Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6 The Epidemiology of Chronic Pain and Work Disability This chapter examines the frequency and distri- bution of chronic pain in the population and its relation to work disability and medical care utilization. The epidemi- ological literature on chronic pain, like that in some other disciplines, suffers from a lack of consensus about basic definitions and from inconsistencies in measurement which make it difficult to compare studies and to generate precise numbers. Keeping in mind that among the claimants who are most troublesome for the disability insurers are those who have pain and associated impairment that cannot fully be accounted for by clinical findings, there are several pieces of epidemi- ological data that would be useful. What are the numbers and characteristics of people with: ~ chronic pain (and what kinds of pain do they have)? · chronic pain not correlated with objective clinical findings? · severe chronic pain with important dysfunction and impairment? and · of those with any of the three abovementioned conditions, how many apply for Social Security Administration (SSA) disability bene- fits, how many get them, and what are their characteristics? Such data would allow estimates of the size of the population at risk, and hence provide a basis for estimating the costs associated with any proposed changes in the way pain is handled by the SSA. Although there are no complete answers to the abovementioned questions, studies carried out after 1970 in working age populations in 101
102 THE EXTENT AND COST OF THE PROBLEM the United States and other countries provide some useful informa- tion. Studies of pain occurrence in populations, in groups of patients, and in groups of workers form the basis for the estimates in this chapter. Seventeen pertinent studies from the United States, Canada, and Western Europe were located, 15 of which examined low back pain alone, the most common chronic pain. These studies are summarized in Table 6-1. METHODOLOGICAL ISSUES Variations in operational definitions of chronic pain, measures of its severity, sampling frameworks, data collection techniques, and survey response rates make comparisons between epidemiological surveys difficult. Definitions and Measures of Chronic Pain The many inconsistent definitions of pain, chronic pain, and severity of pain used in epidemiological surveys reflect the uncertainty about pain and its nature discussed throughout this volume. Pain was defined by the patient's perception or recognition in all of the studies reviewed in this chapter. People reported verbally whether or not they "had" pain. This is consistent with the notion that pain is a subjective phenomenon, most accurately measured by the verbal communications or nonverbal behaviors of the individual in pain. Most of the studies reviewed here attempted to obtain measures of pain occurring during the whole previous year, and some even reported the prevalence of pain during the entire lifetime. The majority of these papers neglect to mention that such recalled historical information may greatly distort estimates of pain occurrence. One obstacle to measurement that is basic to the nature of pain is the completeness with which pain is often forgotten once it is over. Whereas current pain may be measurable with some validity, the threshold of recall of previous pain probably varies with the recentness of the episode, the severity of the symptom when it was present, associated events such as surgery, and the persistence of the probing questions in the interview- ing process (Bierring-Sorensen, 1982~. There is no agreed-upon operational definition of chronic pain in the studies reviewed. Terms such as "frequent" (Reisbord and Greenland, 1985) or "serious" (Enquete Sante Canada, 1981) pain are poor proxies for chronicity. Such definitions are likely to have Tow reliability because individuals will vary in their concepts of these terms. In the
THE EPIDEMIOLOGY OF CHRONIC PMN 103 Nuprin Pain Report (Louis Harris, 1985) respondents were asked to report roughly how many days during the previous 12 months they had different kinds of pain. (Data are given for 1-5 days, ~10 days, 11-30 days, 31-100 days, and >100 days.) From these data chronic pain could be arbitrarily defined as pain on more than 30 days or more than 100 days. The frequency of work disability due to pain may be estimated from the Nuprin study from the reported number of days during the previous 12 months respondents had such severe pains that they "could not work or engage in routine activities." Different aspects of pain and disability were measured in the various surveys. The two national health surveys (Drury, 1984; Enquete Sante Canada, 1981) asked respondents whether they had a number of specific conditions usually associated with pain, among them impair- ment of the back or spine. The Nuprin Pain Report, on the other hand, although not as rigorous as the National Health Surveys in its survey methods, contains a considerable amount of information on the fre- quency of pains of different kinds (Louis Harris, 19851. Study Samples Data in the studies reviewed here are drawn from samples of national populations, communities, patients, and workers. Results from these four types of samples are not strictly comparable because the category "workers" does not include some people who already are disabled and out of the labor force, the category "patients" exclude those who do not go to doctors, and people in particular geographic communities may have relevant attributes not found in other places. Methods of Data Collection All 17 surveys were cross-sectional in nature. Data were collected by several means. The American and Canadian national surveys were done by interviewing respondents in their homes (Drury, 1984; Enquete Sante Canada, 19811. Information was obtained by telephone interviews in the Nuprin study (Louis Harris, 19851. These kinds of surveys are likely to lead to underreporting (Kelsey et al., 19791. Among the regional studies, pain data were obtained from surveys conducted for other purposes in Columbus and Dayton, Ohio (Nagi et al., 1973; Reisbord and Greenland, 1985) and from interviews and actual physical examinations in Glostrup, Denmark (Bierring- Sorensen, 1982), Goteborg, Sweden (Svensson and Andersson, 1982; Svensson, 1982), and Zoetermeer, the Netherlands (Valkenburg and
104 ._ Ct C) m ~0 Ct ._ Ct o a 1 Cal I= C) _ C) Cat o CO Ct a; . o . - . Cat CS 00 g O C, ~ _ q) O ~ V C£5 (Q ED ~ Cal O ~ ~ ~ ~ ~ ° ~ ~ ° ~ ~ I O _ .= ~ ~ ~ ~ A 5, Al ~ Al ~ ~ ~ ·o ~ ~ E ~ .= ~ ~ ~ ~ ~ ~ ~ E i,, ° ~ ~ ~= = ~ ~ E E ~ 3 mm ~ ~ ~ Pa ; ~ o g ·= .~ ,,, 5 2 . ~ ~ ~ ,., 3 3 E ~ ~ E ~ Q ~ ~ N ~ ~ ~ ~ ~ _ E ~ ;: E - E :~ -° <_ t_ 00 c, ~ E c D
105 U: ~ ~ ~ d4 CD ~ Cal CC o ~ Do ~ ~ ~ 1 _ ~ ~ ~ 3 ~ ~ ~ ~ o ~ ~ ~ ~ :- C) ~ ~ ~ Ct ~ ~ go,= ~ ~ o ~ ~ ~ = 0 Ce do ~ ~ ~ ~ ~ ~ ~ ~ ~ _ o ~ ~ a& ~ - ~ Y ~ ~ .o E E = · I ~ e e c° 3 ~ ~ t~ ~ 0 ct sol ~ ~ ~ ~ ~ ~ _ ~ ~ PA ~ ~ ~ .= i= ._ ~ . ~ ~ 5- c) ce ~ ~ ~ 3 o Ct ~ o TV ~ 5: ~ 5- o ~ ~ Sly, cd ~ E , ,, ~ ~ ~ c .~ ~ rev o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ . . ~ '3 Ct ° ° A ~ ~) I bC ~ ~5 .O Z ~ oo tD a~ ~ ^ ~ ' - =: _ ~ ~ o~~ c~ A °~ ¢ o ,s ° ° ^ z £ :s :5 £ 0 £ = ° (~) £ ~ u ~ = 0
106 o CD :e =~ ~ _ 5- ~ ;~;, ~- sit o Ct .= As Pi o o . - ._ as es o o o ~ ~ - o ~ v G) so C: U2 O ~ ~ en ~ O ~ _ CS: Cal ~ o0 Cal Cal CO ~ Cal ~P . . . . . . . . . . . . . . . . . . . Cal ~4 CC Con ~ ~ He Cal CO ~ Cal ~1 ~ Cal O d' ~ ~ N (D ~ _ ~ Car CO _ ~ ~ ~ Cal ~ _. ~ _ ~ =0 .= ~ ~ O .= ~ ~ ~ ~ ~ ~ ~ ~ O - c ~ ~ ' ' ~ ~ ' ~ j ~ ~ 1 ~ . i ~ ~ E 3 3 ~ C) _ ~ ._ i~ ' I ~ . ~ S ~ ~5 ~ ~ E ~ ~ x ~ 3 c ~ ~ 2 ~ 0 ~ ~ ~ ~ ~ 3 _ ~ ~ ~ ~ , C~ o ~ ° ° ~ ct ~ ~ A~
107 00 00 ~ ~ ~ CD Cal . . . . . . . Cal ~ CD O ~ Cal CD ~ CD cat ~ cat ' ~ ~ ~ o O ~ e e -E ~ O ~ ~ ~ e ;> ~ ~ :> ~ a; ~ ~ ~ ° 0 ~ 0 ,,0 0 0 ~ ·,, ·,, 0 3 ~ O ~ V ~ to ~ ~ ~ ._ ~ ~ ~ ~ ~ ~ ._ b,0 Ct he ~ ~ to C;S · - · C-) a, ~ ~ :t Ct I) Ct ~ Ct ~ ~ O C) ID & SO cq C) ~ O) 9 ;3 ~ it ~~ L. ~ ~ ~ 9 =: ~ S S 3 c O ~ ~, £ .~: ~ ~ c' 3 9 9 ~ '~ ~ ~ ~ ~ ~= O o o X ~ ~ ." ~ ~ ,o ~ ~ ~ E~ O E5Q ~ e V -. e -. e -= , s j _ °O v°?O°~ i! ~_ P: ~ ~ ~ C - ° _ -, ^, O . a, (,5 S o _ - E C `:a 5 ~ °0 ~ ~ ~ ~ ~ ~ ~ ^ - ~ ~ ~ ~ ~ A E == _ j .= 0 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ O ~ ~ ~ , ~ .o E~ ·°- V tX) ~ ¢ ~ V - V _ · V V £ ~ ° ~ = Z ~ 3 C: ~ cs ~Z ~ ~ Z ~ ~ ~ ~ _ ~ . ~ 3 ¢ ~ G) ~ -~= ~ ~ 0 ~ ~ ~ ~ ~ ~ =-= ~ ~ O ~ ~ O
108 5) _ o U. Ct ._ o o ._ . Cal Ct ho ~ = ._ o C) C) ~ _ ~ o ~ V - CS - ~: c) Ct . o _ o R$ ~ ~ ~ ~ ~ ~ S o ~ o ~ - ° Ad'= i., : ~ He o ~ =5 ,~ Em o c; ~ ~ 3 v c, ~ v c, ~5 (D ~ _ .~ ~ ~ _ ~ g ~ ,~ ~ I, ~ ~ ~~=
THE EPIDEMIOLOGY OF CHRONIC PIN 109 Haanen, 19821. The Glostrup study is unique in that all of the subjects were followed up by questionnaires a year later. The Goteborg study is also noteworthy, because it compared respondents with nonres- pondents using the register of the Public Health Insurance Office. More nonrespondents (59 percent) than respondents (49 percent) were found to have been off work during the preceding 3 years. A sizeable proportion of respondents, moreover, had been "sicklisted" for Tow back pain when they said that they had not been (27 percent). The authors concluded that Tow back pain incidence and prevalence rates are underestimated when they are based on interview findings. PAIN AND ITS CONSEQUENCES Some information can be gleaned from the epidemiological studies about the kinds of pain people experience, the incidence and preva- lence of pain and its associated outcomes, and the frequency of health services utilization for pain complaints. What Kinds of Chronic Pain Do People Have? Info~-~ation about what kinds of pain people have is found primarily in the Nuprin Pain Report and in one Canadian study (Crook et al., 19841. Most of the other studies are restricted to back pain. The common types of pain are headache, backache, muscle pain, joint pain, stomach pain, premenstrual or menstrual pain, and dental pain. The Nuprin study classified only 2 percent of people as having other types of pains. Chronic pain was arbitrarily defined as having pain for more than 30 days during the year. Table 6-1 shows that back pain, joint pain, and headache each occurred in about 1 in 7 people (1~14 percent), whereas chronic muscle pain occurred in 1 in 10 people, and other types of chronic pain occurred in less than 1 in 20 people (2~ percent). Findings from the Burlington, Ontario, patient survey are in agree- ment (Crook et al., 19841. Chronic pain was defined as "persistent" pain and was categorized by body area. In terms of occurrence, back pain ranked first, lower extremity pain second, and head or face pain third. Eighty-three percent of people reporting persistent pain said they had had the pain longer than a year. One of the critical questions about back pain is what proportion of people with it go on to chronicity and work disability. The Quebec Task Force (Spitzer and Task Force, 1986) reviewed the records of 3,000 workers in Quebec (a random sample of the entire 1981 cohort of 58,000 cases of "disorders of the vertebral column") who reported an
110 THE EXTENT AND COST OF THE PROBLEM incident of occupational back problems. This was done in order to see what happened to these people over the next 4 years. Within 1 month, 74 percent of the sample was back at work and used no further medical care for their back problems. Between the 2nd and 5th months after the onset of back pain, an additional 19 percent returned to work. Most important, only 7.4 percent of the patients were disabled for more than 6 months; but this small group of chronic pain patients accounted for 70 percent of lost work days in 1981, for 73 percent of medical care costs, and for 76 percent of compensation payments made to beneficia- ries who had problems associated with the vertebral column. Incidence and Prevalence of Relevant Outcomes* An estimated 1~15 percent of adults have some work disability due to back pain in any given year. The leading causes of disability in people in their working years are musculoskeletal conditions such as low back pain, joint pain, arthritis, and rheumatism (Kelsey et al., 1979; U.S. Department of Health and Human Services, 19801. Swedish National Health Insurance data show that between 9 and 19.5 percent of all sickness-absence days from work are associated with back complaints (Svensson and Andersson, 1982; Svensson, 19821. This percentage is unquestionably higher than for any other class of health problems and is unlikely to be explained by methodological inconsis- tencies or shortcomings. Almost 17 percent of workers' compensation claims in the United States and in Quebec are due to back problems (Klein et al., 1984; Spitzer and Task Force, 1986~. The authors of the Nuprin Pain Report projected the number of days lost from work or usual occupation to the total U.S. adult population (including those more than 65 years of age, which inflates the figures somewhat). Back pain ranked first at 1.3 billion person-days lost, joint pain was second at 1.0 billion person-days lost, and headache was third at 0.6 billion person-days lost. The incidence of new back pain was measured in three surveys. A relatively low annual incidence rate (0.6 percent) is reported in the National Health Interview Study (Drury, 19841. The present incidence of new back pain was found to be 2.1 percent in the G6teborg survey (Svensson and Andersson, 1982; Svensson, 19821. The annual inci- *Incidence refers to the number of new cases of a disease or condition occurring in the population during a specified period of time. Prevalence refers to the total number of cases of the condition present in the population at a particular time.
THE EPIDEMIOLOGY OF CHRONIC PIN 111 dence of new back pain was reported to be 6.3 percent in the GIostrup survey (Bierring-Sorensen, 19821. The validity of incidence rates derived from cross-sectional surveys is questionable. We have no information about work disability caused by new back pain. Estimates of annual prevalence rates for work disability caused by back pain were reported in several surveys (Svensson and Andersson, 1982; Svensson, 1982; Drury, 1984; Louis Harris, 1985; Nagi et al., 19731. Work disability attests to a certain degree of pain severity. Furthermore, because absence from work is an objectively verifiable behavior rather than a subjectively remembered sensation, it is a reasonably reliable proxy for other outcomes of interest and comple- ments them. Unfortunately, most of the surveys use "one or more bed days" or self-reports of the number of days per year respondents were unable to work or engage in routine activities as proxies for calculat- ing the prevalence of work disability. Neither measure adequately describes a chronic problem and both are subject to problems of recall. The National Health Interview Survey of the adult U.S. population reported a 1.3 percent annual prevalence of back impairment with one or more days in bed (Drury, 1984~. This value is considerably lower than the 14.2 percent of Americans "unable to work or engage in routine activities one or more days per year" given in the Nuprin report (Louis Harris, 19851. The Columbus, Ohio, survey (Nagi et al., 1973) reported that 2.6 percent of people had severe work limitations linked to back pain and a further 4.2 percent reported moderate work limitations. Both the National Health and the Columbus surveys, however, were conducted for many purposes. Their estimates could be low because of respondents' recall failure when attention was not focused on back pain. Data from the 1978 Goteborg survey, on the other hand, were collected expressly for the purpose of investigating back pain; they yield high estimates of back pain prevalence. "Moderate" interference with work associated with back pain occurred in 13.9 percent of the sample and"severe" interference or "complete disablement" occurred in another 7.8 percent (Svensson and Andersson, 1982; Svensson, 19821. Back Pain and Medical Care As discussed elsewhere in this report, some people have considerable pain in the absence of clinical findings accounting for it, whereas others have clear anatomical abnormalities without pain. This discrep- ancy between objective abnormalities and existing complaints is shown
112 THE EXTENT AND COST OF THE PROBLEM clearly in two studies. ~ one study, disc herniation and vertebral fracture accounted for only 3 percent of sickness absence episodes for back pain (Svensson and Andersson, 1982; Svensson, 19821. A considerable number of people in another survey had radiological abnormalities of the spine but no complaints of pain (Valkenburg and Hansen, 19821. Disc degen- eration and rigid lumbar segments normally occur with aging, but the reported relation of back pain with age is inconsistent. Mechanic and Angel (in press) report that although clinical findings increase with age, the relation between age and self-reported pain, controlling for clinical findings, decreases with age. This suggests that pain may be elected, more acceptable, and hence less noteworthy in the context of aging. (See Chapter 8 for a discussion of psychosocial influences on the perception of symptoms and illness behavior.) The magnitude of the back pain problem compared with other illnesses is highlighted by the National Medical Surveys (Cypress, 1983; National Center for Health Statistics, 1983, 19841. Pain and back pain are among the leading symptomatic reasons for visits to physicians. In 1977-1978 and 1980, 2.S percent and 1.8 percent of visits, respectively, were for back symptoms. In 1980-1981,6.1 percent of physician visits- 70 million were attributed to new pain, of which 10 million were for new back pain. Low back pain was the chief reason for visits to the Oxford Regional Pain Relief Unit in 1982 (Frymoyer et al., 1983~. Although 30 percent of people with more than occasional backache (>5 days per year) did not consult a doctor for their pain, 41 percent saw one or two doctors, and 29 percent saw three or more doctors (Louis Harris, 1985~. Descriptive Epidemiology of Chronic Pain and Disability Trends in Time There appears to be an increase on the relative frequency of back pain-caused work disability over time. In England, a 22 percent increase in the number of episodes of back trouble and a 30 percent increase in their duration was noted between 1961 and 1967 (Wood, 19701. In Quebec, physiotherapy treatments have increased dramati- cally; about 40 percent of these treatments are for conditions affecting the spinal column (Spitzer and Task Force, 19861. In the United States, disability as reported in the National Health Interview Surveys increased substantially from 1966 to 1976, and musculoskeletal disor- ders were responsible for a considerable portion of this increase (Colvez and Blanchet, 19811.
THE EPIDEMIOLOGY OF CHRONIC PIN 113 Demographic Factors, Pain, and Work Disability The incidence of new low back trouble was highest ire relatively young adults in the GIostrup survey, the sole source of age- and sex- specific incidence rates in a general population. Eleven percent of 30- year-olds developed new back trouble in the course of a year. In compar- ison, only 3 6 percent of men and women 40, 50, and 60 years old developed new back pain. This difference remained when incidence rates were calculated excluding those who had had back pain earlier in life. In the two studies in Ohio, age-specific prevalence rates of back pain were 3~70 percent higher in people aged 3~64 years than in those aged 18-34 years (Nagi et al., 1973; Reisbord and Greenland, 19851. Back pain and chronic joint pain for more than 30 days per year was highest in the 50- to 64-year-old respondents in the Nuprin Pain Survey. This increasing prevalence of back pain with age was found only in women, however, in the Glostrup and Zoetermeer surveys. Combining all age groups, women appear to have a slightly higher prevalence of back pain than men (Louis Harris, 1985; Nagi et al., 1973; Reisbord and Greenland, 19851. Age- and sex-specific rates of disablement for work associated with back pain are also given by the survey of workmen's compensation claims in Quebec (Spitzer and Task Force, 19861. Although not representative of the general population, demographic data on workers is pertinent to the question of who claims Social Security disability benefits. Claiming compensation for at least 1 day's absence from work for back pain during 1981 in all age groups in Quebec were 0 5 percent of women workers and 1.9 percent of men. Although three times more claims were made by men than by women overall, in the 45- to 64-year-old age group, women made almost as many claims as men. The majority of claims were made by men and women younger than 45 years of age (82 percent). A sizeable proportion (28 percent) were submitted by people under 25 years old. In summary, most people have back pain at some time in their lives. Young adults have the highest incidence of new back pain. Older women, on the other hand, have the highest prevalence of chronic back pain and of chronic joint pain. Few data are available on the frequency of work disability related to pain, but it appears from the Quebec data that compensation for at least 1 day's absence from work for back pain is claimed by men more than by women, and mostly by people who are quite young. Data on the rate of occurrence of back pain by race are given In three U.S. surveys (Louis Harris, 1985; Nagi et al., 1973; Reisbord and
114 THE EXTENT AND COST OF THE PROBLEM Greenland, 19851. Striking differences are not indicated by any of the data, although blacks in the Nuprin study had less chronic back pain than whites and somewhat less chronic joint pain. No information is available on pain-caused work disability by race. Marital status was associated with recurring back pain in two American studies (Nagi et al., 1973; Reisbord and Greenland, 1985) but was not related to the occurrence of back pain in the Swedish study (Svensson and Andersson, 1982; Svensson, 19821. In both the Colum- bus and Dayton surveys 18 percent of married people had frequent back pain, whereas higher rates were found in separated, widowed, and divorced people (2~30 percent in Columbus, 2~37 percent in Dayton) and lower rates occurred in never-married people (9 percent in Columbus, 11 percent in Dayton). Marital status was not reported in the Nuprin study. Physical Factors Certain individuals may be predisposed to chronic back pain because of the shape of their bodies. We have little information on such possibly predisposing factors as constitution, height, degree of obesity, physical fitness, and flexibility or suppleness. Tallness is associated with back pain in some studies (Andersson, 19811. The association of physical effort with back pain also is unclear, although many studies report the frequency of back pain in various work places (Spitzer and Task Force, 1986; Anderson, 1976; Anders- son, 19811. Associations have been reported between back pain and six types of physical work, all of which affect the load on the spine (Andersson, 1981~. These six factors are physically heavy work in general, static work postures, frequent bending and twisting, lifting and forceful movements, repetitive work, and vibrations. Because several of these factors often are present at the same time, multi~rariate analyses are required to know which associations are important. Such methods were used to analyze the work characteris- tics of the 40- to 47-year-old men in the Goteborg survey (53vensson and Andersson, 19831. Although the reported physical demand, posture, and amount of lifting at work were all associated with a history of back pain in univariate analyses, only lifting was significantly associated with back pain in the multivariate analysis. Yelin and his colleagues (1980) also used multivariate methods in their study of work disability in 180 people after diagnosis of rheuma- toid arthritis. No one physical characteristic of a job correlated significantly with whether these people remained employed. Physical
THE EPIDEMIOLOGY OF CHRONIC PAN 1 15 effort of work was not an important predictor of work disability in people with rheumatoid arthritis. Socioeconomic Status Socioeconomic factors were associated with back pain in the surveys in which they were measured. Although not entirely consistent, the data indicate that back pain may be more likely to become a chronic problem in people of lower socioeconomic status. People with less than a high school education had more chronic back pain in three American surveys (Louis Harris, 198~; Nagi et al., 1973; Reisbord and Green- land, 19851. In the Nuprin sure ey (Louis Harris, 1985) 23 percent of respondents with less than a high school education had more than 30 days of back pain in the year compared with 11-13 percent of high school and college graduates. Similarly, in both Columbus (Nags et al., 1973) and Dayton (Reisbord and Greenland, 1985), about twice as many respondents with less than a high school education had frequent back pain as compared with those with a high school education or more (2~30 percent vs. 12-16 percent and 23-25 percent vs. 12-18 percent, respectively). Back pain was not associated with educational level in either the Goteborg survey of 40- to 47-year-old men (Svensson and Andersson, 1982; Svensson, 1982) or in the Zoetermeer survey of adult men and women (Valkenburg and Haanen, 19821. When accompanied by objec- tive clinical signs such as radiating leg pain and muscle spasm, back pain did tend to be associated with education level in Zoetermeer. (In this survey one-quarter of the men and women having back pain had such signs.) Men and women with less education (advanced primary education or less) tended to have disc prolapse more frequently than those with more education (~3 percent vs. ~2 percent); they also tended to have clinically evident lumbago more frequently (~7 per- cent vs. 0~ percent). The disparity between American and European studies could be due to the different definitions of back pain. Only frequent or chronic back pain appears to be associated with lack of education. Low occupational status was also related to back pain. In most surveys (Nagi et al., 1973; Reisbord and Greenland, 1985; Svensson and Andersson, 1982; Svensson, 1982) laborers and blue-collar work- ers were about one-and-one-half to two times more likely to have back pain than white-collar workers. It should be noted that socioeconomic factors are not independent in their influence. Those with lower levels of education are more likely to have jobs of Tower status, and those jobs
116 THE EXTENT AND COST OF THE PROBLEM are more likely to require physical activity that may be more condu- cive to back pain or injury. Psychosocial Factors As discussed elsewhere in this report (see especially Chapters 8 and 9), a host of psychosocial factors influence the course and outcomes of chronic pain. Some epidemiological work includes exploration of these associations. In the Columbus survey (Nagi et al., 1973), people with frequent back pain more often reported that they had trouble getting to sleep, were bothered with nervousness, felt restless and tense, had trouble getting up in the morning, and had trouble with sweating hands. From this cross-sectional study, however, it cannot be deter- mined whether these factors were present before or began after the onset of back pain. There is some fairly good evidence that the psychosocial nature of the work environment is linked both to back pain and to work disability. In the Goteborg survey, several work environment factors were found to be associated with a history of Tow back pain in univariate analyses (Svensson and Andersson, 19831. The factors reported were monotonous or boring work, diminished work satisfac- tion, decreased potential to influence the work situation, and less demand on concentration. Monotonous and/or boring work, as graded by the workers on a four-level scale, remained significantly associated with a history of back pain in a multivariate analysis. In the study by Yelin and his colleagues (1980) of work disability after the diagnosis of rheumatoid arthritis, the social characteristics of the workplace predicted cessation of employment twice as well as personal or medical factors. Two important workplace factors were control over the pace of work and self-employment. Two rival hypoth- eses are suggested to explain these findings: "Because of. . . fluctua- tions in symptoms, the ability to fit the work schedule around the illness through control over the pace and time of work and activities of the job is crucial to the continued employment of persons with rheumatoid arthritis." Alternatively, "it is interesting to speculate as to whether those who control the pace of work, or have flexible schedules of work, are more satisfied with their jobs and are, therefore, more willing to push on to continue working, whereas those who lack control on the job are more willing to take the disability route." These findings suggest that in addition to individually targeted medical interventions, more attention should be paid to the social characteris- tics of the workplace.
THE EPIDEMIOLOGY OF CHRONIC PAIN 117 RESEARCH AGENDA Considering the number of people affected with chronic pain and the magnitude of its personal and social consequences, surprisingly little valid information is available. Some possible reasons for the paucity of longitudinal studies, interdisciplinary research, and research specifi- cally focused on the nexus between chronic pain and disability include institutional and disciplinary constraints as well as lack of funding. Much work remains to be done before rational policies based on scientific studies can be proposed to ameliorate the problem of chronic pain and related work disability. Epidemiological Studies Most of the descriptive epidemiological work and the surveys re- ported to date are cross-sectional studies. As a result, virtually nothing can be said about cause and effect. The single most important recom- mendation made in the context of a research agenda is to begin now with at least one prospective, longitudinal study of a nationally representative sample. This might be a single uncontrolled cohort (like the Framingham study of cardiovascular diseases), or a comparison study with two or more carefully delineated cohorts. These could be: 1. An unselected geographically defined population cohort. Such a cohort would include individuals free of disease or injury or symptoms related to pain and disability. The geographically defined cohort could permit inclusion of individuals with a history of or current minor injury, disability, or disease because strict exclusion of people from the population simply on grounds of a history of pain or disability might create a highly selected, unrepresentative cohort for purposes of generalization. 2. A clinically defined cohort. Such a cohort might include a sample of people with particular diseases, new injuries, or episodes of disabil- ity. These persons would then be followed to determine the natural history of chronic patterns of symptoms, disease, or disability. A combination of geographical and clinical criteria might be used (i.e., all new cases of neck injuries in three states). 3. Special approaches to cohort delineation. The cohort could be defined by occupation groups known to be at high risk for chronic disabling pain, given the interest in work-related disability and the need for follow-ups of considerable length. Cohorts might also be defined by age group or compensation status.
118 THE EXTENT AND COST OF THE PROBLEM Concurrent, prospective epidemiological studies would help to iden- tify determinants of chronicity and impairment before such un- favorable outcomes are manifested. In a clinically defined inception cohort, for instance, we could determine which individuals with low back pain eventually develop chronic pain and chronic disability. Today we know only that fewer than ~ percent of such individuals are likely to develop chronic pain. Faced with a new case, the clinician or investigator currently has no way of predicting whether that person is likely to go on to chronic status or not. Cohort studies can help answer some of these questions and generate focused hypotheses for other types of studies. Most important, they could point the way to designing preventive measures and directing therapeutic efforts toward those people most likely to need and benefit from such inter- ventions. Because longitudinal follow-up studies require considerable time and money, complementary methods should be pursued as well. Case-referent (or case-control) studies, in which "cases" are defined by manifestation of the outcome of interest and "referent" subjects are defined by its absence, permit exploration of the role of suspected risk factors for chronic pain. If two or more well-conducted case-referent studies point in the same direction and are sustained by evolving data from cohort work, the concordance might permit better judgment of the effectiveness of preventive measures and the importance of risk factors and risk markers among people who first experience a problem potentially leading to a chronic condition. It should be noted that the cross-sectional studies that have been done, although leaving many questions unanswered, have provided what little valuable information we now have. New cross-sectional studies can and will answer specific questions and help to generate hypotheses for other strategies of investigation. Methodological Research The development and standardization of concepts of chronic pain measurement are critical. The assessment of phenomena such as pain that have few or no directly observable components, the standardiza- tion of measurement of relevant clinical phenomena, the validation of indexes and scales of attributes such as quality of life, and even the establishment of minimum standards for recording relevant clinical and social information are neglected in much of the epidemiological and clinical literature. It is not glamourous to do or support method- ological research, but without such work one risks conducting
THE EPIDEMIOLOGY OF CHRONIC PAIN 119 uninterpretable and inconclusive investigations. Priority should be given to developing a small but valid annamentarium of data-gather- ~ng instruments and measuring techniques with as much care as is used in the laboratory sciences. Innovative dependent variables need to be developed and measured in standardized ways. Health status measures and indexes of the quality of life, measures of concordance or discordance of seeking compensation with objective evidence of physical disorder or disability, and measures of the style and content of clinicians' interventions are needed. Lastly, the very large array of interventions for disorders associated with chronic pain and disability, and the even larger array of combinations and permutations of such interventions, should be standardized in a relatively small set of packages that can be defined operationally, so that interventions are replicable and comparable from one study to the next. REFERENCES Afacan, A.S. Sickness absence due to back lesions in coal miners. Journal of Social and Occupatronc~l Medicine 32:26 31, 1982. Anderson, J.A.D. Back pain in industry. In: The Lumbar Spine and Back Pain (Jayson, M., ed.). New York: Grune & Stratton, 1976. Andersson, G.B.J. Epidemiologic aspects of low-back pain in industry. Spine 6:53 60, 1981. Bierring-Sorensen, F. Low back trouble in a general population of 30-, 40-, 50-, and 60-year-old men and women. Danish Medical Bulletin 29:289-299, 1982. Colvez, A., and Blanchet, M. Disability trends in the United States population 196~76: analysis of reported cases. American Journal of Public Health 71:4644471, 1981. Crook, J., Rideout, E., and Browne, G. The prevalence of pain complaints in a general population. Pain 18:299~14, 1984. Cypress, B.K. Characteristics of physician visits for back symptoms: a national perspec- tive. American Journal of Public Health 73:389~95, 1983. Drury, T.F. National Center for Health Statistics. Socioeconomic aspects of chronic pain. Presentation to the First Annual Meeting of the American Association of Algology, November 9-10, 1984. Enquete Sante Canada. La sante des Canadiens. Ministre des Approvisionnements et Services Canada. Catalogue 82-538F, 1981. Frymoyer, J.W., Pope, M.H., Clemens, J.H., Wilder, D.G., MacPherson, B., and Ashikaga, T. Risk factors in low-back pain: an epidemiological survey. Journal of Bone and Joint Surgery 65A:213-218, 1983. Kelsey, J.L., White, A.A., Pastides, H., and Bisbee, G.E. The impact of musculoskeletal disorders on the population of the United States. Journal of Bone and Joint Surgery 61A:959~964, 1979. Klein, B.P., Jensen, R.C., and Sanderson, L.M. Assessment of workers' compensation claims for back strains/sprains. Journal of Occupational Medicine 26:443~48, 1984. Louis Harris and Associates. Nuprin Pain Report. New York, 1985.
120 THE EXTENT AND COST OF THE PROBLEM McQuay, H.J., Machin, L., and Moore, R.A. Chronic non-malignant pain: a population prevalence. Practitioner 229:1109-1111, 1985. Mechanic, D., and Angel, R. Some factors associated with the report and evaluation of back pain. Journal of Health and Social Behavior, in press. Nagi, S.Z., Riley, L.E., and Newby, L.G. A social epidemiology of back pain in a general population. Journal of Chronic Disability 26:769-779, 1973. National Center for Health Statistics. The management of new pain in office-based ambulatory care: National Ambulatory Medical Care Survey, 1980 and 1981. NCHS Advance Data No. 97, June 13, 1984. National Center for Health Statistics. 1981 Summary: National Ambulatory Medical Care Survey. NCHS Advance Data No. 88, March 16, 1983. Reisbord, L.S., and Greenland, S. Factors associated with self-reported back-pain prevalence: a population-based study. Journal of ChronicDisabzlity 38:691-702, 1985. Spitzer, W.O., and Task Force. Rapport du Groupe de Travail Quebecois sur les Aspects Cliniques des Affections Vertebrales Chez les Travailleurs. L'Institut de Recherche en Santd et en Securite du Travail du Quebec, February 1986. (Forthcoming in Spine, 1987.) Svensson, H.O. Low back pain in forty to forty-seven year old men. II. Socio-economic factors and previous sickness absence. Scandinavian Journal of Rehabilitation Med- icine 14:5~60, 1982. Svensson, H.O., and Andersson, G.B.J. Low-back pain in forty to forty-seven year old men: work history and work environment factors. Spine 8:272-276, 1983. Svensson, H.O., and Andersson, G.B.J. Low back pain in forty to forty-se~ren year old men. I. Frequency of occurrence and impact on medical services. Scandinavzarz Journal of Rehab~litation Medicine 14:47~3, 1982. U.S. Department of Health and Hurr,an Services. Social Security Administration, Disability Survey 72. SSA Pub. No. 13-11812, 1980. Valkenburg, H.A., and Haanen, H.C.M. The epidemiology of low back pain. In: Symposium on Idiopathic Low Back Pain (White, A.A., and Gordon, S.L., eds.). St. Louis: Mosby, 1982. Wood, P.H.N. Statistical appendix~igest of data on the rheumatic diseases: 2. Recent trends in sickness absence and mortality. Annals of the Rh~urnatzc Diseases 29:324, 1970. Yelin, E., Nevitt, M., and Epstein, W. Toward an epidemiology of work disability. Milbank Memorial Furld Quarterly/Health and Society 58:388415, 1980.