- Email: [email protected]
Do psychosocial factors predict disability and health at a 3-year follow-up for patients with non-acute musculoskeletal pain?
Available online at www.sciencedirect.com
European Journal of Pain 12 (2008) 641–649 www.EuropeanJournalPain.com
Do psychosocial factors predict disability and health at a 3-year follow-up for patients with non-acute musculoskeletal pain? ¨ rebro Musculoskeletal Pain Screening A validation of the O Questionnaire ¨ hrvik a, P. Wahle´n a, J. Leppert a A. Westman a,b,*, S.J. Linton c, J. O a
Centre for Clinical Research, Uppsala University-Va¨stera˚s Hospital, Sweden b Psychosomatic Medicine Clinic, Va¨stera˚s, Sweden c ¨ rebro University, O ¨ rebro, Sweden Center for Health and Medical Psychology, Department of Behavioural, Social and Legal Sciences-Psychology, O Received 20 October 2006; received in revised form 5 October 2007; accepted 21 October 2007 Available online 20 December 2007
Abstract Purpose: Early identification and intervention with those that run the risk of developing long-term disability would offer a great ¨ rebro Musculoskeletal opportunity for reducing costs and personal suffering associated with long-term work absenteeism. The O ¨ Pain Screening Questionnaire (OMPSQ) has been used and validated in several studies for participants with mainly acute pain prob¨ MPSQ for patients with non-acute pain problems (e.g. 1–6 months sick leave) and lems. The aim of this study was to validate the O compare to other relevant questionnaires. Method: One hundred and fifty-eight patients with musculoskeletal pain and disability recruited to a multidisciplinary rehabilitation project completed a battery of questionnaires at baseline and at 3-year follow-up visits. The main analysis involved the relationship between risk levels in the questionnaire and sick leave and perceived health after 3 years. ¨ MSPQ predicted future sick leave and health and was found to have six factors. The function and pain factors Results: The O were the best predictors of sick leave after 3 years, while the distress factor was the best predictor of perceived mental health and return to work-expectancy was borderline significant. Perceived physical health at 3 years was best predicted by the function and pain factors with the fear-avoidance factor being marginally significant. ¨ MPSQ are related to work disability and perConclusion: The results demonstrate that psychosocial factors as measured by O ¨ MSPQ was a good predictor of outcome. ceived health even 3 years after treatment for patients with non-acute pain problems. The O Ó 2007 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights reserved. Keywords: Return to work; Sick leave; Function; Musculoskeletal pain; Risk factors; Perceived health; Screening questionnaire
1. Introduction
* Corresponding author. Address: Psychosomatic Medicine Clinic, Karlsgatan 17 A, S-722 14 Va¨stera˚s, Sweden. Tel.: +46 (0)21 176322; fax: +46 (0)21 176330. E-mail addresses: [email protected], [email protected] (A. Westman).
Today pain is viewed as a multi-dimensional phenomenon, comprising biological, psychological, social and existential elements. Psychological factors have been shown to be good predictors of long-term disability including sick leave. Research during the last decade has shown a link between psychosocial factors and the
1090-3801/$34 Ó 2007 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ejpain.2007.10.007
642
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
development of chronic pain and disability, in that the psychosocial risk factors play an important role for the transition from acute pain to chronic pain problems (Turk, 1997; Linton, 2000; Pincus et al., 2002). In a study of predictor variables for future problems containing a sample of acute and sub-acute back pain patients Burton et al. showed that the best variables were almost exclusively psychosocial (Burton et al., 1995). Many studies have shown the specific importance of fear avoidance beliefs in the development of chronic pain and disability (Vlaeyen and Linton, 2000; Leeuw et al., 2007). However, in a systematic literature review of prospective investigations of patients with acute low back pain. Pincus et al. concluded that there is limited evidence to link such fear states with poor prognosis (Pincus et al., 2006). On the individual level persistent musculoskeletal pain significantly affects the patient’s quality of life. For example, the pain is usually continuous and disrupts a variety of functions (Nachemson et al., 2000; Linton, 2002). However, not all patients develop chronic problems and only a small percentage (3–10%) develops long-term work absence after an acute bout of back pain. This small number nevertheless consume about 75–85% of the recourses (Reid et al., 1997; Nachemson et al., 2000). The prevention of the development of persistent musculoskeletal pain would be greatly enhanced if patients most in need of treatment and rehabilitation could be identified at an early point in time. However, since musculoskeletal pain occurs so frequently it would be costly, and unnecessary to provide every patient suffering from musculoskeletal pain with secondary preventive interventions. Early identification and intervention with those that run the risk of developing long-term disability would offer a great opportunity for reducing costs and personal suffering associated with long-term work absenteeism. Early identification often involves screening procedures. Multiple questionnaires are available for the assessment of chronic low back pain and disability (Main and Waddell, 1991; McCracken et al., 1992; Waddell et al., 1993; Gatchel and Gardea, 1999) but there is a paucity of those that are specific to evaluating psychosocial risk factors for chronicity. ¨ rebro Musculoskeletal Pain Screening QuesThe O ¨ MPSQ) has been used and validated in sevtionnaire (O eral studies for participants with mainly acute pain problems. Results have shown that the total score is related to future sick leave and functional ability; the higher the score, the higher the risk for long-term sick leave and the development of chronic functional problem. About 80% of the acute back pain patients in these studies that go on to develop a long term sick leave can be correctly classified using this screening instrument (Hurley et al., 2000, 2001; Boersma and Linton, 2002; Linton and Boersma, 2003). The instrument has been
developed for possible use in a variety of primary health care settings as a complement to the clinical examination, where patients with acute or recurrent pain may seek care. To date, however, reports are only available for patients with acute or short-term problems. Yet, about 25% or more of primary health care patients have recurrent or persistent problems (von Korff, 1999). Con¨ MPSQ with these sequently, there is a need to evaluate O patients and determine if the results are helpful as compared to other standardised instruments.
2. Aim ¨ MPSQ for The aim of this study was to validate O patients with non-acute pain problems (e.g. 1–6 months sick leave) and compare to other questionnaires such as Job Strain, the Coping Strategies Questionnaire (CSQ), the Pain Catastrophizing Scale (PCS) and the Tampa Scale for Kinesiophobia (TSK). The main purpose of this paper was to study the relationship between risk lev¨ MPSQ and sick leave and perceived els found from the O health 3 years later. ¨ MPSQ would Our specific hypothesis was that the O predict future sick leave and health, but that because the current sample had a longer history of a problem at recruitment, that this relationship would be moderate. A second hypothesis was that the other questionnaires included in the study would be predictive of outcome, ¨ MPSQ. but to a lesser degree than the O
3. Method 3.1. Design During the period 1998–2000, a rehabilitation project was carried out in the Primary Health Care settings in the county of Va¨stmanland in Sweden. One hundred and fifty-eight patients with musculoskeletal pain and disability were recruited to the project. Eighty-seven patients from two primary health care units participated in a multimodal rehabilitation programme and 71 patients with the same inclusion criteria from four other primary health care units received treatment ‘‘as usual”. Because patients normally receive a variety of treatments in primary health care we chose to do the data analysis for the entire group (n = 158). An evaluation of the rehabilitation programme will be carried out in a forthcoming study. 3.2. Patients The inclusion criteria were those between 18 and 65 years old, sick listed P28 days 6180 days and/or had consulted the doctor about the same problem P3 times the last 12 months according to information from the
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
referring physicians. The participants either had permanent employment or were at the disposal of the labour market. Patients requiring orthopaedic surgery and patients with a psychiatric disorder or a substance abuse problem were excluded. Furthermore, participants had to be able to speak Swedish sufficiently well in order to describe their symptoms and understand the information given. The observation period of this study was 3 years. At the 3-year follow-up there were 11 drop-outs. The dropout analysis shows a gender difference where 71% of the participants in the study group were women, compared with 54% of the drop-outs who were women (Table 1).
643
3.3. Measure At baseline and at the 3-year follow-up visit, the participants completed a battery of questionnaires concerning sick leave, work, function, and quality of life, as well as a screening questionnaire for the early identification of people at risk for developing long-term problems. Supplemental questionnaires (e.g., CSQ, TSK, PCS, and life events) were used in order to see if those had ¨ MPSQ. better predictive value compared with O
4. Assessments ¨ rebro Musculoskeletal Pain Screening 4.1. The O ¨ MPSQ) Questionnaire (O
Table 1 Baseline characteristics of the patients (n = 158) Frequency
(%)
Age (mean) Sex (men/women)
47 (range 24–65) 48/110
30/70
Educational level Elementary school only High school University
71 63 14
48 43 10
Occupational status Employed Unemployed Students Other
115 29 4 9
73 19 3 6
Primary pain site Neck Shoulder/shoulder-joint Back – upper part Back – lower part Leg Other symptoms
97 105 48 85 55 57
62 70 31 85 35 37
Pain duration in weeks (n = 156) 0–11 weeks 12–23 weeks 24–52 weeks P53 weeks
24 23 24 85
15.3 14.7 15.4 54.5
Sick leave at baseline 25% sick leave 50% sick leave 75% sick leave 100% sick leave Missing
125 5 15 4 98 3
79 4.1 12.3 3.3 80.3 4
Sick leave, days previous 12 months 0–30 days 31–60 days 61–90 days 91–180 days
49 46 21 40
31 30 13 26
Disability benefits at baseline
16
10.5
Mean ¨ MPSQ) Screening index at baseline (O Experimental group 121 Control group 124
SD 21.4 22.7
¨ MPSQ was developed as a tool for clinicians in the O early identification of people at risk for developing longterm problems. The questionnaire contains 25 items divided into five groups (function, pain, psychological factors, fear avoidance, and miscellaneous) [i.e., sick leave, age, gender, nationality, monotonous or heavy work, job satisfaction]. Twenty-one of the 25 items are scored on a 0–10 scale, yielding a total score range of 0–210. Scores are tallied to form a total score. The actual experience of pain is explored in several questions about site, intensity, duration, and frequency of pain. Ability to perform daily activities is assessed with five items. Participants rate their current ability to carry out light work, walk for an hour, do household chores, the weekly shopping and sleep. Perceived ability to cope with pain and perceived job satisfaction are appraised with two single items. Fear-avoidance beliefs are estimated with three assertions where the patient rates the degree to which she/he agrees. The patient’s own perception of their risk is measured with two items developed especially for this questionnaire. 4.2. Utility In the first study conducted with the instrument, the screening instrument was found to have satisfactory test–retest reliability (0.83) and validity in a study of 142 patients where outcome was sick absenteeism (Linton and Hallde´n, 1997; Linton and Hallde´n, 1998). When using a cut-off score of 105 (the maximum being 210), the specificity was found to be 0.75 and the sensitivity 0.88. At the same cut-off point the chance of accurately classifying someone as ‘‘at risk” for future sick leave would be 88%. Kendall (1999) tested the instrument on a population of more than hundred acute-pain patients in New Zealand and found that a cut-off score of 105 was related to future disability. Hurley et al. (2000) investigated the instruments predictive ability with regard to
644
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
return-to-work after physical therapy. They reported that a cut-off point of 112 correctly identified 80% of patients failing to return to work at the end of treatment (sensitivity), while at the same time 59% of those returning to work were identified (specificity). Boersma et al. showed in a review study 2002 that the questionnaire appears to be most helpful in predicting future functional problems as well as sick absenteeism. The predictive validity of the instrument for pain was low, relative to the predictive validity for function and absenteeism. In an Australian study, Dunstan et al. ¨ MPSQ in a compensable injury (2005) evaluated the O population and found that the scores were related to long-term disability. The findings had significant relevance to the recommendation of routine screening for the early identification of injured workers at risk of long-term disability. 4.3. The Short Form-36 Health Survey The generic measure, SF-36, was developed in the United States. It covers both functioning and wellbeing and has proved suitable for clinical research, patient monitoring, and health care planning. The SF-36 is a 36-item questionnaire that assesses healthrelated quality of life from the point of view of the health care recipient. Ware et al. developed the SF-36 as a comprehensive, psychometrically sound, and brief, health outcome measure. The research also designed it to be generic so that it could be applied to a wide variety of health conditions. The SF-36 describes both the physical and mental components of health. Eight primary SF-36 scales from distinct physical and mental health clusters are based on factor analyses. For each of the eight scales, scores range from 0 to 100, with higher scores reflecting better self-reported health status. Continuous method studies and empirical testing have shown that the scales in SF-36 can be summed up to two overall health indexes – Physical Health Score (PCS) respective Mental Health Score (MCS) (Ware et al., 1994). 4.4. Job Strain This instrument contains 11 items (graded 1–4) concerning demands and control. Indices for work demand and control are calculated. High scores correspond to high demands and high control. By dividing demands by control, a measure of job strain is obtained for each patient (Theorell et al., 1988). 4.5. Coping Strategies Questionnaire (CSQ) The questionnaire assesses eight different coping strategies for pain. Each strategy consists of six different items. On a scale ranging from 0–6, subjects are asked to
indicate how often they use a particular item when they experience pain (0 = never use it, 6 = always). A total score of 36 can be obtained on each of the eight coping strategies. In this study we restricted the questionnaire to assess the coping strategy: Coping self-statements (Rosenstiel and Keefe, 1983; Jensen and Linton, 1993). 4.6. Pain Catastrophizing Scale (PCS) The PCS instruction asks participants to reflect on past painful experiences and to indicate the degree to which they experienced each of 13 thoughts or feelings when experiencing pain on a five-point scale from 0 (not at all) to 4 (all the time) (Sullivan et al., 1995; Osman et al., 2000). 4.7. Tampa Scale for Kinesiophobia (TSK) The Scale for Kinesiophobia (TSK) developed by Miller et al. as a measure of fear of movement/(re)injury. Each of the 17 items is provided with a four-point Likert Scale with scoring alternatives ranking from ‘‘strongly disagree” to ‘‘strongly agree” (Vlaeyen et al., 1995a). In this study, we used a short version with 12 items. In accordance with the psychometric work done by Vlaeyen et al. (1995b) five items that loaded poorly in factor analysis were left out, and a total score was used. 4.8. Life events A continuous variable ‘‘life events” was computed from answers regarding 17 possible major events which could have occurred during the preceding 12 months (12 negative, 1 positive, 4 neutral) (Roll and Theorell, 1987; Theorell and Emlund, 1993). 4.9. Sick leave Information on sick leave was obtained at baseline and at the follow-ups by the patients reporting their own current sick leave. Previous research has shown that these ratings highly correlate to register data from the National Social Insurance office (Linton et al., 1995; Hensing et al., 1998). ‘‘Improved” is defined as a patient who has decreased her/his sick leave level. ‘‘Impaired” is defined as a patient who maintains or increases her/his sick leave level at the follow-up.
5. Statistics The SPSS 11.0 statistical programme was used for the data analysis. Sensitivity and specificity are important properties when evaluating a questionnaire. The relationship between sensitivity and specificity can be studied by constructing the Receiver Operator Characteristic
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
(ROC) curve. Principal Component Analysis was used ¨ MPSQ in order to reduce the dimenfor the items in the O sionality and find relevant factors. Only factors with an eigen value P1 were used (Everitt and Dunn, 1991). To assess the predictive power of different factors on sick leave, stepwise logistic regression was used. Adjustments were made for the variables age, sex, education level and previous sick leave if their p-value were less than 0.2. The predictive power of the different factors on perceived mental and physical health was studied by General Linear Model (GLM). Continuous variables were presented as mean and standard deviation, and categorical data as counts and percentages. A two sided p-value <0.05 was considered significant. Since there were no convincing outcome difference between the control group and the experimental group the data analyses were conducted on the whole group.
6. Ethics The Human Research Ethics Committee at the University of Uppsala, Sweden approved the study.
645
months, on average, how bad was your pain?” (0.79) ‘‘How often would you say that you have experienced pain episodes, on average, during the 3 past months?” (0.81). (III) Distress: ‘‘How tense or anxious have you felt in the past week?” (0.86) ‘‘How much have you been bothered by feeling depressed in the past week?” (0.89). (IV) Fear-avoidance: ‘‘Physical activity makes my pain worse” (0.50). ‘‘An increase in pain is an indication that I should stop what I am doing until the pain decreases” (0.82). ‘‘I should not do my normal work with my present pain” (0.76). (V) Return to work expectancy: ‘‘In your view, how large is the risk that your current pain may become persistent? (0.61), ‘‘In your estimation, what are the chances that you will be working in 6 months?” (0.79) ‘‘If you take into consideration your work routine, management, salary, promotion possibilities and work mates, how satisfied are you with your job?” (0.76). (VI) Coping:‘‘Based on all the things you do to cope, or deal with your pain, on an average day, how much are you able to decrease it?” (0.92). 7.2. Predictive ability of the factors
7. Results 7.1. Total scores of all questionnaires (Table 2) ¨ rebro Musculoskeletal 7.1.1. The factor analysis of the O ¨ MPSQ) Pain Screening Questionnaire (O In order to reduce the items in the questionnaire the structure was examined by Factor Analysis. The items were reduced to six disjoint factors (factor loading within brackets). (I) Function: ‘‘I can do light work for an hour” (0.67), ‘‘I can walk for an hour” (0.66), ‘‘I can do ordinary household chores” (0.68), I can go shopping” (0.68), ‘‘I can sleep at night” (0.47). (II) Pain: ‘‘How would you rate the pain that you have had during the past week?” (0.70), ‘‘In the past 3
To study which of the factors were important for long-term outcome we performed separate analyses for each factor. An overview is presented in Fig. 1 and shows the predictive power of each factor. 7.3. Factors predicting of sick leave Because the outcome was dichotomous (‘‘improved”, ‘‘impaired” sick leave) stepwise multiple logistic regression was used to analyse the predictive ability of the factors. Adjusting for age and earlier sick leave (p less than 0.2) factor I (function) and factor II (pain) significantly predicted sick leave after 3 years as seen in Table 3. This model had a predictive ability of correctly classifying 71% of the participants with a sensitivity of 63% and a specificity of 77%. No other factors were significant.
Table 2 Scores of all questionnaires, mean and standard deviation, at baseline and at 3-year follow-up Questionnaires
n
Baseline
3-Year follow-up
Mean
SD
Mean
SD
¨ MPQ total score O Job Strain Coping Strategies Questionnaire (CSQ) Pain Catastrophizing Scale (PCS) Tampa Scale for Kinesiophobia (TSK) Life events SF-36, Mental Health Score (MCS) SF-36, Physical Health Score (PCS) Psychosomatic symptoms
149 61 116 113 125 110 111 111 124
123 0.9 2.9 18.4 26.3 2.7 38.8 40.8 1.3
22 0.3 1.2 10 6.6 2.2 8.4 5.8 0.7
94 0.8 3.3 16.9 12.7 2.5 45.1 37.4 1.5
43 0.2 1.1 10.8 7.1 2.1 12.1 10.2 0.6
646
A. Westman et al. / European Journal of Pain 12 (2008) 641–649 Table 5 Prediction of self-reported health (SF-36, Physical Health Score) analyzed by General Linear Model Regression coefficient Factor I (function) Factor II (pain) Factor IV (fear-avoidance) Age
95% CI
p-Value
2.056 1.405 0.963
3.024 to 1.087 2.638 to 0.173 0.039 to 1.964
0.269
0.445 to
<0.001 0.026 0.059
0.092
0.003
Fig. 1. Increased sick leave or maintained total work disability. Unadjusted odds ratio and 95% CI for the different factors. Table 3 Prediction of sick leave at 3-year follow-up analysed by stepwise multiple logistic regression analysis
Factor I (function) Factor II (pain) Age Earlier sick leave
Adjusted OR
95% CI
p-Value
1.4 1.3 1.04 1.15
1.152–1.762 1.017–1.696 1.002–1.079 0.946–1.403
0.001 0.037 0.04 0.16
7.4. Predictors of perceived health – Mental Health Score (MCS) Because this measure consists of continuous outcome variables, General Linear Model (GLM) was used for the prediction of perceived health. After adjusting for age, Factor III (Distress) predicted perceived health significantly and Factor V (return to work expectancy) predicted perceived health almost significantly (p < 0.001 and 0.082, respectively), as measured by SF-36 (Mental Health Score). No other factors were significant (Table 4). 7.5. Predictors of perceived health – Physical Health Score (PCS) Adjusting for age, factor I (function) and factor II (pain) predicted perceived health as measured by SFTable 4 Prediction of self-reported health (SF-36, Mental Health Score) analyzed by General Linear Model Regression coefficient Factor III (distress) Factor V (return to work expectancy) Age
1.509 0.922 0.241
95% CI 2.296 to 0.721 1.963 to 0.120 0.035 to 0.447
p-Value <0.001 0.082 0.022
¨ MPSQ Fig. 2. Receiver Operator Characteristic (ROC) curve for O total score’s ability to predict the sick leave status at the 3-year followup.
36 (Physical Health Score) significantly (p < 0.001 and 0.026, respectively). Factor IV was borderline significant (p = 0.059). No other factors were significant (Table 5). ¨ MPSQ score 7.6. Predictive ability of the overall O 7.6.1. Receiver Operator Characteristic (ROC) curve Cut-off scores are another way to assess the predictive ¨ MSPQ. A Receiver Operator Charactervalidity of the O istic curve (ROC) shows the specificity and sensitivity of ¨ MPSQ as the measure for selected total scores on the O seen in Fig. 2. A cutoff ‘‘at-risk” score of 117 correctly classified (sensitivity) 78% of the poor outcomes (failed to reduce sick leave) and a cut-off score of 139 correctly classified 44% of those who failed to reduce their sick leave. For the same score levels 49% and 89% of those who succeeded in reducing their sick leave were correctly classified (specificity). 7.6.2. Other questionnaires The questionnaires Job Strain, Coping Strategies Questionnaire (CSQ), Pain Catastrophizing Scale (PCS), Tampa Scale for Kinesiophobia (TSK) and Life events, did not predict sick leave, or perceived health, significantly at the 3-year follow-up.
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
8. Discussion This study demonstrates that psychosocial factors as ¨ MPSQ are related to work disability and measured by O perceived health even 3 years after treatment in primary care. Moreover, the screening questionnaire had discriminative power even for patients with non-acute or ¨ MPSQ had better recurrent pain problems. Indeed, the O predictive power than any of the questionnaires included in the study. This study shows that among the factors, pain and function are the factors most strongly related to sick leave 3 years later. ¨ MPSQ The prediction of future sick leave with the O is based on the total score. As in earlier studies the results showed that the total score of the screening questionnaire was related to future sick leave and functional ability; the higher the score, the higher the risk for long term sick leave and developing of chronic problems (Linton and Hallde´n, 1997; Linton and Hallde´n, 1998; Hurley et al., 2000; Boersma and Linton, 2002). The factor analysis indicated six clear factors that reflect the original theoretical conceptualization of the questionnaire (Linton and Hallde´n, 1998). However, the results of the long-term follow-up do emphasize the importance of psychosocial factors in prediction. Function with focus on daily living, sleep capacity and pain experience had the most powerful predictive value concerning sick leave at 3 years. While earlier studies have shown that emotional and cognitive variables such as distress and fear avoidance beliefs have been strong predictors for 6–12-month outcomes, the best predictor in this study is having problems functioning. This probably reflects the length of the follow-up and suggests that different variables may be predictive at various stages in the process of chronification. Even though it is recognized that psychological variables are influential factors, little is known about how and when these variables interact in the process toward disability. Furthermore, psychological variables might operate differently for different people and at different time points. Boesrma et al., for example, showed how four distinct cluster profiles varied in their development of problems over time (Boersma and Linton, 2006). The most powerful predictors of mental health at the 3-year follow-up were the variables distress and work expectancy. The fact that distress (consisting of anxiety and mood) predicts mental health is not surprising, but does highlight the role of emotional variables in the development of chronic problem. More surprising, is the fact that work expectancy predicts mental health at the follow-up. Although work expectancies predicted future mental health, it is interesting that they were not very good predictors of actual sick days which is also contrary to previous findings (Linton and Hallde´n, 1998). To achieve reasonable predictability, we needed to have a cut-off score of 117. This is slightly higher than
647
some other studies where similar prediction levels have been obtained with scores of 105–112. The slightly higher cut-off point is not surprising since the participants had longer histories of pain problems. One explanation might be that this sample is different than those in other studies, e.g. having more disability at inclusion. However, this mirror the fact that work is an important factor in mental health. Another explanation may be that the work expectancy factor taps into negative expectancies in general which in turn are related to future mental health. Physical health was significantly predicted by the function and pain factors. Although the pain factor was marginally significant, the result is consistent with a large body of literature showing that current pain conditions predict future pain conditions (Pulliam and Gatchel, 2002). Likewise, function has previously been found to predict future health problems such as being off work and functional problems. Thus, this finding is quite consistent with previous work in the field. Earlier prospective studies have shown that painrelated fear is a predictor of future disability in patients suffering from an acute episode of back pain (Klenerman et al., 1995; Fritz and George, 2002). Moreover cross-sectional studies have shown that pain catastrophizing is associated with psychological distress and pain related disability as well as social functioning and general health inpatients with back pain (Severeijns et al., 2005). However, in our study these variables did not show significant power. One reason could probably be that the items about function and pain are simply such strong predictors that they eliminate fear and catastrophizing. Since the outcome variable is sick leave, function and pain may be directly related and, thus, reflect the reason for being off work rather than the cause. Another possibility mentioned above is that fear and catastrophizing may be important earlier in the developmental process while function becomes important at a later stage. The patient group in the study had longer and more severe condition as compared to several other studies ¨ MPSQ has been assessed. This may also where the O be an explanation as to why factors like fear did not show as much predictive value as in earlier studies. Another interesting finding concerns the role of sick leave as a predictor. This variable has consistently been shown to be a strong predictor of long-term problems and future sick leave. Consequently, we adjusted our analyses for earlier sick leave and this might be of importance in interpreting the result in our study. Function and pain experience was the most predictive variable in our study. The high disability level at baseline probably played a role whereby questions about daily living and pain experiences probably had a more comprehensive importance and showed the strongest predictive value. However, the genesis of musculoskeletal
648
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
pain is complex and often of multifactorial origin. The research during the last decades has focused on several important variables predicting long-term problems. For the general practitioner who often see many patients with a variety of symptoms it is a demanding task to observe patients with risk for long-term problems. This study has certain methodological facts that should be kept in mind when interpreting the results. Firstly, the subjects consisted of an experimental and a control group with different kinds of treatments. However, because an analysis of the effects of treatment did not produce any significant differences, we chose to pool the data to gain power and clarity in interpreting the results. Secondly, the long follow-up period is, perhaps, worthy of consideration. While this offers advantages it also opens the door for other events that may influence outcome. Thus, a long follow-up probably reduces the size of the impact of the predictor variables. Earlier studies have suggested that the instrument could be of value in isolating patients in need of early interventions and in promoting the use of appropriate interventions for patients with psychological risk factors. The results of our study are in line with other studies though with lower predictive value. One conclusion must be to emphasize the value of using the instrument ¨ MPSQ earlier in the course of the pain problem. The O questionnaire could be of importance for the general practitioner and other caregivers in Primary care as a complement for patients with musculoskeletal pain and disability. Moreover the instrument appears to be most helpful in predicting future functional problems as well as sick absenteeism.
Acknowledgements The County Council of Va¨stmanland, Sweden and the Social Insurance Office, in co-operation, supported this study.
References Boersma K, Linton SJ. Early assessment of Psychological factors: the ¨ rebro Screening Questionnaire for Pain. In: Linton SJ, editor. O New avenues for the prevention of chronic musculoskeletal pain and disability. Amsterdam: Elsevier Science B.V.; 2002. p. 205–13. Boersma K, Linton SJ. Psychological processes underlying the development of a chronic pain problem: a prospective study of the relationship between profiles of psychological variables in the fear-avoidance model and disability. Clin J Pain 2006;22:160–6. Burton AK, Tillotson KM, Main CJ, Hollis S. Psychosocial predictors of outcome in acute and subchronic low back trouble. Spine 1995;20:722–8. Dunstan DA, Covic T, Tyson GA, Lennie IG. Does the Orebro Musculoskeletal Pain Questionnaire predict outcomes following a work-related compensable injury? Int J Rehabil Res 2005;28: 369–70.
Everitt BS, Dunn G. Appl Multivar Data Anal 1991. Fritz JM, George SZ. Identifying psychosocial variables in patients with acute work-related low back pain: the importance of fearavoidance beliefs. Phys Ther 2002;82:973–83. Gatchel RJ, Gardea MA. Psychosocial issues: their importance in predicting disability, response to treatment, and search for compensation. Neurol Clin 1999;17:149–66. Hensing G, Alexanderson K, Allebeck P, Bjurulf P. How to measure sickness absence? Literature review and suggestion of five basic measures. Scand J Soc Med 1998;26:133–44. Hurley DA, Dusoir TE, McDonough SM, Moore AP, Linton SJ, Baxter GD. Biopsychosocial screening questionnaire for patients with low back pain: preliminary report of utility in physiotherapy practice in Northern Ireland. Clin J Pain 2000;16:214–28. Hurley DA, Dusoir TE, McDonough SM, Moore AP, Baxter GD. How effective is the acute low back pain screening questionnaire for predicting 1-year follow-up in patients with low back pain? Clin J Pain 2001;17:256–63. Jensen I, Linton SJ. Coping Strategies Questionnaire (CSQ): reliability of the Swedish version of the CSQ. Scand J Behav Ther 1993;22:139–45. Kendall N. Screening and early intervention: the New Zealand experience. In: Paper presented at the international association for the study of pain; 1999. Klenerman L, Slade PD, Stanley IM, Pennie B, Reilly JP, Atchison LE, et al. The prediction of chronicity in patients with an acute attack of low back pain in a general practice setting. Spine 1995;20:478–84. Leeuw M, Goossens ME, Linton SJ, Crombez G, Boersma K, Vlaeyen JW. The fear-avoidance model of musculoskeletal pain: current state of scientific evidence. J Behav Med 2007;30:77–94. Linton SJ. A review of psychological risk factors in back and neck pain. Spine 2000;25:1148–56. Linton SJ. Why does chronic pain develop? A behavioral approach. In: Linton SJ, editor. New Avenues for the prevention of chronic musculoskeletal pain and disability. Pain research and clinical management, vol. 12. Amsterdam: Elsevier Science B.V.; 2002. p. 67–82. Linton SJ, Boersma K. Early identification of patients at risk of developing a persistent back problem: the predictive validity of the Orebro Musculoskeletal Pain Questionnaire. Clin J Pain 2003;19:80–6. Linton SJ, Hallde´n K. Risk factors and the natural course of acute recurrent musculoskeletal pain: developing a screening instrument. In: Jensen TS, Turner JA, Wisenfeld-Hallin, editors. Progress in pain research and management. Seattle: IASP Press; 1997. p. 527–36. Linton SJ, Hallde´n K. Can we screen for problematic back pain? A screening questionnaire for predicting outcome in acute and subacute back pain. Clin J Pain 1998;14:209–15. Linton SJ, Hallde´n K, Hellsing A. The reliability of self-reported sick absenteeism: a pilot study. Scand J Behav Ther 1995;24:145–50. Main CJ, Waddell G. A comparison of cognitive measures in low back pain: statistical structure and clinical validity at initial assessment. Pain 1991;46:287–98. McCracken LM, Zayfert C, Gross RT. The Pain Anxiety Symptoms Scale: development and validation of a scale to measure fear of pain. Pain 1992;50:67–73. Nachemson A, Wedell G, Nordlund AI. Epidemiology of neck and low back pain. In: Nachemson A, Jonsson E, editors. Neck and back pain. The scientific evidence of causes, diagnosis and treatment. Philadelphia: Lippincott Williams & Wilkins; 2000. p. 165–87. Osman A, Barrios FX, Gutierrez PM, Kopper BA, Merrifield T, Grittmann L. The Pain Catastrophizing Scale: further psychometric evaluation with adult samples. J Behav Med 2000;23:351–65. Pincus T, Burton AK, Vogel S, Field AP. A systematic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain. Spine 2002;27:E109–20.
A. Westman et al. / European Journal of Pain 12 (2008) 641–649 Pincus T, Vogel S, Burton AK, Santos R, Field AP. Fear avoidance and prognosis in back pain: a systematic review and synthesis of current evidence. Arthrit Rheum 2006;54:3999–4010. Pulliam CB, Gatchel RJ. Employing risk factors for screening of chronic pain disability. In: Linton SJ, editor. New avenues for the prevention of chronic musculoskeletal pain and disability. Amsterdam: Elsevier Science B.V.; 2002. p. 183–204. Reid S, Haugh LD, Hazard RG, Tripathi M. Occupational low back pain: recovery curves and factors associated with disability. J Occup Rehab 1997;7(1):1–14. Roll M, Theorell T. Acute chest pain without obvious organic cause before age 40 – personality and recent life events. J Psychosom Res 1987;31:215–21. Rosenstiel AK, Keefe FJ. The use of coping strategies in chronic low back pain patients: relationship to patient characteristics and current adjustment. Pain 1983;17:33–44. Severeijns R, Vlaeyen JW, van den Hout MA, Picavet HS. Pain catastrophizing and consequences of musculoskeletal pain: a prospective study in the Dutch community. J Pain 2005;6: 125–32. Sullivan MJL, Bishop SRB, Pivik J. The Pain Catastrophizing Scale: development and validation. Psychol Assess 1995;7:524–32. Theorell T, Emlund N. On physiological effects of positive and negative life changes – a longitudinal study. J Psychosom Res 1993;37:653–9.
649
Theorell T, Perski A, Akerstedt T, Sigala F, Ahlberg-Hulten G, Svensson J, et al. Changes in job strain in relation to changes in physiological state. A longitudinal study. Scand J Work Environ Health 1988;14:189–96. Turk DC. The role of demographic and psychosocial factors in transition from acute to chronic pain. JASP Press; 1997, p. 185– 213. Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. Pain 2000;85: 317–32. Vlaeyen JW, Kole-Snijders AM, Boeren RG, van Eek H. Fear of movement/(re)injury in chronic low back pain and its relation to behavioral performance. Pain 1995a;62:363–72. Vlaeyen JW, Kole-Snijders AM, Rotteveel A, Ruesink R, Heuts PHTG. The role of fear of movement/(re)injury in pain disability. J Occup Rehab 1995b;5:235–52. von Korff M. Pain management in primary care: an individualized stepped-care approach. The Guiford Press; 1999, p. 360–73. Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A FearAvoidance Beliefs Questionnaire (FABQ) and the role of fearavoidance beliefs in chronic low back pain and disability. Pain 1993;52:157–68. Ware JE, Kosinski M, Keller SD. SF-36 physical and mental health summary scales: a user’s manual. MA: New England Medical Center, The Health Institute; 1994.
European Journal of Pain 12 (2008) 641–649 www.EuropeanJournalPain.com
Do psychosocial factors predict disability and health at a 3-year follow-up for patients with non-acute musculoskeletal pain? ¨ rebro Musculoskeletal Pain Screening A validation of the O Questionnaire ¨ hrvik a, P. Wahle´n a, J. Leppert a A. Westman a,b,*, S.J. Linton c, J. O a
Centre for Clinical Research, Uppsala University-Va¨stera˚s Hospital, Sweden b Psychosomatic Medicine Clinic, Va¨stera˚s, Sweden c ¨ rebro University, O ¨ rebro, Sweden Center for Health and Medical Psychology, Department of Behavioural, Social and Legal Sciences-Psychology, O Received 20 October 2006; received in revised form 5 October 2007; accepted 21 October 2007 Available online 20 December 2007
Abstract Purpose: Early identification and intervention with those that run the risk of developing long-term disability would offer a great ¨ rebro Musculoskeletal opportunity for reducing costs and personal suffering associated with long-term work absenteeism. The O ¨ Pain Screening Questionnaire (OMPSQ) has been used and validated in several studies for participants with mainly acute pain prob¨ MPSQ for patients with non-acute pain problems (e.g. 1–6 months sick leave) and lems. The aim of this study was to validate the O compare to other relevant questionnaires. Method: One hundred and fifty-eight patients with musculoskeletal pain and disability recruited to a multidisciplinary rehabilitation project completed a battery of questionnaires at baseline and at 3-year follow-up visits. The main analysis involved the relationship between risk levels in the questionnaire and sick leave and perceived health after 3 years. ¨ MSPQ predicted future sick leave and health and was found to have six factors. The function and pain factors Results: The O were the best predictors of sick leave after 3 years, while the distress factor was the best predictor of perceived mental health and return to work-expectancy was borderline significant. Perceived physical health at 3 years was best predicted by the function and pain factors with the fear-avoidance factor being marginally significant. ¨ MPSQ are related to work disability and perConclusion: The results demonstrate that psychosocial factors as measured by O ¨ MSPQ was a good predictor of outcome. ceived health even 3 years after treatment for patients with non-acute pain problems. The O Ó 2007 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights reserved. Keywords: Return to work; Sick leave; Function; Musculoskeletal pain; Risk factors; Perceived health; Screening questionnaire
1. Introduction
* Corresponding author. Address: Psychosomatic Medicine Clinic, Karlsgatan 17 A, S-722 14 Va¨stera˚s, Sweden. Tel.: +46 (0)21 176322; fax: +46 (0)21 176330. E-mail addresses: [email protected], [email protected] (A. Westman).
Today pain is viewed as a multi-dimensional phenomenon, comprising biological, psychological, social and existential elements. Psychological factors have been shown to be good predictors of long-term disability including sick leave. Research during the last decade has shown a link between psychosocial factors and the
1090-3801/$34 Ó 2007 European Federation of Chapters of the International Association for the Study of Pain. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ejpain.2007.10.007
642
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
development of chronic pain and disability, in that the psychosocial risk factors play an important role for the transition from acute pain to chronic pain problems (Turk, 1997; Linton, 2000; Pincus et al., 2002). In a study of predictor variables for future problems containing a sample of acute and sub-acute back pain patients Burton et al. showed that the best variables were almost exclusively psychosocial (Burton et al., 1995). Many studies have shown the specific importance of fear avoidance beliefs in the development of chronic pain and disability (Vlaeyen and Linton, 2000; Leeuw et al., 2007). However, in a systematic literature review of prospective investigations of patients with acute low back pain. Pincus et al. concluded that there is limited evidence to link such fear states with poor prognosis (Pincus et al., 2006). On the individual level persistent musculoskeletal pain significantly affects the patient’s quality of life. For example, the pain is usually continuous and disrupts a variety of functions (Nachemson et al., 2000; Linton, 2002). However, not all patients develop chronic problems and only a small percentage (3–10%) develops long-term work absence after an acute bout of back pain. This small number nevertheless consume about 75–85% of the recourses (Reid et al., 1997; Nachemson et al., 2000). The prevention of the development of persistent musculoskeletal pain would be greatly enhanced if patients most in need of treatment and rehabilitation could be identified at an early point in time. However, since musculoskeletal pain occurs so frequently it would be costly, and unnecessary to provide every patient suffering from musculoskeletal pain with secondary preventive interventions. Early identification and intervention with those that run the risk of developing long-term disability would offer a great opportunity for reducing costs and personal suffering associated with long-term work absenteeism. Early identification often involves screening procedures. Multiple questionnaires are available for the assessment of chronic low back pain and disability (Main and Waddell, 1991; McCracken et al., 1992; Waddell et al., 1993; Gatchel and Gardea, 1999) but there is a paucity of those that are specific to evaluating psychosocial risk factors for chronicity. ¨ rebro Musculoskeletal Pain Screening QuesThe O ¨ MPSQ) has been used and validated in sevtionnaire (O eral studies for participants with mainly acute pain problems. Results have shown that the total score is related to future sick leave and functional ability; the higher the score, the higher the risk for long-term sick leave and the development of chronic functional problem. About 80% of the acute back pain patients in these studies that go on to develop a long term sick leave can be correctly classified using this screening instrument (Hurley et al., 2000, 2001; Boersma and Linton, 2002; Linton and Boersma, 2003). The instrument has been
developed for possible use in a variety of primary health care settings as a complement to the clinical examination, where patients with acute or recurrent pain may seek care. To date, however, reports are only available for patients with acute or short-term problems. Yet, about 25% or more of primary health care patients have recurrent or persistent problems (von Korff, 1999). Con¨ MPSQ with these sequently, there is a need to evaluate O patients and determine if the results are helpful as compared to other standardised instruments.
2. Aim ¨ MPSQ for The aim of this study was to validate O patients with non-acute pain problems (e.g. 1–6 months sick leave) and compare to other questionnaires such as Job Strain, the Coping Strategies Questionnaire (CSQ), the Pain Catastrophizing Scale (PCS) and the Tampa Scale for Kinesiophobia (TSK). The main purpose of this paper was to study the relationship between risk lev¨ MPSQ and sick leave and perceived els found from the O health 3 years later. ¨ MPSQ would Our specific hypothesis was that the O predict future sick leave and health, but that because the current sample had a longer history of a problem at recruitment, that this relationship would be moderate. A second hypothesis was that the other questionnaires included in the study would be predictive of outcome, ¨ MPSQ. but to a lesser degree than the O
3. Method 3.1. Design During the period 1998–2000, a rehabilitation project was carried out in the Primary Health Care settings in the county of Va¨stmanland in Sweden. One hundred and fifty-eight patients with musculoskeletal pain and disability were recruited to the project. Eighty-seven patients from two primary health care units participated in a multimodal rehabilitation programme and 71 patients with the same inclusion criteria from four other primary health care units received treatment ‘‘as usual”. Because patients normally receive a variety of treatments in primary health care we chose to do the data analysis for the entire group (n = 158). An evaluation of the rehabilitation programme will be carried out in a forthcoming study. 3.2. Patients The inclusion criteria were those between 18 and 65 years old, sick listed P28 days 6180 days and/or had consulted the doctor about the same problem P3 times the last 12 months according to information from the
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
referring physicians. The participants either had permanent employment or were at the disposal of the labour market. Patients requiring orthopaedic surgery and patients with a psychiatric disorder or a substance abuse problem were excluded. Furthermore, participants had to be able to speak Swedish sufficiently well in order to describe their symptoms and understand the information given. The observation period of this study was 3 years. At the 3-year follow-up there were 11 drop-outs. The dropout analysis shows a gender difference where 71% of the participants in the study group were women, compared with 54% of the drop-outs who were women (Table 1).
643
3.3. Measure At baseline and at the 3-year follow-up visit, the participants completed a battery of questionnaires concerning sick leave, work, function, and quality of life, as well as a screening questionnaire for the early identification of people at risk for developing long-term problems. Supplemental questionnaires (e.g., CSQ, TSK, PCS, and life events) were used in order to see if those had ¨ MPSQ. better predictive value compared with O
4. Assessments ¨ rebro Musculoskeletal Pain Screening 4.1. The O ¨ MPSQ) Questionnaire (O
Table 1 Baseline characteristics of the patients (n = 158) Frequency
(%)
Age (mean) Sex (men/women)
47 (range 24–65) 48/110
30/70
Educational level Elementary school only High school University
71 63 14
48 43 10
Occupational status Employed Unemployed Students Other
115 29 4 9
73 19 3 6
Primary pain site Neck Shoulder/shoulder-joint Back – upper part Back – lower part Leg Other symptoms
97 105 48 85 55 57
62 70 31 85 35 37
Pain duration in weeks (n = 156) 0–11 weeks 12–23 weeks 24–52 weeks P53 weeks
24 23 24 85
15.3 14.7 15.4 54.5
Sick leave at baseline 25% sick leave 50% sick leave 75% sick leave 100% sick leave Missing
125 5 15 4 98 3
79 4.1 12.3 3.3 80.3 4
Sick leave, days previous 12 months 0–30 days 31–60 days 61–90 days 91–180 days
49 46 21 40
31 30 13 26
Disability benefits at baseline
16
10.5
Mean ¨ MPSQ) Screening index at baseline (O Experimental group 121 Control group 124
SD 21.4 22.7
¨ MPSQ was developed as a tool for clinicians in the O early identification of people at risk for developing longterm problems. The questionnaire contains 25 items divided into five groups (function, pain, psychological factors, fear avoidance, and miscellaneous) [i.e., sick leave, age, gender, nationality, monotonous or heavy work, job satisfaction]. Twenty-one of the 25 items are scored on a 0–10 scale, yielding a total score range of 0–210. Scores are tallied to form a total score. The actual experience of pain is explored in several questions about site, intensity, duration, and frequency of pain. Ability to perform daily activities is assessed with five items. Participants rate their current ability to carry out light work, walk for an hour, do household chores, the weekly shopping and sleep. Perceived ability to cope with pain and perceived job satisfaction are appraised with two single items. Fear-avoidance beliefs are estimated with three assertions where the patient rates the degree to which she/he agrees. The patient’s own perception of their risk is measured with two items developed especially for this questionnaire. 4.2. Utility In the first study conducted with the instrument, the screening instrument was found to have satisfactory test–retest reliability (0.83) and validity in a study of 142 patients where outcome was sick absenteeism (Linton and Hallde´n, 1997; Linton and Hallde´n, 1998). When using a cut-off score of 105 (the maximum being 210), the specificity was found to be 0.75 and the sensitivity 0.88. At the same cut-off point the chance of accurately classifying someone as ‘‘at risk” for future sick leave would be 88%. Kendall (1999) tested the instrument on a population of more than hundred acute-pain patients in New Zealand and found that a cut-off score of 105 was related to future disability. Hurley et al. (2000) investigated the instruments predictive ability with regard to
644
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
return-to-work after physical therapy. They reported that a cut-off point of 112 correctly identified 80% of patients failing to return to work at the end of treatment (sensitivity), while at the same time 59% of those returning to work were identified (specificity). Boersma et al. showed in a review study 2002 that the questionnaire appears to be most helpful in predicting future functional problems as well as sick absenteeism. The predictive validity of the instrument for pain was low, relative to the predictive validity for function and absenteeism. In an Australian study, Dunstan et al. ¨ MPSQ in a compensable injury (2005) evaluated the O population and found that the scores were related to long-term disability. The findings had significant relevance to the recommendation of routine screening for the early identification of injured workers at risk of long-term disability. 4.3. The Short Form-36 Health Survey The generic measure, SF-36, was developed in the United States. It covers both functioning and wellbeing and has proved suitable for clinical research, patient monitoring, and health care planning. The SF-36 is a 36-item questionnaire that assesses healthrelated quality of life from the point of view of the health care recipient. Ware et al. developed the SF-36 as a comprehensive, psychometrically sound, and brief, health outcome measure. The research also designed it to be generic so that it could be applied to a wide variety of health conditions. The SF-36 describes both the physical and mental components of health. Eight primary SF-36 scales from distinct physical and mental health clusters are based on factor analyses. For each of the eight scales, scores range from 0 to 100, with higher scores reflecting better self-reported health status. Continuous method studies and empirical testing have shown that the scales in SF-36 can be summed up to two overall health indexes – Physical Health Score (PCS) respective Mental Health Score (MCS) (Ware et al., 1994). 4.4. Job Strain This instrument contains 11 items (graded 1–4) concerning demands and control. Indices for work demand and control are calculated. High scores correspond to high demands and high control. By dividing demands by control, a measure of job strain is obtained for each patient (Theorell et al., 1988). 4.5. Coping Strategies Questionnaire (CSQ) The questionnaire assesses eight different coping strategies for pain. Each strategy consists of six different items. On a scale ranging from 0–6, subjects are asked to
indicate how often they use a particular item when they experience pain (0 = never use it, 6 = always). A total score of 36 can be obtained on each of the eight coping strategies. In this study we restricted the questionnaire to assess the coping strategy: Coping self-statements (Rosenstiel and Keefe, 1983; Jensen and Linton, 1993). 4.6. Pain Catastrophizing Scale (PCS) The PCS instruction asks participants to reflect on past painful experiences and to indicate the degree to which they experienced each of 13 thoughts or feelings when experiencing pain on a five-point scale from 0 (not at all) to 4 (all the time) (Sullivan et al., 1995; Osman et al., 2000). 4.7. Tampa Scale for Kinesiophobia (TSK) The Scale for Kinesiophobia (TSK) developed by Miller et al. as a measure of fear of movement/(re)injury. Each of the 17 items is provided with a four-point Likert Scale with scoring alternatives ranking from ‘‘strongly disagree” to ‘‘strongly agree” (Vlaeyen et al., 1995a). In this study, we used a short version with 12 items. In accordance with the psychometric work done by Vlaeyen et al. (1995b) five items that loaded poorly in factor analysis were left out, and a total score was used. 4.8. Life events A continuous variable ‘‘life events” was computed from answers regarding 17 possible major events which could have occurred during the preceding 12 months (12 negative, 1 positive, 4 neutral) (Roll and Theorell, 1987; Theorell and Emlund, 1993). 4.9. Sick leave Information on sick leave was obtained at baseline and at the follow-ups by the patients reporting their own current sick leave. Previous research has shown that these ratings highly correlate to register data from the National Social Insurance office (Linton et al., 1995; Hensing et al., 1998). ‘‘Improved” is defined as a patient who has decreased her/his sick leave level. ‘‘Impaired” is defined as a patient who maintains or increases her/his sick leave level at the follow-up.
5. Statistics The SPSS 11.0 statistical programme was used for the data analysis. Sensitivity and specificity are important properties when evaluating a questionnaire. The relationship between sensitivity and specificity can be studied by constructing the Receiver Operator Characteristic
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
(ROC) curve. Principal Component Analysis was used ¨ MPSQ in order to reduce the dimenfor the items in the O sionality and find relevant factors. Only factors with an eigen value P1 were used (Everitt and Dunn, 1991). To assess the predictive power of different factors on sick leave, stepwise logistic regression was used. Adjustments were made for the variables age, sex, education level and previous sick leave if their p-value were less than 0.2. The predictive power of the different factors on perceived mental and physical health was studied by General Linear Model (GLM). Continuous variables were presented as mean and standard deviation, and categorical data as counts and percentages. A two sided p-value <0.05 was considered significant. Since there were no convincing outcome difference between the control group and the experimental group the data analyses were conducted on the whole group.
6. Ethics The Human Research Ethics Committee at the University of Uppsala, Sweden approved the study.
645
months, on average, how bad was your pain?” (0.79) ‘‘How often would you say that you have experienced pain episodes, on average, during the 3 past months?” (0.81). (III) Distress: ‘‘How tense or anxious have you felt in the past week?” (0.86) ‘‘How much have you been bothered by feeling depressed in the past week?” (0.89). (IV) Fear-avoidance: ‘‘Physical activity makes my pain worse” (0.50). ‘‘An increase in pain is an indication that I should stop what I am doing until the pain decreases” (0.82). ‘‘I should not do my normal work with my present pain” (0.76). (V) Return to work expectancy: ‘‘In your view, how large is the risk that your current pain may become persistent? (0.61), ‘‘In your estimation, what are the chances that you will be working in 6 months?” (0.79) ‘‘If you take into consideration your work routine, management, salary, promotion possibilities and work mates, how satisfied are you with your job?” (0.76). (VI) Coping:‘‘Based on all the things you do to cope, or deal with your pain, on an average day, how much are you able to decrease it?” (0.92). 7.2. Predictive ability of the factors
7. Results 7.1. Total scores of all questionnaires (Table 2) ¨ rebro Musculoskeletal 7.1.1. The factor analysis of the O ¨ MPSQ) Pain Screening Questionnaire (O In order to reduce the items in the questionnaire the structure was examined by Factor Analysis. The items were reduced to six disjoint factors (factor loading within brackets). (I) Function: ‘‘I can do light work for an hour” (0.67), ‘‘I can walk for an hour” (0.66), ‘‘I can do ordinary household chores” (0.68), I can go shopping” (0.68), ‘‘I can sleep at night” (0.47). (II) Pain: ‘‘How would you rate the pain that you have had during the past week?” (0.70), ‘‘In the past 3
To study which of the factors were important for long-term outcome we performed separate analyses for each factor. An overview is presented in Fig. 1 and shows the predictive power of each factor. 7.3. Factors predicting of sick leave Because the outcome was dichotomous (‘‘improved”, ‘‘impaired” sick leave) stepwise multiple logistic regression was used to analyse the predictive ability of the factors. Adjusting for age and earlier sick leave (p less than 0.2) factor I (function) and factor II (pain) significantly predicted sick leave after 3 years as seen in Table 3. This model had a predictive ability of correctly classifying 71% of the participants with a sensitivity of 63% and a specificity of 77%. No other factors were significant.
Table 2 Scores of all questionnaires, mean and standard deviation, at baseline and at 3-year follow-up Questionnaires
n
Baseline
3-Year follow-up
Mean
SD
Mean
SD
¨ MPQ total score O Job Strain Coping Strategies Questionnaire (CSQ) Pain Catastrophizing Scale (PCS) Tampa Scale for Kinesiophobia (TSK) Life events SF-36, Mental Health Score (MCS) SF-36, Physical Health Score (PCS) Psychosomatic symptoms
149 61 116 113 125 110 111 111 124
123 0.9 2.9 18.4 26.3 2.7 38.8 40.8 1.3
22 0.3 1.2 10 6.6 2.2 8.4 5.8 0.7
94 0.8 3.3 16.9 12.7 2.5 45.1 37.4 1.5
43 0.2 1.1 10.8 7.1 2.1 12.1 10.2 0.6
646
A. Westman et al. / European Journal of Pain 12 (2008) 641–649 Table 5 Prediction of self-reported health (SF-36, Physical Health Score) analyzed by General Linear Model Regression coefficient Factor I (function) Factor II (pain) Factor IV (fear-avoidance) Age
95% CI
p-Value
2.056 1.405 0.963
3.024 to 1.087 2.638 to 0.173 0.039 to 1.964
0.269
0.445 to
<0.001 0.026 0.059
0.092
0.003
Fig. 1. Increased sick leave or maintained total work disability. Unadjusted odds ratio and 95% CI for the different factors. Table 3 Prediction of sick leave at 3-year follow-up analysed by stepwise multiple logistic regression analysis
Factor I (function) Factor II (pain) Age Earlier sick leave
Adjusted OR
95% CI
p-Value
1.4 1.3 1.04 1.15
1.152–1.762 1.017–1.696 1.002–1.079 0.946–1.403
0.001 0.037 0.04 0.16
7.4. Predictors of perceived health – Mental Health Score (MCS) Because this measure consists of continuous outcome variables, General Linear Model (GLM) was used for the prediction of perceived health. After adjusting for age, Factor III (Distress) predicted perceived health significantly and Factor V (return to work expectancy) predicted perceived health almost significantly (p < 0.001 and 0.082, respectively), as measured by SF-36 (Mental Health Score). No other factors were significant (Table 4). 7.5. Predictors of perceived health – Physical Health Score (PCS) Adjusting for age, factor I (function) and factor II (pain) predicted perceived health as measured by SFTable 4 Prediction of self-reported health (SF-36, Mental Health Score) analyzed by General Linear Model Regression coefficient Factor III (distress) Factor V (return to work expectancy) Age
1.509 0.922 0.241
95% CI 2.296 to 0.721 1.963 to 0.120 0.035 to 0.447
p-Value <0.001 0.082 0.022
¨ MPSQ Fig. 2. Receiver Operator Characteristic (ROC) curve for O total score’s ability to predict the sick leave status at the 3-year followup.
36 (Physical Health Score) significantly (p < 0.001 and 0.026, respectively). Factor IV was borderline significant (p = 0.059). No other factors were significant (Table 5). ¨ MPSQ score 7.6. Predictive ability of the overall O 7.6.1. Receiver Operator Characteristic (ROC) curve Cut-off scores are another way to assess the predictive ¨ MSPQ. A Receiver Operator Charactervalidity of the O istic curve (ROC) shows the specificity and sensitivity of ¨ MPSQ as the measure for selected total scores on the O seen in Fig. 2. A cutoff ‘‘at-risk” score of 117 correctly classified (sensitivity) 78% of the poor outcomes (failed to reduce sick leave) and a cut-off score of 139 correctly classified 44% of those who failed to reduce their sick leave. For the same score levels 49% and 89% of those who succeeded in reducing their sick leave were correctly classified (specificity). 7.6.2. Other questionnaires The questionnaires Job Strain, Coping Strategies Questionnaire (CSQ), Pain Catastrophizing Scale (PCS), Tampa Scale for Kinesiophobia (TSK) and Life events, did not predict sick leave, or perceived health, significantly at the 3-year follow-up.
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
8. Discussion This study demonstrates that psychosocial factors as ¨ MPSQ are related to work disability and measured by O perceived health even 3 years after treatment in primary care. Moreover, the screening questionnaire had discriminative power even for patients with non-acute or ¨ MPSQ had better recurrent pain problems. Indeed, the O predictive power than any of the questionnaires included in the study. This study shows that among the factors, pain and function are the factors most strongly related to sick leave 3 years later. ¨ MPSQ The prediction of future sick leave with the O is based on the total score. As in earlier studies the results showed that the total score of the screening questionnaire was related to future sick leave and functional ability; the higher the score, the higher the risk for long term sick leave and developing of chronic problems (Linton and Hallde´n, 1997; Linton and Hallde´n, 1998; Hurley et al., 2000; Boersma and Linton, 2002). The factor analysis indicated six clear factors that reflect the original theoretical conceptualization of the questionnaire (Linton and Hallde´n, 1998). However, the results of the long-term follow-up do emphasize the importance of psychosocial factors in prediction. Function with focus on daily living, sleep capacity and pain experience had the most powerful predictive value concerning sick leave at 3 years. While earlier studies have shown that emotional and cognitive variables such as distress and fear avoidance beliefs have been strong predictors for 6–12-month outcomes, the best predictor in this study is having problems functioning. This probably reflects the length of the follow-up and suggests that different variables may be predictive at various stages in the process of chronification. Even though it is recognized that psychological variables are influential factors, little is known about how and when these variables interact in the process toward disability. Furthermore, psychological variables might operate differently for different people and at different time points. Boesrma et al., for example, showed how four distinct cluster profiles varied in their development of problems over time (Boersma and Linton, 2006). The most powerful predictors of mental health at the 3-year follow-up were the variables distress and work expectancy. The fact that distress (consisting of anxiety and mood) predicts mental health is not surprising, but does highlight the role of emotional variables in the development of chronic problem. More surprising, is the fact that work expectancy predicts mental health at the follow-up. Although work expectancies predicted future mental health, it is interesting that they were not very good predictors of actual sick days which is also contrary to previous findings (Linton and Hallde´n, 1998). To achieve reasonable predictability, we needed to have a cut-off score of 117. This is slightly higher than
647
some other studies where similar prediction levels have been obtained with scores of 105–112. The slightly higher cut-off point is not surprising since the participants had longer histories of pain problems. One explanation might be that this sample is different than those in other studies, e.g. having more disability at inclusion. However, this mirror the fact that work is an important factor in mental health. Another explanation may be that the work expectancy factor taps into negative expectancies in general which in turn are related to future mental health. Physical health was significantly predicted by the function and pain factors. Although the pain factor was marginally significant, the result is consistent with a large body of literature showing that current pain conditions predict future pain conditions (Pulliam and Gatchel, 2002). Likewise, function has previously been found to predict future health problems such as being off work and functional problems. Thus, this finding is quite consistent with previous work in the field. Earlier prospective studies have shown that painrelated fear is a predictor of future disability in patients suffering from an acute episode of back pain (Klenerman et al., 1995; Fritz and George, 2002). Moreover cross-sectional studies have shown that pain catastrophizing is associated with psychological distress and pain related disability as well as social functioning and general health inpatients with back pain (Severeijns et al., 2005). However, in our study these variables did not show significant power. One reason could probably be that the items about function and pain are simply such strong predictors that they eliminate fear and catastrophizing. Since the outcome variable is sick leave, function and pain may be directly related and, thus, reflect the reason for being off work rather than the cause. Another possibility mentioned above is that fear and catastrophizing may be important earlier in the developmental process while function becomes important at a later stage. The patient group in the study had longer and more severe condition as compared to several other studies ¨ MPSQ has been assessed. This may also where the O be an explanation as to why factors like fear did not show as much predictive value as in earlier studies. Another interesting finding concerns the role of sick leave as a predictor. This variable has consistently been shown to be a strong predictor of long-term problems and future sick leave. Consequently, we adjusted our analyses for earlier sick leave and this might be of importance in interpreting the result in our study. Function and pain experience was the most predictive variable in our study. The high disability level at baseline probably played a role whereby questions about daily living and pain experiences probably had a more comprehensive importance and showed the strongest predictive value. However, the genesis of musculoskeletal
648
A. Westman et al. / European Journal of Pain 12 (2008) 641–649
pain is complex and often of multifactorial origin. The research during the last decades has focused on several important variables predicting long-term problems. For the general practitioner who often see many patients with a variety of symptoms it is a demanding task to observe patients with risk for long-term problems. This study has certain methodological facts that should be kept in mind when interpreting the results. Firstly, the subjects consisted of an experimental and a control group with different kinds of treatments. However, because an analysis of the effects of treatment did not produce any significant differences, we chose to pool the data to gain power and clarity in interpreting the results. Secondly, the long follow-up period is, perhaps, worthy of consideration. While this offers advantages it also opens the door for other events that may influence outcome. Thus, a long follow-up probably reduces the size of the impact of the predictor variables. Earlier studies have suggested that the instrument could be of value in isolating patients in need of early interventions and in promoting the use of appropriate interventions for patients with psychological risk factors. The results of our study are in line with other studies though with lower predictive value. One conclusion must be to emphasize the value of using the instrument ¨ MPSQ earlier in the course of the pain problem. The O questionnaire could be of importance for the general practitioner and other caregivers in Primary care as a complement for patients with musculoskeletal pain and disability. Moreover the instrument appears to be most helpful in predicting future functional problems as well as sick absenteeism.
Acknowledgements The County Council of Va¨stmanland, Sweden and the Social Insurance Office, in co-operation, supported this study.
References Boersma K, Linton SJ. Early assessment of Psychological factors: the ¨ rebro Screening Questionnaire for Pain. In: Linton SJ, editor. O New avenues for the prevention of chronic musculoskeletal pain and disability. Amsterdam: Elsevier Science B.V.; 2002. p. 205–13. Boersma K, Linton SJ. Psychological processes underlying the development of a chronic pain problem: a prospective study of the relationship between profiles of psychological variables in the fear-avoidance model and disability. Clin J Pain 2006;22:160–6. Burton AK, Tillotson KM, Main CJ, Hollis S. Psychosocial predictors of outcome in acute and subchronic low back trouble. Spine 1995;20:722–8. Dunstan DA, Covic T, Tyson GA, Lennie IG. Does the Orebro Musculoskeletal Pain Questionnaire predict outcomes following a work-related compensable injury? Int J Rehabil Res 2005;28: 369–70.
Everitt BS, Dunn G. Appl Multivar Data Anal 1991. Fritz JM, George SZ. Identifying psychosocial variables in patients with acute work-related low back pain: the importance of fearavoidance beliefs. Phys Ther 2002;82:973–83. Gatchel RJ, Gardea MA. Psychosocial issues: their importance in predicting disability, response to treatment, and search for compensation. Neurol Clin 1999;17:149–66. Hensing G, Alexanderson K, Allebeck P, Bjurulf P. How to measure sickness absence? Literature review and suggestion of five basic measures. Scand J Soc Med 1998;26:133–44. Hurley DA, Dusoir TE, McDonough SM, Moore AP, Linton SJ, Baxter GD. Biopsychosocial screening questionnaire for patients with low back pain: preliminary report of utility in physiotherapy practice in Northern Ireland. Clin J Pain 2000;16:214–28. Hurley DA, Dusoir TE, McDonough SM, Moore AP, Baxter GD. How effective is the acute low back pain screening questionnaire for predicting 1-year follow-up in patients with low back pain? Clin J Pain 2001;17:256–63. Jensen I, Linton SJ. Coping Strategies Questionnaire (CSQ): reliability of the Swedish version of the CSQ. Scand J Behav Ther 1993;22:139–45. Kendall N. Screening and early intervention: the New Zealand experience. In: Paper presented at the international association for the study of pain; 1999. Klenerman L, Slade PD, Stanley IM, Pennie B, Reilly JP, Atchison LE, et al. The prediction of chronicity in patients with an acute attack of low back pain in a general practice setting. Spine 1995;20:478–84. Leeuw M, Goossens ME, Linton SJ, Crombez G, Boersma K, Vlaeyen JW. The fear-avoidance model of musculoskeletal pain: current state of scientific evidence. J Behav Med 2007;30:77–94. Linton SJ. A review of psychological risk factors in back and neck pain. Spine 2000;25:1148–56. Linton SJ. Why does chronic pain develop? A behavioral approach. In: Linton SJ, editor. New Avenues for the prevention of chronic musculoskeletal pain and disability. Pain research and clinical management, vol. 12. Amsterdam: Elsevier Science B.V.; 2002. p. 67–82. Linton SJ, Boersma K. Early identification of patients at risk of developing a persistent back problem: the predictive validity of the Orebro Musculoskeletal Pain Questionnaire. Clin J Pain 2003;19:80–6. Linton SJ, Hallde´n K. Risk factors and the natural course of acute recurrent musculoskeletal pain: developing a screening instrument. In: Jensen TS, Turner JA, Wisenfeld-Hallin, editors. Progress in pain research and management. Seattle: IASP Press; 1997. p. 527–36. Linton SJ, Hallde´n K. Can we screen for problematic back pain? A screening questionnaire for predicting outcome in acute and subacute back pain. Clin J Pain 1998;14:209–15. Linton SJ, Hallde´n K, Hellsing A. The reliability of self-reported sick absenteeism: a pilot study. Scand J Behav Ther 1995;24:145–50. Main CJ, Waddell G. A comparison of cognitive measures in low back pain: statistical structure and clinical validity at initial assessment. Pain 1991;46:287–98. McCracken LM, Zayfert C, Gross RT. The Pain Anxiety Symptoms Scale: development and validation of a scale to measure fear of pain. Pain 1992;50:67–73. Nachemson A, Wedell G, Nordlund AI. Epidemiology of neck and low back pain. In: Nachemson A, Jonsson E, editors. Neck and back pain. The scientific evidence of causes, diagnosis and treatment. Philadelphia: Lippincott Williams & Wilkins; 2000. p. 165–87. Osman A, Barrios FX, Gutierrez PM, Kopper BA, Merrifield T, Grittmann L. The Pain Catastrophizing Scale: further psychometric evaluation with adult samples. J Behav Med 2000;23:351–65. Pincus T, Burton AK, Vogel S, Field AP. A systematic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain. Spine 2002;27:E109–20.
A. Westman et al. / European Journal of Pain 12 (2008) 641–649 Pincus T, Vogel S, Burton AK, Santos R, Field AP. Fear avoidance and prognosis in back pain: a systematic review and synthesis of current evidence. Arthrit Rheum 2006;54:3999–4010. Pulliam CB, Gatchel RJ. Employing risk factors for screening of chronic pain disability. In: Linton SJ, editor. New avenues for the prevention of chronic musculoskeletal pain and disability. Amsterdam: Elsevier Science B.V.; 2002. p. 183–204. Reid S, Haugh LD, Hazard RG, Tripathi M. Occupational low back pain: recovery curves and factors associated with disability. J Occup Rehab 1997;7(1):1–14. Roll M, Theorell T. Acute chest pain without obvious organic cause before age 40 – personality and recent life events. J Psychosom Res 1987;31:215–21. Rosenstiel AK, Keefe FJ. The use of coping strategies in chronic low back pain patients: relationship to patient characteristics and current adjustment. Pain 1983;17:33–44. Severeijns R, Vlaeyen JW, van den Hout MA, Picavet HS. Pain catastrophizing and consequences of musculoskeletal pain: a prospective study in the Dutch community. J Pain 2005;6: 125–32. Sullivan MJL, Bishop SRB, Pivik J. The Pain Catastrophizing Scale: development and validation. Psychol Assess 1995;7:524–32. Theorell T, Emlund N. On physiological effects of positive and negative life changes – a longitudinal study. J Psychosom Res 1993;37:653–9.
649
Theorell T, Perski A, Akerstedt T, Sigala F, Ahlberg-Hulten G, Svensson J, et al. Changes in job strain in relation to changes in physiological state. A longitudinal study. Scand J Work Environ Health 1988;14:189–96. Turk DC. The role of demographic and psychosocial factors in transition from acute to chronic pain. JASP Press; 1997, p. 185– 213. Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. Pain 2000;85: 317–32. Vlaeyen JW, Kole-Snijders AM, Boeren RG, van Eek H. Fear of movement/(re)injury in chronic low back pain and its relation to behavioral performance. Pain 1995a;62:363–72. Vlaeyen JW, Kole-Snijders AM, Rotteveel A, Ruesink R, Heuts PHTG. The role of fear of movement/(re)injury in pain disability. J Occup Rehab 1995b;5:235–52. von Korff M. Pain management in primary care: an individualized stepped-care approach. The Guiford Press; 1999, p. 360–73. Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A FearAvoidance Beliefs Questionnaire (FABQ) and the role of fearavoidance beliefs in chronic low back pain and disability. Pain 1993;52:157–68. Ware JE, Kosinski M, Keller SD. SF-36 physical and mental health summary scales: a user’s manual. MA: New England Medical Center, The Health Institute; 1994.