Health profiles of adults with ASD: Differences between women and men

Abstract

The purpose of the present study was to investigate the hypothesis that women with autism have poorer health compared to men with autism, and compared to women without autism. Utilizing electronic health records drawn from a single health care system serving over 2 million individuals, 2,119 adults with diagnosed autism spectrum disorders were compared to age- and sex-matched controls. When considering health care utilization, we found evidence of multiplicative risk for conditions within some domains (i.e., nutrition conditions, neurologic disease, psychiatric conditions, sleep disorders) such that women with ASD experienced double jeopardy – meaning they had greater rates of health care utilization within a domain than what would separately be expected by virtue of being a woman and having ASD. For other domains (i.e., endocrine disorders, gastrointestinal disorders), the risk was additive such that being a female and having ASD were both associated with higher health care utilization, but there were no significant interaction effects. It was only with respect to one domain (cardiovascular) that rates of health care utilization were reflective of neither ASD diagnosis nor sex. Overall, our findings suggest that women with ASD are a vulnerable subgroup with high levels of health care utilization.

Lay Summary:

This study asked whether women with autism have poorer health compared to men with autism, and compared to women without autism. To answer this question, we used data from electronic health records. We found that women with ASD were at greatest risk for health problems such as nutrition conditions, neurologic disease, psychiatric conditions, and sleep disorders. More research on health of women with ASD is needed.

Despite the fact that autism spectrum disorder (ASD) is a lifelong condition, much more research is needed about the implications of having an ASD for health and mental health in adulthood. The limited available knowledge suggests there are elevated health risks (Croen et al., 2015), but these patterns may be more characteristic of men than women because ASD is 2–5 times more common in males than in females (Lai et al., 2014). Further, as females with ASD tend to be diagnosed at later ages (often not until adulthood) than males, the research and clinical literature is weighted by observations of males (Lai et al., 2015). There is thus a need for focused research on females with ASD not only during childhood but across the life course.

The purpose of the present study is to advance our understanding of the potentially unique health profiles of women with ASD. The analysis is based on electronic health records drawn from a single health care system serving over 2 million individuals, including 2,119 adults with diagnosed ASDs. Comparing those with ASD with matched population controls, and furthermore comparing males and females with and without a diagnosis of ASD, the present study investigates whether women with ASD have poorer health compared to men with ASD, and compared to women without ASD.

Women in the general population have higher rates of health problems and health care utilization than men (Avdic et al., 2019; Owens, 2008), particularly for mental health domains (Altemus et al., 2014; Piccinelli et al 2000). Past research also has indicated that individuals with ASD have higher rates of a range of physical and mental health problems and greater utilization of healthcare services compared to individuals without ASD (Croen et al., 2015; Kohane et al., 2012; Vohra et al., 2016; Weiss et al., 2017). However, much of the literature on health outcomes for individuals with ASD have not included an examination of sex differences, leaving question regarding the extent to which sex differences in health in the general population are mirrored among individuals with autism.

Nevertheless, there is some suggestion that females with ASD may be at particularly high levels of health risk. For example, in a survey of service needs of adolescents and adults with ASD, Tint and colleagues (2017) found that females with ASD were more likely than males with ASD to receive psychiatric services and emergency department services, but there were no differences in other health services examined (e.g., family doctor, neurology, behavior therapy, physiotherapy); however, there was no typically-developing control group in this study. Similarly, a 25-year follow-up study of a population cohort of adults with ASD in Utah found that females had a higher number of medical conditions than males; however, this study also did not have a control group and thus could not test to see if the degree of sex differences were comparable to those in the general population (Jones et al., 2015).

Even among studies that include general population control groups, questions remain regarding whether women with ASD experience double jeopardy by virtue of both their ASD diagnosis and being female. In other words, such studies have not examined the degree to which differences observed between men and women with ASD are the same magnitude as sex differences as in the general population. For example, in a secondary analysis of Scotland’s 2011 census data, Rydzewska and colleagues (2018) examined differences in survey responses of health outcomes (including hearing loss, vision loss, intellectual disability, mental health conditions, and physical disability) between individuals with and without autism. Adults with autism (men and women) were more likely than those without autism to report having each of the probed conditions and, in general, women were more likely than men to report having each of the conditions. Within the autism sample, female sex was associated with greater rates of all conditions. Although sex odds ratios were larger in the autism sample than the non-autism sample, differences in odds ratios were not tested statistically.

Similarly, three studies using the large Kaiser Permanente Northern California electronic health records data base used medical records to examine both sex differences and ASD versus control differences, although these studies did not directly examine whether women with autism are at double jeopardy. These three studies compared males and females with ASD to matched controls in the general population, overwhelmingly finding that youth and adults with ASD have higher rates of physical and mental health conditions (Croen et al., 2015; Davignon et al., 2018) and higher costs of health care services (Zerbo et al., 2019) relative to controls. These studies also provide some evidence that, relative to men with ASD and women in the general population, women with ASD are at higher risk for poor health. In the Croen et al. (2015) study, which analyzed health records during a five-year period, women with ASD were diagnosed with physical and mental health conditions more frequently than men with ASD, and the ASD/control odds ratios for conditions tended to be larger for females than males (but they were not compared statistically). Similarly, in the Zerbo et al. (2019) study, females with ASD had higher heath care costs compared to males with ASD. In the Davignon et al. (2018) study of transition-aged youth, those with ASD were more likely to have most health conditions compared to controls from the general population and females were more likely to have most conditions compared to males, particularly for immune conditions, musculoskeletal conditions, neurological conditions, psychiatric conditions, and infections. In combination, these studies suggest that women with ASD appear to be at higher risk for health conditions and have higher health care costs, even beyond what would be expected by virtue of being female and having an ASD. However, no studies to our knowledge have directly tested whether the sex differences in ASD samples significantly differ in magnitude from sex differences in the general population (i.e., a sex x diagnosis interaction effect).

Present Study

The present study aimed to extend current knowledge by examining if sex differences in health for adults with ASD were similar to or greater than sex differences in the general population using data from an independent health care system that has digitized medical records spanning a 40-year period. In contrast to previous work on rates of health conditions, which primarily focused on the presence/absence of a given health domain (termed prevalence in the Croen et al. research), we also examined health care utilization in specific health domains in the present study. For the present study, we defined health care utilization as the number of ICD-9 codes assigned to a patient within a given health domain. We focus on the domain level of analysis (i.e., category of health problems) rather than on individual codes to avoid multiple testing effects from the very large number of possible codes within the EHR. Our key hypothesis was that there would be statistically significant interactions between sex and ASD status, suggesting that the effects of sex and ASD are multiplicative and not additive. Specifically, we hypothesized that women with ASD would have rates of health problems and greater health care utilization above and beyond the additive effects of being a woman and having an ASD.

Method

Data and Participants

We analyzed diagnostic codes contained in the electronic heath records (EHRs) drawn from the Marshfield Clinic, a multi-specialty group practice with more than 700 physicians providing integrated, comprehensive care to over 300,000 people across more than 50 locations in northern, central, and western Wisconsin. The Marshfield Clinic covers approximately 97% of residents, and captures 99% of deaths, 95% of hospital discharges, and 90% of outpatient visits in the region (Greenlee, 2003). Incidence and prevalence rates of clinically detected diseases in Marshfield Clinic EHR data compare well to previously reported data in the medical literature (Greenlee, 2003). The Marshfield Clinic digitized all of their health records back to 1979, yielding 40 years of EHR data. Cumulatively, Marshfield includes EHRs for over 2 million patients, including those who are deceased or no longer living in the area, as well as active patients.

Participants in the present analysis include adults (18 years of age or older) who had been diagnosed with ASD as well as age- and sex-matched community controls. The patients with ASD were identified via ICD-9 codes of autism (299.0), Asperger’s disorder (299.8), or pervasive developmental disorder not otherwise specified (299.9). In order to rule out potential participants who received an ASD diagnosis in error, to be included in this study, patients had to have had at least two diagnoses of ASD on different days recorded in their EHR. There were 2,187 adult patients with ASD in the Marshfield EHR data who met these criteria and were therefore included in this study. More than three-fourths were males (78.54%, n = 1716), mirroring the sex ratio of ASD in the literature (Lai et al., 2014). An advantage of having such a large sample of individuals with ASD is that the EHR data set contains a substantial number of females with ASD (n = 471). Participants in the ASD sample were born between 1929 and 2000 (M = 1986, SD = 14.1), and were between the ages of 18 and 87 (M = 30.6, SD = 13.5) in 2018, when the EHR data were analyzed (for those who had died before 2018, age of death was used).

The comparison group of community controls included 21,870 age- and sex-matched individuals without disabilities who were patients at the Marshfield Clinic, with no codes in their EHR indicating ASD, selected randomly with a ratio of 10 controls for every one patient with ASD. We selected this ratio as it was the approach of Croen et al. (2015). The comparison group members were 78.5 % male (N = 17,160). They were born between 1929 and 2000 (M = 1,986, SD = 14.1) and were between the ages of 18 and 89 (M = 30.6, SD = 14.0) in 2018, when the EHR data were analyzed (age of death was used for those who died before 2018).

Although the preponderance of patients in the Marshfield EHRs were white, nearly half of the full analytic sample (33%) were receiving Medicaid or Medicare, reflecting socioeconomic and age diversity. About the same percentage (43%) had ten years or more of data in their EHRs, reflecting the low geographic mobility of the population served by the Marshfield Clinic. Descriptive information for cases is presented in Table 1. Institutional review board (IRB) approval for this research was obtained by the Marshfield Clinic and the University of Wisconsin-Madison.

Table 1.

Descriptive Summary Statistics of Samples by ASD Status and Sex

Variables	ASD		Control		F- values (ANOVA)
	Female (n=471)	Male (n=1716)	Female (n=4710)	Male (n=17,160)	ASD	Sex	ASD X Sex
Age (in 2018)	33.1 (15.7)	29.9 (12.7)	33.4 (16.4)	29.9 (13.2)	0.21	75.0***	0.10
Birth year	1983 (16.5)	1987 (13.3)	1983 (16.5)	1987 (13.3)	0.0	77.5***	0.0
Being in EHR 10 years or longer ( = 1)	0.756	0.713	0.395	0.415	611.8***	0.78	5.34*
Deceased case(= 1)	0.083	0.043	0.018	.021	113.8***	20.9***	26.6***
Medicaid or Medicare in record	82.17% (387)	74.65% (1281)	34.76% (1637)	26.88% (4612)	1514.47***	39.61***	0.02

Measures and Analysis Plan

We extracted specific conditions (defined by ICD-9 codes) from the EHR and then grouped the conditions into broader health domains. We analyzed the seven domains based on the same clusters of the ICD-9 codes that were previously analyzed in Croen et al. (2015): cardiovascular disease, endocrine disorders, gastrointestinal disorders, neurologic diseases, nutrition conditions, psychiatric conditions, and sleep disorders. The ICD-9 codes used to define each health domain are presented in Supplementary Materials.

Using their entire Marshfield electronic health record, two variables were computed for each patient with respect to each of the seven health domains: the prevalence of disease in each domain and health care utilization within each domain. First, to measure prevalence of disease for each domain, we checked each condition within a domain for each patient, and if the patient had at least two ICD-9 codes within that domain, the patient was given a code of 1 for that domain, but if the patient did not have at least two codes within that domain, the patient was given a code of 0. Second, to measure health care utilization, for individuals who had at least two codes within a domain, the cumulative number of codes in the EHR included in each of the domains was determined. We note that to meet the requirement of the presence of two codes, it could be either that the same specific ICD-9 code appeared more than once in the record or that two different codes within the domain appeared in the record. Finally, we also computed a total health care utilization score by summing the number of codes across all seven health domains.

For prevalence of disease, logistic regressions were conducted for each of the seven health domains, with ASD status, sex, and their interaction entered as predictors. For health care utilization, data were analyzed using two (ASD status) by two (sex) analyses of variance for each of the seven health domains and the total health care utilization score.

Results

Prevalence of Disease

Results of the logistic regression analyses for prevalence of disease, including odds ratios and predicted probabilities, are presented in Table 2.

Table 2.

Predicting Presence of Health Conditions by ASD Status and Sex

	Odds Ratio [95% C.I.]			Predicted Probability
	ASD	Sex (Female)	ASD X Sex	ASD		Control
	eb	eb	eb	Female	Male	Female	Male
Cardiovascular disease	2.89*** [2.55, 3.28]	1.25*** [1.13, 138]	0.94 [0.72, 1.21]	.28	.25	.13	.10
Endocrine disease	4.67*** [3.88, 5.61]	3.53*** [3.06, 4.08]	0.82 [0.61, 1.10]	.25	.10	.08	.02
Gastrointestinal disorders	2.78*** [2.51, 3.08]	1.31*** [1.23, 1.40]	1.06 [0.85, 1.33]	.69	.61	.42	.36
Neurologic diseases	6.24*** [5.61, 6.96]	1.35*** [1.27, 1.46]	1.39* [1.07, 1.81]	.81	.69	.33	.27
Nutrition conditions	7.02*** [6.32, 7.79]	1.40*** [1.29, 1.52]	1.03 [0.82, 1.29]	.66	.57	.21	.16
Psychiatric conditions	13.8*** [12.2, 15.7]	1.05 [0.98, 1.14]	0.98 [0.69, 1.19]	.81	.81	.25	.24
Sleep disorders	4.12*** [3.62, 4.68]	1.18** [1.04, 1.33]	0.75+[0.57, 1.00]	.20	.22	.07	.07

+ p < .10,

^*p < .05,

^**p < .01,

^***p < .001

There was a statistically significant Sex X ASD status interaction for one domain: neurological conditions, whereby patients with ASD had a greater likelihood of having a condition within this domain than community controls, with a different pattern of sex effects within each group. Descriptively, whereas women with ASD were much more likely to have neurological disorders than men with ASD (.81 versus .69), for the community controls the sex difference was less pronounced (.33 versus .27).

For the other six domains, there were significant main effect differences in disease prevalence between those who had ASD and the matched community controls (ps <.001), with those who had ASD more likely than controls to be diagnosed with conditions in the domains of cardiovascular disease, endocrine disease, gastrointestinal disorders, nutrition conditions, psychiatric conditions, and sleep disorders (see Table 2). There were also main effects of sex for five of the domains, with women more likely than men to have been diagnosed with conditions in the domains of cardiovascular disease, endocrine disease, gastrointestinal disorders, nutrition conditions, and psychiatric conditions.

Health Care Utilization

Table 3 presents the means and standard deviations of the number of ICD-9 codes in each domain as well as the total across the seven domains, reflecting the health care utilization. Importantly, only those with at least two codes in a given domain were included in the analysis. There were significant Sex X ASD status interactions for the measure of health care utilization for four of the seven domains: nutrition conditions, neurologic diseases, psychiatric conditions, and sleep disorders. For these four domains, there were greater male-female differences in the number of codes assigned within a domain for the ASD group than for the community control group, with women with ASD having a greater health care utilization than men. Particularly striking was the interaction of Sex X ASD status for psychiatric conditions, where women with ASD had an average of 72 codes for these conditions and men with ASD had an average of 54 codes, in contrast with fewer than 20 codes for both male and female community controls. The significant interaction effect for neurologic diseases was similarly divergent (i.e., 29 codes for women with ASD vs. 20 for men with ASD, as compared with fewer than 10 for both male and female community controls). Thus, for these domains as well as for nutrition conditions and sleep disorders, there was an increased rate of health care utilization for women with ASD above and beyond what would be expected for being a woman and for having ASD. Consistent with this pattern, there was also a significant Sex X ASD interaction for the total health care utilization score.

Table 3.

Number of ICD-9 Codes within Domains by ASD Status and Sex: [n], Mean, (Standard Deviation), [min, max], Median, <IQR (InterQuartile Range)>

Number of Codes	ASD		Control		F-values (ANOVA)
	Female	Male	Female	Male	ASD	Sex	ASD X Sex
Cardiovascular diseases	[123] 24.85 (52.35) [2, 472] 8 <20>	[390] 20.84 (38.77) [2, 413] 8 <18>	[561] 28.23 (67.02) [2, 796] 8 <21>	[1683] 22.51 (50.25) [2, 749] 6 <16>	2.60	0.70	0.08
Endocrine disorders	[102] 14.78 (21.41) [2, 167] 7 <13>	[131] 13.04 (13.50) [2, 73] 8 <15>	[262] 12.16 (16.50) [2, 145] 5 <12>	[246] 11.53 (14.88) [2, 113] 6 <11>	2.55	0.84	0.19
Gastrointestinal disorders	[284] 21.87 (43.59) [2, 443] 9 <15>	[858] 16.29 (50.26) [2, 1099] 7 <12>	[1537] 11.87 (30.37) [2, 934] 5 <9>	[4312] 8.32 (14.72) [2, 341] 4 <6>	80.16***	20.67***	1.02
Nutrition conditions	[249] 16.15 (25.01) [2, 256] 8 <14>	[757] 11.27 (17.24) [2, 243] 6 <10>	[623] 9.04 (15.55) [2, 214] 4 <6>	[1559] 7.00 (10.11) [2, 122] 4 <5>	78.90***	29.10***	4.93**
Neurologic diseases	[330] 33.63 (61.30) [2, 615] 12 <31>	[957] 24.52 (63.93) [2, 1246] 8 <19>	[1045] 13.30 (35.16) [2, 642] 4 <7>	[2771] 10.77 (27.94) [2, 481] 4 <6>	127.74***	14.91***	4.77*
Psychiatric conditions	[359] 76.05 (211.21) [2, 3491] 30 <73>	[1321] 57.36 (89.37) [2, 1097] 28 <60>	[967] 21.80 (36.98) [2, 526] 9 <18>	[3386] 20.58 (38.64) [2, 733] 8 <17>	311.85***	14.91***	11.48***
Sleep disorders	[68] 11.69 (18.34) [2, 129] 5 <10>	[258] 9.28 (13.75) [2, 113] 4 <8>	[206] 6.88 (10.45) [2, 100] 4 <5>	[596] 8.65 (11.54) [2, 95] 4 <8>	7.69**	.11	4.53*
Total codes across domains	[445] 121.88 (226.08) [2, 3500] 61 <122>	[1601] 84.32 (131.48) [2, 2301] 10 <27>	[2403] 33.70 (83.64) [2, 2197] 45 <86>	[7965] 24.82 (52.59) [2, 912] 8 <19>	888.98***	89.91***	33.51***

For gastrointestinal disorders, there were significant main effects of both ASD status and sex, such that individuals with ASD had greater health care utilization than controls, and women had greater health care utilization than men; the interaction term was non-significant. There were no significant main effects or interactions for the domains of cardiovascular disease and endocrine disease (See Table 3).

The distributions of codes by group are graphically displayed in Figure 1. The X-axis shows the number of codes and Y-axis shows the percentage of participants who had a specific number of code, which vary by domain. For example, 87 percent of females with autism had less than 50 diagnostic codes in neurological disease domain and 8 percent had between 50–100 codes. Almost all females in the control group had less than 50 codes related to neurological diseases and only a few had more than 50 codes, meaning that females with autism have a larger number of medical encounters compared to the female controls. Similar patterns were observed in other diagnostic domains.

Figure 1 –

Distribution of Codes for Females and Males.

As a follow-up analysis, we included number of years of EHR as a covariate in all models. The pattern of significant findings remained the same as with the unadjusted models (data available from first author).

Discussion

Consistent with the growing literature on comorbid health conditions of individuals with ASD (Croen et al., 2015; Kohane et al., 2012; Vohra et al., 2016; Weiss et al., 2017), the present study found that adults with ASD experienced a higher prevalence of each of the seven health domains we investigated (i.e., cardiovascular disease, endocrine disease, gastrointestinal disorders, neurologic disorders, nutrition conditions, psychiatric conditions, and sleep disorders) compared to population controls. There was also a greater rate of health care utilization (defined by more ICD-9 codes within a domain) for individuals with ASD than controls for each of these domains except cardiovascular disease.

Importantly, we found that women with ASD were particularly at-risk for these poor health outcomes. Specifically, when considering health care utilization, we found evidence of multiplicative risk for conditions within some domains (i.e., nutrition conditions, neurologic disease, psychiatric conditions, sleep disorders) such that women with ASD experienced double jeopardy – meaning they had greater health care utilization within a domain than what would separately be expected by virtue of being a woman and having ASD. For other domains (i.e., endocrine disorders, gastrointestinal disorders), the risk was additive such that being a female and having ASD were both associated with higher health care utilization, but there were no significant interaction effects. It was only with respect to one domain (cardiovascular) that rate of health care utilization was reflective of neither ASD diagnosis nor sex. Overall, our findings suggest that women with ASD are a subgroup who are vulnerable for poor health.

The current findings raise critical questions regarding why women with ASD would have heightened rate of health care utilization, even compared to individuals who have the same health conditions. It may be that biological differences lead women with ASD to have more chronic or severe forms of conditions relative to men. For example, females who have an ASD diagnosis have been found to have a heavier “genetic load” than males with the diagnosis (Zhang et al., 2020). The greater number of genetic variants necessary for a female to develop ASD might lead to poorer health. There is also some evidence to suggest that females with ASD might need to have more severe autism symptoms than males to receive an ASD diagnosis (Dworzynski et al., 2012). The greater severity in autism manifestation for diagnosed females might also be accounting for the pattern of more chronic or severe health conditions.

Alternatively, differences in EHR profiles may reflect the unique experiences men and women with ASD have with the health care system. Women with ASD tend to experience more challenges than men during the autism diagnostic process, as reflected by later age at first ASD diagnoses for females compared to males (Begeer et al., 2013; Shattuck et al., 2009). Women also are more likely than men to camouflage their autism symptoms (Green et al., 2019; Lai et al., 2017; Parish-Morris et al., 2017). In a similar fashion, it could be that women with ASD may be perceived differently in the health care system, and thus need additional health visits before a correct diagnosis or course of treatment is determined. It also may be that society responds to women with ASD differently than to men with ASD (Bargiela et al., 2016; Kanfiszer et al., 2017), perhaps leading to more chronic stress for women and ultimately more health conditions and greater health care utilization. Or, it could be that women do not respond equivalently to the same treatments as men, thus requiring more medical encounters, possibly due to their historic underrepresentation in medical research. Future work will need to consider not only the specific health problems of women with ASD, but also how interactions with society and societal expectations may influence health outcomes.

The present study also highlights the value of examining health care utilization in addition to prevalence. As an example, although there was not a statistically significant interaction for the prevalence of conditions within the psychiatric domain, women with ASD by far had a greater rates of health care utilization in this domain. This was also the case for other domains, e.g., nutrition conditions, neurologic diseases, and sleep disorders. Taken together, similar to studies of sex differences in other areas of adult life, such as employment (Taylor et al., 2015; Taylor & Mailick, 2014), our findings highlight how specificity in measurement is critical in the study of sex differences in ASD.

It is important to note the limitations of the present study. Although the inclusion of health records from a large rural area was a strength, the racial and ethnic diversity of the patient population was limited. As with any study utilizing health records, data were collected for clinical and billing use rather than for research purposes. As such, we were limited in the range of questions that could be asked and we cannot account for variation in billing coding practices between providers across time. We do not know the length or quality of relationship that individual patients may have had with providers and if these relationships may have differed by sex and autism status. However, this is also a limitation in health care utilization studies using data collected in other ways unless the quality of the provider-patient relationship is directly measured. Additionally, we grouped codes within a domain to replicate the approach of Croen et al. (2015), and by doing so the prevalence within domains was not analyzed in the present study.

Additional research is needed into other domains that may be particularly salient for women with ASD such as inflammatory or autoimmune diseases. We also note that the age range was wide for the current study, such that individuals with ASD reflect different eras of diagnostic and treatment practices. Importantly, however, the current study was able to examine health objectively for a large number of diverse individuals with autism both in terms of age and level of functioning. Often research that utilizes self-report excludes some individuals with autism and this study addresses that gap. Future research is needed with other diverse groups of women with ASD to address questions regarding both how these sex differences emerge, and also how to remediate them. Finally, although our cases and controls were matched on year of birth, women with autism were disproportionately more likely to be decedents than men with autism or women without autism. We note that this was not a study of differential mortality, which has been shown in past research to be higher in individuals with ASD than community controls (Bishop-Fitzpatrick et al., 2018; DaWalt et al., 2019). Additional study is needed to understand early death among individuals with ASD, particularly women.

In conclusion, we found that women with ASD had significantly elevated risk for poor health. Consistent with prior literature (e.g., Croen et al., 2015; Jones et al., 2015), we found effects of having ASD and of being female, such that individuals with ASD had poorer health than those without ASD, and that females had a greater risk for health problems than males. Most importantly in the present analysis, we found evidence of double jeopardy for nutrition conditions, neurologic conditions, sleep disorders, and psychiatric conditions, as health care utilization in these domains was significantly greater for women with ASD than what one would expect for being a woman and for being on the autism spectrum. This consistent pattern of higher health care utilization for women with ASD compared to men (and to women without ASD) highlights the pressing need for further investigation into the unique needs of women on the autism spectrum. New research is needed to elucidate potential modifications to treatments and services to better support the health of women with ASD.