Radiographically detectable lumbar degenerative changes as risk indicators of back pain. A cross-sectional epidemiologic study of concrete reinforcement workers and house painters.

Radiographi cally detectable lumbar degenerative changes as risk indicators of back pain: a cross-sectional epidemio logic study of concrete reinforcement workers and house painters. Scand J Work Environ Health 1989;15:280-285. The association between radiographically detectable degenerative changes in the lum bar spine and back symptoms was studied, along with the possible effect of occupational work load. The subjects were 216 concrete reinforcement workers and 201 house painters. A questionnaire provided in formation on work history and earlier back accidents, and a standardized interview produced data on back symptoms. The occurrence of disc space narrowing, anterior and posterior spondylophytes, and end plate sclerosis was recorded separately for each intervertebral space from lateral lumbar radiographs. Moder ate to severe degenerative changes were associated with increased risk of sciatic pain but not with the occurrence of lumbago or nonspecific back pain. The different types of degenerative changes provided no further information. In a multivariate logistic regression analysis degenerative changes and earlier back accidents were significant independent predictors of sciatic pain. When these two variates were allowed for, the effect of occupation was not significant.

Lumbar disc degeneration is considered a major cause of back symptoms (1-3). However, the most common means of obtaining information on the degenerative status of the spine, the conventional radiograph, has been subjected to considerable criticism. It has been reported that only one in 2500 spinal radiographic examinations reveals clinically unsuspected positive findings (I).
Because back pain troubles people of optimal work age, measures have been sought to identify individuals susceptible to back pain in industrial settings. Spinal radiographs have been used for preemployment screening purposes, but this practice has been found unsuitable and unjustified (4)(5)(6)(7)(8). The specificity and sensitivity of lumbar radiographs have proved unsatisfactory in detecting symptomatic backs (9,10). Nevertheless, many studies have shown that back symptoms are related to radiographically detectable degenerative changes of the spine (9)(10)(11)(12)(13)(14)(15)(16), but there are also those in which no such relationship was found (17,18

280
In the present study men in two physical occupations, concrete reinforcement work and maintenance house painting, had lumbar spinal radiographs and a history of their back symptoms taken in 1977. Concrete reinforcement work has been shown to burden the back more heavily than house painting (19), but in most other respects workers in these two occupations are comparable and represent stable groups of skilled construction workers with a similar socioeconomic status. Concrete reinforcement workers have been reported to have a higher prevalence of sciatic pain (20), and also a higher prevalence of radiographically detectable lumbar degenerative changes, than house painters (21). The aim of our present study was to ascertain the relation between back symptoms and radiographic signs of lumbar spinal degeneration and to determine whether this relationship is affected by occupational work load.

Subjects and methods
This study was restricted to active male workers, aged 25 to 54 years, who had at least 5 years' experience in their current occupation. All 258 concrete reinforcement workers in these age and seniority categories listed in the regional trade union of Uusimaa Province were enrolled in the study. The reference group comprised 235 house painters who were selected from the register of the painters' local trade union of the Helsinki region of Uusimaa Province. The house painters were frequency-matched with the concrete reinforcement workers according to five-year age strata. Two hundred and sixteen concr ete rein for cement workers (84 % ) and 201 hou se painters (86 070) participated in the study. The mean age of the concrete reinforcement workers was 37.7 (SO 6.6) years, and that of the painters 38.6 (SO 6.6 years).
Th e data on work experience, back accid ents, and smo king were gathered with a self-administered qu estionnaire which was checked for completeness by a physiotherapist. Th e workers were asked to list all pre vious occupations which the y had had fo r at least a year, and also th e number of years in these occup ation s. On the avera ge, the concrete rein forcement workers reported 9.0 years in other occupations (4.2 years in the construction.industry, 4.2 years in agricultural or fore st work) and the hou se pain ters 3.7 years (0.5 years in the con stru ction industry, 2.2 years in agricultur al or fore st work). The number of back accidents or str ains was asked for the past 12 months and fo r the time before th at separately.
The ph ysiotherapist mea sured the height and th e weight of th e workers and cond ucted a standa rdized interview concerning back symptoms. The mean height of the concrete reinforcement workers was 174.9 (SO 6.6) em, and that of the paint ers 174.1 (SO 6.6) em. The respective weights of the two groups were 79.3 (SO 11 .0) kg and 77.3 (SO 12.2) kg. In this report the symptom data are based on the responses (yes, no) to each part of the follo wing que stion: During the past 12 months have you had (i) sciatic pain (defined as back pain rad iatin g to a leg), (ii) lumb ago (defined as a sudden ba ck pain cau sing constrained po sture of th e back), (iii) other backache or pain (here called nonspecific back pain )?
Lateral lumbar radiographs were taken. The infor med co nsent of the subjects was obtained befor e th e examination. The subjects stood 1.5 m from the X-ra y tube, positioned at the level of the iliac crest. The radi ographs were read jointly by two radi ologists (TM and A Z) who did no t kno w the age or occup ation of the sub jects. The occurrence of disc space narrowing , marginal an terior and posterior osteophytes of the vertebra l bodies (spond ylophytes), and end-plate sclerosis was recorded sepa rately for each of the lumba r interve rtebral spaces acco rding to the following four-grade classification: 0 = no change, I =slight change , 2 = moderate change, 3 = severe change. The gra ding was based on visua l judgment suppo rted by a pre selected set of reference films.
Th e reliability of th e classificati on was evaluated in a pilot study that preceded the final readin g of the radiogra phs. The two rad iologists read 50 radio graphs independently, and weighted kappa coefficients were calculated to measure th e agreement beyond cha nce between the two raters. The coefficients varied between 0.4 2 and 0.85 for disc-space narrowing in different inter vertebral spac es, 0.45 and 0.88 for anterior spond ylophytes, 0. 23 and 0.56 for po steri or spo ndy lo-phytes, and 0.46 and 0.80 for end-plate sclerosis. Thu s the agreement was satisfactory, except for posterior spo ndy lo p hytes , which were, on the oth er hand, detected the mo st infrequently (th e pre valence being 18 % for the concrete reinforcement workers and 14 % for the painters).
In the following text, when the findings are described for the entire lumbar spine (Ll-Sl), disc space narrowing, spondylophytosis, and end-plate sclerosis are given the maximal grade of their occurrence; no distinction is mad e betwe en anterior and posterior spondyloph ytes. The grade for th e genera l occurrence of degenerative changes (" all degenerative changes") in the lumbar spine is th e maximum of th e gra des for th e three items.
The ordinary ch i-square test and the Mantel-Haenszel pro cedure (22) were used in the statistical testing of the frequency data.
The relation between radio graphically detectable degenerative changes and sciatic pain was further investigated with multi vari ate logistic regression mod eling according to th e GLIM3 pr ogram (23) supplemented with a macro introduced by Wacholder (24); the macro allows the calculation of the estimates of risk ratios (RR) and the ir 95 % confidence inte rval s (95 It/ o CI) from the regression coefficie nts and th eir sta ndard error s. The dependent variable was the 12-month prevalence of sciatic pain. Mod eling was carried out with indicator variables for degen erative changes (grade 0, grade I, grade 2-3) and covariat es [occupation (hou se painting, concrete reinforcement work); reported ea rlier back accidents (no, yes); age (25-29, 30-34, 35-39,40-44,45-49,50-54 years); height ($169, 170-179,~180 em); body mass index ($23.9, 24.0-27.9,~28.0 kg/rrr '): smoking (non smok ers, exsmo kers, smo ke rsj]. Mult iple disc degeneration ha s been considered to be related to back symptoms (I) , and so the number of narrowed intervertebral space s (0, I , 2 o r more) was also regarded as a determinant. Th e number of years in present occupation was at first included in the set of covari ates, but becau se of the high correlation with age the effe ct of these two variables could not be separated. Thus the number of years in present occupation was excluded from the final modeling, and onl y occupa tion was used as the indicator of occupational exposure. First, a saturated model with main effect terms was constructed , after which we tested for the significa nce of th e term s by excluding them from the model on e at a time and ob serving the significance in the change of the mod el fit. We also looked for significant interactions betw een th e covariates by adding first-de gree product term s to the model and testing for their significance. No significa nt interaction s emerged.

Results
The pre valen ce of gra de I degener ativ e changes in th e radio gra phs was simila r for the concrete reinforcement workers and the house painters, but grade 2-3 changes were more common in the former group (table I).
The relationship between the 12-month prevalence of sciatic pain and the degenerative changes is presented in table 2. Lumbago seemed to occur indepe ndently of lumbar degeneration . Nonspecific back pain showed an increasing trend as the severity of degeneration increased, but the trend lacked statistical significance. Sciatic pai n was dist inctly associated with the degenerat ive changes. The age-adjusted risk ratios for the three types of degenerative changes of grade 2-3 ranged between 1.6 and 1.8 (table 3). The risk ratio increased to 2.2 (95 0,70 CI 1.4-3.4) when all degenerative change s were combined .
The univariate risk ratios for the covariates included in the mult ivariate analysis are presented in tab le 4. Even tho ugh height, bod y mass index, an d smoking were not significant ly related to sciatic pain, they were a Maxi mum of the grades for disc space narrowing, spondylophytes, and end-plate sclerosis. Table 3. Relation between radiographically detectable degenerative changes in the lum bar spine and the 12-month prevalence of sciatic pai n.

Ris k rat io
Grade of the degen erative c hanges· · b Table 5. Relat ion bet w een radiographicall y det ectable degene rative changes in th e lumbar sp ine and the t z-montn prevalence of sc iatic pai n.

Discussion
Rad iographically detectable mod erate or severe degenerative changes in the lumbar spine were associated space nar ro wing. Wh en both covariates were simultan eou sly includ ed as determinants in mult ivariate modelin g, neither was significant conditionally on the oth er , and no significant interaction between the two was detected. The relation s between degenerat ive changes at different levels of the lumbar spine a nd the prevalen ce of sciatic pain ar e presen ted in tabl e 6. In th e presacral space, disc space narrowing was associated with an increase in the pre valen ce of sciatic pain (age-adj usted RR 1.6, 95 % CI 1.2-2.1), as well as spondyloph ytes (age-adjusted RR 1.5,95 070 CI 1.1-2.0) and end-plate sclerosis (age-adjusted RR 1.5,95 % CI 1.1-2.1). For the L4-LS space, the associations were wea ker for disc space narrowing (age-adj usted RR lA, 95 070 CI 1.1-1.8) and for sp ond yloph ytes (age-adj usted RR 1.3,95 070 CI 1.0-1.7). There was a similar tren d also at oth er levels.
Crude Ad ju sted fo r age Adjusted for all covariates v ' Table 4. Risk rati os and 95 % conf idence in terv als (95 % CI) for the univar iate eff ect s of cova riat es on the 12-month prevalence of sciatic pain.
included in the models to allow for their potential confounding effect.
T he mult ivariate analysis yielded similar resul ts for the three types of degenerative chang es as determinants of sciati c pain separately. When all thr ee were simultaneously included in the model, none of them retained its significa nce conditio nally on the other two, and no evidence emerged that any of them was a more powerful determinant of sciatic pain than the others.
Thus the co mbined variable of all degenerati ve .changes was tak en as th e determ inan t for the final modeling (tabl e 5). Degener ation and earlier back accidents proved to be signifi cant independent det erminants of sciatic pain; the risk ratios adj usted for the other covariates were 1.9 (95 % C I 1.2-2.9) for degeneration of grade 2-3 and 1.8 (95 % CI 1.3-2.5) for ear lier back accidents . When these two determinant s were allowed for, neithe r occup ation no r age retained its significance.
No significant interaction was detected between occupatio n and degeneration. In order to reassure the non existence of a mod ifying ef fect by occupat ion, analogo us multivariate logistic regression models were created for the two occupa tiona l gro ups separa tely; these mod els yielded simila r relations between scia tic pa in and degen eration and earlier back accidents.
Twenty-two percent of the concrete reinforcement work ers and 13 % of the pai nters had two or more narr owed inte rvertebral spaces . The effec t of multiple disc space na rrowing on the l2-month prevalenc e of sciatic pain was similar to th at of grade 2-3 disc Risk 95 % con fi dence a 95 % confldence interval in parentheses . Covar iate b Maximum of the grades for disc space narrowing, spondylorat io interval phytes, and end-plate sc lero sis.
Occupation C Earl ier bac k accid ents , occ upation, age, height, bo dy mass index, and smoking. Multivariate log ist ic regress ion analysis. Hou se pain ting 1.0 Con cr ete reinforce me nt wor k 1.4 1.1-1.9 Earli er back accidents with an increase in the risk for sciatic pain but not for other types of back pain. Differentiation between various types of degenerative changes provided no further information. This result seems reasonable because back pain radiating to the legs is an indication of neural irritation, often due to structural derangements in the spine, but lumbago or other back pain may arise from soft tissues, ie, muscles or tendons. The fact that degenerative changes, even disc herniations, are not always symptomatic is well established on the basis of both clinical and epidemiologic experience. Possible contributory factors are the localization of the changes in reference to the neural elements and narrowness of the canal or the foramina of the nerve roots, which may be congenital or due to encroachment caused by degeneration (25,26). The relationship between sciatic pain and degenerative changes was the strongest in the two lowest intervertebral spaces, which is in accordance with prior experience (27).
In a cross-sectional study, health-based selection is a matter of concern. Age groups older than 55 years were excluded from this study in order to minimize the effect of possible selection bias. Selection among the subjects hardly has a bearing upon the observed relationship between lumbar degeneration and back symptoms, but it might have caused a negative bias in the effect of occupation on this relationship.
Disc degeneration must be fairly advanced before it can be visualized as a narrowed disc space in a plain radiograph (28). In most earlier studies a radiographic diagnosis of disc degeneration has been employed with varying criteria; spondylophytes and disc space narrowing either alone or in different combinations with end-plate sclerosis have been used. However, spondylophytes do not always occur in connection with a degenerated disc, and it seems erroneous to describe spondylophytes without unequivocal disc space narrowing as being representative of disc degeneration (29).
In the interview, sciatic pain and lumbago were given distinct definitions in order to assure a uniform understanding of these concepts by the subjects. The 12-month prevalences of back symptoms were used as outcome variables because recall error tends to be greater for longer periods of observation. In addition, even in a cross-sectional setting, it seemed reasonable to study the role of lumbar degeneration as a predictor for recent symptoms because degeneration is a slowly advancing process.
In a study of miners, manual workers, and office workers Kellgren & Lawrence (13) found, in accordance with this study, that the relation between radiographic changes and back symptoms was much the same in the three groups. The higher occurrence of back pain among miners was concluded to be more closely related to degenerative changes than to the arduous nature of the work, except insofar as the work was responsible for the degeneration. In two Swedish 284 studies of selected worker groups (II, 12), the following three types of back symptoms were characterized: sciatica, lumbago, and insufficiency. A definite association between back trouble and radiographic disc degeneration was found in both studies, but the three different back symptoms provided no further information. In a Danish sample of the general population of 60-year-olds, L5 (RR 1.5) and L4 (RR 1.8) disc degeneration was associated with an increased risk for lO-year occurrence of low-back pain radiating to the legs. For other low-back pain the associations were weaker [RR 1.0 and 1.2, respectively, as estimated from the data presented by Biering-Serensen et al (9)J.
In this study lumbar degeneration and earlier back accidents had an independent effect on the occurrence of sciatic pain. The data on back accidents were based on self-report. There may be positive bias in the observed association between back symptoms and back accidents because those with back trouble may be more prone than others to recall their past accidents. It is difficult to obtain reliable and accurate data on back accidents, especially retrospectively. In this study no information on the type or severity of the accidents was available. Episodes of back pain arising suddenly during work are likely to become reported as accidents. According to a previous study (30), only some of such events are true accidents, a considerable proportion are nonaccidental injuries, and for some no cause can be attributed. It seems conceivable that the association between sciatic pain and earlier back accidents may reflect the true injurious effects of accidents, but it may also in part reflect the predictive power of prior episodes for the future attacks, as reported by several authors (31)(32)(33)(34).
According to our study, plain lumbar radiographs seem to provide valuable information for the occupational epidemiology of back ailments. Radiographically detectable changes are permanent, and their occurrence rates are not affected by subjective experience or motivation, unlike data on symptoms or clinical examination. Of course, exposure to radiation is a matter of concern in epidemiologic studies. In the future, magnetic resonance imaging could offer means with which to study the degenerative process of the discs without known adverse effects on the subjects. At present, its costliness and lack of capacity are factors limiting the use of this method.
Even though lumbar degenerative changes were found to be a risk indicator for sciatic pain, the fact remains that the specificity and sensitivity of radiographic signs of lumbar degeneration in the detection of symptomatic backs are not high enough to justify the use of spinal radiographs for preemployment screening purposes. However, the twofold increase in the risk for sciatic pain among workers with moderate or severe lumbar degenerative changes should warrant measures of secondary prevention to be directed towards such workers in occupational health practice.