A cross-sectional study of back belt use and low back pain amongst forklift drivers

ARTICLE IN PRESS

International Journal of Industrial Ergonomics 37 (2007) 505–513 www.elsevier.com/locate/ergon

A cross-sectional study of back belt use and low back pain amongst forklift drivers Darren M. Jouberta,, Leslie Londonb a

Curtin University of Technology, School of Public Health, GPO Box U1987, Perth, Western Australia 6845, Australia b School of Public Health and Family Medicine, University of Cape Town, South Africa Received 1 June 2006; received in revised form 16 November 2006; accepted 8 February 2007 Available online 28 March 2007

Abstract Purpose: To determine the association between back belt usage and back pain amongst forklift drivers exposed to whole-body vibration (WBV). Method: Cross-sectional analytical study design amongst 158 drivers using back belts and 39 controls. Back pain was assessed using a Standardised Nordic Questionnaire for musculoskeletal disorders and WBV was measured on a sample of forklifts as per ISO 2631. Results: Compliance with belt usage was 90%. Eighty-nine percent of drivers reported back pain ever; and back pain after driving was associated with vibration intensity and work area. Belt usage was associated with back pain after driving on multivariate analysis. When restricting analysis to back belt users alone, frequency of usage was associated with increased rates of back pain. Conclusions: In a high WBV-exposed group, back belt usage was not associated with decreased risk of LBP. Users appeared to have increased LBP, although the relationship may be due to selection bias due to non-random assignment of back belt condition. Relevance to industry: back belt use for WBV exposed professional drivers should not be considered as a valid control measure to reduce the prevalence and intensity of LBP. r 2007 Elsevier B.V. All rights reserved. Keywords: Whole-body vibration; Back belts; Drivers; Back pain

1. Introduction Musculoskeletal disorders are amongst the most common cause of occupational morbidity. For example, in the USA musculoskeletal disorders affect 7% of the general population and account for 14% of physician visits and 19% of hospital stays (Rosenstock, 1997). The lifetime prevalence of lower-back pain amongst the general population has been estimated at 60–80% for industrialised countries (Hulshof et al., 2002). In a two year study in Australia, 58% of work related problems reported to general practitioners were related to musculoskeletal disorders with 42% related directly to the back (National Occupational Health and Safety Commission, 2001). Back injuries are also the most highly compensated injury type at the workplace, accounting for almost one Corresponding author. Tel.: +618 9266 7029; fax: +618 9266 2958.

E-mail address: [email protected] (D.M. Joubert). 0169-8141/$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ergon.2007.02.005

third of all compensation in the USA. The precise cost of musculoskeletal disorders is unknown, although a conservative estimate made by the National Institute for Occupational Safety and Health put it at $13 billion annually for the USA (Rosenstock, 1997). Few equivalent data are available for developing countries mainly due to under diagnosis and serious limitations in reporting systems. It is estimated that in Latin America between 1% and 4% of all occupational illnesses are reported (World Health Organisation, 1999), a problem likely to be widespread in developing countries in Africa. Whole-body vibration (WBV) is recognised as an important risk factor for occupational low back pain in a variety of occupational groups (Bogadi-Sare, 1993; Brendstrup and Biering-Sorensen, 1987; Futatsuka et al., 1998; Malchaire et al., 1996; Pope and Novotny, 1993; Pope et al., 1987; Riihimaki et al., 1989; Wikstrom, 1993; Wilder et al., 1996). At least four European countries have placed WBV injury on their scheduled lists of occupational

ARTICLE IN PRESS 506

D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513

diseases (Hulshof et al., 2002). Exposure to the mechanical energy produced whilst driving vehicles, particularly with an increase in mechanisation and long working hours, can result in WBV levels that exceed the ability of the natural protective mechanisms of the body to offer adequate protection. Some evidence suggests that LBP arising from WBV is related to early degeneration of the lumbar spinal system, and herniated lumbar discs (Dinardi, 1997), and to irritation of the nerve endings supplying the various structures of the spine such as the outer annulus, ligaments or related muscles, and compression of the nerve roots passing through the motion segments (Bongers and Boshuizen, 1990). However, many of the mechanisms of damage and effects of vibrational energy on the spinal column and surrounding tissues are not fully understood. Despite the substantial morbidity from LBP at the workplace, few proven interventions are available to prevent this problem. Back belts, alternatively called kidney, lifting, abdominal or back support belts have been used in occupational settings for the prevention of injuries due to lifting activities. The devices themselves vary in design, but are generally manufactured from layers of elasticised material, with Velcro ends for attachment and adjustment, and may or may not have additional vertical support stays. These devices are popularly believed to offer support to the lumbar area, by increasing the intraabdominal pressure and supporting the spinal column, and surrounding organs. Where empirical evaluations of back belts have been conducted in other studies, this has usually been carried out in the setting of exposure to manual lifting activities and not to WBV. Even so, their efficacy at preventing lower back injuries from lifting is the subject of much debate with some studies finding no significant benefits to using back belts (Mitchell et al., 1994; Reddell et al., 1992; van Poppel et al., 1998; Walsh and Schwartz, 1990) and, others indicating some benefit (Kraus et al., 1996, 2002; Thompson et al., 1994). Various studies that investigated the biomechanical effects of back belts during lifting found an increase in intra-abdominal pressure, and low back compression, but did not come to any clear conclusions as to whether the belts actually reduced fatigue, increased support of the back muscles and spinal structures (Ciriello and Snook, 1995; Harman et al., 1989; Lander et al., 1992; McGill and Norman, 1987; McGill et al., 1990; Nachemson et al., 1986). In addition other studies (Hunter et al., 1983; Rafacz and McGill, 1996) indicated that blood pressure and heart rates were increased with belt wearing, and this indicated a potential increased loading on the cardio-vascular system. Notably, not one occupational hygiene agency or governmental organisation makes mention of the use of belts as a protective device against WBV exposure (Joubert, 2000) and no evidence exists in the literature outlining a possible protective mechanism for back belts in relation to the consequence of WBV for back pain. However, despite a complete absence of evidence for their efficacy in preventing the adverse consequences of WBV

(Joubert, 2000), many South African workplaces have been increasingly adopting back belts as preventive measures for low back pain arising from WBV. Much of the perception regarding the benefit of back belts appears to stem from aggressive advertising that associates prevention of occupational morbidity with similar ‘preventive’ usage of back belts in sporting activities such as motor bike racing (Canavan, 1999; Kevco/Stubbs, 1999) or that conflates usage for lifting activities with usage amongst forklift drivers exposed to WBV. One such workplace to adopt back belts without an evidence base for their effectiveness was the port of Durban in South Africa in response to the high number of complaints amongst forklift drivers concerning lower back pain and other musculoskeletal injuries and associated high levels of absenteeism. This situation presented an opportunity to evaluate the effectiveness of back belts in decreasing the intensity and prevalence of back pain in two groups of forklift drivers, one using back belts and one control group. 2. Materials and methods Management and the workers’ trade unions adopted a back belt program at the port in two of the three main areas as a preventive measure. Because back belts were used in only two of the three operational areas, a ‘natural experiment’ was possible, allowing a test ‘‘back belt group’’ (Areas 1 and 2) to be compared to a control group (Area 3). The total study population comprised all drivers of 3, 4, 4.5, and 5 ton forklifts in the permanent employment at the port authority (n ¼ 291). Because of pressures from employees and their trade union, the belts were introduced before a cohort could be established, necessitating a cross-sectional design. Data collection occurred 10 months after the issuing of back belts. The facility at Durban is a high-volume container port with 31 million pieces of cargo handled in 1998 from a total of 4677 vessels. Drivers operate on an 8-h revolving shift cycle, with approximately a third of the drivers on a shift at one time. Labour force turnover is low because of the semiskilled nature of the work. At the port overtime work is common, as a result of which, drivers often have extended exposure periods of up to 12 h. Sources of high levels of vibration exposure at the port included both irregular road surfaces and outdated forklifts, ranging in load capacity from 3 to 40 ton. Many of the older forklifts have inadequate seating, with poor ergonomic design, absence of any adjustment capacity and almost no vibration damping capabilities. Questionnaire data were collected using an instrument based on the standardised Nordic Questionnaire for the Analysis of Musculoskeletal Symptoms (Kuorinka et al., 1987), translated into English and Zulu. Additional questions were added to adapt for local operating conditions at the port including questions on shift work, rest periods, forklift types and personal behaviours such as sporting and other extra mural activities. Exposure to the

ARTICLE IN PRESS D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513

intervention was based on questionnaire reporting of being issued with, and using, back belts as well as their frequency of use. Potential confounders measured included age, prior occupations, reported typical driving speeds, reported use of rest breaks, and other reported control measures used (e.g. support pillows). Because a small number (n ¼ o20) of workers completed the questionnaire without interview, self-administration of questionnaire was also considered as a potential confounder. Prevalence of back pain as an outcome was estimated in four ways: ever, in the past year; current and pain after driving (pain in the lower back experienced during or shortly after driving). The intensity of back pain was evaluated using a 101point pain numerical rating scale (NRS-101) (Jensen et al., 1986). The NRS-101 scale involves asking the respondent to rate his or her perceived level of back pain intensity on a numerical scale from 0 to 100, with 0 representing one extreme (e.g., no pain), and the 100 representing the other extreme (e.g., pain as bad as it could be). Respondents’ ratings of pain intensity were for their back pain generically, irrespective if it was current, past or in the last year. The number stated by the respondent as representing the level of pain intensity is the basic datum for the NRS-101 scale. The NRS-101 is simple to administer and score, and the scale offers greater room for variability of response categories (Jensen et al., 1986). A similar pain intensity and disability scale has been included in the vibration injury network health surveillance questionnaire (Vibration Injury Network, 2001) and has been used in various other studies for pain intensity (Von Korff et al., 1992). To explore the intensity of lower back pain at its worst and least, NRS-101 scores were recoded to five equal categories ranging from 0 (least ¼ 0%) to 5 (maximum ¼ 475%). If the intensity of pain at its worst scored 2 or more greater than the intensity at its least, it was classified as variable back pain. If the pain levels were similar, i.e., the pain at its worst differed by one point or not at all, it was classified as constant back pain. The site of pain (lower, middle or upper back/shoulders) was also identified. A disability scale was not included in this study as used by Von Korff et al. (1992) to grade pain severity. Intensity of back pain was also gauged from a history of consulting a doctor, taking medication and time off work. Workers reporting back pain during or shortly after driving were asked to give the duration of pain (in hours, days, weeks or months). Driver’s participated in the study on the basis of informed consent with the option to withdraw at any time. WBV was measured on a sub-sample of forklifts (n ¼ 9) to develop prediction equations for WBV based on surface area, driving speeds and forklift type. Measurement of vibration was conducted by using a Bruel & Kjaer (B&K) tri-axial piezo-electric accelerometer (Model 4322) mounted on the forklift seat in a deformable rubber disc shaped pad, which followed the seat contour. All equip-

507

ments were set up and used as per ISO 2631 (International Organisation for Standardisation, 1997) and as specified in the SABS General requirements for the competence of calibration and testing laboratories (South African Bureau of Standards, 1990) for accredited laboratories, calibrated on an annual basis. Mean WBV measured in the sub-sample of forklifts over all of the driving surfaces and seat adjustments was 1.18 m/s2. Out of a total of 32 mean vibration levels evaluated for the different combinations of area, seat adjustment and driving conditions, only two were less than the EEC Machinery Directive (Council of the European Union, 2002), action limit of 0.5 m/s2, and 47% (n ¼ 15) exceeded the recommended 8 h exposure level of 1.15 m/s2 (Council of the European Union, 2002) (details in Appendix 2). WBV levels were higher in work areas where back belts were issued both in rough and smooth operational conditions (rough mean ¼ 1.60 m/s2 (SD ¼ 0.67) and smooth mean ¼ 0.94 m/s2 (SD ¼ 0.48) compared to control areas (rough mean ¼ 0.93 m/s2 (SD ¼ 0.26) and smooth mean ¼ 0.67 m/s2 (SD ¼ 0.11). Predicted vibration levels were estimated for each area from the results of the regression equation using the variables most typical of actual operational conditions, i.e., forklifts driven on a rough surface with the seats unadjusted for weight (details in Appendix 2). 2.1. Statistical analyses Basic demographic and driving characteristics were compared among the sites using chi-square tests for categorical data and analysis of variance for numeric variables. Because no significant differences were found amongst the two back belt groups (Table 1), the combined back belt group was used for all subsequent comparisons to the control. The effectiveness of the back/kidney belts was assessed based on different low back pain outcome measures, including prevalence and intensity of back pain compared between back belt and control groups. Multivariate logistic regression was used to identify significant predictors of back pain, including cumulative WBV exposure, controlling for known confounders. Because compliance may be related to the effectiveness of the belt, sensitivity analyses were conducted alternatively including and excluding non-compliant drivers. Usage (yes/no) and frequency of usage of back belts were therefore the independent variables of interest in the analyses. Analyses were also conducted amongst back belt users only to control for the potential confounding effect of work area. In the latter analyses, exposure to back belt usage was characterised based on reported frequency of usage. A stepwise regression procedure with backward elimination was used in the analyses to identify the most parsimonious model, detect independent risk factors associated with back pain and adjust for confounding. The likelihood ratio chi-square and Hosmer–Lemeshow goodness of fit statistics were used to determine the best

ARTICLE IN PRESS 508

D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513

Table 1 Demographic data and occupational history of forklift drivers Back belt group

Age in years: mean (SD) Race: (Black African %) Prior occupation (%) General worker Driver Other History of forklift driving in years: mean (SD) History of forklift driving at the port in years: mean (SD) a

Control group

p-value

Area 1 (n ¼ 108)

Area 2 (n ¼ 50)

Total (n ¼ 158)

Area 3 (n ¼ 39)

47.6 (6.8) 94

46.2 (7.7) 100

46.9 (7.3) 96

49.1 (7.2) 97

0.17a 0.33b

96 0 4 14.1 (8.2)

96 2 2 13.9 (9.5)

96 1 3 14.0 (8.9)

95 5 0 12.3 (10.1)

0.15b

14.0 (8.1)

13.8 (9.6)

13.9 (8.9)

12.0 (10.2)

0.48a

0.55a

Two-way ANOVA comparing means. Fishers exact test to compare proportions.

b

model. Some of the independent variables were highly correlated, e.g., length of driving and cumulative vibration exposure, and they were not used together in one multivariate model but were alternated sequentially in order to ascertain their individual effects.

Table 2 Prevalence of back pain (ever, today and last 12 months)

2.2. Sampling and response rates All drivers in the back belt areas (Areas 1 and 2) were included in the study, while a sample of 51 subjects were included from the control area (Area 3) based on a sample size calculation (1 control: 4 exposed) assuming a prevalence difference of 25%, a error ¼ 0.05 and b ¼ 0.8. Response rates were 100% and 96% in Areas 1 and 2, respectively. After exclusion criteria were applied, 39 control subjects from Area 3 remained in the study. The final sample therefore consisted of 197 forklift drivers from the three sites. No data could be collected on non-responders to assess the extent of non-response bias. However, the control group was thought to be similar to the back belts group in as much as they had also previously requested back belts but were never issued any. Drivers in the control group who had previously worn back belts were excluded from the analysis. 3. Results Demographic data are summarised in Table 1. No differences were noted in mean age or driving experiences, or in race or prior occupation. Of the 158 drivers in the back belt areas, 14 reported never being issued with belts. These subjects were included in the analysis on the intent to treat principle as exclusion was seen as a possible source of bias in regards to compliance and effectiveness of back belts. Sub-analysis was, however, conducted amongst the user group to take account of actual belt usage. Amongst those with belts, frequency of usage was high, with 76% reporting use all the

Back pain ever Back pain today Back pain in last 12 months Lower back pain after driving

Back belt group

Control group

Total (%)

Prevalence odds ratio for back pain: back belt vs. no back belts (95% CI)

Total (n ¼ 158) (%)

Area 3 (n ¼ 39) (%)

92 42 91

80 49 80

89 43 89

1.2 (1.0–1.4) 0.8 (0.2–2.9) 1.2 (1.0–1.4)

83

62

79

1.4 (1.0–1.7)

time, 18% sometimes and only 6% rarely. Eighty one percent of back belt users believed that the belts were effective in reducing back pain. No formal training was given to the drivers on the correct manner of fitting and use of the back belts which were a standard elasticised wrap around back belt worn around the waist and fastened in the front by Velcro fasteners. In the entire sample 89% of all drivers reported ever having previous back pain (Table 2) and reported prevalence’s for different categories of back pain were high for both back belt groups and controls. Back pain ever (92% vs. 80%; POR ¼ 1.2; 95% CI 1.0–1.4) and in the past year (91% vs. 80%; POR ¼ 1.2; 95% CI 1.0–1.4) were more common in the back belt group, while current back pain (today) was not significantly different between exposed and controls (42% vs. 49%; POR ¼ 0.8; 95% CI 0.2–2.9). Lower back pain after driving was also recorded with the back belt group showing an increased prevalence compared to controls (83% vs. 62%; POR ¼ 1.4; 95% CI 1.0–1.7). The latency before onset of back pain after starting as a professional driver (in years) was not significantly different

ARTICLE IN PRESS D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513 Table 3 Pain intensity distribution for those with lower back pain Pain intensity rating scale

509

Table 6 Model fitting: predictors of back paina after driving amongst forklift drivers in the three areas (n ¼ 175)

Back belt group (n ¼ 131) Control group (n ¼ 24) Variablesb Leasta (%)

Worstb (%) Leasta (%)

Worstb (%)

18 72 8 2 100

0 28 46 26 100

0 0 54 46 100

OR (95% CI)

Constant 0–24 25–49 50–74 75–100 Total

0 75 25 0 100

Work area

Area 2 vs. Area 3 Breaks

Other more frequent breaks vs. every 4–6 h

Watching TV

Yes vs. no

Vibration exposurec

High vs. low

a

Comparison of pain at its least—Wilcoxon rank sum test. P value for back belts vs. no back belts ¼ 0.008. b Comparison of pain at its worst—Wilcoxon rank sum test. P value for back belts vs. no back belts ¼ 0.004. Table 4 Duration of back pain after driving How long does the pain last

Age Control groupa

Back belt group

Area 1 vs. Area 3

3.4 (CI: 1.3–8.6) 7.9 (CI: 2.2–28.2) 0.2 (CI: 0.07–0.6) 0.6 (CI: 0.2–1.3) 1.1 (CI: 0.47–2.7) 0.97 (CI: 0.9–1.0)

a

Dependent variable: low back pain (5-point scale). Independent variables entered in model: work area, gardening, watching TV, speed of driving, frequency of breaks, vibration exposure, and age. c Proxy of estimated cumulative exposure to WBV (length of service multiplied by proxy vibration value for each area). High ¼ X23 and low p22. b

Several hours Several days Several weeks Several months Always present

(n ¼ 145) (%)

(n ¼ 31) (%)

35 37 4 3 21

57 27 3 0 13

a

Comparing distribution of back pain duration between controls and back belt group; p ¼ 0.03 on Fishers Exact test.

Table 5 Chronicity: variability of back pain as related to the pain duration Pain duration

Acute/ variable pain (%)

Constant/chronic pain (%)

Several hours (n ¼ 68) Several days (n ¼ 61) Several weeks or months or always present (n ¼ 44)

76 64 43

24 36 57

between the groups (p ¼ 0.4; ANOVA used to compare means) with the back belt group reporting a mean of 9.05 years (SD ¼ 6.8), and the non-belt wearers reporting pain onset after 9.6 years (SD ¼ 8.2). There was a significant difference in the intensity of lower back pain levels between the intervention and control groups for pain at its least (p ¼ 0.008; Wilcoxon rank sum test) and for pain at its worst (p ¼ 0.004 Wilcoxon rank sum test). The back belt group experienced pain that was less severe than the non-belt wearers. Thus although the back belt group experienced more back pain (Table 2), the pain at both its worst and least, was reported as less severe than that reported by the controls (Table 3). More drivers reported pain classified as variable (56%) than constant pain (44%) based on a intensity rating of pain at its worst compared to pain at its least. There was no significant

difference in this pattern between back belt and control groups (p ¼ 0.85; chi-square test). However, the back belt group showed pain of a significantly longer duration than controls (Table 4). The distribution of variable vs. constant pain was associated with reported pain duration in the direction expected (Table 5). When the duration of pain between the two classifications (variable and constant) was compared it was shown that the percentage of pain that was reported as constant increased as the duration of pain increased, 24% for pain lasting several hours, 36% for pain lasting several days and 57% for pain lasting weeks or longer. For variable pain the opposite was seen with most drivers (76%) having pain lasting several hours, 64% for several days and 43% several weeks or longer. This suggests that the biaxial classification by variability and duration may be a useful measure to supplement the characterisation of chronicity of back pain. Multivariate logistic modelling of back pain after driving identified work area (OR ¼ 3.4; 95% CI 1.3–8.6) and higher frequency of breaks (OR ¼ 0.2; 95% CI ¼ 0.07–0.6) as significant associations with back pain after driving. There were no associations with age, vibration exposure and sedentary activities (as measured by watching of TV) (Table 6). Alternative models using other outcomes such as back pain ever, or back pain in the past year gave similar results for significant predictors, both in direction of association and strength of association. Restricting analyses to the group of back belt users alone, (i.e. excluding the control group) and basing exposure to back belts on self-reported frequency of use showed that drivers who wore the back belts all of the time had an odds ratio of 2.3

ARTICLE IN PRESS 510

D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513

Table 7 Model fitting: predictors of back paina after driving amongst drivers in back belt area only (n ¼ 149) Variablesb

Authors

WBV source

Setting

Mean vibration levels (m s2)

Forklifts

(CI: 1.6–9.7) 0.2

Bovenzi et al. (2002)

Harbour companies Trans shipment company Urban bus drivers Port setting

0.8

2.3 (CI: 0.8–7.4) 4.0

Boshuizen et al. (1990) Bongers et al. (1990) Bovenzi (1996)

0.22–0.90

Port authority

1.1–1.9

OR (95% CI)

Constant Frequency of use

All the time vs. never Sometimes/rarely vs. never

Breaks

Other more frequent breaks vs. every 4–6 h

Watching TV

Yes vs. no

Vibration exposurec

High vs. low

Age

Table 8 Whole-body vibration levels reported in occupational studies

(CI: 0.06–0.6) 0.5 (CI: 0.2–1.1) 1.2 (CI: 0.5–2.8) 0.96 (CI: 0.9–1.0)

a

Dependent variable: low back pain (5-point scale). Independent variables entered in model: frequency of use (excludes non belt wearers and control group), gardening, watching TV, driving speeds, frequency of breaks, vibration exposure and age. c Proxy of estimated cumulative exposure to WBV (length of service multiplied by proxy vibration value for each area). High ¼ X23 and lowp22. b

for back pain after driving (non-significant, 95% CI 0.8–7.4) compared to drivers that never wore the back belts; and drivers who wore back belts sometimes had a significantly increased risk of back pain after driving (OR: 4.0; 95% CI 1.6–9.7). The dose–response effect was inconsistent, with the highest risk in drivers using the back belts sometimes (Table 7). Alternative models that substituted vibration levels for work area, driving speeds, and duration of pain, respectively, in place of back pain did not produce improved models as judged from the likelihood ratio chi-square. Subanalysis was attempted on drivers recently employed with less than 1 year of employment to ascertain the presence of any potential healthy worker effects. However, due to the small numbers involved (n ¼ 23 in the intervention group and n ¼ 9 in the control group) these analyses did not yield any meaningful results. Analyses substituting compliance (yes/no) for frequency of use and years of driving for estimated cumulative vibration exposure yielded broadly similar results to those presented in Tables 6 and 7. 4. Discussion The port site in Durban, South Africa, is one where forklift drivers generally are exposed to high levels of WBV consistent with or higher than levels found in other studies involving a range of vehicles (Table 8). Measured WBV levels at the port far exceeded the European Union Machinery Directive (Council of the European Union, 2002) requirements for vibration in the

Joubert (2000)

Wheelloaders Buses Straddle carriers, forklift trucks and cranes Forklifts

1.4 0.24–0.71

majority of empirical measurements under local conditions. It is therefore not surprising that a large percentage (89%) of the drivers in all three areas had experienced back pain at some stage in their lives. These results are consistent with findings from various other studies where equivalent rates have been reported as 79% (Brendstrup and BieringSorensen, 1987), 90% (Riihimaki et al., 1989), 68% (Bongers and Boshuizen, 1990), 51% (Bongers et al., 1990), and slightly over 80% by Bovenzi (1996). Back pain in the past year (89%) was also broadly similar to levels reported in the literature. Point prevalence rates (pain today) in this study were slightly lower than would have been expected from the literature but may have been confounded by variable times of administration of the questionnaire in relation to commencement/completion of a work shift. No data were collected on time of questionnaire administration, which would have enabled analysis for confounding by the time of day of interviewing. The control group was interviewed prior to the shift and exposure to WBV. For this reason, the use of lower back pain outcomes other than those based on point prevalence is preferable in interpreting these data. Consistency between findings for lower back pain ever, lower back pain in the past year, and lower back pain after driving adds credibility to this assumption. For all back pain outcomes, back belt usage was not associated with a decrease in back pain. Indeed, back belt usage was associated with a statistically significant increase in the prevalence of some back pain outcomes (ever and after driving), even when controlling for breaks taken during the working shift, watching television, WBV exposure and age. The possibility of confounding in the relationship between back belt usage and back pain must be seriously considered. Confounding by manual lifting is unlikely to have been a factor because other workers at the port carry out this task separately, and there is no evidence that this practice differs by study area. Confounding, by WBV exposure which may have arisen with differentially increased exposure by area, is an important consideration for interpreting the results. However, the extent of confounding by WBV would have had to

ARTICLE IN PRESS D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513

be substantial to counter the positive association (Table 7) between back belt usage and back pain after driving. Moreover, when compliance with belt use was examined amongst those in the back belt group in the multivariate model, thereby controlling for area-adjusted WBV exposure, it was found that users were still more likely to report back pain than non-users, controlled for frequency of rest breaks taken during the working shift, watching television, gardening speed of driving, and length of driving, and that frequency of use of back belts bore no dose response relationship to back pain prevalence. The models for all the outcomes gave similar predictors. At the very least, no protective effect of back belts is discernible for LBP as an outcome and back belt usage appears to be associated with an increased risk of low back pain. However, this finding should be interpreted with caution due to the limitations of the cross-sectional study design, lack of individual WBV cumulative exposure levels and potential misclassification of WBV exposure; hence additional prospective studies would be warranted to be able to generate more definitive findings. Selection or information bias may explain the statistical association of back pain and back belt usage. The back belt group could have had more pain to start with than the control group and thus more reason to use back belts in an attempt to resolve their problem. Alternatively, the issuing of back belts could have drawn the attention of the drivers who wore belts to their back problems, leading them to report more back pain than controls. Thirdly, the psychological effect of back belt use on the reporting of pain cannot be discounted. The Healthy Worker Effect was unlikely to play any significant role in this study as drivers tended to stay in their job for long periods of time and driver turnover rates were low. Odds ratios for taking medication (OR ¼ 1.5; 95% CI 0.7–3.3) and for work absenteeism related to back pain (OR ¼ 1.5; 95% CI 0.5–4.8) were both mildly and nonsignificantly elevated, representing weak associations that are consistent with a psychological basis for the back belt—back pain association. However, without base-line data and a follow up experimental study design using randomisation, the presence of a selection effect due to non-random assignment of the back belt condition cannot be adequately estimated or controlled. All these explanations are therefore speculative and cannot be adequately tested within the inherent limitations of a cross-sectional design. Nonetheless, compliance with the use of the back belts, frequency of use, and drivers’ opinions as to the effectiveness of the belts in reducing lower back pain was high (90%; 76% and 81%, respectively). The high opinion of effectiveness is consistent with the high compliance and frequency of use, as a positive attitude and opinion would increase the use of the devices. Some bias of workers towards the use of the back belts would be expected, as they had requested them through their trade unions originally and would thus have a positive attitude about their effectiveness. Drivers who took rest breaks more frequently then the official breaks were less likely to report pain after driving

511

Table 9 Chronicity classification by duration and variability (n ¼ 174) Pain duration

Weeks or more Days or less

Pain variability Constant

Variable

Chronic pain: 11% Acute pain: 53%

Chronic pain with exacerbations: 14% Acute pain with fluctuations: 22%

(OR ¼ 0.2; 95% CI 0.07–0.6), a finding that could possibly be explained by increased time for rest and recuperation of the back. If this result represents a valid association, control programs to reduce back pain amongst forklift drivers could benefit from building in regular and frequent rest periods. The watching of television as a sedentary activity would be expected to increase risk for back pain yet; the model suggested that this activity reduced the risk of back pain after driving. However, this may be a spurious association based on confounding by socio-economic status, since workers owning a television may have a higher socioeconomic status than those without (Latza et al., 2000). In this study, the use of differences in pain scores at its worst and at its best to classify back pain into variable vs. constant pain was strongly correlated with responses on the duration of pain. This approach (Table 9), helped to distinguish differences in acute, acute with fluctuations, chronic and chronic pain with exacerbations. Most drivers at the port experienced pain for several hours to several days (74%) slightly higher than in the study conducted by Bongers et al. (1990), which recorded 68.9%. However, almost three times as many drivers (17%) reported pain was always present, compared to the 6.8% reported by Bongers et al. (1990). This may be attributed to a number of possible factors such as the condition of the vehicles and the lack of vibration protection offered by the seats of the majority of forklifts at the port, or possibly due to the extended shifts that drivers at the port are expected to drive. This association is, however, unclear. There was no significant pattern that emerged when the intensity of pain was compared to years of driving, with the pain staying relatively constant over the exposure periods. There is thus no indication that the intensity of back pain follows a linear dose–response curve. 4.1. Intensity and location of back pain Notably, the back belt group reported experiencing less severe pain than the control group, despite having higher prevalence’s of LBP. These results may have been influenced by the psychological effect of the belts given that other specific indicators of the presence of pain such as consulting a doctor (p ¼ 0.9), taking medication (p ¼ 0.15) or taking time off (p ¼ 0.15) were not statistically associated with back belt usage. The location of back pain as predominantly low back pain was consistent in both groups (83% vs. 75%). The

ARTICLE IN PRESS D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513

512

two groups did differ significantly with regards to shoulder pain experienced (11% in back belt users vs. 38% in controls), a finding perhaps explained by the high prevalence of lower back pain amongst belt users masking pain experienced and reported in other parts of the back. Similar psychological differences may therefore underlie reported differences in pain duration and intensity, since it is difficult to identify any biologically plausible explanation for these associations.

per the example (Back belt area 1) shown below:

5. Conclusion

Predicted vibration value ¼ Antilog ð0:6456Þ

Back belt area 1: Log ðpredicted vibration levels in ms2 Þ ¼ 0:60 þ 0:56 ðBack belt area 1Þ þ 0:49 ðrough surfaceÞ þ 0:0 ðunadjusted seatÞ þ 0:002  97:8 ðSEAT % valueÞ. ¼ 0:6456.

¼ 1:9 m s2 . Based on a consideration of the Bradford Hill Criteria for causal inference in epidemiology (Bradford-Hill, 1996), evidence for the effectiveness of back belts in reducing the prevalence and intensity of back pain in this study is lacking, and back belt usage may well increase back pain amongst users of these devices.

Results from the other two areas were: Back belt area 2 : Predicted vibration value ¼ 1:3 m s2 . Control area : predicted vibration value ¼ 1:1 m s2 .

This study was made possible by funding from the National Research Foundation (South Africa) and the Ernest Oppenheimer Memorial Trust. A special thanks goes to Cathy Connolly for statistical assistance, as well as Stephanie Samuels, Sihle Khumalo, Sean Du Plessis and the management and drivers at Portnet, Durban who facilitated and participated in this study.

Estimated cumulative exposure to WBV was calculated as the product of the predicted vibration level per area and the number of years driving for at the port (ms2—years). This was used as a proxy measure of vibration cumulative exposure since individual driver exposure data were not available. The cumulative WBV exposure variable had a bimodal distribution and was categorised as low (22 m s2— years) and high (23 m s2—years) across the median value. As expected, this proxy measure was highly correlated with years driving (r ¼ 0.92) and also associated with work area (p ¼ 0.02).

Appendix A

References

Table showing whole-body vibration root mean squared values (m s2) for combined driving surface and combined seat adjustment results for individual test areas.

Bogadi-Sare, A., 1993. The effect of whole-body vibration: an unrecognized medical problem. Arhiv za Higijenu Rada i Toksikologiju 44, 269–279. Bongers, P., Boshuizen, H., 1990. Back disorders and whole-body vibration at work. Offsetdrukkerij Kanters bv, Alblasserdam, University of Amsterdam. Bongers, P., Boshuizen, H., Hulshof, C., 1990. Self reported back pain in drivers of wheelloaders. Offsetdrukkerij Kanters bv, Alblasserdam, University of Amsterdam, pp. 207–220. Bovenzi, M., 1996. Low back pain disorders and exposure to whole-body vibration in the workplace. Seminars in Perinatology 20, 38–53. Bradford-Hill, A., 1996. The environment and disease: association or causation? Proceedings of the Royal Society of Medicine 58, 295. Brendstrup, T., Biering-Sorensen, F., 1987. Effect of fork-lift truck driving on low-back trouble. Scandinavian Journal of Work Environment and Health 13, 445–452. Canavan, T., 1999. Product Review. Kevco/Stubbs Kidney Belt, California, USA Available from: URL: /http://cgi.motorcycle.com/ mo/mcdirt/stubbs.htmlS (Online, accessed 21-04-99). Ciriello, V., Snook, S., 1995. The effect of back belts on lumbar muscle fatigue. Spine 20, 1271–1278. Council of the European Union, 2002. The minimum health and safety requirements regarding the exposure of workers to the risks arising from physical agents (vibration). Official Journal of the European Communities (article 16 of the directive 2002: 89/391/EEC). Dinardi, S., 1997. The Occupational Environment—its Evaluation and Control. American Industrial Hygiene Association, AIHA Press, Fairfax, Virginia. Futatsuka, M., Maeda, S., Inaoka, T., Nagano, M., Shono, M., Miyakita, T., 1998. Whole-body vibration and health effects in agricultural machinery drivers. Industrial Health 36, 127–132.

Acknowledgements

Test condition

Work area Vibration RMS value (m s2)

Driving surface

Seat adjustment

Mean SD

Min

Max

Rough

Adjusted

1.47 1.42 0.80 2.07 1.45 1.05 1.11 0.60 0.61 1.32 0.72 0.72

1.18 1.33 0.78 1.17 1.36 0.74 0.82 0.48 0.58 0.82 0.47 0.57

1.91 1.51 0.82 3.05 1.54 1.36 1.45 0.75 0.64 2.04 0.97 0.84

Smooth

1 2 Control Unadjusted 1 2 Control Adjusted 1 2 Control Unadjusted 1 2 Control

0.39 0.13 0.23 1.03 0.13 0.31 0.32 0.19 0.04 0.60 0.35 0.14

Appendix B. Predicted vibration levels This was carried out for each of the three areas from all vibration results under different operational conditions, as

ARTICLE IN PRESS D.M. Joubert, L. London / International Journal of Industrial Ergonomics 37 (2007) 505–513 Harman, E.A., Rosenstein, R.M., Frykman, P.N., Nigro, G.A., 1989. Effects of a belt on intra-abdominal pressure during weight lifting. Medicine and Science in Sports Exercise 21, 186–190. Hulshof, C., Van Der Laan, G., Braam, I., Verbeek, J., 2002. The fate of Mrs. Robinson: criteria for recognition of whole-body vibration injury as an occupational disease. Journal of Sound and Vibration 253, 185–194. Hunter, G., McGuirk, J., Mitrano, N., Pearman, P.B.T., Arrington, R., 1983. The effects of a weight training belt on blood pressure during exercise. Journal of Applied Sport Science Research 3 (1), 13–19. International Organisation for Standardisation, 1997. Mechanical vibration and shock-evaluation of human exposure to whole-body vibration. Part 1: general requirements, Geneva, Switzerland. Jensen, M., Karoly, P., Braver, S., 1986. The measurement of clinical pain intensity: a comparison of six methods. Pain 27, 117–126. Joubert, D., 2000. Risk factors and the effectiveness of back belts in the prevention of back pain amongst forklift drivers subject to whole-body vibration exposure. Masters thesis, University of Cape Town, South Africa. Kevco/Stubbs, 1999. The Press, California, USA. Available from: URL: /http://www.kevco-stubbs.com/press.htmlS (Online, accessed 16-04-99). Kraus, J., Brown, K., McArthur, D., Peek-Asa, C., Samaniego, L., Kraus, C., 1996. Reduction of acute low back injuries by use of back supports. International Journal of Occupational and Environmental Health 2 (3). Kraus, J.F., Schaffer, K.B., Rice, T., Maroosis, J., Harper, J., 2002. A field trial of back belts to reduce the incidence of acute low back injuries in New York City home attendants. International Journal of Occupational and Environmental Health 8, 97–104. Kuorinka, I., Jonsson, B., Kilborn, A., Vinterburg, H., Biering-Sorenson, F., Andersson, G., Jorgensen, K., 1987. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Applied Ergonomics 18, 233–237. Lander, J.E., Hundley, J.R., Simonton, R.L., 1992. The effectiveness of weight-belts during multiple repetitions of the squat exercise. Medicine and Science in Sports Exercise 24, 603–609. Latza, U., Kohlmann, T., Deck, R., Raspe, H., 2000. Influence of occupational factors on the relation between socioeconomic status and self reported back pain in a population based sample of German adults with back pain. Spine 25 (11), 1390–1397. Malchaire, J., Piette, A., Mullier, I., 1996. Vibration on forklift trucks. Annals of Occupational Hygiene 40, 79–91. McGill, S.M., Norman, R.W., 1987. Reassessment of the role of intraabdominal pressure in spinal compression. Ergonomics 30, 1565–1588. McGill, S.M., Norman, R.W., Sharratt, M.T., 1990. The effect of an abdominal belt on trunk muscle activity and intra-abdominal pressure during squat lifts. Ergonomics 33, 147–160. Mitchell, L.V., Lawler, F.H., Bowen, D., Mote, W., Asundi, P., Purswell, J., 1994. Effectiveness and cost-effectiveness of employer-issued back belts in areas of high risk for back injury. Journal of Occupational Medicine 36, 90–94.

513

Nachemson, A., Andersson, G., Schultz, A., 1986. Valsalava manoeuvre biomechanics: effects on lumbar trunk loads of elevated intraabdominal pressures. Spine 11 (5), 476–479. National Occupational Health and Safety Commission, 2001. Work related aspects of patient presentations to general practitioners in Australia: analysis of the first two years of BEACH data 2001, Canberra, Australia. Pope, M., Novotny, J., 1993. Spinal biomechanics. Journal of Biomechanical Engineering 115, 569–574. Pope, M., Wilder, D., Jorneus, L., Broman, H., Svensson, M., Andersson, G., 1987. The responses of the seated human to sinusoidal vibration and impact. Journal of Biomechanical Engineering 109, 279–284. Rafacz, W., McGill, S., 1996. Abdominal belts increase diastolic blood pressure. Journal of Occupational and Environmental Medicine 38 (9), 925–927. Reddell, C.R., Congleton, J.J., Dale Huchingson, R., Montgomery, J.F., 1992. An evaluation of a weightlifting belt and back injury prevention training class for airline baggage handlers. Applied Ergonomics 23, 319–329. Riihimaki, H., Tola, S., Videman, T., Hanninen, K., 1989. Low-back pain and occupation. A cross-sectional questionnaire study of men in machine operating, dynamic physical work, and sedentary work. Spine 14, 204–209. Rosenstock, L., 1997. Report of Written Testimony from the Director National Institute for Occupational Safety and Health to the SubCommittee on Workforce Protection 1997. Available from: URL: /http://www.cdc.gov.niosh/nioshfin.htmlS (Online, accessed 21-041999). South African Bureau of Standards, 1990. General requirements for the competence of calibration and testing laboratories. Code of Practice, SABS 0259:1990, Pretoria. Thompson, L., Davidson, H., Hirsh, D., 1994. Attitudes and back belts in the workplace. Work 4 (1), 22–27. van Poppel, M.N., Koes, B.W., van der Ploeg, T., Smid, T., Bouter, L.M., 1998. Lumbar supports and education for the prevention of low back pain in industry: a randomized controlled trial. Journal of American Medical Association 279, 1789–1794. Vibration Injury Network, 2001. Guidelines and questionnaires for wholebody vibration health surveillance. EC Biomed II concerted action BMH4-CT98-3251 2001. Appendix W1A to final report. Von Korff, M., Ormel, J., Keefe, F.J., Dworkin, S.F., 1992. Grading the severity of chronic pain. Pain 50, 133–149. Walsh, N., Schwartz, R., 1990. The influence of prophylactic orthoses on abdominal strength and low back injury in the workplace. American Journal of Physical Medicine & Rehabilitation 69, 245–250. Wikstrom, B., 1993. Effects from twisted postures and whole-body vibration during driving. International Journal of Industrial Ergonomics 12, 61–75. Wilder, D.G., Pope, M.H., Magnusson, M., 1996. Mechanical stress reduction during seated jolt/vibration exposure. Seminars in Perinatology 20, 54–60. World Health Organisation, 1999. The Burden of Occupational Illness: UN agencies sound the alarm. Press release WHO/31, Geneva, Switzerland 1999. Available from: URL: /http://www.who.int/ inf-pr-1999/en/pr99-31.htmlS (Online, accessed 28-02-2003).