Sleep Patterns in Adults with a Diagnosis of High-Functioning Autism Spectrum Disorder

Abstract

Study Objectives:

To examine sleep patterns and sleep problems and their relationship with daytime functioning in adults with a diagnosis of an autism spectrum disorder and no comorbid intellectual disability (high-functioning autism spectrum disorder [HFASD]) compared to neurotypical (NT) adults.

Design:

Cross-sectional.

Setting:

Home-based study.

Participants:

36 adults with HFASD and 36 age-, intelligence quotient- and sex-matched NT adults.

Measurements:

Participants completed an online questionnaire battery including the Pittsburgh Sleep Quality Index (PSQI), a 14-d sleep wake diary and 14-d actigraphy data collection.

Results:

Adults with HFASD had significantly more general sleep disturbances and higher scores on the PSQI, longer sleep onset latencies (actigraphy), and poorer sleep efficiency (diary) and these results remained significant after accounting for the False Discovery Rate. Those adults with HFASD who did not have a comorbid diagnosis of anxiety/depression had significantly shorter total sleep time (diary and actigraphy) compared to NT adults. Compared to NT adults, the HFASD group self-reported significantly poorer refreshment scores upon waking in the morning and higher scores on the daytime dysfunction due to sleepiness subscale of the PSQI.

Conclusions:

These findings support the notion that problems related to sleep, in particular insomnia, continue into adulthood in individuals with high-functioning autism spectrum disorder.

Citation:

Baker EK, Richdale AL. Sleep patterns in adults with a diagnosis of high-functioning autism spectrum disorder. SLEEP 2015;38(11):1765–1774.

INTRODUCTION

Autism spectrum disorder (ASD) is characterized by deficits in social interaction and communication as well as a restricted range of interests and behaviors.¹ Sleep problems are one of the most common comorbid conditions experienced by individuals with ASD. However, most sleep research to date has been conducted with children and adolescents, with limited research focusing solely on adult populations. Nevertheless, sleep disturbances appear to persist into adulthood, indicating that they are a lifelong condition in individuals with ASD.^2–4

To date, few research groups have investigated sleep problems in adults with ASD without intellectual disability (ID), generally referred to as high-functioning ASD (HFASD), or adults with ASD and comorbid ID. Combined samples of adults and adolescents have been used,^4,5 which is problematic given the distinct differences in sleep patterns of adolescents and adults.⁶ Additionally, small sample sizes have been used; Tani et al.^2,7,8 had a small control group (n = 10) compared to 20 adults with Asperger syndrome (HFASD) whereas Hare et al.⁹ only had 10 adults with HFASD in their second study. Consequently, sleep problems have not been well characterized in adults with ASD and the effect sleep disturbance may have on their daytime functioning is relatively unknown.

In the study of 168 adults with ASD and ID,³ 44.7% had a sleep problem based on the Diagnostic Assessment of the Severely Handicapped-II¹⁰ (DASH-II), which is an informant-based screening tool containing items related to sleep problems. The DASH-II was completed by day-shift, direct-care staff members who worked closely with each of the participants. However, because the informants were day-shift staff members, they may not have been fully cognizant of specific sleep problems experienced by the individual and the severity of these problems. Compared to an older sample of 17 adults with ID only, 14 adults with ASD and ID did not significantly differ on measures of sleep quantity or sleep quality.¹¹

In HFASD, consistent with childhood literature, longer sleep onset latency (SoL), more frequent nocturnal awakenings, and poorer sleep efficiency (SE%) was found in 27 adolescents and adults (16–27 y) with HFASD compared to 78 neurotypical (NT) controls of similar age and Full Scale Intelligence Quotient (FSIQ). These findings were supported by polysomnography (PSG) in a subgroup of 16 HFASD and 16 NT participants.⁴ Hare et al.⁹ also reported longer SoL, more fragmented sleep, and poorer SE% in their 10 adults with HFASD. Similarly, using sleep questionnaires and diaries, and free description,² 20 adults (mean age 27 y) with HFASD reported frequent insomnia. Using PSG, a higher proportion of these participants with HFASD had wake after sleep onset (WASO) ≥ 30 min compared to 10 NT adults.⁷ In contrast, no significant group differences on sleep parameters were found for these adults using actigraphy; however adults with HFASD had a mean SoL of 43 min compared to 19 min in the NT group⁸; SoL > 30 min is problematic.

The literature in NT populations unequivocally indicates that poor sleep quality is related to poor daytime functioning and medical health problems.^12–15 Although anxiety and mood disorders are highly prevalent¹⁶ and physical health has recently been reported to be poor in individuals with ASD,¹⁷ the effect of poor sleep on health and daytime functioning in adults with ASD is relatively unknown. Tani et al.² reported that their participants with HFASD and insomnia had co-morbid psychopathology, whereas a recent report on 17 adolescents and young adults (16–27 y) with HFASD showed cognitive performance deficits following poor sleep.⁵ Research in children and adolescents with ASD indicates that poor sleep quality is associated with psychopathology, behavioral difficulties, and academic performance.¹⁸ Further, in addition to an increased prevalence of sleep disorders, adults with ASD have been found to have significantly higher rates of medical and psychological disorders including depression, anxiety, diabetes, gastrointestinal disorders, epilepsy, hyper-tension, and obesity.^17,19,20 Although the relationships between sleep problems and medical or psychological disorder remains largely unexamined for adults with ASD, it may be speculated that they are interrelated given what is known in the general population.²¹ The relationship between sleep and psychopathology is particularly pertinent given the high prevalence of anxiety and depression in individuals with ASD,¹⁶ and the associations found between anxiety and depression and sleep in the general population.²² It is plausible that heightened anxiety and/or the presence of depression may be causative factors for sleep disturbances in ASD.

In summary, sleep problems appear to continue into adulthood for individuals with ASD; however, these sleep problems remain poorly characterized. There are few studies, small sample sizes, and both adolescents and adults are included within the one group. Furthermore, the effect poor sleep may have on daytime functioning is largely unexplored. In a recent review,²³ the need for research in adults with ASD has been emphasized in order to determine the prevalence of sleep problems as well as to gain an understanding of the factors that may be associated with the manifestation and continuation of the sleep problems. It is important to understand the nature of sleep disturbances in this population given the strong links between sleep problems and poor health outcomes.²¹ Therefore, this article presents data assessing sleep disturbances in a well-characterized sample of adults with HFASD compared to age-, IQ-, and sex-matched NT controls. We evaluated whether adults with HFASD experienced reduced sleep quantity and quality, using both objective and subjective sleep measures, and assessed the impact of poor sleep on fatigue and daytime sleepiness.

METHOD

Participants

Thirty-six adults with a clinical diagnosis of HFASD and 36 NT adults participated in this study. The majority of participants in both groups lived with their parents. Although most participants in each group either worked full- or part-time, the groups did significantly differ on employment status; one-fourth of the participants with ASD were unemployed but only one NT participant was unemployed. Demographic details for the two groups are provided in Table 1.

Table 1.

Comparison of demographic variables between the high-functioning autism spectrum disorder and neurotypical groups.

The participants with HFASD were recruited through various Australian Autism Associations, adult ASD support groups, the Olga Tennison Autism Research Centre Participant Registry, and advertisements placed in clinics specializing in adults with ASD, as well as via the general public. The NT sample was recruited primarily from advertisements placed at La Trobe University and via the general public as well as from the La Trobe University School of Psychological Science Participant Registry.

The Wechsler Abbreviated Scale of Intelligence (WASI)²⁴ was administered to most of the participants. Nine participants with HFASD and three NT participants did not complete the WASI; five adults with HFASD had recently completed the WAIS-IV²⁵ as part of their diagnostic process and their FSIQ scores were obtained from their reports. Other reasons for noncompletion of the WASI were that three participants with HFASD and two NT participants resided in distant Australian states and there was no funding support for travel to those states, one participant with HFASD only wanted to complete the online components of the study, and one NT participant was a professional who regularly administers the WASI. These participants were asked to report their highest level of education attained in their screening questionnaire; one NT participant and one with HFASD had attained a diploma, three participants with HFASD and one NT participant had attained or were currently completing a bachelors degree, and one NT participant had attained a postgraduate degree; thus, these individuals were considered high functioning. All participants who completed the WASI/WAIS-IV had a FSIQ > 80 and the two groups did not differ on FSIQ (Table 1).

This study was part of a larger study in which participants provided saliva samples for the analysis of melatonin and cortisol. Thus, participants with a diagnosis of schizophrenia were excluded given that abnormalities are frequently seen in cortisol levels of individuals with schizophrenia.²⁶ Further, individuals taking any type of sleeping medication (e.g., melatonin) were also excluded. Two individuals with HFASD were taking melatonin prior to the study; however, with approval from their general practitioner they ceased melatonin use for the duration of the study. Due to the high rate of comorbid anxiety and depression reported for individuals with ASD, participants with HFASD with these comorbidities were not excluded. Potential NT participants were excluded if they had a first-degree relative with a diagnosis of an ASD, as abnormalities in the genes associated with melatonin have been reported in first-degree relatives of individuals with ASD.²⁷ NT participants were also excluded if they had a diagnosis of an anxiety or mood disorder, of if they exceeded the ASD cutoff score on the Autism Quotient²⁸ (AQ; see next paragraphs). As participants were required to provide self-reports of their sleep and daytime functioning, individuals who were low functioning (FSIQ < 80) and would require assistance with completing the study were also excluded.

Confirming ASD Diagnosis

Individuals with HFASD were screened into the study by self-reporting a diagnosis of ASD. All participants had either an Autism Diagnostic Observation Schedule (ADOS) assessment or provided a copy of their diagnostic report and most participants had both. Thirty-four participants with HFASD provided a copy of their diagnostic report or a statement from their clinician (one participant), confirming their diagnosis. Thirty-two participants with HFASD also completed Module 4 of the Autism Diagnostic Observation Schedule 2^nd Edition (ADOS-2)²⁹ (mean [M] = 11.59, standard deviation [SD] = 4.13). The ADOS-2 was administered by the first author, a trained assessor demonstrating ≥ 80% coding reliability on the ADOS modules (EB is currently undertaking her training to also become an ADOS Trainer). Fifteen percent of video recordings were taken to reliability coding meetings to confirm classifications. Reasons for nonadministration of the ADOS-2 are the same as those described for the WASI. Of those who completed the ADOS-2, 25 met criteria on the recently published revised algorithm.³⁰ This ADOS classification rate is consistent with previous reports for adults with ASD.³¹ All of those participants who did not meet criteria on the ADOS-2 provided a copy of their diagnostic report confirming their diagnosis, and received the diagnosis from an Australian clinician specializing in ASD. All NT participants fell below the exclusionary cutoff of 26³² on the AQ (see Table 1).

Psychopathology and Medication Use

Fifteen participants with HFASD reported a comorbid clinical diagnosis of an anxiety or mood disorder and were taking associated medications for these diagnoses; four had anxiety, five had depression, and six had both anxiety and depression. One NT participant reported using a tricyclic antidepressant; however, this was for pain management, not a mood disorder. Six NT participants and three with HFASD were taking an oral contraceptive or had an Implanon® (Merck & Co. Inc., White-house Station, NJ) whereas five participants with HFASD and one NT participant were using an asthma medication (e.g., Salbutamol). Table 2 presents a full list of medications taken by participants during the 2-w data collection period, their reason for use, and potential effects on sleep.

Table 2.

Medication use and their potential effect on sleep.

Materials

Screening Questionnaire

All participants completed a screening questionnaire prior to entering the study. Demographic information including first language, ASD diagnosis or family history, comorbid diagnoses, employment and marital status, living arrangements, and medication use were collected with this measure.

Sleep Measures

Participants completed the Pittsburgh Sleep Quality Index (PSQI).³³ This is a retrospective, self-report questionnaire used to measure the quality and patterns of sleep in adults over the past month. It consists of 19 self-rated items and five questions rated by the bedpartner or roommate, which are not included in the calculation of the total or subscale scores; however, as questions were answered online the latter five questions were not included. Scores range from 0 to 21, with higher scores indicating worse sleep quality. The PSQI has previously shown high internal consistency (α = 0.83).³³ In the current study the PSQI showed good internal consistency for both groups (HFASD α = 0.68; NT α = 0.62).

Participants completed an online 14-d sleep/wake diary every morning and evening for the 14-d data collection period. Sleep diary variables were averaged and the following sleep parameters were calculated: total sleep time (TST), SoL, WASO, and SE%. In addition, participants reported their waking method each morning by selecting one of the following options: Alarm Clock/Radio, Someone whom I asked to wake me (e.g., partner, parent), Noises, or Just woke up.

Participants wore an actigraph monitor (Actiwatch 2, Respironics Inc., Murraysville, PA) on their nondominant wrist for the duration of the 14-d data collection period. Data were digitized in 1-min epochs with a sensitivity of 0.025 g and a bandwidth of 0.35–7.5 Hz, and analyzed with Respironics Actiware 6 software. As in the diary several sleep parameters were evaluated including: TST, SoL, SE%, WASO, and the fragmentation index (FI). Sleep onset was defined as the first of five consecutive epochs of actigraphic sleep at the beginning of the scoring interval. Sleep offset was defined as the last of at least five consecutive epochs of actigraphic sleep at the end of the scoring interval.

Insomnia

We used information from the PSQI, sleep diary, and actigraphy to assess insomnia symptoms in both our groups. According to the International Classification of Sleep Disorders, Third Edition (ICSD-3),³⁴ insomnia is characterized by difficulties initiating and/or maintaining sleep despite adequate opportunity to sleep with resultant daytime consequences. These symptoms should occur at least three times per week and persist for at least 3 mo. Participants had to have reported experiencing a SoL greater than 30 min and/or WASO/early morning waking on the PSQI at least three times per week. In addition, their average SoL from the sleep diary and actigraphy had to be greater than 30 min and/or their WASO had to be one standard deviation above the mean of the NT group. Last, participants had to have scores on the Epworth Sleepiness Scale (ESS) and/or Flinders Fatigue Scale (FFS) that were above the clinical cutoff or one standard deviation above the mean of the NT group, respectively.

Daytime Functioning Measures

Daytime sleep propensity and fatigue were examined via the ESS³⁵ and FFS,³⁶ respectively. The ESS consists of eight different situations that most people engage in during their daily lives. Participants rate their probability of falling asleep during each situation on a four-point Likert scale. Scores range from 0 to 24, with higher scores indicating greater daytime sleepiness. The ESS has good internal consistency (α = 0.88)³⁷; in the current study, HFASD α = 0.79; NT α = 0.67.

The FFS is a retrospective seven-item self-report questionnaire that measures the severity, consequences, and frequency of fatigue experienced by the participant over a 2-w period. Six items are presented in Likert format, ranging from 0 (not at all) to 4 (extremely). Item five requires the respondents to determine the time of day when fatigue is experienced using a multiple-item checklist (e.g. “early morning”). More than one response for this item is acceptable and it is scored as the sum of all the times of day indicated by the respondent. FFS scores range from 0 to 31, with higher scores indicative of greater fatigue. This scale has previously been shown to have good internal consistency (α = 0.91)³⁶; in the current study, the FFS demonstrated high internal consistency for both groups with α = 0.84 in each group.

Participants were instructed to complete these questionnaires during the 2-w data collection period. The majority of participants in each group completed the questionnaires on day 1 (ASD = 12, NT = 13) or in the remainder of the first week (ASD = 11, NT = 12). The remaining participants completed the questionnaire either in the second week (ASD = 2, NT = 5) or in the 2 w following the data collection period, as they had forgotten to complete it during the 2-w data collection period (ASD = 8, NT = 9); the two groups did not differ on the timing of completion, χ² (3, 72) = 1.43, P = 0.699, Φ = 0.14. Within the sleep diary, sleep refreshment upon waking was assessed by asking participants to respond to the following question, “On the scale below, please rate how refreshed you felt after your sleep last night”; scores ranged from 1 to 9, with higher scores indicating better refreshment.

Procedure

Institutional ethics approval was granted (Approval #HEC12-018). Participants who responded to an advertisement were contacted and completed the screening questionnaire. At the initial appointment participants signed a consent form, completed the WASI and ADOS-2 (HFASD only) assessments, and received their actigraphy monitor. Prior to this appointment, each participant was sent a link to their online questionnaires and online 14-d sleep/wake diary. Participants were instructed to follow their typical sleep patterns during the 2-w data collection period. The first author (EB) logged in daily to check participants' diaries for completion and accuracy. Actigraph monitors were retrieved following the 14-d data collection period and data were downloaded via the Respironics Actiware 6 software (Respironics Inc., Murraysville, PA). All data were then entered into the IBM Statistics Software Package (IBM Corp., Armonk, NY)³⁸ for analysis.

Data Analysis

Preliminary Analyses

Given the known relationship between psychopathology and sleep in the general population, preliminary analyses were conducted to determine if there were differences between the 15 participants with ASD who were medicated for a comorbid diagnosis of anxiety and/or depression (ASD Comorbid) and those participants with ASD with no comorbid psychopathology diagnoses (ASD Only). Mann-Whitney U tests were conducted for variables that were not normally distributed in one or both of the ASD groups and Independent samples t-tests were used for variables that were normally distributed in both groups. The results revealed that the two ASD groups did not differ on any variables except TST measured by both the diary and actigraphy. The ASD Only group obtained less sleep compared to the ASD Comorbid group, effect sizes were moderate (see Tables S1–S4 in supplemental material). In addition, the two ASD groups did not differ significantly on age (ASD Only: M = 34.27, SD = 6.56; ASD Comorbid: M = 34.61, SD = 6.67), t (34) = 0.153, P = 0.879, η² = 0.00. Both groups also had a similar male to female ratio (ASD Only: 52.4% male; ASD Co-morbid: 40.0% male), χ² (1, 36) = 0.156, P = 0.693, Φ = −0.122, and did not differ on employment status, χ² (4, 36) = 2.658, P = 0.617, Φ = 0.272.

Thus, given the two ASD groups did not differ significantly on relevant demographic variables or sleep variables except TST, these groups were collapsed and compared to the NT group for all variables except TST, which was analyzed separately.

Data Screening and Cleaning

Screening revealed random missing data for sleep diary variables. When SoL and/or WASO were not reported by the participant, TST and SE% were not calculated so as not to overestimate these parameters. This information was only missing for 11 recorded days across the entire sample (0.01%). As this study was part of a larger study some participants were required to adhere to a sleep schedule for saliva collection on 1 night of the data collection period; therefore, this night was removed prior to calculating these participants' average scores on sleep variables for both the sleep diary and actigraphy. Some participants did not collect data (diary and/or actigraphy) on all possible days. The primary reason for noncompletion of the diary or missing actigraphy data was participants forgetting to complete the diary or to put the actigraphy monitor back on after showering. One NT participant removed the actigraphy monitor for 2 d due to irritation. Thus, number of available nights ranged from 7 to 14 (diary) and 9 to 14 (actigraphy) in the ASD group and 9 to 14 (diary) and 10 to 14 (actigraphy) in the NT group. One participant with HFASD only completed 3 d of the diary, and consequently their diary data were removed from any calculations.

Random outliers were found for several sleep variables in both groups; however, there were two female participants with ASD who were outliers on several sleep variables measured by actigraphy and the diary. As abnormality or atypicality in the sleep patterns of some individuals with HFASD was hypothesized, participants presenting as outliers were retained for the analyses.

In both groups all daytime functioning variables were normally distributed, as well as TST and SE% (diary). WASO (diary) and SE% (actigraphy) and all subscales of the PSQI were not normally distributed for either group. In addition WASO and SoL (actigraphy) were not normally distributed in the HFASD group and SoL (diary) and the Fragmentation Index (FI) were not normally distributed in the NT group. Global PSQI scores were normally distributed in the ASD group, but not in the NT group.

Analyses

Independent samples t-tests were used to compare differences between the two groups on all sleep and daytime functioning variables that were normally distributed in both groups; effect sizes (η²) and 95% confidence intervals (CIs) of the difference are also reported. Where the assumption of homogeneity of variance (Levene test for equality of variance) has been violated, the more conservative P values are reported. Mann-Whitney U tests were used to compare the differences between the two groups on sleep variables that violated the assumption of normality; effect sizes (r) are also reported. Last, the chi-square test for independence was used to compare the proportion of participants who met criteria for insomnia in each group.

Given the number of comparisons made we encourage the reader to focus on the effect sizes and CIs (where relevant) rather than primarily focusing on P values. Effect sizes and CIs provide clinically meaningful information regarding group differences, as they provide the magnitude of the difference between the two groups.^39,40 The M and SD are also reported for the Mann-Whitney U tests as these can be informative when comparing across other studies. In addition, the False Discovery Rate (FDR)⁴¹ method has been used to control for the number of comparisons made. The FDR controls for the expected proportion of falsely rejected hypotheses and is a more powerful method than traditional false positive rate methods.⁴²

Last, to determine the relationship between sleep and daytime functioning variables Pearson correlations were used for variables that were normally distributed, whereas Spearman rho correlation analyses were used for variables that violated the assumption of normality.

RESULTS

Questionnaire Measures (Table 3)

Table 3.

Comparison of questionnaire measures between the groups using Mann-Whitney U tests and independent samples t-tests.

Pittsburgh Sleep Quality Index

Participants with HFASD had significantly higher scores on the PSQI compared to NT adults; the effect size was moderate. Participants with HFASD also had significantly higher scores on the sleep disturbances, and daytime dysfunction due to sleepiness subscales with moderate effect sizes. These differences remained significant after applying the FDR. No significant differences were found for the SoL, duration of sleep, or SE% subscales; however, there was significantly more variability in the HFASD group on the sleep duration (P = 0.005), SE% (P = 0.012) and sleep disturbance (P = 0.023) subscales based on Levene test for equality of variance.

Daytime Functioning

Adults with HFASD reported significantly lower refreshment scores with a large effect size and more daytime fatigue with a moderate effect size compared to NT participants; however, after applying the FDR only refreshment scores remained significantly different. Although there was a trend for more daytime sleepiness in the HFASD group, this was not significant. Daytime fatigue and refreshment scores were significantly correlated in both groups (HFASD, r = −0.508, P < 0.01; NT, r = −0.619, P < 0.001). No other significant correlations were found for daytime functioning measures in either group.

Sleep Diary (Table 4)

Table 4.

Comparison of diary and actigraphy variables between the groups using Mann-Whitney U tests and independent samples t-tests.

Adults with HFASD reported significantly poorer SE% compared to NT adults with a moderate effect size, and those adults with HFASD and no comorbid psychopathology diagnosis had significantly shorter sleep than NT adults with moderate effect size (Table 5). Both of these findings remained significant after controlling for multiple comparisons. In addition, there was a trend for longer SoLs. Again, there was significantly more variability in the HFASD group on all sleep parameters (Levene test for equality of variance P < 0.01). The two groups did not differ on the average number of days they woke up to an alarm; HFASD (M = 4.29, SD = 3.96) NT (M = 5.39, SD = 3.42), t (69) = 1.258, P = 0.213, η² = 0.02.

Table 5.

Comparison of diary and actigraphy total sleep time between the two autism spectrum disorder groups and neurotypical group using independent samples t-tests.

Actigraphy (Table 4)

Similar to the diary, adults with HFASD had significantly longer SoL, poorer SE%, and more fragmented sleep; effects sizes were small. After controlling for multiple comparisons only SoL differed significantly between the two groups. Adults with HFASD and no comorbid psychopathology also had significantly shorter sleep duration; the effect size was moderate (Table 5). This finding remained significant after applying the FDR. Adults with HFASD also had significantly more variable SoL (P = 0.001) and SE% (P = 0.049) compared to the NT group. No other significant differences were found.

Associations between Daytime Functioning and Sleep Variables

Daytime fatigue was only correlated significantly with diary-reported SoL in the HFASD group (r_s = 0.351, P = 0.039). Diary SoL was also moderately correlated with sleep refreshment; however, this finding was not significant (r_s = −0.303, P = 0.076). In the NT group, sleepiness was correlated significantly with diary WASO (r_s = 0.358, P = 0.032) and SE% (r_s = −0.378, P = 0.023). Diary SoL was also moderately correlated with sleepiness; however, this was not significant (r_s = 0.310, P = 0.065). A nonsignificant moderate correlation was seen between sleep refreshment and diary SE% in the NT group (r_s = 0.311, P = 0.065). Only weak correlations were shown with actigraphy variables across both groups.

Insomnia

Using the aforementioned criteria, ten of the HFASD adults (27.8%) and two NT adults (5.56%) met criteria for insomnia; this difference was significant with moderate effect size, χ² (1, n = 72) = 4.90, P = 0.027, Φ = −0.30.

DISCUSSION

To our knowledge this is one of the first studies to compare sleep using a retrospective questionnaire, 14-d sleep diary, and actigraphy in a sample comprising solely of adults with HFASD. The key findings of this study are that adults with HFASD have more general disturbances associated with sleep (PSQI), take a significantly longer time to fall asleep with resultant poorer SE%, and poorer refreshment upon waking compared to an age-, sex-, and IQ-matched NT control group. The findings are discussed in detail in the next paragraphs.

Pittsburgh Sleep Quality Index

Our adults with HFASD had significantly higher total scores on the PSQI, as well as higher scores on the sleep disturbances and daytime dysfunction due to sleepiness subscales, with moderate effect sizes. These findings remained significant after correcting for multiple comparisons. Moreover, the average score for the HFASD group was above the PSQI clinical cutoff of 5, which is associated with poor sleep quality, whereas the NT group average was 5. Examining the items that contribute to the sleep disturbance subscale, more adults with HFASD endorsed the items “cannot breathe comfortably,” “feel too cold,” and “other” compared to the NT adults. Common themes reported in the “other” category among both groups were children or the bedpartner disturbing sleep and anxiety/ stress. Despite there being no differences in the number of participants who were parents in either group more NT adults (5) reported children disturbing sleep compared to adults with HFASD, (1) although this may be related to the age of the participants' children. Ten adults with HFASD and six NT adults reported anxiety/stress disrupted their sleep. In addition, three adults with HFASD reported that noises disturbed their sleep, which was not reported by any NT adults. Sensory issues are now part of the diagnostic criteria for ASD¹ and are common: Noise sensitivity may be a reflection of these issues. Sensory issues may also explain the adults reporting temperature issues (e.g., feeling too cold). Although breathing issues may typically be associated with sleep disordered breathing diagnoses, sleep disordered breathing is not commonly reported in the ASD literature. Nevertheless, given the increased prevalence of obesity¹⁷ in adults with ASD and the associations between obesity and sleep disordered breathing,⁴³ this warrants further investigation.

Sleep Diary and Actigraphy

Adults with HFASD had significantly longer actigraphy SoL and more fragmented sleep (actigraphy) with resultant poorer SE% (diary and actigraphy) compared to the NT group; however, only actigraphy SoL remained significantly different between the two groups after controlling for multiple comparisons. On average, the HFASD group took between 8 (actigraphy) and 13 (diary) min longer to fall asleep. In relation to TST, adults with HFASD and no comorbid psychopathology had significantly shorter sleep duration compared to NT adults. These findings remained significant even after controlling for multiple comparisons. On average, the HFASD Only adults obtained 30 min (actigraphy) to 37 min (diary) less sleep than NT adults, whereas those adults who were medicated for co-morbid anxiety and depression did not differ significantly from NT adults on measures of TST. In fact, those adults medicated for psychopathology diagnoses obtained slightly more sleep on average than NT adults (diary: 10 min; actigraphy: 7 min). Considering that short sleep is also reported for children with ASD,^44,45 short sleep may be a characteristic of ASD. However, a comorbid diagnosis of anxiety and/or depression appears to result in adequate sleep duration but sleep quality remains poor. These individuals, particularly those with anxiety, may be more inclined to follow strict routines around bed and wake times, resulting in more opportunity for sleep. Rigidity and adherence to routines are characteristic of individuals with ASD and it has been argued that individuals engage in these behaviors in order to reduce anxiety.⁴⁶ However, those individuals without anxiety may be less inclined to follow such protocols. Depression has also been associated with hypersomnia with individuals experiencing prolonged sleep at night, which may also account for the increased TST noted in these adults with HFASD.¹

In the current study, SE% as measured by actigraphy was < 85% for both groups, which is considered problematic.⁴⁷ These findings are consistent with previous PSG findings⁴ as well as our data on adolescents with HFASD, in which adolescents with HFASD had significantly longer SoL (actigraphy) and poorer SE% (diary) compared to age- and sex-matched NT adolescents.⁴⁸ In contrast, using actigraphy, Tani et al.⁸ found no significant sleep differences between young adults with Asperger syndrome and young NT adults; however, they did find reduced SE% on sleep diaries.²

Insomnia symptoms are the most commonly reported problems in individuals with ASD¹⁸. Tani et al.² reported that between 75% and 90% of participants met criteria for insomnia based on the Basic Nordic Sleep Questionnaire (BNSQ), 5- to 7-day sleep diary, and free description. In this study, 27.8% of the adults with HFASD and 5.56% of NT adults had symptoms associated with insomnia over the past month, which is the time period measured by the PSQI. Although these estimates are lower than the previous rates reported in adults with ASD, we have used a more stringent classification compared to that of Tani and colleagues.² It is likely that previous estimates² are inflated given their self-report nature and the fact that daytime consequences were not included in the criteria. In the current study, an additional five adults with HFASD and one NT participant met all the insomnia criteria expect for the daytime dys-function component; thus, we may have underestimated. This may also represent a reporting bias, particularly if participants are chronically sleep deprived and are used to feeling sleepy and fatigued throughout the day. This may be particularly true of the participants with HFASD who may have chronic sleep difficulties from childhood. It should be noted that ICSD-3 criteria requires that the symptoms should be present for 3 mo and our estimates are based on a 1-mo retrospective questionnaire and 2-w diary/actigraphy, and thus caution should be used when interpreting these results; however, it is likely that insomnia symptoms are ongoing given the persistent nature of sleep problems in individuals with ASD.

Adults with HFASD had more fragmented sleep compared to NT adults; however, this was no longer significant after correction. More fragmented sleep in the HFASD group is in line with the findings of Limoges and colleagues⁴ and Hare et al.,⁹ who found more frequent nocturnal awakenings using a sleep questionnaire and more fragmented sleep via actigraphy, respectively.

Daytime Functioning

In line with our previous research in adolescents,⁴⁸ our adults with HFASD had more daytime fatigue compared to NT adults, although this did not remain significant after correction. In the current study, adults with HFASD also reported significantly poorer refreshment upon waking in the morning, with a large effect size. Although there were no differences between the two groups on the ESS, adults with HFASD had significantly higher scores on the daytime dysfunction due to sleepiness subscale of the PSQI with a moderate effect size. The ESS measures an individual's propensity to fall asleep during everyday situations whereas in the PSQI, two items load onto the sleepiness subscale; “difficulty staying awake” and “enthusiasm.” It is possible that the “enthusiasm” item taps into feelings of fatigue and/or depression, rather than sleepiness. Thirty-three adults with HFASD endorsed the “enthusiasm” item, whereas only 23 of them endorsed the “difficulty staying awake” item. Thus, the difference seen between the two groups on the daytime dysfunction subscale of the PSQI is likely to be driven by fatigue or depressive symptoms rather than sleepiness.

Despite adults with HFASD reporting more fatigue and there being problematic sleepers in both groups, fatigue was only significantly correlated with diary SoL within the HFASD group. Although fatigue may be associated with sleep in some individuals, particularly sleep onset, it is possible that other factors such as ASD diagnosis, depression, employment, and looking after young children are causes of fatigue in both adults with HFASD and NT adults. The last item of the FFS asks respondents to rate how much their fatigue was related to poor sleep. This item is rated on a five-point Likert scale, ranging from not at all to entirely. Nine participants in each group stated not at all whereas four participants in each group stated entirely. Eleven participants in the HFASD group endorsed the almost entirely category compared to seven NT adults. Thus, it is evident that although sleep contributes to feelings of fatigue in both groups, there are other factors that could also be involved.

Only sleep diary variables were correlated with daytime sleepiness and refreshment upon waking in each group. This suggests that subjective feelings of poor sleep are related to poor self-reported daytime functioning. Thus, it is likely that sleep quality plays a role in daytime functioning and future research should use objective as well as subjective measures of both sleep and daytime functioning to gain further insight.

Variability in the HFASD Group

There was much more variability in the HFASD group for most of the sleep variables. ASD is a very heterogeneous disorder and it is likely that subgroups of individuals exist in relation to their sleep profiles and etiology of their sleep problems. The childhood literature also appears to suggest this, with children being classified as early morning wakers, night wakers (increased WASO), increased SoL, or short sleepers. More research is needed to further understand the variability observed in the sleep patterns of individuals with ASD and to determine the underlying causes of differing profiles. This will enable more targeted interventions and treatments to be developed.

Strengths and Limitations

This study has two main strengths. First, it extends previous literature examining sleep in adults with ASD. Specifically we have a larger group comprising solely adults who have been carefully matched with NT adults. We report on three measures of sleep, questionnaire, diary, and actigraphy and have found significant group differences on all these measures. Second, we have controlled for multiple comparisons using the FDR and still observed significant differences between the two groups; previous studies in this area have generally not controlled for multiple comparisons.

It should be noted that medications taken by individuals with HFASD may affect their sleep patterns and therefore, it would be beneficial in future research to recruit a larger sample of individuals who do not have comorbid psychopathology diagnoses. However, our data were collected across a 2-y period across six of the eight Australian states and territories, indicating that it is difficult to recruit adults with HFASD and particularly those with no comorbid psychopathology. Nevertheless, our two groups of adults with HFASD did not differ significantly on any measures of sleep and daytime functioning, except TST. Given the increased prevalence of psychopathology in individuals with ASD and the known effect psychopathology has on sleep in the general population, it is important to gain a better understanding of the relationship between sleep and psychopathology in individuals with ASD. It is also important to note that poor sleep quality was still evident in both our HFASD groups, with similar means and score ranges observed across the two groups.

The phrasing of the questions and retrospective nature of the PSQI may explain why no differences were observed for the SE%, SoL, and sleep duration subscales whereas differences were observed on the diary as well as actigraphy. For example, within the PSQI participants are asked how frequently they “cannot get to sleep within 30 minutes” with responses ranging from “not during the past month” to “three or more times per week,” while the diary and actigraphy are the average of all SoLs across the 14-d period. A participant who has sporadic difficulties with sleep onset, only three times per week, will have a sleep diary and actigraphy profile very different from that of someone who has difficulties with sleep onset every night of the week; however, they will score similarly on the PSQI. In addition, the majority of participants completed the questionnaire battery within the first week of the data collection period; therefore, there was little overlap with the actigraphy assessment. Future research studies should aim to control the timing of the completion of retrospective questionnaires to coincide with the objective measures of sleep.

CONCLUSION

Few research groups have previously examined sleep in adults with HFASD, and adolescents have been included, which is problematic.¹¹ Participants in our study were well characterized and carefully matched. Our results support that poor sleep persists into adulthood in individuals with ASD, with reduced sleep quality, lengthy SoLs, and poor SE% negatively affecting daytime functioning of these individuals. Future research should be directed toward understanding the cause of sleep disturbance to prevent its persistence across the lifespan, in particular the relationship between comorbid psychopathology and sleep. Research should also include individuals with ASD and comorbid intellectual disability. It is important to gain an understanding of the issues experienced by adults with ASD so that tailored interventions and supports are developed. Across all areas of ASD research, attention toward adults is limited. Sleep is a critical issue to understand in adults with ASD given the effect it can have on so many facets of life, many of which tend to be exacerbated by ASD.

DISCLOSURE STATEMENT

This was not an industry supported study. Financial support was provided by a PhD grant from The Apex Trust for Autism. The authors have indicated no financial conflicts of interest.

SUPPLEMENTAL MATERIAL

Table S1

Comparison of Pittsburgh Sleep Quality Index variables between the two autism spectrum disorder groups using Mann-Whitney U tests and independent sample t-tests.

Table S2

Comparison of diary and actigraphy variables between the two autism spectrum disorder groups using independent samples t-tests.

Table S3

Comparison of diary and actigraphy variables between the two autism spectrum disorder groups using Mann-Whitney U tests.

Table S4

Comparison of daytime functioning variables between the two autism spectrum disorder groups using independent samples t-tests.