The purpose of this study was to systematically review prospective cohort studies to gain insights into risk factors for the development of non-specific neck pain in office workers as well as to assess the strength of evidence. Publications were systematically searched from 1980 - March 2011 in several databases. The following key words were used: neck pain paired with risk or prognostic factors and office or computer or visual display unit or visual display terminal. Relevant studies were retrieved and assessed for methodological quality by two independent reviewers. The strength of the evidence was based on methodological quality and consistency of the results. Five high-quality and two low-quality prospective cohort studies investigating the predictive value of 47 individual, work-related physical and work-related psychosocial factors for the onset of non-specific neck pain in office workers were included in this review. Strong evidence was found for female gender and previous history of neck complaints to be predictors of the onset of neck pain. Interestingly, for a large number of factors that have been mentioned in the literature as risk factors for neck pain, such as high physical leisure activity, low social support, and high psychosocial stress, we found no predictive value for future neck pain in office workers. Literature with respect to the development of non-specific neck pain in office workers is scant. Only female gender and previous history of neck complaints have been identified as risk factors that predict the onset of neck pain.
- Visual display unit
- neck pain
- systematic review
- health promotion
- exposure assessment
Statistics from Altmetric.com
- Visual display unit
- neck pain
- systematic review
- health promotion
- exposure assessment
Non-specific neck pain is neck pain (with or without radiation) without any specific systematic disease being detected as the underlying cause of the complaints.1 Neck pain is a significant health problem in workers with office workers among those with the highest frequency of neck pain.2 Between 42% and 69% of office workers experienced neck pain in the preceding 12 months3–6 and about 34%–49% reported a new onset of neck pain during a 1-year follow-up.7–9 Neck pain is viewed as an episodic occurrence over a lifetime with variable recovery between episodes.10 In a working population, 60%–80% of workers with neck pain also report neck pain 1 year later.11 Neck pain causes considerable personal suffering due to pain, disability and impaired quality of work and life in general, which can be a great socio-economic burden on both patients and society.2 ,12–14
Evidence suggests that neck pain in workers is non-traumatic and assumed to be of multi-factorial origin.2 The relationship between risk factors and neck pain is a complex one, meaning that neck pain is likely to be caused by multiple serial exposures rather than by the direct effect of a single exposure.2 ,15 In the past 10 years, a number of systematic reviews have been conducted about risk factors for neck pain in non-specific groups of population.2 ,16–18 The Neck Pain Task Force2 proposed that different occupations are exposed to different working conditions and the nature of work has influenced the health of workers. Thus, predisposing factors for neck pain are likely to be occupation-specific.
Office work is sedentary work, which mainly involves computer use, participation in meetings, giving presentations, reading and telephoning.19 Office work may require sitting for long hours on a computer, working in awkward positions or performing repetitive manual tasks. Studies have identified several individual factors associated with neck pain in office workers, including older age, female gender, high body mass index, lack of physical exercise, smoking, alcohol consumption and previous symptoms.3 ,8 ,20 Work-related risk factors, such as accumulated computer usage, sitting for long periods or with forward head posture, and poor workstation ergonomics, have been linked to increased risk of neck pain.3 ,6 ,8 ,20–22 Some psychosocial problems, such as high stress, high job demands, job strain and low coworker support, have also been associated with neck pain.7 ,8 ,21–24 However, a number of these studies were cross-sectional in design,3 ,6 ,20 ,21 ,23 ,24 which only allowed for the association between exposures and outcome to be examined. It is therefore not possible to establish the causal relationship between exposures and outcome. Research to identify the risk factors of neck pain requires longitudinal research design, which permits the tracking of study participants over time.25
Thus, the aim of this paper is to systematically review prospective cohort studies to gain insights into risk factors for the development of neck pain in office workers as well as to assess the strength of evidence. Such information would be of value for policy makers and healthcare providers to determine effective prevention measures for decreasing the incidence and burden of neck pain in the workplace.
Data sources and search strategy
Online searches were conducted on PubMed, CINAHL Plus with full text, The Cochrane Library, ScienceDirect, PEDro, ProQuest and Scopus databases from 1980 to March 2011 using the following keywords: neck pain paired with risk or prognostic factors and office or computer or visual display unit or visual display terminal. Articles were initially screened on the basis of title and abstract, and full text copies were then retrieved of articles that met all inclusion criteria. Subsequently, full text copies were read in order to make a final decision regarding inclusion or exclusion. The search and full inclusion process was performed by one reviewer (AP). After inclusion of the articles based on the selection criteria, references were searched for additional articles.
Selection of studies
A reviewer (AP) selected relevant articles from the articles retrieved using the search strategy. The selection criteria were:
The study population was office workers or those working with computers or visual display units or visual display terminals.
Study samples were free from neck pain at baseline assessment. Studies in a population with specific underlying pathology, such as tumours, fractures, infection, inflammatory disorders and osteoporosis, were excluded.
The study design was a prospective cohort study with a follow-up period of 1 year or more. Experimental studies were excluded.
The onset of neck pain was assessed separately from other musculoskeletal symptoms.
Non-specific neck pain, that is, neck pain (with or without radiation) without any specific systematic disease being detected as the underlying cause of the complaints, was assessed in the study. Studies on whiplash-associated disorder were excluded.
The article was a full, peer-reviewed report published in English. Letters, abstracts, books, conference proceedings and posters were excluded.
Quality assessment of studies
The articles that met the selection criteria were independently evaluated by two reviewers (AP and NP) to determine methodological quality. The methodological quality of each study was assessed by using the 21-item checklist for quality appraisal developed by van der Windt et al26 and Ariëns et al16 (table 1). The checklist was divided into two parts, the internal validity (11 items) and descriptive quality (10 items) of studies. Each item was scored as positive (1), negative (0) or unclear (if insufficient information was available for a specific item) (0). The scoring for each item of the two reviewers was compared. Disagreements between the reviewers on individual items were identified and discussed in an attempt to achieve consensus. The inter-rater agreement of this quality assessment was derived by calculating the percentage agreement as well as Cohen's κ for categorical items, both before and after the consensus discussion. If agreement could not be reached, a third reviewer (PJ) was consulted to achieve a final judgement. Studies scoring a minimum of 6/11 (>50%) for internal validity with a total score of 11/21 (>50%) or greater were deemed ‘high quality’.16 ,26–29
Data extraction was performed by the first reviewer (AP). For each article, the first author and year of publication, study population, sample size, drop-out rate, outcome measured (pain, disability), duration of follow-up, risk factors, the strength of the association between risk factors and the onset of neck pain in terms of OR, HR or RR with their 95% CI were extracted.
The strength of evidence for risk factors associated with the development of non-specific neck pain was assessed by defining five levels of evidence based on the number of studies and the quality score of studies:17
Strong evidence: consistent findings from two or more high-quality cohorts.
Moderate evidence: consistent findings from at least one high-quality study and one or more low-quality cohorts.
Limited evidence: findings of one high-quality study or consistent findings in one or more low-quality studies.
Conflicting evidence: inconsistent findings irrespective of study quality.
No evidence: no studies found.
A risk factor association was considered positive only if it was statistically significant and was derived from multivariate results. A risk factor association was considered negative only if it was statistically insignificant and was derived from multivariate results. Statistical significance was concluded if the reported p value was <0.05 or if the 95% CIs around a RR or similar statistic (such as OR or HR) did not cross 1.
Sensitivity analysis was conducted to assess how sensitive the results of the review were in relation to the way it was performed. First, the effect of the cut-off point used in the methodological quality assessment for qualification as a high-quality study on the synthesised results was assessed by shifting the cut-off point from >50 to >60% or shifting the cut-off point from >50 to >70%. Second, the effect of the inclusion of low-quality studies on the synthesised results was assessed by repeating the analysis using only high-quality studies.
Selection of studies
The initial search of the computerised databases yielded 7982 citations (figure 1). After the screening of abstracts and titles, 35 full text articles were read in full. Twenty-eight articles were excluded because they did not meet the selection criteria. A total of seven articles were judged to meet the selection criteria and were included in the methodological quality assessment.7–9 ,22 ,30–32
Methodological quality assessment
The scoring of two reviewers of the included studies before discussion had an agreement rate of 85% (125/147). The overall inter-rater agreement was k=0.66 with an SE of measurement of 0.07. After discussion, the two reviewers had an agreement rate of 99% (145/147). Then, the overall inter-rater agreement was k=0.98 with an SE of measurement of 0.02. This represents very good agreement between the two reviewers.33 Disagreements were often related to reading errors or interpretation of the quality criteria list. These disagreements were resolved during a consensus meeting. However, disagreements persisted on two items (items 9 and 18) in the studies from Brandt et al31 and Hush et al7 A third reviewer (PJ) made the final decision in these cases.
The results of the methodological quality appraisal are presented in table 2. The scores for the methodological quality of the studies ranged from 10 to 14 points (48%–67%). The median score was 14 points (67%). Five studies were scored as high-quality studies,9 ,22 ,30–32 while two studies were scored as low-quality studies.7 ,8 The items in the criteria list rated as negative in most studies were participation rate (item 1: 29%), assessment of physical load at work (item 3: 14%), quality of assessment method for physical load at work (item 15: 14%), assessment of exposure during leisure time (item 18: 14%) and frequency of data collection during follow-up period (item 20: 29%).
All included studies were conducted on office workers or computer users (table 3). The sample sizes varied greatly from 53 to 6943. The drop-out rate during follow-up ranged from 0% to 23%. Five studies defined incident cases as those experiencing neck pain or discomfort for the duration of at least 1–8 days during the study period, whereas one study defined incident cases as those experiencing neck pain in the past 7 days and pain in the past year with at least moderate disability. The remaining one study did not specify the duration of experiencing neck pain. Four studies followed up for 12 months and the remaining three followed up for 17–24 months.
Summary of risk factors
Risk factors were divided into three groups: individual, work-related physical and work-related psychosocial risk factors (table 4). A majority of factors (74%) were investigated by only one study. There was strong evidence that female gender and previous history of neck complaints are predictors of the onset of neck pain. Strong evidence was also found that high keyboard usage time, poor perception of computer placement and low social support have no predictive value for the onset of neck pain. Moderate evidence was found that high physical leisure activity and high psychosocial stress have no predictive value for the onset neck pain. There was limited evidence that pain started after an accident, irregular head and body posture, duration of employment in same job <1 year (for males only), poor computer skills (for males only), distance of the keyboard from the edge of the table <15 cm, high task difficulty, low influence at work (for female subjects only) and high muscular tension are associated with the onset of neck pain. There was also limited evidence that high/low body mass index, chronic diseases, smoking, cervical flexion–extension or lateral flexion mobility, arm support during mouse and keyboard use, poor perception of office equipment position, poor physical work environment, awkward body posture, high average mouse activity per 10 min, high average keyboard activity per 2 min, high mouse or keyboard speed, low micro-pauses per min (for mouse or keyboard use), high work flow, high physical exposure, sitting duration before break >1 h, poor social network, non-adjustable chair and desk, low decision authority, low skills discretion, low control and Type A behaviour have no predictive value for the onset of neck pain. Conflicting evidence was found for factors, such as older age, daily computer use, high mouse usage time, screen height above eye level, high job strain and high demand.
Changing the cut-off point from >50 to >60% would not have altered our conclusions at all. With a cut-off point of >70%, there would have been no study in high-quality status. By excluding low-quality studies (with a cut-off point of >50%), several conclusions would be altered including:
The level of evidence for high physical leisure time activity and high psychological stress would change from moderate ‘No’ to limited ‘No’.
The level of evidence for smoking, cervical flexion–extension mobility, cervical lateral flexion mobility, poor physical work environment and sitting duration before break >1 h would change from limited ‘No’ to no evidence.
The level of evidence for distance of the keyboard from the edge of the table <15 cm would change from limited to no evidence.
This review summarised the results of five high-quality and two low-quality prospective cohort studies investigating the predictive value of 47 individual, work-related physical and work-related psychosocial factors for the onset of non-specific neck pain in office workers. Because of heterogeneity among studies mainly regarding case definition, risk factors, outcome measure and follow-up duration, the analysis of the results was limited to a qualitative summary. Strong evidence was found for female gender and previous history of neck complaints and limited evidence for pain started after an accident, irregular head and body posture, duration of employment in same job <1 year, poor computer skills, distance of the keyboard from the edge of the table <15 cm, high task difficulty, low influence at work and high muscular tension as predictors for new-onset neck pain in office workers. Interestingly, for a large number of factors that have been mentioned in the literature as risk factors for neck pain, we found no predictive value for future neck pain in office workers, such as high physical leisure activity, low social support and high psychosocial stress.
In this review, studies solely investigating neck pain were included. The area of neck is usually defined according to the standardised Nordic questionnaire34 or a region bounded superiorly by the superior nuchal line, laterally by the lateral margins of the neck and inferiorly by an imaginary transverse line through the T1 spinous process.35 Clinically, symptoms in the shoulder region may be the result of injuries in the neck and/or shoulder regions. Evidence suggests that risk factors for neck and shoulder pain in the general population are not identical.17 ,26 Thus, an exclusion of those studies investigating neck and shoulder symptoms as a single region from this review would increase homogeneity among included studies, increasing internal validity of the findings.
Of the seven included studies, the items in the criteria checklist rated as negative in most studies were participation rate, assessment method of physical load at work and its quality, assessment of exposure during leisure time and frequency of data collection during follow-up period.
Of the seven included studies, only two studies had a participation rate of ≥80%.8 ,9 The participation rates of the remaining five studies varied considerably, ranging from 1% to 73%.7 ,22 ,30–32 In general, studies with low levels of participation may be more vulnerable to self-selection bias than those with high participation.36 Therefore, a low participation rate in a population survey may threaten the internal validity of studies.37
Common methods for the assessment of physical exposures at work include subjective judgement, systematic observation and direct measurement.38 ,39 Most studies employed a self-reported questionnaire to assess physical load at work.7–9 ,22 ,30 ,31 Only one of the seven included studies using a software program to assess physical load at work.32 Many of the subjective methods, particularly those non-standardised methods of acceptable quality, had problems with test–retest reliability, which may have led to a poor validity of exposures.40 ,41 All included studies using self-reported questionnaires did not report the test–retest reliability of their measurement tools. Future research should attempt to use a reliable systematic observation or objective measurement, instead of subjective judgement, to evaluate physical load at work.
Of the seven included studies, only one study measured physical exposures during leisure time.24 The remaining six studies did not measure exposures during leisure time9 ,22 ,30 ,32 or did not clearly state when exposures were measured.7 ,8 Apart from work time, exposures during leisure time should be assessed and included as part of the cumulative dose that an individual is exposed to. Hildebrandt et al42 demonstrated the association between physical activity during leisure time and neck pain in the working population, especially in office workers. Future research should consider measuring exposure during work and leisure time in order to be more representative of an individual's exposure.
The frequency of data collection of neck pain incidence during the follow-up period for the included studies varied considerably, ranging from 2 weeks to 24 months. Of the seven studies, one collected data every fortnight7 and one collected data every month.9 The rest of the studies collected data at the beginning and the end of study only.8 ,22 ,30–32 A longer recall period regarding the incidence of neck pain during follow-up period may increase recall bias.43 This bias may be pronounced in studies in which detailed information, such as the duration of experiencing pain and/or pain intensity, was required. Future studies should pay more attention to the frequency of data collection during their follow-up period, and it is recommended that data are collected at least every 3 months or are obtained from a continuous registration system.
To date, there is a lack of consensus over the definition of a new episode of neck pain.17 In this review, the onset of neck pain was considered to be the onset of any reported neck symptoms, regardless of severity of symptoms, duration of symptoms and level of disability. This pragmatic choice was made because of the fact that this review focused on a specific group of population and only a small number of studies were qualified to be included in the review.
Of the seven studies, four were followed up for 12 months7 ,8 ,31 ,32 and the remaining three were followed up between 17 and 24 months.9 ,22 ,30 The predictive value of any exposure depends on the duration of follow-up as well as the disease of interest. A long duration of follow-up is generally considered a strength in prospective cohort studies, as it usually results in a larger number of cases and thereby increases the power of the statistical analysis.44 Thus, a long duration of follow-up is likely to enhance the internal validity of the study.
Evidence of risk factors for the onset of neck pain in office workers
In the general population, McLean et al17 systematically reviewed prospective cohort studies and found strong evidence that female gender, older age (for men only), high job demands, low social/work support, being an ex-smoker, a history of low back problems and a history of neck problems were risk factors for new-onset neck pain in the general population. In the working population, Côté et al2 in their systematic review of prospective cohort and randomised controlled studies found strong evidence for older age, previous musculoskeletal pain, high quantitative job demands, low social support at work, job insecurity, low physical capacity, poor computer workstation design and work posture, sedentary work position, repetitive work and precision work. The predictive value of several factors identified in previous reviews could not be confirmed in the present review, which only showed strong evidence for female gender and history of neck complaints as predictors of the onset of neck pain among office workers. The observed variation in the results among studies may be due to the limited number of studies in a population of office workers. However, the findings shed some light on the notion that risk factors for the onset of neck pain in a subpopulation may be a subset of risk factors identified in a general population or occupation-specific.2 To gain further insight into risk factors for the development of neck pain, future studies should consider the investigation of risk factors in a more specific group of population. Although both gender and history of neck complaints are non-modifiable risk factors, this information is useful for clinicians to identify office workers at risk, which would mean the enhancement of resource allocation to those most in need and most likely to benefit from it. Otherwise, a large number of people would receive intervention, which is likely to compromise its effectiveness.45 ,46 Most of the variables included in the review have been supported by evidence from one high-quality study only. Thus, this review was limited in its ability to draw conclusions about the predictive nature of these variables and the conclusions may change or modifiable risk factors will be identified when new studies become available in the future. In addition, there are still several other variables that have not been investigated at all. Therefore, further prospective studies to investigate biopsychosocial risk factors for the development of neck pain in office workers are still required.
Since all high-quality studies had total scores of greater than 60%, changing the cut-off point from >50 to >60% would not have altered our conclusions at all. However, shifting the cut-off point from >50 to >70% would have led to no study qualifying as a high-quality study.
By excluding low-quality studies, several conclusions about risk factors with moderate and limited evidence, namely, high physical leisure time activity, high psychological stress, smoking, cervical flexion–extension mobility, cervical lateral flexion mobility, poor physical work environment, sitting duration before break >1 h, and distance of the keyboard from the edge of the table <15 cm, would alter.
This variation in the level of evidence reflects the fact that there have been a small number of very good quality studies investigating risk factors for the development of neck pain in office workers. Thus, further study is required before firm conclusions can be drawn.
Strengths and limitations of the study
The major strength of this review is that the studies were systematically searched, evaluated for their methodological quality by two independent reviewers, extracted and synthesised based on the number of studies and the quality score of studies. However, two main methodological limitations are noteworthy. First, the search strategy was limited to full reported publications in English. The possibility of publication and selection bias cannot be ruled out. This may have affected the results of this review. Second, the researchers summarised the results from studies with substantial heterogeneity in study characteristics. This may explain the observed variation in the results among studies. Future research is required to indicate whether differences in these aspects affect the effectiveness of exercise intervention before direct comparisons among different programmes can be conducted.
Five high-quality and two low-quality prospective studies on the association between risk factors and the onset of non-specific neck pain in office workers were reviewed and analysed. The findings showed strong evidence for female gender and previous history of neck complaints as risk factors of the onset of neck pain. Furthermore, we found strong evidence for the following factors not having predictive value: high keyboard usage time, poor perception of computer placement and low social support. The results of this review need to be interpreted with caution because most variables have been investigated by only one study. More high-quality studies in this area are needed.
Funding This work was funded by Chulalongkorn University Centenary Academic Development Project (12) and Faculty of Allied Health Sciences, Chulalongkorn University.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.