Are Autistic Females at Greater Risk of Suicide? A Call for Clarity to Advance Suicide Prevention for the Whole Community

Abstract

Understanding suicide risk is critical for supporting prevention. A growing body of evidence shows autistic people are at greater risk for multiple suicide-related outcomes than non-autistic people. This commentary is in response to an observed pattern of miscommunication in scientific and community spaces about autistic females having higher risk of suicide. However, it is not always clear who they are being compared with in these statements. To address this confusion, we summarize the current population-based evidence on autistic suicide risk, highlighting findings related to sex similarities and differences, which actually indicates comparable rates of suicide death among autistic males and females, and mixed findings related to sex differences in risk of other suicidal behaviors. We call for greater clarity in suicide risk communication moving forward focused on outcomes, measurement, sampling methods, and comparison groups to reflect accurate conclusions about existing evidence. Further research is needed about the full range of suicide-related outcomes for autistic people, including a greater understanding of sex differences as well as potential gender differences to include transgender and nonbinary autistic people. However, studies of sex and gender differences should not overshadow the compelling need for efforts to understand and address the elevated risk of suicidal thoughts, behaviors, and death among autistic people across sex and gender boundaries.

Lay Summary:

It is important to understand and address the elevated risk of suicide among the autistic population. Some studies have looked at sex differences and the most consistent finding is that autistic males and females have similar risk of dying by suicide. To give accurate information about research findings, communication should always be clear about the suicide outcomes, measurement, sampling methods, and comparison groups that were studied.

Suicide is a critical concern for people on the autism spectrum. In fact, according to one study, suicide is one of the leading causes of premature death among autistic people (Hirvikoski et al., 2016). Further, there is evidence that suicidality is a common concern identified among autistic individuals in clinical settings (Jachyra et al., 2022). While this topic received little research attention until the last decade, recent findings on community priorities (Benevides et al., 2020) and several studies demonstrating elevated risk of suicide death and other suicidal behaviors (e.g., Cassidy et al., 2014; Hirvikoski et al., 2016; Kolves et al., 2021) have emphasized the need for focused clinical and research attention on suicide risk factors and suicide prevention for autistic people.

Sex and gender were also long-overlooked in autism research. Autism has historically been written about as a predominantly male condition, with consistently higher reported prevalence of autism among boys (Maenner et al., 2021). However, understanding has begun to shift over the last decade and there is growing interest in and attention on understanding autism and autistic experiences among women and girls (e.g., Lockwood Estrin et al., 2021; Ratto et al., 2018). Available evidence suggests some important sex differences (Bölte et al., 2023), such as in premature mortality (Lunsky et al., 2022). Further, beyond viewing gender as synonymous with birth sex and as binary (Torgrimson & Minson, 2005), there has also been increased awareness of and attention to the needs and experiences of autistic transgender and nonbinary people (Strang et al., 2020; Warrier et al., 2020).

The intersection of these two topics—autistic suicide risk and sex/gender—has become a focus of much interest and discourse as a result of some intriguing published study findings. However, we have noticed repeated patterns of misunderstanding and miscommunication about the existing evidence. In particular, we see misinterpretations about risk of suicide among autistic females when reviewing journal manuscript submissions, attending conferences, and communicating with various types of stakeholders. The purpose of this commentary is to address these common misunderstandings, highlight what current evidence shows about sex differences in autistic suicide risk, and identify areas in need of future work to advance the science of autism and suicide and ultimately reduce suicide risk in the autistic community. Leading into this discussion, we offer questions that can be used to guide readers when interpreting studies on suicide risk to avoid miscommunications about study findings; these may be useful to journal editors, reviewers, researchers, clinicians, journalists, and other community members. Enhanced clarity and accuracy in reporting and understanding about suicide risk are needed to help advance the field toward suicide prevention for the whole community.

Interpreting Studies on Suicide Risk in Autistic Communities

A growing body of evidence has examined questions related to suicide risk and autism in a variety of ways. When interpreting the literature on suicide risk and autism—as with all studies—it is important to first clearly understand the specific details of the study designs in terms of what was measured and with whom. A wide variety of studies examine autistic suicide risk in different ways. Two recent meta-analyses of autism studies of suicidal behavior found that reported prevalence rates range widely (Huntjens et al., 2023; Newell et al., 2023). Moreover, subgroup analyses indicated that the large heterogeneity in prevalence rates of suicidal thoughts and of suicidal behaviors vary along with the range of sampling approaches and measures used (Huntjens et al., 2023). Therefore, we propose that the following four questions are critical to consider when interpreting evidence to draw accurate and comprehensive conclusions about autistic suicide risk. Consideration of the following aspects of study design would help avoid the miscommunications and misinterpretations we have witnessed regarding sex/gender differences in suicide risk in autistic people:

1. How is suicide risk defined? Suicide risk is a general phrase that is often used to refer to the risk of a wide range of suicidal behaviors. Existing evidence about autistic suicide risk includes studies using a variety of metrics including death by suicide (e.g., Hirvikoski et al., 2016; Kirby et al., 2019; Lai et al., 2023), suicide attempts (i.e., actions taken with the intent of dying; e.g., Hirvikoski et al., 2020; Kolves et al., 2021), and suicidal thoughts or feelings (e.g., Cassidy et al., 2014). Not all suicide metrics have the same prevalence or associations, so it is imperative that the studied outcome is defined.

2. How is suicide risk measured? The methods of assessment (e.g., diagnostic records reviews, clinical interviews, self-report questionnaires) and the reporter (e.g., clinician, self, proxy/caregiver) must be taken into consideration. Some emerging research suggests differences in whether autistic individuals endorse suicidal thoughts and behaviors based on the method of assessment (i.e., self-report versus clinician interview; Schwartzman et al., 2023). Studies in the general population also demonstrate inconsistencies in rates of endorsement across reporters (e.g., Deming et al., 2021; Lewis et al., 2014). Therefore, clear communication about measurement is critical as these factors can influence conclusions regarding prevalence and sex/gender differences.

3. Who is the focus of study and how was the sample recruited? Population-based studies involving secondary data analyses have primarily used ICD codes to identify an autistic sample (e.g., Kolves et al., 2021). Other studies have focused on clinical samples (drawn from one or multiple clinical sites such as outpatient therapy centers, inpatient facilities, or emergency rooms; e.g., Cassidy et al., 2014; Jachyra et al., 2022), community samples (volunteers from a defined region who meet set criteria; e.g., Schwartzman et al., 2023) or surveys of autistic people (often widely available to volunteers recruited online; e.g., Pelton et al., 2020). Another group of studies has looked more broadly at people with autistic traits (e.g., Cassidy et al., 2020); in these cases, it is critical to be clear that these are not necessarily autistic people, and thus these studies should not be interpreted to be reflective of autistic risk but of risk associated with autistic traits (Sasson & Bottema-Beutel, 2021). Descriptions of risk should be summarized with reference to the studied population.

4. What groups are being compared? Several studies have compared autistic and non-autistic populations, and have compared autistic males to non-autistic males and autistic females to non-autistic females (e.g., Hirvikoski et al., 2016; Kirby et al., 2019). However, misinterpretations have often been made assuming that the comparisons of autistic and non-autistic females showing autistic females at greater risk imply differences between autistic females and males. It is critical to specify which groups are being compared when a result is interpreted to avoid misconstruing what “higher risk” actually means.

Understanding the Influence of Sex on Autistic Suicide Risk

The most common confusion we see regarding sex-related results in studies on autistic suicide risk is about elevated risk for autistic females. There is often discussion about autistic females being at “higher risk.” However, as indicated above, it is important to understand: higher risk compared to whom?

To help clarify the existing evidence, we summarized key results from existing population-derived studies on autistic suicide risk that looked at sex differences. We focus on summarizing findings from population-based studies as opposed to clinical and convenience samples, because they can risk selection bias, as noted above (Rubenstein & Furnier, 2021). No known studies have used population-based methods to examine gender differences, which is likely reflective of limited population-level data collection of gender vs. sex (Goodman et al., 2019). In Table 1, we present comparisons published in the papers, as well as comparisons that were not specifically analyzed within the respective papers but that we were able to make using information provided in the papers. For example, neither of Hirvikoski et al.’s papers (2016, 2020) specifically compared autistic males to autistic females. However, the authors reported adequate raw data to generate confidence intervals and determine if there were significant differences between males and females. When determining if there were significant differences in suicide risk, we used 95% confidence intervals (either provided in the published work or calculated). To examine the significance of odds ratios and risk ratios, we examined if the confidence intervals spanned 1.0. When comparing proportions (percentages) between sexes, we examined if the confidence intervals overlapped with one another.

Table 1.

Summary of Population-Based Findings Related to Autistic Suicide Risk and Sex

Study	Suicide Metrics	Sample Identification	Comparators	Key Sex-Related Findings & Conclusions
Croen et al. (2015)	Suicide attempt	ICD autism codes from a healthcare insurance company population in one state (U.S.) ≥ 18 years of age	Non-autistic population	• Autistic females & males at higher risk of attempt than non-autistic comparators, respectively:  ○ Female OR (95% CI): 6.68 (2.35, 19.02)  ○ Male OR (95% CI): 4.32 (1.93, 9.68) • Autistic females & males have similar risk:  ○ Female percent: 2.72% (1.36, 4.81%)*  ○ Male percent: 1.45% (0.83, 2.35%)*
Hirvikoski et al. (2016)	Suicide death	ICD autism codes from a national population (Sweden) Any age	Non-autistic population	• Autistic females & males at higher risk of death than nonautistic comparators, respectively:  ○ Female OR (95% CI): 13.05 (8.73, 19.50)  ○ Male OR (95% CI): 6.28 (4.79, 8.23) • Autistic females & males have similar risk:  ○ Female percent (95% CI): 0.32% (0.21, 0.47%)*  ○ Male percent (95% CI): 0.30% (0.23, 0.39%)*
Chen et al. (2017)	Suicide attempt	ICD autism codes from a national population (Taiwan) 12–29 years of age	Non-autistic population	• Autistic females & males at higher risk of attempt than non-autistic comparators, respectively:  ○ Female HR (95% CI): 4.80 (2.86, 8.06)  ○ Male HR (95% CI): 6.55 (4.78, 8.98) • Direct comparison of autistic males & females not possible
Kirby et al. (2019)	Suicide death	ICD autism codes from a state population (U.S.) ≥ 5 years of age	Non-autistic population	• Autistic females at higher risk of death than non-autistic females; no difference between autistic & non-autistic males:  ○ 2013–2017 Female RR (95% CI): 3.42 (1.63, 7.20)  ○ 2013–2017 Male RR (95% CI): 1.01 (0.66, 1.55) • Autistic females & males have similar risk:  ○ 2013–2017 Female (95% CI): 0.17% (0.08, 0.38%)  ○ 2013–2017 Male (95% CI): 0.16% (0.10, 0.25%)
Hand et al. (2020)	Suicidal ideation Suicide attempt	ICD codes from Medicare claims data (U.S.) 18–59 years of age	None	• Autistic females & males at similar risk of ideation:  ○ Female OR (95% CI): 1.14 (0.96, 1.36) • Autistic females & males at similar risk of attempt:  ○ Female OR (95% CI): 1.12 (0.96, 1.31)
Hirvikoski et al. (2020)	Suicide attempt Suicide death	ICD autism codes from a national population (Sweden) Any age	Non-autistic population	• Autistic females & males without ID or ADHD at higher risk of attempt than non-autistic comparators:  ○ Female OR (95% CI): 6.27 (5.72, 6.88)†  ○ Male OR (95% CI): 3.06 (2.82, 3.32)† • Autistic females without ID or ADHD at higher risk of attempt than males:  ○ Female percent (95% CI): 13.98% (13.22, 14.76%)*  ○ Male percent (95% CI): 5.80% (5.45, 6.16%)* • Autistic females & males without ID or ADHD at higher risk of death than non-autistic comparators:  ○ Female OR (95% CI): 12.05 (6.85, 21.21)†  ○ Male OR (95% CI): 7.19 (5.31, 9.73)† • Autistic females & males without ID or ADHD have similar risk of death:  ○ Female percent (95% CI): 0.52% (0.37, 0.71%)*  ○ Male percent (95% CI): 0.62% (0.50, 0.75%)*
Kõlves et al. (2021)	Suicide attempt	ICD autism codes from a national population (Denmark) ≥ 10 years of age	Non-autistic population	• Autistic females & males at higher risk of attempt than non-autistic males:  ○ Female adjusted IRR (95% CI): 8.51 (7.63, 9.49)  ○ Male adjusted IRR (95% CI): 1.93 (1.71, 2.18) • Autistic females at higher risk of attempt than males:  ○ Female adjusted IRR (95% CI): 4.41 (3.74, 5.19) • Autistic females & males at higher risk of death than non-autistic males:  ○ Female adjusted IRR (95% CI): 2.63 (1.46, 4.76)  ○ Male adjusted IRR (95% CI): 3.48 (2.57, 4.74) • Autistic females & males at similar risk of death:  ○ Female adjusted IRR (95% CI): 0.75 (0.39, 1.46)
Martini et al. (2022)	Self harm (included suicide-related codes, but not exclusively)	ICD autism codes from a national population (Sweden) ≥ 16 years of age	Non-autistic population	• Autistic females at higher cumulative risk of self-harm than autistic males, non-autistic females, and non-autistic males:  ○ Autistic female incidence (95% CI): 0.19 (0.18, 0.20)  ○ Autistic male incidence (95% CI): 0.08 (0.07, 0.09)  ○ Non-autistic female incidence (95% CI): 0.03 (0.03, 0.04)  ○ Non-autistic male incidence (95% CI): 0.03 (0.03, 0.03)
Lai et al. (2023)	Self harm (included suicide-related codes, but not exclusively) Suicide death	ICD autism codes from a province population (Canada) Any age	Non-autistic population	• Autistic females & males at higher risk of self-harm than non-autistic comparators:  ○ Autistic female percent (95% CI): 8.01% (7.64, 8.40%)*  ○ Non-autistic female percent (95% CI): 2.22% (2.12, 2.33%)*  ○ Autistic male percent (95% CI): 3.71% (3.55, 3.87%)*  ○ Non-autistic male percent (95% CI): 1.17% (1.13, 1.22%)* • Autistic females at higher risk of self-harm than males:  ○ Autistic female percent: 8.01% (95% CI: 7.64, 8.40%)*  ○ Autistic male percent: 3.71% (95% CI: 3.55, 3.87%)* • Autistic females & males at similar risk of death than non-autistic comparators:  ○ Autistic female percent: 0.09% (95% CI: 0.06, 0.15%)*  ○ Non-autistic female percent (95% CI): 0.05% (0.03, 0.06)*  ○ Autistic male percent: 0.12% (95% CI: 0.09, 0.15%)*  ○ Non-autistic male percent (95% CI): 0.08% (0.07, 0.10%)* • Autistic females & males at similar risk of death:  ○ Autistic female percent: 0.09% (95% CI: 0.06, 0.15%)*  ○ Autistic male percent: 0.12% (95% CI: 0.09, 0.15%)*
Tsai et al. (2023)	Suicide death	ICD codes from a national population (Taiwan) Any age	Non-autistic population	• Autistic males at higher risk of death than non-autistic males:  ○ Male HR (95% CI): 3.81 (2.37, 6.13) • Autistic and non-autistic females at similar risk of death:  ○ Female HR (95% CI): 3.19 (0.96, 10.63) • Direct comparison of autistic males & females not possible

Notes. ICD, international classification of disability. ID, intellectual disability. OR, odds ratio. HR, hazard ratio. RR, relative risk. CI, confidence interval. ADHD, attention-deficit hyperactivity disorder. IRR, incidence rate ratio.

^*indicates that the confidence intervals were not presented in the published paper but we calculated them using the published study results.

^†indicates that the reported ORs are the reported crude rates; adjusted rates were also presented in the paper (see Hirvikoski et al., 2020) and also show significantly elevated odds relative to non-autistic comparators.

The study summaries presented in Table 1 demonstrate that there is a fair amount of consistency across population-derived studies looking at sex differences in autistic suicide risk. The five studies that have looked at differences in risk of suicide death (Hirvikoski et al., 2016, 2020; Kirby et al., 2019; Lai et al., 2023; Tsai et al., 2023) have consistently shown that autistic people (analyzed with sex groups combined) are at greater risk than their non-autistic counterparts. However, these studies also consistently show that within the autistic population, autistic males and females have similar risk of dying by suicide (i.e., no significant differences between autistic male and female risk, demonstrated by overlapping confidence intervals). Two studies that looked at sex differences in risk of suicide attempt also found similar risk between autistic males and autistic females (Croen et al., 2015; Hand et al., 2020). However, two large and more recent studies identified differences in suicide attempt risk with autistic females at greater risk than autistic males (Hirvikoski et al., 2020; Kolves et al., 2021). Martini et al. (2022) and Lai et al. (2023) also both found higher risk for autistic females than males when examining self-harm codes. Hand et al. (2020) found no differences between autistic males and females when looking at suicidal ideation. Across all the presented studies, we see evidence of elevated risk of multiple suicide outcomes when compared with non-autistic comparators. Each study looked at a unique age range, with some looking across all ages and others focusing on adults or another specific range. Although not all studies presented the age range of suicide-related outcomes, the available evidence indicates the importance of considering risk across the lifespan.

Taken together, these studies suggest that autistic populations (with a documented ICD code) are at significantly greater risk for suicide death, suicide attempt, and self-harm than non-autistic comparison populations of the same sex. Although rates of suicide death do not differ between autistic males and females, there are mixed findings related to sex-specific risk of suicide attempt, with two studies from Scandinavia demonstrating elevated risk of suicide attempt among autistic females compared with autistic males and two studies from the U.S. showing similarities in attempt risk.

Future Directions on Autistic Suicide Risk and Sex/Gender

The most consistent evidence with regard to suicide risk among the autistic population is elevated risk for multiple suicide-related outcomes (i.e., suicide death and suicide attempt) in comparison to the non-autistic population. Thus, there is a clear need to evaluate reasons for this elevated risk and to improve community and clinical efforts to mitigate risk of suicide attempt and death among autistic people.

Research studies to understand differences between autistic and non-autistic people are critical. Existing studies point to several potential reasons for group differences in suicidal thoughts and/or actions such as co-conditions (e.g., depression, attention deficit hyperactivity disorder), dissatisfaction with living arrangements, masking, loneliness, lack of meaningful social connections, perceiving oneself to be a burden on others, and having unmet support needs (Cassidy et al. 2014, 2018; Hand et al., 2019; Hedley et al. 2018a, 2018b; Pelton et al. 2020). These are all areas of potential action to improve services and supports for autistic people.

Continuing to expand clinical supports is dependent on rigorous research and accurate interpretations of research findings., Special care must be taken in selecting and describing sampling methods, including describing the limitations on generalizing findings from non-representative samples. While studies using population-based records based on ICD codes can miss those who have been mis-diagnosed, not yet diagnosed, and/or who have limited access to the healthcare system, online convenience samples (from registries, prior participants, and/or advertised in autistic online spaces) risk over-generalization from a biased sample (Rubenstein & Furnier, 2021). Specifically, many studies on autistic suicide risk have relied upon online convenience samples (e.g., Cassidy et al., 2018; Conner et al., 2020; Moseley et al., 2023). Although these are a convenient means to acquire large samples and eliminate access barriers (and we have ourselves utilized online samples), it is crucial to note that these studies have often recruited participants that are majority female, high education level, high socioeconomic status, and late diagnosed (Newell et al., 2023). While including females—who are traditionally under-represented in autism research—is positive, it is important to recognize that such findings are not representative of the whole autistic community and thus should not be used to draw conclusions about the prevalence of suicidality overall or by sex or gender. Prospective data collection efforts can address some of these limitations by using focused recruitment strategies to achieve population-representative samples and promote inclusion of individuals who are currently underrepresented in research.

Understanding sex-related differences could contribute to increased understanding of risk and protective factors that are important for clinical work. For example, sex differences in different suicide outcomes could potentially reflect biological differences and thus different avenues for treatments such as pharmaceutical interventions (Bölte et al., 2023). Alternatively, or additionally, they may reflect social differences, such as differences in how individuals were reared or treated, different societal expectations, or different clinical interpretations of behaviors. Thus, there is potential to use explorations of sex differences for theory generation that could produce new studies to further examine risk and protective factors. That two studies from Scandinavia identified differences between autistic males and females in suicide attempt rates, while two studies in the U.S. did not identify differences, could potentially suggest systemic differences in diagnosis and documentation of suicide attempts or other healthcare provider practices. There could also be other social factors differing between these two regions of the world. Thus, research that explores cross-cultural differences could also offer additional insights to help understand risk of suicide attempt.

Future studies exploring gender differences may also be informative. Research shows that transgender and gender nonconforming people are at higher risk of experiencing suicidal thoughts and behaviors (Herman et al., 2019). For future research, it is important to collect detailed information about sex and gender identity, as there is also evidence of an overrepresentation of transgender and gender nonconforming identities among autistic people (Warrier et al., 2020). Therefore, it will be important for research to explore if autistic transgender and gender nonconforming individuals are uniquely at risk for various suicide metrics than people who share just one of those identities (e.g., Strauss et al., 2021). These types of studies may further elucidate potential societal influences that contribute to suicide-related outcomes, and inform suicide prevention strategies.

Based on the current population-based evidence, the most pressing clinical consideration is to recognize risk for multiple suicide-related outcomes (e.g., suicide death, suicide attempts) across sexes and to provide individualized care for support and prevention. Ultimately, it will be necessary to identify if the same types of treatments and approaches to suicide prevention and mental health support are effective for autistic people generally, or if sex and/or gender differences warrant unique approaches. Currently, autistic people have extremely limited access to mental health services and have high needs for support (Adams & Young, 2021). Therefore, it stands to reason that the most critical needs are for autism-tailored services and mental health providers with specific skills to work with autistic clients of multiple sexes and gender identities. Finally, although we highlighted sex and gender in this commentary due to common misconceptions in the field, it is important to recognize a need for future research to consider suicide risk among autistic individuals with additional forms of intersectionality such as marginalized racial and ethnic identities which have received hardly any research attention to date.

Conclusions

Research findings on autistic suicide risk have drawn a great deal of public attention. However, we have noticed patterns of mischaracterization of those findings across multiple settings. While population-derived studies show equivalent rates of suicide death among autistic males and females, and mixed findings related to risk of other suicidal behaviors, we have often seen emphasis placed on findings related to female risk. These may be easy errors to make because the available findings that show sex differences are intriguing and suggest divergence from general population evidence. Further, for a field that so long focused almost exclusively on males, studies showing high needs among females are important as they highlight significant gaps in our prior understanding. However, while it is true that autistic females have been understudied and need much more research focus, it is critical that we as a field maximize clarity of research findings. Moving forward, we hope that all people citing studies on autistic suicide risks can clearly indicate the suicide outcomes, measurement, sampling methods, and comparison groups to draw accurate conclusions based on the research. Researchers and scientific journalists must take the lead to avoid inaccurate information being disseminated to community members and nonscientific audiences. Furthermore, there is a need for more research to understand underlying risk and to broaden this area of research to explore potential gender differences as well as sex differences. Conscious efforts for accuracy and expansion will help advance scientific and community understanding of risk, and aid in working toward suicide prevention for the whole autistic community.

Data Availability

There is no data associated with this manuscript.