Journal of Behavioral Medicine

, Volume 32, Issue 2, pp 162–173

A measure for quality of life assessment in chronic pain: preliminary properties of the WHOQOL-pain

Authors

    • Department of Psychology, WHO Field Centre for the Study of Quality of LifeUniversity of Bath
    • Psychological SciencesUniversity of Worcester
  • S. M. Skevington
    • Department of Psychology, WHO Field Centre for the Study of Quality of LifeUniversity of Bath
  • M. Osborn
    • Palliative Care Team, Royal United Hospital
Article

DOI: 10.1007/s10865-008-9187-y

Cite this article as:
Mason, V.L., Skevington, S.M. & Osborn, M. J Behav Med (2009) 32: 162. doi:10.1007/s10865-008-9187-y

Abstract

Chronic pain has a considerable impact on patient-reported outcomes such as quality of life (QoL). To assess QoL in people with chronic pain, a pain and discomfort module (PDM) was developed for use with the WHOQOL-100 and its psychometric properties assessed. Sixteen items covered four facets on pain relief; anger and frustration; vulnerability, fear and worry; and uncertainty. Chronic low back pain patients (n = 133) (age 56; pain duration 85 months; 65% female) completed the WHOQOL-100 and PDM, McGill Pain Questionnaire, and SF-12. The PDM showed good internal consistency reliability (α = .88) (alphas .66–.81). Except for anger, all facets associated most strongly with their ‘parent’ domain. Pain and poorer QoL were strongly associated, supporting construct validity. The SF-12 physical health component associated strongly with pain relief, and the mental health score with other facets, indicating good concurrent validity. Discriminant validity tests showed that PDM scores distinguished between ill and well patients, and between those reporting different health statuses. The PDM has fair to good psychometric properties indicating its value as a patient-reported outcome measure for clinical trials.

Keywords

Chronic low back painQuality of lifeWHOQOL-100WHOQOL-painReliabilityValidity

Introduction

Assessing subjective outcomes such as quality of life (QoL) is an essential component of evaluating health care outcomes in chronic conditions and the WHOQOL group have contributed extensively to addressing this aim (e.g. Skevington and Wright 2001; Ackerman et al. 2006). Chronic pain not only has a considerable burden on the community (Brooks 2006) and health care (Shipton and Tait 2005), but also impacts widely on a person’s physical, psychological and social well-being (Skevington 1998). Instruments to measure this impact should be meaningful and acceptable to patients, in addition to having good psychometric properties. Instruments tend to be either generic, such as the SF-36 (Ware and Sherbourne 1992) and its abbreviated form the SF-12 (Ware et al. 1994) or condition-specific, for example the Cervical Spine Outcomes Questionnaire (CSOQ) (BenDebba et al. 2002) or the Arthritis Impact Measurement Scale (AIMS) (Meenan et al. 1980). These instruments are widely used and have been applied to, and validated in a variety of contexts. However, instruments assessing the specific impact on the QoL of patients presenting with a range of chronically painful conditions are fewer in number. Furthermore, despite its dominance in the literature, and the frequency with which studies report the use of instruments such as the SF-36 as assessing QoL, it assesses health status rather than QoL per se (Bowden and Fox-Rushby 2003). Consequently, this limits the range of factors that are assessed and does not provide a comprehensive assessment beyond patient functioning.

Until recently, few randomised controlled trials testing the efficacy of interventions to treat and manage pain included QoL as an outcome measure. Since 2003 however, the US Food and Drugs Administration required QoL as a secondary outcome measure in all clinical trials (Apolone 2003). This is consistent with a movement towards measuring QoL as a patient-reported outcome, rather than focusing on other outcomes such as depression and anxiety. Whilst these have clear advantages in terms of assessing changes in symptoms for example, they only focus on one or two aspects of patient QoL rather than taking a more comprehensive view. Such a comprehensive view is assessed through the use of QoL instruments which allow the broad subjective views of patients to be considered.

The WHOQOL instrument was developed to assess subjective QoL across 25 broad facets pertaining to different life dimensions and is scored in 6 domains (physical; psychological; level of independence; social relationships; environment; spirituality) (WHOQOL Group 1998a, b). This instrument is underpinned by the definition of QoL as “an individual’s perception of their position in life, in the context of their culture and value systems in which they live and in relation to their goals, expectations, standards and concerns” (WHOQOL Group 1998a, b). Although one of the 25 facets addresses pain and discomfort, this facet was found to under represent the impact of pain on QoL (Skevington 1998). In response to this, and given the potential advantages of appending condition or disease-specific modules to a generic core instrument, the PDM was developed to address the specific impact of chronic pain on the lives of patients (Mason et al. 2004, 2008). Whilst there are notable advantages of generic assessment, there are particular advantages of being able to assess both generic and specific QoL dimensions simultaneously. For example it is possible to compare QoL across different conditions and between people who are ill and well whilst also benefiting from assessing the specific impact of a condition by obtaining more relevant and sensitive measurements. The rationale for developing a pain-specific module was that it would improve measurement by enhancing the specificity and sensitivity of the core instrument when used with patients with chronic pain.

Whilst assessing dimensions of the pain itself such as its intensity, severity or location are important (with, for example the McGill Pain Questionnaire, Melzack 1987), measuring the impact of pain on QoL is part of ensuring comprehensive assessment of the pain patient to enable appropriate care and management to be provided. In view of this, the novelty of the PDM lies in its ability to capture the specific consequences of the impact of pain. Furthermore, whilst the instrument was developed for use with a range of patients with chronic pain, it is anticipated that its widest application might be with patients with musculoskeletal pain given the prevalence of this in relation to other chronically painful conditions or syndromes (Elliott et al. 1999).

Given the need to establish the validity of new and existing instruments, condition or symptom-specific outcome measures can be administered concurrently to provide a means of assessing criterion validity (Juniper et al. 1996). Additionally, in the absence of a ‘gold standard’ instrument, it is possible to use global ratings such as the core WHOQOL items assessing general QoL and health to provide support for construct validity (Juniper et al. 1996) to ensure that the concepts being measured are actually assessing QoL. Being able to distinguish between different groups of patients known to differ on particular characteristics is also an important aspect of demonstrating the validity of instruments (Campbell 1960). For example, the ability to discriminate between patients reporting different levels of pain, functioning or health is a distinct advantage particularly in the context of demonstrating treatment effectiveness in patients with chronic pain.

The aim of this study was to examine the psychometric properties of a pain and discomfort module used in conjunction with the WHOQOL-100 to assess QoL relating to chronic pain. Specifically, we report on the internal consistency reliability and construct, concurrent and discriminant validity of the instrument.

Method

Design

This study used a cross-sectional, observational design to investigate consecutive admissions to two clinics to determine the reliability and validity of a pain and discomfort module designed to be used in conjunction with the core WHOQOL-100.

Sample

Adult patients over the age of 18 with a diagnosis of low back pain or sciatica were recruited from two district general hospitals in South-West England. Two hundred and twenty-eight consecutive patients attending consultant and nurse led clinics at site 1 were invited to participate (an outpatient’s pain clinic, staffed by a multidisciplinary team, specialising in chronic pain and its management). At site 2, 30 new patients attending the clinic were invited to take part (a day surgery clinic where a consultant rheumatologist carries out lumbar epidural steroid injections). Patients were excluded from the study by the clinical psychologist (MO) if they were known to have a personality disorder or substance abuse. For different reasons, reliable responses to these subjective measures could be distorted by these conditions. This screening was done prior to patient selection for recruitment and numbers of patients excluded on the basis of these criteria were not recorded. It is not known how many of the patients included in the study had anxiety or depression.

Procedure

Patients received a letter from the clinic inviting them to attend an appointment with the consultant, with an additional letter inviting them to participate in the study. Two weeks prior to their appointment, patients were sent an information sheet, consent form and questionnaire battery (see below) and requested to bring the completed questionnaire to their appointment (patients were subsequently followed-up and this will be reported elsewhere). Patients were assured of confidentiality and anonymity, and could decline to take part. Ethical approval was granted by the Local Research Ethics Committees and participating patients provided written informed consent.

Measures

WHOQOL-pain

WHOQOL

The UK WHOQOL-100 is represented by 25 facets of QoL (for example, energy and fatigue; self esteem; mobility; social support; recreation and leisure; spirituality) which are subsumed within six broad domains of QoL (physical; psychological; level of independence; social relationships; environment; spirituality). It contains 102 items (100 items (four per facet) with two additional national items) and 26 importance items which rate the importance of separate facets of QoL (Skevington 1999). To score the instrument, 41 of 102 core items require reverse scoring (Szabo et al. 1997). Twenty-five facet scores are calculated from the mean of each of the 4-item facets. Six domain scores are calculated from the mean of facet scores representing each domain. Its psychometric properties have been reported previously (The WHOQOL Group 1995; 1998a, b) and it has been validated for use with chronic pain patients (Skevington 1998; Skevington et al. 2001).

PDM

The PDM was developed following focus groups and a web survey of people with chronic pain who derived 68 items representing 10 facets (Mason et al. 2004). Subsequent pilot testing using cognitive interviewing and a cross-sectional survey reduced the number of items to 16 representing four facets (Mason et al. 2008). It contains four facets of QoL; pain relief; anger and frustration; vulnerability, fear and worry; uncertainty. The WHOQOL-pain therefore consists of 118 items: the core WHOQOL-100 (102 items), plus the PDM (16 items). Patients also completed 31 importance items: 26 from the core instrument and the importance of the four new facets of the PDM (5 items—two representing the dual components of pain relief, and one each representing the remaining three facets).

Consistent with the core instrument, the PDM is scored by calculating facet means and 11 of the 16 PDM items require reverse scoring so that high scores always represent a good QoL. The qualitative work showed that conceptually, pain relief belongs to the physical domain; anger and frustration and vulnerability, fear and worry to the psychological domain; and uncertainty with the independence domain. On a Likert scale of 1–5, the four PDM facets have been rated as important or very important: pain relief (4.5); anger and frustration (4.2); vulnerability/fear/worry (4.2); uncertainty (4.0), providing further evidence for their inclusion in the module and this is reported elsewhere (Mason et al., in preparation). Although the completion time was not measured, the 16 PDM items and 5 importance items took approximately 10 min to complete and they are presented in Appendix 1.

Short-form 12 (SF-12)

The SF-12 is an abbreviated form of the SF-36 Health Status Questionnaire, consisting of a 12-item sub-set of questions, which can provide two summary scores: physical (PCS) and mental health component scores (MCS). Although instrument precision is compromised due to single items (Ware et al. 1996) this is offset by minimal respondent burden, as it only takes 2–3 min to complete (Ware et al. 1994; Sturgis et al. 2001; Jenkinson and Layte 1997). It has been validated for use in patients with low back pain (Deyo et al. 1998), osteoarthritis (Theiler et al. 2002) and Ankylosing Spondylitis (Haywood et al. 2002). Four of the 12 items require reverse scoring so that a high score always represents best health.

Short-form McGill Pain Questionnaire (SF-MPQ)

The SF-MPQ (Melzack 1987) was derived from the long form of the McGill Pain Questionnaire (LF-MPQ, Melzack 1975). The instrument can be used with adults with pain in multiple settings and was based on the earlier work of Melzack and Torgerson (1971). It captures dimensions of the pain experience including its quality and intensity, and enables quick assessment (Melzack and Katz 1994, 1999). The SF-MPQ contains 15 pain words and has been shown to correlate highly with LF-MPQ scores (Melzack 1987; Dudgeon et al. 1993) and to be sensitive to clinical changes brought about by analgesics (Melzack 1987; Harden et al. 1991). Although it was designed to distinguish between different pain syndromes (Melzack 1987), this has not been reliably demonstrated (Melzack and Katz 1999). It takes 2–5 min to complete (Wilkie et al. 1990), and therefore minimises respondent burden. Five scores were derived from the SF-MPQ: the Sensory Pain Rating Index (S-PRI) (11 pain adjectives), the Affective Pain Rating Index (A-PRI) (4 adjectives), the Total Pain Rating Index (T-PRI) (sum of S-PRI and A-PRI), the Present Pain Intensity-Visual Analogue Scale (PPI-VAS) (VAS from no pain to worst possible pain scored by dividing the line into 10 × 1 cm sections), and the evaluative overall intensity of total pain experience (scored 0–5, where 0 is no pain and 5 is excruciating).

Analysis

The analysis aimed to determine the psychometric properties of the PDM. Overall, it was predicted that the PDM would demonstrate good internal consistency reliability (Cronbach’s alpha (α) = 0.70–0.90) to show the cohesiveness of the items representing each facet. To demonstrate construct validity, it was predicted that the PDM would be significantly negatively associated with pain (−0.4 to −0.8) (Streiner and Norman 1995), as assessed by the SF-MPQ, so that more severe or intense pain would be associated with a poorer QoL. To demonstrate concurrent validity it was hypothesised that the PDM would be significantly positively associated with health status, as measured by the SF-12 and with the general QoL and health facet of the core WHOQOL. It was also expected that the PDM would be able to distinguish between those defining themselves as ill or well, and between people with different reported levels of health, where higher scores on the PDM (or good QoL) would be associated with being well and having good health, and vice versa. The psychometric properties of the core instrument have been reported elsewhere (Skevington et al. 1999; Skevington 1999). The purpose of this study was not to describe the QoL of this group of chronic pain patients per se, but to test the performance of the PDM facets and the properties of the instrument.

Results

One hundred and thirty-three questionnaires (response rate 52%) were returned by patients: 118 out of 228 from site 1 (52% response rate) and 15 out of 30 from site 2 (50% response rate). Mean age was 55.9 (SD 16.87, range 17–92) and most of the participants were female (65.4%), married (57.1%), educated to secondary level (48.9%) and around a third were currently employed (32.3%). The majority described themselves as being in poor health (40.6%) and ill (51.1%). Patients came from a range of diagnostic groups and pain was most commonly reported in the lower back and spine (87.2%). Mean duration of pain was 84.6 months (SD 120.9, range 3–720) and patients most often described their pain as continuous (70.7%) (followed by intermittent 23.3%; brief 2.3%) and discomforting (33.1%) (followed by distressing 26.3%; horrible 24.8%; excruciating 7.5%; mild 3.8%; no current pain 0.8%) in nature. Patients from both sites were comparable in terms of age, health, education, marital status, pain duration or intensity in that no significant differences were found between these (p > .21).

Data quality

To ensure that ceiling and floor effects are not observed in PDM items, the whole range of scores should be used, so that at least 10% of participants use each of the five scale points. Where there is a low percentage for 5 (highest QoL), this suggests a floor effect which means that respondents may be treating the scale as a four-point scale, and improvements may not be detected if they perceive the verbal label for the 5-anchor point as too extreme. The reverse is true for items where 1 (poorest QoL) has not been used. Of the 16 PDM items, satisfaction with control of pain, being able to find a comfortable position, and concern about experiencing pain had no responses at ‘5’, showing that the anchor point representing the best QoL for these items was not being utilised. The item about the extent to which vulnerability interfered had no response for ‘1’, showing that the anchor point representing the poorest QoL was not used for this item. The most frequently missed question was about the extent to which having treatment has improved QoL (pain relief facet).

Reliability of the PDM

Mean PDM facet ratings are shown in Table 2. QoL was highest for vulnerability/fear/worry, followed by anger and frustration, then uncertainty, and pain relief. Internal consistency reliability for the PDM was found to be good (standardised item α = .88). For the PDM facets, anger and frustration (α = .81) and uncertainty (α = .79) demonstrated good internal consistency reliability. Pain relief (α = .66) and vulnerability/fear/worry (α = .67) were only marginally acceptable.

Inter-item correlations

For pain relief, inter-item correlations ranged from .12 to .43 (Table 1). Cronbach’s α was lowest with the removal of the item asking about satisfaction with control of pain, reflecting the important contribution of this item to the reliability of this facet. For the anger and frustration facet, inter-item correlations ranged from .39 to .72 and α was lowest with the removal of the item asking how often pain makes a person feel angry. For vulnerability/fear/worry, inter-item correlations ranged from .10 to .64 and α was lowest with the removal of the item asking how much fear bothers a person. For uncertainty, inter-item correlations ranged from .34 to .70 and α was lowest with the removal of the item asking about difficulty planning.
Table 1

Ranked mean facet ratings, inter-item correlation range and Cronbach’s α for the PDM

Facet

Mean (SD)

Inter-item correlation range

α

Vulnerability/fear/worry

13.14 (2.74)

.10–.64

.67

Anger & frustration

12.00 (3.26)

.39–.72

.81a

Uncertainty

11.08 (3.19)

.34–.70

.79a

Pain relief

10.20 (2.44)

.12–.43

.66

aAcceptable α levels

The inter-item correlations for anger and frustration, and uncertainty were acceptable, although the relatively low inter-item correlations for pain relief and vulnerability/fear/worry explain the low facet α levels. The relative strength of these correlations reflects the cohesiveness of the underlying constructs and suggests that anger and frustration and uncertainty are more internally consistent than the pain relief and vulnerability/fear/worry facets.

Item-facet correlations

All item-facet correlations exceeded .4 with the exception of two items from pain relief and three from vulnerability/fear/worry (Table 2). The two items within the pain relief facet were marginally low (.36 and .38) and the item within the vulnerability facet was low (.25), suggesting that this item was least related to its parent facet. For item-total correlations, all but one item was less than .4 for pain relief (.28) and vulnerability/fear/worry (.29), suggesting that these items were least related to the whole PDM scale. However, the change on overall α when these items were deleted was negligible.
Table 2

Item-facet and item-total correlations and Cronbach’s α for the PDM facets

Domain, facet and item

Item-facet correlation

α if item deleted

Item-total correlation

α if item deleted

Physical domain—pain relief facet

.52

.80

Treatment improved QoL

.38

.64

.28

.89

Satisfied control of pain

.57

.49

.45

.88

Cope with level of pain

.36

.64

.46

.88

Comfortable

.46

.57

.45

.88

Psychological domain—anger & frustration facet

.66

.73

Anger interfere

.64

.76

.58

.87

Frustration interfere

.57

.79

.62

.87

Pain angry

.68

.73

.60

.87

Pain irritable

.63

.76

.62

.87

Psychological domain—vulnerability/fear/worry facet

.52

.80

Vulnerability interfere

.52

.58

.52

.88

Fear bother

.62

.50

.55

.87

Worry about treatment

.48

.61

.29

.88

Concern experiencing pain

.25

.73

.36

.88

Level of independence domain—uncertainty facet

.80

.65

Uncertainty interfere

.61

.73

.72

.87

Difficulty planning

.75

.65

.70

.87

Pain limit life

.51

.77

.63

.87

Satisfied make future plans

.52

.77

.58

.87

Item-domain correlations

Pearson’s correlations between PDM items and each of the core WHOQOL-100 domains were calculated (Table 3). If it is located under in the most appropriate domain, each item should correlate more highly with its parent domain. For pain relief, three out of four items were more strongly associated with the parent physical domain than with other domains. However, the item asking about whether treatment had improved QoL was associated more strongly with the independence domain. One anger and frustration item was more strongly associated with the parent psychological domain; two with the physical domain and one with independence. For vulnerability/fear/worry, two items were more strongly associated with the parent psychological domain; for the other two items one was more associated with the environment, and another with independence. For uncertainty, two items correlated more with the parent independence domain and 2 with the psychological domain. Overall, the PDM items were most strongly associated with the physical, psychological and independence domains of QoL.
Table 3

Item-domain and facet-domain Pearson’s correlations (r) (2-tailed)

Domain, facet and item

Domain

Physical

Psychological

Level of independence

Social relationships

Environment

SRPB

Physical domain—Pain relief facet

.65*a

.40*

.62*

.42*

.41*

.19

Treatment improved QoL

.26

.19

.32*a

.31*

.21

.08

Satisfied control of pain

.59*a

.30*

.53*

.35*

.33*

.19

Cope with level of pain

.42*a

.30*

.42*

.27

.24

.09

Comfortable

.60*a

.38*

.49*

.26

.42*

.14

Psychological domain—anger & frustration facet

.53*a

.51*

.51*

.32*

.41*

.12

Anger interfere

.41*

.48*a

.30*

.35*

.44*

.16

Frustration interfere

.46*

.46*

.53*a

.35*

.41*

.10

Pain angry

.38*a

.31*

.37*

.10

.19

.02

Pain irritable

.49*a

.38*

.46*

.20

.26

.11

Psychological domain—vulnerability/fear/worry facet

.39*

.53*a

.44*

.32*

.52*

.01

Vulnerability interfere

.35*

.54*

.41*

.37*

.56*a

.05

Fear bother

.34*

.60*a

.34*

.30*

.47*

.08

Worry about treatment

.13

.23a

.19

.12

.24

−.05

Concern experiencing pain

.31*

.13

.32*a

.13

.19

−.06

Level of independence domain—uncertainty facet

.65*

.63*

.72*a

.45*

.55*

.15

Uncertainty interfere

.43*

.50*a

.45*

.30*

.46*

.14

Difficulty planning

.52*

.51*

.57*a

.34*

.45*

.04

Pain limit life

.61*

.37*

.74*a

.33*

.35*

.02

Satisfied make future plans

.50*

.58*a

.55*

.42*

.46*

.25

p < .001

aHighest item/facet-domain correlation

Facet-domain correlations

All facets were more strongly associated with their parent domain with the exception of anger and frustration, which was more strongly associated with the physical rather than the psychological domain (Table 3). The PDM was least associated with the SRPB facet.

Construct validity of the PDM

The relationship between the PDM facets and pain severity, quality, intensity and subjective QoL was examined. As predicted, Pearson correlations (1-tailed) between PDM facet scores and the five MPQ scores were significantly negatively associated (Table 4). Pain relief is most closely associated with present pain intensity, and anger and frustration with the sensory qualities of pain which is consistent with the strong association between anger and frustration and the dimensions of QoL assessed by the WHOQOL physical domain. Uncertainty was most highly associated with the total pain-rating index, and vulnerability/fear/worry with the sensory pain rating. Overall, vulnerability/fear/worry had the lowest correlations with each MPQ score. Every PDM facet was also significantly positively correlated (1-tailed) with the core pain and discomfort facet, particularly pain relief and uncertainty, suggesting that subjective QoL relating to pain and discomfort is associated with these pain-related aspects of QoL assessed by the PDM. Overall, higher pain intensity and severity is associated with poorer QoL assessed by the PDM facets.
Table 4

Pearson correlations between PDM facet scores and MPQ scores, the core pain and discomfort facet (1-tailed), and with the PCS, MCS and overall QoL (2-tailed)

Facet

Sensory pain

Affective pain

Total pain

Present pain

Overall pain intensity

Core pain & discomfort

Physical health component scorea

Mental health component scorea

Overall quality of lifea

Pain relief

−.36*

−.35*

−.40*

−.57*

−.42*

.63*

.46*

.37*

.60*

Anger & frustration

−.42*

−.23*

−.40*

−.32*

−.33*

.48*

.28

.55*

.49*

Vulnerability/fear/worry

−.27*

−.17

−.27*

−.23*

−.21*

.32*

.33*

.36*

.43*

Uncertainty

−.45*

−.37*

−.47*

−.45*

−.45*

.61*

.44*

.57*

.64*

p < .01

a2-tailed

Concurrent validity of the PDM

The relationship of the PDM facets to health status assessed by the SF-12 (MCS and PCS subscales) and general QoL (G) was examined (Table 4). Each PDM facet was significantly correlated with the PCS and MCS of the SF-12. Pain relief was most strongly associated with the PCS and anger and frustration, vulnerability/fear/worry and uncertainty with the MCS. Uncertainty and pain relief were most highly associated with G, suggesting that perception of QoL in these areas is strongly associated with rating of overall QoL. Overall, a higher QoL assessed by the PDM is associated with better health status assessed by the SF-12 and a higher QoL measured by the global rating of QoL.

Discriminant validity of the PDM

All facets show significant differences between participants defining themselves as ill and well, where those perceiving themselves to be ill reported poorer pain-related QoL (Table 5). Given the small number of patients reporting either very poor (n = 9) or very well (n = 2), the five categories representing perceived health were aggregated into three categories: poor, neither poor nor good, and good. For all facets, QoL was found to better for those rating their health as good rather than poor showing that the PDM facets are able to distinguish between those reporting a different health status measured by the single item enquiring about health. One exception to this was the slightly higher QoL reported by those health was neither poor nor good for the vulnerability, fear and worry facet.
Table 5

Differences between PDM facet means for well and ill groups and for patients reporting different levels of health

Facet

Well

Ill

F

p

Poor

Neither poor nor good

Good

F

p

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Mean (SD)

Pain relief

11.06 (2.75)

9.39 (1.81)

15.41

.001

9.40 (1.93)

9.54 (2.01)

12.27 (2.55)

21.15

.001

Anger & frustration

13.12 (3.07)

11.30 (3.01)

10.50

.002

11.00 (3.32)

11.79 (2.51)

13.85 (3.44)

8.69

.001

Vulnerability/fear/worry

13.98 (2.73)

12.82 (2.57)

5.64

.019

12.64 (2.73)

12.59 (2.34)

14.73 (2.71)

8.09

.001

Uncertainty

12.42 (2.97)

10.05 (2.89)

19.20

.001

9.67 (2.61)

10.53 (2.81)

14.00 (2.63)

27.62

.001

In summary, the PDM shows moderate to good psychometric properties including a good overall scale alpha. Two facets show acceptable alphas, and two facets have alphas approaching acceptable levels. When assessed concurrently with other instruments, the PDM demonstrated good construct and concurrent validity. In addition it was able to distinguish between those who were ill and well demonstrating discriminant validity.

Discussion

We announce a tool for the assessment of QoL in chronic low back pain having tested its psychometric properties and found these to be fair to good. The value of the PDM module is that it is used with the WHOQOL-100 to comprehensively assess QoL. Due to its patient-centered bottom-up development process and careful item selection, it contains components that are relevant, acceptable and important to those with chronic low back pain. This report focuses primarily on QoL assessed by the PDM module which showed that QoL was reasonably good for patients in terms of the lack of impact from perceived vulnerability, fear and worry; anger and frustration or uncertainty, and was poorest for QoL relating to pain relief. The latter finding might have been expected from a sample who were seeking pain relief treatment, but also shows that complete coverage of relief from suffering in this patient group still has some way to go.

The PDM performed well in terms of construct, concurrent and discriminant validity, and demonstrated adequate internal consistency reliability. The relatively low levels of missing data suggest that there were few problems completing items, reflecting the acceptability of the instrument. This confirms the value of deriving items from focus groups of potential users and in the case of items which form the facets of the PDM, extra previous testing stages using cognitive interviewing to support face validity, comprehensibility, feasibility and acceptability (Mason et al. 2004, 2008). Although completion time was not collected, an additional advantage is the approximate completion time of around 10 min, in addition to completion time of the core WHOQOL instrument (around 20 min for the WHOQOL-100, 5 min for the WHOQOL-Bref), although this might be longer for patients with conditions characterised by fatigue.

Internal consistency reliability for the 16-item module was good and facet alphas for uncertainty and anger and frustration were acceptable, but less good for pain relief and vulnerability/fear/worry due to low inter-item correlations. Although these facets showed lower internal consistency, they were approaching acceptable levels (.70). When tested with chronic low back pain patients, the pain relief and vulnerability/fear/worry facets were less cohesive, and the facet items more heterogeneous than those representing the uncertainty, and anger and frustration facets which suggests that some items are restricting the internal consistency of facets. Such an observation could be due to the relatively small sample or reflect the broad nature of the constructs these facets measure. It may also reflect the fact that some items specifically mention pain, whereas others are broader. This was based on the language of focus group participants (Mason et al. 2004) rather than any deliberate attempt to vary how specific or general the questions were. The methodology adopted for the development of the PDM is consistent with the development of the WHOQOL where facets contain four items (except in abbreviated forms). Consequently, unless developing a short-form, it is not possible to carry out further item deletion. However, future work needs to pay particular attention to the items that contribute most to low reliability. This will be addressed through testing the instrument in larger and more heterogeneous populations where it may be possible to modify or substitute items to enhance the overall reliability of the instrument.

Construct validity was moderately good. Items were usually more strongly associated with their parent domain than with other QoL domains. An exception was that half of the uncertainty items were associated with the psychological rather than independence domain. In addition to raising issues about the conceptual integrity of the facets, the strength of these associations also reflect the complex relationship and interconnectedness of the concomitants of pain and that the items representing these facets might affect QoL in different ways, despite belonging to the same concept. Facet level analysis confirmed that three of the four PDM facets were strongly associated with their parent domain, although anger and frustration was marginally more associated with the physical domain. This highlights the salience and possible co-occurrence of these emotions for people with pain and is consistent with the strong negative association between the sensory dimension of pain assessed by the MPQ with QoL relating to anger and frustration. Although it appears that perceptions of QoL relating to pain relief and sensory aspects of pain are associated with high levels of anger and frustration, this cross-sectional study does not allow direction of causation to be established. However, previous studies have shown that anger is a consequence of living with chronic pain (Wade et al. 1990). Irrespective of whether they co-occur or are causally related, there is potential clinical value in targeting pain relief to minimise anger, and in targeting anger and frustration to bring about improvements in reported QoL. Given that the association between anger and frustration is only marginally higher than with the psychological domain and that conceptually they are psychological constructs, further testing would be required before moving this facet to the physical domain would be justified.

Construct validity was also investigated through examining the relationship between the PDM and reports of pain (assessed by the SF-MPQ). As predicted, higher pain intensity and severity was associated with poorer QoL in these aspects. Better QoL assessed by the PDM was also associated with better health status, providing support for concurrent validity. Furthermore, the pain relief facet and the physical functioning score of the SF-12 were strongly associated, as were the anger and frustration, vulnerability/fear/worry and uncertainty facets with the mental health score. This demonstration of concurrent validity supports the place of the pain relief facet within the physical domain and of anger and frustration, vulnerability, fear and worry, and uncertainty within the psychological and independence domains. The PDM facets also distinguished well between participants defining themselves as ill or well, and between those reporting different levels of health, where QoL was significantly lower for ill participants and for those reporting poorer health. This is important for two reasons; firstly, it provides evidence of discriminant validity which is an important psychometric property because it demonstrates that scores can sensitively discriminate between different clinical conditions. Secondly, it highlights the potential role of beliefs around ‘wellness’, ‘illness’ and ‘health’ and their relationship to the way people rate their pain-related QoL.

In terms of data quality, although some floor and ceiling effects were observed, given the relatively poor health status of patients, it is not surprising that the anchor point for good QoL was not used in some cases. For the analysis itself, although two-tailed tests are more stringent, the one-tailed tests were performed because a negative association between pain and QoL was expected, where high pain was associated with poor QoL. Many of these associations were also found to be significant at the more stringent level of p < .01. Although some of the correlations between the PDM facets and other measures of functioning that were used (MPQ, PCS, MCS, QOL) are weak (.20–.40 range), four are above .60 and all are significant. Given that none of the instruments assessed QoL, only modest correlations in the absence of a ‘gold standard’ pain QoL instrument were predicted. As we would not expect very strong associations, the strength of these correlations provides modest support for the validity of the PDM. The PDM also assesses subjective self-reports of the impact of pain, not of the pain itself (assessed by the MPQ) which would also lessen the strength of the association.

The overall aim of developing the PDM was to increase the specificity and sensitivity of the generic core WHOQOL-100 instrument, because the pain and discomfort facet in the original instrument was found to under-represent the impact of chronic pain on QoL (Skevington 1998). The PDM facets relate most strongly to the physical, psychological and level of independence domains, and less to social relationships, the environment or spirituality. That QoL relating to pain is not associated with spirituality is consistent with previous research that spiritual QoL does not appear to be affected by the presence of pain (Skevington 1998). However, although not significant overall, it is possible that spirituality is important to some patients with chronic pain as previous studies show that it may be related to coping in terms of using prayer or seeking spiritual support as part of a repertoire of strategies to manage pain (Wachholtz et al. 2007).

This was a pragmatic study designed more to investigate the properties of the PDM than report on the QoL of patients with chronic low back pain. Now that the instrument is available and confirmed to have some properties of reliability and validity it can be used more confidently for this and other pressing clinical research. Given the fairly homogenous group that this patient sample was drawn from in terms of pain located primarily in the lower back and spine, further work should examine the PDM in other groups of chronic pain patients and address test-retest reliability and sensitivity to change, which are crucial properties of instruments. Although it is not known how many of the patients in this study had other co-morbid conditions such as common mental disorders (e.g. anxiety and depression), those patients who where known to have a psychological disorder were excluded from this study. Consequently, the instrument should also be tested to ensure that the scale is appropriate for use with such individuals. In addition, the response rate naturally impacts on the extent to which the findings generalise to the wider clinical population and it was not possible to collect information about non-responders which means systematic differences cannot be ruled out. Work to address these issues is on-going and will enable an examination of the robustness of the PDM facets for generic use across a broad range of diagnostic categories. In addition, given that the modest sample size might explain some of the observed associations between facets and domains, further work should add to this preliminary report of the psychometric properties of the module by applying the module to larger, heterogeneous pain populations. As the sample size (n = 133) was not considered sufficient for reliable exploratory factor analysis (Comrey and Lee 1992) this would also enable factor analytic techniques to be applied to the data.

The relatively small amount of missing data suggests that the instrument is comprehensible and acceptable to patients. The most frequently missed item addressed the extent to which having treatment has improved QoL which may have been due to the fact that not all patients were receiving treatment at the time and could not make a judgement about this. Of the 16 items, the full range of scores was only used in 12, and none of these had 10% of respondents using each anchor point of the scale. However, given that the sample was drawn from a relatively homogenous group, the distribution of scores may reflect this. Moreover, given the chronicity of the conditions, it is not surprising that the anchor point for best QoL was not always utilised.

Given concerns about respondent burden, the development of an abbreviated form of the PDM module for use with the WHOQOL-Bref (The WHOQOL Group 1998a, b), would be an important development as this shorter instrument has already been validated for patients with rheumatoid arthritis (Taylor et al. 2004). Further data from heterogeneous pain populations would enable us to reduce the number of items in addition to refining the instrument by substituting problematic items to optimise the psychometric properties. Other key issues relate to a closer inspection of the relative importance of aspects of QoL to patients with chronic pain, and investigating the relationship between pain-related QoL and other patient-reported outcomes. Lastly, this work provides a blueprint for collaborative international work on the WHOQOL-Pain using the innovative WHOQOL methodology for cross-cultural research (Skevington et al. 2004), so that the universality of these constructs may be eventually established.

Acknowledgments

The Economic and Social Research Council (UK) funded this research. The authors thank Sister Liz Phelps, Dr Monica Baird, Dr Steven Hill, Dr Andrew Souter, Dr Ellis, Dr Arora, Staff Nurse Sarah Morgan, Beth Mathias and all those in pain who participated in this study. The foundation work of the WHOQOL Group is acknowledged.

Copyright information

© Springer Science+Business Media, LLC 2008