All children’s baseline data will be collected, including demographic characteristics, peripheral hearing, intelligence level, autism symptoms, and auditory processing and language skills. The auditory processing and language skills will be reevaluated after 1 year and 2 years.
(1) Demographic characteristics include children’s gender, age, mother’s pregnancy and childbirth history, birth status, handedness, chronic otitis media, ear disease or trauma, parents’ educational level, monthly family income, parents’ occupation, etc.
(2) Peripheral hearing assessment includes pure tone audiometry. The instrument is Conera diagnostic audiometer from Denmark. Using plug-in headphones, the hearing thresholds of binaural 500, 1000, 2000, and 4000 Hz are tested respectively.
(3) Gesell Developmental Schedules system includes adaptability, gross motor, fine motor, language, personal social interaction, and other five aspects, which are assessed using the development quotient to measure their cognitive level.
(4) Autism Diagnostic Observation Scale-2 (ADOS-2): The ADOS-2ASD diagnostic scale assesses communication and social interaction in cases with ASD. The evaluation includes measuring the ability to play and use objects imaginatively, along with observing personal stereotyped and repetitive behaviors. An evaluator selects the module based on the age and language development of the individual, conducting approximately 40 minutes of game assessment or dialogue.
(5) Childhood Autism Diagnostic Scale (CARS): The CARS assessment comprises 15 items, encompassing aspects, such as interpersonal relationships, imitation, emotional response, physical application ability, relationship with non-living objects, adaptation to environmental changes, visual response, auditory response, proximity sensory response, anxiety response, verbal communication, nonverbal communication, and intelligence. Evaluators score based on the oddity, frequency, severity, and duration of observed behaviors, employing methods, such as observation, inquiries, and data collection from the individual’s medical history.
(6) Auditory processing assessment: The ASD children’s subjective and objective auditory processing skills are assessed using the preschoolers’ auditory processing assessment scale and speech-ABR.
1) Preschool Auditory Processing Assessment Scale (PAPAS)[28]: It was developed by Professor Hong Qin and used to assess the auditory processing function of children aging 3–6 years. It includes the following five dimensions: auditory decoding, auditory attention, communication, hyperactivity impulse, and visual attention. Cronbach's alpha coefficient for the scale was determined to be 0.941, indicating strong reliability and validity. A regional norm specific to Jiangsu Province was established during this study. Total scores on the scale were recorded, with higher scores reflecting an elevated risk of abnormal auditory processing.
2) Speech-ABR: It reflects the processing of speech temporal characteristics and records brainstem activation in response to stimuli. It is an important tool for evaluating the development of brainstem auditory speech coding ability and pathological research, possessing high reliability, stability, and reproducibility. In this study, the BioMARK, a commercialized Speech-ABR tool, is utilized to assess the objective auditory processing skills of the children under examination. The testing apparatus is the AEP 7.0 version of the auditory evoked potential system from Bio-logic (United States). The recorded outcome index for this test is the BioMARK total score, which is derived from five parameters: V-wave latency, A-wave latency, V/A slope, first formant frequency, and higher formant frequency. The higher the total score, the more serious the impairment of auditory processing skills.
(7) Language assessment: Language Development Assessment Scale for Children Aged 1–6 is used to assess ASD children’s language skills[29]. This diagnostic language development assessment scale is designed specifically for children aging 1–6 years, evaluating language understanding, language expression, and story comprehension using physical objects, picture albums, and audio recordings. The scale demonstrates robust reliability and validity and has established a regional norm specific to Jiangsu Province.
Quality control
(1) Quality control of appraisers
Before the evaluation, evaluators will receive unified and standardized training, including the purpose, significance, evaluation procedures, operation specifications, and precautions of the study.
(2) Quality control of hearing and electrophysiological tests
The test environment is an electromagnetic shielding sound insulation room, and the background noise is less than 30 dBA. Two experienced audiologists mark the position of the speech-ABR waveform.
(3) Quality control of scale collection
In this study, 2–3 trained researchers will collect the general information questionnaire and the PAPAS, and use unified guidelines to retrieve the questionnaire on the spot. If any missing or wrong filling is found, feedback on the questionnaire is promptly provided to the preparer to ensure the optimal response rate and completeness during the survey.
(4) Quality control of follow-up
After the baseline assessment, WeChat account of children’s parents is added, and the follow-up records are collected. The follow-up time, contact information, and precautions are recorded in detail to ensure the traceability of the follow-up. For those who cannot timely attend in the hospital to participate in the follow-up, they are contacted and asked the reasons, and the follow-up rate is highly guaranteed under the principle of subjects’ voluntary participation. During the follow-up, researchers patiently answer the subjects’ questions and improve their compliance.
(5) Quality control of data processing
Two researchers independently import data to ensure the accuracy of the data.
(6) Quality control of data analysis
In the process of statistical analysis of data, the analysis is carried out by researchers who are skilled in statistical analysis to ensure the robustness of analysis and correctness of results.
Statistical analysis
In line with Hypothesis 1, which posits a correlation between auditory processing and language in ASD children at baseline (T1), 1 year later (T2), and 2 years later (T3), SPSS 25.0 software (IBM, Armonk, NY, USA) is utilized. Pearson correlation analysis is employed to examine the association among subjective auditory processing assessment results (total score of the PAPAS), objective auditory processing assessment results (total score of BioMARK), and the three facets of language assessment (language understanding, language expression, and story understanding) across different time points.
For Hypothesis 2, suggesting diverse developmental trajectories in ASD children’s auditory processing, the latent class growth model (LCGM) is fitted using Mplus 8.3 software. Principal component analysis reduced the dimensions of the simultaneously collected total scores of the PAPAS and BioMARK, extracting a common factor termed "auditory processing." The auditory processing data at three time points may serve as the model-fitting input, with indicators, such as Akaike information criterion (AIC), Akaike Bayesian information criterion (BIC), sample size adjusted BIC (aBIC), entropy, Vuong Lo Mendell Rubin likelihood ratio test (VLMR), and Bootstrapped likelihood ratio test (BLRT). A model is considered superior if it exhibits higher entropy and lower AIC, BIC, and aBIC values, along with statistically significant P-values for VLMR and BLRT (P < 0.05). The final model selection considers both practical significance and statistical criteria.
According to Hypothesis 3 (ASD children’s auditory processing development trajectory has a predictive effect on their language damage), the one-way analysis of variance (ANOVA) is used to analyze the differences in language assessment results (language understanding score, language expression score, and story understanding score) among different auditory processing developmental trajectories. Furthermore, a hierarchical multivariate linear regression model is established, in which language comprehension score, language expression score, and story comprehension score are taken as dependent variables into account, and baseline data (e.g., gender, age, listening, intelligence, and other variables) are utilized to establish Model 1. On the basis of Model 1, different auditory processing developmental trajectories are incorporated, and Model 2 is therefore developed to explore the impact of different auditory processing developmental trajectories on ASD children’s language development.