Académique Documents
Professionnel Documents
Culture Documents
ORIGINAL INVESTIGATION
Abstract
Aim. To examine if perfectionism predicts self-reported sleep disturbances over time. Methods. The HewittFlett
Perfectionism Scale was used to assess self-oriented, socially-prescribed (SPP) and other-oriented perfectionism. Sleep
disturbance was evaluated with two items: difficulty in falling asleep and waking up many times during the night. Out of 870
students who participated at baseline, 592 and 305 completed the same measures 1 year (T1) and 2 years later (T2),
respectively. Results. Subjects who reported insomnia at baseline, T1 and T2 (persistent insomnia) had significantly higher
scores of baseline SPP (T1 M 51.5, SD 15.8; T2 M55.0, SD 19.0) than subjects reporting, in all stages of the study,
never/rarely having had sleep problems (T1 M41.9, SD 11.4; T2 M 42.2, SD 12.3, P B0.001 in both cases).
For personal use only.
Regression analyses showed that baseline SPP was the only significant positive predictor of difficulties in falling asleep at T1
and T2 (T1 partial R0.187; T2 partial R0.196, P B0.001) and of difficulties maintaining sleep (T1 partial R0.116;
T2 partial R0.244, P B0.001). Conclusion. SPP was found to be the most reliable predictor of sleep disturbances over
time, which constitutes a new important finding.
Key words: Personality, perfectionism, sleep disturbance, longitudinal study, university students
Correspondence: Maria Helena Azevedo, MD, PhD, Institute of Medical Psychology, Faculty of Medicine, University of Coimbra, Rua
Larga, 3004-504 Coimbra, Portugal. Tel: 351 239 857759. Fax: 351 239 823170. E-mail: mazevedo@fmed.uc.pt
whereas SPP was directed toward avoiding disap- insomnia. Vincent and Walker (2000) used both
proval by others. MPS instruments in a small sample of 32 adults who
Frost and colleagues (1990), conversely, proposed responded to an advertisement placed in a commu-
that six perfectionism facets should be differentiated nity newspaper offering treatment for chronic in-
in their instrument (F-MPS): Personal Standards somnia and 26 healthy controls No significant
(PS, reflecting the setting of high personal associations were found between any of the H&F-
demands); Organization (O, emphasizing that per- MPS dimensions and insomnia, although insomniac
fectionists value order and precision); Concern over subjects tended to report more SPP than healthy
Mistakes (CM, indicating the extent to which the controls. Regarding F-MPS, subjects with chronic
person is concerned with making mistakes); Doubts insomnia were more likely to report DA, frequent
about Actions (DA, reflecting the person indecisive- PC, and CM. However, delayed sleep-onset latency
ness about one’s performance quality). The instru- was only related to the perception of heightened PC.
World J Biol Psychiatry Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 03/04/15
ment presents two additional subscales of an In a previous study with undergraduate students of
interpersonal nature which are circumscribed to both genders we found an association between sleep
the familial environment: Parental Expectations (PE) disturbances and perfectionism (Azevedo et al.
and Parental Criticism (PC) (addressing the great 2007). Further correlational and categorial analyses
importance that perfectionists attach to evaluations revealed that SPP was the only dimension associated
by their parents). According to Flett and Hewitt with sleep disturbance in subjects of both genders.
(2002) this more restricted scope is inadequate. The Males with the highest levels of SPP were approxi-
authors believe that it is important to consider other mately twice more likely to report sleep disturbances
people’s role (i.e. peers and teachers) and socio- than those with less SPP. Similar results were found
cultural factors. within the female sample. However, a positive small
The recognition of perfectionism as a multidimen- correlation between SOP and sleep measures was
sional phenomenon, with both adaptive and mala- observed in females.
daptive components, has enhanced the compre- The literature is sparse on the long-term effects of
For personal use only.
when exams periods were not taking place. Baseline verify if the former groups were representative of the
was considered the first stage, the first year of the whole cohort subjects. Subjects followed up at T1
follow-up was Time 1 (T1) and the second year of had higher baseline SOP than subjects who were not
the follow-up was Time 2 (T2). The cooperation followed up (M 82.5, SD 14.90 vs. M 79.10;
was voluntary, in all stages, and confidentiality was SD 16.79, t 2.820; P 0.005). Apart from this
ensured. Most students returned the questionnaires. observation, no other significant differences were
Participation refusal occurred when questionnaires identified between groups.
were not handed.
Measures
Participants
Perfectionism. Hewitt and Flett-MPS (1991) was
Table I presents participants characteristics at each
used to measure Self-oriented perfectionism,
World J Biol Psychiatry Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 03/04/15
(never/rarely/sometimes responses) either at T1 ‘‘often, very often and/or always’’ was mentioned
or T2. by 5.2% at baseline, 5.4% at T1 and 6.8% at T2 (see
Table III). Both DIS and/or DMS were experienced
Data analyses and statistics at baseline, T1 and T2 by 23.6, 22.2 and 25.6%,
respectively.
To investigate if perfectionism groups (previously
defined) diverged in DIS, DMS and SDI at different
stages of the study (baseline, T1 and T2), Student’s Perfectionism and sleep variables correlations between
t-test were applied. Odds ratio analyses were addi- the same variables at different follow-up time points
tionally performed to explore the likelihood of a
Correlations (Pearson’s Moment Product Correla-
subject with high scores in total perfectionism or in
tions) between Perfectionism variables at baseline
particular perfectionism dimensions experiencing
and at T1 and T2 revealed large correlation coeffi-
World J Biol Psychiatry Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 03/04/15
the three perfectionism dimensions (baseline scores) baseline revealed significantly higher mean scores in
explained sleep difficulties (DIS, DMS and SDI) at DMS (M 1.33, SD 0.91) and SDI (baseline)
T1 and T2. According to Cohen (1992), correla- (M 3.20, SD 1.86) than the group with lower
tions of 0.10 represent small effect, 0.30 medium, perfectionism total score (baseline) (DMS, M
and 0.50 large effect. 1.01, SD 0.82, P 0.018; SDI, M2.60, SD
1.56, P 0.024).
Results The former group also revealed significantly
higher mean scores in all sleep variables (DIS,
Descriptive data DMS and SDI) at T2 (2 years after baseline
Perfectionism and sleep variables mean scores and measurements) comparatively to the group with
standard deviations at baseline, T1 and T2 are lower total perfectionism scores (at baseline) (see
presented in Table II. Table V).
Difficulty initiating sleep (DIS) described as With respect to SPP, the group with high scores in
‘‘often, very often and/or always’’ was reported by this dimension (baseline) showed significantly more
18.4% at baseline, 16.8% at T1 and 18.9% at T2. DIS, DMS and SDI in baseline, T1 and T2 than the
Difficulty maintaining sleep (DMS) reported as group with low scores in this dimension (Table VI).
Table II. Perfectionism and sleep variables mean scores and standard deviations (SD).
Perfectionism
SOP 81.41 (15.60; 18117) 79.61 (14.89; 35119) 77.88 (16.07; 26119)
SPP 44.78 (12.28; 1594) 43.22 (12.12; 1596) 43.37 (12.38; 1897)
OOP 44.84 (6.98; 1863) 45.29 (7.32; 1663) 45.31 (7.16; 2462)
Total scale 186.13 (27.48; 97265) 183.32 (27.18; 103302) 181.92 (29.01; 87277)
Sleep
DIS 1.70 (1.60; 05) 1.61 (1.04; 05) 1. 67 (1.28; 05)
DMS 1.18 (0.80; 05) 1.23 (0.80; 04) 1.13 (0.84; 05)
SDI 2.89 (1.64; 010) 2.84 (1.59; 09) 2.07 (1.39; 06)
SOP, Self-oriented perfectionism; SPP, Socially-prescribed perfectionism; OOP, Other-oriented perfectionism; DIS, Difficulties initiating
sleep; DMS, Difficulties maintaining sleep; SDI, Sleep disturbance index; SD, Standard deviation; T1, Time 1; T2, Time 2.
480 M.H. Azevedo et al.
Table III. DIS and DMS frequency at baseline and follow-up.
NK, Not known; DIS, Difficulties initiating sleep; DMS, Difficulties maintaining sleep; T1, Time 1; T2, Time 2.
World J Biol Psychiatry Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 03/04/15
Considering SOP and OOP dimensions, only a results were only found for SPP scores. The persis-
small number of significant results were found. The tent insomnia group showed significantly higher SPP
group with high SOP values revealed higher mean values (n 23, M 55.0, SD 19.0) than the good
scores in DMS in T2 than the group with low SOP sleepers group (n 74, M42.2, SD 12.28, P
(M 1.47, SD 0.98 vs. M1.09, SD 0.74, P 0.02).
0.033). The group with high OOP revealed less
DMS (M 1.16, SD 0.80) than the group with
low OOP (M 1.39, SD 0.86). Correlational and regression analyses
Exploring SPP in detail, subjects with high values Pearsons’ correlations between perfectionism di-
in this dimension at baseline were approximately 3 mensions and sleep variables showed that SOP at
times more likely to have overall sleep problems baseline was significantly and positively correlated
(x2 9.164; P 0.002; OR 2.696 CI 95%, 1.454
For personal use only.
Time 2 1.07 (.74) 1.54 (1.04) 2.535 0.013 showed consistently having more DIS, DMS and
SDI SDI than subjects with low levels of SPP through all
Baseline 2.60 (1.56) 3.20 (1.86) 2.274 0.024 assessment times/stages. These students were 3
Time 1 2.72 (1.56) 3.06 (1.91) 1.225 0.222 times more likely to have overall sleep problems or
Time 2 1.76 (1.28) 2.50 (1.41) 2.628 0.010 DIS, 2 times more likely to have DMS 1 year later
DIS, Difficulties initiating sleep; DMS, Difficulties maintaining (T1), and 7 times more likely to report overall sleep
sleep; SDI, Sleep disturbance index; SD, Standard Deviation; t, disturbance 2 years later (T2) than subjects with low
Student’s t-test. SPP. Additionally, students with persistent insomnia
showed higher scores on SPP, at T1 and T2, than
line scores) explained 4% of the total variance of the good sleepers.
SDI [r2 0.040, F(3, 541) 8.505, P B0.001], One of the earliest and most enduring insomnia
3.4% of the total variance of DIS [r2 0.034; F(3, conceptualisations refers to psychophysiological
541) 7.409, P B0.001] and 2.6% of the total arousal (Buysse and Dorsey 2002). From the several
For personal use only.
variance of DMS [r2 0.026, F(3, 542) 5.882, reasons for hyperarousal, psychological distress (PD)
P 0.001] at T1. At T2 the three perfectionism and psychiatric disorders are among the most salient
variables explained 6.4% of the total variance of the causes of insomnia (Hauri 2002). It is well known
SDI [r2 0.064, F(3, 277) 7.274, P B0.001], that the aetiology of PD and psychiatric syndromes
5.6% of the total variance of DIS [r2 0.056, F(3, involves the interplay of many biological (e.g.,
277) 6.516, P B0.001] and 8.1% of the total genetic) and environmental factors (Tandon and
variance of DMS [r2 0.081, F(3, 277) 9.150, McGuffin 2002; Rijsdijk et al. 2003). Therefore, as
P B0.001]. Considering the partial correlations it would be expected in complex transmission
between perfectionism and sleep variables, baseline disorders, the pathway between genetic/develop-
SPP was a significant predictor of DIS, DMS and mental risk factors and the behavioural phenotypes
SDI at T1 and T2. DMS at T1 (1 year after
baseline) could be predicted (positive association) Table VI. Comparison between baseline Socially prescribed
perfectionism groups (SPP) in sleep variables at baseline, Time
by SPP baseline and by OOP (negative association). 1 and Time 2.
Sleep groups
1 2 3 4
Good Sleepers Remission Onset Persistent
Baseline n 189 n 54 n48 n 57
Perfectionism Mean (SD) Mean (SD) Mean (SD) Mean (SD) P Comparisons (1)
SOP 81.4 (14.6) 84.6 (12.4) 85.5 (14.6) 84.3 (17.2) 0.231
SPP 41.9 (11.4) 44.2 (9.8) 46.2 (12.1) 51.5 (15.8) 0.000** 4 1**,2* (2)
OOP 45.2 (7.0) 46.1 (7.2) 46.1 (6.1) 42.8 (8.0) 0.05
Total scale 183.5 (26.3) 189.8 (22.8) 193.4 (27.0) 193.0 (30.5) 0.038* 1 B 3*, 4* (3)
is not straightforward. Instead, it is a long and (Stöber and Otto 2006), specifically to depression,
devious road with many crossroads and dead-ends. anxiety, eating and OC disorders (Maia et al. in
Some alternatives ways have been proposed to over- preparation).
come this complexity and to clarify the underlying It is worth mentioning that the main reason for
mechanisms by which genetic susceptibility influ- student’s self-referral in the outpatient clinic for
ence psychopathological conditions. One consists in psychological treatment and counselling of Coimbra
using alternative phenotypic measures (e.g., endo- University is stress-related problems, namely social
phenotypes), including personality traits such as anxiety and psychological distress associated with
perfectionism. In a near future, we may discover as evaluative concerns (Pereira 2004). Likewise, the
For personal use only.
Insel and Collins (2003) have stated about the study of Cabrita et al. (2004), based in a large
aetiology of anorexia nervosa that its genotype is sample of Lisbon University students found that
not linked to a specific disorder but to a perfectio- stress-related complaints (anxiety, depression and
nistic, risk-aversive personality style that confers insomnia) were the most frequently stated health
vulnerability to many syndromes. There is some problems to justify the use of psychoactive drugs
evidence of the involvement of genetic factors in during the fortnight before the interview. Tranqui-
perfectionism (Lilenfeld et al. 2000; Woodside et al. lizers (BZD; valerian) were taken by 7.2% students
2002; Tozzi et al. 2004). Moreover, an association
(with valerian accounting for 35.5% of the total
between perfectionism and insomnia has system-
number of medicines consumed) and 1.7% used
atically been found in the literature (Lundh et al.
antidepressants.
1994; Vincent and Walker 2000; Azevedo et al.
Therefore, it can be argued that the association
2007). Therefore, based in the existing literature, we
between SPP and future insomnia observed in our
believe that perfectionism can be seen as an inter-
study can be a result of PD experienced by students.
mediate phenotype associated with sleep difficulties.
It is noteworthy that, in our study, SPP mean values
To support the case of perfectionism as an
intermediate phenotype of sleep disturbances, it is for the insomnia group and for good sleepers were
also important to insure a strong temporal stability similar to the ones found by Vincent and Walker
of the influence of this trait measure and, also to (2000) in their chronic insomnia and healthy control
assure that it is not predominantly state-related group. In addition, our insomnia group SPP mean
(Rice and Aldea 2006). Research on depression values at T1 were equal to the scores of psychiatric
and sleep disturbances supports that perfectionism outpatients with a clinical diagnosis of depression
is a personality trait with both relative stability and and/or anxiety, and similar, at T2, to the scores of a
state dependence effects (Zuroff et al. 1999; Cox mixed sample of outpatients with eating and/or
and Enns 2003; Rice and Aldea 2006; Maia et al. in obsessivecompulsive disorders found in a previous
preparation). study of our group (Maia et al. in press).
Specific personality traits, including perfectionism It is particularly relevant that certain persona-
have been associated with cognitive arousal and lity traits correlate with particular cognitive aspects,
psychological distress (PD; Kales et al. 1983; Kales which play a central role in the conceptual
and Kales 1984; Shafran and Mansell 2001). Some models of PD (Beck 1976; Clark 1999) and in the
dimensions of perfectionism (e.g., SPP, CM, DA) explanation of complex mechanisms involved in the
have been consistently related to higher levels of PD interplay between personality, PD and insomnia.
Perfectionism and sleep: longitudinal study 483
SOP, Self-oriented perfectionism; SPP, Socially-prescribed perfectionism; OOP, Other-oriented perfectionism; r, Pearson moment-product correlation coefficients; Partial R, partial correlation
According to the cognitive model of insomnia,
0.231**
Partial R
0.013
0.000
worriers and ruminators are more likely to react to
life stressors creating cognitive arousal which, in
T2
turn, will affect sleep, leading to insomnia (Perlis
et al. 2005). There is empirical evidence that some
0.271**
0.133*
perfectionism negative facets (e.g., SPP) are asso-
0.016
r
ciated with worry, rumination and autonomic arou-
sal (Flett et al. 1995, 1998). Worriers have an
Table VIII. Correlation coefficients and partial correlations between baseline Perfectionism and sleep disturbances scores at Time 1 (n 592) and Time 2 (n 305).
0.181**
elevated and intense fear of failure that makes
Partial R
0.015
0.062
them hesitant about making mistakes. In other
words, they need to be absolutely sure that they
are doing the ‘‘right thing’’ before a response can be
T1
World J Biol Psychiatry Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 03/04/15
0.042
0.003
0.031
0.048
0.049
0.049
0.023
0.053
insomniacs (see Harvey 2002). To our knowledge Flett GL, Hewitt PL, Blankstein K, Mosher SW. 1991. Perfec-
tionism, self-actualization, and personal adjustment. J Soc
there is only one prospective study where these
Behav Pers 6:147160.
pathways were analysed. Results showed an associa- Flett GL, Russo FA, Hewitt PL. 1994. Dimensions of perfection-
tion between perfectionism, PD and insomnia ism and constructive thinking as a coping response. J Ration
(Jansson-Fröjmark and Linton 2007). However, Emot Cogn Behav Ther 12:163179.
when emotional distress was accounted for, none Flett GL, Hewitt PL, Endler NS, Tassone C. 1995. Perfectionism
of the perfectionism subscales (CM and PS) con- and components of state and trait anxiety. Curr Psychol
13:326350.
tributed significantly to explain pre-existing and
Flett GL, Hewitt PL, Blankstein KR, Solnik M, Van Brunschot
future insomnia. M. 1996. Perfectionism, social problem-solving ability, and
Future work should examine the mechanisms psychological distress. J Ration Emot Cogn Behav Ther
linking SPP to insomnia/sleep disturbances, specifi- 14:245275.
cally exploring the role of psychological distress, Flett GL, Hewitt PL, Blankstein KR, Gray L. 1998. Psychological
World J Biol Psychiatry Downloaded from informahealthcare.com by University of Ulster at Jordanstown on 03/04/15
worry and rumination. distress and the frequency of perfectionistic thinking. J Pers Soc
Psychol 75:13631381.
Frost RO, Marten P, Lahart C, Rosenblate R. 1990. The
dimensions of perfectionism. Cogn Ther Res 14:449468.
Acknowledgements Frost G, Heimberg R, Holt C, Mattia J, Neubauer A. 1993. A
comparison of two meaasures of perfectionism. Pers Indiv
The data for this report were drawn from a research
Differ 14:119126.
on Perfectionism and ObsessiveCompulsive Spec- Hamachek D. 1978. Psychodynamics of normal and neurotic
trum Disorders, funded by Fundação para a Ciência perfectionism. Psychology 15:2733.
e Tecnologia (FCT no. 37569/PSI). The co- Harvey AG. 2002. A cognitive model of insomnia. Behav Res
operation of Professors and Students is gratefully Ther 40:869893.
acknowledged. Hauri PJ. 2002. Psychological and psychiatric issues in the
etiopathogenesis of insomnia. Primary care companion. J Clin
Psychiatry 4:1720.
Statement of interest Hewitt PL, Flett GL. 1991. Perfectionism in the self and social
For personal use only.