Test–retest reliability and validity of the Pittsburgh Sleep Quality Index in primary insomnia Jutta Backhausa,*, Klaus Junghannsa, Andreas Broocksa, Dieter Riemannb, Fritz Hohagena a b
Department of Psychiatry and Psychotherapy, University Hospital of Luebeck, Luebeck, Germany Department of Psychiatry and Psychotherapy, University Hospital of Freiburg, Freiburg, Germany Received 19 July 2001; accepted 5 February 2002
Abstract Objective: Psychometric evaluation of the Pittsburgh Sleep Quality Index (PSQI) for primary insomnia. Methods: The study sample consisted of 80 patients with primary insomnia (DSM-IV). The length of the test – retest interval was either 2 days or several weeks. Validity analyses were calculated for PSQI data and data from sleep diaries, as well as polysomnography. To evaluate the specificity of the PSQI, insomnia patients were compared with a control group of 45 healthy subjects. Results: In primary insomnia patients, the overall PSQI global score correlation coefficient for test – retest reliability was .87. Validity analyses showed high correlations between PSQI and sleep log data and lower correlations with polysomnography data. A PSQI global score > 5 resulted in a sensitivity of 98.7 and specificity of 84.4 as a marker for sleep disturbances in insomnia patients versus controls. Conclusion: The PSQI has a high test – retest reliability and a good validity for patients with primary insomnia. D 2002 Elsevier Science Inc. All rights reserved.
Keywords: Pittsburgh Sleep Quality Index; Primary insomnia; Sleep; Reliability; Validity
Introduction Since its introduction in 1989, the Pittsburgh Sleep Quality Index (PSQI)  has gained widespread acceptance as a useful tool to measure sleep quality in different patient groups (for overview of sleep questionnaires, see Refs. [2,3]). Its good reliability and validity could be shown for patients with psychiatric and sleep disorders [1,4], for patients with different somatic diseases , for nursing home residents , and for healthy elderly subjects , whereas data on its validity and reliability for primary insomnia are scarce. Primary insomnia is a sleep disorder that is caused neither by an organic nor by any other psychiatric disorder. There is a chronic difficulty in initiating or maintaining sleep, which causes clinically significant distress or impairment in social, occupational, or other important areas of functioning. Doi et al.  evaluated the reliability of a Japanese version of the PSQI in a sample of psychiatric patients, which included * Corresponding author. Klinik fur Psychiatrie und Psychotherapie, ¨ Universitaetsklinikum Luebeck, Ratzeburger Allee 160, D-23538 Luebeck, Germany. Tel.: +49-451-500-2922; fax: +49-451-500-2603. E-mail address: firstname.lastname@example.org (J. Backhaus).
a small subsample of 14 patients with primary insomnia. They found a high reliability with an overall Cronbach’s alpha of .77 (n = 14). There are no data published yet on test – retest reliability and validity of the PSQI with sleep logs for patients with primary insomnia. The study presented here serves this goal.
Methods PSQI The PSQI is a self-rating questionnaire resulting in a global score between 0 and 21, which consists of seven subscores (see Table 1). The questionnaire is easy to handle and can be completed within 5 min. The whole questionnaire is published by Buysse et al. . The German version has been translated from English into German and then retranslated for comparison with the original version. In the German version, the estimate period was reduced from 4 to 2 weeks , because the German PSQI served as an instrument to evaluate a short-term therapy for primary insomnia , for which the 4-week period seemed too long.
0022-3999/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S 0 0 2 2 - 3 9 9 9 ( 0 2 ) 0 0 3 3 0 - 6
J. Backhaus et al. / Journal of Psychosomatic Research 53 (2002) 737–740
Table 1 Correlation coefficients for test – retest reliability in patients with primary insomnia Separate test – retest correlation analyses in patients with primary insomnia Overall test – retest correlation in patients with primary insomnia (N = 76) r Global score Subscores Sleep quality Sleep onset latency Sleep duration Sleep efficiency Sleep disturbances Use of sleeping medication Daytime dysfunction .87 .53 .79 .80 .81 .69 .88 .81 P .000 .000 .000 .000 .000 .000 .000 .000 Short interval (2 days, n = 45) r .90 .76 .87 .88 .90 .84 .92 .90 P .000 .000 .000 .000 .000 .000 .000 .000 Longer interval (45.6 ± 18 days, n = 31) r .86 .23 .71 .71 .74 .27 .83 .59 P .000 .211 .000 .000 .000 .138 .000 .001
Study sample and study design The study sample consisted of 80 patients with primary insomnia according to the criteria of DSM-IV  and 45 healthy controls. All patients were referred by their physicians or were self-referred to one of the sleep laboratories at the Departments of Psychiatry of the Universities of Freiburg or Luebeck. Controls participated in different studies and performed the PSQI additionally for this study. After a thorough medical examination to rule out any somatic or psychiatric disorder underlying sleep disturbances, the patients had two nights of polysomnography in the sleep laboratory including measurement of sleep apneas and periodic leg movements. Standard electroencephalographic (C3/A1, C4/A2), submental and tibial electromyographic, electrooculographic, electrocardiographic, and breathing parameters were recorded over 8 h (23:00 to 7:00 h). For correlation analyses, the sleep stages of the second night were used, which were scored according to standard criteria . Seventy-six of the insomnia patients received the PSQI twice: 45 of them during their examination in the sleep laboratory with a test –retest interval of 2 days, and 31 received the PSQI twice over an interval of 45.6 ± 18 days prior to and after a waiting time for treatment. Thirty-two patients filled out daily sleep logs over a period of 7 to 14 days (mean 11.5 ± 3.3) directly before or after they received the PSQI. Only data from controls were used for the calculation of specificity. Therefore, they filled out the PSQI but had no polysomnography. The insomnia patients had a mean age of 46.3 ± 15 years, the controls of 43.3 ± 9.5 years; the difference between the groups was not statistically significant (t test: df = 123, P =.171). The sex ratio in both groups was equal, with a female rate of 58.8% in the sample of insomnia patients and 51.1% in the control (chi-square: P =.409). Statistics Statistical analyses were performed using Pearson’s correlations for test –retest reliability of the PSQI global score
and subscores. For validity analyses, total sleep time and sleep onset latency according to the PSQI and the sleep logs were correlated. Furthermore, sleep EEG data were correlated with the PSQI global score. The t tests for independent samples and a chi-square test were calculated to compare insomnia patients and healthy controls. All tests were calculated with a two-tailed alpha of 5%. Reliability, validity, and sensitivity analyses were made with the primary insomnia patient group. For the specificity analysis, patients were compared with the age- and sex-matched control group of healthy subjects.
Results The mean PSQI global score was 12.5 ± 3.8 for insomnia patients and 3.3 ± 1.8 for controls ( P = .000). All PSQI subscores were also significantly different between both groups ( P = .000), with higher scores for the insomnia patients. The recommended PSQI global score > 5 as an indicator of relevant sleep disturbances  resulted in a sensitivity for primary insomnia of 98.7. The specificity was 84.4, with 7 out of 45 healthy subjects having a PSQI global score of 6 and none having a global score beyond 6. A cutoff point > 6 in the PSQI global score thus resulted in a slightly lower sensitivity of 93.4% and a specificity of 100%. In primary insomnia patients, the overall PSQI test – retest reliability was high for the global score and for the subscores with the exception of the subscores ‘‘sleep quality’’ and ‘‘sleep disturbance,’’ which were in the middle range (see Table 1). The test – retest reliability for the short interval (2 days) was high for the global score and all subscores. For the longer interval (45.6 ± 18 days), the test –retest reliability was low for the subscores ‘‘sleep quality’’ and ‘‘sleep disturbance’’ whereas it remained high for the global score and four of seven subscores (see Table 1). The PSQI homogeneity was high with a Cronbach’s alpha of .85. Patients estimated their sleep duration at 298.1 ± 101.6 min in the PSQI and at 328.9 ± 81.8 min in the sleep
J. Backhaus et al. / Journal of Psychosomatic Research 53 (2002) 737–740 Table 2 Correlation coefficients for PSQI global score and polysomnographic data in patients with primary insomnia Mean ± S.D. Total sleep time Total sleep time without Stage 1 Sleep efficiency (% sleep of bedtime) Sleep onset latency (minutes until first epoch of Stage 2) Number of nightly awakenings % of Stage 1 % of Stage 2 % of Stages 3 and 4 % of REM 388.2 ± 66.9 349.5 ± 72.0 80.3 ± 12.9 20.6 ± 21.7 22.6 ± 9.1 6.8 ± 4.2 50.7 ± 9.5 3.4 ± 5.5 17.9 ± 5.1 r À .27 À .32 À .32 .28 .12 .24 À .33 À .20 À .13
P .074 .037 .034 .063 .427 .107 .028 .190 .385
Patients with primary insomnia (n = 45) performed the PSQI during their examination in the sleep laboratory.
diaries. The estimate of sleep onset latency was 51.8 ± 69.0 min in the PSQI and 38.1 ± 28.9 min in the sleep diaries. For both parameters, the correlations between the PSQI and sleep log data were high and significant (r =.81, P =.000 for sleep duration and r =.71, P =.000 for sleep onset latency). The correlations between PSQI and polysomnographic data were much lower, but also significant for sleep efficiency, sleep onset latency, percentage of Stage 2, and total sleep time without Stage 1 (see Table 2).
Discussion For patients with primary insomnia, the overall test – retest reliability for the PSQI global score is high for a shortterm interval of 2 days, as well as for a longer interval of several weeks. The fact that some subscore test – retest coefficients diminish when the time interval between the tests is longer underlines the variability of sleep across time. The correlation of PSQI data with sleep logs demonstrates the good validity of the PSQI. The retrospective evaluation of several weeks’ sleep leads to a more global estimate than a daily log. This might lead to a bias in the PSQI due to memory distortion. Indeed, the global estimate of sleep duration was lower and that of the sleep onset latency was higher in the PSQI than in the sleep diaries. This probably reflects a tendency of insomnia patients to focus on bad nights and therefore to underestimate sleep duration and overestimate sleep onset latency across several nights’ sleep. Sleep diaries therefore seem indispensable to a more differentiating approach to sleep disorders in which the distorted view of the insomnia patient is a focus of diagnosis or treatment, e.g., in cognitive – behavioral psychotherapy. It is well known by sleep researchers that subjective and objective measures of sleep are discrepant in patients with insomnia . The high correlations between PSQI and daily logs demonstrate that the PSQI can be a very timeeffective and useful first-line questionnaire to qualify sleep disturbances especially in settings like general practice where time to look for special aspects in detail is often
lacking. However, it should be kept in mind that the PSQI score alone does not allow for a diagnostic differentiation of different kinds of sleep disorders and sleep disturbances due to psychiatric disorders or somatic diseases. The PSQI contains some questions with respect to restless legs syndrome, sleep apnea, and parasomnia, which do not yield to the PSQI score. The validity of these questions will have to be evaluated further. The correlations between PSQI and the sleep EEG parameters’ total sleep time without Stage 1, sleep efficiency, sleep onset latency, and percentage of Stage 2 were significant though considerably lower than for the correlations between PSQI data and sleep logs. The PSQI has a high sensitivity and specificity for insomnia patients in comparison to healthy controls, thus underscoring that it is a good measure for differentiating between good sleepers and patients suffering from sleep disturbances. Our data suggest, however, that the cut-off score should be set to 6 in order to maximize specificity while only modestly reducing sensitivity. In sum, the PSQI proved to be a valuable adjunct to clinical work on insomnia and is a useful first-line, easy-to-handle, and time-efficient questionnaire to evaluate sleep disturbances.
 Buysse DJ, Reynolds CF, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res 1989;28:193 – 213.  Chesson A, Hartse K, Anderson WM, Davila D, Johnson S, Littner M, Wise M, Rafecas J. Practice parameters for the evaluation of chronic insomnia: an American Academy of Sleep Medicine report. Standards of Practice Committee of the American Academy of Sleep Medicine. Sleep 2000;23:237 – 41.  Sateia MJ, Doghramji K, Hauri PJ, Morin CM. Evaluation of chronic insomnia: an American Academy of Sleep Medicine review. Sleep 2000;23:243 – 308.  Doi Y, Minowa M, Uchiyama M, Okawa M, Kim K, Shibui K, Kamei Y. Psychometric assessment of subjective sleep quality using the Japanese version of the Pittsburgh Sleep Quality Index (PSQI-J) in psychiatric disordered and control subjects. Psychiatry Res 2000;97: 165 – 72.
J. Backhaus et al. / Journal of Psychosomatic Research 53 (2002) 737–740 tiveness of a short-term cognitive – behavioral group treatment for primary insomnia. Eur Arch Psychiatry Clin Neurosci 2001;251:35 – 41.  American Psychiatric Association. Diagnostic and statistical manual of mental disorders, fourth edition. Washington (DC): American Psychiatric Association, 1994.  Rechtschaffen A, Kales A. A manual of standardised terminology, techniques and scoring system for sleep stages of human subjects. Beheads: Neurological Information Network, 1968.  Carskadon MA, Dement WC, Mitler MM, Guilleminault C, Zarcone VP, Spiegel R. Self-reports versus sleep laboratory findings in 122 drug-free subjects with complaints of chronic insomnia. Am J Psychiatry 1976;133:1382 – 8.
 Carpenter JS, Andrykowski MA. Psychometric evaluation of the Pittsburgh Sleep Quality Index. J Psychosom Res 1998;45:5 – 13.  Gentili A, Weiner DK, Kuchibhatla M, Edinger JD. Test – retest reliability of the Pittsburgh Sleep Quality Index in nursing home residents. J Am Geriatr Soc 1995;43:1317 – 8.  Buysse DJ, Reynolds CF, Monk TH, Hoch CC, Yeager AL, Kupfer DJ. Quantification of subjective sleep quality in healthy elderly men and women using the Pittsburgh Sleep Quality Index (PSQI). Sleep 1991;14:331 – 8.  Riemann D, Backhaus J. Behandlung von Schlafstoerungen. Ein psychologisches Gruppenprogramm. Weinheim: Beltz, 1996. pp. 94 – 103.  Backhaus J, Hohagen F, Voderholzer U, Riemann D. Long-term effec-