Interpreting Scores On The K10

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Measuring Risk Article

Interpreting scores on the Kessler


Psychological Distress Scale (K10)

Abstract Gavin Andrews and Tim Slade


Objective: To provide normative data on World Health Organization Collaborating Centre in Evidence for Mental
the Kessler Psychological Distress Scale Health Policy and School of Psychiatry, University of New South Wales
(K10), a scale that is being increasingly
at St Vincent’s Hospital
used for clinical and epidemiological

I
purposes.
n 1997, the Australian Bureau of Statis- 16 domains from depression, anxiety and
Method: The National Survey of Mental
Health And Well-Being was used to provide tics conducted a household survey on worry to the physical symptoms of arousal,
normative comparative data on symptoms, mental disorders and Mathers et al. fatigue and thoughts of death. These 45 items
disability, service utilisation and diagnosis found that mental disorders accounted for were administered by mail to a national sam-
for the range of possible K10 scores. 13.5% of the burden of disease, ranking third ple of 1,401 respondents, further ref ined and
Results: The K10 is related in predictable behind heart disease and cancers.1,2 A brief 32 items were administered to another na-
ways to these other measures. screening scale was needed to identify vari- tional sample. They used psychometric mod-
Implications: The K10 is suitable to assess ations in the population. When developing els based on item response theory to analyse
morbidity in the population, and may be the national survey the authors sought for a the results of both surveys. On the basis of
appropriate for use in clinical practice. brief symptom scale that might be suitable each item’s difficulty and discrimination they
(Aust N Z J Public Health 2001; 25: 494-7) for use in other health surveys and might selected a 6- and 10-item set that represented
inform clinicians about the level of morbid- the entire range of distress and was discrimi-
ity in individual patients. They chose the nating along that continuum.5
scale of psychological distress (K10) devel- The 10-item (or K10) scale is:
oped by Kessler et al.3-5 In this paper we In the past 30 days how often ...
present comparative normative data on that 1. Did you feel tired out for no good rea-
scale from the Australian national survey. son.
2. Did you feel nervous.
3. Did you feel so nervous that nothing
Method could calm you down.
Development of the K10: Kessler and 4. Did you feel hopeless.
Mroczek repor ted the development of a 5. Did you feel restless or fidgety.
screening battery for non-specific psycho- 6. Did you feel so restless that you could
logical distress for use in the annual US Na- not sit still.
tional Health Interview Survey.3,4 They 7. Did you feel depressed.
acknowledged that a number of excellent 8. Did you feel that everything was an
scales were in existence but argued that de- effort.
velopments in item response theory would 9. Did you feel so sad that nothing could
permit a shorter scale with more informa- cheer you up.
tion than has previously been achieved. They 10.Did you feel worthless.
assembled a large item pool from 22 sources The scale used a f ive-value response
(list available from GA). The 500 or so items option for each question – all of the time,
in those sources were then reviewed and re- most of the time, some of the time, a little of
duced to 45 non-redundant items. They en- the time, and none of the time – that were
sured that the items were relevant to special scored from f ive through to one. The
populations (adolescents, US racial and eth- maximum score is therefore 50, indicating
nic groups, rural populations) and covered severe distress, and the minimum score is

Submitted: April 2001 Correspondence to:


Professor Gavin Andrews, UNSW at St Vincent’s Hospital, 299 Forbes Street, Darlinghurst,
Revision requested: September 2001
NSW 2010. Fax: (02) 9332 4316; e-mail: gavina@crufad.unsw.edu.au
Accepted: October 2001

494 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2001 VOL. 25 NO. 6
Measuring Risk The Kessler Psychological Distress Scale (K10)

10, indicating no distress. Items 3 and 6 are not asked if the re- Table 1: Correspondence between psychological
sponse to the preceding question was ‘none of the time’. distress as measured by the K10, and scores for
The K10 scale was a part of the National Survey of Mental Health respondents in that category on the GHQ, the SF-12 (all
measured with respect to the last 30 days), and the
and Well Being conducted in Australia in 1997.1 This was a strati-
frequency of consultations for a mental health problem
fied household survey of a population sample of persons 18 years
over the past year.
and older. The response rate was 78% and 10,641 people were in-
terviewed, all of whom completed the K10. The Composite Inter- K10 % in GHQ SF-12 n of
national Diagnostic Interview v2.06 was used to provide DSM-IV score category score score consultations
(weighted)
diagnoses of the common anxiety, affective and substance use dis-
10-14 67.5 0.3 (0.0) 55.6 (0.1) 0.2 (0.0)
orders. In addition, screening instruments were used for personal-
ity disorders, psychosis and neurasthenia. Disability was measured 15-19 20.4 1.4 (0.1) 48.3 (0.0) 0.9 (0.1)

by the SF-127 and psychological distress by the 12-item GHQ8 and 20-24 7.2 2.4 (0.1) 41.5 (0.4) 1.9 (0.2)

by the K10. Service utilisation was also measured. 25-29 2.7 4.0 (0.2) 34.3 (0.6) 3.4 (0.3)

All scores on all variables were weighted to reflect the compo- 30-34 1.3 5.4 (0.4) 31.2 (1.1) 5.8 (0.9)

sition of the whole population. Routine data analytic procedures 35-39 0.5 5.4 (0.4) 29.2 (1.4) 8.0 (1.3)
were used, but as a result of the complex sample design and weight- 40-50 0.4a 7.6 (0.9) 26.2 (1.5) 10.2 (1.7)
ing the SUDAAN package was used to estimate standard errors Overall mean NA 0.9 (0.0) 52.0 (0.1) 0.7 (0.0)
around prevalence estimates.9 Overall skew NA 3.0 -1.4 6.7
Correlation
with K10 (rho) NA 0.5 -0.6 0.3

Results Notes:
(a) 0.4% equals 45 respondents before weighting back to the population.
The K10 had a mean of 14.2, a median of 12, range 10-50, with
68% of respondents scoring under 15 and 3% scoring 30 and
above (skewness = 2.2). The distribution, like all measures of current DSM-IV or ICD-10 case by the CIDI is respectively dis-
psychological distress is heavily skewed (skew 2.2) and the played in Tables 2 and 3 in respect to any anxiety, affective, sub-
majority of people report little or no distress. The mean for fe- stance use or personality disorder and for any disorder identif ied
males was higher than that for males; 14.5 vs. 13.9 (p<0.001). in the survey. Similar data in respect to cases that met criteria for
The K10 scores were banded into seven categories to facilitate one of these diagnoses at some point in the previous 12 months
comparison with the GHQ (a symptom measure that can predict are displayed in Tables 4 and 5. The sensitivity and specif icity of
casedness), the mental health component score of the SF-12 (a the K10 as a marker for diagnosis of any current DSM-IV or ICD-
measure of disability), the number of consultations for a mental 10 anxiety or affective disorders is displayed in Table 6.
health problem in the past year (a measure of met need), and the
probability of meeting criteria for a diagnosis either currently or
in the past 12 months. The results comparing the K10 Scores with Discussion
those of the GHQ, the SF-12 and the frequency of mental health There is a significant association between scores on the K10
consultations are displayed in Table 1. The correspondence be- and scores on the GHQ and SF-12, measures of symptoms and
tween K10 scores and the probability of being identified as a disability respectively, and between the K10 and the number of

Table 2: Correspondence between psychological distress as measured by the K10 and the probability of meeting
criteria for a current DSM-IV mental disorder (exclusion criteria operationalised) for respondents scoring in each K10
category.

K10 score Anxiety disorder Affective disorder Substance use Any study disorder
(weighted) prevalence % (SE) prevalence % (SE) disorder prevalence prevalence
% (SE) % (SE)
10-14 0.5 (0.1) 0.2 (0.1) 1.8 (0.2) 5.4 (0.3)
15-19 3.8 (0.4) 3.0 (0.4) 4.8 (0.5) 17.8 (0.9)
20-24 12.6 (1.3) 10.4 (1.2) 8.4 (1.1) 35.7 (2.0)
25-29 27.3 (2.4) 27.5 (3.0) 10.6 (1.8) 58.9 (3.5)
30-34 39.9 (3.1) 54.2 (5.6) 10.0 (2.7) 76.3 (5.1)
35-39 55.5 (8.1) 62.5 (7.2) 16.8 (5.4) 84.0 (5.9)
40-50 59.1 (8.9) 73.2 (8.8) 16.3 (4.8) 87.5 (7.3)
Overall prevalence 3.8 (0.2) 3.5 (0.2) 3.4 (0.2) 13.1 (0.4)
Chi square df=6 p<0.0001 441.02 381.11 166.40 848.57

2001 VOL. 25 NO. 6 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 495
Andrews and Slade Article

Table 3: Correspondence between psychological distress as measured by the K10 and the probability of meeting
criteria for a current ICD-10 mental disorder (exclusion criteria operationalised) for respondents scoring in each K10
category.

K10 score Anxiety disorder Affective disorder Substance use Personality Any study
(weighted) prevalence prevalence disorder disorder disorder
% (SE) % (SE) prevalence prevalence prevalence
% (SE) % (SE) % (SE)
10-14 1.1 (0.2) 0.2 (0.1) 1.4 (0.2) 1.7 (0.2) 5.4 (0.4)
15-19 6.8 (0.5) 3.5 (0.4) 3.8 (0.5) 7.1 (0.6) 19.0 (0.8)
20-24 18.7 (1.7) 11.8 (1.3) 8.2 (1.3) 16.4 (1.2) 41.2 (1.7)
25-29 31.8 (3.4) 30.3 (3.4) 9.4 (1.5) 26.5 (3.2) 61.4 (3.6)
30-34 42.1 (3.1) 54.9 (6.2) 10.5 (2.6) 30.2 (4.7) 77.1 (5.0)
35-39 65.1 (7.5) 62.9 (6.6) 17.6 (4.9) 42.2 (5.6) 86.1 (6.1)
40-50 75.0 (7.3) 74.5 (8.7) 15.1 (4.9) 43.6 (7.7) 89.1 (5.5)
Overall prevalence 5.5 (0.3) 3.8 (0.2) 2.8 (0.2) 5.2 (0.2) 13.9 (0.5)
Chi square df=6 p<0.0001 529.82 359.56 187.41 274.66 966.19

consultations for a mental problem in the previous 12 months. cutoff points given in www.crufad.org for clinical decision mak-
These findings support the validity of the K10 as a measure of ing are correct.
psychological distress. The K10 is to be preferred over the GHQ The questions in the K10 focus on anxiety and depression, the
because of the larger range and because the K10 is in the public usual focus of scales that measure psychological distress. There
domain and may be used without charge, whereas the GHQ is are no questions directed at identifying people with psychosis
proprietary and is paid for each time it is used. The K10 is to be simply because identifying people with psychosis by brief ques-
preferred over the SF-12 because it was designed as a measure of tionnaire is difficult. This would be important if the K10 was
distress not disability and because it is simpler to administer and used as an outcome measure in people with psychosis, but when
score. The K10 is to be preferred over the number of consulta- used with people with the common mental disorders this is not a
tions because consulting is a measure of met need, not morbidity. problem. If used to estimate the needs of a population for com-
There is a strong association between a high score on the K10 munity mental health services, the K10 may still be appropriate,
and a current CIDI diagnosis of anxiety and affective disorders simply because people with psychosis do get distressed. The preva-
and a lesser, but still significant association between the K10 and lence of psychosis (0.4%) is so low that estimates based on the
other mental disorder categories, or the presence of any current K10 would not seriously under-estimate population need. There
mental disorder. The associations between the K10 and mental are no data in the present study to inform about the test retest
disorders identified in the previous 12 months are also signifi- reliability of the K10. There are no data to show that the K10 is
cant and similar to the corresponding category of current disor- sensitive to change, whether that is occurring naturally or due to
der. The sensitivity and specificity data make it clear that the K10 the effects of treatment. Both sets of work are required.
is appropriate as a screening instrument to identify likely cases in There have been a number of consequences of the inclusion of
the community but further work is need to determine whether the the K10 in the national survey. The K10 is promoted on a website

Table 4: Correspondence between psychological distress as measured by the K10 from symptoms in the past 30 days
and the probability of meeting criteria for a DSM-IV mental disorder (exclusion criteria operationalised) sometime in the
previous 12 months. Includes current cases for respondents in each K10 category.

K10 score Anxiety disorder Affective disorder Substance use Any study disorder
(weighted) prevalence % (SE) prevalence % (SE) disorder prevalence prevalence
% (SE) % (SE)
10-14 1.4 (0.2) 1.8 (0.2) 4.7 (0.3) 10.4 (0.5)
15-19 6.6 (0.6) 7.8 (0.7) 11.6 (0.8) 29.2 (1.2)
20-24 17.4 (1.1) 19.2 (1.8) 18.2 (1.8) 48.5 (2.3)
25-29 32.1 (3.5) 37.9 (3.4) 17.8 (2.7) 69.4 (3.1)
30-34 47.7 (4.3) 56.8 (5.6) 19.3 (4.6) 82.4 (3.8)
35-39 61.8 (7.0) 68.5 (6.5) 23.3 (4.9) 88.9 (4.6)
40-50 68.1 (8.7) 77.8 (8.2) 34.9 (7.0) 94.0 (4.2)
Overall prevalence 5.6 (0.2) 6.6 (0.3) 7.9 (0.2) 20.2 (0.5)
Chi square df=6, p<0.0001 664.00 446.16 299.86 1298.48

496 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 2001 VOL. 25 NO. 6
Measuring Risk The Kessler Psychological Distress Scale (K10)

Table 5: Correspondence between psychological distress as measured by the K10 from symptoms in the past 30 days
and the probability of meeting criteria for a ICD-10 mental disorder (exclusion criteria operationalised) sometime in the
previous 12 months. Includes current cases for respondents in each K10 category.

K10 score Anxiety disorder Affective disorder Substance use Personality Any study
(weighted) prevalence prevalence disorder disorder disorder
% (SE) % (SE) prevalence prevalence prevalence
% (SE) % (SE) % (SE)
10-14 3.7 (0.3) 2.0 (0.2) 4.3 (0.3) 2.2 (0.2) 12.0 (0.6)
15-19 12.6 (0.7) 9.0 (0.8) 11.6 (0.9) 9.1 (0.7) 33.0 (1.1)
20-24 27.8 (2.0) 21.2 (2.0) 18.3 (1.7) 20.0 (1.7) 55.2 (2.3)
25-29 43.6 (3.8) 41.3 (3.8) 18.6 (2.6) 30.3 (3.5) 74.8 (3.3)
30-34 51.1 (4.6) 57.9 (6.2) 19.3 (4.4) 37.3 (6.0) 84.7 (3.6)
35-39 78.3 (5.0) 69.9 (5.8) 25.4 (5.4) 42.9 (5.8) 91.1 (4.6)
40-50 86.2 (5.8) 79.1 (8.3) 33.3 (7.2) 51.6 (7.2) 99.0 (1.0)
Overall prevalence 9.6 (0.3) 7.2 (0.3) 7.6 (0.3) 6.5 (0.3) 22.7 (0.6)
Chi square df =6, p<0.0001 501.63 351.23 286.96 278.75 1029.88

(www.crufad.org) as a self-report measure to identify need for Table 6: Sensitivity and specificity of the K10 in
treatment. The K10 scale has been chosen for routine public health identifying people who met CIDI criteria for any current
telephone surveys in a number of Australian States, for the ABS anxiety or affective disorder (prevalence 7.1%).
regular survey of Australian health and for routine use in all
K10 score Sensitivity Specificity
patients in contact with the mental health services in NSW. It is greater than (hit rate) correct (rejection
being considered as a routine outcome measure in other States. It or equal to rate)
is widely recommended as a simple measure of outcome follow- 14 0.94 0.63
ing treatment for the common mental disorders. To illuminate these 15 0.90 0.72
uses this paper has presented normative data on the K10 but ad- 16 0.86 0.78
ditional data are required from sources other than the National 17 0.81 0.83
Survey to support the present wide use of the K10. 18 0.77 0.87
19 0.71 0.90
20 0.66 0.92
References 21 0.60 0.94
1. Andrews G, Henderson S, Hall W. Prevalence, comorbidity, disability and
service utilisation: an overview of the Australian National Mental Health 22 0.55 0.95
Survey. Br J Psychiatry 2001,178:145-53. 23 0.50 0.97
2. Mathers C, Vos T, Stevenson C. Burden of Disease and Injury in Australia.
Canberra: Australian Institute of Health and Welfare, 1999. AIHW Cat No.: 24 0.45 0.97
PHE17. 25 0.41 0.98
3. Kessler R, Mroczek D. An Update of the Development of Mental Health Screen-
ing Scales for the US National Health Interview Study [memo dated 12/22/ 26 0.36 0.98
92]. Ann Arbor(MI): Survey Research Center of the Institute for Social Re- 27 0.33 0.99
search, University of Michigan, 1992.
4. Kessler R, Mroczek D. Final Versions of our Non-Specific Psychological Dis- 28 0.31 0.99
tress Scale [memo dated 10/3/94]. Ann Arbor(MI): Survey Research Center 29 0.27 0.99
of the Institute for Social Research, University of Michigan, 1994.
5. Kessler RC, Andrews G, Colpe L, et al. Short Screening Scales to Monitor 30 0.24 0.99
Population Prevalences and Trends in Nonspecific Psychological Distress 31 0.21 1.00
[paper]. Cambridge(MA): Department of Health Care Policy, Harvard Medi-
cal School, 2000 April. 32 0.18 1.00
6. World Health Organization. Composite International Diagnostic Interview.
Version 2.0. Geneva: WHO, 1997.
7. Ware JE, Kosinski M, Keller SD. A 12 item short form health survey. Med
Care 1996;34:220-33.
8. Goldberg DP. The Detection of Psychiatric Illness by Questionnaire. Lon-
don: Oxford University Press, 1972.
9. Shah BV, Barnwell BG, Biegler GS. SUDAAN User’s Manual. Research Tri-
angle Park(NC): Research Triangle Institute, 1997.

2001 VOL. 25 NO. 6 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 497

You might also like