Objectives To evaluate if occupational mechanical exposures are associated with an increased risk of surgery for varicose veins (VV) in the lower extremities.
Methods We conducted a longitudinal study of persons from the Musculoskeletal Research Database at the Danish Ramazzini Centre who were 18–65 years old when they provided baseline questionnaire data during 1993–2004. Exposure estimates were obtained from a job exposure matrix based on expert ratings. The register information on first-time surgery for VV was retrieved. We used Cox regression analyses.
Results During 416 317 person-years of follow-up among 38 036 persons, 851 first-time operations for VV occurred. Using standing/walking <4 h/day and uncommon lifting as references, exposure–response relationships with risk of surgery were found for men. For women, the risk increased too, but without clear exposure–response patterns. The adjusted HRs for ≥6 h/day spent standing/walking were 3.17 (95% CI 2.06 to 4.89) and 2.34 (95% CI 1.72 to 3.19) for men and women, respectively. For high lifting exposures (≥1000 kg/day), the adjusted HRs were 3.95 (95% CI 2.32 to 6.73) for men and 2.54 (95% CI 1.95 to 3.31) for women. Other risk factors were increasing age for men and parity for women. Minimal leisure-time physical activity, a high body mass index and smoking were not associated with increased risk.
Conclusions The results suggested an increased risk of surgery for VV in relation to prolonged standing/walking and heavy lifting and a preventive potential of more than 60% of all cases in exposed occupations.
Statistics from Altmetric.com
What this paper adds
Varicose veins of the lower extremities are common in Western populations. Prolonged standing/walking has been implicated as a risk factor, primarily based on cross-sectional studies, which have used exposure measures obtained by self-report. Few studies have explored associations with heavy lifting.
This longitudinal study applied a job exposure matrix based on expert ratings, thus avoiding inflated risk estimates due to recall bias. The risk of surgery for varicose veins was associated with prolonged standing/walking and heavy lifting at work.
Varicose veins (VV) of the lower extremities are dilated, palpable, subcutaneous veins with a diameter of >3–4 mm. VV can often be identified by inspection.1 ,2 In the clinical, aetiological, anatomical and pathophysiological classification of venous disease, VV correspond to class C2.2 Traditional theories regarding the development of VV have focused on mechanical factors such as valve incompetence and venous wall dilation due to venous hypertension. Prevailing theories now emphasise molecular changes in the extracellular matrix of vessel walls and valves leading to valve incompetence.1 ,3 These changes may be due to the mechanical stretching of the veins and/or increased venous pressure.1 ,3 In a resting standing posture, the venous pressure at the ankles increases to approximately 90–100 mm Hg2 ,4 from 50–60 mm Hg when sitting4 and approximately 10 mm Hg when lying down.4 When the calf muscle pump is used during walking, the venous pressure decreases to approximately 10–30 mm Hg2 ,4 and returns to the resting standing level after 20–30 s when standing still.2 ,4
The prevalence of VV has been reported to be 2–56% among men and <1–73% among women.3 ,5 To a large extent, these considerable variations can be explained by varying disease definitions and methods of diagnosing VV as well as differences in the study populations’ age and other characteristics.3 ,5 Annual incidence rates (IR) from 1.4 to 2.3 per 100 person-years have been found based on repeated clinical examinations.6–8 Symptoms of VV include cosmetic concerns, oedema, a feeling of heaviness, pain and discomfort.1 ,2 Treatment options include compression stockings and surgery.9 Well-established but rare indications for surgery are complications such as advanced skin changes, ulceration and variceal haemorrhage. Surgery may also be indicated due to cosmetic concerns, to obtain symptom relief, and to prevent disease progression.9 Even though up to one-third of patients do not seek medical assistance for their VV,9 ,10 surgery for VV is one of the most common operations in Denmark.11 Surgery rates for VV vary quite considerably across European countries with Danish rates being close to the average.12
Parity is a risk factor for VV, and the prevalence of VV increases with age.2 ,3 ,5 ,13 The association between incidence and age is less clear.3 ,5 ,8 ,10 A high body mass index (BMI),3 ,5 ,13 ,14 low dietary fibre intake,3 ,5 ,15 female sex,3 ,5 ,15–17 familial predisposition2 ,3 ,5 ,17–19 and smoking3 ,5 ,8 ,18 have been suggested as risk factors, but the findings have been inconsistent. Studies of the aforementioned personal risk factors have focused on the self-reported presence of VV16 ,19 or clinically diagnosed VV,14 ,15 ,17 ,18 not surgery for VV. Occupational mechanical exposures in terms of prolonged standing/walking and heavy lifting have also been implicated.3 ,5 ,13 ,20 Prolonged standing/walking at work has been examined in a number of studies, of which some found a positive association with VV,11 ,15 ,16 ,18 ,19 ,21 ,22 while others did not.6 ,14 ,17 ,23 ,24 Apart from three longitudinal studies,6 ,11 ,21 all of these studies had a cross-sectional14–17 ,19 ,22–24 or case–control design,18 and apart from three studies,11 ,17 ,22 all of the studies used self-reported exposure estimates. The use of self-reported exposure estimates in cross-sectional or case–control studies implied a risk of inflation bias because individuals with symptoms might exaggerate their exposures.3 ,5 ,20 In some studies, sitting and standing were combined as a measure of physical activity at work25 or at work and during leisure time8; hence, these studies did not address the influence of prolonged standing/walking at work on the risk of VV.20 Two cross-sectional studies found an association between heavy lifting and VV when both sexes were analysed together17 or specifically among men.23 Other findings regarding heavy lifting have been negative.6 ,14 ,15 ,21 We are not aware of previous studies that have focused on the risk of surgery for VV (as distinct from self-reported VV,16 ,19 clinically diagnosed VV6 ,14 ,15 ,17 ,18 ,22–24 and hospitalisation due to VV11 ,21) in relation to occupational mechanical exposures. The aim of this study was to evaluate the hypothesis that prolonged standing/walking and heavy lifting at work is each associated with an increased risk of surgery for VV.
Design and data sources
We performed a longitudinal study using the Musculoskeletal Research Database (MRD) at the Danish Ramazzini Centre, which was established for the study of occupational risk factors for musculoskeletal disorders treated at hospitals.26 At present, the database comprises 39 590 persons who have provided questionnaire data in at least one of nine original studies (conducted successively during 1993–2004) of musculoskeletal symptoms in Danish working populations. The proportion who responded was ≥70% in all the original studies.26 We identified cases of surgery for VV in the Danish National Patient Register (NPR),27 where admissions to public somatic hospitals have been registered since 1977 and outpatient contacts since 1995. Visits to private somatic hospitals have been registered since 2001, but this was not mandatory until 2003.27 From the Danish Medical Birth Register,28 which contains computerised information on all births in Denmark since 1973, we obtained information on parity; numbers of previous live births and stillbirths are included at the first entry of each woman in the register. The Danish Register for Evaluation of Marginalisation (DREAM)29 was used to obtain information on education level. Information on death and emigration was obtained from the Danish Civil Registration System.30
A total of 265 persons participated in two of the original nine studies in the MRD and one person participated in three of the original studies. For these persons, we selected the most informative questionnaire data set. The study cohort was limited to persons who were 18–<65 years old at the date of participation in the original study (baseline). We excluded persons with prior VV surgery and persons with missing information on both job title and D-ISCO 88 code (D-ISCO 88 is the Danish version of the International Standard Classification of Occupations from 1988).
For all members of the study cohort, we identified first-time operations under a main diagnosis of VV between 1 January 1980 and 31 December 2011 in NPR. The diagnostic codes were 454×× (International Classification of Diseases, Eighth Revision, ICD-8, for the period 1977–1994) and I83 (ICD-10, from 1995 onwards). We used the following codes in the Danish Classification of Surgical Procedures: 8790–8804 (for the period 1980–1988) and 87 529, 87 532, 87 969–88 289 (for the period 1988–1995), and the following Danish Nordic Medico-Statistical Committee (NOMESCO) codes: PHB10-PHB14, PHB99, PHD10-PHD15, PHD99 (from 1996 onwards). In general, surgical codes in the NPR have been found to have a high validity.31
Information on job title was obtained from the baseline questionnaires and transformed into an occupational title represented in D-ISCO 88. Individual estimates of exposures to the lower extremities were obtained by connecting the occupational title with the newly developed Lower Body Job Exposure Matrix (JEM).32 The JEM cross-tabulates 121 job groups with expected similar exposure patterns, and five experts’ ratings of number of hours per day spent standing/walking and total load lifted per day (kg). The JEM includes exposure estimates for 689 occupational titles out of the 2227 occupational titles in D-ISCO 88—the remainder were either considered rare or obsolete (n=117) or minimally exposed (n=1421).32 If an occupational title was missing while a D-ISCO 88 code was available (n=6), exposure estimates were calculated by averaging across the original JEM-exposures for occupational titles with the same D-ISCO 88 code. For minimally exposed occupational titles, we set all exposure estimates to zero, although a truly non-exposed group hardly exists. Standing/walking was categorised as <4 (reference), 4–6 and ≥6 h/day. The total load lifted per day was categorised as 0 (reference), >0–<1000, and ≥1000 kg/day.
Lifestyle factors and other covariates
We included age at baseline, categorised as <36 (reference), ≥36–<45, ≥45–<56, and ≥56 years; BMI, calculated as weight/height squared (kg/m2) and categorised as <25 (reference), ≥25–<30, and ≥30 kg/m2; smoking status, categorised as never smoker (reference), ex-smoker and smoker; leisure time physical activity (per week) categorised as<2 h of light activity, ≥2–≤4 h of light activity (reference), >4 h of light activity or ≥2–≤4 h of strenuous activity and >4 h of strenuous activity; and—for women—parity at baseline (with 0 as the reference). Two of the original studies did not ask about smoking and one of these also did not ask about leisure time physical activity, height and weight. As a proxy for socioeconomic status, we assessed education level based on DREAM registrations of unemployment insurance fund membership at baseline. For 638 cohort members, this information was missing and we used their D-ISCO 88 codes to assess education level. We categorised education level as higher/medium, vocational and low;26 persons who received cash benefit were categorised in the low category (cash benefit is a part of the lowest level of the social welfare net in Denmark).
Analyses were performed separately for men and women. Sex-specific correlation coefficients were calculated between the two continuous exposure variables and between each of the two continuous exposure variables and education level; 95% CIs for the IR were calculated using the quadratic approximation to the Poisson log likelihood for the log-rate parameter. Crude and adjusted HRs with 95% CIs were obtained using Cox regression analyses. The proportional hazards assumption was evaluated by means of a global test, which also comprised a variable by variable test.33 Follow-up time was calculated from baseline until the first operation for VV, emigration, death or end of follow-up on 31 December 2011, whichever came first. We analysed one occupational exposure variable at a time, and a priori decided to adjust for age, BMI, physical activity in leisure time, smoking status and—for women—parity. Furthermore, we adjusted for baseline year to account for the original study and for the time trend in surgery rates. It was impossible to stratify by original study because of a lack of exposure contrast within some of the studies, but any effect of original study would be limited to lifestyle factors since exposure and outcome assessment (as well as assessment of parity and education level) was based on JEM and register data, which did not differ across the original studies. As a further precaution, we performed analyses stratified by original study using the three original studies, which had sufficient within-study exposure contrast for both men and women. Test for trend was performed using the exposure categories as a continuous variable. We also performed analyses with mutual adjustment for standing/walking and lifting to examine which of these exposures had the larger influence on the risk of VV surgery. Education level was only included in supplementary analyses due to expected correlations with occupational mechanical exposures. Crude and adjusted analyses were repeated with restriction to original studies which asked about all lifestyle factors.26 We performed interaction analysis between standing/walking (dichotomised as <6, ≥6 h per day) and lifting (dichotomised as <1000, ≥1000 kg/day). The variables were dichotomised by collapsing the minimally and moderately exposed groups because few persons had work which entailed standing/walking without lifting and vice versa.
Any potential for prevention was evaluated by calculating the excess number of cases above the expected number among the exposed, by multiplying the excess fraction, (HRadj−1)/HRadj, for the medium and high exposure category, respectively, with the number of cases within the corresponding exposure category, then summing up the excess numbers and dividing the sum by the total number of VV operations among the exposed and finally converting to per cent. Analyses were performed with Stata V.12 software (StataCorp LP, College Station, Texas, USA).
Of the 39 590 persons in the MRD, 340 were excluded because of age, 959 because of prior surgery for VV, and 255 because of missing information on both the job title and D-ISCO 88 code, resulting in a final study cohort of 38 036 persons; 16 259 men and 21 777 women. A total of 1130 occupational titles were represented, 931 among men and 696 among women. Among men, the most frequent occupational titles with high exposure to standing/walking were slaughterhouse workers, painters and machine-tool setters, and the most frequent jobs with high lifting exposures were slaughterhouse workers, paper-products machine operators and poultry workers. These occupational titles accounted for 54% and 62% of the jobs, respectively, with high occupational mechanical exposures. Among women, the most frequent occupational titles with high exposure to standing/walking were cleaners, shop assistants and poultry workers, and the most frequent jobs with high lifting exposures were nurses’ aides, poultry workers and home-based personal care workers, accounting for 70% and 83% of the jobs, respectively, with high occupational mechanical exposures.
Tables 1 and 2 show characteristics of the cohort at baseline according to categories of standing/walking for men and women, respectively. There was a strong positive correlation between standing/walking and lifting with a correlation coefficient of 0.68 for men and 0.73 for women. Correlation coefficients between standing/walking and education level were 0.30 for men and 0.48 for women, and for lifting 0.27 and 0.43, respectively. None of the cohort members who spent ≥6 h/day standing/walking were classified as minimally exposed to lifting, and almost none of the cohort members who lifted ≥1000 kg/day were classified as standing/walking <4 h/day. The majority of parous women (83%) had given birth to all their children before baseline. The mean total load lifted per day for men in the moderately and highly exposed groups was 490 and 2850 kg/day, respectively. For women, the corresponding mean loads were 506 and 1341 kg/day. Owing to the limited difference between these mean values among women, and in order to obtain comparable mean loads for men and women, we subdivided the highest exposure category to obtain an extra lifting category of ≥3000 kg/day. This resulted in mean loads for the two highest exposure categories of 1656 and 3498 kg/day for men, and 1093 and 3502 kg/day for women. However, only 3.5% of the women were in the highest exposure group, and only 9% of the men were in the second highest exposure group. The mean duration of standing/walking for men was 0.7 in the reference group, 5.3 in the moderately exposed group, and 6.6 h/day in the highly exposed group. The corresponding numbers for women were 0.1, 5.2 and 6.6 h/day. Of note, the mean standing/walking exposures in the reference groups were so small because we artificially set a large proportion of these exposure estimates to zero.
The total follow-up time was 416 317 person-years with a mean of 10.9 years per person, 10.4 for men and 11.3 for women. During follow-up, 851 first-time VV operations were performed, yielding an IR of 20.4 (95% CI 19.1 to 21.9) per 10 000 person-years, 236 operations among men (IR 13.9 (95% CI 12.2 to 15.8) per 10 000 person-years) and 615 among women (IR 25.0 (95% CI 23.1 to 27.0) per 10 000 person-years). The mean age at surgery was 49.2 years (SD 10.1), 49.7 (SD 10.7) for men and 49.0 (SD 9.9) for women. Overall, 4% of the operations were performed at private hospitals, 7% in the group with <4 h of standing/walking per day, 3% in the group with ≥4–<6 h of standing/walking per day, and 3% in the group with ≥6 h of standing/walking per day.
Tables 3 and 4 show the risk of first-time VV surgery in relation to occupational standing/walking and lifting for men and women, respectively. For men, both exposures showed exposure–response patterns when analysed one at a time in unadjusted and adjusted models. For women, an exposure–response pattern was seen for lifting, whereas the risk was more than doubled for both medium and high exposures to standing/walking. Additional analyses with an extra lifting category ≥3000 kg/day did not change the patterns (results not shown). With mutual adjustment for the two exposure variables, both standing/walking and lifting remained as risk factors among men, although the HRs were smaller (results not shown). For women, only standing/walking remained as a risk factor, and again the HRs were smaller (results not shown). Interaction analysis yielded HRs of 2.17 (95% CI 1.39 to 3.40) for prolonged standing/walking, 1.71 (95% CI 0.88 to 3.32) for heavy lifting and 1.27 (95% CI 0.54 to 3.00) for the interaction term for men. The corresponding HRs for women were 1.52 (95% CI 0.99 to 2.33), 2.03 (95% CI 1.56 to 2.64) and 0.86 (95% CI 0.49 to 1.50), respectively. For men, increasing age was the only other factor associated with surgery for VV. For women,<2 h/week of light leisure time physical activity, increasing age, and smoking at baseline showed small negative associations, while increasing parity showed a positive association with VV surgery. To explore the negative association between increasing age and VV surgery, we repeated the analysis which was restricted to women who were nulliparous, and found no association with age (results not shown).
In supplementary analyses adjusted for education level, the results hardly changed for men. For women, the HRs for occupational exposures were reduced by almost 25% (results not shown).
Since information on smoking and BMI was missing for a substantial part of the study population, particularly in the most highly exposed groups (tables 1 and 2), we compared results with and without adjustment for smoking and BMI in analyses, where we restricted the population to participants in those original studies which asked about these factors. Adjusting for smoking and BMI hardly changed the HRs for the exposure variables (results not shown). Analyses stratified by original study using the three original studies, which had sufficient within-study exposure contrast for both men and women, yielded adjusted HRs for ≥6 h/day spent standing/walking of 2.46 (95% CI 1.41 to 4.29) and 2.07 (95% CI 1.46 to 2.94) for men and women, respectively. For high lifting exposures, an adjusted HR of 3.29 (95% CI 1.88 to 5.75) was reached for men and 1.77 (95% CI 1.27 to 2.48) for women.
An estimated excess of 134 cases among exposed men (68%) and 256 cases among exposed women (58%)—altogether 61%—could in theory be prevented according to the adjusted standing/walking models (tables 3 and 4).
In this longitudinal study, we found exposure–response relationships between standing/walking and lifting at work and the risk of surgery for VV among men. Among women, we also found an exposure–response pattern for lifting, whereas the risk was more than doubled for both medium and high exposures to standing/walking. For standing/walking, the excess fraction of cases among the exposed was 61% (68% among men and 58% among women). Other risk factors were increasing age for men and increasing parity for women.
So far, this is the largest longitudinal study which has examined VV in relation to both lifestyle factors (and other covariates) and occupational standing/walking and lifting. We used a JEM with exposure estimates that were assessed independently of the outcome. In this way, we ensured that the observed associations were not inflated by recall bias. We are aware of only two previous studies of VV that have used a JEM approach.11 ,17 Our exposure estimates disregarded exposure duration, but the Danish labour market is quite stable, so most likely the results reflected associations with cumulative exposures. A limitation of the Lower Body JEM was that it did not differentiate between standing and walking. Previous studies based on self-reported exposure estimates have tried to separate standing and walking,6 ,14 ,15 ,24 but it is unknown to what degree they have succeeded, and findings regarding the relative importance of standing and walking have been inconsistent. A higher after-work venous pressure has been found among surgery room workers with standing work postures than among outpatient department nurses with combined standing and walking at work,34 and since the venous pressure is higher when standing than when walking,2 ,4 it seems plausible that standing is the more relevant exposure with regard to VV.
Exposure estimates based on the Lower Body JEM have shown good predictive validity with respect to inguinal hernia repair among men (standing/walking and lifting),35 and total hip replacement due to primary osteoarthritis among men (lifting),36 but not with respect to pelvic pain among pregnant women (lifting).37 Two11 ,21 out of three6 ,11 ,21 previous longitudinal studies have suggested an association between VV and occupational standing/walking. The negative study6 was a 13-year follow-up study of 555 participants who were sampled from the working age population and free of VV at baseline. The negative findings of that study may well be explained by the small study size and minimal exposure contrast with respect to baseline information on the percentage of the daily working hours spent standing/walking across the entire lifetime (<50% vs ≥50%).14 In our study, low leisure time physical activity was negatively associated with the risk of surgery for VV among women; we have speculated that spending time with the legs elevated after a day with prolonged standing/walking at work might be a protective factor, but other explanations—including chance—may be equally plausible.
We have not identified studies that have measured venous pressure at the ankle in relation to occupational lifting, but a positive association seems plausible, maybe through a link of increased intra-abdominal pressure.15 ,38 The JEM provided quantitative estimates of total load lifted per day, which has not been estimated in any of the previous studies of VV in relation to lifting.6 ,14 ,15 ,17 ,21 ,23 However, for jobs that entailed lifting of persons, the total load lifted per day may have been particularly difficult for the experts to assess, which may imply exposure misclassification in the JEM, especially for women. Women in jobs with high lifting exposures may also perform less exposed tasks than their male colleagues, and since the JEM did not provide different lifting estimates for men and women within the same occupation, this would lead to an underestimation of associations. Moreover, high lifting exposures were relatively uncommon among women. Maybe these factors explain why the exposure–response relationship for lifting was weaker among women than among men. We have not been able to think of the likely errors that can explain away the exposure–response relationship that we observed between lifting and VV among men even in the analyses with mutual adjustment for lifting and standing/walking. Interaction analysis also showed effects of each exposure. We are aware of only one other longitudinal study which has addressed heavy lifting as a potential risk factor for VV.21 This was a Danish 12-year prospective study of 5647 working age persons who were asked how much of their working day they spent standing/walking and lifting loads weighing >20 kg (<1/4 vs ≥1/4 of working hours).21 Maybe the negative findings of that study can be explained by the limited exposure contrast with respect to total load lifted per day, and the fact that the presented results were adjusted for standing/walking.
Denmark has a public and free healthcare system, and we think that our results can be generalised to other countries with easy access to surgery. Our results suggested a minor differential access to private treatment in relation to time spent standing/walking at work. However, any under-reporting from private hospitals would presumably be non-differential and even a major under-reporting before 2003 would not influence our risk estimates substantially since only 4% of the VV operations were performed in private hospitals. Education level only seemed to confound the results among women. This was most likely due to a stronger correlation between mechanical occupational exposures and education level among women than among men, but it might also have to do with cosmetic concerns. Likewise, it might be speculated that in particular among women, the association with high exposures might be confounded by cosmetic concerns, for example, if female shop assistants with high exposures to standing/walking had more cosmetic concerns than female office workers with low exposures. However, the influence of such a potential tendency on our results would be minimal because shop assistants accounted for only 4% of the VV operations in our study.
We found higher HRs of VV operations with increasing age for men, but not for women. This could not be explained by earlier surgery among women in relation to parity. VV are benign and treatable and therefore constitute an unlikely reason for leaving one's occupation; moreover, it seems unlikely that a healthy worker effect would be particularly strong among women. Another potential explanation might be that men tend to postpone surgery until they have complications,10 whereas women seek treatment earlier.10 ,13 Unfortunately, the NPR does not contain information about the severity of VV, and we identified few records of complications.
Given the easy access to surgery in Denmark, the observed surgery rates may be considered a good proxy for the incidence of developing VV. However, the relationships between occupational mechanical exposures and the risk of surgery for VV could reflect an increased risk of VV or a higher probability of surgery given VV if highly exposed persons were more inclined to have surgery due to symptom aggravation. No matter if our results reflected an increased risk of VV or a higher probability of surgery given VV, our results indicate that occupational preventive actions may reduce rates of surgery for VV. Further research is needed to evaluate whether reductions of the daily time spent standing/walking, walking rather than standing and even use of compression stockings39 ,40 during prolonged standing at work can prevent VV.34 ,40 In conclusion, our findings corroborated our initial hypotheses that the risk of VV surgery was increased in relation to prolonged standing/walking and heavy lifting at work and suggested a preventive potential of more than 60% of all first-time operations for VV in exposed occupational groups.
The authors thank the Danish Working Environment Research Fund supported the establishment of the Musculoskeletal Research Database at the Danish Ramazzini Centre (grant number 31-2009-09). The authors thank Tine Steen Rubak who made the Lower Body Job Exposure Matrix available for the study.
Contributors PF and SWS conceived the study. All authors contributed to the design of the study and to interpretation of the results. ST and PF conducted the analyses. ST drafted the manuscript in collaboration with SWS. All authors revised the manuscript for important intellectual content, and the final version has been approved by all authors.
Funding Danish Working Environment Research Fund.
Competing interests None.
Ethics approval The Danish Data Protection Agency.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.