Thermal Detection Thresholds Are Not Elevated in Children and Adults with Autism Spectrum Disorder

Background: In recent years, sensory features, including an “apparent indifference to pain/temperature” (American Psychiatric Association, 2013) have been included in the diagnostic criteria for autism spectrum disorder (ASD). However, most evidence of atypical pain/temperature perception in ASD is based on clinical descriptions or parent reports rather than experimental studies (Moore, 2015). The few studies to date examining warm and cold detection thresholds in this population have produced mixed and contradictory results (Cascio et al., 2008; Duerden et al., 2015; Fründt et al., 2017; Yasuda et al., 2016). Nevertheless, all prior studies utilized small sample sizes (≤20 ASD participants), and none explicitly controlled for covariates such as age and IQ when comparing groups.

Objectives: To assess thermal detection thresholds in a large sample of children and adults with ASD, comparing them to typically-developing (TD) controls and adjusting for a number of covariates.

Methods: Participants included 143 individuals with average or above-average intelligence (WASI-II FSIQ > 70): 32 ASD adults (22 males, mean age 28.76 years), 24 TD adults (13 males, mean age 29.0 years), 51 ASD children (41 males, mean age 11.24 years), and 36 TD children (26 males, mean age 10.03 years). In the experimental task, a thermode on the palm increased or decreased in temperature at 1°C per second until the participant indicated a sensation of warmth or cold via mouse click. This procedure was repeated over four alternating blocks of five warm or five cold trials (counterbalanced across participants). Warm and cold detection thresholds were estimated using the Harrell-Davis median (Harrell & Davis, 1982) of threshold temperatures from warm and cold trials, respectively. Participants or caregivers also completed measures of IQ (WASI-II), autistic traits (SRS-2), and sensory reactivity (Sensory Profile). Threshold values were analyzed using proportional odds regression models (Liu et al., 2017), with baseline models including ASD status, age, sex, and counterbalance order as predictors. Additional predictors were added to the model using a best-subset approach based on BIC values.

Results: ASD and TD groups in the full sample and child/adult subsamples were matched on age, sex, and performance IQ (ps > 0.118), although verbal IQ was higher in the TD group (p < 0.001). Baseline regression models that included ASD status, age, sex, and counterbalance order as predictors failed to find any significant associations with warm or cool detection thresholds in the combined sample or any subgroup (ps > 0.05). In the best-subset regression, performance IQ was consistently included in all best-fitting models, with higher performance IQ predicting lower warm thresholds and higher cool thresholds (ps < 0.025). Questionnaire measures of autistic traits and sensory reactivity were not found to significantly predict warm or cool thresholds in any model.

Conclusions: Warm and cool detection thresholds did not differ between individuals with ASD and TD controls. Performance IQ was the only consistent predictor of sensory threshold across temperature conditions and age groups. These findings suggest that the “hyposensitivity” to thermal stimuli observed in some individuals with ASD is likely not related to a quantitatively higher sensory threshold.