Estlander, Ann-Mari PhD; Takala, Esa-Pekka MD, DMed; Viikari-Juntura, Eira MD, DMed
The influence of psychological variables on musculoskeletal pain has been the focus of growing interest during the last decade. The assumption has been that, for example, symptoms of psychological distress and depression, the subject's own predictions, and work-related subjective experiences such as work satisfaction determine recovery, the process of chronicity, and treatment outcome (for reviews, see references 1, 2, and 3). Early studies were mainly cross-sectional and based on simple models, and a majority of the recently published prospective studies in this field have been based on clinical populations. The associations in these studies have, however, generally been weak, the predictors have explained only a small amount of the variance in the outcome, and different studies have shown contradictory results. The necessity of performing prospective studies including more complex models to determine cause and effect and interrelationships of various variables has been emphasized by several researchers (eg, references 1, 3, and 4-6).
Epidemiological studies among normal populations or various worker groups have provided some information about the significance of psychological factors for future morbidity. For example, Leino and Magni7 and Magni et al8 found relationships between depression and future musculoskeletal pain, and Leino and Hänninen9 demonstrated associations between psychosocial factors at work and musculoskeletal disorders. Pietri-Taleb et al10 found occupation-specific relationships between personality characteristics, psychological distress, and neck trouble. However, longitudinal studies are also hampered with methodological problems, and the results from different studies are contradictory.3 The interrelationships between the predictors or potential "risk factors," the influence of earlier morbidity or symptoms, the appropriateness and limitations of the statistical methods, and the psychometric properties of the assessment instruments are among the major concerns.
Bongers et al3 presented a review based on 44 cross-sectional and 15 longitudinal studies of psychosocial factors at work and musculoskeletal disease. The authors identified five categories of factors possibly related to musculoskeletal symptoms: (1) psychosocial factors at work (demands and control); (2) psychosocial factors at work(social support); (3) individual characteristics such as psychological dysfunction; (4) stress symptoms such as worry, tension, and exhaustion; and(5) physical and behavioral health indicators such as symptoms of physical illness. Their review showed that, for example, social support, time pressure, low control at work, and high job demands have been related to musculoskeletal symptoms. They presented a model of the mechanisms, concluding that psychosocial work factors may increase work-related stress, especially when coping strategies are inadequate. Stress may lead to a development of musculoskeletal pain either through physiological mechanisms such as increased muscle tone, or to an increased perception of pain. They also concluded that even if several emotional and psychological problems are related to various aspects of musculoskeletal pain, the results from various studies are contradictory and the role of these variables is not clear.
Psychological distress, depression, and cognitive evaluations such as self-efficacy beliefs are presumed to influence several aspects of musculoskeletal symptoms.4,5 Assessment methods have been developed during the last decades for symptoms of distress and depression in persons with musculoskeletal symptoms, especially low back pain. Psychometric tests originally developed for other conditions have been applied to and evaluated in musculoskeletal patients. The Modified Somatic Perception Questionnaire11 and a modification of the Zung Depression Scale12,13 have been evaluated in several studies and found useful, reliable, and valid in back pain populations.14,15 Main et al14 presented a patient classification method to identify distress and evaluate the risk of poor outcome based on the Modified Somatic Perception Questionnaire and the modified Zung depression scale. Based on scores of these two questionnaires, they identified four types of patients: no distress, at risk of developing psychological distress, distressed depressive, and distressed somatic. This classification has later been used and found to be useful in other studies (eg,reference 15). Self-efficacy beliefs, ie, a person's belief in his/her capability to perform a required behavior or manage in a specific situation, have been consistently found to predict health-related behavior.16 Questionnaires for assessing various aspects of self-efficacy in low back patients in particular have been developed during the last decade.17 Estlander et al18 developed a questionnaire for assessing activity-related self-efficacy beliefs in low back pain patients and found that these beliefs predict physical performance. The properties and prognostic value of self-efficacy scales are not well known in healthy populations or in persons reporting some, but not clearly disabling, musculoskeletal problems.
The objective of this study was to determine the predictive value of some psychometric assessment instruments for the development and persistence of musculoskeletal pain in one and two years' follow-up in a working population. The hypotheses were (a) high degree of psychological distress and negative self-efficacy beliefs in healthy persons predict the development of musculoskeletal pain, and (b) psychological distress in persons with musculoskeletal pain predicts persistent pain.
Subjects and Methods
A prospective cohort of 141 women and 311 men was selected based on a questionnaire survey on pain in the low back, neck, and shoulder during the previous 12 months. The outcome was a grouping based on the cumulative duration of these symptoms after one and two years, as inquired with a questionnaire. The predictors were psychometric questionnaire data and other background information at baseline.
A population of 6848 workers participated in a questionnaire survey in a large forest-industry enterprise. Postal surveys with identical questions of musculoskeletal pain were carried out in March of 1992, 1993, 1994, and 1995. Each year, new questionnaires were mailed twice to nonrespondent subjects.
Pain of the low back, neck, and shoulder was investigated using a modified version of the Nordic questionnaire.19,20 The questions regarding the low back comprised three types of pain: radiating pain (sciatic pain), sudden attacks of pain (lumbago), and other kinds of low back pain. Two modalities of neck pain were questioned: radiating and other kinds of pain. Shoulder pain was inquired about with one question. The form of the question was "Estimate the total number of days you have had (this type of) pain (in this anatomic area) during the preceding 12 months?" and the response categories were: "not any," "1 to 7 days," "8 to 30 days," "more than 30 days but not daily," and "daily." In the analysis, the responses were combined into three categories: <8 days, 8-30 days, and >30 days. In the preliminary analysis, the associations between the psychometric variables and the different modalities (radiating vs nonradiating) or anatomic areas(low back vs neck and shoulder) were similar. Therefore, the anatomic areas and all pain modalities were combined. The longest total duration of days with any type of pain in any of these areas during the preceding 12 months represented the pain status of the subject.
Based on the results of the surveys in 1992 and 1993, a set of tests measuring the function of the musculoskeletal system was performed for 141 women and 311 men in the laboratory between February and August 1993. The subjects were sampled in order to have representative groups of subjects as regards contrasts in outcome (musculoskeletal pain) and potential confounding factors (workload, gender, and age). The main selection criteria were duration of pain in the back, neck, and shoulders, and workload. The subject was classified as "nonsymptomatic" if the total number of days with pain in all these body areas during the preceding 12 months was less than 8 days and"symptomatic" if the number of days with pain in any of these locations exceeded 30 days. The physical workload was classified according to the job title as "light" (office work and sedentary tasks; 70 women and 59 men) or "moderately strenuous" (cleaning, various tasks at the paper machine, maintenance tasks; 71 women and 252 men). Subjects older than 54 years were excluded in order to avoid potential dropouts due to retire during the follow-up.
The outcome was based on the responses of musculoskeletal pain in 1994 and 1995 as compared with 1993; ie, one year and two years' follow-up. In retrospective questions, misclassification bias is most probable between adjacent classes of duration of pain: ie, "<8 days" versus "8-30 days" and"8-30 days" versus ">30 days." With the aim of reducing the effects of this potential bias, two kinds of criteria for the changes in the one and two years' follow-up were formed:
"Soft" criteria included subjects who had any change from the baseline class. "Nonsymptomatic" subjects in the beginning of the study were classified as "painfree" if they had pain during <8 days in the follow-up; if they had pain during >8 days they were classified as"contracted pain." Similarly, "symptomatic" subjects in the beginning of the study were classified as "persistent pain" if during the follow-up they had pain during >30 days and as "recovered" if the number of days with pain was <30.
"Strict" criteria included only changes greater than to the adjacent class. The aim of the "strict" criteria was to reduce the possible misclassification bias, but its cost was a reduced number of subjects. The subjects without any change during the follow-up ("painfree" and "persistent pain") were the same in the "soft" and "strict" classification."Non-symptomatic" subjects in the beginning of the study were classified as"contracted pain" if the total number of days with pain was at least 30 in the follow-up. In the beginning of the study, "symptomatic" subjects were classified as "recovered" if their number of days with pain was < 8 in the follow-up.
Tables 1 and 2 show the formation of groups and number of subjects with "soft" and "strict" criteria for one and two years' follow-up.
The classification for two years' follow-up with the "soft" and "strict" criteria was made independently of the results of the first-year follow-up. Of the 365 subjects who had responded to both follow-up surveys, 137 had changed their pain status from the first to the second follow-up. Therefore an additional classification for the whole period of two years was formed in order to take into consideration the data of all three assessments ("trend" criteria). The subjects were classified as "painfree" (n = 96) or"persistent pain" (n = 90) only if they had had no changes after the baseline. The category of "contracted pain" (n = 29) included workers with a change from the baseline "nonsymptomatic" status (<8 days) to the "symptomatic" (>30 days), and the "recovered" group (n = 35) included those with a change from the baseline "symptomatic" (>30 days) status to "nonsymptomatic" (<8 days) in either year.
Of the potential confounders, the distributions of sex, workload, stature, weight, and body mass index (BMI = weight · stature-2; kg · m-2) were similar in the "painfree" vs "contracted pain" groups and in the "persistent pain" vs "recovered" groups. A difference of about three years was seen in the mean age between the "recovered" vs"persistent pain" groups in all classifications. The "persistent pain" subjects had, on average, a slightly higher BMI than those in the other groups. Table 3 shows the mean values of these variables for the one-year follow-up classification with "soft" criteria. The distributions were similar in the other classifications.
The potential confounding variables, as well as the psychological variables, were equally distributed at baseline among the respondents and nonrespondents.
The psychological assessment was based on self-administered questionnaires and covered the following areas: distress symptoms, depression, self-efficacy beliefs, prediction of future ability to work, physical workload, and psychosocial work factors. The subjects filled in these questionnaires during the laboratory visit in 1993, and the same questionnaires were included in the 1995 survey.
The Modified Somatic Perception Questionnaire (MSPQ) developed by Main11 assesses heightened somatic awareness or"somatic anxiety," a form of distress. The psychometric properties and usefulness of the MSPQ in musculoskeletal pain have been investigated in several studies (eg, references 14, 15, and 21). The test includes 13 items describing various somatic symptoms. The subject rates, on a 0- to 3-point scale, low much he/she has experienced the symptoms during the last week (0 = "Not at all," 1 = "A little, sometimes," 2= "A great deal, often," and 3 = "Very much, almost all the time"). A sum score (theoretical range, 0-39) was calculated, a higher sum score indicating more symptoms. For this study, an alternative sum score of the MSPQ (MSPQ-n) was also calculated when the neck pain item was omitted. This was done because neck pain was included in the outcome criteria. The theoretical range of this sum score was 0-36.
The modified Zung depression scale (ZUNG) has been widely used in low back pain populations. It comprises 20 items from the original Zung depression scale12 plus three additional items.13 The scoring used in this study followed the procedure used by Main et al14 (0 = "rarely or none of the time," 1 = "some or little of the time; 1-2 days/week," 2 = "a moderate amount of time; 3-4 days/week," and 3 = "most of the time; 5-7 days/week"). A sum score (range, 0-69) was calculated; a higher score indicating more symptoms of depression.
The Self-Efficacy Scale (SES) was constructed to assess self-efficacy beliefs specifically related to physical activities. The properties of the questionnaire have been described earlier.22-24 The subject was asked to rate for how long he/she believed that he/she would be able to engage in the following activities: walking, running, carrying weights, standing, bicycling, sitting, and working in a forward-leaning posture. For this study, which included non-patients, the answer format was modified. The answer categories were the following: 1 ="Less than 10 minutes," 2 = "10-30 minutes," 3= "30-60 minutes," 4 = "1-2 hours," and 5 = "More than 2 hours." A sum score (theoretical range, 8-40) was calculated, a higher score indicating more positive beliefs.
The psychometric properties of MSPQ, ZUNG, and SES were studied by intercorrelations and Cronbach's alpha. The Cronbach's alpha was 0.83 for MSPQ, 0.85 for ZUNG, and 0.80 for SES. The stability of these variables from 1993 to 1995 was investigated in the groups of persistently painfree and persistent-pain subjects. Test-retest correlation (Pearson's product-moment correlation) of the sum scores ranged between r = 0.52 andr = 0.75.
Subjective work prognosis was assessed by the following question: "Considering the health of your musculoskeletal organs(neck-shoulder region, arms, lower back, legs), how do you think you will do in your present or corresponding work tasks during the next five years?"(3-point scale, with 1 = "I will not have difficulties," 2 = "I might have difficulties," and 3 = "I will have difficulties").
The work characteristics score was based on six questions about possibilities to influence matters concerning oneself, to have support from one's superior, about relationships between colleagues, use of work skills, time pressure, and difficulty in work situations.25 The answer format was a 5-point scale. A factor analysis using Varimax rotation was performed on these items. It resulted in a one-factor solution(Eigenvalue >1), where the first factor accounted for 69% of the variance. Thus a sum score (theoretical range, 6-30) was calculated, a higher score indicating more problems in these areas.
The subjective disability score was based on five questions related to disability due to symptoms of radiating neck pain, other neck pain, radiating low back pain, sudden attacks of low back pain, and other low back pain: "Estimate the total number of days your pain has interfered with your daily activities or disturbed your sleep during the preceding 12 months." The answer format was the following: 0 = "No pain during the preceding 12 months," 1 = "Not any disability," 2 = "1-7 days," 3 = "8-30 days," and 4 = "more than 30 days of disability." A sum score (theoretical range, 0-20) was calculated, a higher score indicating more frequent disability in several locations.
The "contracted pain" group was contrasted with the "painfree" group and the "persistent pain" with the "recovered" group. The distributions of psychometric variables in the five groups of outcome were first compared by univariate statistics. Medians and 25th and 75th percentiles were used in the comparisons because the distributions of most psychometric variables were skewed especially in the "painfree" group. Statistical testing was performed by Mann-Whitney test. The variable "work prognosis" was studied by cross tabulation and χ2 test.
Logistic regression analysis was used in order to assess the predictive strength of the psychometric variables when adjusting for the potential confounders. The methods of maximum likelihood, as well as least squares, were used for the estimation of statistical significance. Two strategies were used for the selection of the variables into the model. First, all of the potential confounders and predictors were included in the model, and the least significant variable was dropped out from the model step by step. The second analysis was performed by adding variables step by step into the model in the order of statistical significance in the univariate analysis.
Finally, we applied the classification presented by Main et al,14 based on MSPQ and ZUNG scores. Because neck pain was a critical item in our selection of the subjects and classification of the outcome, we used the MSPQ score without this item (MSPQ-n) and reduced the proposed discrimination limit of the MSPQ-n score from 12 to 9. Thus the cutoffs for our modified classification were as follows:
Normal: Zung <17
At risk: Zung 17-33 and MSPQ-n < 9
Distressed depressive: Zung >33
Distressed somatic: Zung 17-33 and MSPQ-n > 9
Additional classifications with systematic alteration of the score of ZUNG and MSPQ-n were formed. The predictive value of these classifications was studied by calculating sensitivity and specificity with respect to the different outcomes. In order to find an alternative classification, discriminant analysis using all our psychometric variables was performed.
P values of 0.05 or less were considered to be statistically significant in all tests. The software used were NCSS™ (NCSS Statistical Software, Kaysville, UT) and Statistica™ (StatSoft, Inc., Tulsa, OK).
As an example, Table 4 shows the distributions of the psychometric measures in the one-year follow-up classification based on"soft" criteria. In general, the values for the predictors were similar, using other classifications of the outcome. Table 5 summarizes the statistically discernible differences between the contrasted groups. The initial ZUNG depression score and subjective disability were higher in the "persistent pain" group than in the "recovered" group in all classifications, regardless of the length of the follow-up. Subjective disability was also higher in the "contracted pain" group than in the"painfree" group in all classifications except one. The subjects who contracted symptoms in the second year had a higher initial MSPQ-n score than those who remained painfree, and the subjects who recovered had more positive self-efficacy beliefs (SES) and were more satisfied with their work than those who remained symptomatic The other differences were unsystematic.
Calculations with maximum likelihood and least squares methods gave similar estimates in the modeling. In all classifications, the four potential confounders and six predictors in the logistic models explained less than 15% of the total variation and the variables in the final models less than 10%, when estimated with the least squares method. The following P values are those obtained by the maximum likelihood method.
One year follow-up, "soft" criteria. "Contracted pain" vs"Painfree." Sex and BMI remained statistically significant in the final models. High subjective disability and poor work prognosis were statistically significant predictors for contracting pain when these variables were modeled separately with the confounders. However, when both of these predictors were included simultaneously, they were of borderline significance (P = 0.06 for both variables).
"Persistent pain" vs "Recovered." Age was the confounder and high subjective disability the only statistically significant (P = 0.007) predictor for persistence of pain in the final model. MSPQ-n, together with Zung, age, and subjective disability, was a statistically significant predictor (P = 0.045), but neither MSPQ-n nor Zung was significant when added separately with age and subjective disability into the model.
One year follow-up, "strict" criteria. Poor work prognosis(P < 0.001) predicted contracting pain when adjusted by BMI, sex, and workload in this classification of outcome. High subjective disability (P = 0.022) was the only variable predicting persistence of symptoms.
Two years' follow-up, "soft" criteria. "Contracted pain" vs"Painfree." High subjective disability was only statistically significant variable predicting contracting pain in the final model (P < 0.001).
"Persistent pain" vs "Recovered." High subjective disability(P = 0.040) and many problems, as measured by the work characteristics score (P = 0.047), were significant predictors for persistence of pain when modeled separately with age (P < 0.01 in both models). When both of these variables were modeled together with age, age was the only significant variable.
Two years' follow-up, "strict" criteria. High subjective disability (P < 0.01) adjusted by age was the only statistically significant predictor for both contracting and persistent pain. High work characteristics score was of borderline significance (P= 0.058) for persistent pain.
Two years' follow-up, "trend" criteria. High subjective disability (P = 0.017) and poor work prognosis (P = 0.014), adjusted by gender, predicted contracting pain. Persistent pain was predicted by subjective disability (P = 0.001), adjusted by age.
Discriminative value of the predictors. The sensitivity-specificity combination in our material was close to 0.5 when based on the cutoff scores recommended by Main et al.14 Other combinations of cutoff scores generally did not increase this low discriminative value. The best prediction was yielded in the two years follow-up with "strict" classification criteria, a Zung score of >17 with a MSPQ-n score of >7 predicting an increased risk of persistent pain(sensitivity 0.67, specificity 0.62).
The results of the discrimination analysis were similar to those of the logistic regression analyses.
Our hypothesis that a high degree of psychological distress, depressive symptoms, and negative self-efficacy beliefs predict the development and persistence of musculoskeletal pain was not contradicted by the results of this study. The associations were, however, weak, and the statistically significant results from the univariate analyses generally disappeared when controlling for the confounders, especially age and gender. Our results show that predictions on the individual level cannot reliably be based on the assessment instruments used in this study.
The multivariate analyses showed that initial subjective disability was the strongest predictor of changes in musculoskeletal symptoms, regardless of whether the follow-up time was one or two years and regardless of the stringency of the classification criteria. Older age predicted persistency of symptoms. Subjective work prognosis and problems, as described by the work characteristics score, showed statistically discernible differences in some of the comparisons in an unsystematic way.
Somewhat surprisingly, high disability also predicted the development of symptoms in the subjects who were symptom-free or relatively symptom-free at the initial assessment. The cutoff-line for assigning a subject to the category "nonsymptomatic" at the first assessment was a duration of musculoskeletal pain of less than 8 days during the preceding year. Our results emphasize the significance of perceived disability in subjects with minor musculoskeletal symptoms. On the other hand, a tendency to report subjective disability may be influenced by other factors than musculoskeletal symptoms. Anyhow, all of our subjects were working, so any severe other health problems causing disability were improbable.
Symptoms of distress and depression, self-efficacy beliefs, and psychosocial work factors were weak predictors of pain in this study. Croft et al26 found that symptoms of psychological distress, as measured by the General Health Questionnaire (GHQ), did predict the onset of new episodes of low back pain in healthy persons. In their study, consultation for low back pain was used as the outcome criterion, while the outcome in our study was self-reported neck/shoulder or low back pain of more than 8 days' cumulative duration. The GHQ used in the study by Croft et al was originally developed for identifying symptoms of depression and anxiety in the general population, while the MSPQ was originally developed for patients. Our results indicate that assessment methods developed for and found reliable in clinical populations may not be sensitive enough as "risk indicators" in nonpatients.
In several previous studies, workload has been found to predict musculoskeletal symptoms.27 The subjects included in this study had "medium" or "light" workload; none of our subjects had heavy physical jobs. This probably explains why in our study workload did not predict future symptoms.
The outcome variable in our study was a classification based on self-reported frequency of musculoskeletal pain. Pain is an unspecific measure of outcome,6 and a report of pain is a measure of behavior, not an objective measure of pain experience. As such, it is influenced by a variety of factors regulating behavior. Other aspects of the pain problem, such as intensity or quality, were not assessed; neither were medical diagnoses included for those who reported symptoms. Certain limitations are related to studies that, as this one, are based on self-reports. Their validity and reliability in general can be difficult to evaluate, and the subject's personal style influences how he/she reports symptoms. Bongers et al3 emphasized a problem in studies in which both the dependent and the independent variables are self-reports: a relationship between these can arise from, eg, general dissatisfaction or readiness to report complaints.
Magni et al8 defined chronic pain as a self-report of pain present for most of the days in at least one month of the year preceding the assessment. According to other criteria, pain is chronic when it has lasted for more than seven weeks.28 The subjects in our"recovered" and "persistent pain" groups were initially chronic according to the criteria of Magni et al,8 but they were employed and working even if they reported symptoms. Our classification criteria (less than 8 days/8-30 days/over 30 days) were possibly not sensitive enough, or discriminating enough, in order to distinguish persons with progressively and increasingly more disabling symptoms from persons with recurrent episodes of less disabling pain. Our attempts to classify the subjects based on all three assessments (eg, painfree in 1993 and pain > 30 days both in 1994 and in 1995 as compared with those who were painfree at all three occasions, or pain> 30 days in 1993 and painfree both in 1994 and 1995 as compared with those who reported pain at all three occasions) resulted in very small groups in which proper statistical analyses were not possible.
Mäkelä et al29 found a strong co-morbidity of musculoskeletal symptoms in various sites. We decided to include persons reporting symptoms in different sites (neck and low back) and of different types (radiating and non-radiating) in our outcome categories. An interesting alternative would have been to analyze patients with radiating and non-radiating pain separately, as it has been claimed that determinants of sciatica may be different from determinants of "common" low back pain.30 Our subject material, however, was too small for this kind of type-specific analysis.
In this study, as in most similar studies, the individual predictors explained less than 10% of the total variance. Thus even if the specific variables seem to have some relevance, their contribution to the outcome is marginal. One explanation for this may be related to the traditional research paradigm, which relies on linear relationships. The "independent variables" are presumed to directly or indirectly-modified or mediated by other variables-influence the outcome variable. This paradigm does not take into account the fact that human beings are active, dynamic, reflective, and continuously interactive. The development and persistence of symptoms, as well as recovery from symptoms, is a dynamic process regulated by numerous individual and continuously changing factors. The individual is during his/her whole lifetime in a continuous reciprocal relation with other individuals and socioeconomic systems. These reciprocal relationships interact with, influence, and are influenced by the individual's subjective beliefs, interpretations, and attributions, his/her affective reactions, ways of coping, and other forms of behavior. In view of this explanatory model, it is evident that focusing on one or a few possibly relevant factors in cross-sectional, or repeated cross-sectional, study designs is not very informative. Scientific research based on this dynamic process model requires other methods of analysis than the traditional ones.
This study was financially supported by the Finnish Work Environment Fund.
1. Frymoyer JW, Cats-Baril W. Predictors of low back pain disability. Clin Orthop. 1987:89-98.
2. Mendelson G. Psychological and social factors predicting responses to pain treatment. In: Bond MR, Charlton JE, Woolf CJ, eds.Proceedings of the VIth World Congress on Pain. Amsterdam: Elsevier Science Publishers; 1991:193-206.
3. Bongers PM. De Winter C, Kompier MAJ, Hildebrandt VH. Psychosocial factors at work and musculoskeletal disease. Scand J Work Environ Health. 1993;19:297-312.
4. Gamsa A. The role of psychological factors in chronic pain. I. A half century of study. Pain. 1994;57:5-15.
5. Gamsa A. The role of psychological factors in chronic pain. II. A critical appraisal. Pain. 1994;57:17-29.
6. Riihimäki H. Hands up or back to work-future challenges in epidemiologic research of musculoskeletal diseases.Scand J Work Environ Health. 1995;21:401-403.
7. Leino P, Magni G. Depressive and distress symptoms as predictors of low back pain, neck-shoulder pain, and other musculoskeletal morbidity: a 10-year follow-up of metal industry employees. Pain. 1993;53:89-94.
8. Magni G, Moreschi C, Rigatti-Luchini S, Merskey H. Prospective study on the relationship between depressive symptoms and chronic musculoskeletal pain. Pain. 1994;56:289-297.
9. Leino PI, Hänninen V. Psychosocial factors at work in relation to back and limb disorders. Scand J Work Environ Health. 1995;21:134-142.
10. Pietri-Taleb F, Riihimäki H, Viikari-Juntura E, Lindström K. Longitudinal study on the role of personality characteristics and psychological distress in neck trouble among working men.Pain. 1994;58:261-267.
11. Main CJ. The modified somatic perception questionnaire(MSPQ). J Psychosom Res. 1983;27:503-514.
12. Zung WWK. A self-rating depression scale. Arch Gen Psychiatry. 1965;12:63-70.
13. Main CJ, Waddell G. The detection of psychological abnormality in chronic low back pain using four simple scales. Curr Concepts Pain. 1984;2:10-15.
14. Main CJ, Wood PLR, Hollis S, Spanswick CC, Waddell G. The distress and risk assessment method: a simple patient classification to identify distress and evaluate the risk of poor outcome. Spine. 1992;17:42-52.
15. Mannion AF, Dolan P, Adams MA. Psychological questionnaires: do "abnormal" scores precede or follow first-time low back pain? Spine. 1996;21:2603-2611.
16. Holden G. The relationship of self-efficacy appraisals to subsequent health related outcomes: a meta-analysis. Soc Work Health Care. 1991;16:53-93.
17. French DJ, Gauthier JG, Doyon J, Savard J. La perception de l'efficacite personelle et la doleur. [The perception of self-efficacy and pain.] Comportement Human. 1992;6:147-158.
18. Estlander A-M, Mellin G, Weckström A. Influence of repeated measurements on isokinetic lifting strength. Clin Biomechanics. 1992;7:149-152.
19. Kuorinka I, Jonsson B, Kilbom Å, et al. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Appl Ergonomics. 1987;18:233-237.
20. Viikari-Juntura E, Riihimäki H, Takala E-P, et al.Factors Predicting Pain in the Neck, Shoulders, and Upper Limbs in Forestry Work. [In Finnish.] Helsinki: Finnish Institute of Occupational Health; 1993.
21. Deyo RA, Walsh NE, Schoenfeld LS, Ramamuthy S. Studies of the modified somatic perceptions questionnaire (MSPQ) in patients with back pain. Spine. 1989;14:507-510.
22. Estlander A-M, Vanharanta H, Moneta GB, Kaivanto K. Anthropometric variables, self-efficacy beliefs, and pain and disability ratings on the isokinetic performance of low back pain patients.Spine. 1994;19:941-947.
23. Kaivanto KK, Estlander A-M, Moneta GB, Vanharanta H. Isokinetic performance in low back pain patients: The predictive power of the self-efficacy scale. J Occup Rehabil. 1995;5:87-99.
24. Gibson L, Strong J. The reliability and validity of a measure of perceived functional capacity for work in chronic back pain.J Occup Rehabil. 1996;6:159-175.
25. Elo A-L, Leppänen A, Lindström K, Ropponen T.Occupational Stress Questionnaire: User's Instructions. Helsinki: Finnish Institute of Occupational Health; 1992.
26. Croft PR, Papageorgiou AC, Ferry S, Thomas E, Jayson MIV, Silman AJ. Psychologic distress and low back pain: evidence from a prospective study in the general population. Spine. 1996;20:2731-2737.
27. Scientific Committee for Musculoskeletal Disorders of the International Commission on Occupational Health (ICOH). Musculoskeletal disorders: work-related risk factors and prevention. Int J Occup Environ Health. 1996;2:239-246.
28. Spitzer WO, LeBlanc FE, Dupuis M. Scientific approach to the assessment and management of activity-related spinal disorders.Spine. 1987;12:9-59.
29. Mäkelä M, Heliövaara M, Sievers K, Knekt P, Maatela J, Aromaa A. Musculoskeletal disorders as determinants of disability in Finns aged 30 years or more. J Clin Epidemiol. 1993;46:549-559.
30. Heliövaara M, Mäkelä M, Knekt P, Impivaara O, Aromaa A. Determinants of sciatica and low-back pain.Spine. 1991;16:608-614.
From Dust to Dust?
The volcanic ash that devastated Pompeii in A.D. 79 also saved it from time's ravages. Now modern man seems to be finishing what Mount Vesuvius began: ruining the ancient city for good. Although progress has been made in cleaning up sloppily excavated sites, Pompeii's ruins remain poorly protected. Guards have been known to sleep on duty while visitors defaced walls with graffiti. [Recently] vandals lopped off the heads of "the fugitives," a family group found in 1921. Little of the gate receipts from the site's two million annual visitors goes to preservation efforts. Now most scholars want to see excavation stopped to prevent further deterioration… A proposed solution would distribute the city in blocks to different universities, to complete and record earlier work. Many scholars favor reburying sections of Pompeii. After all, for almost two millennia dirt did a better job of protecting the site than do modern-day caretakers.
From Periscope. Newsweek, September 1, 1997, p 8.
© Williams & Wilkins 1998. All Rights Reserved.