‘I feel trapped in my safe clothes’: The impact of tactile hyper-sensitivity on autistic adults

Participants

Eighty-six participants were recruited through opportunity sampling using social media; a university participant pool including students, staff and general public; and organisations supporting autistic people which shared the study via their social media, newsletter or website. Participants were eligible to take part if they were 18 years old or older, lived in the United Kingdom and formally identified as Autistic. Participants who did not meet the criteria and those who reported having an eating disorder (n = 1) were excluded from the analysis. Negative self-evaluation and body image disturbance are key features of eating disorders (American Psychiatric Association, 2013) and inclusion of participants with these conditions could skew the appearance-related results. There was no previous research available to guide the sample size calculation. Sample size was based on an a priori estimate to detect a medium effect size (f² = 0.15; Cohen, 1992) with α = 0.05 and 0.8 power for a regression with a single predictor, which indicated 55 participants were required (G*Power 3.1.9.7; Faul et al., 2007). The final sample of 86 participants was sufficient to detect a medium effect for the simple regression and moderation analyses. Due to time restrictions on the recruitment window, it was not possible to obtain a larger sample sufficient for detecting a small effect size (f² = 0.02, N = 395; Cohen, 1992).

During recruitment, there were two instances of a sudden, large influx of responses submitted simultaneously or within a few minutes, suggesting spamming of the survey. To ensure data was trustworthy, responses which met at least two of the following criteria were removed prior to data analysis: an IP address located outside of the United Kingdom; an IP address which completed the study multiple times; completing the study in under 7 min; responding inconsistently to multiple items (e.g. strongly agreeing that they are satisfied with their appearance, then strongly agreeing that they dislike the way the look); submitting identical complex sentences to the open-ended questions as another participant; or reporting a significant language barrier. This led to data from 57 different IP addresses being deleted.

The final sample included 49 female, 25 male and 11 non-binary participants. One participant did not report their gender. The mean age was 32.12 (SD = 12.41) years. Participants reported their ethnicity as: White (n = 76, 88.4%), Asian (n = 2, 2.3%), mixed ethnic background (n = 6, 7.0%) and other ethnic groups (n = 2, 2.3%). Regarding employment, 32 participants (37.2%) were part-time or full-time students, 37 (43.0%) were in part-time or full-time employment, six (7.0%) were unemployed, one was retired (1.2%) and 10 (11.6%) were unable to work. Data on socioeconomic status and educational attainment level was not collected.

Participants’ scores on the three ‘social bifactor’ subscales of the Comprehensive Autism Trait Inventory (CATI; English et al., 2021) were above the recommended cut-offs: Social Interaction (Median = 28, Range = 10–35), Communication (Median = 23, Range = 12–32) and Camouflage (Median = 27, Range = 9–35). These subscales were used for characterising the sample rather than control variables in analyses as the current study focuses on a non-social autistic trait dimension (tactile sensitivity) and previous research (English et al., 2021) suggests relatively weak correlations between the social bifactor subscales and the CATI Sensory Sensitivity subscale. Thirty-eight participants had additional mental or neurodevelopmental conditions. The most frequent were anxiety disorders (n = 14), attention-deficit hyperactivity disorder (n = 14) and depression (n = 13). Other co-occurring conditions included Tourette’s syndrome, obsessive-compulsive disorder, emotionally unstable personality disorder, post-traumatic stress disorder, sensory processing disorder, dyspraxia, developmental language disorder, specific learning difficulties and conduct disorder.

Ethics

This project was granted ethical approval by the Department of Psychology at Royal Holloway, University of London. All participants read a participant information sheet and provided written informed consent prior to participating. In exchange for taking part in the study, participants could enter a draw to win one of two £25 Amazon vouchers. Participants’ contact details were kept separate from the data and were deleted after the draw.

Materials

Procedure

The survey was completed online via Qualtrics (https://www.qualtrics.com). After reading the information sheet and providing consent, participants completed the demographics questionnaire, Clothing Questionnaire, CARVAL and CARSAL (Moss & Rosser, 2012), RSES (Rosenberg, 1965), CATI (English et al., 2021), the closed question on importance of fashion style and four open-ended questions on clothing-related tactile sensitivity. Participants also completed the Anxiety Scale for Autism-Adults (Rodgers et al., 2020) and WHOQOL-BREF (The WHOQOL Group, 1998) which were not analysed in this study. The median time to complete all questionnaires was 17.68 min. Finally, participants were given the opportunity to provide their email address to participate in the prize draw.

Analysis

The quantitative data was analysed using IBM SPSS Statistics 28 and the PROCESS macro v4.2 for SPSS (Hayes, 2017). The prevalence of tactile hyper-sensitivity in this sample was calculated by summing the number of participants who responded ‘often’ or ‘always’ for the six items of the tactile hyper-sensitivity scale. Response frequencies for the closed question on the importance of having a good fashion style were also calculated. As the responses to the closed question were considered ordinal data, non-parametric Spearman’s correlations exploring the relationship between importance of fashion style and appearance awareness and dissatisfaction were conducted. Descriptive statistics, bivariate Pearson’s correlations, simple regression analysis and moderation analyses were conducted to test the three study hypotheses. Initial screening of the data determined that it met the assumptions of linearity, homoscedasticity, independence and normality.

Analysis of the open-ended questions was undertaken by the first author (A.F.K.) using Braun and Clarke’s (2006, 2022) methodology for reflexive thematic analysis. An inductive approach was used to capture semantic and latent patterns of meaning across the dataset. The six phases of analysis followed were: data familiarisation by re-reading transcripts; coding data; generating initial themes by clustering codes into broad patterns of meaning; developing and reviewing themes; defining and naming themes; and writing the report. The development and interpretation of codes and themes was discussed with the second author (D.B.) throughout the process.

Positionality statement

This research project was conducted by non-autistic researchers. The authors’ views align with the neurodiversity paradigm, which conceptualises neurodivergence as a natural and valuable variation of human cognition (Walker, 2014).

Participatory methods statement

The research questions were developed in line with the research priority of the autistic community to better understand sensory processing differences which can negatively affect autistic people in order to create more sensory-friendly environments (Autistica, 2016), and based on first-person accounts of autistic people found online. An autistic researcher provided feedback on the initial interpretation of the results.

Quantitative results

Participants of all genders reported a range of different opinions towards how important a good fashion style was to them (see Figure 1). Further analyses exploring gender differences were not possible due to low participant numbers when split into subgroups. There was a significant correlation between responses to the importance of fashion style and appearance awareness (r_s = 0.60, p < 0.001) confirming that individuals who rated fashion as important to them were more self-aware of their appearance. There was no significant correlation between appearance dissatisfaction and importance of fashion style (r_s = −0.10, p = 0.343).

OPEN IN VIEWER

Figure 1.

Importance of fashion style broken down by gender.

Regarding the frequency of hyper-sensitivity among participants, 73.2% reported they regularly avoid wearing certain types of clothing, 50% dislike having a haircut, 55.8% cut labels off clothes, 46.5% avoid wearing clothing items with seams that will contact their skin, 48.8% have difficulties adapting to new clothing items and 48.8% avoid wearing constricting clothes or shoes.

The zero-order correlation matrix for the key variables and the descriptive statistics are presented in Table 1.

OPEN IN VIEWER

Table 1.

Descriptive statistics and correlation matrix for predictor and outcome variables.

Variable	M	SD	1	2	3	4
1. Tactile hyper-sensitivity	20.87	5.00	-
2. Appearance dissatisfaction	30.98	9.42	0.21*	-
3. Appearance awareness	19.29	5.70	-0.07	-0.11	-
4. Self-esteem	23.53	6.36	-0.22*	-0.64**	-0.16	-

*

p < 0.050, **p < 0.001.

A simple linear regression examined the predictive relationship between tactile hyper-sensitivity and appearance dissatisfaction. The analysis indicated that the model was significant (F(1, 84) = 3.95, p = 0.05), with tactile hyper-sensitivity explaining 3.4% (R² adjusted = 0.034) of the variance of appearance dissatisfaction. The effect size (f² = 0.05) indicated a small effect. Higher tactile hyper-sensitivity was associated with higher appearance dissatisfaction (β = 0.40, p = 0.050).

A moderation analysis with centred variables examined the relationship between the predictor variable ‘tactile hyper-sensitivity’, dependent variable ‘self-esteem’ and moderator ‘appearance dissatisfaction’. The moderation model was significant (F(3, 81) = 22.52, p < 0.001) and explained 45.5% of the variance in self-esteem (R² = 0.455). The interaction was significant (b = −0.026, 95% CI = [−0.048, −0.003], t = −2.28, p = 0.025), indicating that there is a predictive relationship between hyper-sensitivity and self-esteem moderated by appearance dissatisfaction, with a large effect size (f² = 0.83). When appearance dissatisfaction is low (b = 0.074, 95% CI = [−0.191, 0.338], t = 0.55, p = 0.581) or at the mean value (b = −0.169, 95% CI = [−0.387, 0.049], t = −1.55, p = 0.126), there is a non-significant relationship between hyper-sensitivity and self-esteem. However, when appearance dissatisfaction is high, there is a significant negative relationship between hyper-sensitivity and self-esteem (b = −0.412, 95% CI = [−0.752, −0.073], t = −2.42, p = .018), indicating that the negative predictive relationship between hyper-sensitivity and self-esteem only emerges in people with higher levels of appearance dissatisfaction, not those with low or mean levels.

Another moderation analysis examined the relationship between the predictor ‘appearance dissatisfaction’, dependent variable ‘self-esteem’ and moderator ‘appearance awareness’. The moderation model was significant (F(3, 81) = 24.79, p < 0.001), explaining 47% of the variance. However, the interaction was non-significant (p = 0.134) indicating that appearance awareness does not influence the relationship between appearance dissatisfaction and self-esteem. Due to this, another model without the interaction was conducted. This hierarchical regression was found to be significant (F(2, 82) = 35.48, p < 0.001) and accounted for 45.1% (R² adjusted = 0.451) of the variance in self-esteem, with a large effect size (f² = 0.86). Higher appearance dissatisfaction (β = −0.67, p < 0.001) and higher appearance awareness (β = −0.23, p = 0.006) were significant predictors of lower self-esteem.

Qualitative results

Three main themes, with five sub-themes, relating to the impacts of clothing-related tactile sensitivity on autistic adults’ lives were generated. ‘Negative consequences’ captures the direct negative impacts of clothing-related tactile sensitivity. ‘Managing clothing-related tactile sensitivity’ captures how autistic adults cope with and adapt to living with clothing-related tactile sensitivity. ‘The emotional value of clothing’ captures how clothing has an important emotional role for autistic adults and positively impacts aspects of their well-being despite the various negative impacts of hyper-sensitivity.

Strengths and limitations

The present study builds on previous research identifying clothing which autistic adults find comfortable and uncomfortable by exploring their perception of clothing and the impacts of clothing-related sensitivity. Research on the impacts of clothing-related hyper-sensitivity on well-being and daily functioning is extremely limited and mostly qualitative. A mixed-methods design was used in this study to ensure autistic adults’ experiences were captured as accurately as possible while also examining these constructs quantitatively.

The sample was diverse in terms of gender, age, employment, co-occurring conditions and opinions regarding the importance of fashion. However, autistic people of colour were under-represented. Autistic adults without a formal diagnosis were excluded due to concerns around the reliability and accuracy of findings, as well as those with an intellectual disability. There was no community involvement during the design or implementation of this study due to time constraints and lack of funding – which future research should address and include. Notably, the study captured the views and experiences of 86 autistic adults through the analysis of open-text responses, allowing for a variety of experiences of clothing-related tactile sensitivity within the autistic community to be reflected and mitigating the risk of bias. Although clothing-related tactile sensitivity was common among participants, not all of them were impacted equally by this. While some participants – such as those dissatisfied with their appearance – were significantly affected by this emotionally and in their day-to-day life, others were less affected. Moreover, participants preferred and disliked very different clothing items. For example, some participants preferred loose clothing as they found tight clothing ‘suffocating’, whereas other participants found loose clothing uncomfortable as this ‘moves about too much’ and can be ‘tickly’. Furthermore, tactile sensory stimuli can also be enjoyable and comforting for autistic adults when this has a physical feel they like (MacLennan, O’Brien, et al., 2022; Robertson & Simmons, 2015). This highlights the individualistic nature of sensory sensitivity and, consequently, the importance of autistic people and those supporting them understanding their unique sensory profile.

Due to time constraints and a lack of measures examining clothing-related hyper-sensitivity, items were adapted from the GSQ 70-item version (Robertson, 2012) to examine this construct in this study. Although this had acceptable reliability and items aligned with the experiences of participants in this and previous studies (e.g. MacLennan, O’Brien, et al., 2022), a more rigorous measure needs to be created to ensure findings in this research area are accurate and generalisable. In particular, great care must be taken when formulating questions about participants’ experiences of highly subjective concepts such as ‘a good fashion style’, as a participant may interpret this in several ways including fitting socio-cultural standards of appearance or meeting their desired personal style.