Perfect Enough to Sleep? Perfectionism and Actigraphy-Determined Markers of Insomnia

ABSTRACT

Objectives

Perfectionism is an important factor in insomnia development and maintenance. Previous studies exploring the relationship between perfectionism and insomnia have predominantly relied on self-reported sleep measures. Therefore, this study sought to assess whether actigraphy-measured sleep parameters were associated with perfectionism.

Methods

Sixty adults (85% females, mean age 30.18 ± 11.01 years) were sampled from the Australian general population. Actigraphy-derived objective sleep measures, subjective sleep diary measures, the Frost Multidimensional Perfectionism Scale (FMPS), Hewitt-Flett Multidimensional Perfectionism Scale (HFMPS) and Depression, Anxiety and Stress Scale 21 (DASS-21) were collected.

Results

High perfectionism levels were associated with poor sleep, but these relationships differed between objective and subjective measures. Perfectionism via FMPS total score and subscales of Concern over Mistakes, Doubts about Actions, Personal Standards and Self-oriented Perfectionism correlated with subjective sleep onset latency and sleep efficiency with moderate effects (r = .26 to .88). In contrast, perfectionism via HFMPS total score and subscales of Socially Prescribed Perfectionism and Parental Expectations predicted objective sleep onset latency and sleep efficiency. Additionally, stress mediated the relationships between objective sleep efficiency and Concern over Mistakes and Doubts about Actions.

Conclusions

Perfectionism demonstrated stronger associations with subjective than objective sleep measures. Higher Parental Expectations and Socially Prescribed Perfectionism may increase one’s vulnerability to objectively measured poor sleep. Therefore, perfectionism may be important in preventing and treating insomnia.

Introduction

Insomnia is a prevalent and significant sleep disorder. Insomnia and inadequate sleep affect almost 15% and 45% of Australians respectively (Adams et al., 2017; Reynolds et al., 2019). Sleep problems can affect mood, memory, energy level and general wellbeing, leading to increased use of healthcare services, higher personal and economic costs (Attard et al., 2019; Bin et al., 2012; Deloitte Access Economics, 2017). Insomnia symptoms include difficulty initiating and/or maintaining sleep, and/or early morning awakening despite adequate sleep opportunity. Insomnia symptoms are listed as a diagnostic criterion for a variety of disorders (e.g., depression, post-traumatic stress disorder) in the Diagnostic and Statistical Manual of Mental Disorders; however, insomnia can also be classified as a sleep-wake disorder in its own right (American Psychiatric Association, 2022). Furthermore, insomnia is identified as a transdiagnostic factor in mental health given its bidirectional relationship and high comorbidity with disorders (e.g., depression, anxiety, bipolar disorder, alcohol use disorder, post-traumatic stress disorder, schizophrenia and psychosis), contributing to their onset, maintenance, and relapse (Dolsen et al., 2014; Harvey, 2008; Hertenstein et al., 2019).

Perfectionism is an important factor in various etiological models of insomnia that propose cognitive, emotional, and physiological arousals disrupt the automatic process of sleep onset (Espie et al., 2006; Harvey, 2002; Harvey et al., 2014; Riemann et al., 2010; van de Laar et al., 2010). Perfectionism, a multidimensional construct defined by excessive high standards and self-criticism, is commonly measured by the Frost Multidimensional Perfectionism Scale (FMPS, Frost et al., 1990) and/or the Hewitt-Flett Multidimensional Perfectionism Scale (HFMPS, Hewitt & Flett, 1991). The FMPS measures six dimensions of perfectionism: Concern over Mistakes (reacting negatively to and viewing mistakes as failures), Doubts about Actions (doubting one’s own performance), Parental Expectations (perceiving parents as setting very high goals), Parental Criticism (perceiving parents as being highly critical), Personal Standards (setting high standards for self-evaluation) and Organisation (preferring order and organization). The HFMPS measures three dimensions of perfectionism: Self-oriented Perfectionism (setting high standards for oneself), Socially Prescribed Perfectionism (perceiving others set high standards for oneself), and Other-oriented Perfectionism (setting high standards for others).

Evidence for cross-sectional and longitudinal relationships between insomnia and dimensions of perfectionism is presented in Table 1. Notably, there is a paucity of research in this area in Australia. Lundh et al. (1994) found that Concern over Mistakes, Doubts about Actions and Personal Standards were higher in patients with insomnia, and were positively correlated with sleep problems in a general population sample. The authors theorized that perfectionism amplifies the negative mood and performance worries caused by common negative life events, which increases the risk of transient sleep problems. The impact of transient sleep problems on daytime performance subsequently becomes another source of worry. Subsequent rumination then drives dysfunctional sleep beliefs and perpetuates sleep initiation and maintenance difficulties in a vicious cycle that often results in chronic insomnia (Lundh & Broman, 2000). As a transdiagnostic process in the development and maintenance of psychopathology, treating perfectionism may be clinically relevant to improving the efficacy of cognitive-behavioral therapy for insomnia (CBT-I) in individuals who have insomnia and are highly perfectionistic (Akram et al., 2023; Egan et al., 2011; Johann et al., 2018; Limburg et al., 2017; Rodriguez-Seijas et al., 2015). Perfectionism may therefore play an important role in the sleep interpretation process relevant to both insomnia development and treatment. Furthermore, this relationship may be mediated by factors such as depression and/or anxiety, arousability, stress perception, emotional regulation, and counterfactual thoughts (Table 1).

Table 1. Overview of studies examining perfectionism and insomnia and/or sleep difficulties using subjective sleep measures.

Study	Sample	Scale used	Key findings
Akram et al. (2020)	624 community adults and students in the United Kingdom	FMPS	DA, PC and PE: Positive association with and predict insomnia but mediated by anxiety and dysfunctional beliefs. O: Negative association and predicts insomnia.
Akram et al. (2017)	39 adults with insomnia and 39 normal sleepers in the United Kingdom	FMPS HFMPS	CM and DA: Insomnia group score higher than normal sleepers. Predict insomnia but mediated by anxiety. PC: Insomnia group score higher than normal sleepers. Predict insomnia, but partially mediated by anxiety.
Akram et al. (2015)	76 community adults in the United Kingdom	FMPS HFMPS	DA and PC: Positive association with insomnia but mediated by depression and anxiety. Baseline insomnia predicts future DA but mediated by baseline anxiety and current insomnia. Baseline insomnia predicts future PC.
Andersson et al. (2005)	256 tinnitus patients in Sweden	FMPS	CM and PC: Positive association with ISI.
Azevedo et al. (2010)	870 university students in Portugal	HFMPS	SPP: Insomnia group score higher than good sleepers group score at T1 and T2. Baseline SPP predicts difficulties initiating sleep and difficulties maintaining sleep at T1 and T2.
Azevedo et al. (2009)	1163 university students in Portugal	HFMPS	HFMPS, SOP and SPP: Positive association with difficulties initiating sleep, difficulties maintaining sleep and sleep disturbance index.
Brand et al. (2015)	346 university students in Switzerland	FMPS	CM: Positive association with ISI, SOL and NOA but mediated by stress and emotional regulation. DA and PS: Positive association with ISI and NOA but mediated by stress and emotional regulation.
Jansson-Frojmark and Linton (2007)	1936 community adults in Sweden	FMPS: CM and PS only	CM: Positive association with pre-existing insomnia and future insomnia but mediated by emotional distress.
Lombardo et al. (2013)	819 community adults in Italy	FMPS	CM, DA, PC and O: Positive association with current and future insomnia.
Lundh et al. (1994)	383 community adults and 70 insomnia patients in Sweden	FMPS: Only CM, DA and PS	CM and DA: Positive association with sleeping problems.
Maia et al. (2011)	870 university students in Portugal	HFMPS	SOP: Inconsistent pattern found with small correlations. SPP: Positive association with difficulties initiating sleep and difficulties maintaining sleep across T1, T2 and T3.
Molnar et al. (2020)	335 university students in Canada 296 community adults in America	HFMPS	SOP: Positive associations with good sleep quality. SPP: Positive association with poor sleep quality.
Richardson and Gradisar (2020)	281 adolescents in Australia	Child Adolescent Perfectionism Scale	SOP: Positive association with insomnia but mediated by vulnerability to stress. SPP: Positive association with insomnia.
Schmidt et al. (2018)	180 university students in Switzerland	FMPS	CM and DA: Positive association with and predict insomnia severity but mediated by counterfactual processing. O: Negative association with and predicts insomnia severity but mediated by counterfactual processing.
Trudel-Fitzgerald et al. (2017)	853 cancer patients in Canada	FMPS	CM and DA: Positive association with insomnia in men. PC: Positive association with insomnia in women and men but mediated by arousability.
Vincent and Walker (2000)	32 community adults with insomnia and 26 as healthy controls in Canada	FMPS HFMPS	CM and DA: Clinical group score higher than healthy control group score. PC: Clinical group score higher than healthy control group score. Positive association with SOL in clinical group. PC: Positive association with sleep quality in healthy control group.

ISI, Insomnia Severity Index; TST, total sleep time; SOL, sleep onset latency; WASO, wake after sleep onset; NOA, number of awakenings; FMPS, Frost Multidimensional Perfectionism Scale; CM, Concern over Mistakes; DA, Doubts about Actions; PE, Parental Expectations; PC, Parental Criticism; O, Organisation; PS, Personal Standards; HFMPS, Hewitt-Flett Multidimensional Perfectionism Scale; SOP, Self-oriented Perfectionism; OOP, Other-oriented Perfectionism; SPP, Socially Prescribed Perfectionism.

Most studies examining the relationship between perfectionism and insomnia rely on self-reported sleep measures. Notable exceptions include three studies that assessed perfectionism and insomnia measures using polysomnography. Using punctuality as a proxy for perfectionism in a sample of clinical patients, Spiegelhalder et al. (2012) found that shorter total sleep time, lower sleep efficiency and higher wake after sleep onset were associated with arriving early at a sleep laboratory. Johann et al. (2017) found significant associations between perfectionism (e.g., FMPS-total, Concern over Mistakes and Personal Standards) and markers of poor sleep (e.g., nocturnal awakenings, arousal index; and wake after sleep onset) in the first but not second sleep laboratory night in a sample of clinical patients. The authors explained the unpleasantness of spending the first night in a sleep laboratory as an acute stressor and suggested perfectionistic individuals may be predisposed to developing insomnia when facing acute stressors. In contrast, a recent study by Johann et al. (2023) found FMPS-total, Concern over Mistakes and Personal Standards to be negatively associated with wake after sleep onset, while total sleep time and sleep efficiency were positively associated with Personal Standards. These findings indicate that perfectionism may play a role in the etiology of objectively measured sleep disturbances, but further investigation with objective sleep measures is required.

Actigraphy allows for the estimation of sleep in participants’ natural sleep environment for extended periods of time and is an economical alternative to polysomnography (Martin & Hakim, 2011). Actigraphy-derived longitudinal objective sleep measures may aid in understanding of the relationship between perfectionism and parameters of insomnia and sleep continuity, and may have clinical implications since actigraphy is recommended for insomnia monitoring and treatment evaluation (Akram, 2018; Liguori et al., 2023; Smith et al., 2018). It is currently unknown whether it is the perception of poor sleep or actual physiological and physical markers of poor sleep that are related to perfectionism. It is also unknown if and how the perception of poor sleep progresses into physical markers of poor sleep in the development of insomnia in highly perfectionistic individuals. To our knowledge, no actigraphy study in this research area exists despite potential theoretical and clinical significance.

The aim of this study is to investigate the association between perfectionism and sleep disturbances by: 1) replicating and extending previous findings between perfectionism and subjective sleep measures with an Australian general population sample, 2) exploring if relationships between each dimension of perfectionism and subjective sleep are replicated using actigraphy-derived objective sleep measures, and 3) exploring if depression, anxiety and stress mediate the relationships between perfectionism and actigraphy-derived objective sleep measures. It is hypothesized that high levels of perfectionism will be associated with subjective measures of sleep. No a priori hypothesis is made relating to objective sleep measures given limited past research findings.

Materials and methods

Participants

Adult participants, aged 18 years and above, were recruited from the Research Experience Program (REP) at the University of Melbourne, via social media and word-of-mouth or referred by their treating sleep physician at the Melbourne Sleep Disorder Centre with diagnosis of insomnia. REP participants were undergraduate psychology students, who received course credit for participation and self-selected to the study in response to an online advertisement at the University of Melbourne, Australia. All potential participants regardless of recruitment pathway completed an online screening survey via Qualtrics (https://www.qualtrics.com) to determine eligibility. Exclusion criteria include: (a) current diagnosis of mental illness; (b) central nervous disorder (c) prior head injury; (d) shift work, defined as paid work starting before 7:00AM or after 2:00PM, rotating or including work outside of 7:00AM to 6:00PM; (e) symptoms indicative of sleep disorders other than insomnia (e.g., obstructive sleep apnea, narcolepsy). In this study, no clinical insomnia diagnosis was required, and participants included both clinical and general population participants who were assessed for their insomnia symptomology.

A total of 1132 screening responses were obtained. There were 243 eligible participants after removing incomplete and duplicate responses, and those who declined to participate. Sixty-five participants agreed to participate in the two-week actigraphy sleep component. A final sample of 60 participants were included in this study, after exclusions of participants with incomplete questionnaires and/or actigraphy recording errors.

All participants provided informed written consent online and received an online gift voucher worth $25 upon completion of the actigraphy sleep component. Approval for this research study was granted by the Human Research Ethics Committee at the University of Melbourne.

Sleep measures

Throughout the 14-day actigraphy period, participants completed the Pittsburgh Sleep Diary (PSD; Monk et al., 1994), which was delivered daily to participants’ smartphones via the Smartphone Ecological Momentary Assessment app (SEMA3, Koval et al., 2019). The PSD captured self-reported daily bedtime, lights out time, minutes to sleep, waketime, etc. Subjective total sleep time (TST), sleep onset latency (SOL), wake after sleep onset (WASO), number of awakenings (NOA) and sleep efficiency (SE) were calculated from diary data. Objective daily sleep was assessed using a wrist worn GENEActiv accelerometer (Activinsights, Cambridge, United Kingdom). The accelerometer was set to a sample frequency of 40 Hz. Accelerometer data was extracted using GENEActiv PC Software (version 3.3) and processed with GGIR (version 2.6.0) using R (version 4.0.5). Sleep recordings were extracted using GGIR parameters set to T3A10 where a period of sustained inactivity, as measured by a lack of change in arm angle above 10-degrees for at least 3-minutes, was detected as sleep. This setting was chosen based on GGIR’s validation study against polysomnography showing that shorter time window and higher angle threshold provided better sensitivity in detecting sleep when compared to its default setting (van Hees et al., 2015).

Sleep recordings were evaluated to derive objective WASO, TST, NOA and SE. Objective SOL was derived using a combination of subjective information supplied by participants’ sleep diaries (e.g., lights out time) and objective data extracted from GENEActiv. While data collection was scheduled for 14 days, some participants had missing or unusable data. Therefore, the average score for each variable for each participant was calculated over the number of nights where usable sleep data were recorded. Participants also completed the Insomnia Severity Index (ISI) at the end of the two-week actigraphy sleep study (Bastien et al., 2001; Morin, 1993). The ISI is a self-reported questionnaire that measures the nature, severity, and impact of insomnia over the past fortnight. It consists of seven items that are rated on a 5-point Likert scale from 0 (no problem) to 4 (very severe problem) with a total score range of 0 to 28. These items assess sleep onset and maintenance difficulties, early morning awakenings, sleep satisfaction, and distress, daytime dysfunction and noticeability by others resulting from sleep problem. The ISI has excellent internal consistency with Cronbach’s alphas of .90 for community sample and .91 for clinical sample, and an optimal cutoff score of 10 for detecting insomnia in community sample with sensitivity at 86.1% and specificity at 87.7% (Morin et al., 2011).

Assessment

To assess the different dimensions of perfectionism, both the Frost Multidimensional Perfectionism Scale (FMPS; Frost et al., 1990) and Hewitt-Flett Multidimensional Perfectionism Scale (HFMPS; Hewitt & Flett, 1991) were utilized. The FMPS consists of 35 items, each rated on a Likert scale from 1 (disagree strongly) to 5 (agree strongly). There are six subscales in the FMPS, each with maximum scores as follows: Concern over Mistakes 45, Doubts about Actions 20, Parental Expectations 25, Parental Criticism 20, Personal Standards 35 and Organisation 30. Total FMPS score ranges from 29 to 145 as it excludes the Organisation subscale which comprises 6 items. Higher levels of perfectionism are reflected by higher scores. Internal consistency for each subscale of the FMPS was good with Cronbach’s alphas ranging .77 to .93 and reliability coefficients ranging .77 to .94 (Frost et al., 1990).

The HFMPS consists of 45 items, each rated on a Likert scale from 1 (disagree) to 7 (agree). There are three subscales in the HFMPS: Self-oriented Perfectionism, Other-oriented Perfectionism and Socially Prescribed Perfectionism. The maximum total score of each subscale is 105 with higher scores representing higher levels of perfectionism. Internal consistency for each subscale of the HFMPS was good with Cronbach’s alphas ranging from .74 to .88 (Hewitt et al., 1991).

Participants also completed the Depression, Anxiety and Stress Scale 21 (DASS-21; Lovibond & Lovibond, 1995). The DASS-21 consists of three scales with each comprising seven items measuring levels of Depression, Anxiety, and Stress and rated on a 4-point Likert-type scale from 0 (Did not apply to me at all) to 3 (Applied to me very much or most of the time). Internal consistency is high with Cronbach’s alphas reported as: .91 for Depression, .80 for Anxiety and .84 for Stress (Sinclair et al., 2012).

Study protocol

Prior to the start of the actigraphy component, participants attended a consent meeting with the lead researcher (S.O.) either in-person or online via Zoom (https://zoom.us). During this meeting, explanations about the actigraphy watch and study protocol were provided, and participants completed the FMPS and HFMPS. Over the 14-day actigraphy study period, participants wore the accelerometer on their non-dominant wrist 24 hours a day, but removal was permitted during contact sports, swimming or if required by work. Actigraphy watches were distributed during the in-person consent meeting, or via post prior to the online consent meeting.

Participants also completed the PSD daily when prompted by SEMA3, which sent out reminders via notifications within each data collection input timeframes. At the end of the 14-day actigraphy study period, participants attended a debrief to return the actigraphy watch, complete the DASS-21 and ISI, and received remuneration.

Statistical analysis

The statistical software package R (version 4.3.0; http://www.R-project.org/) was used for all analyses. The relationships between perfectionism and sleep were assessed with correlation, regression, and mediation analyses. Firstly, the associations between perfectionism and ISI, subjective and objective sleep parameters were analysed using Pearson correlations, which is consistent with past research (Akram et al., 2015; Brand et al., 2015; Lundh et al., 1994; Schmidt et al., 2018). Correlations of .10 indicate small effect, .30 medium effect and .50 large effect (Cohen, 1988).

Prior to further analysis, SOL, WASO and NOA were log transformed due to data skewness. Since SE was a proportion between 0 and 1, a logit transform was applied to stabilize the variance of residuals (Warton & Hui, 2011). A series of regression models were then conducted for each combination of perfectionism variable (FMPS-total and HFMPS-total and subscale scores as independent variable) and actigraphy derived objective sleep parameter (TST, logSOL, logWASO, logNOA and logitSE as dependent variable) with age and gender as covariates.

Finally, using Depression, Anxiety, and Stress scores obtained from DASS-21 as mediators, a series of mediation analyses were conducted for each selected combination of perfectionism subscale (Concern over Mistakes, Doubts about Actions, Parental Criticism, Personal Standards and Socially Prescribed Perfectionism) and actigraphy derived objective sleep parameters. Notably, all mediation analyses were conducted even in the absence of a significant total effect as per O’Rourke and MacKinnon (2018).

The alpha level for the two-tailed analyses was set at p < .05. Model assumptions and effects, which included and excluded outliers, were visually inspected and compared. Overall, model assumptions were met and there was no meaningful change in model effects with or without outliers, so outliers were included in the current analyses (see Tables A3 to A10 in the Appendix for models excluding outliers).

Results

Descriptive data

The final sample included 60 participants aged 18–66 years (M = 30.18, SD = 11.01), of whom 85.0% identified as female. Participants reported ISI scores ranging from 0 to 25 (M = 7.48, SD = 5.92). Just over a third of the sample (n = 23, 38.3%) were categorised as a “poor sleeper”, with an ISI score exceeding 9 (Morin et al., 2011). Mean and standard deviations of subjective and objective sleep measures, and perfectionism and DASS-21 are presented in Tables 2 and 3 respectively. As shown in Table 2, there were significant differences between all subjective and objective sleep measures, except for NOA.

Table 2. Mean scores and standard deviations for objective and subjective sleep measures.

	Objective		Subjective		Paired t-test		Effect size
	M	SD	M	SD	t	p	Cohen’s d
TST, mins	424.30	72.65	447.90	59.43	−2.58	.012*	−0.35 (small)
SOL, mins	7.36	8.23	22.97	21.74	−5.24	<.001**	−0.95 (large)
WASO, mins	49.02	34.01	18.65	22.69	5.50	<.001**	1.05 (large)
NOA, no.	1.25	1.48	1.43	0.97	−0.72	.474	−0.14 (negligible)
SE, %	87.52	9.50	91.49	7.45	−2.51	.015*	−0.46 (small)

TST, total sleep time; SOL, sleep onset latency; WASO, wake after sleep onset; NOA, number of awakenings; SE, sleep efficiency. *p < .05. **p < .01.

Table 3. Mean scores, standard deviations, and sample range for measures of perfectionism, depression, anxiety, and stress.

	M	SD	Sample range (Score range)
FMPS	86.93	19.69	41–141 (29–145)
Concern over Mistakes	26.92	8.18	11–45 (9–45)
Doubts about Actions	11.87	3.77	5–20 (4–20)
Parental Criticism	9.37	4.04	4–20 (4–20)
Parental Expectation	13.20	4.54	6–25 (5–25)
Organisation	24.58	3.43	15–30 (6–30)
Personal Standards	25.58	5.65	9–35 (7–35)
HFMPS	177.20	29.85	100–238 (45–315)
Self-oriented Perfectionism	72.30	14.90	33–105 (15–105)
Other-oriented Perfectionism	52.48	13.36	23–81 (15–105)
Socially Prescribed Perfectionism	52.47	13.37	28–105 (15–105)
Depression	4.37	4.74	0–21 (0–21)
Anxiety	3.35	3.57	0–18 (0–21)
Stress	6.47	4.75	0–19 (0–21)

FMPS, Frost Multidimensional Perfectionism Scale; HFMPS, Hewitt-Flett Multidimensional Perfectionism Scale.

Correlation analyses between sleep measures and perfectionism

There were significant positive correlations of medium to large effects between ISI and all subscales of perfectionism, except for Parental Expectations, Organisation and Other-oriented Perfectionism (Cohen, 1988; see Table A1 in Appendix). Results in Tables 4 and 5 showed that the number and strength of significant correlations between subjective sleep and perfectionism measures were generally higher than those found between objective sleep and perfectionism measures. Specifically, significant positive correlations were found between subjective logSOL and FMPS-total (p = .006), Concern over Mistakes (p = .008), Doubts about Actions (p = .012), Personal Standards (p = .004) and Self-oriented Perfectionism (p = .020); and between objective logSOL and Parental Expectations (p = .029). Significant negative correlations were found between subjective logistSE and FMPS-total (p = .016), Personal Standards (p = .006) and Self-oriented Perfectionism (p = .042); and between objective logitSE and Parental Expectations (p = .047) and Socially Prescribed Perfectionism (p = .009). Overall, these relationships were considered to have medium effects (Cohen, 1988).

Table 4. Correlations between log transformed subjective sleep measures and perfectionism.

n = 60	L1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. TST #
2. logSOL	−.15*
3. logWASO	−.27	.50**
4. logNOA	−.04	.16	.68**
5. logitSE	.36**	−.87**	−.80**	−.31*
6. FMPS	−.08	.35**	.17	−.02	−.31*
7. CM	−.02	.34**	.11	.01	−.25	.88**
8. DA	−.09	.32*	−.02	−.17	−.21	.71**	.62**
9. PC	−.16	.15	.15	.12	−.18	.71**	.48**	.42**
10. PE	−.15	.04	.13	.00	−.12	.62**	.30*	.19	.72**
11. O	.03	.13	−.05	−.07	−.03	.09	.07	.08	−.16	.03
12. PS	.06	.37**	.22	−.04	−.35**	.74**	.62**	.44**	.20	.29*	.26*
13. HFMPS	−.07	.20	.14	.03	−.20	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.
14. SOP	.03	.30*	.13	−.03	−.26*	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	.76**
15. OOP	−.01	−.14	.03	.11	.08	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	.56**	.02
16. SPP	−.17	.25	.13	−.02	−.23	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	.83**	.57**	.22

log, log-transformed; TST, total sleep time; SOL, sleep onset latency; WASO, wake after sleep onset; NOA, number of awakenings; SE, sleep efficiency; FMPS, Frost Multidimensional Perfectionism Scale; CM, Concern over Mistakes; DA, Doubts about Actions; PE, Parental Expectations; PC, Parental Criticism; O, Organisation; PS, Personal Standards; HFMPS, Hewitt-Flett Multidimensional Perfectionism Scale; SOP, Self-oriented Perfectionism; OOP, Other-oriented Perfectionism; SPP, Socially Prescribed Perfectionism, N.A., not available. *p < .05. **p < .01. # TST was the only sleep variable that was not log-transformed.

Table 5. Correlations between log transformed objective sleep measures and perfectionism.

n = 60	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. TST #
2. logSOL	−.36**
3. logWASO	−.46**	.57**
4. logNOA	.-.37**	.43**	0.80**
5. logitSE	.61**	−.32**	−.80**	−.74**
6. FMPS	.04	.23	.14	.07	−.16
7. CM	.13	.13	.11	.04	−.08	.88**
8. DA	.15	.03	.19	.05	−.07	.71**	.62**
9. PC	−.23	.15	.14	.09	−.19	.71**	.48**	.42**
10. PE	−.24	.28*	.13	.08	−.26*	.62**	.30*	.19	.72**
11. O	.09	.25	−.07	−.09	.16	.09	.07	.08	−.16	.03
12. PS	.21	.25	.07	.02	−.04	.74**	.62**	.44**	.20	.29*	.26*
13. HFMPS	.01	.23	.18	.02	−.20	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.
14. SOP	.13	.25	.12	.01	−.07	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	.76**
15. OOP	−.01	.02	−.03	−.09	−.03	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	.56**	.02
16. SPP	−.12	.21	.28	.12	−.33**	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	N.A.	.83**	.57**	.22

log, log-transformed; TST, total sleep time; SOL, sleep onset latency; WASO, wake after sleep onset; NOA, number of awakenings; SE, sleep efficiency; FMPS, Frost Multidimensional Perfectionism Scale; CM, Concern over Mistakes; DA, Doubts about Actions; PE, Parental Expectations; PC, Parental Criticism; O, Organisation; PS, Personal Standards; HFMPS, Hewitt-Flett Multidimensional Perfectionism Scale; SOP, Self-oriented Perfectionism; OOP, Other-oriented Perfectionism; SPP, Socially Prescribed Perfectionism, N.A., not available. *p < .05. **p < .01. # TST was the only sleep variable that was not log-transformed.

Regression analyses of objective sleep measures and perfectionism

Regression analyses indicated that Parental Expectations predicted TST, logSOL and logit SE, while HFMPS-total predicted logSOL; and Socially Prescribed Perfectionism predicted logitSE as shown in Table 6. No other perfectionism scale or subscales were significant predictors of the various objective sleep measures.

Table 6. Regression coefficients between objective sleep measures and perfectionism.

n = 60	TST #			logSOL			logWASO			logNOA			LogitSE
	β	CI	p	exp β	CI	p	exp β	CI	p	expβ	CI	p	expβ	CI	p
FMPS	0.06	−0.92–1.05	.90	1.31	0.96–1.79	.09	1.06	0.91–1.23	.44	1.06	0.70–1.59	.78	0.88	0.72–1.08	.23
CM	1.07	−1.32–3.45	.37	1.19	0.86–1.64	.29	1.04	0.89–1.20	.63	1.01	0.66–1.52	.98	0.95	0.78–1.17	.66
DA	2.39	−2.72–7.50	.35	1.02	0.74–1.40	.91	1.03	0.89–1.19	.70	1.03	0.68–1.55	.89	0.95	0.77–1.16	.59
PC	−4.30	−0.88–0.28	.07	1.19	0.87–1.63	.26	1.07	0.93–1.24	.33	1.13	0.76–1.69	.53	0.87	0.71–1.05	.15
PE	−4.18	−8.27–-0.10	.045*	1.39	1.03–1.89	.033*	1.06	0.92–1.23	.43	1.10	0.73–1.64	.65	0.82	0.67–0.99	.043*
O	1.04	−4.66–6.74	.72	1.32	0.97–1.80	.07	0.95	0.82–1.10	.50	0.84	0.56–1.26	.39	1.09	0.89–1.34	.40
PS	2.48	−0.84–5.81	.14	1.32	0.97–1.80	.07	1.03	0.89–1.19	.69	1.00	0.67–1.50	.99	0.95	0.78–1.16	.63
HFMPS	0.05	−0.61–0.71	.88	1.38	1.01–1.88	.043*	1.09	0.94–1.26	.27	1.00	0.66–1.52	.98	0.89	0.73–1.09	.25
SOP	0.50	−0.79–1.79	.44	1.32	0.97–1.80	.08	1.05	0.91–1.22	.48	0.98	0.65–1.47	.90	0.94	0.77–1.15	.53
OOP	0.21	−1.26–1.68	.78	0.10	−0.23–0.42	.55	0.98	0.85–1.14	.82	0.88	0.58–1.33	.53	1.03	0.84–1.27	.75
SPP	−0.62	−2.09–0.85	.40	1.36	1.00–1.87	.05	1.16	1.00–1.34	.05	1.18	0.78–1.79	.42	0.80	0.65–0.97	.027*

exp, standardized exponentiated regression coefficient, for example, a 1.00 standard deviation increase in PE leads to a ratio change of 1.39 or a 39%increase in logSOL, and a 1.00 standard deviation increase in HFMPS leads to a ratio change of 1.38 or a 38% increase in logSOL. log, log-transformed; TST, total sleep time; SOL, sleep onset latency; WASO, wake after sleep onset; NOA, number of awakenings; SE, sleep efficiency; FMPS, Frost Multidimensional Perfectionism Scale; CM, Concern over Mistakes; DA, Doubts about Actions; PE, Parental Expectations; PC, Parental Criticism; O, Organisation; PS, Personal Standards; HFMPS, Hewitt-Flett Multidimensional Perfectionism Scale; SOP, Self-oriented Perfectionism; OOP, Other-oriented Perfectionism; SPP, Socially Prescribed Perfectionism. *p < .05. # TST was the only sleep variable that was not log-transformed.

Mediation analyses using depression, anxiety and stress as mediators in the relationships between objective sleep measures and selected measures of perfectionism

Results in the mediation analyses were broadly consistent with results from the regression analyses. None of the mediation analyses using Depression and Anxiety as mediators were significant (see Tables A5 and A6 in Appendix). Significant results presented in Table 7 indicated that while the total effect of Concern over Mistakes on logitSE was not significant, the mediation effect of stress on the relationship between Concern over Mistakes and logitSE was significant, such that an increase in Concern over Mistakes led to a decrease in sleep efficiency via stress as a mediator. A similar pattern was found in the relationship between Doubts about Actions and sleep efficiency. However, when outliers were removed, this result was no longer significant, suggesting they were driven by the outliers (for full results of mediation analyses using stress as a mediator with and without outliers, see Tables A7 and A12 in Appendix). Furthermore, considering the number of analyses involved in the current study, which could give rise to spurious findings, there was little evidence to support these relationships. Therefore, these results were likely inconclusive.

Table 7. Significant results from exponentiated regression coefficients of mediation analysis using stress as a mediator in the relationship between selected log transformed objective sleep measures and selected subscales of perfectionism (n = 60, simulations = 10,000).

Sleep measure	logitSE
perfectionism	Estimate	CI	P
Concern over Mistakes
- Average causal mediated effect	0.86	0.71–0.99	.036*
- Average direct effect	1.12	0.87–1.43	.40
- Total effect	0.95	0.77–1.20	.68
Doubts about Actions
- Average causal mediated effect	0.90	0.77–1.00	.042*
- Average direct effect	1.06	0.84–1.32	.61
- Total effect	0.95	0.76–1.19	.63

*p < .05. Only significant results presented above. For full results, see Table A9 in Appendix.

Discussion

This study explored the relationship between facets of multidimensional perfectionism and insomnia as measured by self-reported subjective and actigraphy-derived objective sleep measures. It also explored whether this relationship is mediated by depression, anxiety, and stress. Results showed individuals who self-reported higher levels of insomnia also reported higher levels of perfectionism. Specifically, those who reported longer sleep onset latency and lower sleep efficiency also tend to report higher total perfectionism (via FMPS-total), Concern over Mistakes, Doubts about Actions, Personal Standards and Self-oriented Perfectionism. These results support the hypothesis that higher level of perfectionism is associated with self-reported sleep measures of poor sleep. Notably, the relationship between perfectionism and insomnia differed between objective and subjective sleep measures. Participants who reported higher levels of Parental Expectations had longer objective sleep onset latency and those who reported higher levels of Parental Expectations and Socially Prescribed Perfectionism had lower objective sleep efficiency. Furthermore, Parental Expectations and total score on the HFMPS were significant predictors of objective sleep onset latency, while Parental Expectations and Socially Prescribed Perfectionism were significant predictors of objective sleep efficiency. Finally, there is inconclusive evidence to indicate that the associations between Concern over Mistakes and Doubts about Actions with objective sleep efficiency were mediated by stress. Overall, these results suggest certain aspects of perfectionism relate to markers of poor sleep, but some of those relationships differ across objective and subjective markers of poor sleep.

Subjective measures of sleep

Evidence for subjective measures of insomnia being related to perfectionism is further supported by this study with an Australian general population sample. The finding of relationships between subjective measures of insomnia and perfectionism dimensions of Concern over Mistakes, Doubts about Actions, Parental Criticism and Personal Standards is consistent with past literature (Akram et al., 2015, 2017, 2020; Andersson et al., 2005; Brand et al., 2015; Jansson-Frojmark & Linton, 2007; Lombardo et al., 2013; Lundh et al., 1994; Richardson & Gradisar, 2020; Trudel-Fitzgerald et al., 2017; Vincent & Walker, 2000). The current findings relating self-reported insomnia measures to Self-oriented Perfectionism and Socially Prescribed Perfectionism add evidence to support similar findings by Azevedo et al. (2010); Azevedo et al. (2009); Maia et al. (2011) and Molnar et al. (2020) but contradict findings by Akram et al. (2017); Akram et al. (2015) and Vincent and Walker (2000). Notably, these relationships were previously found in large samples (e.g., 335 students by Molnar et al. (2020)), yet they were significant in the current study with a sample size of 60. Overall, this study adds to evidence in the literature supporting theoretical suggestions that perfectionism plays an etiological role in the development of insomnia.

Objective measures of sleep derived from actigraphy

The current study extends the literature as it examined objective measures of sleep using actigraphy over two weeks. Results suggest Parental Expectations is associated with longer sleep onset latency and lower sleep efficiency, while Socially Prescribed Perfectionism is associated with lower sleep efficiency. These results are largely in line with findings by Johann et al. (2017) where higher levels of total perfectionism (via FMPS-total), Concern over Mistakes and Personal Standards were associated with higher levels of nocturnal awakenings and wake after sleep onset. However, the current results differ from the negative associations of total perfectionism (via FMPS-total), Concern over Mistakes and Personal Standards with wake after sleep onset; and positive associations of total sleep time and sleep efficiency with Personal Standards found by Johann et al. (2023). These discrepancies may be due to sampling and methodological differences. The current sample was assessed using actigraphy and comprised of members of the general population with or without insomnia, but without other sleep disorders. In contrast, Johann et al. (2017) sampled clinical patients of various sleep disorders and Johann et al. (2023) sampled clinical patients with insomnia disorder and both assessed sleep using polysomnography. While Johann et al. (2017) suggested that perfectionism may predispose one to objective sleep disturbances in the context of acute stressors, current findings suggest perfectionism may also be a contributing factor to objective sleep disturbances in the context of ongoing stressors arising from the perceived expectations from others.

Both Parental Expectations and Socially Prescribed Perfectionism relate to the perception of meeting others’ expectations of perfectionism, and Socially Prescribed Perfectionism has also been associated with Parental Expectations and Parental Criticism (Frost et al., 1990). Furthermore, the perception of these external demands of high standards in Socially Prescribed Perfectionism extends beyond demands from family and friends to those from acquaintances and society at large (Hewitt, 2020). One possible explanation for the current findings is that individuals with such perfectionistic tendencies attempt to meet those external sleep expectations by spending more time in bed, which inadvertently causes longer sleep onset latency and lower sleep efficiency to worsen sleep in a vicious cycle.

There have been recent calls to increase awareness about the importance of sleep and encourage ways to reduce inadequate sleep to improve sleep health in Australia (Adams et al., 2017; Attard et al., 2019; Australian Government, 2023; Deloitte Access Economics, 2017). However, media messaging about sleep is complex, often presenting good sleep as an ideal to strive for and is ironically, overly idealistic (Breheny et al., 2023). Furthermore, the proliferation of messages about sleep in the media may promote and perpetuate common dysfunctional sleep beliefs and myths (Katz & Lazersfeld, 1955; Robbins et al., 2019). Consequently, increasing awareness about the impact of inadequate sleep through public health messaging combined with a low level of sleep health education can be detrimental to those who have unrealistic expectations about needing to have “perfect” sleep as it can cause some level of distress leading to insomnia (Ekirch, 2015; Meaklim et al., 2020). The current findings suggest this process may be exacerbated in those who are highly perfectionistic, particularly those high in Socially Prescribed Perfectionism. This is because such individuals may be more vulnerable to the perceived demands at societal levels, particularly with excessive exposure to social media, which is increasingly a part of digital technologies employed in public health promotion (Flett et al., 2022; Stark et al., 2022).

In contrast to previous research (Akram et al., 2015, 2017, 2020; Jansson-Frojmark & Linton, 2007), neither depression nor anxiety mediated relationships between perfectionism and sleep in this study. These differences may be explained by the different measures used to assess depression and anxiety between this study and earlier studies. Furthermore, the exclusion criterion of a current mental illness diagnosis excluded participants with comorbid depression and anxiety disorders. Given the divergent severity in symptoms between general and clinical populations, this could have contributed to the lack of a mediation effect in the current study, which could indicate a conservative finding. Although the current evidence is inconclusive, the mediation effect of stress on the relationship between sleep difficulties and perfectionism, specifically in Concern about Mistakes and Doubts about Actions, suggests stress may play a role in the relationship between perfectionism and insomnia. This is in line with the finding by Brand et al. (2015) that perceived stress, as measured by the Perceived Stress Scale, may be an underlying psychological mechanism in the relationship between poor sleep and perfectionism. The current finding strengthens that line of thinking because the same relationships were found despite using different measures of stress and sleep.

Theoretical and clinical implications

Results from the current study support theoretical conceptualizations and empirical literature that perfectionism may increase one’s vulnerability to insomnia. Perfectionism is suggested to amplify common negative life events and cause arousals to disrupt the automatic sleep process, which becomes another source of stress and/or worry that leads to insomnia development (Akram et al., 2023; Lundh & Broman, 2000). The current study builds upon past research by replicating previous findings of the relationship between insomnia and perfectionism using subjective sleep measures obtained from an Australian general population sample, and adding to preliminary evidence that perfectionism may also play a role in the etiology of objectively measured sleep disturbances. Specifically, we suggest individuals who have high Socially Prescribed Perfectionism and Parental Expectations may be more prone to developing chronic insomnia as they increase sleep efforts to try to meet sleep expectations prescribed by others and perceived to be important, even if overly idealistic. Moreover, stress may be an underlying mechanism linking perfectionism to poor sleep such that individuals high on Concern for Mistakes and Doubts about Actions may be more prone to experiencing poor sleep when stressed, which then can lead to the development of dysfunctional sleep beliefs and unhelpful behaviors that cause insomnia (Espie et al., 2006; Morin et al., 1993).

Consequently, modifications to CBT-I to incorporate both perfectionism and stress reactivity as therapy targets should be considered in future trials. A recent CBT-I trial of clinical sample with perfectionism by Johann et al. (2023) indicated both perfectionism and insomnia were positively related to sleep efforts, pre-sleep arousal, and dysfunctional beliefs about sleep and that upon completion of CBT-I, levels of Organisation and Parental Expectations were elevated. Although the treatment was successful, the authors noted perfectionism may undermine sleep restriction therapy, a core component of CBT-I, because it can initially cause excessive daytime sleepiness, fatigue, and low mood and lead to dropouts. The current study findings suggest it may be important to address stress reactivity, Concern about Mistakes and Doubts about Actions prior to sleep restriction therapy to minimize the risk of dropout from CBT-I. Efficacy and impact of incorporating techniques from CBT for perfectionism into CBT-I could be explored in future research.

The current study findings also highlight the importance of improving sleep health education in Australia and taking a nuanced approach in public health messaging about sleep to prevent the proliferation of misperception, dysfunctional beliefs, and myths about sleep. Applying the suggestions relevant to public health messaging for Socially Prescribed Perfectionism by Flett et al. (2022), it may help to shift the focus from average to a range of sleep duration guidelines and normalize the occasional experiences of poor sleep to reduce the likelihood of overaction when one falls short of those expectations. For example, emphasizing that 6 to 10 hours of sleep may be appropriate for adults aged 26 to 64 years to feel good and function well, in addition to or even replacing the usual recommendation of 7 to 9 hours. Overall, public sleep health messaging should be developed with consideration of impact upon insomnia development and maintenance.

Strengths, limitations, and future directions

This is the first study to examine the relationships between perfectionism and markers of poor sleep relating to insomnia using actigraphy. Although actigraphy allows the collection of ecological sleep data, it provides only binary sleep/wake detection and does not measure sleep architecture or capture physiological sleep data, such as, brain activity, eye movement, or muscle tone. With technological advances and the introduction of new wearable devices to measure sleep (e.g., Somfit by Compumedics, Melbourne, Australia), future studies could explore if there are ecological physiological markers of poor sleep related to perfectionism. While it was noted there were significant differences between objective and subjective sleep measures in the current sample, no further analysis was done. It remains unknown if sleep-wake state discrepancy might explain the differences found in those relationships between perfectionism and sleep, and whether dysfunctional beliefs play a role in those differences and interrelationships. Another key limitation of the current study relates to its cross-sectional data, which does not allow for causality to be determined and calls for future longitudinal studies. Additionally, the small sample size of the current study is a limitation, and confirmatory work will be required to investigate any specific hypotheses that were generated. Lastly, this study used an actigraphy setting optimized to detect sleep. Other actigraphy settings which may be more optimal for detecting wake can be investigated in future studies.

Conclusions

Research into perfectionism as an important factor in insomnia has been lacking. The current study indicates having high Parental Expectations and Socially Prescribed Perfectionism may increase one’s vulnerability to objectively measured poor sleep and insomnia. Additionally, stress may be a mechanism linking perfectionism and insomnia for individuals with high levels of Concern for Mistakes and Doubts about Action. These findings indicate that perfectionism may be an important factor in preventing and treating insomnia.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by the Melbourne School of Psychological Sciences.