Eye discomfort and work with visual display terminals

B. Eye discomfort and work with visual display terminals. Scand J Work Environ Health 1994;20:27- 33. OSJF.<;TIVES - The aim of this study was to investigate the rela tionships between eye discomfort symptoms and work with visual display terminals among routine office workers. METHODS - Three hundred and twenty-seven office workers and their work stations were investigated by means of questionnaires and worksite investigations. The data were subjected to multivariate logistic regression analyses. R ESULTS - The occurrence of eye discomfort increased as the extent of VDT work increased, as did the specific symptoms of sensitivity to light and smarting, gritty feeling, or redness. The use of spectacles during visual display terminal work, age, stomach stress reaction, distances between the eye and different visual task objects, as well as the vertical po sition of the terminal also influenced certain symptoms. Having (i) the terminal at about eye level during prolonged terminal work, (ii) using monofocal glasses during terminal work in situations with large distance variations to visual task objects, or (iii) being elderly with prolonged terminal work and reporting stomach stress all led to increased odds ratios for certain eye discomfort symptoms. CONCLUSIONS- The use of a visual display terminal in routine office work is associated with an in creased occurrence of certain eye discomfort symptoms. This association is affected also by the pres ence of certain other individual and ergonomic factors.

Eye discom fort among visual dis play terminal (VDT) wo rkers was fir st describ ed in 1973 by Hultgren & Kn ave ( I). Sinc e then severa l stu dies have investiga ted this associatio n. Our stud y is part of an in vestigation of a coh ort of routine office workers in Stockh olm , Swed en , with emphas is o n VOT wo rk . In a first cross-sectio na l exa minatio n in 1981, eye discomfort was mor e common amo ng VOT user s than am ong non -VOT users (2). Between 1981 and 1987, the cumulative inc idence of eye discom fort increased with the extent of VOT work (3) . The question s elucidated in thi s study -a seco nd cross-sectional investi gati on in 1987 of the same cohortare "Wha t indivi dual, ergon omi c, and organiz ational varia bles are associated with eye discom fort among routine office workers , do VOT user s experience a higher prevalence of eye di scom fort than non-VOT users, after adj ustme nts for the po ssibl e confounding effec ts of the se other variables , and is the effect of VOT work on various eye discomfort symptoms -if any -modi fied by these variables?"

Subjects and methods
Cohort description and data acquisition In 1981, 588 office wo rkers with or witho ut V OT work in seve n companies were se lected for inclu sion in the study. Most (N == 535 , 9 1%) of them participat ed. By 1987 , 353 individ ua ls remai ned in the group -the othe rs had quit working in the respective co mpa ny, retired, or the like . For 150 (82%) of the dropouts, data were obtained that de scribed whether var ious typ es of discomfort (including eye discom fort) or VDT wor k had co ntributed to their decision to quit th e workplace. Onl y one individua l rep orted that eye discomfort had influenced the decision.
A que stionn aire with detai ls on eye discomfort , individua l fac tors, and wo rk co nditio ns was distributed to the 353 persons who remained in the study, with a 93% response rate (N == 327). Ergonomic conditions were measur ed for 228 (88%) of the 260 VDT user s. Th e rel ative humidity at the workplaces was obt ained fo r 203 (78 %) of the VOT users.

Variable s describin g eye discomf ort
Th e respond ent s we re as ked about the occurrenc e of eight different eye sy mptoms. "Any discomfort" was the occurrence of an y of the se symptoms , while the vari able "moderate discomfort" described the occurrence of only tho se sy mptoms that were mo re frequent or intense. (For a more-detailed description , see refere nce 2.) The occurrence of smarting (S), gritty feel ing (G), or redness (R) (regardless of frequenc y and intensity) were collected into the varia ble SGR, on the assumption that these sy mpto ms rep resent identical or similar etiologies. Oc currences of itching , ac hes , sens itivity to light , teariness, and dryne ss sy mptoms were ana lyzed separately .

27
Scand J Work Environ Health 1994, vol 20. no I Variables describing work with visual display terminals "No VDT use" indicated that the respondent did not use a VDT or that his or her use was normally less than 5 h a week. The VDT users were divided into those with normall y 5-20 h a week and more than 20 h a week, The type of VDT work was categorized as data entry (data entry or word processing or both) and interactive work (data acquisition, interactive work, and mixed work or progr amming or both). Interactive work implied longer viewing time towards the screen than data entry work. Thus, when the highest odds ratio for a discomfort was found for those doing interactive work for more than 20 h a week, the effect of the total time lookin g at the screen ("VDT viewing time") was analyzed. This estimate was based on VDT work types, the total VDT work time, and repo rted estimates of the fract ion VDT viewing timeNDT worktime for different VDT work types (2,(4)(5)(6).

Other variables
The variab le "VDT glasses" described the use of spectacles or contact lenses in the VDT work situation and was separated into none, monofoc al, or others (bifocal or progre ssive glasses and contact lenses) . "Stomach stress" reactions were based on responses indicating an upset stomach, while "tiredness" reactions were related to the inabilit y to relax and other such symptoms (7) . "Negative affectivity" corresponded to a personality trait expressed as "anger, disgust, scorn, guilt, fearfulness, and depression" that has been suggested to influen ce the reporting of discomfort (8)(9)(10). Age, gender, and smoking (yes or no) were also used in the analysis .
"Organizational influence" was the opportunity to take initiat ive, influence planning, develop oneself at work, and other such factors . "Rest-break opportunity" was based on the ability to take unscheduled breaks. "Flexibility of work task" described whether workers could shift their work tasks to another day, avai lability of several tasks, and the like. "Work pace" summed the perceived demands for concentration, time pressure, inability to take a breather, and the like (7). Other variables were "co mpany type," "ability to change work if tired," "overtime frequency," and "overtime on short notice ." The angle between the horizo ntal plane from the eyes and the line betwe en the eyes and the center of the VDT was measured for the vertical position of the VDT. "High" meant an angle between 0 and 20 degrees below the horizontal line. "Low" meant a VDT at a lower level, for about 90% the level being between 20 and 30 degrees . The distances were measured between the eyes and the followi ng three task objects: the VDT , the keyboard, and the document or document holder. The maximum difference between these three distances was also noted. In addition, observations were made regarding the oper-28 ator' s head position (bent forward or not) and the presenc e of glare on the VDT. The observations and measurements corresponded to "normal" positions in the most common work situation. The average relative humidity during a workweek was also measured; for details see reference II.

Analysis
For each eye discomfort variable, a multivariate model was constructed which included "VDT use," as well as important individual and organizational variables, in order to adjust for their possible confounding effects (table I). The procedure for the construction of this model was that, at first, six variables describing individual conditions that were considered relevant for eye discomfort were selected, and only those showing a substantial association with the symptom in questi on were retained and combined in a multi variate model. This model was then extended by including the varia ble "VDT use" and also eig ht selected organizational variables. The number of variables in the model was reduced through the elim ination of variables with low adjusted odds ratios at each step in order to increase the precision, but only insofar as odds ratios of other variable s were unaffected by this elim ination ( 12). Amon g the VDT users, the impacts of eight ergono mic variables and negative affectivity were also evaluated (table 2). The procedur e for the construction of this seco nd model was analogous with that of the first.
Some interactions were investigated, with the aim of detecting whether a variable describing VDT work conditions and another variable had a combined effect on eye discomfort in excess of the sum of the effect of each variable in isolatio n (13). The selection of interactions to be eva luated was based on plausibility of the interaction s.
Estimates of univariate (crude) and stratum-specif ic odds ratios and their 95% precision-based confidence intervals (95% CI) were computed. (For both theoretic al and practical reasons, the odds rati os rather than the prev alence ratios were computed.) Unconditional maximum likelihood estimates were obtained from the multi variate logistic regression analysis. The FREQ and CATMOD procedures in the SAS system (14) were used for these analyses. When the number of individuals was small, the results were verified by either reducing the number of variabl es in the model (in the absence of confounding) or using exact methods (Fisher). Linear trends for odds ratios were calculated as sugges ted by Rothm an ( 13).

Results
Tables I and 2 show the associations found between the various types of eye discomfort and VDT work with control for indi vidual , organizati onal and (for table 2) ergonomic variables. Increasing odds ratios with increasing VDT use in hours per week were observed for any discomfort, sensitivity to light, and possibly also for SGR. For sensitivity to light, the VDT users performing interactive work had higher odds ratios than those perfo rming data entry work. Therefore, an analysis of VDT viewing time was performed among the VDT users. The precisio ns of the odds ratios for VDT viewing time were lower, a linear trend for the odds ratio of sensitivity to light as a function of VDT viewing time having a slope of 0.09 (-0.06 to 0.24) per hour of viewi ng time.
For three of the six individual variables (age, VDT glasses , and stomach stress), odds ratios were elevated for cer tain sympto ms (as seen in tables I and Scand J Work Env iron Health 1994, vol 20, no I 2). For gender, the crude analys is resulted in odds ratios between 1.5 and 2 (women vers us men) for some symptoms, but these odds ratios were considerably reduced in the multivariate models, leadi ng to the elimination of the gender indicator from the models. Smoking and tiredness reactions did not have eleva ted crude odds ratios. For work pace , elevated odds ratios appeared in some models (table I). In table 2, this variable was not retai ned due to the rcduction of these odds ratios by the introduction of the variable negative affectivity. Other organizational variab les had odds ratios of less than 1.5 in the multivariate models and were not retained. Six ergonom-   9 Compared w ith no use of gl asses during VOT work .
n Com pared wit h tra vel agency. i Compared wi th a relative hum idit y of 2:30% . I Ano ther model was used to evaluate eye-obj ect differenc es and glare (apart from relativ e humidity, terms of th e mod el shown were pres ent with si milar odds ratios). k Compa red wit h di stance diff erences of < 10 cm . f Compared wit h no glare.
ic variables (VDT vertical position, distance to VDT, distance to keyboard, maximum difference in eyetask object distances, specular glare, and relative humidity) were retained for at least some symptoms (table 2). The other two were eliminated due to low odds ratios.
Some interactions in the excess of additivity were indicated betwee n a variable describin g VDT work and certain individual or ergonomic variables, albeit with low precisions. Having a VDT viewing time of more than 9 h . week-I and having the VDT placed at about eye level was assoc iated with an excess occurrenc e of sensitivi ty to light (fig ure I). Likewise, working more than 20 h a week at a VDT and 30 having the VDT in a high vertical position was associated with an exc ess occurrence of any discomfort, with an odds ratio of 5.2 (N = 191, 95% CI 1.7-15.9).
Individuals who used monofocal glasses durin g their VDT work and had fairly large (> 10 em) differences between distances from their eyes to different task objec ts had an excess occurrence of dryness symptoms (figure 2). The odds ratio for the factor combination was 6.7 (95% CI 0.7-68.0). The number of individuals available for this analysis was limited, especially in the reference category (N =14).
In an a posteriori model , users of monofocal glasses and large object differenc es had an odds ratio of 2.2 . Associations between dryness symptoms and combinations of the use of monofocal glasses during visual display terminal (VOT) work and the maximum difference between distances from the eyes to the VOT, the keyboard , or the manuscript. All of the comparisons were made within the VOT user group (N =114). The group using monofocal glasses at the VOT and having at least a to-em differen ce in the distances from the eyes to different task objects had an odds ratio of 6.7 (95% CI 0.7-68.0) compared with the reference group (shown in black ). The odds ratio in excess of addit ivity is indicated. Compared with all of the othe r groups in combination , the odds ratio of this group was 5.2 (95% CI 1.6-16.9).

Discussion
Eye discomf ort and work with visual displa y terminals For any discomfort, an association was found with the curre nt extent of VDT use. Several studies have found a higher prevalenc e of eye discomfort or asthenopia among VDT operators compared with non-VDT operators, or a prevalence that increa sed with VDT working time (15)(16)(17)(18)(19)(20). Other studies have failed to do so (5, 2 1, 22), although a nonsignifi cant increase was seen at least in one study (22). In a few  N = 191). All comparisons were made within the VOT user group. In the group having a VOT view ing time of more than 9 h a week and the VOT placed 0-20 degrees below eye level, the odds ratio in excess of additivity is indicated. A few individuals had a vert ical glance direction downwards greater than 30 degrees; they have been included in the 20-to 3D-degree group .
(95% CI 0.5-10.2) if less than 40 years of age, and 4.9 (95% CI 1.9-13.0) if at least 40 years of age, compared with those lacking the combination of both fact ors.
Older (~5 0 years) individual s who reported stomach stress reacti ons and who worked for more than 20 h a week at a VDT had an excess occurren ce of teariness, with an odds ratio of 9.0 (95% CI 2.2-36.4) comp ared with younger pers ons «40 years) who reported stomach stress and worked less than 20 hours a week at a VDT. The odds rati os of the other VDT user groups reporting stomach stress were lower, but they increased both with age and with the amount of VDT use. Among individu als who did not report stomach stress, the odds ratios did not increase with age or VDT use.
Odds ratio for sensitivi ty to light studies, mixed results appeared (23,24). Thus our study, in agreement with the majority of other studies, suggests an increase in eye discomfort with VDT work or VDT work duration, also when a number of pos sible confounding variables are taken into account .
Several other studies have reported on specific eye discomfort symptoms with mixed result s. Rossignol and her co-workers (18) found an increase in redness but not in irritating symptom prevalence s with VDT work duration. Coe et al [cited from Collin s et al (16)] found an increase in reported sensitivity to light ("glare") among VDT operator s comp ared with referents. Belisario et al (15), on the other hand, failed to find associations between redness or sensitivity to light and VDT work. Teariness or itching was not significantly associated with VDT work or VDT work duration (15,18,22). Coe et al [cited from Collins et al ( 16)] found an increase in the reporting of aches among VDT operators compared with that of referen ts. In summary, symptoms of smarting, gritty feeling, or redness, as well as sensitivity to light, were found to be associated with VDT work in our study, with some limited previous support from other studies.

Influ ence of other fa ctors on eye discomf ort
Our study has provided data suggestin g the involvement of several individu al and ergonomic variables with eye discomfort. Some of the variables acted as confounders of the VDT use-eye discomfort relationships. The most consistent confounder was the variable VDT glasses, followed by age and stomach stress. For the four two-factor interactions described in the Results section, the combined effect was larger in this study than the sum of the effe cts of each factor in isolation. While this result could have been due to chance, the plausibility of the noted interactions supports their general credibility.
Older age was assoc iated -with limited precision -with some specific symptoms (sensitivity to light and teariness), while dryness symptoms appeared to be associated with younger age. Reported stomach stress was associated with most of the symptoms in our study. These findings correspond with those of some other studies (16,17,19,25). (Most of the reviewed studies have utili zed age and gender as possible confounding factors to be controlled for, but they have less often reported specific associations.) For teariness, both age and stomach stress were relevant factor s, and they appeared to interact with the effect of VDT work duration. Contrary to expectations, gender was not retained as an important variable, since its effect was strongly reduced when other variables were introduced into the models containing individual and organizationa l factors. It should be noted that this study population included a fairly large number of older men doing routine VDTwork.

32
There were no indications of an impact of negative affectivity on the odds ratios between VDT work and eye discomfort. The confounding effect of negative affectivity on the odds ratio of a perceived high work pace (see the Results section) could conceivably have been due to the perception of a high work pace being influenced by the negative affectiv ity trait.
A reduced blinking rate and widening of the ocular surface due to the raisin g of the viewing angle during VDT work would increase the drying of the eyes, causing eye discomfo rt (26)(27)(28). This hypothesis appears to be consistent with our findin gs that certain types of discomfort were associated with a high position of the VDT (at about eye level). The high odds ratios obtained for prolonged VDT use or viewing time in combination with a high VDT position (see, eg, figure 1) further strengthens the proposed hypothe sis.
The use of glasses during VDT work was another important factor for itching , aches, and dryness symptoms in our study. Eye symptoms have been found to be associated with small refractive errors (29-3 1), and it is arguable that such error s would be more common among those that have found it necessary to use spectacles or contact lenses. Using contact lenses may also be a problem in relat ion to blink rate and tear fluid , as has alread y been discussed (27). Other researchers have, on the other hand, not found differences attributable to the use of glasses or contact lenses (18,25). It should be noted that the population of our study was a fixed cohort, with a fairly high median age (47 years). This situation may make comparability with some other studies less clear-cut [eg, that of Rossignol et al ( 18), in which the median age was 34 years] and may also partly explain the larger impact of the use of glasses during VDT work in our study. In our earlier investiga tion in 1981, the impact of VDT glasses on the then six-year younger cohort appeared to be considerably less (32).
A low relat ive humidity was important for some symptoms, as has been seen also in another study ( 19). Another important factor for eye discomfort was short distances between the eyes and the VDT, consistent with the findings of some studies (33,34) that suggeste d an increase in fatigue with short distances, especially in individuals with a disparity betwee n the (shorter) viewing distance and the individual's (longer) dark convergence. For dryness, there was an apparent interaction in our study between large variations in the distance to the eye task objects and the use of monofocal glasses during VDT work. This interaction appeared to be stronger with age and could have been related to the reduced accommodative capacity of older age.