The influence of Spanish language on autism spectrum disorder screening: a systematic review

Abstract

Objective

In the United States, autism spectrum disorder (ASD) is significantly less diagnosed in Spanish-speaking populations compared to other racial/ethnic groups. Structural barriers such as transportation, insurance, and time have been documented within this population. However, many families also report language as a barrier. The present systematic literature review aims to evaluate research findings on the influence of Spanish language and culture on the ASD screening process.

Method

Five databases were searched using key words relating to the aim of the paper to identify potential articles to review. Inclusion criteria included: (1) sample of participants diagnosed with ASD/being evaluated for ASD, (2) Latinx populations that included Spanish-speaking participants, (3) a reference to language included as a research aim or study outcome, and (5) a focus on the screening process or measures for ASD. A total of 132 articles were initially identified.

Results

After reviewing for inclusion criteria, 22 articles were included. In total, six countries were represented across all included papers. Major themes emerged regarding barriers to care, cultural awareness, translation barriers, intervention, knowledge/awareness, and screener evaluation.

Discussion

Findings indicate that language and sociocultural differences can affect knowledge development, communication within healthcare settings, and result in a limited number of translated resources available for the heterogeneous population of Spanish-speakers.

Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by impairments in social communication and the presence of stereotyped behaviors, interests, and activities (American Psychiatric Association (APA) 2013). In the United States (U.S.), the average age of ASD diagnosis is four years old (Christensen et al. 2016); however, early diagnosis, which can occur as early as 18 months of age, is critical for prognostic outcomes, as it allows access to interventions targeting language and behavioral development (MacDonald et al. 2014). Consequently, ASD screening procedures have become an important facet of developmental surveillance. The American Academy of Pediatrics (AAP) recommends formal ASD screening at ages 18 and 24 months (Hyman et al. 2020), and an estimated 63% of pediatricians report using developmental screening tools in their practice (Lipkin et al. 2020).

ASD has a prevalence rate of 1 in 44 children, and prevalence is similar across racial and ethnic groups (Maenner et al. 2021). However, Latinx, individuals with Latin American heritage, children with ASD are more likely to be under-identified (Maenner et al. 2021) or misdiagnosed (Magaña et al. 2013, Christensen et al. 2016) compared with non-Latinx Black and White children. Furthermore, Latinx children receive ASD diagnoses at an even later age than non-Latinx Black and White children (Jo et al. 2015). Without timely and appropriate diagnostic identification, beginning with appropriate screening, Latinx children are at a disadvantage with respect to accessing treatment and services, consequently impacting health equity for this population.

A plurality of factors may contribute to this differential rate of diagnosis, including different symptom presentation patterns (Magaña and Smith 2013) and socio-structural barriers (Bornstein and Cote 2004). Latinx families voice that limited ASD information, stigma, poverty, lack of empowerment to utilize services, dismissal of parental concern by providers, and length of diagnostic process are predominant barriers to seeking ASD-related care (Zuckerman et al. 2014). Families report that these barriers have led to dismissal of concerns, normalization of red flag behaviors, and perpetuation of mistrust in the medical system (Zuckerman et al. 2014).

The literature has traditionally ignored language when explaining factors influencing ASD detection among primarily Spanish-speaking individuals and Latinos, preferring instead to focus on the above cultural factors (Troxel et al. 2018). However, evidence suggests that language plays a central role in health outcomes. For example, a population-based study of Latinx individuals demonstrated those with high levels of acculturation (i.e. effective English communication and endorsing mainstream ideas related to U.S. current events) reported greater ease in accessing care, understanding ASD, and receiving ASD treatment compared to Latinx individuals with low levels of acculturation (Voelkel et al. 2013). The present study aims to address this gap by evaluating the influence of language in the ASD screening process within the context of cultural factors. Therefore, the goal of this review is to strengthen the understanding of how the Spanish language and Spanish culture combined may impact ASD screening with Spanish-speaking families. As screening represents a critical first step in the pathway to early diagnosis and intervention, it is essential to understand the ways in which use of Spanish language interacts with and affects the screening process for both families and providers. The present review provides an evaluation of Spanish language from a sociocultural and relational perspective rather than an examination of linguistics like grammar, or syntax. Specifically, this systematic review aims to synthesize research findings on the influence of Spanish language on the ASD screening process. This review is exploratory in nature; however, the following research questions were considered: 1) How does Spanish language impact the relationship between patients and providers? 2) What are some of the barriers related to screening for ASD in Spanish? 3) What current measures exist for screening for ASD in Spanish, how are they administered, and how is the quality of those screeners? 4) How can we improve ASD screening in the U.S. for Spanish-speaking populations? These questions were addressed by summarizing evaluative studies of Spanish-translated ASD screeners including psychometrics, strengths and weaknesses, and bias risk. Although a major aim of the study includes improving ASD screening services for Spanish-speaking families in the U.S., studies were included from other countries to increase representation and broaden interpretation of current Spanish screeners. This review also addresses the clinical implications related to Spanish language on ASD screening and provides recommendations for use of these screeners in clinical settings. Finally, research on the steps of translating, adapting, and validating pre-existing ASD screeners for use in the heterogeneous Spanish-speaking population both in the U.S. and internationally is discussed.

Method

Search strategy

The search strategy was developed in collaboration with the lead author and a medical research librarian. The search strategy included the following databases: PubMed, Scopus, PsycINFO, and PsycArticles. The final search was completed on February 1st, 2023. The Cochrane Database of Systematic Reviews (Wiley) was explored for existing systematic reviews and protocols. Records were deduplicated using EndNote and then loaded into Covidence for further screening of duplicates. The review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al. 2009) and was registered before data extraction on the international prospective register of systematic reviews PROSPERO website (registration number CRD42021266041). Detailed search terms are included in the supplementary materials.

Inclusion and exclusion criteria

Inclusion criteria included (1) sample of participants diagnosed with ASD/being evaluated for ASD, (2) Latinx populations that included Spanish-speaking participants, (3) a reference to language included as a research aim or study outcome, and (5) a focus on the screening process or measures for ASD. Conference abstracts, dissertations/theses, and book chapters, were not considered due to them not being peer reviewed in nature. Systematic reviews and meta-analyses were excluded as well; however, these papers served as a tool to find any additional studies for the present review. In addition, papers were excluded if they used non-Spanish-speaking bilingual samples (i.e. German-English bilingual groups, etc.), or if they focused on Spanish cultural themes without discussing Spanish language. Based on the inclusion criteria, some studies were excluded if they focused on an ASD screening tool in Spanish but did not include examination of language in the study or if they instead focused on race, ethnicity, and culture without mention of language as a primary or secondary aim. Additionally, studies examining parent or provider perspectives of a Spanish screening tool were excluded if language was not included as an aspect of the evaluation.

Study selection

Literature search results were downloaded and organized in EndNote and exported to Covidence. The first and second authors each independently screened the titles and abstracts from the initial search results using inclusion criteria. Full-text screening was also conducted independently by the first two authors. Discrepancies were resolved through mediated discussion by the first author and second author.

Risk of bias and quality assessment

The risk of bias was assessed using standardized risk of bias assessment tools for qualitative and case control studies. The Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2; Whiting et al. 2011) was used to evaluate risk of bias and applicability concerns related to patient selection, index tests, reference standard, and flow and timing in cross-sectional studies evaluating ASD screening measures.

Data extraction and synthesis

The third and fourth authors extracted data independently from each eligible study using the Covidence extraction form developed by reviewers before formal data extraction (Appendix A). Discrepancies were resolved through mediated discussion by the first, third, and fourth author. For each article, data were identified for the following variables: screener, reliability and validity data, administration language, translation methodology, sample size, inclusion/exclusion criteria, study location, gender distribution, racial and ethnic distributions, diagnoses, age data, advantages of the screeners, limitations of the screeners, and clinical implications.

Data analysis

Analyses were conducted in three steps. First, interrater reliability was conducted and evaluated during the screening process between the first and second authors. Second, descriptive statistics, such as frequency data and measures of central tendency, were compiled to assess screener psychometrics and other clinically-relevant details. Third, a brief thematic review modeled from Braun and Clarke (2006) characterized the clinical implications of each study into specific themes. Given the brevity of the data included for the thematic analysis, all themes were established via discussion between the first and second author. The first two authors completed the first three steps of the brief thematic review: (1) review the main findings of each included paper a minimum of three times to improve understanding of the data, (2) identify preliminary themes that best characterized the findings, and (3) refined and finalize the themes that best represent the observed clinical implications. The final step was completed by the first author and involved pulling the main findings from each article, identifying specific examples that best represent elements of the themes, interpreting the themes in the results and discussion sections of this paper, and organizing the themes in a table.

All reported results including strengths and limitations are summaries of the information provided by the authors of each cited article. Therefore, when describing these components in the results section, the current authors are providing an overview of the information listed in each article. Elaboration on these summaries is provided in the discussion section of this article in which the authors of the current study provide reflection on the results.

Results

As shown in the PRISMA Flow Chart (Figure 1), the initial search yielded 132 results. Following deduplication, 112 unique articles remained. Study selection via title and abstract screening and full text review yielded 24 articles that met inclusion criteria. Of these 24 studies, only 22 were included in the final analysis based on data availability. Figure 1 details the process of screening records, retrieving reports, and eligibility determination according to PRISMA guidelines. Table 1 summarizes study characteristics for all 22 articles included in this review, described in more detail below. The results section is organized into three parts. First, we describe the characteristics of the reviewed studies. Second, we report on the results of the Spanish-translated screeners. In this section, we discuss psychometrics, strengths and limitations of the screeners, barriers and facilitators to implementation, and clinical implications. Within the strengths and limitations subsection, we further separate each type of screener to clearly highlight specific strengths and weaknesses attributable to the screener. Third, we document the QUADAS-2 bias and quality assessment across the papers.

Figure 1.

PRISMA flow diagram.

From: Page et al. (2021).

Table 1.

Study sample characteristics.

Study	Sample description	Sample Size (N)	Age Mean(SD)	Gender (% Female)	Race/Ethnicity (%)	Country	Inclusion and Exclusion Criteria
Albores-Gallo et al. (2012)	Two groups of Mexican children and their caregivers: 1) typically developing and 2) ASD diagnoses	456	Children: 4.46 years Mothers: 31.79 years Fathers: 35.79 years	41.9	Not reported	Mexico	Children from both sexes between 18 to 72 months of age Presumptive diagnosis of ASD Children from the general community with the same age range than the clinical group Excluded if children had comorbid chronic conditions that would bias the MM-CHAT scores
Atwood et al. (2015)	Peruvian children with documented developmental disabilities and challenging behaviors and their parent/caregiver Diagnoses included: ASD, Down Syndrome, Cerebral Palsy, cognitive deficiencies, and genetic syndromes	71 dyads	Children: 41 months Parents: Not reported	36.6	Not reported	Peru	Some inclusion criteria cited from Mayo-Ortega et al. (2012) Participation in prior mass screening project Parental consent to participate in follow-up study No exclusion criteria reported
Bölte et al. (2018)	Representatives from 21 European countries involved in COST-ESSEAa	16 language groupsb	Not reported	Not reported	Not reported	21 European Countries c	Country involvement in COST- ESSEA No exclusion criteria included
Canal-Bedia et al. (2011)	Stage 1: Children aged 18–36 months living in Salamanca and Zamora and high-risk children ages 18–24 Stage 2: Children ages 18–36 months living in Madrid	Stage 1: 2480 children Stage 2: 2055 children	Not reported	Stage 1: 47.5 Stage 2: 46.2	Not reported	Spain	Stage 1: High risk children with developmental age 18–24 months, maximum chronological age 48 months, living in Salamanca and Zamora No exclusion criteria reported Stage 2: Children ages 18–36 months, living and receiving medical care in Madrid No exclusion criteria reported
Coelho-Medeiros et al. (2019)	Chilean children aged 16–30 months, recruited from the UC Christus Health network	120 children	22.47 months	35.8	Not reported	Chile	For the ASD group, children needed to present with high clinical suspicion of ASD after an evaluation by specialists Exclusion criteria included severe medical conditions or mental health concerns
Cuesta-Gómez et al. (2016)	Toddlers living in Buenos Aires	420 children	Not reported	Parents: 77.6 Children: 44.8	Not reported	Argentina	Consumers of most visited medical centers Representative of CABA neighborhoods No exclusion criteria reported
DuBay et al. (2021a)	Spanish-speaking primary caregivers	25 caregivers	Parents: 33.2 years Children: 11.8 months	Parents: 96.0 Children: 44.0	Hispanic 100 96.0 born in Latin America 1.0 born in US	United States	No inclusion criteria reported No exclusion criteria reported
DuBay et al. (2021b)	Community sample of caregivers who took the M-CHAT-R either in Spanish or in English 3724 records reviewed	Risk Analyses n = 2591 Missingness Analyses n = 1024	Risk Analyses Parents: not reported Children: 21.9 months Missingness Analyses Parents: not reported Children: 22.1 months	Risk Analyses Parents: not reported Children: 48.75 Missingness Analyses Parents: not reported Children: 49.55	74.3 (English-speaking) 25.7 (Spanish-speaking)	United States	Records excluded if patients outside of the 16–30 month age range or did not contain age data Records excluded if not English or Spanish Records excluded if used the older M-CHAT-R Records excluded if a duplicate child
DuBay et al. (2022)	US-based Latin American Spanish-speaking caregivers of children 8–16 months of age	380	Parents: 31 years Children: 11.3 months	Parents: 93.6 Children: 48.3	Range of countries of origin, including almost all Latin American countries where Spanish is considered an official languaged	United States	Spanish-speaking families No exclusion criteria reported
Eugenin et al. (2015)	Caregivers and their toddlers representative of medium-low and low-income residents of a major Chilean city	200 dyads	Mother-SJ: 32 years Mother-LP: 27 years Father-SJ: 34 years Father-LP: 29 years Child-SJ: 19.9 months Child-LP: 22.1 months	Parents: Not reported Children-SJ: 39.8 Children-LP: 47.6	Not reported	Chile	No inclusion criteria reported No exclusion criteria reported
Fombonne et al. (2012)	Children with/without diagnosis of Pervasive Developmental Disorder	563 children	8(2.3) years	Parents: 91.3 Children: 32.9	Not reported	Mexico	Diagnosis of PDD-NOS from CLIMA clinic (case group) Age- and gender-match to case (control group) No exclusion criteria included
Kimple et al. (2014)	Toddlers seeking primary care	Not reportede	Not reported	Not reported	Not reported	United States	All children who presented for 18- or 24-month well-visits between July 2010 and July 2012 No exclusion criteria
Kuhn et al. (2021)	Low-income, racial/ethnic minority parents and their child who screened positive on the M-CHAT-R/F	309 parents and children	Child age at screening 21 months Parent age at screening 30 years	Children: 30.4 Parents: Not reported	Child race/ethnicity not reported Parents: 55.8 = Black, non-Hispanic 29.2 = Hispanic, any race 8.6 = Other, non-Hispanic 6.3 = White, non-Hispanic	United States	Needed to screen positive on the M-CHAT-R/F No exclusions were based on language or comorbid conditions
Lugo-Marín et al. (2019)	Spanish native adults	N = 294 46 adults with ASD 41 ASD relatives 17 adults with SSDf 190 non clinical adults	36.77 years	51.3	All participants were born in Spanish territory	Spain	ASD group – needed a prior evaluation and ASD diagnosis; below average IQ was exclusion criteria No inclusion/exclusion criteria reported for ASD-Relatives group SSD group – needed a diagnosis of SSD and could not have acute psychotic symptoms, intellectual difficulties, or ASD Non-ASD group – not having received an ASD diagnosis, nor any other psychiatric disorder diagnosis prior to the evaluation
Magán-Maganto et al. (2020)	Children in Salamanca, Zamora, and Valladolid, Spain presenting for Well Baby Check-ups	6625 children	(months) 14–22 range: 18.2(.72) 23–36 range: 24.5(1.23)	14–22 rangeg: 47.31 23–36 range: 48.56	Not reported	Spain	Children born in study catchment areas between data collection dates and who attended routine 18- and 24-month well baby screenings Excluded if caregiver did not complete the FUI or evaluation due to communication
Miller et al. (2011)	Toddlers with a 2006 date of birth, between the ages of 14 and 24 months at the start of the screening period and between 21 and 32 months of age at the end. Diagnoses include: ASD or non-ASD	796 children	Not reported	Not reported	hBlack (3.2) Pacific Islander (4) Asian (4.2) Native American (2) White Non-Hispanic (55) Other (0.2)	United States	Child between 14 and 30 months of age seeking care from community pediatric practice during 6-month screening period No exclusion criteria included
Morales-Hidalgo et al. (2017a)	Spanish sample of children from community and clinical settings between the ages of 4–12 years	1476	7.51 years	40.17	Autochthonous (86.32)	Spain	No inclusion criteria reported No exclusion criteria reported
Reyes et al. (2021)	Mothers of school-aged children (Kindergarten through Grade 12) who spoke either English or Spanish Children were either at high or low risk for internalizing, externalizing, or tic disorders. Diagnoses include: ASD, non-ASD	199 mothers English: 104 Spanish: 95	Spanish Male: 10.6 years Spanish Female: 10.52 years English Male: 10.66 years English Female: 11.04 years	Spanish: 46.32 English: 43.27	Hispanic/Latino (58.3) African American (11.7) White Non-Hispanic (14.6) Mixed (11) Asian American (2) Native American (1.5) Other (.5)	United States	A random sample of school-aged children enrolled in PLAY-MHi whose teachers completed a BASC-2 and whose caregiver agreed to a follow-up interview Excluded if follow-up interview was completed by someone other than biological mother
Rosenberg et al. (2018)	Children between the ages of 30 and 68 months of age enrolled in the SEEDj Study, recruited via clinical/educational sources due to previous developmental concerns (CE) or birth certificates to represent the general population (BC) Diagnoses included: ASD, DD, no ASD/DD	2,557 children CE: 1653 BC: 904	(months) CE Children: 55.8(7.2) BC Children: 55.6(7.6)	Parents: 99 Children: 33.9	Maternal Race: Black (17.7) White (71.2) Other (1.2) Maternal Ethnicity: Hispanic (11.5) Non-Hispanic (88.5)	United States	Born between September 1, 2003 and August 31, 2006 in a study catchment area Child resided with caregiver at least 18 years of age and who consistently cared for child since they were ≤ 6 months Caregiver able to speak English (or, in California and Colorado, English or Spanish)
Scarpa et al. (2013)	Toddlers in Southwest Virginia presenting for well-child visits	447 toddlers	Not reported	Parents: 86.4 Children: 45.2	Hispanic (27.1) Black (25.3) White (25.1) Mixed (8.9) Asian (1.8) Native American (0.7) Not reported (11.2)	United States	Caregiver and their child attending 18- or 24-month well visit Between ages 16 and 32 months Excluded if outside of age range and/or if primary language of caregiver was neither English nor Spanish
Windham et al. (2014)	Primarily low-income Latinx families covered by government-funded insurance	1760 families	Parents not reported	Parent gender not reported Children: 50.1	Not reported	United States	Child between 14 and 30 months of age Seeking care from community pediatric practice during 6-month screening period No exclusion criteria included
Zuckerman et al. (2013)	Primary Care Pediatricians (PCPs)	267 PCPs	Not reported	Not reported	Not reported	United States	Excluded if provided invalid contact information

^a COST-ESSEA: European Cooperation in Science and Technology; (www.cost.eu) action ‘Enhancing the Scientific Study of Early Autism’ (ESSEA; for details see www.cost-essea.com) participated. COST-ESSEA is a network of over 60 scientists from 23 European countries, including three LAMI countries (Romania, FYR Macedonia, and Turkey), aiming to improve function in early autism research.

^b These 16 language/cultural groups in practice typically use the same versions of adapted diagnostic tools: Austria/Germany/Switzerland (German), Belgium/The Netherlands (Dutch), Czech Republic, Finland, France (including information on French-Canadian adaptations of tools), Hungary, Iceland, Italy, Israel, Macedonia, Poland, Portugal (including information on Brazilian adaptations of tools), Spain, Sweden/Norway (Scandinavia), Romania, and UK/Ireland (English; even US versions of instruments included here).

^c 21 European Countries.

^d Countries of origin include: Mexico (N = 80), United States (N = 64), El Salvador (N = 48), Honduras (N = 33), Guatemala (N = 28), Nicaragua (N = 4), Costa Rica (N = 2), Panama (N = 1), Puerto Rico (N = 20), Cuba (N = 13), Dominican Republic (N = 8), Colombia (N = 28), Venezuela (N = 11), Peru (N = 10), Bolivia (N = 8), Chile (N = 3), Argentina (N = 1), Ecuador (N = 1), and Paraguay (N = 1).

^e only analyzed screener results from medical record, 338 questionnaires from pre-transition and 251 from post transition to WHS (589).

^f Schizophrenia spectrum disorders.

^g these values are (n) and not (%).

^h only reported for 524 of the 796 children screened.

ⁱ PLAY-MH: Project to Learn About Youth-Mental Health, a multi-site (Colorado, Florida, Ohio, South Carolina) study aiming to estimate the prevalence, treated prevalence, and co-occurrence of internalizing, externalizing, and tic disorders in school-aged youth (ages 5 to 18).

^j SEED: Study to Explore Early Development – Phase I, a multi-site (California, Colorado, Georgia, Maryland, North Carolina, Pennsylvania), case-control study exploring phenotypes and determinants of ASD.

Characteristics of reviewed studies

Of the 22 included studies, 11 were conducted in the United States (DuBay et al. 2021a, DuBay et al. 2021b, DuBay et al. 2022, Kimple et al. 2014, Kuhn et al. 2021, Miller et al. 2011, Reyes et al. 2021, Rosenberg et al. 2018, Scarpa et al. 2013, Windham et al. 2014, Zuckerman et al. 2013), and the other 11 were conducted in Latin American (Albores-Gallo et al. 2012, Atwood et al. 2015, Coelho-Medeiros et al. 2019, Cuesta-Gómez et al. 2016, Eugenin et al. 2015, Fombonne et al. 2012, Lugo-Marín et al. 2019, Morales-Hidalgo et al. 2017a) or European countries (Bölte et al. 2018, Canal-Bedia et al. 2011, Magán-Maganto et al. 2020). In total, six countries were represented across all included papers.

A total of 7 studies involved parents and their children (Albores-Gallo et al. 2012, Atwood et al. 2015, Eugenin et al. 2015, Kuhn et al. 2021, Reyes et al. 2021, Rosenberg et al. 2018, Windham et al. 2014). The remainder of the studies (n = 15) involved either only parents (n = 4), only children (n = 8), or only non-parent adults (n = 3; e.g. primary care pediatricians) and reported the demographics for each group, respectively. Of the 22 included studies, only 14 studies included participant data related to age (Albores-Gallo et al. 2012, Atwood et al. 2015, Coelho-Medeiros et al. 2019, DuBay et al. 2021a, DuBay et al. 2021b, DuBay et al. 2022, Eugenin et al. 2015, Fombonne et al. 2012, Kuhn et al. 2021, Lugo-Marín et al. 2019, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017b, Reyes et al. 2021, Rosenberg et al. 2018). Of the studies that included age data, the reported age range for children was 11.3 months to 18 years), while the reported age range for adults was 26 years to 37 years. Specific details on participant racial and ethnicity data were frequently omitted; however, race and ethnicity was often broadly discussed when describing the study setting to suggest the majority of samples were Spanish-speaking and from Latin American countries of origin. Only eight studies provided specific breakdowns of race and ethnicity (DuBay et al. 2021a, DuBay et al. 2021b, Kuhn et al. 2021, Lugo-Marín et al. 2019, Miller et al. 2011, Morales-Hidalgo et al. 2017a, Reyes et al. 2021, Rosenberg et al. 2018). When gender was reported, most studies reported relatively even distributions (50% females and 50% males). Yet, four studies did not provide any gender details (Bölte et al. 2018, Kimple et al. 2014, Miller et al. 2011, Zuckerman et al. 2013).

A total of 14 studies did not identify diagnoses present before or after screening (Bölte et al. 2018, Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, Dubay et al. 2021a, DuBay et al. 2021b, DuBay et al. 2022, Eugenin et al. 2015, Kimple et al. 2014, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017a, Reyes et al. 2021, Scarpa et al. 2013, Windham et al. 2014, Zuckerman et al. 2013), and three studies did not clearly identify inclusion and exclusion criteria (Dubay et al. 2021a, Eugenin et al. 2015, Morales-Hidalgo et al. 2017a). Two studies did not explicitly evaluate a Spanish-translated screener, but rather analyzed the potential impact of demographic factors, including Spanish maternal language and appropriate application of interpreters, on screening results (Kuhn et al. 2021, Rosenberg et al. 2018).

Spanish-translated screener results

The Modified Checklist for Autism in Toddlers (M-CHAT) was the most frequently studied screener (n = 14; Albores-Gallo et al. 2012, Bölte et al. 2018, Canal-Bedia et al. 2011, Coelho-Medeiros et al. 2019, Cuesta-Gómez et al. 2016, Dubay et al. 2021b, Eugenin et al. 2015, Kimple et al. 2014, Kuhn et al. 2021, Magán-Maganto et al. 2020, Miller et al. 2011, Scarpa et al. 2013, Windham et al. 2014, Zuckerman et al. 2013), followed by the Ages and Stages Questionnaire (ASQ; n = 2; Windham et al. 2014, Zuckerman et al. 2013), First Year Inventory version 3.1 (FYIv.3.1; n = 2; DuBay et al. 2021a, DuBay et al. 2022), and the Social Communication Questionnaire (SCQ; n = 2; Reyes et al. 2021, Rosenberg et al. 2018).

Translation procedures differed across studies, with some studies utilizing native Spanish-speakers or fluent Spanish-speakers for translation or administration (n = 8; Atwood et al. 2015, DuBay et al. 2021a, DuBay et al. 2021b, Eugenin et al. 2015, Fombonne et al. 2012, Lugo-Marín et al. 2019, Morales-Hidalgo et al. 2017b, Windham et al. 2014) and others hiring formal interpreters or translators (n = 9; Canal-Bedia et al. 2011, Coelho-Medeiros et al. 2019, Cuesta-Gómez et al. 2016, DuBay et al. 2022, Eugenin et al. 2015, Kuhn et al. 2021, Magán-Maganto et al. 2020, Scarpa et al. 2013, Windham et al. 2014). Many studies applied or evaluated the forward-back translation process (n = 7; Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, DuBay et al. 2021a, DuBay et al. 2021b, Fombonne et al. 2012, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017b), some used an informal translation approach (n = 2; Albores-Gallo et al. 2012, Atwood et al. 2015), and one applied a translation with cultural adaptation (TCA) approach (Lugo-Marín et al. 2019). Additionally, many studies also used a cognitive interview or pilot interview stage in the translation process to help further clarify any confusion and interpretation challenges among caregivers (n = 7; Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, DuBay et al. 2021a, DuBay et al. 2021b, DuBay et al. 2022, Eugenin et al. 2015, Magán-Maganto et al. 2020). Moreover, three studies did not describe a translation process (Bölte et al. 2018, Miller et al. 2011, Zuckerman et al. 2013) and five studies used already translated versions of screeners and did not provide explicit details on the original translation procedure (Kimple et al. 2014, Kuhn et al. 2021, Scarpa et al. 2013, Windham et al. 2014).

Of the 22 included studies, seven screeners were administered in written format (Albores-Gallo et al. 2012, Coelho-Medeiros et al. 2019, Fombonne et al. 2012, Kimple et al. 2014, Miller et al. 2011, Scarpa et al. 2013, Zuckerman et al. 2013), six were administered in oral format (Atwood et al. 2015, Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, DuBay et al. 2021a, Reyes et al. 2021, Rosenberg et al. 2018), and 12 included a combination of written and oral administration (Albores-Gallo et al. 2012, Bölte et al. 2018, Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, DuBay et al. 2021b, Eugenin et al. 2015, Kuhn et al. 2021, Lugo-Marín et al. 2019, Magán-Maganto et al. 2020, Miller et al. 2011, Morales-Hidalgo et al. 2017b, Windham et al. 2014). In summary, a variety of translation approaches and administration approaches were applied across the included studies.

Psychometric properties

Not all studies included reliability or validity statistics, and it was unclear in some studies whether a comparison screener was used to validate Spanish screeners. Cronbach’s alpha statistics were provided for 10 studies (Atwood et al. 2015, Coelho-Medeiros et al. 2019, Cuesta-Gómez et al. 2016, Fombonne et al. 2012, Lugo-Marín et al. 2019, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017b, Reyes et al. 2021, Rosenberg et al. 2018, Scarpa et al. 2013). Factors related to education and income levels appeared to impact Cronbach’s alpha for Scarpa et al. (2013), such that Cronbach’s alpha was lower in the lower income sample. Further, Rosenberg et al. (2018) found that Cronbach’s alpha was lower among a sample recruited from the general population compared to a sample recruited from sources with previous developmental concerns. Additionally, test-retest statistics were obtained for three studies (Canal-Bedia et al. 2011, Lugo-Marín et al. 2019, Windham et al. 2014) and internal reliability or consistency was assessed for four studies (Albores-Gallo et al. 2012, Fombonne et al. 2012, Morales-Hidalgo et al. 2017a, Coelho-Medeiros et al. 2019).

Many studies used a range of validity statistics: concurrent (Coelho-Medeiros et al. 2019), discriminant (Albores-Gallo et al. 2012, Coelho-Medeiros et al. 2019, Fombonne et al. 2012), convergent (Albores-Gallo et al. 2012, Lugo-Marín et al. 2019, Morales-Hidalgo et al. 2017b), criterion (Albores-Gallo et al. 2012), and construct validity (Albores-Gallo et al. 2012). Additionally, sensitivity and specificity data were reported for seven studies (Canal-Bedia et al. 2011, Coelho-Medeiros et al. 2019, Lugo-Marín et al. 2019, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017b, Rosenberg et al. 2018). Each study incorporated these statistics into their findings slightly differently. For example, Albores-Gallo et al. (2012) evaluated convergent validity with Spearman correlations, assessed discriminant validity with t-tests and chi-square analyses, and established criterion validity with kappa coefficients. Their findings indicated that the Mexican-M-CHAT showed strong discriminant validity between typically developing and autism populations and moderate internal consistency and convergent validity with the CBCL and ADI-R. Further, criterion validity for the Mexican-M-CHAT demonstrated a higher concordance with the nonverbal dimension on the ADI-R.

Some studies did not report on reliability and validity but instead reported on false positives, completion rates, missingness, and measurement variance (DuBay et al. 2021b, DuBay et al. 2022). Studies that reported on false positives, completion rates, and missingness often demonstrated significant differences in positive screenings between English-speaking and Spanish-speaking samples (DuBay et al. 2021b) as well as differences in translation approaches (DuBay et al. 2022). Measurement and item-specific variance findings also provided details on specific items that may be more or less challenging for Spanish-speaking families to comprehend (DuBay et al. 2021b, DuBay et al. 2022). Together, these psychometric statistics enhance the need to adequately translate and culturally adapt ASD screeners for Spanish-speaking families. See Table 2 for a more detailed breakdown of the psychometric properties for the relevant studies.

Table 2.

Measure/screener characteristics.

Study	Study Design	Name of Screener	Translation Process	Psychometric Statistics	Risk of Bias/QUADAS-2
Albores-Gallo et al. (2012)	Case control	Mexican Modified Checklist for Autism in Toddlers (MM-CHAT)a	Spanish translation of each item Adding minor cultural adjustments	Internal consistency of MM-CHAT total (MM-CHAT-T) using Kuder-Richardson coefficient (KR) = .76 Internal consistency of MM-CHAT for 6 critical items (MM-CHAT-6ci) using KR = .70 Convergent validity using Spearman correlation between MM-CHAT-T, MM-CHAT-6ci, CBCL subscales, and the ADI-R varied and ranged from .18-.86. Discriminant validity: MM-CHAT-T was significantly higher for the ASD group compared to the TD group overall, in the younger age group, and in the older age group. Construct validity: Kappa coefficient (κ) was calculated with the following criteria: MM-CHAT 2/6 or greater. In the ASD group, κ between the MM-CHAT and the ADI-R dimensions was .17.	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Atwood et al. (2015)	Cross sectional study	Communication Complexity Scale (CCS)	Spanish translation Written translation Native Spanish speakers from Peru involved in translation	Inter-rater reliability during measureb development (ICC = .98) 92% scores from independent raters fell within 1 point of each other Kappa values for each collection site = .44, .77, and .78. SPANISH TRANSLATION CCS correlates with CSBS (r = .685, p<.000)	Patient Selection Bias = High Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = High
Bölte et al. (2018)	Cross sectional study	CBCL, Q-CHAT/M-CHAT, SCQ, CDI, SRSc	Translated versions from each respective country’s language	Not reported	No Risk of Bias analysis
Canal-Bedia et al. (2011)	Cross sectional study	Modified Checklist for Autism in Toddlers (M-CHAT)	Spanish translated version for use in Spain Written translation Three interpreters/translators included	Sensitivity = 1 Specificity = 0.98 PPV = 0.19 NPV = 1	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Coelho-Medeiros et al. (2019)	Case control study	Modified Checklist for Autism in Toddlers – Revised (M-CHAT-R) Chilean Version; Modified Checklist for Autism in Toddlers – Revised with Follow-up (M-CHAT-R/F) Chilean Versiond	Phase 1 of this study:e Cultural adaptation from the M-CHAT-R/F in its European Spanish version A content equivalence, semantic, conceptual, and technical comprehension analysis was completed in a sample from the UC Health network users	Cronbach’s alpha = 0.889 Concurrent validity comparing the results of the M-CHAT-R/F and ADOS-2 = r = 0.849, p = 0.0001 Sensitivity = 100% Specificity = 83.3% Discriminant validity estimated by comparing high-risk clinical diagnosis of ASD with the M-CHAT-R/F = 100%	Patient Selection Bias = Medium Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Medium
Cuesta-Gómez et al. (2016)	Cross sectional study	M-CHAT	Spanish translation Written translation	Test-retest Kappa index = 1 Cronbach’s alpha = 0.76 Father-mother Kappa = 0.79	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Unclear Patient Flow Bias = Low
DuBay et al. (2021a)	Qualitative research	First year inventory (FYI) 3.1 (FYv.3.1)	Spanish translation Clinicians spoke the language fluently and did not use interpretation services Written translation 3 clinicians included during forward translation, 2 during back translation	Not reported	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Dubay et al. (2021b)	Retrospective Medical Chart Review	M-CHAT-R Spanish-Western Hemisphere Version (SWH)	Translated specifically for Latin American Spanish-speaking populations 1 forward translator 1 back translator additional translators for same- language comparison and revisions Translators recruited to represent multiple Spanish dialects	No reliability or validity statistics reported Risk-Score Analyses Spanish-speakers (SS) had significantly higher total risk scores English-speakers (ES) had a significantly higher floor rate ES had a significantly lower initial screen-positive rate Six items remained significantly different in proportions of at-risk responses with SS more likely to endorse the at-risk response for 5 of the 6 items Missingness Analyses SS left more items blank SS were more likely to contain missing data for 7 items	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Dubay et al. (2022)	Mixed Methods	FYIv3.1e	Forward-Back One professional translator translated the tool into Spanish Another translator translated the Spanish version back to English An expert in ASD then compared the English versions and items with notable differences were re-translated by the professional translator Translation with Cultural Adaptation Developed using multiple translators from a range of backgrounds Preliminary pre-testing Back translations Additional pre-testing Consensus meetings Cognitive Interviews Participants were shown both translations (beginning with their randomization) Read items aloud, rephrased it, and explained their response Misinterpretations were clarified by interviewers Shown alternative translation – asked to identify differences between the versions and semantic impact of those differences Chose the version they found easiest to understand	Completion Rates Those who responded to the TCA version were more likely to discontinue the survey prior to completion. Missingness Not significant different between the groups Variability More evenly distributed selection of response among the TCA group Measurement Invariance Non-invariance was identified at the first level of invariance testing suggesting that the item factor loadings were not equivalent between the two groups No reliability or validity statistics reported	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Eugenin et al. (2015)	Quantitative and cross sectional study	M-CHAT	Spanish translation Clinicians spoke the language fluently and did not use interpretation services	Not reported	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Fombonne et al. (2012)	Case control study	Social Responsiveness Scale (SRS)	Spanish translation Clinicians spoke the language fluently and did not use interpretation services	Internal consistency: Parent SRS Cronbach’s alpha was 0.97, Teacher SRS Cronbach’s alpha was 0.97 Interrater reliability: Moderate correlation between parent-teacher reports (r = 0.49) Discriminant validity: Parent report AUC =0.96, teacher report AUC = 0.96 No difference in AUC for males/females or younger/older children	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Kimple et al. (2014)	Cross sectional study	M-CHAT	English version Spain Spanish Version (SS) Western Hemisphere Spanish Version (WHS) Written translation	Not reported	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Kuhn et al. (2021)	Randomized Control Trial	M-CHAT-R/F	Research staff systematically administered the M-CHAT-R/F by telephone to every family within 1 week of study referral Interpreters were employed as needed 3 staff who administered the M- CHAT-R/F used a phone interpreter for all non-English languages 1 conducted Spanish administrations using a previously translated versions	Not reported	No Risk of Bias analysis
Lugo-Marín et al. (2019)	Case control study	Spanish Autism-Spectrum Quotient (AQ-Short)f	Translated into Spanish after obtaining permission from the original author Translated completed by a native Spanish speaker Translator was blind to the construct that was intended to be measured Translator was told to translate with the consideration of semantic, linguistic, and cultural equivalence First translation was revised by the first author of the article to verify items corresponded to the items of the original version Bilingual psychologist then translated the Spanish version back into English Panel of 4 bilingual experts compared the original version and the translated version, discussed points of discrepancy and reached agreements for corrections	Cronbach’s alpha = .79 to .88 Test-retest analysis = .90 to .97 Convergent validity = r = .734, significant Accuracy = 71.77% Sensitivity = .98 Specificity = .84	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Magán-Maganto et al. (2020)	Cross sectional study	M-CHAT-R/F (revised/follow-up)	Spain Spanish Version (SS) version Written translation	Cronbach’s alpha = 0.62 for 14 to 22 months Cronbach’s alpha = .72 for 23 to 36 months Sensitivity = 0.79 Specificity = 0.99	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = High
Miller et al. (2011)	Cross sectional study	M-CHAT and the Infant Toddler Checklist (ITC)	English version Spain Spanish Version (SS) Written translation	Not reported	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = High Patient Flow Bias = Low
Morales-Hidalgo et al. (2017b)	Cross sectional study	Spanish version of the Childhood Autism Spectrum Test (CAST)g; Full version = all items; Reduced version = 3 developmental items removedg	Back-translation method First translated from English into Spanish and then back into English Two native speakers served as translators Some items adapted for culture	Full CAST Sensitivity = 83.9% Specificity = 92.5% Positive Predictive Value = 0.63 Cronbach’s alpha = 0.826 Reduced CAST Sensitivity = 85.7% Specificity = 91.16% Positive Predictive Value = 0.61 Cronbach’s alpha = 0.839 Convergent validity was high between the full and reduced versions and the ADI-R, ADOS-2, and CBCL scales.	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = High Patient Flow Bias = Low
Reyes et al. (2021)	Cross sectional study	Social Communication Questionnaire (SCQ)	English version Spanish version of SCQ	Cronbach’s alpha = .70 for English group Cronbach’s alpha = .83 for Spanish group DIF analysis showed non-uniform responses on 5 items	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Unclear Patient Flow Bias = Low
Rosenberg et al. (2018)	Case control study	SCQ	English Version Method of Spanish translation not reportedh	Cronbach’s alpha = 0.89 for CEi group Cronbach’s alpha = 0.77 for BC group Sensitivity = .88 for CE group Specificity = .74 for CE group Sensitivity = .78 for BC group Specificity = .95 for BC group	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Unclear Patient Flow Bias = Low
Scarpa et al. (2013)	Cross sectional study	M-CHAT	English version Western Hemisphere Spanish Version (WHS) Written translation 2 translators (back-translated)	Reliability by education groups: >12th grade education (alpha = .729) 12th grade/GED education (alpha = .320) <12th grade/no GED (alpha = .428) Reliability by income groups: <$20,000/year (alpha = .585) >$20,000/year (alpha = .284)	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Low Patient Flow Bias = Low
Windham et al. (2014)	Cross sectional study	M-CHAT; ASQ	Spanish translated M-CHAT for Mexican-Spanish speakers Spanish version of ASQ Clinicians spoke the language fluently and did not use interpretation services	Rescreened children: 40 % scored positive on the M-CHAT a second time Another 11 % were only ASQ-positive Re-screening children who initially passed both screeners: Additional 95 (25%) scored positive on at least one measure	Patient Selection Bias = Low Index Bias = Low Reference Standard Bias = Unclear Patient Flow Bias = Low
Zuckerman et al. (2013)	Observational, cross-sectional study	ASQ, PEDS, Denver, CDI, MCHAT, ASSQ, SCQj	English version Spanish version	Not reported	Patient Selection Bias = Low Index Bias = Unclear Reference Standard Bias = Unclear Patient Flow Bias = Low

^a Other measures evaluated in this study include: The Autism Diagnostic Interview-Revised (ADI-R) and the Child Behavior Checklist (CBCL).

^b The original paper evaluating the screener obtains reliability and validity statistics (Brady et al. 2012).

^c Q-CHAT: Quantitative Checklist for Autism in Toddlers. Other measures evaluated in this study include: ADOS, ADI-R, CBCL, Social Communication Questionnaire (SCQ), Children’s Depression Inventory (CDI), Development and Well-Being Assessment (DAWBA), Vineland Adaptive Behaviour Scales (VABS), Infant Behavior Questionnaire-Revised (IBQ-R), Early Childhood Behavior Questionnaire-Revised (ECBQ-R), Social Responsiveness Scale (SRS), Infant Toddler Sensory Profile (ITSP), Repetitive Behavior Scale-Revised (RBS-R), Mullen Scales of Early Learning (MSEL).

^d Other measures evaluated in this study include: Autism Diagnostic Observation Schedule (ADOS-2).

^e Phase 1 of this study includes details on the translation process for this screener. Refer to those papers for additional details (Coelho-Medeiros et al. 2019, Robins et al. 2017).

^f Refer to Dubay et al. (2021a) for a detailed description of the translation process.

^g Other measures include a WISC-IV, ADI-R, CBCL, ADOS-2.

^h Families for whom Spanish was the primary language only comprised of 3.8% of the total sample. Further, only the California and Colorado sites were able to recruit families who primarily spoke Spanish. Therefore, the authors do not describe how the materials given to these 3.8% of participants were translated.

ⁱ The majority of SCQ’s given in this sample were in English (<4% were given in Spanish). These are not sensitivity/specificity metrics for the Spanish-translated SCQ.

^j ASQ, Parents’ Evaluation of Developmental Status (PEDS), Denver Developmental Screening Test Autism Spectrum Screening Questionnaire (ASSQ).

Strengths and limitations

Among studies evaluating Spanish-translated screeners (n = 16; Albores-Gallo et al. 2012, Atwood et al. 2015, Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, Coelho-Medeiros et al. 2019, DuBay et al. 2021a, DuBay et al. 2021b, DuBay et al. 2022, Fombonne et al. 2012, Kimple et al. 2014, Kuhn et al. 2021, Lugo-Marín et al. 2019, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017b, Reyes et al. 2021, Scarpa et al. 2013), strengths and limitations were characterized for the following screeners: Spanish version of the Childhood Autism Spectrum Test (CAST), Communication Complexity Scale (CCS), First Year Inventory (FYI) 3.1, M-CHAT (Spain Spanish Version [SS], Western Hemisphere Spanish Version [WH], Argentinian Version, Mexican Version), M-CHAT-Revised (M-CHAT-R) or M-CHAT-R with Follow-Up [M-CHAT-R/F]), Social Responsiveness Scale (SRS), and Social Communication Questionnaire (SCQ). See Table 3 for a more detailed analysis of each study’s main findings categorized by theme.

Table 3.

Main findings, clinical implications, and themes addressed by included studies.

Study	Main findings	Themes
Albores-Gallo et al. (2012)	The MM-CHAT demonstrated overall moderate to strong psychometrics. In concordance with other international studies, some parents did not understand all of the MM-CHAT questions and additional cultural adaptations were needed to assist parents in understanding items on the M-CHAT. Parenting and social behavior styles could be influencing these cultural biases.	Screener evaluation Cultural awareness Translation barriers
Atwood et al. (2015)	The CCS may represent a new tool to assess pre-symbolic communication in Spanish speaking children with intellectual disability for researchers outside of the United States.	Intervention Screener evaluation
Bölte et al. (2018)	The limited access to evidence-based diagnostic and screening tools among middle-income European countries and/or countries with small populations/rarely spoken languages may lower the rates of ASD awareness, identification, and screening/diagnosis for these individuals. Inequalities persist regarding access to translated and well-validated screeners and diagnostic tools for autism across Europe (especially middle and low income countries). Several Spanish translated measures exist; few have been evaluated for reliability and validity or culturally adapted for Spain/Spanish-speakers.	Knowledge/awareness Barriers to care Screener evaluation Translation barriers Cultural awareness
Canal-Bedia et al. (2011)	Specifically when using Spain M-CHAT, the false-positive rate may be lowered if a positive screen was deemed following 5 failed items. The M-CHAT can be integrated into primary care settings during a critical period for ASD diagnosis and evaluation.	Screener evaluation Intervention
Coelho-Medeiros et al. (2019)	The M-CHAT-RF Chilean Spanish version demonstrated adequate internal consistency, strong discriminant validity, and high levels of sensitivity and specificity with the ADOS-2. The work in this study has led to the implementation of training for health professionals in primary care to implement this screener in the Well-Child Care Program for 18-month-old children at risk for ASD.	Screener evaluation Intervention Knowledge/awareness
Cuesta-Gómez et al. (2016)	The Argentinian version of the M-CHAT facilitates earlier ASD detection among children in Buenos Aires. May provide an easily accessible and adaptable ASD screener for pediatricians in Argentina. Some language and comprehension concerns may need to be clarified for mothers not from Argentina.	Screener evaluation Intervention Knowledge/awareness Translation barriers Cultural awareness
DuBay et al. (2021a)	Rigorous translation methods are essential when translating screeners; translation and comprehension errors can present during various stages of translation. The First Year Inventory (FYI) 3.1 may serve as a screener for ASD in children 8 to 16 months with Spanish-speaking caregivers living in the United States, following careful revision.	Screener evaluation Translation barriers Intervention
DuBay et al. (2021b)	The M-CHAT-R SWH translation does not perform the same way as the original English M-CHAT-R. Authors suggest that the differences are a reflection of the potential cultural factors not addressed adequately in the traditional translation procedures.	Screener evaluation Translation barriers Cultural awareness
DuBay et al. (2022)	Qualitative interviews with participants demonstrated that participants chose the translated term in the TCA version 67% of the time and the term used in the FB version 33% of the time. Word order also made a difference and led to a preference of the TCA version 67% of the time. Participants reported a preference of items to be short, simple, and direct. Some noted a need for additional details to help understand the question.	Screener evaluation Translation barriers Cultural awareness
Eugenin et al. (2015)	Caregivers with lower education levels and/or from non-Western cultures may benefit from revised versions of the M-CHAT. Suggestions such as pairing M-CHAT items with pictograms may increase understanding and comprehension of behaviors. Latinx cultures may interpret implications of social communication deficits differently (e.g. social challenges may only apply to strangers rather than mothers). The importance of family among Latinx patients must not be underestimated by providers. The new version of the M-CHAT-R – modifies the two particularly problematic questions on this M-CHAT – and/or screening only at-risk children may increase accessibility and feasibility. Prior to screening, it is necessary to consider whether there are available services to accommodate and treat ASD should it be identified – especially whether services are available in the patient’s primary language.	Screener evaluation Barriers to care Knowledge/awareness Translation barriers Cultural awareness Intervention
Fombonne et al. (2012)	SRS is valid and reliable for screening in Spanish-speaking populations. Parent- and teacher-report is not always necessary to administer given that there is limited additional benefit noted for the teacher report. Consider using parent-only if the circumstances suggest difficulty obtaining a teacher-report.	Screener evaluation Intervention Barriers to care
Kimple et al. (2014)	This screening tool may not be as effective or culturally appropriate across ethnic groups. The M-CHAT-SS is abnormal more often than the M-CHAT in English, despite lower national prevalence rate of autism in Latinx populations. The M-CHAT-SS and M-CHAT-WH demonstrated similar patterns of discrepancies.	Screener evaluation Translation barriers Cultural awareness
Kuhn et al. (2021)	Hispanic families were significantly less likely to receive both an evaluation and a diagnosis. Children screened directly in the parent’s preferred language (no interpreter) were more likely to receive an ASD diagnosis. These findings support previous research indicating problems with M-CHAT-R/F accuracy among minority populations. Potential issues with the telephonic language interpretation process for parent checklist screening approaches – greater reliance on word-for-word translation which leads to increased confusion and misunderstanding, resulting in poorer positive predictive power of the screening process. Authors emphasize the importance of hiring and training native speakers to directly administer screening assessments in families’ preferred languages.	Translation barriers Cultural awareness Barriers to care
Lugo-Marín et al. (2019)	The AQ-Short is a reliable instrument for the screening of ASD in Spanish native-speaker adults. It shows very good internal structure and a good convergence with diagnostic ASD gold-standard measures. It clearly differentiates ASD from other specific populations (SSD, ASD-relatives), and is thus a useful instrument in the screening stage. Clinical groups (ASD, SSD) are less coherent in their responses, suggesting the need to identify the possible difficulties of these groups when understanding the items.	Screener evaluation Cultural awareness
Magán-Maganto et al. (2020)	The M-CHAT-R/F is an effective tool for screening young children in the general Spanish-speaking population. In order to improve early detection, it is necessary to train pediatricians to identify early signs of ASD and provide them with a standardized tool to screen further.	Screener evaluation Intervention Knowledge/awareness
Miller et al. (2011)	Partnerships between pediatric providers and ASD providers are important for making progress in developing and adapting formal screening procedures and facilitating earlier and more accurate identification. Inclusion of both English and Spanish screeners is feasible in a pediatric primary care setting as long as there is relevant expertise involvement. This integrated approach may successfully identify more children at risk than traditional clinic observation or reported parent concern. Screening children for ASD at sick visits rather than only well-child visits may further improve early identification.	Intervention Knowledge/awareness Cultural awareness Barriers to care
Morales-Hidalgo et al. (2017a)	The CAST was demonstrated to be a reliable and valid test for screening for ASD among Spanish school-aged children. Boys showed higher scores than girls on the CAST. The effect size was medium. Questions on general development did not appear to make the screener more readily understandable or improve the psychometric properties. The authors propose a reduced Spanish version of the CAST.	Screener evaluation Cultural awareness
Reyes et al. (2021)	No differences in SCQ scores were found between the English and Spanish versions in a nonclinical sample. Spanish-speaking mothers with less educational attainment reported more ASD symptoms compared to Spanish-speaking mothers with more educational attainment. Spanish-speaking mothers exhibited different response patterns on 5 SCQ items. Among the English-speaking sample only, teacher BASC-2 scores predicted higher SCQ scores. Among the Spanish-speaking sample only, decreased maternal education and teacher BASC-2 scores predicted higher SCQ scores.	Screener evaluation Translation barriers
Rosenberg et al. (2018)	Spanish maternal language did not have a significant impact on SCQ score. Male child sex, lower household income, Black and Other maternal race, and lower maternal education were associated with higher SCQ scores among both children with pre-identified developmental concerns (CE group) and the general population (BC group). Using a cut-point of 11, the SCQ may be more sensitive in detecting ASD among young children with pre-identified developmental concerns, but more specific in rejecting ASD for children in the general population.	Cultural awareness Knowledge/awareness
Scarpa et al. (2013)	Modifications of the Spanish M-CHAT may be necessary to improve internal consistency and accuracy across ethnic and educational groups in low SES rural areas. Alternative administrations may be considered for low SES parents (e.g. read aloud, items explicitly reviewed for comprehension, paired with pictures). The Spanish M-CHAT is not recommended until further research better evaluates its properties in the United States.	Screener evaluation Translation barriers Cultural awareness Intervention
Windham et al. (2014)	Additional resources, beyond screeners, will be required to successfully assess children at-risk for ASD from underserved backgrounds. Similar to Kimple et al. (2014), a higher screen positive rate was detected in predominantly Latinx groups compared to non-Latinx groups despite lower prevalence rates of ASD.	Intervention Cultural awareness
Zuckerman et al. (2013)	Provider-related factors – such as bias, limited Spanish language resources, or limited exposure – may contribute to some diagnostic disparities among Latinx and Spanish-speaking children. Primary care providers identified the following barriers to ASD care for Latinx children: limited access to specialists, language and cultural differences between providers and patients, and limited access to care. Two suggestions are made to address these barriers: increase training for working with this population and increase connections between developmental/ASD specialists and primary care pediatricians.	Barriers to care Translation barriers Cultural awareness Intervention

Spanish version of the Childhood Autism Spectrum test (CAST)

The CAST demonstrated to be a reliable and valid test for ASD screening in Spanish school-aged children. In particular, the screener showed high values of sensitivity and specificity that align closely with previous research (Scott et al. 2002). Notably, both a full and reduced version of the CAST was evaluated by Morales-Hidalgo et al. (2017b). The authors report that the reduced version possesses many strengths, including better psychometric properties and higher correlations with the ADI-R, ADOS-2 and CBCL specific scales. All limitations of this screener pertain to the full version, which identified a few items that showed low discriminative poor between children with ASD and without ASD. Similarly, the general development questions included in the screener are considered a limitation due to evidence demonstrating that these items did not enhance comprehension or improve the screener’s psychometric properties. Therefore, the authors recommend the reduced version as a solution to the limitations evident in the full version. Ultimately, findings support application of the adapted full Spanish version and the reduced version of the CAST for screening for milder or subtler manifestations of ASD in a Spanish population.

Communication Complexity Scale (CCS)

Strong, positive correlations with the Communication and Symbolic Behavior Scales Developmental Profile Infant-Toddler Checklist (CSBS-ITC) and limited reliance on parent report were considered strengths of the Spanish-translated CCS (Atwood et al. 2015). Some limitations include that item comprehension appeared to depend on the syntax or pronunciation used by the administrator, and that translation may reflect cultural/linguistic differences across cities and regions rather than generalize to Peru (the origin country). Further, the validation sample was not balanced between gender, and the measure has only been validated in Peruvian children, thereby limiting generalizability. Overall, findings support the CCS as a valid screening measure of communication behaviors among native Spanish-speaking children in Peru.

First years Inventory (FYI) v. 3.1

A few studies evaluated the FYI v. 3.1. Across all studies, the Spanish-translated FYI’s strengths included a rigorous translation and cultural adaptation process. Limitations for one study included a lack of representation for some Latin American Spanish dialects or cultures among both translators and participants as well as a disparity in administration of the screener, such that it was delivered as an interview rather than independent parent-report (DuBay et al. 2021a). Alternatively, in a follow-up paper (DuBay et al. 2022), representation was achieved across a wide range of countries of origin despite the study taking place in the United States. Further, DuBay et al. (2022) successfully garnered strengths and limitations reported by caregivers completing the surveys and discerned important nuances in item recognition, length of item descriptions, and language preferences (e.g. ‘hijo/hija’ instead of ‘hijo’ only to represent a child). The translated measure is praised for its rigorous translation and adaptation approach, and although not a limitation of the screener, it is acknowledged that this level of rigor is challenging, time consuming, and resource intensive to develop. These barriers increase risk of error, which could unintentionally impact the reliability of the screeners.

Modified Checklist for Autism in Toddlers (M-CHAT)

Many studies included the M-CHAT. We first discuss the studies that examined its dialectic variations, then we separately review the study that evaluated the Argentinian M-CHAT as unique strengths and limitations emerged from that research, and finally we discuss a revised version of the M-CHAT.

When considering strengths and limitations of the M-CHAT, it is necessary to review and compare the studies that examined its dialectic variations. Five studies (Albores-Gallo et al. 2012, Canal-Bedia et al. 2011, Eugenin et al. 2015, Kimple et al. 2014, Magán-Maganto et al. 2020) evaluated the M-CHAT in this manner. The following strengths were described: high values of sensitivity and specificity that were similar to the original M-CHAT (Coelho-Medeiros et al. 2019, Magán-Maganto et al. 2020), flexibility with phone-administration that may increase access to screening (Canal-Bedia et al. 2011), minimal evidence of differences between the M-CHAT SS and the M-CHAT WH when used with Mexican/Central American caregivers in the United States (Kimple et al. 2014), simplicity and ease of answering across SES levels (Magán-Maganto et al. 2020), incorporation of implicit references to developmental milestones for teaching purposes (Eugenin et al. 2015), and inclusion of questions that resemble and require reflection of daily activities familiar to Spanish-speaking parents (Eugenin et al. 2015). Limitations included difficulties comprehending questions (Albores-Gallo et al. 2012, Eugenin et al. 2015) – especially regarding repetitive or unusual behaviors (Eugenin et al. 2015), increased time and financial resources for follow-up (Canal-Bedia et al. 2011, Magán-Maganto et al. 2020), lower positive predictive value than the original M-CHAT (Canal-Bedia et al. 2011), potential barriers to in-person screening (Canal-Bedia et al. 2011), and difficulty with blinding and preventing caregiver bias when evaluating the efficacy of the screeners (Coelho-Medeiros et al. 2019).

Second, the study reporting on the development of the Argentinian M-CHAT identified the following strengths: robust translation and methodology to support content, semantic, and technical equivalence, high test-retest reliability and item-response agreement when comparing mothers and fathers, efficiency as a screening tool for pediatricians, and inclusive, elementary-level reading for parents with lower literacy (Cuesta-Gómez et al. 2016). These strengths emphasize the utility of this translated screener when considering its ease of use for providers and reporters, and its similarities to the original screener. However, this study was conducted in Buenos Aires, which may limit generalizability to a specific subset of the Argentinian population. An additional limitation includes that a pediatrician must discern the screener’s administration method based on the parent’s literacy level (i.e. self-administered or administered with assistance). Further, the locational specificity of the screener comes at the cost of reduced clarity and relatability for users not originally from or currently living in Argentina. These factors limit widespread utility given the practical challenges and more limited scope of the translation.

Last, the M-CHAT-R and M-CHAT-R/F are revised and updated Spanish versions of the previous M-CHAT (Coelho-Medeiros et al. 2019, DuBay et al. 2021b, Magán-Maganto et al. 2020). Various versions of the M-CHAT-R and M-CHAT-R/F exist with multiple different Spanish translations. Strengths of the M-CHAT-R/F included accurate detection of ASD, reduction of follow up appointments, and adequate sensitivity and specificity scores. Limitations highlight that psychometric properties may be especially dependent on the age of the child and their education level. More specifically, more accurate results were noted for younger children and for parents with higher education levels. These limitations are important when considering AAP recommends formal ASD screening at ages 18 and 24 months (Hyman et al. 2020). Additionally, internal consistency may be on the lower end of acceptability. Further, concerns remain regarding how the translated versions compare to the English versions: DuBay et al. (2021b) demonstrated that the M-CHAT-R Spanish Western Hemisphere translation does not perform the same way as the original English M-CHAT-R and instead is more sensitive to the ASD diagnosis and positive screening of ASD symptoms in Spanish-speaking populations in the United States.

Social Responsiveness Scale (SRS). The Spanish-translated SRS demonstrated strengths in both its teacher and parent forms (Fombonne et al. 2012). For both forms, excellent discriminant validity was documented when discerning children diagnosed with Pervasive Developmental Disorder (PDD) from typically developing children that did not receive a PDD or ASD-related diagnosis. This is a key finding for increased detectability of Spanish-speaking children showing signs of ASD, which can reduce diagnostic gaps and improve access to early intervention. Interrater reliability statistics were also strong, evidenced by the moderately high correlations observed between parent and teacher ratings. Noted limitations include lengthy administration time, limited age range for validation (ages 4–13), lack of data determining concurrent validity with other screening measures, and lack of discriminant validity data differentiating ASD from other clinical presentations among Spanish-speaking populations. These limitations indicate that use of this screener may impede early identification of ASD (i.e. before age 4) among Spanish-speaking children. Additionally, the Spanish-translated SRS was studied in 2012 before the changes for an autism diagnosis were released in the DSM. Therefore, the Spanish SRS may be suited as a general screener for potential autism-specific behaviors until restudied utilizing the current DSM-5 criteria.

Social Communication Questionnaire (SCQ). Strengths of the Spanish-translated SCQ include its ease and low cost of implementation, as well as its ability for use across a wide age range. This may be especially relevant when screening for ASD symptoms among school-aged children who may have immigrated to the United States past the age when developmental surveillance measures intended for younger children may have flagged them. A limitation of this measure is its lack of comparison norms for Spanish-speaking populations. Reyes et al. (2021) noted other limitations of the Spanish-translated SCQ when used in a nonclinical sample. Translated items may not reflect the intended meaning of the original measure, resulting in the potential misinterpretation of items reduced reliability and validity. Moreover, SCQ results may reflect co-occurring behavior and emotional problems masked as ASD symptoms rather than ASD symptoms themselves, further reducing its potential validity.

Barriers and facilitators to implementation and interpretation

Additional studies focused on assessing successes and barriers to implementation of screeners (n = 6; Bölte et al. 2018, Kimple et al. 2014, Reyes et al. 2021, Scarpa et al. 2013, Windham et al. 2014, Zuckerman et al. 2013). The findings from these studies highlight important differences in screener performance between low- and higher-income or low- and high-education Spanish-speaking families (Bölte et al. 2018, Reyes et al. 2021, Scarpa et al. 2013, Windham et al. 2014, Zuckerman et al. 2013). Screeners were less effective when presented to Spanish-speaking families with lower income levels and lower education histories. Reyes et al. (2021) found that Spanish-speaking mothers with less education attainment reported more ASD symptoms compared to Spanish-speaking mothers with more education attainment. Additionally, income level affects screener accessibility (Bölte et al. 2018, Windham et al. 2014, Zuckerman et al. 2013). For example, middle-income European countries are at the highest disadvantage for availability of translated measures compared to high- or low-income European countries (Bölte et al. 2018). Further, Spanish-speaking families in low-income communities are underserviced due to the high costs of translated screening tools in predominantly English-speaking communities or countries like the U.S. (Zuckerman et al. 2013).

When examining differences in education level, Spanish-speaking respondents with fewer years of completed education reported more ASD symptoms among their children compared to Spanish-speaking respondents with higher educational attainment, despite there being no differences in overall symptom reporting between English- and Spanish-speaking respondents (Reyes et al. 2021). Additionally, respondents with lower education attainment demonstrated more difficulty with screeners due to literacy and comprehension challenges (Scarpa et al. 2013). Further, slight differences were noted between Spanish-translated screeners of different Latinx dialects (Kimple et al. 2014), demonstrating the heterogeneity of Latinx culture and Spanish language and the complexity of translated screening tools. In particular, Kimple et al. (2014) measured response differences when comparing the M-CHAT SS and M-CHAT WH, which indicated distinctions in cultural interpretation rather than translation differences among the Spanish-speaking population. Although there were limited differences in abnormal screening rates for the M-CHAT SS and M-CHAT WH, both measures produced higher abnormal screening rates for Spanish-speakers compared to English-speakers using the original M-CHAT. This finding further highlights the importance of proper Spanish translation and interpretation in ASD screening. See Table 3 for a more detailed breakdown of each included study’s main findings.

Clinical implication themes

All studies offered clinical implications relevant to the study question. In this section, the current authors summarize the clinical implications provided by the authors of the cited articles. In order to provide broader conceptualizations of the implications of utilizing Spanish screeners for ASD, the authors of the current study read and grouped implications listed in each article into themes using a brief thematic analysis protocol (Braun and Clarke 2006): barriers to care (n = 6), cultural awareness (n = 15), translation barriers (n = 12), intervention (n = 12), knowledge/awareness (n = 7), and screener evaluation (n = 16). Regarding barriers to care, authors discussed issues related to limited access to primary or specialty care providers for Spanish-speaking individuals, especially families from low-to-middle income countries or rural communities (Bölte et al. 2018, Scarpa et al. 2013, Zuckerman et al. 2013). Additional barriers to care included price and opportunity cost related to time for follow-up visits and interviews for some screeners (Eugenin et al. 2015, Fombonne et al. 2012, Miller et al. 2011), and availability of post-screening services for Spanish-speaking families (Eugenin et al. 2015, Zuckerman et al. 2013).

Themes of cultural awareness and translation barriers centered around challenges associated with comprehending differences between cultural groups within the Latinx community (Albores-Gallo et al. 2012, Cuesta-Gómez et al. 2016, Eugenin et al. 2015, Reyes et al. 2021, Scarpa et al. 2013, Zuckerman et al. 2013). For example, generalizability concerns were noted for the Argentinian M-CHAT, as non-Argentinian mothers might need assistance understanding certain items (Cuesta-Gómez et al. 2016). Additionally, for the Mexican M-CHAT, it was noted that the translation process involved making small cultural adjustments to the screener, such as describing the ‘peek-a-boo’ game, since Mexican mothers do not know the explicit name for it. Further, issues with patient/provider interactions, such as miscommunication between patients and providers due to language barriers, highlighted the importance of multicultural training and teaching for providers working with Spanish-speaking populations (Zuckerman et al. 2013). The intervention theme consisted of recommendations for screener implementation outside of the U.S. (Atwood et al. 2015), screener integration into primary care clinics (Miller et al. 2011, Zuckerman et al. 2013), and adaptation of validated screeners for appropriate age and cultural groups (DuBay et al. 2021a, Scarpa et al. 2013, Windham et al. 2014). The importance of incorporating Latinx families in the screening and diagnostic process was also considered (Eugenin et al. 2015).

Within the knowledge/awareness theme, a general lack of ASD knowledge and awareness was noted in Spanish-speaking or Latinx populations, thus emphasizing the importance of developing these screeners as tools for improved parental education and awareness for early detection. Last, the screener evaluation theme captured interpretations of the screener included in the study (e.g. Kimple et al. 2014) and suggestions for future more rigorous evaluation of future screeners (e.g. Scarpa et al. 2013). See Table 3 for a more detailed breakdown of each study’s main findings categorized by theme.

QUADAS-2 bias and quality assessment

Mixed results emerged regarding risk of bias and quality assessment. Most studies included were cross sectional study designs (n = 12; Atwood et al. 2015, Bölte et al. 2018, Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, Fombonne et al. 2012, Kimple et al. 2014, Magán-Maganto et al. 2020, Miller et al. 2011, Morales-Hidalgo et al. 2017a, Reyes et al. 2021, Scarpa et al. 2013, Windham et al. 2014, Zuckerman et al. 2013), some were case control designs (n = 5; Albores-Gallo et al. 2012, Coelho-Medeiros et al. 2019, Fombonne et al. 2012, Lugo-Marín et al. 2019, Rosenberg et al. 2018), one was a qualitative design (DuBay et al. 2021b), and one was a mixed methodology design (DuBay et al. 2022). Risk of bias was considered relatively low, but a few studies had higher risk of bias due to lack of clarity regarding selection bias, detection bias, and reporting bias (Table 2). No study was excluded or considered high risk due to the risk of bias analysis. An adapted version of the QUADAS-2 evaluated bias risk for the 16 screener evaluation studies (Albores-Gallo et al. 2012, Atwood et al. 2015, Canal-Bedia et al. 2011, Cuesta-Gómez et al. 2016, Coelho-Medeiros et al. 2019, DuBay et al. 2021a, DuBay et al. 2021b, DuBay et al. 2022, Fombonne et al. 2012, Kimple et al. 2014, Kuhn et al. 2021, Lugo-Marín et al. 2019, Magán-Maganto et al. 2020, Morales-Hidalgo et al. 2017b, Reyes et al. 2021, Scarpa et al. 2013). Quality assessment results paralleled risk bias outcomes and demonstrated similar, moderate concern for bias. Concerns typically noted lack of information on participant flow, diagnostic knowledge and screening process, index text psychometrics, and identification of a comparison test (Table 2).

Discussion

This review of 22 studies serves to synthesize known information on the influence of Spanish language on ASD screening. Primarily Spanish-speaking individuals experience unique barriers in detection and screening for ASD. The studies included in the present review provide evidence to support the use of ASD screeners in Spanish but demonstrate some limitations, including challenges of generalizability due to variability in Spanish dialects across populations and insufficient psychometrics for some screeners. The included studies also provide insight on the implementation of Spanish screeners in healthcare settings and emphasize the relevance of client characteristics in validation of screeners (e.g. income level and education history), importance of multicultural training for providers working with Spanish-speaking clients, and the general lack of ASD knowledge and awareness noted in Spanish-speaking or Latinx populations.

However, there were limitations that must be addressed. Most evaluative studies did not report standardized translation processes, reliability and validity statistics, or factors related to accessibility and receipt of services outside of linguistic barriers. This missing information complicates replication, adoption, and dissemination of the screeners. Future research should gather the appropriate psychometric data, replicate the findings, and examine the feasibility of dissemination in settings that foster access to care. A second limitation refers to the heterogeneity of Latinx culture, which suggests that validation techniques utilized may not apply to all Latinx populations and limits generalizability. Future research must take this heterogeneity into account. With additional studies examining more representative samples, the procedures may gain the necessary empirical support for widespread use. Additionally, results regarding risk of bias and quality of assessment were mixed. Future research would benefit from the application of more rigorous research designs, like Randomized Controlled Trials and diagnostic accuracy studies, to address concerns related to bias, psychometric interpretations, and generalizability. Moreover, rigorous research designs will not have the intended impact if best-practice methods are not used to ensure cultural and linguistic relevance. The translation, adaptation, and validation (TAV) method is recommended as it differs from common translation and validation techniques (i.e. forward-back translation) and includes adaptation (DuBay et al. 2021). Without adaptation, direct translation of materials normed in English-speaking cultures will not be relevant to the heterogeneous Spanish-speaking population. A final limitation refers more to the methodological approach of the review. Although it is common to exclude grey literature (Paez 2017) in a systematic review, given that much of the work is new and less studied, some research may be captured in dissertations or conference presentations rather than published, peer-reviewed papers. Future research on this topic should consider including grey literature, especially if completing a more comprehensive analysis (e.g. a meta-analysis).

Clinical implications

The Latinx population is a diverse population with many different sub-cultural traditions, levels of multilingualism, and, in the U.S., differing layers of acculturation. Several steps can be utilized to begin moving towards linguistically and culturally-competent care for Latinx, Spanish-speaking populations. First, researchers should identify specific ASD screening and assessment measures and linguistic and cultural adaptations that have been shown to be useful and effective for this population. This review provides a first step in this process, yet much of the necessary details to build on this research was notably missing. An additional, recently published review by Alonso Esteban et al. (2020) further supports the need for a concentrated and collaborative effort to develop reliable and valid instruments for the early identification of ASD in Spanish-speaking populations at an international level. Part of this process includes simply identifying and reporting culture and language backgrounds of participants included in their research. This is necessary for informing the creation of specific Spanish-language screening and assessment measures as well as evaluation and treatment procedures included later on in the ASD diagnostic process. Second, studies evaluating the efficacy and validity of screeners must report psychometric data relevant to the validity, generalizability, and reliability of the measure. In order to provide appropriate ASD screenings to individuals, it is necessary to report suitable statistics to support those claims. Third, when examining cultural and linguistic differences in Latinx and non-Latinx populations, research recruitment must focus on building a representative sample of Latinx individuals to increase the field’s dialectical and cultural knowledge.

Conclusion

Overall, this systematic literature review demonstrates the important role Spanish language plays in the ASD screening process. Namely, language differences – especially considering the indelible relationship between language and culture – may: affect knowledge and awareness of development and developmental milestones due to a lack of available resources and supports that are linguistically- and culturally-relevant for the wide range of populations within the Latinx culture; level and ease of communication between families and providers; and result in translated resources validated for use in specific sub-groups of the heterogeneous population of Spanish-speakers potentially impacting generalizability. Additional measures may be needed to capture all concerns and symptoms. The findings from this systematic literature review provide support to improve the current ASD screening process for Spanish-speaking families. The current review provides an ideal foundation for the growing body of Latinx ASD literature and offers a source of clinically applicable information for both ASD providers and Spanish-speaking families.

Appendix A.

Data Extraction Sheet

• Title of Article

• Authors (last name, first/middle initials)

• Year of publication

• Journal

• Volume/Issue #

• Page numbers

• Lead author contact details

• Country in which the study conducted

• Language(s) in which the paper was published

• Aim of study

• Study design

• Start date

• End date

• Study funding sources

• Possible conflicts of interest for study authors

• Name of screener(s) evaluated

• Reliability Statistics

• Validity Statistics

• Administration of screener (e.g. English version, Spanish version, translated version?)

• Translation methods

• Population description

• Participant Type(s)

• Inclusion criteria

• Exclusion criteria

• Country of Study Location (if possible, include geographical region [i.e. US, Midwest])

• Method of recruitment of participants

• Total number of participants

• Were parents included?

  • Total number of parents
  • Age range of parents: Youngest age
  • Age range of parents: Oldest age
  • Mean age of parents
  • SD age of parents
  • Sex: Number of females (parents)
  • Sex: Number of males (parents)
  • Race – parents (n, % [if possible]; please indicate if not reported)
  • Ethnicity – parents (n, % [if possible]; please indicate if not reported)
• Were children included?

  • Total number of children
  • Age range of children: Youngest age
  • Age range of children: Oldest age
  • Mean age of children
  • SD age of children
  • Sex: Number of females (children)
  • Sex: Number of males (children)
  • Diagnoses of children included in the study
  • Race – children (n, % [if possible]; please indicate if not reported)
  • Ethnicity – children (n, % [if possible]; please indicate if not reported)

• Total number of interpreters/translators

• Education level of sample

• Advantages of Screeners

• Limitations of Screeners

• Clinical implications

Funding Statement

This work was funded by National Institute of Child Health and Human Development; Center for Clinical and Translational Science, University of Alabama at Birmingham; Department of Health and Human Services Administration for Children and Families; National Institute of Child Health and Human Development; National Institutes of Health training grant through the Heart, Lung and Blood Disease Institute; Dwight David Eisenhower Transportation Fellowship Program.

Disclosure statement

No potential conflict of interest was reported by the authors.