Analysis of real‐world visual ergonomics at the visual display unit

In this interventional study, the ergonomic workplace set‐up and the impact of character size on subjectively estimated working productivity and computer vision syndrome (CVS) were evaluated in the field.

covering ergonomics of human-computer interaction. It contains 54 parts as single standards. Visual ergonomics for work with screens, such as recommended character size, viewing distances and gaze angles are mainly represented in the standards ISOs 9241-303:2011 8 'Requirements for electronic visual displays' and ISO 9241-5:1998 9 'Workstation layout and postural requirements'. Both standards were adopted by The European Committee for Standardization (CEN) and are therefore valid in their member states, including the United Kingdom, Germany, France, Spain, Italy, Sweden and the Swiss Confederation. In addition to these international standards, there are other valid national standards, such as in the USA 10 and Canada. 11 While these standards are generally not legally binding, many industrialised countries have laws and directives which were specified in their national guidelines on the ergonomic design of VDU workstations. While these often refer to specific standards, such as IOS 9241-303:2011, they are usually not a compulsory prerequisite and are often inconsistent. 12 Table 1 compares international and national standards as well as the respective current guidance documents from various industrialised nations regarding visual factors at the VDU.
Visual ergonomic factors have also been part of research. For instance, Bababekova et al. 17 determined that the font sizes on smartphones and tablets are generally very small. Legge et al. 18 showed that reading speed slowed down significantly when the font size was set smaller than 18 angular minutes. However, in a study by Bartha et al., 19 the average character height was 14.6 angular minutes (n = 47). They hypothesised that physical and visual discomfort would be reduced if the font size was set between 16 and 20 angular minutes. Ko et al. 20 demonstrated an increase in productivity and task accuracy with increasing font size. Fostervold et al. 21 demonstrated that there were fewer musculoskeletal complaints with a viewing angle of 30° compared to 15° below the horizontal viewing axis (fixed at the eye-ear line). Bartha et al. 19 stated that presbyopic computer operators required a significantly lower monitor position than non-presbyopes, and they stated that this could only be achieved in part by reducing the height of the desk, which would most likely correspond to a nonergonomic intervention in the workplace design. Both Weidling and Jaschinski 22 and Jaschinski et al. 23 showed that even small reductions in gaze angle lead to lower levels of musculoskeletal complaints.
However, Long et al. 24 determined that modern workplace designs change rapidly. For example, the introduction of handheld devices such as tablets and smartphones make most guidelines no longer applicable, and newer standards should at least attempt to anticipate future developments. As an extreme example, in 2019, Hewlett-Packard and ASUS introduced two ultrawide monitors having a 65″ screen diagonal, which offer more than seven times the screen area of a 24″ monitor in 16:9 format.
The purpose of this field study was to obtain up to date information on workplace ergonomics and VDU set-up, considering factors such as the number of displays, the type and size or head inclination. In addition, these findings were compared with the existing standards and guidelines related to visual factors. In a previously conducted study, 25 the authors assumed that the lower limit of the recommendation for character size (i.e., 22 angular minutes) was too large and did not correspond to real field conditions. Therefore, the primary aim of this study was an analysis of the character size for the habitual tasks of computer workers in their offices and comparison to the recommended limits. The interventional part of this study evaluated the impact on subjectively estimated work productivity when computer workers switched to the limits of the recommended character sizes.

METHODS
The study was conducted between February and August 2019. This research was reviewed on 21 January 2019 by the independent ethical review board of the University Hospital Jena, Germany (2018-1262_1-BO) and conformed with the principles and applicable guidelines for the protection of human subjects in biomedical research. The study followed the tenets of the Declaration of Helsinki.

Study population
In total, 152 VDU workers (47 males and 105 females) with a mean (SD) age of 41 years (11) participated in this study. Sixty-eight subjects (44.7%) were older than 43 years and hence categorised as presbyopic. Twenty-two presbyopic subjects (32.4%) wore general purpose progressive addition lenses, 13 used specialised progressive addition lenses for computer work (19.1%), 1 wore multifocal contact lenses and the remaining 32 subjects (47.1%) did not wear any correction for presbyopia.
Computer workers were recruited from various branches of local and national companies (e.g., hospitals, industry,

Key points
• In the workplace, the habitual font size is set both statistically and clinically significantly too small. • Increasing the character size to 22 angular minutes produced a significant decrease in subjectively rated productivity. • No significant correlation was found between symptoms of computer vision syndrome and failing to meet the recommendations for character size.
information technology, logistics and engineering offices) as well as public institutions (e.g., patent offices, universities and labour offices). All subjects had normal binocular vision and worked for at least 4 h/day and 4 days/week with a regular VDU. Exclusion criteria were: • chronic back pain (including reported whiplash within the last 12 months, cervical disk herniation and lumbar disk herniation); • manipulative, osteopathic or physical treatment concerning musculoskeletal disorders within the last 12 months; • temporomandibular joint dysfunction; • self-reported or diagnosed repetitive strain injury; • dizziness and/or ear noise; • diabetes mellitus; • previously diagnosed mental illness (e.g., depressions, anxiety and insomnia).

Study procedures
Individual workplace characteristics and ergonomic conditions were recorded in a single visit at the subject's workspace by a member of the study team (an optometrist; Figure 1). Demographic data, the optical correction worn in the workplace and self-reported glare due to light reflections from the screen surface were collected using a non-standardised questionnaire during an interview with the subject. The number of monitors, their arrangement relative to the eyes (frontal/side), position relative to the window, physical size (diagonal dimension in inches) and resolution used (vertical and horizontal controllable pixels) including aspect ratio and the properties of their surface (matte, anti-reflective coating, glossy) were recorded by a trained optometrist. All distances and heights were measured with a laser distance meter (S2-40 m, Tacklife, tackl ifeto ols.com) by the study team member.
To measure head inclination, a photograph was taken of each participant and analysed with an image processing program (Inkscape 0.92.4, inksc ape.org). To avoid parallax, a Canon CEOS M100 camera (canon.com) was enhanced with a laser pointer and a height adjustable tripod. Head inclination was defined as the angle between the eye-ear line (running through the right external acoustic meatus and the juncture of the right eyelids) and the horizontal plane when the subject was viewing the display. 22 One standard is the Frankfurt line of head inclination, which is 11° below the eye-ear line. 26 The visual display resolution was obtained from the system settings.
The size of an uppercase E character as used by the computer operator was measured by the study team member using a ruler template 27 (Figure 2) in their habitually used software, that is, the software which was used predominantly by the computer worker. Computer vision syndrome was evaluated using the CVS-Q questionnaire. 28 T A B L E 1 Ergonomic recommendations related to visual factors, depending on country-based international or national standards or national guidelines. Subjectively perceived productivity in the habitual setting was assessed by the subject using a visual analogue scale during an interview. The scale consisted of a slider that could be moved freely by the subject on a scale from 1 (low) to 100 (high) without any intermediate markings. The visual analogue scale was used due its superior metrical characteristics compared with discrete scales. 29 If the character size was below 22 angular minutes, then it was increased to 22 angular minutes (on all displays in case of multi-monitor set-ups) to achieve the lower limits of the recommendation range of the German guideline (Table 1). If the character size exceeded 22 angular minutes, then no intervention was performed. In case of intervention, 2 weeks after the initial visit, a second nonstandardised interview with closed and semi-open questions was performed and recorded as follows:

Standards
• Whether the main work task changed between the first and second visits (closed question: 'Yes' or 'No'; for example, due to changes in the job level; exclusion criteria).
• Whether the subject continued to use the modified character size (closed question: 'Yes' or 'No'

Statistical analysis
Data were analysed using IBM SPSS statistics version 23.0 (ibm.com). Descriptive statistics were used to analyse individual workplace characteristics and ergonomic conditions. Due to the large sample size, differences were evaluated using the independent t-test and the dependent t-test for paired samples. Associations were analysed using Spearman correlations. An α level of 0.05 was applied to all data to determine significant differences. The effect size d z and the power were calculated individually from differences (mean and SD of difference) using a post hoc analysis with the software package GPower (psych ologie.hhu.de/arbei tsgru ppen/allge meine -psych ologi e-und-arbei tspsy cholo gie/gpower). 30

Prevalence of computer vision syndrome
The total score for the CVS-Q ranged from 0 to 18 with a mean (SD) of 6.1 points (3.8). Values greater than or equal to the cut-off score of 6 points were found in 52% of the subjects. These individuals were considered to be symptomatic.

Workplace set-up
Display position, type, number, height and head inclination Descriptive data of the number of displays and their size, aspect ratio, type of surface, position of the window and location relative to the eye are shown in Table 2 Figure 3). However, in 23% of cases, the top line displayed on the monitor was above eye level and therefore too high. The CVS-Q score for subjects with displays above eye level was 1.8 points higher than for subjects with displays below eye level. This difference was not significant (95% CI; −3.69 to 0.166, t(150) = 1.81, p = 0.07, unpaired t-test).   Glare was perceived subjectively in 27.6% of all display surfaces; most frequently with glossy surfaces (85.7%) followed by those with an anti-reflection coating (31.8%) and matte surfaces (23.6%). Perceived glare was linked with a 3.4-point higher score on the CVS-Q (95% CI; 1.66-5.16, t(150) = 3.85, p < 0.001, unpaired t-test).

Desk
Descriptive data of the desk and chair are shown in Table 3.
In the 136 subjects using a chair with armrests, on average, the armrest was 41.3 mm lower than the height of the [Correction added on 16 June 2023, after first online publication: The above heading level has been changed from "Desk and illuminance" to "Desk"]

Character size
The means (SD) character size was 14.29 angular min (3.53), which is significantly smaller than all cited recommendations/guidelines (Tables 1 and 4).
Half of the sample (76 subjects) assumed compliance with the recommended limits. However, compliance was only confirmed in 4% of all cases (Figure 4).
The difference between the individually measured character size and the lower limits of the German guidelines (22 angular minutes) was defined as the deviation in character size. It was analysed when a smaller character size, and therefore an increased deviation, led to higher values in the CVS-Q. However, there was no statistically significant correlation (r = −0.06; p = 0.45; N = 152) between the character size and the CVS-Q score. The same analysis for the age clustered subgroups of non-presbyopic (≤42 years; r = −0.01; p = 0.92; N = 84) and presbyopic (≥43 years; r = −0.14; p = 0.26; N = 68) computer users showed comparable results. Further, there was no significant correlation between the display size and character size (Pearson: r = −0.03, p = 0.75).

Intervention and impact on subjectively perceived productivity
In 52 cases, the character size of the habitually used software could not be manipulated by the end user. In the T A B L E 2 The number of displays, display size (diagonal dimension), aspect ratio, type of surface, position of the window and location relative to the eye (N = 152, unless otherwise specified). remaining 100 cases, it was possible to increase the character size to achieve the recommended range. After this intervention, the mean subjectively estimated productivity decreased from 90.2 to 66.8 points (95% CI; 16.90-30.06, t(99) = 7.08, p < 0.001; Figure 5). There was no correlation between the loss of subjectively perceived productivity and display size (r = −0.04, p = 0.70) or age (r = −0.11, p = 0.27). Additionally, there was no significant difference in the loss of subjectively perceived productivity in relationship to the onset of presbyopia (2.31, 95% CI; −11.1 to 15.7; t(98) = 0.34; p = 0.96).
After working for 2 weeks with an increased character size, 54% of the 100 subjects receiving this intervention found working more comfortable, and 48% retained the adjusted character size. There is a significant relationship between both variables (χ 2 (1, N = 100) = 27.6, Cramer's V = 0.53). In all other cases, subjects decreased the character size within the test period. The main reason for returning to smaller sizes was an incompatibility with the work task (71% of all cases).

Character size and subjectively perceived productivity
The habitual character size fell below the minimum ergonomic dimensions as well as the recommendations of the ISO 9241-303:2011 8 and ANSI/HFES 100-2007 10 and the national guidelines of Sweden 16 and Germany. 13 This finding is in line with the study of Bartha et al. 19 However, a large percentage of the users were not aware that they were undercutting the recommended range. It is also alarming that in about one third of cases, the software did not allow a change in font size or the use of zoom functions. Users did not have the option to work within the recommended limits. This discrepancy between believed compliance with recommendations from guidelines or standards for character size and the actual working conditions shows a marked need for training in the ergonomic set-up of individual VDU workstations.
There is no fixed relationship between the physical size (diagonal in inches) of the display and the addressable T A B L E 3 Descriptive data regarding the height of the desk and chair (N = 152, unless otherwise specified).

Lower limit of recommendation (angular minutes)
One-sample t-Test number of points on the screen. There are some particularly common combinations, for example, 24″ and 1920 × 1080 pixels, and often the number of addressable pixels increases with an increase in the screen diagonal, for example. 27″ size and 2560 × 1440 pixels, so that the physical size of each pixel and therefore the number of pixels per inch (PPI) remains constant. For some years now, there have been high-resolution office displays with a significantly larger number of addressable pixels (3840 × 2160 pixels), and thus, a much higher PPI, especially for screen diagonals >24″. With these monitors, there is the risk that the content be displayed at a smaller angle without the user's intervention. Microsoft Windows (Micro soft.com), for example, offers a function called 'DPI scaling' so that the characters can be displayed correspondingly larger on high-resolution displays. It seems that these functions help to keep the character size constant across different sized monitors. However, users with larger monitors are able to increase the character size further, as the indentation of the overview is smaller here due to the larger display area. But no such trend was found in these data. Bartha et al. 19 showed that a character size between 14.2 and 16.8 angular minutes was associated with less selfreported eye strain compared with visual angles between 8.5 and 12.6 angular minutes. However, in the present study, there was no significant correlation between the validated CVS-Q score and character size.
Approximately, half of the sample perceived working with character sizes within the recommendations as more comfortable and would permanently consider using a larger font size, albeit slightly smaller than the suggested 22 angular minutes. Given that 37 out of 100 subjects rated this intervention as being incompatible with their work task, the findings suggest that a character size of 22 angular minutes is too large for readability. The greatest difficulties were reported by subjects working with spreadsheet processing programs, such as Microsoft Excel (Micro soft.com).

Workplace set-up
The standard computer workstation in this study consisted of two non-glare (matte) 24″ widescreen monitors that were located approximately 73 cm (primary) and 76 cm (secondary) in front of the eyes. Most of the subjects were working under the ergonomic recommendations of ISO 9241-5. 9 Although this (currently valid) ISO standard is over 20 years old, there are still many workplaces that failed to meet the recommendations: • Approximately, 20% of the sample positioned the workstation/monitor so that they had a window either in front of or behind them. There is a risk of glare from reflections even on non-glare type displays, which leads to compensatory and forced postures that were linked with an increase in CVS in this study. • Approximately, 23% set their monitor height too high and increased the risk of working in a forced posture due to a need for an excessive reclined position. • On average, armrests were set significantly too low in this study. The height of the armrest, just like table height, should be set so that the bent arms rest loosely. This means that both parameters should have the same height. A lower armrest poses a risk of contact stress, as the forearm presses against the (usually) hard edge of the desk.   • Although VDU work is linked with a higher risk of neck, shoulder, arm, wrist and hand musculoskeletal disorders, 31 only 25% of the sample had a height-adjustable desk, which has been shown to reduce the severity of musculoskeletal discomfort for most upper body regions. 32 While presbyopic computer user may be restricted by the zone of clear vision of their corrective lenses, there was no difference in the distance between the eye and the monitor for presbyopic and non-presbyopic subjects. This could be due to compensatory backward tilting of the head to use the zone of clear vision for their monitor, as found in previous studies. 23,33 Limitations of this study One aim of this study was to collect a snapshot of the visual-ergonomic state. There are some ergonomic variables that probably do not or never change over the course of the day (e.g., desk height, number of monitors, resolution and display size). Other metrics are likely to be subject to some variability: • Character size may vary due to the use of different software, zoom functions or changes in the work task. However, it should be noted that approximately 33% of the sample did not have the ability to adjust the font size in their habitually used software. • Sitting position, and therefore the distance between the eye and monitor may change markedly due to the ease in resolving the task. However, this study only examined a snapshot of the primary work task. Other demands may involve smaller or larger characters and thus influence the sitting position. There is a risk that the use of characters that are too small will cause the worker to move closer to the screen, resulting in an unergonomic forward-leaning posture. • An objective measurement of productivity based on standardised work tasks, as well as a time measurement would have allowed a more precise statement about the actual change in productivity. Such measurements are not trivial in the field. The additional effort involved would have been disproportionately high for a secondary end point. Therefore, the authors opted for subjective estimation, even though this was not a validated method.
For more precise information on highly variable ergonomic factors, long-term measurements are desirable. The evaluation procedure cannot guarantee that VDU workers will occupy exactly the same distance or head inclination in a non-observed situation. Therefore, the distance between the eye and the monitor as well as head inclination could be determined with proximity and position sensors, which are attached to a spectacle frame temple. Modern eye trackers in the form of glasses provide additional information on gaze deflection without fundamentally changing the habitual posture. Character size and illuminance levels could be determined several times a day.
The reported changes in productivity refer to a subjective estimation by the subjects. It is conceivable that an objective change in productivity would have a different magnitude. Evaluating a productivity change in the field objectively, with such a large and, moreover, heterogeneous group is anything but trivial. Nevertheless, a follow-up study that measures and evaluates the productivity change due to differences in character size objectively using a standardised work task would be desirable.
The data collected in this study were obtained before the Coronavirus (COVID-19) pandemic. However, the onset of this disease has had a massive impact on labour practices. Protective measures such as stay-at-home orders increased the percentage of people working exclusively from home in the USA from 8.2% in February 2020 to 35.2% in May 2020. 34 Similar data were collected in a representative survey of 1503 participants in Germany 35 : the respective proportion of employees exclusively and partially working from home was 3% and 15% before the pandemic, 25% and 20% during the pandemic and is expected to be 8% and 27% after the pandemic. There is strong evidence that workplace ergonomics are much worse in the home environment. 36 According to Davis et al., 37 only 58% of their sample used an office chair, with the remainder sitting on a dining chair, couch or even in bed. In Germany, working from home is often not contractually regulated (partly due to the short-term nature of the intervention). 38 As a result, in the domestic environment, workplace design is the responsibility of the employees, for whom a lack of financial resources may make ergonomic workplace design difficult. 38 The shift of work to the home environment makes it difficult to check and adapt conditions to comply with the standards and guidelines analysed in this paper. Therefore, it is recommended that VDU workers increase their knowledge of ergonomic workplace design and self-organisation. 38

CONCLUSION
The study identified the components that comprise a VDU workstation in the field and how it they are set-up ergonomically. Most VDUs met the published recommendations. However, a considerable number of workplaces violated basic ergonomic principles, such as proper monitor and armrest height.
The results clearly show that the habitual font size of computer workers is both statistically and clinically significantly below the recommendation range of ISO 9241-303:2011 and any other cited standard or national guidelines. Many subjects worked with larger characters sizes following their participation in this study, although still smaller than the recommended 22 angular minutes of ANSI/HFES 100-2007 or the German guidelines. Therefore, character size should not generally be increased to this recommendation limit. A font size of 22 angular minutes seems to be too large for readability in the context of real tasks and is often incompatible with the work demand. This might be reasoned historically since the requirements stem from an era of low-resolution electronic displays and are mostly based on legibility studies. 39 Everyday screen work often requires the use of more than one type of software. Even individuals whose primary role is data processing write reports, e-mails or search the Internet between tasks. Therefore, the worker should always check whether the current task can be performed at higher magnification using the software's zoom function to minimise strain on the visual system. The real visual conditions observed in this study can be used to conduct investigations on model VDUs in the laboratory under more realistic field conditions.

AC K N O W L E D G E M E N T S
Open Access funding enabled and organized by Projekt DEAL.

FU N D I N G I N FO R M AT I O N
This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

CO N F L I C T O F I N T E R E S T S TAT E M E N T
All the authors have no conflicts of interests to declare.