965

sorders

JAD17310.1177/1087054711428965

v

Articles

Associations Between Sleep Characteristics, Seasonal Depressive Symptoms, Lifestyle, and ADHD Symptoms in Adults

Journal of Attention Disorders 17(3) 261­–275 © 2011 SAGE Publications Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/1087054711428965 jad.sagepub.com

Denise Bijlenga1, Kristiaan B. van der Heijden2, Minda Breuk1, Eus J. W. van Someren3, Maria E. H. Lie4, A. Marije Boonstra1, Hanna J. T. Swaab2, and J. J. Sandra Kooij1

Abstract Objective: The authors explored associations between ADHD symptoms, seasonal depressive symptoms, lifestyle, and health. Method: Adult ADHD patients (n = 202) and controls (n = 189) completed the ASESA questionnaire involving lifestyle, eating pattern, and physical and psychological health, and validated measures on ADHD and sleep. ASESA is the Dutch acronym for Inattention, Sleep, Eating pattern, Mood, and General health questionnaire. Results: Indication for delayed sleep phase syndrome (DSPS) was 26% in patients and 2% in controls (p < .001). Patients reported shorter sleep, longer sleep-onset latency, and later midsleep. Shorter (R2 = .21) and later (R2 = .27) sleep were associated with hyperactivity, male gender, younger age, and seasonal depressive symptoms. Seasonal depressive symptoms were related to hyperactivity, female gender, unemployment, and late sleep (pseudo R2 = .28). Higher body mass index (BMI) was associated with shorter sleep in patients (ρ = −.16; p = .04) and controls (ρ = −.17; p = .02). Longer sleep showed lower odds for indication of metabolic syndrome (OR = −0.17; p = .053). Conclusion: DSPS is more prevalent in ADHD and needs further investigation to establish treatment to prevent chronic health issues. ( J. of Att. Dis. 2013; 17(3) 261-275) Keywords ADHD, sleep, circadian rhythm, seasonal depressive symptoms, sleep-onset latency, sleep-onset insomnia, delayed sleep phase syndrome

Introduction The lifetime neuropsychiatric disorder ADHD is characterized by inattention and/or hyperactive/impulsive symptoms and presents with three clinical subtypes: the inattentive, the hyperactive/impulsive, and the combined subtypes (American Psychiatric Association [APA], 1994). ADHD commences in childhood and often persists into adulthood (Biederman, 1998). International prevalence estimates of adult ADHD range from 1% to 7% (M = 3.4%), with a majority of the combined subtype (Faraone, Sergeant, Gillberg, & Biederman, 2003; Fayyad et al., 2007). Around three quarters of adults with ADHD has at least one psychiatric comorbid disorder (Biederman et al., 1993; Kooij, Aeckerlin, & Buitelaar, 2001). Circadian rhythm disturbances are caused by a misalignment between the endogenous timing system in the brain and the external 24-hr environment (Bjorvatn & Pallesen, 2009). The behavioral domains that are affected in ADHD

(hyperactivity, inattention, and impulsivity) seem closely related to disturbances in the circadian rhythm (Lecendreux & Cortese, 2007). Sleep disorders such as sleep-related movement disorders, parasomnias, hypersomnias, and circadian rhythm disorders are prevalent in most ADHD patients (Gau & Chiang, 2009; Gau et al., 2007; Gruber et al., 2009; Middelkoop, Gils, & Kooij, 1997; Owens, 2009; Philipsen et al., 2005; Philipsen, Hornyak, & Riemann, 2006; Stein et al., 2002; Surman et al., 2009; Van der Heijden, Smits, 1

PsyQ, Expertise Center Adult ADHD, The Hague, Netherlands Leiden University, Center for the Study of Developmental Disorders, Leiden, Netherlands 3 Netherlands Institute for Neuroscience, Amsterdam, Netherlands 4 Academic Hospital Maastricht, MedPsych Center, Netherlands 2

Corresponding Author: Denise Bijlenga, PsyQ Program Adult ADHD, Carel Reinierszkade 197, 2593 HR, The Hague, Netherlands. Email: [email protected]

262 Van Someren, & Gunning, 2005; Walters, Silvestri, Zucconi, Chandrashekariah, & Konofal, 2008). Around 80% of ADHD patients may have a delayed sleep, which is prevalent in children and adolescents as well as in adults with ADHD (Van der Heijden, Smits, & Gunning, 2005; Van Veen, Kooij, Boonstra, Gordijn, & Van Someren, 2010). The delayed sleep phase syndrome (DSPS) is characterized by a delay of the sleep/wake cycle, with high activity level in the late evening and night, sleep-onset insomnia (SOI) when trying to get asleep early, and a preference for late sleep and late rise, which causes impairment in daily functioning. These people are often referred to as “night owls.” In our previous study by Van Veen et al. (2010), we measured salivary melatonin level in patients with ADHD with and without SOI every hour between 09:00 p.m. and 01:00 a.m. We found that ADHD patients with SOI reached the threshold for dim-light melatonin onset (DLMO) at least 01:41 hr later than healthy controls, and 01:15 hr later than ADHD patients without SOI (Van Veen et al., 2010). The delayed sleep in ADHD thus may be characterized by a delayed DLMO, and possibly in some even a completely lacking nocturnal melatonin peak because in many patients no melatonin increase could be observed at 01:00 a.m., the time of the final measurements (Van der Heijden, Smits, Van Someren, et al., 2005; Van Veen et al., 2010). The results are comparable with those found in children with ADHD with and without SOI (Van der Heijden, Smits, Van Someren, et al., 2005). Also in other studies the relationship between ADHD, delayed sleep, and sleep-onset insomnia have been reported (Chiang et al., 2010; Lecendreux & Cortese, 2007). Delayed sleep can lead to sleep debt and daytime sleepiness because of the out-of-phase circadian rhythm of the patient in relation to the social environment that demands early wake-up time for school or work (Shirayama et al., 2003). Also, eating at socially demanded time, which may be out of phase with the patient’s biological clock for appetite and digestive hormone releases, is hypothesized to lead to eating at unpreferred times, skipping breakfast, and binge eating (Copinschi, 2005). On the long term, the eating and sleeping at undesirable times can contribute to the development of, among others, mood disorders, obesity, cardiovascular disease, and immune suppression (Copinschi, 2005; Fonken et al., 2010; Lewy, 2009; Shirayama et al., 2003). Studies show that hyperactivity as well as short sleep are related to obesity (Fuemmeler, Ostbye, Yang, McClernon, & Kollins, 2010; Lauderdale et al., 2009). We also know that a disrupted melatonin release leads to increased risk of cardiovascular disease and cancer (Blask, 2009; Puttonen, Harma, & Hublin, 2010; Viswanathan & Schernhammer, 2009). Overall, multiple studies have demonstrated that circadian sleep disturbances may be a hazard to public health, and therefore, research is needed to further develop prevention and treatment methods.

Journal of Attention Disorders 17(3) A common comorbid disorder among individuals with ADHD is the seasonal affective disorder (SAD), which is characterized by episodes of major depression that recur during specific times of the year, usually in winter (Lurie, Gawinski, Pierce, & Rousseau, 2006). SAD is prevalent in 19% to 27% of adult ADHD patients (Amons, Kooij, Haffmans, Hoffman, & Hoencamp, 2006; Levitan, Jain, & Katzman, 1999). Patients with SAD have increased prevalence of high caloric intake, hypertension, obesity, diabetes, metabolic syndrome, and low physical exercise (Rintamaki et al., 2008). SAD and ADHD may share a common origin: In a relatively small sample, SAD and ADHD were more prevalent among women with a serotonin 2A-receptor gene (HTR2A) polymorphism (Levitan et al., 2002). Like ADHD, SAD is also linked to DSPS, seasonal bright light deficiency, and subjective sleep deficiency (Lewy, Lefler, Emens, & Bauer, 2006; Oyane, Ursin, Pallesen, Holsten, & Bjorvatn, 2008; Rintamaki et al., 2008). SAD can be effectively treated with bright light therapy in which the biological clock can be “reset” (Gruber, Grizenko, & Joober, 2007; Rastad, Ulfberg, & Lindberg, 2008). Bright light therapy is presumed to act on the biological clock of the brain localized in the hypothalamic suprachiasmatic nucleus (SCN). The SCN receives information about the time of day from the light intensity measured by the retina of the eye and controls, among others, the melatonin secretion in the pineal gland of the brain (Van Veen et al., 2010). In an open study, bright light therapy decreased the ADHD symptoms in patients with SAD and ADHD (Rybak, McNeely, Mackenzie, Jain, & Levitan, 2006). This notion, together with the extreme high prevalence of sleep disorders among patients with ADHD, led us to hypothesize an overlap in the neurobiological origin of ADHD, SAD, and DSPS, suggesting that delayed sleep, seasonal depressive symptoms, and severity of ADHD symptoms are associated, even among people without diagnosed ADHD. To understand associations between the circadian rhythm disorders and sleeping problems that are associated with ADHD and comorbid SAD, we obtained self-reported sleep/wake characteristics of ADHD patients and controls. We explored the association between general characteristics, lifestyle factors, physical and mental health, and sleep characteristics. We were especially interested in associations between sleep-onset latency (the time to fall asleep after lights out), bedtime, time of midsleep, seasonal depressive symptoms, BMI, and ADHD symptoms. We determined these associations among the patients and the control group to gain insight into its generalizability. This may help us to understand the impact of circadian rhythm disorders on health consequences on the long term and eventually to gain insight into appropriate prevention or treatment.

263

Bijlenga et al.

Method Participants A total of 202 patients with ADHD between 18 and 65 years of age, diagnosed by a trained psychologist or psychiatrist with childhood-onset and persistent ADHD between 2008 and 2009, were recruited from the PsyQ outpatient clinic, Program Adult ADHD, in The Hague, the Netherlands. The control group consisted of 75 adult participants who were recruited by the researchers in various locations such as public libraries and municipal buildings. The control group also included 114 students attending a 3rd-year bachelor course in Child Studies at Leiden University, The Netherlands, and one of their acquaintances above the age of 30 years to a maximum of 65 years of age. Participants were not informed about the objective of the study and participation in the study was voluntary.

Procedure The patients were diagnostically assessed for ADHD using the semistructured diagnostic interview for ADHD in adults (DIVA; Kooij, 2010). A diagnosis of ADHD required participants to have experienced at least six out of nine Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; APA, 1994) symptoms of inattention and/or hyperactivity/impulsivity in childhood, a chronic persisting course of symptoms and impairment, and at least four out of nine DSM-IV symptoms of either or both domains in adulthood (Kooij, 2010). These diagnostic criteria were in accordance with the literature and with the proposed cutoff of the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-V; http://www.dsm5.org/; Barkley, 1997; Kooij et al., 2005). In addition, a standard checklist was used to systematically assess psychiatric comorbidity as well as physical diseases. During the initial consultation for diagnostic assessment of ADHD and comorbidity, the patients were invited to fill out the ASESA questionnaire by pen and paper. ASESA is the Dutch acronym for Inattention, Sleep, Eating pattern, Mood, and General health questionnaire. The controls were given oral and written information about the study by the researchers. Controls filled out the informed consent and gave their email addresses. The bachelor students in the control group were recruited by email invitation from their lecturers (K.B.H., H.J.S.B.) and were asked to forward the email to their acquaintance so he or she could enter the online questionnaire as well. The researchers emailed the link to the online questionnaire to all recruited controls. The controls filled out the identical ASESA questionnaire, however implemented for internet assessment using a free open-source web-based survey tool (http://www .limesurvey.org).

Measures The ASESA questionnaire was in Dutch and involved background characteristics and questions about current lifestyle, such as daily activities, working hours, number of hours of daylight, use of light at night, nicotine, caffeine, alcohol and drugs use, use of medication, eating pattern, and physical and psychological morbidities and complaints. An unstable eating pattern was defined as an eating pattern in which meals are often skipped and/or in case of regular binge meals. The ASESA also contained the following measures: the validated ADHD Rating Scale (ADHD-RS; DuPaul, Power, Anastopoulos, & Reid, 1998; Kooij et al., 2005), a Dutch questionnaire based on the validated Horne and Östberg Morningness–Eveningness Questionnaire (“Vragenlijst Ochtend/Avondmens”; VOA; Horne & Ostberg, 1976; Kerkhof, 1984; Natale, Esposito, Martoni, & Fabbri, 2006), the modestly validated Munich Chronotype Questionnaire (MCTQ; Zavada, Gordijn, Beersma, Daan, & Roenneberg, 2005), and the Seasonal Pattern Assessment Questionnaire–Global Seasonality Score (SPAQ-GSS; Mersch, Middendorp, Bouhuys, Beersma, & van den Hoofdakker, 1999; Mersch et al., 2004). The ADHD-RS is a self-evaluation of the DSM-IV ADHD symptoms hyperactivity and inattention that are scored for adulthood and childhood separately, each involving 23 items. The VOA is a 7-item survey that was used to determine the participants’ chronotype on a 5-point scale varying from extreme early chronotype to extreme late chronotype. The MCTQ involves 32 items on bedtime, rise time, sleep latency, daytime naps, and light exposure. It is scored for both workdays and free days, and is used to determine sleep duration, midsleep, and chronotype. The midsleep on free days, corrected for sleep debt on workdays (MSFsc) is a standardized measure which is predominantly used in chronobiology as a parameter for chronotype classification. The later one’s chronotype the later his midsleep, and vice versa. The MSFsc was calculated from the MCTQ following Roenneberg et al. (2004). From the MCTQ, we also calculated sleep length, defined as the difference between bedtime and wake up time subtracted by sleeponset latency (Allebrandt et al., 2010). A variable length of sleep onset was defined when participants reported their sleep-onset latency to be different everyday. The SPAQGSS is a 6-item survey on seasonal changes in sleep length, social activities, mood, appetite, weight, and energy. Seasonal depressive mood was determined using the SPAQGSS score, which was based on the summed scores of the subscales that were each rated 0 (no change during the seasons) to 4 (extreme change during the seasons). A cutoff score of 11 out of 24 is needed for an indication of SAD (Mersch et al., 1999; Mersch et al., 2004). We defined a composed measure of an indication of DSPS, which consisted of a sleep-onset latency of 30 min or more in combination

264 with a bedtime of 00:30 a.m. or later on workdays, or an “extreme evening type” score on the VOA. The completion time of the total ASESA questionnaire was 20 to 30 min.

Analysis We compared sleep parameters between patients and controls. Because chronotype is age dependent (Roenneberg et al., 2007), we analyzed the sleep parameters within two age groups: ≤30 years and >30 years. Sleep parameters were checked for normality using the normal probability plot and compared between patients and controls using ANOVA for normal distributions and Mann–Whitney for non-normal distributions. To determine explained variances (β) of sleeping problems, we used linear regression with MSFsc as the dependent value, and background characteristics, ADHD characteristics, and lifestyle factors as independent variables. We applied two multinomial logistic regressions with SPAQ-GSS indication for SAD and short versus long sleep-onset latency as dependent values to determine associations between background characteristics, ADHD symptoms, lifestyle factors, and sleep/wake characteristics. We calculated relative risks (RRs) and odds ratios (ORs) of ADHD characteristics, lifestyle factors, and sleep/ wake characteristics on SPAQ-GSS score within the patient and control groups. We calculated correlations between BMI and sleep duration and also between BMI and sleeponset latency for both groups separately. Using binary logistic regressions we calculated the change in OR of various diseases per increased unit of BMI, with correction for age. We used SPSS 18.0 (Chicago, Illinois) for data analysis. An alpha level of ≤.05 was used for statistical significance.

Results General Characteristics We included 391 adult participants, of which 202 were patients and 189 were controls. General characteristics are listed in Table 1. The patients (47% females) were between 18 and 64 years old (M = 34.9 years; SD = 10.6). The controls (65% female) were between 17 and 65 years old (M = 33.0 years; SD = 13.6). Clinical diagnoses according to the DSM-IV criteria of the patients were 83.2% ADHD– combined subtype, 16.3% ADHD–inattentive subtype, and 0.5% ADHD–hyperactive/impulsive subtype. As stated in Table 1, 76% of the patients had at least one comorbid disorder. The most frequent diagnosed comorbid disorders were depression (33%), anxiety disorder (21%), and sleep disorder (21%). In the group of controls, 4.3% met the criteria for ADHD based on their self-reported ADHD-RS

Journal of Attention Disorders 17(3) scores. The ADHD patients reported more mental and physical problems than the controls (Table 1).

Validated Measures In Table 2, the ADHD-RS scores, SPAQ-GSS score, VOA chronotype, and sleep/wake characteristics are compared between the patients and controls and between the (≤30 and >30) age groups. The patients were more often evening chronotypes (both age groups p < .001) and more extreme evening chronotypes (≤30 years, p < .001; >30 years, p = .003). The age groups did not differ in respect to SPAQ-GSS indication for SAD.

Sleep/Wake Characteristics There were differences between patients and controls on all sleep parameters, with less desirable values in the patient group (see Table 2). We also found that 25.8% of the patients and 2.1% of the controls (p < .001) met the criteria for the indication of DSPS. Within the patient group, the participants with the indication of DSPS, compared with the patients without indication of DSPS, more often skipped meals (89.7% vs. 64.9%; p = .003). Within the control group, the participants with indication of DSPS more often had a high score (≥23) on the ADHD-RS (50.0% vs. 7.1%; p = .002), more often three or more hyperactivity/impulsivity symptoms on the ADHD-RS (25.0% vs. 3.8%; p = .038). However, we did not find significant correlations between indication of DSPS and obesity, physical complaints, smoking, or indication for SAD in either group. Concerning light at night, we found a correlation between the duration of watching TV after 11.30 p.m. and the sleep-onset latency (ρ = 0.225; p = .002).

Linear Regressions Table 3 shows the multivariate linear regression for the impact of background variables, ADHD-RS score, seasonal depressive symptoms, use of medication, and lifestyle factors on midsleep and sleep length. Factors that had significant impact on midsleep were an ADHD diagnosis (β = 20.53; p = .048), having more than three hyperactive/impulsive symptoms on the ADHD-RS in adulthood (β = 0.94; p = .029), female gender (β = −37.61; p < .001), age (per year, β = −2.18; p < .001), and an indication for SAD (β = 21.73; p = .023). Unemployment and an unstable eating pattern had negative impact on midsleep but did not reach statistical significance. Factors that had significant impact on sleep length were ADHD diagnosis (β = 41.79; p = .002), age (per year, β = −2.04; p < .001), and unstable eating pattern (β = −22.66;

265

Bijlenga et al.

Table 1. General Characteristics of the Group of Adults With ADHD (Patients) and Healthy Adults (Controls) With Tests for Statistical Differences Between Groups Using Chi-Square for Binary Data, Mann–Whitney for Ordinal Data, and t test for Interval and Ratio Data (N = 391). Characteristics

Patients (n = 202)

Female, n (%) 95 (47.0) Age in years; M (SD) 34.9 (10.6) Diagnosed DSM-IV ADHD subtype  Hyperactive/impulsive, n (%) 1 (0.5)  Inattentive, n (%) 33 (16.3)  Combined, n (%) 168 (83.2) Diagnosed DSM-IV current psychiatric comorbidity at time of assessment   Depressive disorder, n (%) 66 (32.8)   Seasonal affective disorder (SAD), n (%) 16 (8.0)   Anxiety/panic disorder, n (%) 42 (20.9)   Posttraumatic stress disorder (PTSD), n (%) 12 (6.0)   Personality disorder, n (%) 23 (11.4)   Sleep disorder, n (%) 43 (21.4)   Eating disorder, n (%) 10 (5.0) 46 (22.9)   Alcohol or drug use disorder,a n (%) Self-reported current psychiatric comorbid symptoms at time of assessment   (Seasonal) depressive symptoms, n (%) 37 (18.3)   Anxiety/panic symptoms, n (%) 8 (4.0)   Stress/burnout/chronic fatigue, n (%) 11 (5.4) Self-reported lifetime morbidity   Respiratory disorder, n (%) 63 (31.2)   Cardiovascular disorder, n (%) 86 (42.6)   Nervous system disorder, n (%) 88 (43.6)   Digestive system disorder, n (%) 66 (32.7)   Urinary system disorder, n (%) 26 (12.9)   Metabolic disorder, n (%) 25 (12.4)   Immune system disorder, n (%) 14 (6.9)   Skin disorder, n (%) 90 (44.6)   Skeletal disorder, n (%) 101 (50.0)  Cancer, n (%) 5 (2.5)   Injury after accident, n (%) 48 (23.8) BMI (n = 386); M (SD) range 24.8 (4.9) 17-45  BMI < 18.5 (underweight), n (%) 8 (4.1)  BMI ≥ 25 (overweight), n (%) 93 (47.2)  BMI ≥ 30 (obese), n (%) 33 (16.8)  BMI ≥ 40 (morbidly obese), n (%) 2 (1.0) Vocational status (n = 388)   Elementary school, n (%) 24 (12.0)   Lower vocational training/lower secondary education, n (%) 60 (30.0)   Higher secondary school/intermediate vocational training, n (%) 56 (28.0)   Preuniversity education, n (%) 13 (6.5)   Higher professional school, n (%) 31 (15.5)  University, n (%) 16 (8.0) Main occupation (n = 387)  Employed, n (%) 104 (52.3)  Unemployed, n (%) 54 (27.1)  Student, n (%) 39 (19.6)  Retired, n (%) 2 (1.0)

Controls (n = 189)

p

123 (65.1) 33.0 (13.6)