Taking Off the Mask: Investigating Autism Diagnosis and Camouflaging in Adult Women

Abstract

Background:

Awareness and recognition of autism have grown significantly in recent years. Despite this progress, research on the diagnosis and camouflaging behaviors in adult autistic women remains limited. This study seeks to address this gap by examining the characteristics of autism in adult women, with a particular focus on understanding the role of camouflaging in the challenges these women face in obtaining a formal diagnosis.

Methods:

Through a cross-sectional study involving 253 autistic participants, we administered a comprehensive battery of psychological assessment instruments to collect data on autistic traits, camouflage behaviors, stress, anxiety, depression, and diagnosis/suspicion-related factors.

Results:

Our findings revealed that the traits and characteristics of autism did not significantly differ between women and men or non-binary individuals. However, there were significant delays in the formal identification of autism in women, who had a lower rate of early autism diagnosis compared with other genders. In contrast, women often received prior diagnoses of other psychological conditions before their autism was identified. While camouflage behaviors were significantly more common among women than men, these patterns did not directly explain the delay in suspicion and diagnosis in this population. Although camouflaging predicted higher levels of anxiety, stress, and depression, especially in women, there were no consistent associations between masking measures and early autism identification.

Conclusions:

These findings suggest that while camouflaging behavior may exacerbate psychological distress and make autism more invisible for women, it is not the primary factor contributing to the delayed identification. Instead, gender stereotypes and the inadequacy of clinical assessment tools tailored to more typically feminine presentations of autism are likely major obstacles in the diagnostic process.

Community Brief

Why is this an important issue?

Autistic traits are pivotal aspects of individuals’ identities, yet they are frequently misunderstood, especially in girls and women. Recognizing and diagnosing autism in women are crucial for providing self-awareness and appropriate support, when necessary, and improving their quality of life.

What was the purpose of this study?

This study aimed to understand the characteristics of autism in adult Brazilian women, including how they might mask their traits and the challenges they face in obtaining a formal identification.

What did we do?

We conducted a survey involving 253 autistic adults and used several psychological assessment tools to collect data on stress, anxiety, depression, camouflage behaviors, autism-related characteristics, and the processes of suspicion and diagnosis. By comparing data from women with those of men and non-binary individuals, we aimed to provide a comprehensive understanding of autism across genders.

What were the results of the study?

Our research revealed that autism traits were relatively consistent across women, men, and non-binary individuals. However, women experienced more significant delays in receiving an autism diagnosis compared with other genders. It was notably more common for women to receive prior diagnoses of other psychological conditions before being identified as autistic, compared with men. Women frequently employ camouflaging behaviors more than men, but these behaviors do not appear to directly account for the delays in autism identification, although they are linked to increased feelings of anxiety, stress, and depression, especially among women.

What do these findings add to what was already known?

These findings highlight the complexity of identifying and diagnosing autism in women and challenge previous assumptions about the role of camouflage in diagnosis delays. The study emphasizes the need for improved awareness among health care professionals and more sensitive assessment tools to accurately identify autism in women. By addressing these challenges, we can ensure timely and accurate diagnoses, leading to better support and outcomes for autistic women.

What are potential weaknesses in the study?

While the study focused primarily on autistic women, its limitations provide insights into other overlooked populations. The small sample size for non-binary individuals limits the generalizability of the findings to this group. Future research should aim to include larger and more diverse samples to better understand the experiences of non-binary individuals on the autism spectrum.

How will these findings help autistic adults now or in the future?

In summary, this study contributes to a growing body of research aimed at improving the recognition and support of autistic individuals, particularly women, in Brazil and beyond. By raising awareness of the challenges women face in receiving an autism identification, we can pave the way for better support and understanding, ultimately enhancing the quality of life for all individuals on the autism spectrum.

Introduction

Autism has increasingly been recognized as a natural variation in human neurodevelopment, and while understanding these characteristics can be essential for the well-being of autistic individuals, the benefits of its early identification have only been investigated in health contexts. Despite being historically rooted in a purely medical perspective, the formal identification of autism, also referred as autism spectrum disorder in this context, has undergone numerous adjustments in its “nosological characteristics” since it was first identified in the mid-1940s by psychiatrist Leo Kanner.^1–4 This progress has reached the present day, as the concept of autism has grown more inclusive, emphasizing neurodiversity over merely clinical definitions; this shift has contributed to a reduction in the stigma faced by autistic individuals, enabling a greater number of people to embrace their identities within the autism community.⁵ Thus, a consensus on the global prevalence and incidence of autism remains elusive. Going further, most studies are conducted in the Northern Hemisphere,^6–8 leading to a lack of understanding of this population in other worldwide regions, where data on autistic minorities and cultural nuances are still scarce.⁹ Alongside this knowledge gap, there are misconceptions about a supposed “autism epidemic,” fueled by a significant rise in identifications and diagnoses.^5,10 Previous estimates suggested that 1 in every 68 American children would be diagnosed with autism; however, more recent data now indicate a ratio closer to 36:1.¹¹ It is not only this proportion of diagnosis that is under scrutiny, but gender-related prevalence is also undergoing recent revisions.

In terms of gender-related attributes, there is a significantly higher number of diagnoses in the male population, with evaluations suggesting that for every five diagnosed children, four will be boys, and one will be a girl.¹² Despite these data, the gender imbalance in autism incidence has been questioned through new research, which suggests that the autistic female population might be significantly larger than the initial projections indicated. This hypothesis suggests that autism diagnosis protocols, primarily developed from research conducted predominantly with boys, may introduce biases that overlook or underestimate the specific characteristics of autistic girls and women; as a result, the manifestation and identification of autism in this population may often go unnoticed or be misdiagnosed.^7,8,13 According to Rynkiewicz and colleagues,⁸ diagnostic assessments are not sensitive enough to the clinical presentations of women with traits of autism, leading to the oversight of female autism and these women’s characteristics by diagnostic services and the health system.

Additionally, a complementary hypothesis is that autistic girls develop greater skills in masking neurodivergent traits. In other words, they can use a set of strategies to hide or minimize their autistic traits in order to seek greater adaptability and integration into social settings—a phenomenon known as autistic masking or camouflaging.¹⁴ Although masking may seem beneficial initially, enhancing conformity to social expectations and avoiding judgment, rejection, and discrimination, it can result in increased stress and emotional strain, as there is a constant effort by the individual to suppress their natural behaviors, aiming to conform to a supposedly desirable behavioral pattern.¹⁵ To differentiate masking from typical behavior, it is recognized that individuals engaging in masking exhibit noticeable differences between their social behavior with peers and their internal behavior, including the imitation of body language and facial expressions of others, for example.^15,16 This camouflaging tendency can contribute to the underdiagnosis of females, as autistic girls and women may be more skilled at disguising their characteristics compared with males, potentially further exacerbating the already underestimated rates of autism among females driven by diagnostic criteria.

As a result of the masking process or the criteria utilized, the formal identification of female autism often occurs late, sometimes in adolescence or adulthood.^6,12 Late diagnosis impacts the exhaustive pursuit of self-awareness and an accurate understanding of traits and behavioral patterns, encountering various obstacles, such as the significant lack of knowledge among health care professionals regarding the specificities of this female population. This lack of awareness can lead to misdiagnoses (e.g., borderline personality disorder and bipolar mood disorder) and affect treatment and prognosis, when necessary.¹⁷ Another hypothesis suggests that these difficulties encountered may be related to a kind of “female autism phenotype”,^6,14 a pattern of female-specific traits that differ from those found in typical assessments of autism, which are based on a predominantly male perspective, leading to a gap in understanding how characteristics of autism presents in girls and women. According to this model, women generally exhibit greater social awareness, display stronger motivation for forming friendships and engage more actively in group activities and reciprocal conversations, use more nonverbal gestures, and have fewer repetitive or restrictive behaviors compared with autistic men.¹⁴ Thus, these behavioral patterns often go unnoticed or are undervalued by parents, caregivers, teachers, and health professionals in the early years of life.¹⁸ This occurs either due to the currently used diagnostic criteria, established on stereotypes and characteristics typical of “male autism,” or due to masking behaviors; both factors potentially contribute to the delayed identification. In contrast, with the diagnosis, many girls and women realize they need not conform to neurotypical standards, thereby enhancing their self-compassion and self-awareness.¹³ Additionally, formal diagnosis serves as a way to prevent psychological distress, minimizing anxious and depressive symptoms and, in more severe cases, potential suicidal behaviors.¹⁴ Although late diagnosis is common among autistic women,^6,19 studies on the implications of this delay in the diagnosis of female autism remain scarce.

In light of the above, the importance of formal identification, even if delayed, for the well-being and quality of life of autistic girls and women requiring minimal support is acknowledged. Over time, it has become evident that a gender-specific perspective on autism is essential and a priority to comprehend the unique experiences of autistic women, which may differ from those of men. Furthermore, by understanding the peculiarities of this population, more studies can be conducted to comprehend their characteristics, traits, and consequences, such as camouflaging, and refine outdated assessments, allowing for a restructuring of health care services with a focus on fair and appropriate medical and psychological support tailored to the needs of all autistic populations.¹³ Hence, this study aims to investigate the characteristics of autism in adult women, more specifically in those with more subtle autistic traits, by comparing their behavioral and affective patterns with those of other genders. This approach seeks to understand the roles of these variables, particularly autistic camouflaging, in the challenges encountered in obtaining a formal diagnosis for this population.

Methods

Design, procedure, and participants

For this cross-sectional study, we recruited 253 individuals exhibiting autistic traits, whether formally diagnosed or self-identified, at support level 1, irrespective of gender, aged 18–75, domiciled in the country, and/or holding Brazilian nationality. Sampling of men and non-binary individuals aimed to establish reference groups for comparisons with the women. Inclusion criteria for recruitment were declaring oneself as neurodivergent and requiring minimal support (level 1 autism, encompassing the former classification of Asperger syndrome). We chose this criterion due to this population’s invisibility within the autistic community itself despite serious consequences, such as high suicide rates.²⁰ Furthermore, only participants devoid of conditions that could affect the completion of psychological assessments were included: incomplete elementary education, inattention during data collection (as verified through attentional check), and prior emotional sensitivity (immediately after obtaining consent, participants were asked if they were feeling particularly anxious or “nervous” at the moment; if affirmative, they were invited to participate on another occasion). To ensure proper understanding of the items and/or the validity of the instruments used, we included only native Brazilian Portuguese speakers and/or permanent residents of the country, as well as individuals up to 75 years of age. The recruitment strategy aimed to reach individuals who consider themselves part of the autistic community, meaning those who self-perceive as autistic, whether with a formal diagnosis by a health care professional or with just an informal evaluation. However, to confirm that the participants we sampled genuinely exhibited significant neurodivergent traits, we checked their total scores on the Autism Spectrum Quotient (AQ). We recruited participants by sending invitations and disseminating them through groups, communities, associations, institutes, and nongovernmental organizations engaged in disseminating information about autism within the Brazilian context, including local and/or national groups of students, family, friends, and/or mutual support, both online and offline.

Initially, 449 individuals agreed to participate in the study. Participants who did not complete all data collection steps (187), failed the attentional check (2), or did not indicate that they self-declared as autistic (7) were excluded from the data analysis, leaving a total of 253 participants. Thus, we collected data through SoSci Survey, a specialized cloud-based platform for online survey development. After volunteers willingly agreed to take part in the research, providing prior consent via the Informed Consent Form, they accessed a comprehensive sociodemographic and health questionnaire, followed by a battery of psychological tests: AQ,²¹ Camouflaging Autistic Traits Questionnaire (CAT-Q),¹⁵ Need to Belong Scale (NTBS),²² and Depression, Anxiety, and Stress Scale (DASS).²³ All procedures followed guidelines and standards for research involving human participants from the National Health Council. This study was approved by the Research Ethics Committee of the Federal University of Rio Grande (approval code 69182923.4.0000.5324).

Instruments

Sociodemographic and health questionnaire

The sociodemographic and health questionnaire collected general information, such as gender, age, and region of the country of residence, as well as additional details regarding the autism diagnostic process. This included whether there was a diagnosis/identification with formal testing, the age at which they received the diagnosis, suspicions of autism, and any secondary condition or previous diagnoses.

Autism Spectrum Quotient

Developed by Baron-Cohen and colleagues,²¹ this tool aims to identify traits and key characteristics of autism. The adult version of the scale, also known as the AQ, is a self-report instrument comprising 50 items across five subscales: social skills, imagination, communication, attention switching, and attention to detail. Respondents rate each statement on a 4-point Likert scale, with higher scores indicating a greater presence of autistic characteristics. The scale’s authors suggest a cutoff score of 32 as a useful threshold for distinguishing between individuals likely to be neurodivergent and neurotypical.²¹ Psychometric studies have confirmed the suitability of the scale for use in the Brazilian population, with various studies demonstrating its reliability and validity across diverse samples.^24,25 It serves as a psychological measurement tool for screening behavioral patterns related to autism, although it does not aim to establish diagnoses.²⁴

Camouflaging Autistic Traits Questionnaire

The CAT-Q is a self-report measure designed to assess camouflaged autistic social behaviors in adults. Developed by Hull and colleagues,¹⁵ the questionnaire consists of three subscales: Compensation (strategies to overcome difficulties associated with autism), Masking (strategies used to hide autistic characteristics in public), and Assimilation (strategies to avoid standing out negatively in social interactions). In total, the CAT-Q comprises 25 items rated on a 7-point Likert scale, with a maximum score of 175 points, indicating a higher degree of camouflage of autistic traits. The CAT-Q demonstrates high levels of internal consistency and reliability, with Cronbach’s alpha coefficient of 0.94 for the total scale and alpha above 0.85 for each factor of the subscales.

Need to Belong Scale

The NTBS is an instrument conceived by Leary and colleagues,²² adapted and validated for use in Brazil by Gastal and Pilati.²⁶ Its purpose is to measure an individual’s motivation toward establishing social connections and the value they attribute to the social acceptance received from others. This measure consists of 10 items, arranged in a unifactorial manner, where respondents indicate the degree of agreement with the statements provided. Higher scores on this scale indicate greater sensitivity to social cues, reflecting an individual’s ability to perceive others’ emotions and identify situations of social rejection. Consequently, individuals with higher scores tend to exhibit more socially desirable behaviors for the maintenance of relationships and emotional bonds.²² According to Gastal and Pilati,²⁶ the Brazilian version of the NTBS has shown good reliability (i.e., α = 0.82), as well as validity and practicality in its application.

Depression, Anxiety, and Stress Scale

The 21-item version of DASS (DASS-21) is a concise screening tool composed of 7 items for each dimension: anxiety, depression, and stress. Respondents rate the extent to which each item applies to them over the past week using a Likert scale ranging from 1, “Did not apply to me at all,” to 4, “Applied to me very much, or most of the time.” Derived from the original 42-item DASS, the abbreviated version of DASS was developed by Lovibond and Lovibond²³ and was translated and validated for use in Brazil by Machado and Bandeira.²⁷ Research has shown that both the full 42-item DASS and its abbreviated 21-item version are valid and reliable measures of depression, anxiety, and stress in populations of adults, children, and adolescents across various cultures.²⁸

Data analysis

The data collected from the psychological assessment battery underwent preprocessing, which involved cleaning the database to remove incomplete or inaccurately filled-out questionnaires. Subsequently, we calculated scores and performed descriptive and exploratory analyses to assess the statistical models’ assumptions and estimate the data’s frequencies, means, and standard deviations. For comparisons among genders, after verifying the assumptions, we employed generalized linear models (GzLM) with a binomial distribution and a logit link function to estimate the main effects on binary variables, while we used models with a normal distribution and an identity link for linear-response data. These analyses were conducted for all comparisons among genders. We also carried out gender-specific models to evaluate the predictive role of the CAT-Q on other response variables, we also used GzLM with a normal distribution, conducted separately for each gender. However, these models were only conducted for the women and men groups, as the non-binary sample size in our study was too small to yield accurate inferences in these gender-specific models. In all GzLM tests, we used canonical link functions, a significance level of 0.05, and two-tailed hypothesis testing.

Results

The sample comprised 253 participants, from all regions of the country, who self-identified as autistic. Among the participants, the majority were women (67.6%), with an average age of 33.1 years (standard deviation [SD] = 10.1; ♀, mean [M] = 33.9; SD = 10.3), had high levels of education, with some form of postgraduate education (31.6%; ♀ = 36.8%), and were single (55.7%; ♀ = 50.9%). Regarding the presence of an autism identification among participants, approximately 63.6% (♀ = 61.4%) reported having received a formal diagnosis from a health care professional. The average age of suspicion of autism was 26.65 years (SD = 11.7) for women, 24.3 years (SD = 11.8) for men, and 20.1 years (SD = 9.6) for non-binary individuals (and for the total sample, M = 25.6; SD = 11.7). In turn, the formal diagnosis was received by women participants, on average, only at the age of 32.1, which is descriptively later than for the men (M = 29.1; SD = 11.2) and non-binary groups (M = 24.0; SD = 8.1) (and for the total sample, M = 30.7; SD = 10.1). In absolute numbers, the difference in years from the age of suspected autism to the formally diagnosed age was wider for women (5.5 years) compared with men (4.8 years) and non-binary individuals (3.9 years). The findings descriptively showed that, on average, women take 2.3 years longer than men to suspect autism and three years longer to receive the correct diagnosis of the condition. Moreover, women also receive more diagnoses of psychological disorders in general, with about 84.8% of respondents claiming to have one or more conditions diagnosed before autism. This percentage decreases to 63.5% for men and increases to 88.2% for the non-binary participants. The descriptive statistics can be seen in more detail in Table 1.

Table 1.

Descriptive Statistics Categorized by Female, Male, and Non-Binary Samples

	Female (n = 171)			Male (n = 63)			Non-binary (n = 17)
	M ± SD	Min	Max	M ± SD	Min	Max	M ± SD	Min	Max
Age (years)	33.97 ± 10.31	18	72	32.73 ± 9.58	18	60	25.71 ± 7.38	19	47
Suspicion (years)	26.59 ± 11.74	2	68	24.30 ± 11.83	1	48	20.12 ± 9.64	5	41
Formal diagnosis (years)	32.13 ± 9.39	13	62	29.12 ± 11.21	7	50	24.00 ± 8.05	16	46
AQ Attention to detail	6.94 ± 1.85	1	10	6.17 ± 2.03	1	10	6.59 ± 1.91	3	10
AQ Social Skills	7.78 ± 1.87	1	10	7.84 ± 1.85	2	10	7.88 ± 1.90	3	10
AQ Attention switching	8.67 ± 1.36	4	10	8.64 ± 1.27	5	10	8.77 ± 1.25	6	10
AQ Communication	7.39 ± 1.99	2	10	7.29 ± 1.92	2	10	7.53 ± 1.63	4	10
AQ Imagination	5.21 ± 2.30	0	10	5.44 ± 2.39	1	10	5.29 ± 1.93	1	8
AQ Total	35.98 ± 6.34	16	47	35.38 ± 5.50	22	47	36.06 ± 4.24	29	43
CAT-Q Masking	38.90 ± 7.97	11	56	35.86 ± 9.85	8	51	42.82 ± 4.95	32	49
CAT-Q Assimilation	44.49 ± 7.88	19	56	42.67 ± 8.00	18	56	44.47 ± 5.99	31	53
CAT-Q Compensation	47.48 ± 9.60	15	63	44.33 ± 10.29	19	63	50.65 ± 8.86	35	62
CAT-Q Total	130.86 ± 20.16	46	173	130.86 ± 20.16	72	166	137.94 ± 15.21	110	157
DASS-21 Stress	20.15 ± 5.18	8	28	20.46 ± 4.90	10	28	21.29 ± 4.07	14	28
DASS-21 Anxiety	16.66 ± 5.63	7	28	15.95 ± 6.04	7	28	17.53 ± 5.06	11	28
DASS-21 Depression	18.64 ± 6.01	7	28	20.07 ± 6.32	7	28	19.24 ± 5.89	10	28
NTBS	31.15 ± 7.75	14	47	31.38 ± 8.04	18	47	34.82 ± 7.16	17	45

Suspicion: age at which suspicion of autism began; formal diagnosis: age at which the formal diagnosis was made.

AQ, Autism Spectrum Quotient; CAT-Q, Camouflaging Autistic Traits Questionnaire; DASS-21, 21-item version of Depression, Anxiety, and Stress Scale; M, mean; NTBS, Need to Belong Scale; SD, standard deviation.

Inferential comparisons among genders

In order to understand the relationships established between gender and variables related to the diagnostic process, we used GzLM with normal or binomial distribution (see Table 2). Thus, initially, we tested whether there are significant gender differences in the age of formal diagnosis, which was confirmed by our model (Wald’s χ² = 9.53; p = 0.009). The women (reference group) differed significantly from the non-binary group (p = 0.005) but did not reach the level of significance when compared with men (p = 0.088). We also sought to verify the occurrence of autism identification during childhood and adolescence (i.e., before the age of 20) and its relationship with the gender of autistic individuals. Notably, the gender variable was significantly related to the earlier diagnosis of autism (Wald’s χ² = 8.24; p = 0.016), wherein women significantly differed from both males (p = 0.010) and non-binary individuals (p = 0.031). Similarly, gender was a determining factor for receiving diagnoses of other psychological conditions before the formal identification of autism (Wald’s χ² = 12.90; p = 0.002), particularly contrasting women with men (p = 0.001), while no significant difference was observed with the non-binary sample (p = 0.705). We also investigated whether early suspicions of autism were linked to the gender of autistic individuals. Indeed, although gender did not reach the level of significance, there was a discernible tendency indicating that women exhibited the lowest rates of early suspicions, whether considering the age of suspicion (Wald’s χ² = 5.86; p = 0.053) or the occurrence of suspicion before the age of 20 (Wald’s χ² = 4.66; p = 0.097); it is worth noting that we used two-tailed hypotheses for all analyses. Therefore, overall, our data confirm that gender is related to the early identification of autism, as well as it may play a role in clinical suspicion of other conditions.

Table 2.

Coefficients Fitted by the Generalized Linear Models for Gender Differences (n = 253)

			Comparisons of parameter estimates
	Model effects		Intercept (female)^a	Male		Non-binary
	W	p	B (SE)	B (SE)	p	B (SE)	p
Formal diagnosis (years)	9.53	0.009	32.13 (0.95)	−3.01 (1.76)	0.088	−8.13 (2.86)	0.005
Formal diagnosis (under 20)	8.24	0.016	−3.15 (0.39)	1.36 (0.53)	0.010	1.61 (0.74)	0.031
Suspicion (years)	5.86	0.053	26.59 (0.88)	−2.29 (1.71)	0.179	−6.47 (2.94)	0.028
Suspicion (under 20)	4.66	0.097	−0.91 (0.17)	0.36 (0.31)	0.249	1.03 (0.52)	0.045
Previous psychological diagnoses	12.90	0.002	1.72 (0.21)	−1.17 (0.34)	0.001	0.30 (0.78)	0.705
AQ \| Social Skills	0.09	0.956	7.78 (0.14)	0.06 (0.27)	0.816	0.11 (0.47)	0.825
AQ \| Attention switching	0.13	0.937	8.67 (0.10)	−0.03 (0.20)	0.870	0.09 (0.34)	0.770
AQ \| Attention to detail	7.55	0.023	6.94 (0.15)	−0.76 (0.28)	0.006	−0.35 (0.48)	0.470
AQ \| Communication	0.25	0.881	7.39 (0.15)	−0.11 (0.29)	0.711	0.14 (0.49)	0.781
AQ \| Imagination	0.51	0.775	5.21 (0.18)	0.24 (0.34)	0.477	0.09 (0.58)	0.878
AQ \| Total	0.48	0.786	35.98 (0.46)	−0.60 (0.88)	0.500	0.08 (1.52)	0.957
CAT-Q \| Masking	11.40	0.003	38.90 (0.63)	−3.04 (1.22)	0.013	3.93 (2.11)	0.062
CAT-Q \| Assimilation	2.59	0.274	44.49 (0.59)	−1.82 (1.14)	0.112	−0.02 (1.97)	0.994
CAT-Q \| Compensation	7.60	0.022	47.48 (0.74)	−3.15 (1.43)	0.027	3.17 (2.46)	0.198
CAT-Q \| Total	10.37	0.006	130.86 (1.56)	−8.00 (3.00)	0.008	7.08 (5.18)	0.171
DASS-21 \| Stress	0.88	0.643	20.15 (0.39)	0.31 (0.75)	0.679	1.15 (1.28)	0.370
DASS-21 \| Anxiety	1.26	0.534	16.66 (0.44)	−0.71 (0.85)	0.402	0.87 (1.44)	0.547
DASS-21 \| Depression	2.52	0.283	18.64 (0.47)	1.43 (0.90)	0.114	0.60 (1.54)	0.697
NTBS	3.50	0.174	31.15 (0.59)	0.23 (1.14)	0.838	3.68 (1.97)	0.062

Suspicion: age at which suspicion of autism began; formal diagnosis: age at which the formal diagnosis was made.

Reference group.

SE, standard error; W, Wald χ².

To explore which other autistic traits might differ between genders, potentially contributing to the delayed identification of autism in women, differences among genders for autism characteristics were also tested using GzLM with a normal distribution. We examined whether participants’ gender could predict the subfactors of the AQ scale; curiously, the subfactors of the scale did not show significant differences (p’s > 0.775; see Table 2) concerning gender, except for the “attention to detail” factor (Wald’s χ² = 7.55, p = 0.023). Furthermore, within our sample, no gender disparities were observed for the various subscales of the DASS-21 (p’s > 0.283) and for the sense of belonging measure (p = 0.174), as presented in Table 2. However, as expected, gender was a significant factor for camouflage variables: Compensation, Wald’s χ² = 7.60; p = 0.022; Masking, Wald’s χ² = 11.40; p = 0.003; Assimilation, Wald’s χ² = 2.59; p = 0.274; CAT-Q Total Score, Wald’s χ² = 10.37; p = 0.006 (Table 2). With the exception of the assimilation measure, which did not reach the level of significance, all significant contrasts were exclusively between female and male genders (all p’s < 0.027).

Predictive models of camouflaging autistic traits for each gender

We conducted separate models for each gender to test the predictive relationship between camouflage measures and diagnostic outcomes, as well as sense of belonging, and psychological distress (i.e., depression, anxiety, and stress). Due to the limited sample size, we did not carry out such models for the non-binary population. Thus, we used camouflaging autistic traits as predictor variables through GzLM with normal distribution (see Table 3). For women, the overall CAT-Q score consistently predicted anxiety (Wald’s χ2 = 32.16; p < 0.001), stress (Wald’s χ² = 14.79; p < 0.001), depression (Wald’s χ² = 9.74; p = 0.002) symptoms, and sense of belonging (Wald’s χ² = 14.66; p < 0.001). The models for the total CAT-Q score for the men were also significant; however, all models had a considerably smaller effect size (compared with those for women): Anxiety, Wald’s χ² = 16.96; p < 0.001; Stress, Wald’s χ² = 6.89; p = 0.009; Depression, Wald’s χ² = 9.18; p = 0.002; Belonging, Wald’s χ² = 6.69; p = 0.010.

Table 3.

Generalized Linear Models for Each Gender to Test the Predictive Relationship of Camouflage Measures (CAT-Q) with Early Identification of Autism, Previous Diagnoses of Other Psychological Conditions, Sense of Belonging, and Psychological Distress

Camouflaging autistic traits	Model effects
	Female		Male
	W	p	W	p
Formal diagnosis (years)	3.93	0.048	0.04	0.835
Formal diagnosis (under 20)	1.69	0.193	1.38	0.240
Suspicion (years)	0.25	0.614	0.01	0.924
Suspicion (under 20)	0.01	0.912	0.333	0.564
Previous psychological diagnoses	2.93	0.087	0.866	0.352
DASS-21 \| Stress	14.79	1.2e-4	6.89	0.009
DASS-21 \| Anxiety	32.16	1.4e-8	16.96	3.8e-5
DASS-21 \| Depression	9.74	0.002	9.18	0.002
NTBS	14.66	1.3e-4	6.69	0.010

Suspicion: age at which suspicion of autism began; formal diagnosis, age at which the formal diagnosis was made.

Interestingly, the total camouflage score did not predict early diagnosis during childhood or adolescence in both genders (♀, Wald’s χ² = 1.69; p = 0.193; ♂, Wald’s χ² = 1.38; p = 0.240), suggesting that factors other than camouflage may be associated with the delayed identification of autism in women. To reinforce this notion, we conducted a GzLM to investigate whether CAT-Q predicts the age of autism diagnosis, which was weakly confirmed, but only among women: ♀, Wald’s χ² = 3.93; p = 0.048; ♂, Wald’s χ² = 0.04; p = 0.835. Concerning diagnoses of personality and other disorders—the most frequently reported diagnoses were borderline personality disorder, major depressive disorder, and bipolar disorder—we utilized the binomial GzLM to observe if CAT-Q predicts the receipt of other diagnoses before autism. Although the data revealed some differences between genders in the patterns of the results, none reached significance for two-tailed hypotheses, neither for women (Wald’s χ² = 2.93; p = 0.087) nor for men (Wald’s χ² = 0.866; p = 0.352), regarding the relationship between the level of camouflage and other diagnoses. Generally, these data suggest a tendency that camouflaging autistic traits in women may be a subtle risk factor for misdiagnosis or the occurrence of other conditions, albeit in almost negligible proportions.

Discussion

This study aimed to investigate the characteristics of autism in adult Brazilian women, comparing their behavioral and affective patterns with those of other genders, identifying camouflage processes, and examining the challenges encountered within the pursuit of diagnosis in the national context. The data indicate that the female gender receives more diagnoses of other psychological conditions, including borderline personality disorder, depressive disorder, and bipolar disorder. Additionally, women tend to experience a longer waiting time between suspicion and formal autism identification, obtaining significantly fewer early diagnoses than other genders, which may be related to potential difficulties in recognizing and categorizing autism in women. As for autistic masking, it was possible to identify significant gender-related differences. Among adults, autistic women (with a more subtle form of autism) who use camouflage for social integration exhibit heightened levels of anxiety, stress, and depressive symptoms, alongside a diminished sense of belonging to the social circles they are part of. Men also exhibited these aspects related to camouflage but with less expressiveness than women. Interestingly, despite posing challenges in social relationships for individuals, camouflage does not appear to be directly related to delays in autism identification. However, it may be only subtly and indirectly associated with the diagnosis of other psychological conditions in the female population. Our findings suggest that masking does not expressively explain the misdiagnosis and delay in identification of autism in adult women, implying that other factors may be contributing to this context.

The difficulty in identifying, observing, and evaluating autism in women is already discussed by several authors in the international literature^18,29–31 with some proposed explanations for this phenomenon, such as differences in phenotypic characteristics⁶ or the masking of autistic traits.^14,15 Consistent with this pattern, masking or camouflage was indeed higher in our female sample; however, in our study, there were not enough observed gender-related differences to fully support these hypotheses. On the contrary, our findings contribute to partially refuting these explanations: the data we gathered do not indicate robust autistic trait/phenotypic differences between men and women (which, although present, are weak in our data) and emphasize that masking is linked to gender but in a less relevant way for the diagnosis of autism, being commonly observed in both women and men. Similarly, Milner and colleagues,³² in a recently published study, also found that women are diagnosed later than men, even when they report higher levels of autistic traits. Moreover, these authors found no relationship between autistic camouflaging scores and the age of diagnosis in either men or women, nor did they find any significance in the interaction between gender and the CAT-Q in predicting the age of diagnosis.

Thus, it is understood that one of the primary indicators that might contribute to the challenges in identifying autism in women could be the delay of professionals in observing and evaluating the clinical presentation, reflected in the time between suspicion and formal diagnosis, and especially in the age of diagnosis. Despite a significant scarcity of recent advances in psychological tools for autism assessment in developing countries,³³ this inadequacy does not account for the gender disparity in early identification rates and misdiagnosis, since such differences are observed worldwide.³² Historically, the diagnostic bias in autism has resulted in men being diagnosed more frequently than women, with stereotypically male traits being perceived as “more characteristic” of autism.³⁴ That is, gender stereotypes influence how autistic behavioral patterns are perceived and identified by health care professionals. Therefore, it is imperative to investigate the role of gender stereotypes associated with counseling and clinical evaluation, which may directly contribute to the ongoing underreporting of autism cases in girls and women, thereby impacting autism incidence rates. The clinical depiction of autism remains strongly linked to traditional behavioral characteristics more commonly observed and socially accepted in boys (e.g., disruptive behaviors low frustration tolerance),^35,36 making diagnoses in girls unfeasible, as they, for various reasons, including social pressures and sexism, tend to express such behaviors less.

Regarding traditional psychological assessment instruments, it is necessary to scrutinize biases and their actual predictive capabilities, which may be compromised and require adjustments based on gender, age, and context.³⁰ Some studies are evaluating the sensitivity and accuracy of autism assessment and screening tools, such as the Autism Diagnostic Observation Schedule (ADOS),³⁷ and the results notably illuminate gender differences. Merely 21% of autistic girls would meet the criteria for autism of ADOS, whereas 51% of autistic boys would receive the same identification based on the instrument’s scores. This gender bias may not only increase false negatives for autism diagnosis in women but also lead to false positives, misdiagnosing other conditions for them.

We must also take into account other specificities beyond gender bias, such as the presence of coexisting conditions with other neurodevelopmental effects. Recent research indicates that autistic girls with intellectual disabilities are more likely to be diagnosed than those without such differences.³⁸ Within the sample of this study, a significant portion of participants received their diagnosis in adulthood, similar to findings from international studies emphasizing late formal identification of autism as an increasingly common occurrence, especially among women.³¹ While not ideal, a late diagnosis brings the opportunity for self-acceptance and self-awareness.^39,40 Moreover, it facilitates access to essential social rights: in Brazil, for instance, some of these rights include the Autism Identification Card and preferential treatment in leisure, health, and educational services.

The underestimation of the autistic female population becomes evident when considering data on camouflage, the increasing incidence rates, and psychometric instrumental biases. Beyond these factors, the social aspect plays a pivotal role in such prevalence. Social norms and expectations imposed on girls contribute to both typical and atypical girls seeking to fit into desired behavioral patterns from childhood.^41,42 In the case of neuroatypical girls, these expectations result in excessive emotional strain in social interactions, leading to autistic burnout, as they feel compelled to mimic or perform in front of others to gain a sense of belonging.⁴⁰ This social and historical pressure contributes to the camouflage process and complicates diagnosis, as health care professionals may also apply the same social criteria within the clinical context. Without clearly disruptive social behavior, there is a lack of clarity in the clinical diagnosis of autism in girls and women. Thus, by comprehending the aspects of camouflage, psychometric instruments, and the social context, we can identify the factors perpetuating the disparity and underreporting of cases of female autism.

The current study provided valuable insights into camouflage and autistic traits in adult Brazilian women, advancing empirically and shedding light on how these factors may impact the identification of autism in women. However, it is imperative to acknowledge certain limitations of the study here. Generalizations regarding the outcomes obtained from our analyses for the non-binary population and their comparison with the female population are subject to significant constraints. The sample of non-binary respondents was notably smaller than other groups, potentially compromising the confidence in the results for the non-binary population and also hindering the conduct of certain statistical analyses involving this group. Future research endeavors should exert greater effort in recruiting this population in a more substantial manner, enabling more accurate comparisons with binary genders, for example. Alternatively, researchers could focus exclusively on this population, which is considerably larger among autistic individuals than in the general population.^43,44 Additionally, it is important to note that our findings cannot be generalized to encompass the entire diversity of autistic women, given that recruitment in this study focused on adults with minimal support needs. To ensure the applicability of these findings to other autistic women with more pronounced characteristics and traits, it is imperative to undertake further investigations. These subsequent studies should aim to replicate and expand the initial study, identifying both specificities and generalities across the autism spectrum.

This study revealed critical nuances related to the formal identification of autism, shedding light on significant challenges in recognizing the condition in women. The phenomenon of camouflage appears to be more expressive among autistic women compared with men, being associated with feelings of anxiety, stress, and depression—potentially serving as a pivotal factor in autistic burnout—alongside a reduced sense of social belonging. However, contrary to prior research, camouflage does not seem to be the primary culprit behind diagnostic difficulty. Instead, the foremost obstacle to early and accurate diagnosis among Brazilian women seems to stem from health care professionals’ tardiness in recognizing autistic characteristics, likely influenced by traditional autism stereotypes linked to behaviors more common in boys than girls. This warrants thorough exploration in future studies. Indeed, our findings underscore the imperative to scrutinize potential gender biases in counseling and clinical practice, which perpetuate the underreporting of women with autism. Such biases also resonate in the scarcity of comprehensive psychological tools for autism assessment that can effectively discern the subtler traits prevalent among females. Thus, we hope that these results will contribute to future investigations into the identification of autism in adult women.

Footnotes

Acknowledgments

The authors extend their heartfelt gratitude to all their participants for their generosity in sharing their time and experiences. Additionally, they express their sincere appreciation to Dr. Vanessa Andina and Fernanda Tavares for their invaluable insights and feedback on an earlier draft of this article.

AUTHORSHIP CONFIRMATION STATEMENT

The authors of this article confirm that all listed authors have contributed significantly to the conception, design, acquisition of data, analysis, and interpretation of data, drafting the article or revising it critically for important intellectual content, and final approval of the version to be published. The article has been submitted solely to Autism in Adulthood.

Ethics Approval

These studies were performed in accordance with the principles of the Declaration of Helsinki. This research was approved by the Ethics Committee of the Federal University of Rio Grande (approval no. 69182923.4.0000.5324).

Open Practices

Data, scripts, and materials used in these studies can be obtained from the first author upon request.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received in relation to this article.

References

Baron-Cohen

. Leo Kanner, Hans Asperger, and the discovery of autism. Lancet. 2015; 386(10001):1329–1330.

Fernandes

, Tomazelli

, Girianelli

. Diagnóstico de autismo no século XXI: Evolução dos domínios nas categorizações nosológicas. Psicol USP. 2020; 31.

American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders. 5th ed. Text rev. American Psychiatric Association; 2022. doi:10.1176/appi.books.9780890425787

World Health Organization. International statistical classification of diseases and related health problems. 11th ed. 2021. Available from: https://icd.who.int

Lundström

, Taylor

, Larsson

, et al. Perceived child impairment and the ‘autism epidemic’. J Child Psychol Psychiatry. 2022; 63(5):591–598; doi: 10.1111/jcpp.13497

Keller

, Chieregato

, Bari

, et al. Autism in adulthood: Clinical and demographic characteristics of a cohort of five hundred persons with autism analyzed by a novel multistep network model. Brain Sci. 2020; 10(7):416; doi: 10.3390/brainsci10070416

Bargiela

, Steward

, Mandy

. The experiences of late-diagnosed women with autism spectrum conditions: An investigation of the female autism phenotype. J Autism Dev Disord. 2016; 46(10):3281–3294.

Rynkiewicz

, Janas-Kozik

, Słopień

. Girls and women with autism. Psychiatr Pol. 2019; 53(4):737–752.

Almeida

, Neves

. A popularização diagnóstica do autismo: Uma falsa epidemia? Psicol cienc prof. 2020; 40:e180896; doi: 10.1590/1982-3703003180896

10.

Center for Disease Control and Prevention. Prevalence of autism spectrum disorders among children aged 8 years: Autism and developmental disabilities monitoring network, 11 sites, United States, 2010. Surveill Summ. 2014; 63(2):1–21.

11.

Maenner

, Warren

, Williams

, et al. Prevalence and characteristics of autism spectrum disorder among children aged 8 years—autism and developmental disabilities monitoring network, 11 Sites, United States, 2020. MMWR Surveill Summ. 2023; 72(2):1–14; doi: 10.15585/mmwr.ss7202a1

12.

Pearlman-Avnion

, Shlain

. Adult women on the high-functioning autism spectrum. In: Autism in Adulthood. ( Lowinger

., eds.) Springer Nature; 2019;pp. 89–112. doi:10.1007/978-3-030-28833-4

13.

Murphy

, Flower

, Jellett

. Women seeking an autism diagnosis in Australia: A qualitative exploration of factors that help and hinder. Autism. 2023; 27(3):808–821; doi: 10.1177/13623613221117911

14.

Hull

, Petrides

, Mandy

. The female autism phenotype and camouflaging: A narrative review. Rev J Autism Dev Disord. 2020; 7(4):306–317; doi: 10.1007/s40489-020-00197-9

15.

Hull

, Mandy

, Lai

, et al. Development and validation of the Camouflaging Autistic Traits Questionnaire (CAT-Q). J Autism Dev Disord. 2019; 49(3):819–833; doi: 10.1007/s10803-018-3792-6

16.

Lai

, Szatmari

. Sex and gender impacts on the behavioural presentation and recognition of autism. Curr Opin Psychiatry. 2020; 33(2):117–123.

17.

Fusar-Poli

, Brondino

, Politi

, Aguglia

. Missed diagnoses and misdiagnoses of adults with autism spectrum disorder. Eur Arch Psychiatry Clin Neurosci. 2022; 272(2):187–198.

18.

Duvekot

, van der Ende

, Verhulst

, et al. Factors influencing the probability of a diagnosis of autism spectrum disorder in girls versus boys. Autism. 2017; 21(6):646–658.

19.

Green

, Travers

, Howe

, et al. Women and autism spectrum disorder: Diagnosis and implications for treatment of adolescents and adults. Curr Psychiatry Rep. 2019; 21(4):22; doi: 10.1007/s11920-019-1006-3

20.

Cassidy

, Au-Yeung

, Robertson

, et al. Autism and autistic traits in those who died by suicide in England. Br J Psychiatry. 2022; 221(5):1–9; doi: 10.1192/bjp.2022.21

21.

Baron-Cohen

, Wheelwright

, Skinner

, et al. The Autism-Spectrum Quotient (AQ): Evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians. J Autism Dev Disord. 2001; 31(1):5–17; doi: 10.1023/A:100565341147

22.

Leary

. Need to belong scale. J Pers Assess. 2013.

23.

Lovibond

, Lovibond

. The structure of negative emotional states: Comparison of the Depression Anxiety Stress Scales (DASS) with the Beck depression and anxiety inventories. Behav Res Ther. 1995; 33(3):335–343.

24.

Do Egito

JHT

, Ferreira

GMR

, Gonçalves

, et al. Brief report: Factor analysis of the Brazilian version of the adult autism spectrum quotient. J Autism Dev Disord. 2018; 48(5):1847–1853.

25.

Alves

ALC

, Paula

, Miranda

, et al. The autism spectrum quotient in a sample of Brazilian adults: Analyses of normative data and performance. Dement Neuropsychol. 2022; 16(2):244–248.

26.

Gastal

, Pilati

. The need to belong scale: Adaptation and evidences of validity. Psico-USF. 2016; 21(2):285–292.

27.

Machado

, Bandeira

. Adaptação e validação da Depression, Anxiety and Stress Scale (DASS-21) para o Português brasileiro. Manuscript submitted for publication.

28.

Patias

, Machado

WDL

, Bandeira

, Dell’Aglio

. Depression anxiety and stress scale (DASS-21)-short form: Adaptação e validação para adolescentes brasileiros. Psico-usf. 2016; 21(3):459–469.

29.

Lai

, Lombardo

, Suckling

, et al. MRC AIMS Consortium. Biological sex affects the neurobiology of autism. Brain. 2013; 136(Pt 9):2799–2815.

30.

Burrows

, Grzadzinski

, Donovan

, et al. A data-driven approach in an unbiased sample reveals equivalent sex ratio of autism spectrum disorder–associated impairment in early childhood. Biol Psychiatry. 2022; 92(8):654–662.

31.

McCrossin

. Finding the proportion of females with autistic spectrum disorder who develop anorexia nervosa, the true prevalence of female ASD and its clinical significance. Children. 2023; 10(2):272.

32.

Milner

, Colvert

, Hull

, et al. Does camouflaging predict age at autism diagnosis? A comparison of autistic men and women. Autism Res. 2024; 17(3):626–636; doi: 10.1002/aur.3059

33.

Backes

, Mônego

, Bosa

, et al. Propriedades psicométricas de instrumentos de avaliação do transtorno do espectro do autismo: Uma revisão sistemática de estudos brasileiros. J Bras Psiquiatr. 2014; 63(2):154–164.

34.

Brickhill

, Atherton

, Piovesan

, et al. Autism, thy name is man: Exploring implicit and explicit gender bias in autism perceptions. PLoS ONE. 2023; 18(8):e0284013; doi: 10.1371/journal.pone.0284013

35.

De Almeida

RMM

, Cabral

JCC

, Narvaes

. Behavioural, hormonal and neurobiological mechanisms of aggressive behaviour in human and nonhuman primates. Physiol Behav. 2015; 143:121–135; doi: 10.1016/j.physbeh.2015.02.053

36.

Potard

, Pochon

, Henry

, et al. Relationships between school bullying and frustration intolerance beliefs in adolescence: A gender-specific analysis. J Rat-Emo Cognitive-Behav Ther. 2022; 40(1):103–123; doi: 10.1007/s10942-021-00402-6

37.

Lai

, Lombardo

, Pasco

, et al. MRC AIMS Consortium. A behavioral comparison of male and female adults with high functioning autism spectrum conditions. PLoS ONE. 2011; 6(6):e20835.

38.

Gould

. Towards understanding the under-recognition of girls and women on the autism spectrum. Autism. 2017; 21(6):703–705.

39.

Leedham

, Thompson

, Smith

, et al. ‘I was exhausted trying to figure it out’: The experiences of females receiving an autism diagnosis in middle to late adulthood. Autism. 2020; 24(1):135–146; doi: 10.1177/1362361319853442

40.

Zener

. Journey to diagnosis for women with autism. AIA. 2019; 5(1):2–13; doi: 10.1108/AIA-10-2018-0041

41.

Doey

, Coplan

, Kingsbury

. Bashful boys and coy girls: A review of gender differences in childhood shyness. Sex Roles. 2014; 70(7–8):255–266; doi: 10.1007/s11199-013-0317-9

42.

Heilman

. Gender stereotypes and workplace bias. Res Organ Behav. 2012; 32:113–135; doi: 10.1016/j.riob.2012.11.003

43.

Walsh

, Krabbendam

, Dewinter

, et al. Gender identity differences in autistic adults: Associations with perceptual and socio-cognitive profiles. J Autism Dev Disord. 2018; 48(12):4070–4078; doi: 10.1007/s10803-018-3702-y

44.

George

, Stokes

. Gender identity and sexual orientation in autism spectrum disorder. Autism. 2018; 22(8):970–982; doi: 10.1177/1362361317714587