1. Dr. Panayiotis Panayides
Investigating the appropriateness of the Pittsburgh Sleep Quality Index (PSQI) in a non-clinical Cypriot sample
Dr. Panayiotis Panayides

2. The PSQI is a self-rated instrument used to measure sleep quality in clinical samples.
Its development was based on:
psychiatrists’ clinical intuition
experience with sleep disorder patients
a review of other sleep quality questionnaires found in the literature and
clinical experience with the instrument during 18 months of field work.
(Buysse, Reynolds, MonK, Berman & Kupfer, 1989).
In order to:
to provide a reliable, valid, and standardized measure of sleep quality
to discriminate between “good” and “poor” sleepers
to provide an index that is easy for subjects to use and for clinicians and researchers to interpret and
to provide a brief, clinically useful assessment of a variety of sleep disturbances that might affect sleep quality.
(Buysse, Reynolds, MonK, Berman & Kupfer, 1989, p.194)

3. The PSQI Structure

4. The PSQI Items

6. The seven components Component 1: Subjective sleep quality = Score Q6
Q6. During the past month, how would you rate your sleep quality overall?
Very good _____ Fairly good ______ Fairly bad ______Very bad ______
Component 2: Sleep latency = Score Q2 + Score Q5a
Q2. During the past month, how long (in minutes) has it usually taken you to fall asleep each night?
NUMBER OF MINUTES ___________
1 <=15, – 30, – 60, >60
Q5. During the past month, how often have you had trouble sleeping because you . . .
Cannot get to sleep within 30 minutes
0 If sum = 0
1 If sum = 1 – 2
2 If sum = 3 – 4
3 If sum = 5 – 6
Component 3: Sleep duration = Score Q4
Q4. During the past month, how many hours of actual sleep did you get at night?
(This may be different than the number of hours you spent in bed.)
HOURS OF SLEEP PER NIGHT ___________
> 7, – 7, – 6, < 5

7. Component 4: Habitual Sleep Efficiency =
0 If >= 85%, – 84% – 74% < 65%
Component 5: Sleep Disturbances = Sum of score Q5b, …, Q5j (9 items)
0 If sum = 0
1 If sum = 1 – 9
2 If sum = 10 – 18
3 If sum = 19 – 27

8. Component 6: Use of Sleep Medication = Score Q7
Q7. During the past month, how often have you taken medicine to help you sleep
(prescribed or "over the counter")?
Component 7: Daytime Dysfunction = Score Q8 + Score Q9
Q8. During the past month, how often have you had trouble staying awake while driving,
eating meals, or engaging in social activity?
Q9. During the past month, how much of a problem has it been for you to keep up enough
enthusiasm to get things done?
0 If sum = 0
1 If sum = 1 – 2
2 If sum = 3 – 4
3 If sum = 5 – 6

9. Another difference (between the PSQI and other instruments) is that these other questionnaires
have used factor analysis to generate specific factors, while the PSQI components are empirical
and clinical in origin rather than statistical
(Buysse, Reynolds, MonK, Berman & Kupfer, 1989, p.203)

10. Other validation studies with samples of
elderly people (Buysse, Reynolds, Monk, Berman & Kupfer, 1991)
patients with AIDS (Rubinstein & Selwyn, 1998)
patients with panic disorders (Stein, Chartier, & Walker, 1993)
cancer patients (Beck et al., 2004)
patients with primary insomnia
(Backhaus, Junghanns, Broocks, Riemann & Hohagen, 2002)
bone marrow transplant patients and women with breast cancer
(Carpenter & AndryKowski, 1998)
Nigerian students (Aloba, Adewuya, Ola & Mapayi 2007).
Only in one study Chien, Hsu, Tai, Guo and Su (2008) have used Rasch analyses
to validate a revised version of the PSQI (9 items instead of the original 18).

11. The sample
Stratified sampling (By area of residence urban or rural, age and gender)
415 (69.2%) lived in urban areas and 185 (30.8%) in rural areas
+ 15 depression patients (for validation purposes)

12. Rasch Rating Scale Diagnostics
1. Category frequencies
Regular distributions (Uniform, Normal, Bimodal, slightly skewed) are preferable
Low Frequencies do not provide enough observations for estimation of stable
thresholds. Recommended min frequency 10 (Linacre 1999)
2. Average measures (a.m) Average of the ability estimates for all persons who
choose that particular response category on any item in the scale.
(Useful for quick observation of initial problems with the rating scale)
They should increase monotonically, indicating that on average, those with higher
ability / stronger attitude endorse the higher categories
3. Threshold or step calibrations
Thresholds should also increase monotonically across the rating scale, otherwise they are
considered disordered.
Distances between threshold estimates should indicate that each step defines a distinct
position on the variable. (i.e. should be neither too close together nor too far apart)
Between 1.4 – 5 logits to avoid large gaps in the variable. (Linacre, 1999; Bond & Fox, 2001, p. 224)
4. Fit Statistics
Outfit > 2 indicate more misinformation than information (Linacre, 1999), meaning that the
category is introducing noise into the measurement process. Those categories warrant further
empirical investigation and might be good candidates for collapsing with adjacent categories.

13. ANALYSES 1 – Component 5 (Sleep Disturbances, 9 items)

16. ANALYSES 2 – The seven Components
Component 1: Subjective sleep quality
Component 2: Sleep latency
Component 3: Sleep duration
Component 4: Habitual Sleep Efficiency
Component 5: Sleep Disturbances
Component 6: Use of Sleep Medication
Component 7: Daytime Dysfunction

17. 1,
15 out of the 21 3s
4 out of the 5 2s

20. Should component 6 be removed from the PSQI ?
(in the case of using it in a non-clinical sample)

21. ANALYSES 3 – The whole PSQI
7. During the past month, how often have you taken medicine to help you sleep (prescribed or "over the counter")?
8. During the past month, how often have you had trouble staying awake while driving, eating meals, or engaging in
social activity?
5e. During the past month, how often have you had trouble sleeping because you cough or snore loudly

23. 0123

24. 0122

25. 0112

26. Var. explained by Measures 35.1 % 33.6 % 33.8 % 1st factor eigenvector
Original 4-category model category models
PCA of st. residuals
Var. explained by Measures
35.1 %
33.6 %
33.8 %
1st factor eigenvector
2.0
2.1
% of Unexplained variance
12 %
12.2 %
% of Total variance
7.8 %
8.1 %
8.0 %
Item fit
Infit msq
One (1.81)
---
One (1.54)
Outfit msq
Two (1.57, 1.52)
One (1.53)
Reliability
Person Reliability
0.67
0.69
0.71
Person Separation
1.44
1.50
1.56
Strata
2.25
2.33
2.41
Categories
Disordered
Ordered

27. 9 items
5 b) Wake up in the middle of the night or early in the morning
5 c) Have to get up to use the bathroom
5 de) Having breathing problems or snoring
5 fg) Feeling too cold or too hot
5 hij) Have any other problems (like pains, bad dreams, …. )
5 items

28. PSQI 17 items 13 misfitting persons 13 (2.2%) removed 15 (2.5%)
Outfit > 2.30
PCA of st. residuals
Var. explained by Measures
34.7 %
39.0 %
1st factor eigenvector
2.0
% of Unexplained variance
12 %
15.4 %
% of Total variance
7.8 %
9.4 %
Item fit
Infit msq
---
Outfit msq
Reliability
Person Reliability
0.71
0.73
Person Separation
1.58
1.64
Strata
2.44
2.52
Categories
Nicely ordered

29. 17 items
13 items
1.32
1.76

31. FINAL CALIBRATION

32. Use of sleep medication
Staying awake
Sleep efficiency
How long to sleep, Too cold or too hot
To keep enough enthusiasm
Breathing or snoring
Hours actual sleep, Can’t sleep 30 min, Rate Quality of Sleep
Other reason
Bathroom
Wake up middle of night/early morning

33. Distinct Group Comparisons
Depression patients vs Non-clinical subjects
(Doi et al., 2000; Buysse, Reynolds, Monk, Berman and Kupfer, 1989)
p = 0.000
Gender, with higher rate of complaints and worse quality of sleep found in women
(Campbell, Gillin, Kripke, Erikson & Clopton, 1989;
Lindberg, Janson, Gislason, Bjornsson, Hetta & Boman, 1997;
Manni, Ratti, Marchioni, Castelnovo, Murelli, Sartori, Galimberti & Tartara, 1997).
p = 0.000

34. Age, with older people having a higher sleep disorder rate
(Foret, Bensimon, Benoit, & Vieux, 1981; Marquie and Foret, 1999)

35. People living in Urban areas vs Rural areas

36. THANK YOU FOR YOUR ATTENTION