Etiologic fractions for physical work load, sports and overweight in the occurrence of coxarthrosis.

Etiologic fractions for physical work load, sports and overweight in the occurrence of coxarthrosis. Scand J Work Environ Health 1994;20: 184-8. OBJECTIVES - The aim of this study was to estimate the impact of physical work load, sports, and overweight on the incidence of coxarthrosis. M ETHODS - A case-referent study was made of 239 male recipients of a hip prosthesis and 302 men randomly selected from the general population. Information was obtained by means of an interview and questionnaire. Exposures to physical workloadand sportswere measuredas thecumulative number of hours of exposure up to 49 years of age. Overweight was measured as the estimated body mass index at 30 or 40 years of age. RESULTS - The etiologic fraction related to the three risk factors was 40% for physical work load, 55% for sports, and 15% for overweight. Various measures of physical work load were considered, but they all proved to be correlated. CONCLUSIONS - Approximately 80% of the idiopathic coxarthrosis was explained by the presence of the three risk factors.


184
The etiologic frac tion (population attrib utable risk) has been introduce d as an appropriate parameter for quan tifyi ng the di sease-produ cin g role of an etiologic fac tor, and popul ation -based case-refere nt studies might directl y pro vide the mea sures req uired (2, 10, I I).

Subjects and methods
From a case-re fere nt study (12)(13)(14) of firs t-time prosth esis of the hip at four large hospitals in Stockholm , Swede n, the quantitative imp ortance of physical work load , sports, and overweight to the development of cox arthro sis has been studie d. Th e etiologic fractio n was calculated with the use of the fo llow ing operational entities for "physical work load": cum ulative number of hours exposed to dynamic work loads, cumulative number of hours working in a twisted locked pos ition, cu mulative number of ton s lifted , tota l number of heavy lifts , and cum ulative number of jumps . Th e cumulative numbers were collected up to the age of 49 yea rs, as we re the numbers of hou rs spe nt in spo rts act ivities.
The study base com pr ise d all Swedish men who were between the ages of 50 and 70 years and lived in the referra l areas of four large hospit als in Stoc kholm. Information was co llected from 1984 to 1988. The ca ses were the men in the study pop ulation who received a first-time prosthesis of the hip joint as a result of idiopathic os teoar thros is. The referents were randomly selected from the study base. The main as-peet s of the design, material, and meth ods of the study have been pre sented elsewhere (12)(13)(14) . All information about the participants' health status, smoking hab its, education, sports activities, and occ~pational career was obtained in a telephone interview .. According to this inform ation, eac h subje ct's o~cup~tlO.nal career was divided into different periods w ith similar work tasks . For each such peri od the st~dy participants were asked to complete a questionnaire about the specific ph ysical work load s in that type of work. The subj ects wer e asked ho w man ỹ ours per week were spent in different work posinons, how many kilograms were lifted per week, and other such que stions. Information regarding expo sure was co llec ted from the start of the occ upational career to the year of the diagnosis for the case s and to the year of the interview for the referents . Exposure was then aggregated for the men ' s worklife up to 49 yea rs of age. Three exposure groups were defined according to the loads in the reference group. Those unexposed and the 5% least exposed were considered to have low exposure. The rest of the exposed~n were divided into two equally large gro ups, classified as the medium-exposed and the high-exposed group.
The etiologic fra ctions were calculated as follows for the work loads: static + dynamic work , lifted tons, numbe~of lifted hea vy burdens (> 40 kg), and num -?~r of Jump s. The etiologic fraction for spo rts activitie s was ca lculated in the same way as for expo sure from work load . The exposure from spo rts activities was qu ant ified as low, medium, and high according to the tot al amount of hours spent in sports. Th e etiologic fract ions for overw eight were calcula ted so that~he lean group (BMI< mean BMI-I SD ) was cons ider ed unexpo sed and the rest were co nsidered ex posed, with the use of unadjusted odds ratios, or so that the lean group was con sidered the reference gro up and onl y those at least slig htly obe se (BMI > mean B~l + I SD) were con sidered expose d, with the use of adj usted odds ratios.
Th. e e~iologic f~actions were estim ated according to principles outlined by Miettinen ( I I) . The etio-I?gic fraction for all the cases in the study populatron (EFr.) was calculated as LEF, xCF, where CF is the frac tion of cases (fraction of all cases) in each cat egory of the determinant and EF is the etiolog ic fraction for the exposed. EF wa~~ca lculated as (RR -I)/RR, where RR is the adju sted odd s ratio for !hose e~posed in each exposure ca tego ry in comparson with those with the low est ex pos ure. Th e adjusted odds ratio s were obtained from stratified Mantel-Haen szel anal yses (12)(13)(14). Th e 95% confid enc e interval s for the EF we re based upon the sta nda rd error given by w diter ( 15). Wh ere indicated, over all etiologic fractions for two or more factors were calculated by mean s of Miettinen ' s formul a (II ): Scand J Work Environ Health 1994, vol 20, no 3

Results
The etiologic fractions due to occupational exposures ar~p~esented in table I for each of four operati onalizations of work load. Anal yses of cumulative work l~ad up to 29 years of age instead of up to 49 years did not cha nge the etiologic fra ctions by more than 7%. The var ious work loads were highl y correlated ; onl y 3 of the 31 ca ses with low expo sure to static and dynamic forces had experienced exposure to any of the ?ther loads. An analy sis with a reference group compnsed of tho se who had experienced no exposure to any of the variables listed in tabl e I yielded an eti ologic fraction of 40 %. All of the se estimates wer~adju sted for age, body mass ind ex, smoking habits, and sports ac tivi ties ( 12).
The eti ologic fra ction due to sports activities up to 49 year s of age was estimated to be 55 % (table 2). The overall eti ologic fraction due to occupati on and sport can be estimated from tables 7 and 8 in reference 13. With the use of the subjects une xpo sed to both spo rts activities and occupational load s as the reference cat egory (only 14 cases), the combined etiologic fraction for oc cupation or sport or both was est imated to be 66 %. However, in table 7 of reference 13, the numbers of hours of sports activities we~e.c.umulate? only up to 29 years of age. If sports ac tivittes are, ins tead, cumulated up to 49 years of age , the number of unexposed per son s becom es s~a ller, and the relative risks become larg er (13). If thi s larger etio log ic fraction for sports activities (55%) is co mbi ned with the etiologic fraction for occupation (40%) acc ording to Miettin en ' s formula, the combined etio logic fraction for occupation and sport is estimated to be 73 %.
Th e etio logic fra ctions due to overweight are prese nted in table 3. They ra nge from 13 to 17% de-pen~ing on the choice of reference gro up and the cho ice of age at whi ch the bod y mass index was estimated. With the use of the subjects une xpo sed to both sports ac tivities (up to 49 years of age) and overweight (at 30 years of age) as the reference category (onl y 7 cases), the combined etiologic fraction for overweight and sport was estimated to be 82 %. If sp~rts activities up to only 29 years of age were used instead, the es timation would be based on II ca ses and the etiologic fraction would be estimated to be 66 %.
The combined estimates were subj ect, howe ver , to large sampling errors based as they were on very few unexposed cas es . If the etiologic fractions for occ upation (40%), sports activities (55%), and overweight ( 15%) are co mbined according to Miettinen ' s formula, the tot al etiologi c fraction for the se ex pos ures is 77 %.

Discussion
To estimate the fraction of disease that can be avoided by reduci ng or eliminating the population e xpo- sure to a given etiolog ic agent is a prerequisite for evaluating disease-preventing strategi es. However, estimates of etiologic fractions must be interpreted with caution, as they presuppose that the observed relation s between factors and disease are causal relationships, and not merely statistical associations. The associations studied in this paper fulfill most of Bradford Hill' s relevant criteria (16,17) for causality, namely, strength of the association (relative risks above 2, except for overweight), dose-re sponse effect present except for the small group with extreme overweight, lack of temporal ambiguity, the hypothesis about "wear and tear" (common for all three of the studied risk factor s) biologically plausible, supportive experiments on animals available ( 12), consistency of the findings [the association between coxarthro sis and occupational load being repeated in a large, long-term , prospective regi ster study ( 18), the association between coxarthrosis and sports activi-186 ties being repeat ed in the two best of three studies on top athletes ( 13), but the quality of studies on overweight being generally of lower quality and giving inconsistent results].
The sources of bias, misclassification , and confounding in the present study have been discussed earlier (12- 1 4), and they are generally expected to be of minor importance. As far as the given estimates represent causal relationships, they contribute to the discussion on the proportion s of coxarthro sis which might be due to physical work load, sports, and overweight and which thereby could theoretically have been prevented .
As "occup ational exposure " is an inaccur ate description of exposure to a multitude of specific physical work loads during a work career, the estimates of the fraction of a disease due to "occupational exposure" are highly dependent upon the definition and categorization of the exposure variable. Etiolo gic fractions for four different measures of physical work load were presented. These measures were, however, so closely correlated that it was impossible to distinguish the specific effect s of the various risk factors. Thus the data simply indicated that approximately 40% of the coxarthroses might have been prevented if all occupational work loads had been eliminated.
With regard to the external validity of the occupational etiologic fraction , the Greater Stockholm area, in which the study was performed , is somewhat more urbanized than the average for the country. However, one of the hospitals in the study covered a rural region, and, due to migrat ion and urbanization in the 1950s and 1960s, the study base probably comprised a sample of work careers that was fairly represent ative for Sweden. The estimate, however, is historical, as it was based on, and primarily appropri ate for, the generations who worked in the middle of this century. If society had been in steady state, it would cause no problem to apply the obtained estim ate to future generations as well. But some of the physical work loads that were previously common or at high levels are now, to a large extent, controlled (at least in our part of the world). Thus it can be expected that the occupational etiologic fraction will decrease in the future. However, due to similar occupational structures and similar changes in it, the estimate is probabl y applicable to middle-aged and elderly generations in other Nordic countries.
Calculations similar to the ones made in the present paper have been performed in only one oth-er study, a Finnish study based on a representative sample of the population aged 30 years or over and including both men and women (19). In that study the total etiologic fract ion for prior trauma of the lower limbs, physical stress at work, and overweight was 59% (note that sports were not included ). The etiologic fraction s for physical stress at work and overweight can be calculated with the aid of table 2 in reference 19, and the etiologic fractions are 37 and 24%, respectively. The former is very close to the 40% calculated in the present paper, and the 24% for overweight is within the confidence limits of the estimate for overweight presented in the present paper.
No other studies allow similar calculations of an etiologic fraction for coxarthrosis. However, using a case-referent study of disability pension due to musculoskel etal disorder s (20), we were able (with the aid of table 4 in the paper) to calculate an occupational etiologic fraction from the occupational titles for the hip cases (EF = 72%), knee cases (EF = 75%), neck and shoulder cases (EF = 69%), andPlow-back cases (EF =71 %). These Petiologic fractions should, howeve r, be interpreted with great caution , as they may reflect both a causal relationship and the difficulty in continu ing work in certain occupations while suffering from a musculo skeletal disorder. In the cross-sectional HANES I study (21), an occupational etiologic fraction of radiograph ic osteoarth ritis of the knee was estimated to be 32%. Unfortunately the studies of neck and shoulder disorders (7-9) and of low-back pain (22, p 28) are generally of low epidemiologic quality and do not \87 Scand J Work Environ Health 1994, vol 20, no 3 permit etiologic fractions to be estimated. The few good studies, however, provide large odds ratios and suggest that the etiologic fractions may be appreciable.
The etiologic fraction for sports activities of all kinds was 56%. The most hazardous sports seemed to be track and field , racket sports, and soccer (13) . However, the participants in individual sports were few, and the results must be interpreted with caution on this level. The estimate of the etiologic fraction for sports activities in general was, however, so large that it warrants further research into the different loads and possible later deleterious effects of various sports. This finding is important, especially in light of the many positive aspects of sports activities in general.
It should be pointed out that the individual etiologic fractions of two factors contributing to a multifactorial disease are not invalidated, although their sum exceeds 100 % , and each estimate mu st be interpreted as a measure of di sease reduction if the factor in question was th e first to be eliminated (2 2). It is furthermore pos sible to calculate etiologic fractions for preventive strategies covering the si multaneo us elimination of several factors (23). The present study indicates that approximately half of the occurrences of coxarthrosis in the present generation might have been prevented by the elimination of either of the single factors occupational exposure or sports activ ities . Approximately 75 % of the occurrences mi ght ha ve been prevented if ph ysi cal work load, sports, and o verwe ight had all been eliminated.
Although, clinically, idiopathic coxarthrosis ha s been considered a disease "occurring without known cause" (24), we conclude that the occurrence of the disease, from an epidemiologic point of view, turns out to be highly explainable, understandable, and preventable.