Changes in chronic pain severity over time: the Chronic Pain Gradeas a valid measure

Alison M. Elliotta,*, Blair H. Smitha, W. Cairns Smithb, W. Alastair Chambersc

aDepartment of General Practice, University of Aberdeen, Foresterhill Health Centre, Westburn Road, Aberdeen, AB25 2AY, UKbDepartment of Public Health, University of Aberdeen, Medical School, Foresterhill, Aberdeen, UK

cDepartment of Anaesthetics, Aberdeen Royal In®rmary, Foresterhill, Aberdeen, UK

Received 12 October 1999; received in revised form 1 May 2000; accepted 4 May 2000

Abstract

Our understanding of the natural history of chronic pain in the community is limited. This is partly due to the lack of a validated measure of

chronic pain severity known to be responsive to change over time. The Chronic Pain Grade questionnaire has been shown to be valid and

reliable for use in a general population as a self-completion questionnaire. However, its reliability and validity for use in longitudinal studies

and its responsiveness to change over time has not yet been assessed. We undertook a postal survey designed to test the responsiveness and

the validity of the Chronic Pain Grade questionnaire over time. A random sample of 560 chronic pain patients, aged 25 years and over was

drawn from an existing cohort and strati®ed for age, gender and chronic pain severity. Subjects were re-surveyed by a postal self-completion

questionnaire consisting of the Chronic Pain Grade and the SF-36 general health questionnaire, which is known to be responsive to change in

health over time. To test whether changes in CPG scores correlated with changes in SF-36 scores, Spearman's rank correlation coef®cients

were calculated. A response rate of 86% was achieved for the follow-up study. The majority of SF-36 scores changed in the hypothesized

directions. Changes in CPG scores were signi®cantly correlated with changes in most of the SF-36 domains. We concluded that the CPG is a

useful and valid objective instrument for measuring change in severity of chronic pain over time and could be used in longitudinal studies of

chronic pain severity. q 2000 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved.

Keywords: Chronic pain; Pain measurement; Validation; Natural history; Epidemiology; Survey

1. Introduction

Chronic pain is a common problem in the community. A

recent population study in the UK estimated the prevalence

in the general population to be 47% (Elliott et al., 1999). It is

one of the commonest reasons for attending medical care

and places signi®cant demands on the health service (Von

Korff et al., 1990, 1991). It has been suggested that most

chronic pain is currently assessed and managed inade-

quately (Diamond, 1991). However, attempts at evaluation

of treatment and management are confounded by our rela-

tively poor understanding of the natural history of chronic

pain. Previous studies of the natural history of chronic pain

have tended to be relatively small and/or condition-speci®c

(Potter and Jones, 1992; Magni et al., 1993; Macfarlane et

al., 1996; Croft et al., 1998). A further problem has been

with the research methodologies used. There is currently no

validated measure of change in chronic pain over time. This

has meant that previous studies have had to use unvalidated

measurement instruments or have only been able to identify

how many people still have chronic pain, not how their pain

has changed over time. There is a clear need therefore for a

validated measure of change in chronic pain over time.

The Chronic Pain Grade questionnaire (CPG) (Von Korff

et al., 1992) is a seven-item instrument that measures

chronic pain severity in two dimensions: intensity and

disability. It classi®es patients into ®ve hierarchical grades:

Grade 0 (pain free), Grade I (low disability±low intensity),

Grade II (low disability±high intensity), Grade III (high

disability±moderately limiting) and Grade IV (high disabil-

ity±severely limiting). The CPG was originally validated for

use in the USA in interview-based research on patient

samples representing sufferers of back pain, headache and

temporo-mandibular disorder pain. It was recently tested for

use in the UK when it was found to be acceptable, valid and

reliable for use in a general population as a self-completion

postal questionnaire (Smith et al., 1997). However, the relia-

bility and validity of the CPG for use in longitudinal studies

has not yet been assessed and its responsiveness to change

Pain 88 (2000) 303±308

0304-3959/00/$20.00 q 2000 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved.

PII: S0304-3959(00)00337-7

www.elsevier.nl/locate/pain

* Corresponding author. Tel.: 144-1224-554228; fax: 144-1224-

840683.

E-mail address: a.m.purves@abdn.ac.uk (A.M. Elliott).

over time needs to be tested. The questionnaire is easy to

complete and its brevity makes it an attractive instrument if

its use is appropriate. If shown to be valid the CPG would be

a useful instrument for use in studies of chronic pain varia-

bility and the effectiveness of different interventions.

Chronic pain has deleterious effects on general health

(Becker et al., 1997). As well as impairing physical func-

tioning chronic pain also has detrimental effects on social

and psychological well-being (Gureje et al., 1998). The SF-

36 general health questionnaire (Garratt et al., 1993; Ware et

al., 1993) has been extensively tested. A recent study look-

ing at whether the SF-36 is a valid measure of change in

population health found that each of the eight dimensions of

health measured by the SF-36 changed in hypothesized

directions with age, employment status and disease status.

These changes occurred within a mean follow-up period of

36 months in a general population cohort, which had not

been the subject of intervention. The authors concluded that

the SF-36 was valid and sensitive in studies measuring

change over time (Hemingway et al., 1997). While the

SF-36 provides useful information on changes in general

health over time, a validated measure of changes in chronic

pain severity is also needed for studies of the natural history

of chronic pain.

We describe here the results from a prospective study

designed to test the responsiveness and validity of the

Chronic Pain Grade questionnaire over time in a group of

chronic pain patients from a general population.

2. Methods

2.1. Subjects and sampling

The study was set in the Grampian region of North East

Scotland. A random sample of 560 chronic pain patients

(individuals with pain or discomfort for longer than 3

months) was drawn from an existing cohort (n � 6940)

established for a previous study (Elliott et al., 1999). The

sample was strati®ed for sex, age (25±59 or 601 years) and

chronic pain severity (mild chronic pain (CPG I and II) or

severe chronic pain (CPG III and IV)). Lists of sampled

patients were then screened by their own general practi-

tioner to preclude inappropriate or insensitive inquiry (for

example in the case of terminal illness).

2.2. Data collection

Subjects were surveyed by postal self-completion ques-

tionnaire, which consisted of the CPG and an anglicized

version of the SF-36 general health questionnaire. Subjects

were surveyed twice in total. The ®rst administration

formed part of the original study in September 1996 (Elliott

et al., 1999). The second administration took place 18

months later in March 1998. For each mailing patients

were sent a copy of the questionnaire accompanied by a

letter signed by both their own general practitioner and

the general practitioner in the research group. Up to two

reminders were sent at fortnightly intervals to maximize

response rates.

2.3. Assessing validity of the CPG

One method of testing the validity of a new measure is to

test its concurrent validity. This involves correlating a new

measure against one that has already been validated and is

an accepted measure by administering them simultaneously.

In previous work CPG scores were found to correlate

closely with SF-36 scores at a single point in time (Penny

et al., 1999). To determine whether the CPG is a valid

measure of change over time it was compared to the SF-

36 to allow assessment of its concurrent validity.

2.4. Hypotheses

It was hypothesized that median SF-36 scores would

decrease as Chronic Pain Grades became more severe,

increase as Chronic Pain Grades became less severe, and

remain constant when there was no change in the Chronic

Pain Grade. It was also hypothesized that the changes in

Chronic Pain Grades should correlate with changes in the

SF-36 scores in the majority of the domains, in particular the

Bodily Pain, Physical Functioning and Role Physical

domains.

3. Results

3.1. Response rate

Of the 560 individuals sampled, GPs excluded 25 patients

as unsuitable for survey. Five hundred and thirty-®ve ques-

tionnaires were therefore mailed. Four people had died since

the sample had been drawn. Ten people had moved house

and so their questionnaires were returned to us by the Royal

Mail. Seven people were unable to complete their question-

naires due to ill health, current hospitalization, etc. We were

noti®ed of these individuals through the patients' families.

A total of 450 questionnaires were returned completed (see

Fig. 1). This represents a corrected response rate of 87.5%.

There was no signi®cant difference in response rate between

males (85.9%) and females (89.2%) (x2�1d:f:� � 1:29,

P � 0:256) or between the 25±59 age group (84.8%) and

the 601 age group (90.4%) (x2�1d:f:� � 3:76, P � 0:053).

3.2. Changes in Chronic Pain Grades and median SF-36

scores

Table 1 details the baseline and follow-up median SF-36

scores for each of the changes in the CPG, and Table 2

shows the change in SF-36 scores and the direction of

change. For the majority of the changes in Chronic Pain

Grades median SF-36 scores changed in the direction

hypothesized, i.e. SF-36 scores increased as severity

decreased and decreased as severity increased. Scores

A.M. Elliott et al. / Pain 88 (2000) 303±308304

were most likely to change in the hypothesized direction in

the domains of Social Functioning and General Health and

least likely to change in the hypothesized direction in the

domains of Role Emotional and Mental Health.

Some of the SF-36 scores changed dramatically with

changes in the CPG. The biggest changes in scores were

in the domains of Role Physical, Social Functioning, and

Bodily Pain. In contrast some of the domains showed little

change with time, e.g. Role Emotional and Mental Health.

The majority of scores in the Role Emotional domain did

not change at all between baseline and follow-up, even with

considerable change in the Chronic Pain Grade, e.g. a

change from Grade I to Grade IV and from Grade IV to

Grade I.

Fig. 2 shows a summary of the changes in SF-36 scores

based on whether the CPG increased, decreased or remained

constant over time. The baseline SF-36 scores of these three

groups differ in different ways across the domains. For most

of the SF-36 domains there was little difference in baseline

scores between those whose Chronic Pain Grade was going

to increase, stay the same, or decrease. Kruskal±Wallis test-

ing showed a signi®cant difference in the baseline SF-36

scores between the three groups in two of the domains,

namely Role Physical (P � 0:01) and Social Functioning

(P � 0:001). In the domain of Role Physical for example,

the baseline SF-36 score for those whose CPG was going to

increase or stay the same was 50. For those whose CPG was

going to decrease the baseline SF-36 score was 0.

3.3. Correlation between the CPG and SF-36

At baseline there was a signi®cant negative correlation

between the SF-36 and the CPG. Negative correlations

between the measures were found since high scores on the

SF-36 dimensions are associated with well being, whereas

high scores on the CPG indicate severe chronic pain. To test

whether changes in CPG scores correlated with changes in

SF-36 scores Spearman's rank correlation coef®cients were

calculated and are shown in Table 3. All changes in CPG

scores were signi®cantly correlated with changes in SF-36

scores except for the Role Emotional and Mental Health

domains. The largest correlations were with the Bodily

Pain and Role Physical domains as hypothesized.

Although there was a signi®cant correlation between

changes in the CPG and changes in the Role Physical

domain this may not be a true correlation. The baseline

median Role Physical score for those whose CPG was

going to decrease was 0, making it impossible for this

group to report a decrease in their Role Physical score.

The increase in score seen in this group may therefore

simply be the result of regression to the mean. In order to

determine if the correlation between changes in the CPG

and changes in Role Physical scores is valid or is the result

of the increase in Role Physical score seen in this group, a

second Spearman's correlation was carried out. For this

analysis only individuals whose CPG increased or stayed

the same were included (n � 309). The resulting correlation

was 20.170 and was signi®cant at the 0.1 level (two-tailed).

Although the magnitude of the correlation is quite small it is

a signi®cant correlation and supports the initial ®nding of a

signi®cant correlation between changes in the CPG and

changes in the Role Physical domain.

4. Discussion

It was hypothesized that median SF-36 scores would

decrease as Chronic Pain Grades became more severe,

increase as Chronic Pain Grades became less severe, and

remain constant when there was no change in the Chronic

Pain Grade. Most of the SF-36 scores changed in these

hypothesized directions with changes in the Chronic Pain

Grade. The biggest changes in scores were in the domains of

Role Physical, Social Functioning, and Bodily Pain, while

the smallest changes in score were in the domains of Role

Emotional and Mental Health. Our ®ndings are consistent

with those presented by Hemingway et al. (1997) in their

paper looking at changes in SF-36 scores over time. They

found the biggest changes were in the domain of Role Physi-

cal and the smallest changes in the domain of Mental

Health.

It was also hypothesized that changes in Chronic Pain

A.M. Elliott et al. / Pain 88 (2000) 303±308 305

Fig. 1. Diagram of patient ¯ow of individuals sampled for survey.

A.M. Elliott et al. / Pain 88 (2000) 303±308306

Tab

le1

Bas

elin

ean

dfo

llo

w-u

pm

edia

nS

F-3

6sc

ore

sb

yC

PG

chan

ges

bet

wee

n1996

and

1998

a

Ch

ang

ein

CP

G

nP

hysi

cal

Fu

nct

ion

ing

So

cial

Fu

nct

ion

ing

Role

Physi

cal

Role

Em

oti

onal

Men

tal

Hea

lth

Ener

gy

and

Vit

alit

yB

odil

yP

ain

Gen

eral

Hea

lth

Bas

elin

e

SF

-36

Fo

llow

-up

SF

-36

Bas

elin

e

SF

-36

Fo

llo

w-u

p

SF

-36

Bas

elin

e

SF

-36

Foll

ow

-up

SF

-36

Bas

elin

e

SF

-36

Foll

ow

-up

SF

-36

Bas

elin

e

SF

-36

Foll

ow

-up

SF

-36

Bas

elin

e

SF

-36

Foll

ow

-up

SF

-36

Bas

elin

e

SF

-36

Foll

ow

-up

SF

-36

Bas

elin

e

SF

-36

Foll

ow

-up

SF

-36

I±0

61

00

97

.59

3.7

51

00

100

100

100

83.3

386

80

55

77.5

78

100

64.5

82

I±I

10

85

85

10

01

00

100

100

100

100

84

84

65

65

67

62

72

72

I±II

19

75

65

10

08

7.5

100

50

100

100

80

80

65

55

52

42

67

57

I±II

I1

76

06

01

00

75

37.5

25

100

100

84

84

55

60

62

41

67

57

I±IV

85

2.5

35

87

.53

1.2

587.5

066.6

766.6

770

66

52.5

35

57

26.5

53.1

340

II±

01

85

94

88

10

0100

25

100

100

68

64

85

60

61

100

82

92

II±

I2

47

58

08

7.5

10

050

87.5

100

100

78

78

52.5

62.5

51

62

62

72

II±

II3

46

57

57

58

7.5

75

75

100

100

78

78

50

50

41.5

41

67

67

II±

III

12

55

52

.58

7.5

56

.25

37.5

25

100

50

78

72

40

50

31

32

52

42

II±

IV6

52

.54

06

2.5

43

.75

25

12.5

83.3

333.3

370

70

45

10

41

31

37.5

22.5

III±

05

75

90

87

.51

00

75

100

100

100

88

92

65

70

52

100

57

80

III±

I1

76

57

2.5

62

.57

525

75

66.6

758.3

364

72

45

60

41

62

50

72

III±

II1

85

55

7.5

50

62

.525

33.3

3100

83.3

364

74

40

50

41

41

45

58.5

III±

III

28

40

37

.56

2.5

50

00

33.3

333.3

362

68

40

40

41

31.5

35

42

III±

IV2

13

2.5

30

50

37

.50

00

33.3

360

52

35

30

31

31

35

32.5

IV±

00

±±

±±

±±

±±

±±

±±

±±

±±

IV±

I1

75

06

06

2.5

87

.50

50

100

100

72

80

45

50

31

64

46

72

IV±

II1

44

55

55

6.2

56

8.7

50

37.5

100

100

68

76

40

40

31

41

41

52

IV±

III

27

20

22

.22

37

.55

00

033.3

333.3

364

68

30

35

31

32

40

35

IV±

IV6

41

51

52

52

50

00

056

50

20

25

22

22

25

25

aF

or

bo

ldv

alu

esth

eS

F-3

6sc

ore

sch

ang

edin

the

hy

poth

esiz

eddir

ecti

on

bet

wee

nbas

elin

ean

dfo

llow

-up

bas

edon

chan

ges

inth

eC

PG

.

Grades would correlate with changes in the SF-36 scores in

the majority of the domains. Changes in CPG scores were

found to correlate with changes in SF-36 scores for all of the

domains except Role Emotional and Mental Health. The

A.M. Elliott et al. / Pain 88 (2000) 303±308 307

Table 2

Change in median SF-36 scores by CPG changes between 1996 and 1998a

Change in

CPG

n Physical

Functioning

Social

Functioning

Role

Physical

Role

Emotional

Mental

Health

Energy and

Vitality

Bodily

Pain

General

Health

I±0 6 22.5 6.25 0 216.67 26 22.5 22 17.5

I±I 10 0 0 0 0 0 0 25 0

I±II 19 210 212.5 250 0 0 210 210 210

I±III 17 0 225 212.5 0 0 5 221 210

I±IV 8 217.5 256.25 287.5 0 24 217.5 230.25 213.13

II±0 1 9 12 275 0 24 225 39 10

II±I 24 5 12.5 37.5 0 0 10 11 10

II±II 34 10 12.5 0 0 0 0 20.5 0

II±III 12 22.5 231.25 212.5 250 26 10 1 210

II±IV 6 212.5 218.75 212.5 250 0 235 210 215

III±0 5 15 12.5 25 0 4 5 48 23

III±I 17 7.5 12.5 50 28.34 8 15 21 22

III±II 18 2.5 12.5 8.33 216.67 10 10 0 13.5

III±III 28 22.5 212.5 0 0 6 0 29.5 7

III±IV 21 22.5 212.5 0 33.3 28 25 0 22.5

IV±0 0 ± ± ± ± ± ± ± ±

IV±I 17 10 25 50 0 8 5 33 26

IV±II 14 10 12.5 37.5 0 8 0 10 11

IV±III 27 2.2 12.5 0 0 4 5 1 25

IV±IV 64 0 0 0 0 26 5 0 0

a For bold values the SF-36 scores changed in the hypothesized direction between baseline and follow-up based on changes in the CPG.

Fig. 2. Baseline and follow-up median SF-36 scores based on changes in Chronic Pain Grades between 1996 and 1998.

fact that there were small changes in these domains and that

they do not correlate with changes in the CPG may suggests

that the CPG is insensitive to change in certain dimensions

of health status. More realistically it suggests that certain

domains of the SF-36 are not relevant to change in reported

pain experience.

Since most of the SF-36 scores changed in the direction

hypothesized with changes in the CPG over time, and

changes in the CPG and SF-36 scores correlate with each

other, we can infer that changes in Chronic Pain Grades are

not merely due to ¯uctuations but are showing true change

with time.

In order to improve the effectiveness of the treatment and

management of chronic pain, our understanding of the

natural history of chronic pain needs to be improved.

Large-scale longitudinal studies of community-based

cohorts are therefore required. This study has demonstrated

that the CPG is a valid objective instrument for measuring

change in severity of chronic pain over time and could be

used in longitudinal studies of chronic pain severity to

inform our understanding of this important problem.

Acknowledgements

The authors gratefully acknowledge Tenovus, Scotland

for their ®nancial support of the work. We thank Dr Michael

Von Korff for his support and advice on the Chronic Pain

Grade questionnaire. Finally, we would also like to thank

the patients and general practices whose support and work

made this study possible.

References

Becker N, Thomsen AB, Olsen AK, Sjogren P, Bech P, Eriksen J. Pain

epidemiology and health related quality of life in chronic non-malig-

nant pain patients referred to a Danish multi-disciplinary pain center.

Pain 1997;73:393±400.

Croft PR, Macfarlane GJ, Papageorgiou AC, Thomas E, Silman AJ.

Outcome of low back pain in general practice: a prospective study.

Br Med J 1998;316:1356±1359.

Diamond A. The future development of chronic pain relief. Anaesthesia

1991;46:83±84.

Elliott AM, Smith BH, Penny KI, Smith WC, Chambers WA. The epide-

miology of chronic pain in the community. Lancet 1999;354:1248±

1252.

Garratt AM, Ruta DA, Abdalla MI, Buckingham JK, Russell IT. The SF-36

health survey questionnaire: an outcome measure suitable for routine

use within the NHS? Br Med J 1993;306:1440±1441.

Gureje O, Von Korff M, Simon GE, Gater R. Persistent pain and well-being.

A World Health Organisation study in primary care. J Am Med Assoc

1998;280:147±151.

Hemingway H, Stafford M, Stans®eld S, Shipley M, Marmot M. Is the SF-

36 a valid measure of change in population health? Results from the

Whitehall II study. Br Med J 1997;315:1273±1279.

Macfarlane GJ, Thomas E, Papageorgiou AC, Schollum J, Croft PR, Silman

AJ. The natural history of chronic pain in the community: a better

prognosis than in the clinic? J Rheumatol 1996;23:1617±1620.

Magni G, Marchetti M, Moreschi C, Merskey H, Luchini SR. Chronic

musculoskeletal pain and depressive symptoms in the National Health

and Nutrition Examination I. Epidemiologic follow-up study. Pain

1993;53:163±168.

Penny KI, Purves AM, Smith BH, Chambers WA, Smith WC. Relationship

between the chronic pain grade and measures of physical, social and

psychological well-being. Pain 1999;79:275±279.

Potter RG, Jones JM. The evolution of chronic pain among patients with

musculoskeletal problems: a pilot study in primary care. Br J Gen Pract

1992;42:462±464.

Smith BH, Penny KI, Purves AM, Munro C, Wilson B, Grimshaw J, Cham-

bers WA, Smith WC. The chronic pain grade questionnaire: validation

and reliability in postal research. Pain 1997;71:141±147.

Von Korff M, Dworkin SF, Le Resche L. Graded chronic pain status: an

epidemiologic evaluation. Pain 1990;40:279±291.

Von Korff M, Wagner EH, Dworkin SF, Saunders KW. Chronic pain and

use of ambulatory health care. Psychom Med 1991;53:61±79.

Von Korff M, Ormel J, Keefe FJ, Dworkin SF. Grading the severity of

chronic pain. Pain 1992;50:133±149.

Ware JE, Snow KK, Kosinski M, Gandek B. SF 36 health survey. Manual

and interpretation guide, Boston, MA: The Health Institute, 1993.

A.M. Elliott et al. / Pain 88 (2000) 303±308308

Table 3

Correlation between the change in CPG score and change in SF-36 score

Change in SF-36 scores Change in Chronic Pain Grade

Physical Functioning 20.293*a

Social Functioning 20.320*

Role Physical 20.342*

Role Emotional 20.051

Mental Health 20.077

Energy and Vitality 20.211*

Bodily Pain 20.420*

General Health 20.282*

a An astersik indicates correlations signi®cant at the 0.01 level (two-

tailed).