 3 (30.0%)
15m Screens 250 7 19 6 21 2
 Prompt Evaluation ASD 5 (2.0%) 0 0 1 (16.7%) 3 (14.3%) 1 (50.0%)
Non-ASD 52 (20.8%) 4 (57.1%) 8 (42.1%) 1 (16.7%) 6 (28.6%) 0
 Delayed Evaluation ASD 7 (2.8%) 0 1 (5.3%) 0 0 0
Non-ASD 4 (1.6%) 0 1 (5.3%) 0 0 0
 Declined Evaluation --- 182 (72.8%) 3 (42.9%) 9 (47.4%) 4 (66.7%) 12 (57.1%) 1 (50.0%)
18m Screens 181 39 79 42 33 4
 Prompt Evaluation ASD 15 (8.3%) 3 (7.7%) 33 (41.8%) 2 (4.8%) 10 (30.3%) 3 (75.0%)
Non-ASD 56 (30.9%) 4 (10.3%) 17 (21.5%) 4 (9.5%) 4 (12.1%) 0
 Delayed Evaluation ASD 3 (1.7%) 1 (2.6%) 4 (5.1%) 0 0 0
Non-ASD 2 (1.1%) 0 0 0 0 0
 Declined Evaluation --- 105 (58.0%) 31 (79.5%) 25 (31.6%) 36 (85.7%) 19 (57.6%) 1 (25.0%)
1

No concern category includes children for whom providers specifically indicated no concerns as well as children whose provider did not select whether or not they had a concern.

2

Prompt Evaluation indicates child attended an evaluation based on ASD risk from a single timepoint. The single timepoint could have been at either initial screen or any rescreen.

3

Delayed Evaluation indicates that the child was identified by screen positive and/or provider indicated concern at least two timepoints prior to attending an evaluation.

Age of ASD Diagnosis

The average age of ASD diagnosis was 23.62 months; age of ASD diagnosis differed significantly for the three groups (F(2,123) = 9.69, P < .001; ηp2= .14). Toddlers in practices randomized to start screening at later ages received an ASD diagnosis at a significantly older age (15m group: 25.20 ± 7.40 months, P = .003, d = 0.99; 18m group: 25.35 ± 7.91 months; P < .001, d = 0.97), compared with toddlers in practices that were randomized to the 12m group (18.57 ± 5.91 months); there was no statistically significant difference between 15m and 18m groups (P = .937, d = 0.02). In addition, we compared age of ASD diagnosis between prompt evaluation (attended after a single visit indicated ASD risk; this may have occurred at initial or repeat screening ages) and a delayed evaluation (attended after risk indicated at more than one visit). Children enrolled at 12 months showed a significantly earlier age of ASD diagnosis with prompt evaluation (17.56 ± 5.33 months) compared with delayed evaluation (25.05 ± 6.91 months, t(1,28) = −2.52, P = .018, d = 1.21), Similarly, children enrolled at 18 months showed a significant difference in age of diagnosis (prompt evaluation: 24.64 ± 7.60 months; delayed evaluation: 31.27 ± 8.44 months, t(1,72) = −2.31, P = .024, d = 0.83). For children enrolled at 15 months, there was no statistically significant difference in age of ASD diagnosis between prompt (25.58 ± 8.53 months) and delayed evaluation (24.72 ± 6.24 months, t(1,16) = .24, P = .817, d = 0.12).

Psychometric Properties

Table 3 displays the psychometric properties for each group, both for initial visit only and for initial plus repeat visits. Results indicate that initial screening yields good sensitivity (i.e., detecting ASD when truly present) for all ages (71.5%), with slightly greater sensitivity at 15m and 18m compared with 12m, although the difference was not significant. Specificity (i.e., detecting nonASD) is high (above 90.0%) for all groups; overall specificity of the initial screen across groups was 95.9%. Screening at 18 months resulted in significantly higher positive predictive value for ASD (PPVASD) compared with 12m (X2 (1, n=221) = 9.87, P = .002, OR = 2.60) or 15m groups (X2 (1, n=208) = 14.57, P < .001, OR = 3.67). PPV for any developmental disability (PPVDD; likelihood that positive result indicates ASD or another DD) is 59.3% for 12m, 59.0% for 15m, and 89.2% for 18m groups. The likelihood ratio of positive screen (LR+) indicates strong support for ASD screening for all age groups (LR+ 11.28–28.58), with the strongest LR+ in the 18m group.

Table 3.

Psychometric Properties of Initial Screening at 12, 15, and 18 months and Repeated Screening for Each Initial Age

TP FP FN TN Sens (95% CI) Spec (95% CI) PPV NPV LR+
12m initial screen only 20 71 11 1240 .645 (.477–.813) .946 (.933 – .958) .220 .991 11.944
12m+ later screens 25 75 6 1208 .806 (.667–.946) .942 (.929–0.954) .250 .995 13.897
15m initial screen only 13 65 5 954 .722 (.515–.929) .936 (.921–.951) .167 .995 11.281
15m+ later screens 15 66 3 948 .833 (.661–1.00) .935 (.920–.950) .185 .997 12.815
18m initial screen only 55 75 19 2773 .743 (.644–.843) .974 (.968–.980) .423 .993 28.577
18m+ later screens 61 79 13 2729 .824 (.738–.911) .972 (.966–.978) .436 .995 29.429
All groups initial screen 88 211 35 4967 .715 (.636–.795) .959 (.954–.965) .294 .993 17.439
All group any screen 101 220 22 4885 .821 (.753–.889) .957 (.951–.962) .315 .996 19.093

Note. For each screen age, psychometric properties are displayed for initial visit only as well as for initial plus repeat visits directly underneath. TP = True Positive cases, FP = False Positive cases, FN = False Negative cases, TN = True Negative cases, Sens = sensitivity, Spec = specificity, CI = Confidence Interval, PPV = Positive Predictive Value for ASD, NPV = Negative Predictive Value, LR+ = likelihood ratio of positive screen [sensitivity/(1-specificity)].

Repeat screening improves sensitivity for all ages, without notably decreasing specificity (all specificity > 93.5%). Chi-square analyses of sensitivity for initial screen compared with all visits (initial + rescreens) across all age groups revealed a small, but statistically significant difference (X2 (1, n=246) = 3.86, P = .049, OR = 1.83). However, the increase in sensitivity of repeat screening was not significant for any group (all Ps > .155). PPVASD and PPVDD increased at repeat screening (PPVASD:12m = 25.0%, 15m = 18.5%, 18m = 43.6%; PPVDD: 12m = 61.0%, 15m = 59.3%, 18m = 89.1%); differences were not significant across all groups (X2 (1, n=620) = 0.300, P = .584, OR = 1.10), or for any of the specific ages (all Ps > .62). The likelihood ratio of positive screen similarly increased for all ages, with the highest LR+ for the 18m group (29.43).

Discussion

The goal of this study was to examine early and repeated screening for ASD, balancing the benefits of very early detection against anticipated low sensitivity and specificity of screening at young ages. The AAP recommends ASD-specific screening begin at 18 months of age;3, 4 however, our results suggest that screening at the 12 month well-child visit identifies many children at risk. Children screened at 12 months receive a diagnosis of ASD significantly earlier than peers first screened at 15 or 18 months, which can lead to earlier engagement with intervention services. The average age of diagnosis for the entire sample was 23 months, more than 2 years earlier than the national median of 51 months.1 This study adds to the literature demonstrating the efficacy of early, ASD-specific screening in the first years of life.5, 13

A strength of the current study is the ability to assess the incremental utility of repeat screening from 12 months forward using longitudinal data. Our results indicate that repeat screening throughout early development improves sensitivity for detection of ASD without notably decreasing specificity, regardless of the age at which screening began. Although the main focus in this study is on the sensitivity and specificity trade-off, the balance between false positive vs. false negative case identification is also important. Results of this study indicate repeated screening reduced the number of false negative cases by 13 and increased false positive cases by 9. These cases represent 13 additional children who were diagnosed and referred to intervention services prior to the second birthday. Although false positive cases contribute to a greater demand for assessments in the community, seven out of the nine children in this group were diagnosed with another developmental disorder and benefitted from a referral for evaluation. Thus, repeat screening is a relatively low-cost method of enhancing detection for young children which does not disproportionally increase false positive cases.

Researchers have observed variable early symptom onset in infants with ASD,9 and some children may not be readily identified by provider concerns or parent-reported screening at the earliest screening ages. Results from the present study suggest that repeated screening early in life is more effective than one-time screening in identifying children at risk. this study identified that toddlers who screen negative or who do not attend an evaluation after a positive screen at 12, 15, or 18 months of age need at least one additional screen at either 24 or 36 months. Data suggest that the need for repeated screening is higher for younger screen ages, but further studies are needed to confirm this finding, and to specifically compare 24 vs. 36 months for rescreen efficacy. In addition, our results highlight the need for use of standardized screening instruments, in addition to relying on provider concern to maximize early identification of ASD.

Screening at 12 months proved valuable for identifying and diagnosing children before the second birthday, but sensitivity, specificity, and likelihood of a positive screen (LR+) were highest when screening was initiated at 18 months. PPVASD was also significantly higher for the 18m group than younger groups. Thus, children may be more readily identified by screening at this age. Nonetheless, despite lower sensitivity and specificity at 12 months, many children with ASD are detected and will benefit from earlier engagement in ASD-specific interventions. Repeated screening balances the tradeoff of lower sensitivity of 12m screening. It is more difficult to interpret results of the 15m screening as the overall screen positive rate was higher than for the other two groups with lower attendance compared with the 18m group. Provider and parent confidence may have been lower for 15m screens. Future research may need to modify the 15m screeners to determine utility of screening at this visit. Finally, it is important to consider that even 36m screening is not likely to identify all cases of ASD;26 CDC prevalence studies show fewer cases based on record review at age 427 vs. age 8.1, 28 Children with mild symptoms and/or strong compensatory mechanisms may not be detected until early elementary school.

To maximize positive outcomes by capitalizing on early neuroplasticity of the brain,29 early identification of children with autism must be coupled with efforts to reduce the age at which intervention services begin. Treatments for children as young as 12 months of age have been shown to be effective;30 intervention prior to the second birthday has the potential to significantly impact the child’s skill acquisition and developmental outcomes.31 AAP guidelines indicate that children identified to be at risk for ASD by screening tools should be referred for both diagnostic evaluation and early intervention services. The latter are provided through the Individuals with Disabilities Education Act IDEA in the U.S.; some regions begin ASD-related services on suspicion of ASD without a formal diagnosis, to avoid the compounded delay of both evaluation and intervention. This is also recommended for children diagnosed with other conditions, such as global developmental delay.

The recommendation for universal toddler screening during well-visits requires a balance of evidence and capacity. Although our study was completed within research conditions, procedures were embedded within routine primary care visits without additional time or personnel allotted for screening. Exploring solutions to universal screening barriers are encouraged, such as a two-tiered system to alleviate system overload.32 Others have documented near universal screening during primary care visits through screening via electronic health record systems.18 This is promising, as study participants in our sample experienced a shorter wait between a positive screen and the evaluation compared with community waitlists. Further, screening within primary care contributes to an earlier age of diagnosis.18, 33

Although our findings do not point to a single optimal schedule for early and repeated screening, they highlight that universal toddler screening should start early (prior to age 2) and be repeated at least once by age 3. The higher proportion of FP to TP cases at younger ages does not strongly advocate for universal screening at 12 or 15m, but taken together, initial screening prior to second birthday, as well as repeated screening, demonstrate strongest sensitivity and specificity.

These results should be interpreted in light of several limitations. Comparison of screening at different ages is complicated by the age ranges of available screeners. As a result, differences across 12, 15, and 18 months may relate to differences in the screening tools themselves. Future research should examine the performance of individual screeners at multiple ages and different screening tools at the same age, as well as consider alternative thresholds to determine ASD risk. Additionally, many at-risk toddlers did not attend an evaluation even after repeated invitations. This limitation is not due to a lack of availability of evaluations, nor is it unique to this study,5, 13, 34 although the disproportionate attendance rates by age further complicates interpretation. For both cross-sectional and longitudinal studies, further research is needed into how both providers and parents view and act upon the results of screening during infancy, and what factors influence evaluation attendance after a positive screen or provider indicated concern. Provider concern enhanced screening by approximately 16–17% in our sample, representing the minority of identified cases. This is consistent with work suggesting clinical judgment alone is insufficient to consistently identify ASD in toddlers,35 nor should it override a positive screen, given the importance of early diagnosis and start of services. Future research may consider factors that influence provider reported concern in the context of continuous care. Furthermore, our research design did not permit verification of low-risk cases as true negatives because evaluation only occurred for children at risk for ASD and whose parents chose to attend an evaluation. It is likely that some children who screened negative at one or more visits and some children who screened positive but declined evaluation will be identified with ASD at older ages; ongoing follow-up of samples such as this one will be critical to identify prospective sensitivity.36

Conclusion

We suggest that screening for ASD should begin as early as 12 months of age, consider both the results of parent-report screeners and provider concerns and include repeated screening. Initiation of screening for ASD at 12 months of age is not currently recommended by the AAP. However, the benefits of accessing early intervention for children are substantial; repeated screening allows for the detection of children whose symptoms are not identifiable until later in toddlerhood. An early, and multiple time screening schedule may best achieve favorable outcomes for all children.

Supplementary Material

1

Acknowledgements

We thank the healthcare providers, as well as toddlers and their families for participating in the study. In addition, we thank the members of the research team who participated in data collection, including Katherine Sand, who coordinated the multi-site study and created the FileMaker Pro database for the study. We also thank Grace Baranek, PhD and the insp!ire lab team for their aid in use of the FYI-L in this study, and other study advisors for feedback on design and data collection: Paul Dworkin, MD, Jennifer Pinto-Martin, PhD, Ho-Wen Hsu, MD, and Mark A. Greenstein, MD.

Supported by the National Institute of Child Health and Human Development (R01HD039961 [to D.R.]). This report reflects the views of the authors and may not reflect the opinions or views of the NIH. D.R., M.B., and D.F. are co-owners of M-CHAT LLC, which receives royalties from parties that license use of the M-CHAT in electronic products. No royalties were received for any of the data presented in the current study. D.R. serves on the advisory board for Quadrant Biosciences Inc. The other authors declare no conflicts of interest.

Abbreviations:

AAP

American Academy of Pediatrics

ASD

autism spectrum disorder

FYI-L

First Year Inventory, Lite

ICD-10

International Classification of Diseases, 10th Revision

ITC

Infant-Toddler Checklist

LR+

Likelihood ratio of positive screen

M-CHAT-R/F

Modified Checklist for Autism in Toddlers, Revised with Follow-Up

PPV

positive predictive value

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  • 1.Maenner MJ, Shaw KA, Baio J. Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years—Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2016. MMWR Surveillance Summaries. 2020;69(4):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Zwaigenbaum L, Bauman ML, Choueiri R, Kasari C, Carter A, Granpeesheh D, et al. Early intervention for children with autism spectrum disorder under 3 years of age: recommendations for practice and research. Pediatrics. 2015;136(Supplement 1):S60–S81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Hyman SL, Levy SE, Myers SM. Identification, evaluation, and management of children with autism spectrum disorder. Pediatrics. 2020;145(1)doi: 10.1542/peds.2019-3447. [DOI] [PubMed] [Google Scholar]
  • 4.Lipkin PH, Macias MM. AAP Council On Children With Disabilities, Section On Developmental and Behavioral Pediatrics. Promoting optimal development: identifying infants and young children with developmental disorders through developmental surveillance and screening. Pediatrics. 2020;145(1). [DOI] [PubMed] [Google Scholar]
  • 5.Robins DL, Casagrande K, Barton M, Chen C-MA, Dumont-Mathieu T, Fein DA. Validation of the Modified Checklist for Autism in Toddlers, Revised with Follow-Up (M-CHAT-R/F). Pediatrics. 2014;133(1):37–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Pierce K, Gazestani VH, Bacon E, Barnes CC, Cha D, Nalabolu S, et al. Evaluation of the diagnostic stability of the early autism spectrum disorder phenotype in the general population starting at 12 months. JAMA Pediatr. 2019;173(6):578–587. doi: 10.1001/jamapediatrics.2019.0624. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zwaigenbaum L, Bryson S, Garon N. Early identification of autism spectrum disorders. Behav Brain Res. 2013;251:133–46. doi: 10.1016/j.bbr.2013.04.004. [DOI] [PubMed] [Google Scholar]
  • 8.Ozonoff S, Young GS, Landa RJ, Brian J, Bryson S, Charman T, et al. Diagnostic stability in young children at risk for autism spectrum disorder: a baby siblings research consortium study. J Child Psychol Psychiatry. 2015;56(9):988–98. doi: 10.1111/jcpp.12421. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Bacon EC, Courchesne E, Barnes CC, Cha D, Pence S, Schreibman L, et al. Rethinking the idea of late autism spectrum disorder onset. Dev Psychopathol. 2018;30(2):553–569. doi: 10.1017/S0954579417001067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Jashar DT, Brennan LA, Barton ML, Fein DA. Cognitive and adaptive skills in toddlers who meet criteria for autism in DSM-IV but not DSM-5. J Autism Dev Disord. 2016;46(12):3667–3677. doi: 10.1007/s10803-016-2901-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Wetherby AM, Prizant BM. Communication and symbolic behavior scales: Developmental profile. Paul H Brookes Publishing Co.; 2002. [Google Scholar]
  • 12.Wetherby AM, Brosnan-Maddox S, Peace V, Newton L. Validation of the Infant—Toddler Checklist as a broadband screener for autism spectrum disorders from 9 to 24 months of age. Autism. 2008;12(5):487–511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Pierce K, Carter C, Weinfeld M, Desmond J, Hazin R, Bjork R, et al. Detecting, studying, and treating autism early: the one-year well-baby check-up approach. The Journal of pediatrics. 2011;159(3):458–465.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Baranek G, Watson L, Crais E, Turner-Brown L, Reznick J. The First Years Inventory – Lite Version 3.1b (FYI-Lite v3.1b). University of North Carolina at Chapel Hill. 2014. [Google Scholar]
  • 15.Baranek G, Watson L, Crais E, Reznick J. The First Year Inventory. Unpublished instrument Available from: bcrais@ med unc edu The University of North Carolina at Chapel Hill Chapel Hill. 2003. [Google Scholar]
  • 16.JHS Chen Y-JC, Watson Linda R., Crais Elizabeth R., Turner-Brown Lauren, Grace T Baranek V Validation of the FYI-Lite as a Short Screening Tool for Infants at Risk for Autism. Presented at: The American Occupational Therapy Association Annual Conference & Expo; April 19–22, 2018; Salt Lake City, Utah. [Google Scholar]
  • 17.Robins DL, Fein DA, & Barton M The Modified Checklist for Autism in Toddlers, Revised, with Follow-Up (M-CHAT-R/F). Self-published. https://mchatscreen.com. 2009. [DOI] [PMC free article] [PubMed]
  • 18.Guthrie W, Wallis K, Bennett A, Brooks E, Dudley J, Gerdes M, et al. Accuracy of autism screening in a large pediatric network. Pediatrics. 2019;144(4):e20183963. [DOI] [PubMed] [Google Scholar]
  • 19.Lord C, Rutter M, DiLavore P, Risi S, Gotham K, Bishop S. Autism Diagnostic Observation Schedule Second Edition (ADOS-2) Manual (Part 1): Modules 1–4. Torrance, CA: Western Psychological Services. 2012. [Google Scholar]
  • 20.Barton M, Boorstein HC, Cordeaux C, Dumont-Mathieu T, Haisley L, Herlihy L, et al. Toddler ASD Symptom Interview - Version 1. Self-published, current version at https://mchatscreen.com/tasi/. 2015.
  • 21.Rutter M, Le Couteur A, Lord C. The Autism Diagnostic Interview-Revised (ADI-R). Los Angeles, CA: Western Psychological Services. 2003:659–685. [Google Scholar]
  • 22.Mullen EM. Mullen Scales of Early Learning. AGS; Circle Pines, MN; 1995. [Google Scholar]
  • 23.Sparrow S, Cicchetti D, Balla D. Vineland Adaptive Behaviour Scales: Survey Form, 2nd edn (Circle Pines, MN: American Guidance Service; ). 2005. [Google Scholar]
  • 24.Dietz C, Swinkels S, van Daalen E, van Engeland H, & Buitelaar JK. Screening for autistic spectrum disorder in children aged 14–15 months. II: Population screening with the Early Screening of Autistic Traits Questionnaire (ESAT). Design and general findings. Journal of autism and developmental disorders. 2016;36(6), 713–722. [DOI] [PubMed] [Google Scholar]
  • 25.World Health Organization. ICD-10: international statistical classification of diseases and related health problems: tenth revision. 2004.
  • 26.Robins DL. Prevalence counts: commentary on “Prevalence and characteristics of autism spectrum disorder among 4-year-old children in the autism and developmental disabilities monitoring network”. Journal of Developmental & Behavioral Pediatrics. 2016;37(1):80–82. [DOI] [PubMed] [Google Scholar]
  • 27.Christensen DL, Maenner MJ, Bilder D, Constantino JN, Daniels J, Durkin MS, et al. Prevalence and characteristics of autism spectrum disorder among children aged 4 years—Early autism and developmental disabilities monitoring network, seven sites, United States, 2010, 2012, and 2014. MMWR Surveillance Summaries. 2019;68(2):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Baio J, Wiggins L, Christensen DL, Maenner MJ, Daniels J, Warren Z, et al. Prevalence of autism spectrum disorder among children aged 8 years—autism and developmental disabilities monitoring network, 11 sites, United States, 2014. MMWR Surveillance Summaries. 2018;67(6):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Klin A, Micheletti M, Klaiman C, Shultz S, Constantino JN, Jones W. (2020) Affording autism an early brain development re-definition. Development and Psychopathology. 1–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Rogers SJ, Vismara L, Wagner AL, McCormick C, Young G, Ozonoff S. Autism treatment in the first year of life: a pilot study of infant start, a parent-implemented intervention for symptomatic infants. Journal of autism and developmental disorders. 2014;44(12):2981–2995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.MacDonald R, Parry-Cruwys D, Dupere S, Ahearn W. Assessing progress and outcome of early intensive behavioral intervention for toddlers with autism. Res Dev Disabil. 2014;35(12):3632–44. doi: 10.1016/j.ridd.2014.08.036. [DOI] [PubMed] [Google Scholar]
  • 32.Khowaja M, Robins DL, Adamson LB. Utilizing two-tiered screening for early detection of autism spectrum disorder. Autism. 2018;22(7):881–890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Carbone PS, Campbell K, Wilkes J, Stoddard GJ, Huynh K, Young PC, et al. Primary care autism screening and later autism diagnosis. Pediatrics. 2020;146(2). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Miller JS, Gabrielsen T, Villalobos M, Alleman R, Wahmhoff N, Carbone PS, et al. The each child study: systematic screening for autism spectrum disorders in a pediatric setting. Pediatrics. 2011;127(5):866–871. [DOI] [PubMed] [Google Scholar]
  • 35.Gabrielsen TP, Farley M, Speer L, Villalobos M, Baker CN, Miller J. Identifying autism in a brief observation. Pediatrics. 2015;135(2):e330–e338. [DOI] [PubMed] [Google Scholar]
  • 36.Robins DL. How do we determine the utility of screening tools? Autism. 2020;24(2):271–273. doi: 10.1177/1362361319894170. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

RESOURCES