Shoulder tendinitis and its relation to heavy manual work and exposure to vibration.

Three categories of construction industry workers (54 bricklayers, 55 rockblasters, and 98 foremen) were compared in a cross-sectional study. In a structured interview they reported exposure to loads lifted, vibration, and years of manual work. They were also subjected to a clinical investigation including medical history and a detailed shoulder examination. Among the rockblasters 33% had signs of tendinitis in the left and 40% in the right shoulder. Among the bricklayers and foremen 8-17% had signs of shoulder tendinitis. In a multiple logistic regression being a rockblaster compared with being a foreman showed an odds ratio (OR) of 3.33 for left-sided and 1.71 for right-sided shoulder tendinitis. Vibration exposure yielded an OR of 1.84 and 1.66 for the left and right sides, respectively. Vibration exposure or work as a rockblaster seemed to be risk indicators for tendinitis of the shoulders.

Several report s have indicated that stres s and strain in the workplace are factors in the developm ent of occupational shoulder disorders (1-5 ). There arc several conditi ons that can cause shoulder pain, but bicipital tend initis and supraspinatus tendiniti s form the great majority of soft tissue lesions in the shoulders (6).
In the construction industry many workers suffer from shoulder pain. Palpable pain is a common finding on examination in these cases. Together with pain reaction to isometric contraction of the affected muscle, this finding is interpreted as a sign of inflammation of the muscle attachment or the tendon (7) .
In the present study the objective was to determine whether signs of tendiniti s or muscle attachment inflammation in the shoulders was related to different work loads (eg, lifted loads), years of manual work, hours of exposure to vibration, or job title .

Study subjects
The source population, drawn from construct ion industr y workers in the Stockholm region, constituted three subgroups with a high or low degree of the exposures under study. In the Swedish construction industry, bricklayers eith er lay bricks or spend periods plaster ing; they do not cast concrete. Rockbla sters drill holes both underground and on the surface. They charge dynamite and transport rocks. In tunnels they also trim the roof. Construction foremen are engineers and not manu al workers, and they are rarely recruited from the manual worker group . The foremen are supervisors on the constru ction site; they also work in offices, which are often close to the construction place. The three group s thus represent different occupational work loads , although they have a common trade.
Representatives from these study groups were identified and randomly selected from the local union files. First, all of the 75 rockbl aster s in the files were included, and then 75 bricklayers in the same age groups were randoml y selected. Finally, 110 foremen were randomly selected who worked with large companies so that foremen that had been exposed to manual work would be avoided.
All of the subjects were invited to part icipate by letter and reminded by telephone the day before the exa mination. Refusal s were received from 19 bricklayers, 20 rockbl aster s, and 13 foremen, either because they did not wish to participate or because they lived in another part of the country or abro ad. One bricklayer was excluded from the investigation because of language difficulties. One of the bricklayers had previously worked for 18 years as a foreman and only for the last two years as a brickl ayer ; he was, as a consequence, categorized as a foreman. Thus the persons under study, all male, finally comprised 54 brickl ayers, 55 rockbl asters, and 98 foremen.
A summary of the background variables and exposures of the subjec ts is presented in tables I and 2.

Methods
The investigation had the following three parts: exposure assessment. clinical examination, and radiographic examination. The results of the radiographic investigation have been published elsewhere (8).
A questionna ire was developed to record the estimated load lifted during a worklife, exposure to vibration, years of manual work, years as foreman, smoking habits, dexterity , sports activities engaging the arms, and current and previous citizenship. A specially trained nurse interviewed the participants using the questionnai re.
The exposure was analyzed on the basis of (i) sum of the loads lifted during work years, (ii) sum of the hours of expos ure to vibration, (iii) years of manual work, and (iv) job title.
Load lifted was summarized on an annual basis. The participant calc ulated how many days he had worked per year, how many loads he had lifted per day, and their weight. All of the worktoo ls were recorded separately. The weight of a brick and of the trowel and mortar together is approximate ly 3.5 kg. Rockblasters the most frequent ly use jack hammers that weigh 47.2-50.7 kg and are lifted several times an hour. Rockblaste rs also load rocks of differe nt weights, and these loads were also taken into account although their weight was not easi ly estimate d.
Load lifted was categorized into one of the following three classes: 0-709, 7 10-25999, and > 25 999 t. The limits 710 and 26 000 t were chosen because they represented the median values for the foremen and the bricklayers and rockblasters, respectively . The value 710 t correspo nded to the lifting  of 19.7 kg per hour, 8 h a day, 225 d a year for 20  years. Years of manual work were categorized into 0-9, 10-28, and >28 years, the last category representing the median value for the bricklayers and rockblas ters together.
Exposur e to vibration was reported in the protocol for each tool used and expressed as hours of exposure . The reported hours were then multiplied by I, 10, and 100, depen ding on the tool that had been used. These exponents correspond to the three steps of energy emission from the vibrating tools, as suggested by Gemne et al (9). The individual values were then classified into one of the following three categories : 0-8999, 9000-255 199, and > 255 199 h. The limits represented the median values for the foreme n and rockblas ters, respectively.
The age range was 26-70 years. The mean was 50.2 years for the bricklayers, 51.8 years for the rockblasters, and 45.8 years for the foremen (table 2). We dealt with the differences in the distribution of age, smoking habits, dexterity, and sports activities between the three groups by including all of these variables in all of the regression models. isometric contraction in any of the four rotator cuff muscles or the biceps muscles. The results of subjects who had experienced pain in the shoulder during the last year and, on examination, had pronounced pain reaction to palpation or isometric contraction were analyzed in separate logistic regression models (table 3). We combined these variables into "a clinical entity" to demonstrate the relation between clinical tendin itis and physical work load.

Interaction analysis
The background variables of age, smoking habit s, dexterity, and sports activities were fitted into the Scand J Work Environ Health 1993, vo l 19, no 1 unconditional multiple logistic regression analysis in all of the models. The different work exposure variables load lifted, vibration hours, and years of manual work have been included separately in tables 3 and 4. Load lifted and vibration were fitted into the models in the categories previo usly described in table 2. In table 5 the high-and low-exposure groups were compared. In one model in table 5, vibration exposure and load lifted were entered simultaneously. The job categories were bricklayer, rockbla ster, and foreman. In all of the logistic regression models the results were calculated separately for the left and right sides.
The interaction analyses were performed with the EPILOQ@computer program.  Life. hours of sports activities that included movements of the arms were entered in the regression analysis on an hour per hour basis. Weighted hours, that is hours of exposure to each tool multiplied by 1, 10, or 100 corresponding to the vibration energy emitted by the tool according to Gemne et al (9) .

Results
Signs of tendinitis of the shoulder were defined as pronounced palpable pain of the muscle attachment or pronounced pain reaction to isometric contraction in any of the four rotator cuff muscles and the biceps muscles. Among the rockblasters 40% had signs of tendinitis in the right shoulder, and 33% had signs of tendinitis in the left shoulder, whereas among the foremen and bricklayers only 8-15% had positive signs of tendinitis (table I). The bricklayers had lower frequencies of signs of tendinitis than the foremen.
In the unconditional multiple logistic regression analyses for vibration exposure, the odds ratio for signs of tendinitis was 1.66 for the right side and 1.84 for the left side (table 4). Load lifted during worklife and the variable years of manual work yielded odds ratios close to 1.0 for the right side, but for the left side the load lifted yielded an odds ratio of 1.81. Years of manual work yielded an odds ratio of 1.87 for the left side, and job title an odds ratio of 2.26; the 95% confidence interval (95% CI) for job title included 1.0, indicating uncertainty (table 4).
Work as a rockblaster in comparison with work as a foreman yielded a relative risk of 3.33 (95% CI 1.21-9.15) for signs of shoulder tendinitis on the left side and 1.71 (95% CI 0.71---4.17) for the right side. Work as a bricklayer or rockblaster in comparison with work as a foreman yielded an odds ratio of 2.10 (95% CI 0.84-5.24) for the left side and 1.00 (95% CI 0.47-2.16) for the right side. Work as a bricklayer yielded an odds ratio of 0.44 (95% CI 0.16-1.25) for signs of shoulder tendinitis on the right side, but for the left side the coefficient estimates did not converge because of too few cases. Adjustments were made for age, dexterity, smoking, and sports activities.
The entity of tendinitis (ie, the combined variable of symptoms of pain in the shoulder during the last year and either pronounced palpable pain or pronounced pain reaction to isometric contraction) was also more common among the rockblasters than among the foremen and bricklayers (table I).
Vibration yielded even higher odds ratios. The risk estimate was 2.49 for the left side and 1.86 for the right side. For load lifted, years of manual work, and 46 job title the estimated odds ratios were not significant, as indicated by the confidence intervals (table 3).
The findings presented in tables 4 and 5 indicated the possibility of a exposure-response relationship in that a comparison of high-vibration exposure with low-vibration exposure yielded higher odds ratios (table 5) than the comparison of three classes, as in table 4. In table 5 a comparison was demonstrated between low exposure and high exposure without the medium exposure.

Discussion
Exposure to vibrating tools such as jackhammers for long periods of time was demonstrated to raise the relative risk for tendinitis of the shoulder muscles. In some models job title, loads lifted, or years of manual work yielded increased risks for tendinitis of the shoulders, but, according to the 95% confidence intervals, the estimates were uncertain.

Selection bias
A selection bias that may have influenced the outcome is the possibility that the issue under study, tendinitis of the shoulder, makes it impossible to work as a bricklayer or rockblaster but does not influence work as a foreman to an equivalent degree. Such selection would lead to an underestimation of the relative risk, since there would be fewer bricklayers and rockblasters with tendinitis of the shoulder available for the study. There were six foremen who had changed jobs from heavy manual work to foreman but only one of them had changed jobs because of a musculoskeletal disorder of the shoulder.

Misclassification of exposure
Nondifferential misclassification diminishes any truly increased risk (10). In our study, nondifferential misclassification might have been introduced by the use of subjective assessments of exposure. Kilbom et al ( II ) demonstrated that 20% of self-reported lifts did not occur in the workplace and that only 10% of the workers reported correct weights and frequencies of mat erial handl ed. In the pre sent study, the use of many different exposure variables was one way of min imizing the subjec tive influen ce on estimations of occupational stress . Another was to categorize the expo sure parameters in broad clas ses so that the exact values played a less prominent role.
The reported mean hours of vibration exposure for rockblastcrs were almost 22 times larger than for bric klayer s and 17.5 times larger than for foremen . After mul tipli cation by I , 10, or 100 to produce energy-weig hted hours of vibration expos ure, the rockblasters ' values were 14.7 times larger than the bricklayers' and 19.3 times larger than the foremen ' s. This findi ng indicates that the forem en reported many more hours of vibratio n exposure with tools emitting low-ene rgy vibrati on (table 2). Thu s there was no indication that the transformation into energy-weighted hours of vibration overestimated the vibration exposure.
Age can result in a differential miscl assifi cation of exposu re assessment since older peopl e in general may have more difficult ies in recallin g exposure than younge r peopl e. Th is phenom enon would also diminish a truly increased risk .

Misclassification of outcome
Th e pain reactions to pa lpation and isometric contract ion were classified into one of three grades. In the ca lculations only grade 3, pron oun ced pain reactio n, was classified as a positive sign of tend initis or muscl e attachment inflamm ation. The differenc e between gra des 2 and 3 is relatively distinct. In order to minimi ze the risk of miscla ssificati on (eg, by alteri ng criteria by time) eac h grading of any uncertainty was made as a con sensu s judgment by one of the authors (BS) and a physiothera pist.
The demonstrated signs of tendiniti s, palpable pain, and pain reacti on to isometric contrac tion, are not the exact crit eria used for the clinical diagnos is of tendiniti s or muscle attachment inflammation. The clinical entity used in our investigation includ ed symptoms of pain during the last year that could have originated from structures other than the tendons or the muscle attac hments in the sho ulders . From clinical experience we know that tendinitis and muscle attachment inflammation are the mo st common disorders in the shoulders. Therefore it is highl y probable that our "clinica l entity" was closely related to clinica l tendinitis. In other words persons who have experienced pain in their shoulder during the last yea r and on exa mination have pronounce d pain reac tio n to pa lpation or isometric contraction have probably had a disorder in the ide ntified muscle attac hment or tendon that , in clinical practice, would have been classified as inflamm ation or tendinitis.
In clini cal practice patients call on a physician because of pain , and the clinical exami nation verifies the diagno sis tendin itis. In that situation the numb er of tendinitis patients doe s not refl ect a true preva-Scand J Work Environ lI ealth 1993, vol 19, no 1 lence ratio of tendiniti s in the population since there are persons with tend initis who, for var ious reasons, will not see a physician for their symptoms. The situation is that of selection bias. In the present study no such bias was introduced as the study had a crosssec tional design.

Confou nding
Age, smoking habit s, dexterit y, and sports activities, all of which are possible confounders , were included in all of the mult iple logistic regression models. Age was entered as three classe s in the logistic regression analysis, and it yielded odd s ratio s of 2-3 in most of the models for the right side and odds ratios of 1-2.4 for the left side.
Smok ing also see med to have an influence on tendinitis of the shoulder although, in most cases , the 95 % confidence inter vals indica ted uncert ainty. Amon g the bricklayers 83% were smok ers or exsmoke rs, among the rockblasters 80% we re smokers or ex-smokers, whereas among the foremen 63 % were smokers or ex-smokers (table I ). Thu s there was a co-variation between smoking and heavy work.
Sports activities we re entered in all of the models as hour s of sports acti vities without classi fication. Sport s activities , cla ssified as low-, med ium-, and high-l evel , will be an alyzed in relatio n to tendinitis of the shoulder and osteoarthrosis of the acromioclavicular joint, and the results will be reported in another publ ication .
The three groups of subjects were chosen initially to contrast heavy manual work to light work and exposure to vibration to no such exposure in the same study base of constru ction trade employees . Bricklayers work with equa l loads on each hand, and rockblasters also lift their tool s and mater ial with both hand s. Thus dexterit y may have little significance in these situations, but it was treat ed as a confounder in the interaction analyses since there were more lefthanded and ambidextro us workers among the bricklayers and rockblaste rs. Tend initis was more common on the right side in all of the groups (table 1). In all of the interaction analyses however the left side yielded higher odds ratios than the right side for all of the work-re lated variables. One reason for this finding might be that the right hand is used more often in vigorous activities, even outside work, and that the rockblasters are an exce ption because they have to use both hands sim ultaneo usly in vigorous acti vities at work .
Vibration exposure seemed to be a risk indicat or, but, since vibration exposure is indivisible from static load and heavy lifti ng in the present data, all three are possible risk indi cator s or can be looked upon as confounders. In the analysis in table 5, lifted load was entered simultaneo usly with vibration exposure. Vibration still rend ered increase d odd s rat ios (OR 1.87 and 2.64), a finding which supports the fact that vibration without the influence of lifted load still acts Scand J Work Environ Health 1993. vol 19. no I as an indicator of risk for tendinitis of the shoulders. The subjects who had reported the highest exposure to vibration were the rockblasters. Rockblasters also make heavy lifts several times a day as opposed to bricklayers and foremen . Vibration exposure is thus indivisible from exposure to heavy lifts. The collected data did not separate heavy lifts from light ones, and the question of wheth er vibration is a true risk indicator or whether there is another risk factor involved that is strongly linked to vibration exposure, like heavy lifts, remains unanswered.

Mechan ism of injury
The mechanism of injury to a tendon or a muscle attachment in the shoulder includes several factors. The arterial pattern of the rotator cuff has been mapped by Moseley & Goldie (12), and they found no evidence that the "critical zone" for ruptures and calcified deposits was much less vascularized. During raised intramuscular pressure there might however be a lowered blood flow and a possibility for injury (3). Another obvious injurious factor is momentar y overload (14). Yet another possible factor is overload as a result of enhanced muscle contraction by the tonic vibration reflex (15,16).
The our finding that rockblasters had high relative risks for tendiniti s could be explained by all of these theoret ical factors. It is obviou s that working as a rockblaster involves exposure to vibration. Vibration exposure from handheld tools also implies exposure to static load, which has been demon strated to raise internal muscle pressure (13). Rockblasters lift heavy burdens like jackhammers and rocks of different weights. The lifting of heavy weights can cause ruptures of the tendons or muscle attachments (14).
Bricklayers were found to have a low prevalence of clinical entity , 1.8%, which was the same prevalence figure that Herberts et al (2) and Bergenudd et al ( 17) reported. This prevalence is low even though bricklayers lift a mean of 29437 t during their worklife, and it indicates that repetitive loading with small weights is not hazardou s to the shoulder muscles, maybe because there is such a low amount of static load in the work of a bricklayer. It is highly unlikely that bricklayers leave work because of shoulder tendinitis to a higher extent than rockblasters since both job s are considered to be two of the hardest in the construction industry. Therefore selection bias in the present study was unlikel y.

External validity
Signs of tendiniti s in the shoulders (eg, palpabl e pain) were found in 40% of the right shoulders and 33% of the left shoulders of the rockbla sters. The same signs of tendinitis have been investigated by Bergenudd et al (17), and they found that only 2% of a group of health y subjects had palpable pain in the rotator cuff or biceps muscles of the shoulders. Herbert s et al (2) examined welders and found a prevalence of 16-18% for tendiniti s of the shoulders in that group. The authors drew attention to the arm positioning, namely, overhead work , as an important risk factor for shoulder tendinitis. Luopajarvi et al (4) investigated assembly workers and found an increased prevalence of shoulder tendinitis. Static load and overhead work were pointed out as risk factors for tendiniti s of the shoulders.
There have been several studies that have demonstrated a relation between heavy work and shoulder pain, but there are few studies in which quant ified exposure variables within the occupation have been analyzed 0, 3,5). The results of our study support those of previou s studies in the assumption that heavy work is a risk indicator for shoulder tendinitis, but we also found that vibration exposure could to be a true risk indicator. The risk for shoulder tendiniti s on the right side increases by 1.66 for each step between the three classes of vibration exposure (0-8999, 9000-255 199, >255 199 exposure hours), and for the left side the correspondin g increase is 1.84.

Concluding remarks
Exposure to vibration or work as a rockblaster seems to be a risk indicator for muscle attachm ent inflammation or tendinitis in the shoulders. However, since vibration exposure from handheld vibrating tools is interlinked with exposure to static load, and for rockblasters it is also interlinked with exposure to heavy lifts, vibration expo sure is possibly an indicator for either of these risk factor s. Work as a bricklayer indicated a low risk for shoulder tendinitis. Years of manual work and the sum of load lifted during one's worklife did not seem to be risk indicators for tendiniti s of the shoulders.
Further efforts should be made to release humans from acting as a stand for tools that are heavy and transmit vibration to the arms and the rest of the body. The tools should be constructed so that they contain the stand needed for their operation. The stand could be a small caterpillar or some sort of tripod. In this way exposure to vibration and heavy static load could be avoided.