Applied Ergonomics 32 (2001) 559571

Musculoskeletal, visual and psychosocial stress in VDU operators before and after multidisciplinary ergonomic interventions. A 6 years prospective studyFPart II
( sa,*, Gunnar Horgenb, Hans-Henrik Bjrsetc, Ola Rod, Heidi Walsee Arne Aara
b

Alcatel STK A/S, PO Box 60, kern, N-0508 Oslo 5, Norway Kongsberg College of Engineering, Department of Optometry, P.O. Box 235, 3601 Kongsberg, Norway c K. O. Thornesveg 11, 7033 Trondheim, Norway d Premed A/S, P.O. Box 275, kern, 0511 Oslo, Norway e Parexel Medstat, P.O. Box 210, N-2001 Lillestrm, Norway Accepted 22 May 2001

Abstract A prospective epidemiological eld study covering a 2 years period has earlier been published (Appl. Ergon. (1998) 29(5) 335). The study has a parallel group design with two intervention groups (T and S) and one control group (C) of Visual Display Unit (VDU) operators. The present paper covers the period from 2 to 6 years of the study. After 3.5 years, the C group got the same intervention in terms of new lighting system, new workplaces and at last an optometric examination and corrections if needed. The C group reported a signicant reduction in visual discomfort after interventions while the two groups (T and S) continued to report signicant reduction of visual discomfort after 6 years. By supporting the forearm on the table top, the C group reported signicant reduction of shoulder and neck pain while the T group reported signicant reduction in shoulder and back pain after 6 years. Organizational and psychosocial factors at work and outside work did not show any signicant changes during the study period. r 2001 Elsevier Science Ltd. All rights reserved.
Keywords: VDU workplaces; Lighting conditions; Visual conditions; Visual discomfort; Musculoskeletal illness

1. Introduction Eye discomfort and musculoskeletal illness are the main problems reported by visual display unit (VDU) operators. Eye discomfort is shown to be connected to ( s et al., 1998). VDU work (Bergqvist et al., 1992; Aara Further, the prevalence of musculoskeletal illness is found to be higher in VDU work compared with nonVDU work (Punnett and Bergqvist, 1997). Important factors for designing the lighting systems and the workplace, as well as procedure for optometric correc( s et al. (2000). tions of VDU workers are given by Aara In addition, the paper gives a review of the international literature regarding health consequences for VDU workers. This review covers lighting, optometry, and factors related to musculoskeletal illness for VDU
*Corresponding author. Tel.: +47-22638816; fax: +47-22638944. ( s). E-mail address: arne.aaraas@alcatel.no (A. Aara

workers. Punnett and Bergqvist (1997), in their review of epidemiological studies of VDU work, found that VDU work indicated higher risk of neck, shoulder, arm, wrist and hand musculoskeletal illness compared with non-VDU work. In their extensive review only nine intervention studies are reported until 1997. Eight of these studies suggested that a decrease in musculoskeletal illness could be achieved by ergonomic intervention programmes. The follow up time varies for most of the studies between 6 and 18 months. Ong (1984) reported that the prevalence of symptoms in neck, arms and hands decreased by about 67% after giving data entry VDU workers, ergonomic workstation adjustments, longer lunch break, improvements in noise and illumination, as well as improved thermal environment. Kukkonen et al. (1983) combined workstation improvements, training and relaxation exercises for 60 data entry operators. Six months follow up showed that the intervention group had fewer symptoms and physical

0003-6870/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 3 - 6 8 7 0 ( 0 1 ) 0 0 0 3 0 - 8

560

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

ndings than a reference data entry group. At commencement the intervention group had more upper extremity disorders than the reference group. These results were supported by a larger study by Oxenburgh (1985). Shute and Starr (1984) introduced adjustable workstation and chairs to 66 data entry workers. After 58 weeks, the operators reported signicant decrease in the incidence of neck, shoulder and wrist/ hand symptoms. Training the VDU workers to modify the workstation layout as well as taking other measures to improve their working conditions are factors reported to give less pain in two intervention groups compared to the control group both after 4 weeks and after 6 months (Kamwendo and Linton, 1991). There is still need for prospective multidisciplinary intervention studies with long follow up time in order to evaluate the health consequences of VDU work. The background for this study was that the VDU workers reported both visual problems and musculoskeletal discomfort. The interventions were based on the following studies: 1. Laboratory studies had been carried out in order to nd criteria for luminaires and their optimal position relative to the screen (Bjrset, 1986). Luminaires giving both direct and indirect lighting, with one luminaire on each side of the screen, were found to give the best solution regarding visual condition. 2. Dierent types of lenses were studied in the laboratory regarding postural load by Horgen et al. (1989, 1995). Single vision lenses were found to create less muscle load compared with progressive lenses. 3. A laboratory study of muscle load in dierent work positions showed that supporting the forearm on the table top reduced the muscle load of trapezius and erector spina lumbalis (L3 level) for sitting VDU ( s et al., 1997). work (Aara Following these laboratory results a multidisciplinary team was established for the intervention study. Three serial interventions were carried out in two intervention groups (T and S) while one group acted as a control ( s et al., 1998). The interventions group (C), (Aara consisted of the following: rst a new lighting system, then new workplaces and last an optometric intervention. The new lighting system for localized lighting increased maintained illuminance levels from about 300 lux to above 600 lux, increased luminances of the room surfaces from about 30 cd/m2 to about 80 cd/m2, and reduced glare problems. The rst part of the study covered a follow up period of 2 years. The two intervention groups (T and S) reported signicant improvement of the lighting conditions, as well as of the visual conditions, and signcantly reduced visual discomfort and glare. No signicant change was reported in the C group. The optometric intervention

with single vision lenses which were given according to actual visual distances for VDU work, further reduced the visual discomfort while no change was observed in the C group. By redesigning the workplaces, and allowing the operators to support their forearms on the table top for the T and S groups, a signicant reduction of shoulder pain was reported in the S group and a clear tendency was observed in the T group, while no such change was reported in the C group. After two years, signicant dierences in shoulder pain were found between the two former intervention groups T and S and the C group. The second part of this study describes the results reported by the C group when this group got the same interventions as the T and S groups.

2. The aims of the study Will interventions in the C group, consisting of improved lighting, improved workplaces and optometric corrections, inuence the visual discomfort, headache and musculoskeletal pain? In particular,
*

will changing the lighting condition and giving optometric corrections reduce visual discomfort and headache for the VDU operators? will an opportunity to support the forearms on the table top in front of the operator lead to reduced pain in the musculoskeletal system? will the eect reported by the former intervention groups (T and S) after 2 years, still be present after 6 years?

3. Design of the study The study was performed as a prospective, parallel group design. Approximately 50 male subjects partici( s et al., 1998). pated in each of the three groups (Aara The main task in each group was software engineering. Two groups (T and S) were given interventions in terms of new lighting, new workplaces, and optometric corrections, when required. The C group acted as a control group for 3.5 years. That means that the C group continued with the initial lighting system, initial workplaces, and were not given optometric intervention. Points of time for start of the three interventions in the T ( s et al. (1998, Table 2). and S groups are given by Aara The three interventions were, for each participant in the T and S groups, at least 6 months apart. This allowed recording of visual discomfort and pain levels also for an observation period of 6 months. The C group got the same lighting and work place interventions as the T and S groups approximately 3 months before the 4th measurement (3.5 years after the start of the

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

561

project). The implementation of the lighting and workplaces took place simultaneously over a period of about one year. Six months after these interventions, the optometric intervention was performed for those who needed such corrections in the C group. The 5th measurement of the three groups was taken 6 years after the start of the study. At that time a cross-sectional investigation was performed regarding the relationship between visual discomfort and the operators distance from the window. Subjects from all the three groups were included.

plete sample at measurement 5 and one for the complete sample (including the drop-outs). The two statistical analyses were compared for possible systematic biases regarding health parameters. 4.3. Statistical methods Continuously distributed location parameters are presented by both median and means with their 95% condence interval. The standard deviation and range are used as estimates for dispersion. Counts are presented in contingency tables. Continuous variable tests for changes within groups are made by Wilcoxon test and comparisons between groups by Kruskal Wallis test. The chi-square test is used for counts. Possible time eects and interactions between time and intervention are tested by a repeated measurement analysis of variance with the baseline value as covariate. All tests are two-sided using the null hypothesis of equality. P-values less than 5% are regarded as statistically signcant and p-values between 5 and 10% are described as a tendency.

4. Methods and procedures 4.1. Questionnaires Questionnaires which dealt with headache, visual conditions and discomfort, musculoskeletal pain as well as organizational and psychosocial factors were lled in at each measurement (5 times). The factors were measured on a 10 cm visual analogue scale (VAS). A detailed description of the questionnaires and procedures of measurements regarding lighting, optometric examination and psychosocial questions is given by ( s et al. (1998). Many researchers have documented Aara that psychosocial factors inuence the musculoskeletal pain. An extensive review of psychosocial aspects of working with VDU and employees physical and mental health is given by Smith (1997). These variables were tracked to see if there was any change of the psychosocial factors during the study period. The psychosocial questionnaire contained factors at work and at home. The main questions dealt with the amount of VDU work, i.e. how often the VDU was used; total amount of time per day; length of periods without a break in front of the screen. An assessment was done regarding variation of the work as a whole, i.e. other work tasks were compared with VDU work. Job control and the opportunity to make contact with colleagues, self-realization in terms of learning, increased skills and utilization of own capability and satisfaction at work were assessed. Further, basic need satisfaction such as the time at own disposal and work burden at home were also considered. 4.2. Drop-out routine A separate statistical analysis was carried out to investigate possible systematic inuence on the results due to subjects who dropped out during the study period. The test was performed by creating a complete sample for measurement 5 by carrying the last observed value regarding health parameters before they dropped out, forward to measurement 5. Then two separate statistical analyses were performed, one for the incom-

5. Interventions 5.1. Lighting After 3.5 years from the start of the study, the C group got a new lighting system and new workplaces. The new lighting system applies suspended luminaires with a light distribution about 25% upwards and 75% downwards, through an eective, semidiused reector( s et al., 1998). The louvre system (Bjrset, 1997; Aara positioning of the luminaires was one luminaire at each side of the VDU workplace, Fig. 1. 5.2. Intervention of the workplaces ( s et al. The intervention is decribed in detail by Aara (1998). The main feature of the new workplaces was that the operators were allowed to support their forearms on the table top, Fig. 1. 5.3. Optometric intervention 5.3.1. Procedure About 4.5 years after start of the study, when the C group had got new lighting systems and new work places, all operators were given a complete optometric examination, and the corrections given were based on this examination. The examination focused on optimum visual acuity and comfort for the VDU-workers. The individual workplaces were measured for all the dierent viewing distances and the corrections were modied in order to t the visual distance of the work

562

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

6. Results The one month observation period of the end variables (i.e. the average discomfort and pain level) gave similar results as for the 6 months observation period. The gures in this paper give the results with the same observation time of the end ( s et al. (1998) variables as corresponding gures in Aara (1 and 6 months data). This is done for easier comparison of results between the two papers. The results before the interventions in the C group regard( s et al. (1998). ing Figs. 29 are given by Aara This paper will focus on the results in the period after intervention of the C group 3.5 years after the start of the study. 6.1. The lighting and visual conditions The average maintained illuminance of the relevant work areas was increased from about 300 lux to more than 600 lux. The luminance levels of the ceiling and walls were increased from approximately 30 cd/m2 to more than 80 cd/m2. After the C group got lighting and work place interventions, they reported signicant improvements of the lighting and visual conditions ( p 0:000), Fig. 2. At 6 years from the start of the study there are no longer signicant dierences between the three groups ( p 0:62). No signicant changes were reported regarding lighting and visual conditions in the period after optometry to 6 years follow up for the T and S groups. 6.2. The glare conditions The C group reported a signicant reduction of glare problems after lighting and work place interventions ( po0:000), Fig. 3. At the end of the study there are no signicant dierences between the three groups ( p 0:81). No signicant changes were reported regarding glare problems in the period after optometry to 6 years after start of the study for the T and S groups. 6.3. Optometry New corrections were given to 18 subjects in the C group, while 11 subjects did not need new optometric correction. In this part of the study, we did not analyze the changes in refraction as was done in the rst part of (s the study, more details on this are given by Aara et al. (1998). 6.4. Visual problems After lighting and optometric interventions in the C group, a signicant reduction in visual discomfort was

Fig. 1. The workplace with the new table/chair and the new luminaires.

tasks. Binocular problems were evaluated and corrected according to both the amount of xation disparity measured, and phoria measurements (Borish, 1975; Sheedy, 1995). 5.3.2. Criteria for optometric correction Criteria for prescribing optometric corrections are not easily given, and some controversy exists around this topic. The criteria used were based on clinical experience and earlier studies (Horgen et al., 1989, 1995; Methling, 1992). 5.3.3. Spectacle lenses The corrections should be worn while working on the VDU, and single vision lenses were the lenses of choice (Horgen et al., 1989, 1995). There exist several new designs of VDU-lenses (so-called VDU-progressives), which give a longer focusing range, but because of lack of scientic proof that these lenses do not cause increased postural load at the time of intervention, single vision lenses were chosen. Ordinary progressive lenses do intervene with body posture and postural load (Guillon, 1999; Horgen et al., 1989). The lenses recommended were white, organic lenses with antireection coating. Organic lenses were chosen because of lighter weight.

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

563

Fig. 2. Lighting conditions for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 is very bad lighting conditions while 100 is very good lighting conditions. A* indicates a signicant dierence when comparing with the column containing a+.

Fig. 4. Visual discomfort in the last month for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no visual discomfort while 100 is very severe discomfort. A* indicates a signicant dierence when comparing with the column containing a+. A (*) is very close to a signicant dierence.

Fig. 3. Glare conditions for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 is very bad glare condition while 100 indicates no glare. A* indicates a signicant dierence when comparing with the column containing a+.

Fig. 5. Visual discomfort in the last 6 months for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no visual discomfort while 100 is very severe discomfort. A* indicates a signicant dierence when comparing with the column containing a+. A (*) is very close to a signicant dierence.

reported, comparing the results at the 6 years follow up with after optometry in the T and S groups ( p 0:03). This was true for the observation period of one month, Fig. 4. These results are based on analysis of all participants in the C group. No signicant dierences were found between the groups at 6 years follow up ( p 0:1). When dividing the C group, there was a tendency to reduction in

visual discomfort in the last month for those who got new corrections (8,9 mm on a 100 mm VAS as mean group value, p 0:08) while the group who did not get new correction did not report signicant changes (3.5 mm on the same VAS as mean group value, p 0:48). In the T and S groups no signicant changes

564

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

Fig. 6. Headache in the last month for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no headache while 100 is very severe headache. A* indicates a signicant dierence when comparing with the column containing a+.

Fig. 8. Intensity of shoulder pain in the last 6 months for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no pain while 100 is very severe pain. A* indicates a signicant dierence when comparing with the column containing a+. A (*) is very close to a signicant dierence.

Fig. 7. Intensity of neck pain in the last 6 months for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no pain while 100 is very severe pain. A* indicates a signicant dierence when comparing with the column containing a+. A (*) is very close to a signicant dierence.

Fig. 9. Intensity of pain in the forearm and hand in the last 6 months for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no pain while 100 is very severe pain. A* indicates a signicant dierence when comparing with the column containing a+.

in the visual discomfort in the last month were reported in the period from after optometry to 6 years follow up. The results for visual discomfort in the last 6 months (Fig. 5) are similar to the results for the last month (Fig. 4), except for a signicant reduction in the T group after optometry to 6 years follow up ( p 0:04), Fig. 5.

6.5. Dierent types of eye symptoms The results of the eye symptoms analysis are given in Tables 1, 2 and 3. Feeling of tired eyes was signicantly reduced in the T group ( p 0:000) and a clear tendency to reduction of the same symptom was found in the S

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara Table 1 Eye symptoms of the T group Symptoms Before interventions Mean Feeling of tired eyes Stinging or itching and irritation Sensitivity to light Redness of the eyes Gravelly sensation of the eyes Blurred or double vision 40.2 26.5 30.9 21.7 14.4 17.0 95% C.I. 34.845.6 20.832.2 25.036.8 16.527.0 9.818.9 11.822.2 After 6 years Mean 26.8 17.5 21.9 13.3 14.3 16.2 95% C.I. 21.032.6 12.122.8 15.528.3 9.117.6 9.419.1 9.922.4

565

P-value

0.000 0.017 0.002 0.152 0.404 0.494

Table 2 Eye symptoms of the S group Symptoms Before interventions Mean Feeling of tired eyes Stinging or itching and irritation Sensitivity to light Redness of the eyes Gravelly sensation of the eyes Blurred or double vision 37.0 28.2 31.0 18.0 15.5 16.4 95% C.I. 29.844.3 21.135.3 22.539.4 11.324.7 9.221.7 9.723.0 After 6 years Mean 25.9 13.6 20.8 14.1 10.1 12.9 95% C.I. 19.132.8 8.418.9 13.827.8 8.419.7 5.914.4 7.817.9 0.065 0.007 0.046 0.576 0.807 0.909 P-value

Table 3 Eye symptoms of the C group Symptoms Before interventions in the C group Mean Feeling of tired eyes Stinging or itching and irritation Sensitivity to light Redness of the eyes Gravelly sensation of the eyes Blurred or double vision 33.4 25.5 30.0 23.1 22.5 18.1 95% C.I. 26.640.2 17.733.3 21.838.2 16.929.4 15.629.3 11.225.0 After 6 years Mean 31.1 18.3 21.5 15.9 16.6 15.4 95% C.I. 22.040.1 10.426.2 13.229.9 9.821.9 8.225.0 7.523.3 0.527 0.125 0.708 0.009 0.531 0.981 P-value

group ( p 0:065), while no such reduction was found in the C group ( p 0:53) when comparing before interventions with 6 years follow up. Signicant reduction for stinging or itching and irritation of the eyes were also found in the T group ( p 0:007) and in S group ( p 0:007) while the C group reported no signicant reduction ( p 0:125). Sensitivity to light was found signicantly reduced in both the T ( p 0:002) and S group ( p 0:046), but not in the C group (0.71). The C group reported a signicant reduction in redness of the eyes ( p 0:009) while no signicant changes were observed in the T and S groups. No signicant changes were reported regarding gravelly sensation, blurred or double vision in any of the groups.

6.6. Relationship between the distance from the window and visual discomfort A cross-sectional investigation of the relationship between visual discomfort and distance from the window was performed. This was done on the data from the 6 years follow up measurement. When using a sightline to the screen parallel to the window, there is a clear reduction in visual discomfort with increasing distance from the window. A comparison was done by constructing two groups, each consisting of 32 subjects. One group was doing VDU work closer than 1.5 m from the window (2 subjects less than 1 m, 30 subjects between 1 and 1.5 m) and the other group more the

566

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

1.5 m away from the window (28 subjects between 1.5 and 2 m, 4 subjects more than 2 m). A signicantly lower intensity of visual discomfort was reported in the latter group, 10.8 (SD, 18.5), (>1.5 m) versus 22.5 (SD, 24.5) for the other group, (o1.5 m), p 0:03: For the VDU workers having their sightline turned towards window, no relationship was found between visual discomfort and the distance from the window. 6.7. Headache The C group reported no signicant change regarding headache in the last month in the period from after optometry in the T and S groups to 6 years follow up ( p 0:44) or during the whole study period ( p 0:50), Fig. 6. After 6 years follow up the T group reported signicant reduction of headache compared with the start of the study ( p 0:003) while no signicant changes were observed in the S group ( p 0:78). No signicant dierences are found between the three groups after 6 years follow up ( p 0:176). Similar results were found for the headache in the last 6 months. 6.8. Neck pain The average intensity of neck pain during the last 6 months at the ve measurements in the study period are shown in Fig. 7. When comparing the same subjects within the three groups regarding intensity of neck pain in the period from 6 years follow up with after optometry in the T and S groups, no signicant change was reported in the C group ( p 0:16), T group ( p 0:49) or S group ( p 0:79). Comparing at 6 years follow up with the start of the study, there was a tendency to reduction in the C group ( p 0:07) while no signicant changes were reported in the T ( p 0:25) and S group ( p 0:62). Similar results were found for neck pain last month except that the C group reported a signicant reduction when considering the whole study period. 6.9. Shoulder pain Comparing the same subjects regarding intensity of shoulder pain in the last 6 months in the period from 6 years follow up with after optometry in the T and S groups, a signicant reduction was found in the C group ( p 0:03) while no signicant changes were reported in the T group ( p 0:76) or S group ( p 0:32). Considering the period from start of the study to 6 years follow up, a signicant reduction was reported in the C group ( p 0:004) and the T group ( p 0:02), while the S group did not report signicant reduction ( p 0:66), Fig. 8. Similar results were found for the shoulder pain in the last month.

6.10. Pain in the forearm and hand The pain intensity in the last 6 months was very low in the forearm and hand in the T and S groups during the whole study period, Fig. 9. When comparing the same subjects after 6 years follow up with after optometry in the T and S groups, no signicant changes were reported in the C group ( p 0:11) or the T group ( p 0:27) while the S group reported a signicant increase in the forearm pain ( p 0:04). When comparing the follow up at 6 years with the start of the study, no signcant change in the forearm pain was observed in the C group ( p 0:13) while there was a tendency to increase in forearm and hand pain in the T group ( p 0:06) and S group ( p 0:07). When comparing the three groups at 6 years follow up no signicant dierences were found ( p 0:34). Considering the forearm and hand pain the last month, no signicant changes were reported in any of the groups during the study period. 6.11. Back pain in lumbar region When comparing the same subjects regarding intensity of back pain last 6 months in the period from 6 years follow up with after optometry in the T and S groups, no signicant changes were reported in any of the three groups (T group, p 0:36; S group, p 0:34 and C group, p 0:41). When comparing the 6 years follow up with the start of the study, the T group reported a signicant reduction in back pain ( p 0:001) while no signicant changes were observed in the S group ( p 0:18) or the C group ( p 0:32). Similar results were found for the back pain the last month. Fig. 10. 6.12. Organizational and psychosocial factors Such confounding factors may inuence on the dependent variables, for example, pain. Therefore these factors were tracked during the study period. Possible time eects and interactions between time and intervention were tested by a repeated measurement analysis of variance with the baseline value as covariate. For all psychosocial factors, there was no statistical intervention eect or time eect and no interactions between time and intervention were found. The only psychosocial factors which correlate with pain was job control and possibility to discuss problems with the immediate job superior. Job control for the T group had a Pearsons correlation to shoulder pain (r 0:135; p 0:02). For forearm pain the S group had a Pearsons correlation to job control (r 0:165; p 0:04) and possibility to discuss with the immediate job superior (r 0:192; p 0:02). Job control for the C group had a correlation to neck pain (r 0:242; p 0:0006)

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

567

6.13. Work tasks There was a dierence in development of shoulder and back pain in the T and S groups. In the S group there were 6 subjects who worked with installing software in the telephone exchange in Norway in addition to software engineering. These 6 subjects were compared with the rest of the S group regarding pain. There were no signicant changes between the groups during the study period or between the groups before interventions and at 6 years follow up. This additional work task cannot explain the dierence of pain in shoulder and back between the T and S groups. 6.14. Drop-out
Fig. 10. Intensity of back pain last in the month for the T, S and C groups. The values are given as mean, with 95% condence interval, before the interventions, after the lighting intervention, after the optometric intervention, after 3.5 and 6 years. On the ordinate, 0 indicates no pain while 100 is very severe pain. A* indicates a signicant dierence when comparing with the column containing a+. A (*) is very close to a signicant dierence.

and shoulder pain (r 0:162; p 0:02). However, the job control and possibility to discuss problems with the immediate job superior show only a tendency to change during the study period i.e. the eect of the two psychosocial factors on the end variable pain must have been minimal (job control, p 0:07; possibility to discuss problems with the immediate job superior, p 0:08).

Reason for drop-out was a reduction in the number of workplaces/workforce in all three groups. All of the drop-outs got jobs outside the company or took prepension. During the 6 years study period there were 43 subjects who dropped out, 23 in the T group, 6 in the S group and 14 in the C group. In order to create a complete sample for the health parameters (pain level in the neck, shoulder, forearm, back and head as well as visual discomfort) at measurement 5, the last observed pain in dierent body areas was carried forward to measurement 5. The same statistical analyses that were performed on the incomplete sample at measurement 5, were then performed on the complete sample. There were no signicant dierences regarding pain parameters for neck, shoulder, forearm, head and visual discomfort. Regarding back pain, there was a tendency to higher values in the complete sample, particularly for

Table 4 Drop-out analysis. Pain last 6 months at measurement 5 Body area Group Incomplete group Mean Neck T S C T S C T S C T S C T S C T S C 15.5 22.8 18.8 13.1 19.0 10.7 12.5 9.3 8.4 11.2 23.2 10.5 15.6 20.9 22.1 13.2 11.5 21.9 95% C.I. 10.620.4 15.829.8 12.225.5 7.818.5 10.827.2 4.916.6 7.717.3 3.315.3 2.114.6 6.515.9 15.630.8 4.516.5 10.620.6 14.227.5 14.829.4 8.418.0 6.116.8 13.030.7 0.212 P-value Complete group Mean 16.1 21.4 21.2 12.5 20.3 14.0 12.2 11.5 14.1 15.7 21.6 16.6 17.1 18.8 23.8 14.5 13.6 24.4 95% C.I. 12.519.7 15.527.4 16.126.4 8.416.6 13.027.6 8.719.3 8.216.2 5.717.2 8.220.0 11.220.2 14.928.2 11.122.2 13.221.1 13.024.6 18.329.3 10.818.2 8.518.7 17.431.4 0.198 P-value

Shoulder

0.352

0.198

Forearm

0.344

0.587

Back

0.013

0.286

Head

0.164

0.072

Visual discomfort

0.080

0.254

568

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

the C group. The dierence in p-values is due to the increase in the number of subjects in the complete sample. This leads to less variance. The analyses showed no indications of systematic biases regarding health parameters, Table 4.

7. Discussion Prospective epidemiological studies have many pitfalls. In this study the same subjects were followed over a period of 6 years. There were no changes of the work tasks or of the senior leadership in any of the groups during the study period. However, such studies have a lot of confounding factors such as organizational and psychosocial factors. For all psychosocial factors, there was no statistical intervention eect or time eect and no interactions between time and intervention. This means that the eect of the psychosocial factors tracked in this study, may have approximately the same inuence on the dependent variables (i.e. the pain level in dierent body areas) during the study period. The organizational and psychosocial factors may have inuenced the absolute level of pain, but the change in the level of pain due to the intervention factors will not be aected. Optimizing the visual condition seems crucial in order to reduce visual discomfort and eye symptoms. All three groups reported signicant improvements of the visual condition in parallel with a signicant reduction in visual discomfort. This means that lighting and optometric corrections are extremely important in reducing visual discomfort. The importance of increasing the illumination level to more than 500 lux is supported by many studies. The illuminating Society (IES, 1989) has recommended oce lighting levels depending on age and work tasks, from 500 to 1000 lux. Christoersen et al. (1999) found in a study of 1800 VDU workers in Denmark that the illumination levels at their work surface were between 200 and 500 lux. Only 55% of the workers were satised with these illumination levels from the general lighting in the ceiling. Almost 90% of the VDU workers supplemented their general lighting with moveable desk lamps, either at their desk and/or at their VDU. Daylight shows great variability compared to articial lighting. The same researchers found that even on a clear sunny day about 70% had their indoor lighting on. Tenner et al. (1997) found that in order to get 80% of oce workers satised with their illumination level, the illuminance should be approximately 900 lux. Fukuda et al. (1999) found that quality of readability increased with level of illuminance up to more than 500 lux. Further, Sugimoto et al. (1999) found that illumination levels lower than 500 lux or higher than 1000 lux increase the physiological load in terms of increased heart rate variability (HRV). On the

contrary it was found in a study by ODonnell et al. (1999) that illumination levels below 500 lux are rated as insucient by only 13% of VDU workers. Therefore, it seems important to relate lighting conditions to health parameters such as visual discomfort in prospective eld studies. This is supported by the fact that the VDU workers vary greatly in how they set their sucient illumination level (Veitch and Newsham, 1999). Signicantly lower degree of visual discomfort was found in a group sitting more than 1.5 m away from the window compared with a group sitting less than 1.5 m from the window. Both groups had sightline parallel to the window. The light contribution from the window is clearly reduced when the VDU is placed 2 m or more away from the window (i.e. down to 25% of the level at the window), (Bjrset, 1986). The distance from the window seems not to be important for satisfaction of view through the window. Christoersen et al. (1999) found no relationship between satisfaction of view and distance from the workplace to the window. Luminance and its distribution in the room are important for avoiding contrast glare. Luminaires with both direct and indirect lighting give better luminance distribution and better visual conditions compared to only direct light (Bjrset, 1986; Hedge et al., 1995). Visibility may also be reduced if objects with high luminance, such as windows, are reected in the VDU screen (Bjrset, 1986; Hedge et al., 1995). An extensive review of lighting and visual conditions for VDU workers is given by Bjrset (1997). Among the VDU workers having their sightline partly towards window (approximately half turned), no relationship was found between visual discomfort and the distance from the window. The reason for this result may be that all workers had Venetian blinds which were used by almost all VDU workers (meaning that they have no windows to see through). This is in accordance with a Danish study documenting that 70% often or always used Venetian blinds or curtains when the sky was clear (Christoersen et al., 1999). In the same study, 20% of the VDU workers reported that in spite of using Venetian blinds, they had to move the VDU further away from the window and rotate their VDU to avoid screen reections and glare from the window. Further, more than 70% were highly satised with daylight at their desk for normal oce activity, while as many as 63% were unsatised with the daylight conditions when doing VDU work. The luminance from windows will normally be high even without sunshine, i.e. 1000 cd/m2 or more. Already at a background luminance of 600 cd/ m2 visual fatigue is increased and accomodation is aected (Wolska and Switula, 1999). Therefore, if the VDU is placed so that the gaze direction is towards the window, there is an increased possibility for glare.

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara

569

Other eects of lighting were shown in a cross. ller et al. (1999). They found cultural study by Ku the highest occurrence of eyestrain during winter in the northern countries, and particularly when the oce had low illumination level. The same researchers found that sadness, fatigue, sleeping problems and social withdrawal increased with reduced number of daylight hours. Reduction in secretion of cortisol was also found. The adjustment of the workplace was performed so that the sightline to the centre of the screen was approximately 151 below horizontal. The angle was chosen according to study by Nyman and Berns (1996) who found that VDU operators with a head exion of 91 (S.D. 3.3) and viewing angle from the Frankfurt Plane of 20.41 (S.D. 4.5) to the centre of the screen had the lowest prevalence of eye disorder and discomfort of the neck, shoulder and back. The prevalence of these disorders increased for those operators with higher and lower angles. This is supported by Menozzi et al. (1992). A more detailed discussion regarding gaze angle (s to the screen and health consequences is given by Aara et al. (2000). The workplace intervention in this study in terms of giving the VDU workers support for their forearms, did not inuence the pain level to a great extent during the study period. Shoulder pain was signicantly reduced in the C group after intervention. Considering the whole study period the C group reported a signicant reduction of shoulder and neck pain. Electromygrapic (EMG) measurements were not taken in the C group. However, the operators in the C group got the same work positions as the former T and S groups. Results from EMG measurements both in laboratory and eld studies showed that supporting the forearm on the table top reduced the static muscle load in trapezius and ( s et al., 1997, 1998). In the erector spina at L3 level (Aara rst part of this study, less shoulder pain was reported after ergonomic intervention that reduced the static trapezius EMG activity level from 1.5 to 0.25% maximum voluntary contraction (MVC). This is in accordance with a study by Jensen et al. (1998) who found less pain in the passive than the active shoulder, and they recorded static EMG activity levels of the two shoulders of approximately 1.5 and 0.2% MVC. The T group reported a signicant reduction in shoulder and back pain which are also in accordance with the results from the EMG measurement. Small reduction in static muscle load seems important (Westgaard, 2000). However, the S group did not report reduction in the neck and shoulder pain considering the whole study period. This is contradictory to what should be expected from the EMG measurement carried out after workplace ( s et al., 1998). No intervention in the S group (Aara reasonable explanation for the dierences in pain development in the S group compared with the T and

C groups are found, considering the whole study period. A small number of workers in the S group did installing of software in telephone exchanges in addition to VDU work. However, this work did not increase the pain compared with software engineering in the S group, because the pain level for these subjects was in fact lower compared to the rest of the S group. A relationship between visual discomfort before interventions and pain in the neck and shoulder was found (0.30oro0.40), ( s et al., 1998). However, the visual discomfort did (Aara not dier signicantly between the S and T groups during the study period. Organizational and psychosocial factors could hardly be the reason for the observed variation in pain level between the groups. No signicant changes in organizational and psychosocial factors were found during the study period in any of the groups. This means that the eect of these factors may have had approximately the same inuence on the dependent variables, i.e. the pain level in dierent body areas during the study period. The observed variation of pain during the study period, demonstrates the importance of long lasting intervention studies in order to evaluate the eect of interventions (Kilbom, 1988; Shackel, 2000). Regarding the forearm pain, the S group reported a signicant increase in the period after optometry to the 6 years follow up. Considering the whole study period, it was observed that both the T and S groups reported increased forearm pain. In this study the intervention consisted of supporting the forearms on the table top when using the mouse and keyboard. Such intervention seems not to be eective to reduce the forearm pain for VDU workers. The reduction of forearm pain may have been achieved by using a more neutral position of the forearm when operating the mouse. A neutral position of the forearm seems important to reduce pain in the ( s et al., 1999; Aara ( s and upper part of the body (Aara Ro, 2001).

8. Conclusion This 6 years prospective epidemiological study has shown that lighting and optometry are of crucial importance for VDU workers. All three groups reported signicant reduction of visual discomfort after improving visual conditions and giving optometric corrections. Supporting the forearm on the table top, two of the three groups reported positive health eects in terms of reduced shoulder pain.

Acknowledgements This study was supported by grants from The Norwegian Research Council and The Norwegian

570

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara ( s, A., Fagerthun, H., Larsen, S., 1995. Is there a Horgen, G., Aara reduction in postural load when wearing progressive lenses during VDT work over a three-months period? Appl. Ergon. 25 (3), 165171. IES (Illuminating Engineering Society of North America, New York)., 1989. VDT Lighting Recommended Practice for Oce Lighting with Visual Display Units IES RP 24-1989, New York. Jensen, C., Borg, V., Finsen, L., Hansen, K., Juul-Kristensen, B., 1998. Job demands, muscle activity and musculoskeletal symptoms in relation to work with the computer mouse. Scand. J. Work Environ. Health 24, 418424. Kamwendo, K., Linton, S.J., 1991. A controlled study of the eect of neck school in medical secretaries. Scand. J. Rehab. Med. 23, 143152. ( ., 1988. Intervention programmes for work-related neck and Kilbom, A upper limb disorders: strategies and evaluation. Ergonomics 31 (5), 735747. . rvi, T., Riihima . ki, V., 1983. Prevention of Kukkonen, R., Luopaja ( lseth, T.O. (Ed.), fatigue amongst data entry operators. In: Kva Ergonomics of workstation design. Butterworths, London, pp. 2834. . ller, R., Ballal, S.G., Mikkellides, L.B., 1999. Shortness of dayKu light as a reason for fatigue and sadness. A cross-cultural comparison. In: Proceedings of the 24th session of the CIE, Warsaw, June 2430, Vol. I, CIE, Publication 133, ISBN 3900734933, 291294. Menozzi, M., Buol, A.V., Krueger, H., 1992. Fitting Varifocal Lenses: Strain as a Function of the Orientation of the Eyes. Service Optique Physiologique, ESSILOR, F-94000 Creteil. Methling, D., 1992. Standards and methods for visual examination of VDU users-a critical revue. In: H. Luczak, A. E. Cakir, G. Cakir . t Berlin(Eds.), Work With Display Units. Technische universita . r Arbeitswissenschaft, Berlin, B-7. FInstitut fu Nyman, K., Berns, T., 1996. Assessing the viewing angle. A prerequiste for good visual ergonomics. The 25th International Congress on Occupational Health. Book of Abstracts II, OS 419. National Institute for Working Life, S171 84 Solna. ISBN: 91-7045-381-0. ODonnell, B., Raitelli, M., Kirschbaum, C., 1999. Lighting evaluation at work places in subtropical regions. Proceedings of the 24th session of the CIE, Warsaw, June 2430, Vol. I, CIE Publication 133, ISBN 3900734933, 313317. Ong, C. N., 1984. VDT work place design and physical fatigue: A case study in Singapore. In: Grandjean, E (Ed.), Ergonomics and Health in Modern Oces. Taylor & Francis, London. Oxenburgh, M.S., 1985. Musculoskeletal injuries occurring in word processor operators. In: Steenson, M. (Ed.), Readings in RSI. The ergonomic approach to repetition strain injuries. New South Wales University Press, 9195. Punnett, L., Bergqvist, U., 1997. Visual Display Unit Work and Upper Extremity Musculoskeletal Disorders. A Review of Epidemiological Findings. National Institute for Working Life, 17184 Solna, Sweden-Ergonomic Expert Committee Document No. 1. ISBN 917045-436-1, ISSN: 0346-7821, 16. Shackel, B., 2000. People and computersFsome recent highlights. Appl. Ergon. 31, 595608. Sheedy, J.E., 1995. Vision at computer Display. Vision Analysis. 136 Hilcroft Way. Walnut Creed, CA 04596 USA. Shute, S.J., Starr, S.J., 1984. Eects of adjustable furniture on VDT users. Human Factors. 26, 157170. Smith, M.J., 1997. Psychosocial aspects of working with video display terminals (VDTs) and employee physical and mental health. Ergonomics 40, 10021015. Sugimoto, S., Ikeda, I., Noguchi, Y., 1999. Estimation of Physiological Eects of Lighting by Analysis of Heart Rate Variability. Proceedings of the 24th Session of the CIE,

Employer Federation, Division TBL. Thanks to all the participants of the study and the management who made this study possible.

References
( s, A., Fostervold, K.I., Ro, O., Thoresen, M., Larsen, S., 1997. Aara Postural load during VDU work: a comparison between various work postures. Ergonomics 40 (11), 12551268. ( s, A., Horgen, G., Bjrset, H-H., Ro, O., Thoresen, M., 1998. Aara Musculoskeletal, visual and psychosocial stress in VDU operators before and after multidisciplinary ergonomic interventions. Appl. Ergon. 29 (5), 335354. ( s, A., Ro, O., Thoresen, M., 1999. Can a more neutral position of Aara the forearm when operating a computer mouse reduce the pain level for visual display unit operators? a prospective epidemiological intervention study. Int. J. Human-Computer Interaction 11 (2), 7994. ( s, A., Horgen, G., Ro, O., 2000. Work with the visual display Aara unit: health consequences. Int. J. of Human-Computer Interaction 12 (1), 107134. ( s, A., Ro, O., 2001. Will supporting the forearm in a Aara neutral position reduce the musculoskeletal discomfort for computer workers. Review paper. In: Leif Sandsj and Roland Kadefors (Eds.), The 2nd PROCID SYMPOSIUM. Prevention of muscle disorders in computer users: Scientic and Recommendations. 810 March 2001, Gteborg Sweden, pp. 101104. Bergqvist, U., Knave, B., Voss, M., Wibom, R., 1992. A longitudinal study of VDT work and health. Int. J. Human-Computer Interaction 4 (2), 197219. Bjrset, H.-H., 1986. Lighting for visual display unit workplaces. Work With Display Units 86. Proceedings Part II, Elsevier Science Publishers B.V., North Holland, pp. 683687. Bjrset, H.-H., 1997. Visual conditions for VDU workers. In: D. . ck (Eds.), Brune, G. Gerhardsson, G.W. Crockford, D. Nordba The Workplace. Major Industries and Occupations. Vol. 2, ISDN 82-91833-00-1, 191214. Borish, I.M., 1975. Clinical Refraction. Pro. Press Inc., Chicago, USA. Christoersen, J., Johnsen, K., Petersen, E., Hygge, S., 1999. PostOccupancy evaluation of Danish oce buildings. In: Proceedings of the 24th session of the CIE, Warsaw, June 2430, Vol. I, CIE Publication 133, ISBN 3900734933, pp. 333337. Fukuda, R., Shimizu, Y., Fukuda, T., Takahashi, Y., Fuchida, T., 1999. Empirical study of desired illuminance for the elderly: Part 1Fan experimental consideration of indices for readability- In: Proceedings of the 24th session of the CIE, Warsaw, June 2430, Volume I, CIE Publication 133, 8791. ISBN 3900734933. Guillon, M., 1999. Pilot evaluation of head and eye tracker system to study visual behavior with PAL amd single vision lenses. J. Am. Academy Optometry 76, 181. Hedge, A., William Jr., R.S., Franklin, D.B., 1995. Eects of lensed-indirect and parabolic lighting on the satisfaction, visual health and productivity of oce workers. Ergonomics 38 (2), 260280. ( s, A., Fagerthun, H.E., Larsen, S., 1989. The Horgen, G., Aara work posture and the postural load of the neck/shoulder muscles when correcting presbyopia with dierent types of multifocal lenses on VDU-workers. In: Smith, M.J., Salvendy, G. (Eds.), Work with Computers: Organizational Management, Stress and Health Aspects. Elsevier Science Publishers B.V, Amsterdam, pp. 338347.

(s et al. / Applied Ergonomics 32 (2001) 559571 A. Aara Warsaw, June 2430, 1999. Vol. I Part 2, CIE Publication 133, ISBN: 3900734933, CIE Central Bureau, Vienna, Austria pp. 8082. Tenner, A.D., Begemann, S.H.A., Van den Beld, G.J., 1997. Acceptance and Perference of Illuminances in Oces. Lux Europa. The 8th European Lighting Conference, Amsterdam 1114 May 1997, Lux Europa 1999, 130143. Veitch, J., Newsham, G.R., 1999. Preferred luminous conditions in open-plan oces: Implications for lighting quality recommenda-

571

tions. Proceedings of the 24th session of the CIE, Warsaw, June 46, Vol. II, CIE Publication 133 ISBN 3900734933, 313317. Westgaard, R.H., 2000. Work-related musculoskeletal complaints: some ergonomics challenges upon the start of a new century. Appl. Ergon. 31, 569580. Wolska, A., Switula, M., 1999. Luminance distribution on VDT work stands and visual fatigue. Proceedings of the 24th session of the CIE, Warsaw, June 2430, Vol. I, CIE Publication 133, ISBN 3900734933, 4347.