Can we distinguish the roles of demographic and temporal changes in the incidence and prevalence of musculoskeletal disorders? A systematic review

Objectives Musculoskeletal disorders (MSD) represent a major public health issue, affecting more then 40 million European workers in 2017. The overall aging of the working population is expected to increase the burden of disease, but temporal changes in exposures or diagnosis may also drive the global trends in MSD. We therefore conducted a systematic review to summarize the evidence on the role of demographic and temporal changes in the occurrence of MSD. Methods We conducted a systematic review of articles reporting temporal trends in MSD in the general working-age population. Only articles controlling for age in the analysis were included. The risk of bias was assessed. The main indicators extracted were age-controlled time trends in MSD incidence or prevalence. Results Among 966 articles, 16 fulfilled the inclusion criteria, representing 23 results according to the indicators extracted. No study was found with a high risk of bias. Results presenting time trends in prevalence were found in 12 studies and incidence in 11. After controlling for age, the reported temporal trends varied, mostly between non-monotonic changes (N=12/23) and increases (N=10/23). One article also highlighted an increase among women and non-monotonic changes among men (N=1/23). Several factors other than aging were suggested to explain temporal trends in MSD, mainly trends in obesity, changing occupational exposures, and cultural factors regarding pain tolerance. Conclusion This review shows that different kind of factors in addition to aging may contribute to varying or increasing trends in MSD. This review also highlighted the scarcity of evidence regarding time trends in the burden of MSD and their underlying causes.

Checklist item Location where item is reported 13b Describe any methods required to prepare the data for presentation or synthesis, such as handling of missing summary statistics, or data conversions.
13c Describe any methods used to tabulate or visually display results of individual studies and syntheses. Line 140-144 & 223-224 13d Describe any methods used to synthesize results and provide a rationale for the choice(s). If meta-analysis was performed, describe the model(s), method(s) to identify the presence and extent of statistical heterogeneity, and software package(s) used.
13e Describe any methods used to explore possible causes of heterogeneity among study results (e.g. subgroup analysis, meta-regression). Not applicable here 13f Describe any sensitivity analyses conducted to assess robustness of the synthesized results.
Reporting bias assessment 14 Describe any methods used to assess risk of bias due to missing results in a synthesis (arising from reporting biases). Not applicable here Certainty assessment 15 Describe any methods used to assess certainty (or confidence) in the body of evidence for an outcome.

Study selection
16a Describe the results of the search and selection process, from the number of records identified in the search to the number of studies included in the review, ideally using a flow diagram.

Austria
In each survey, the presence of back pain was questioned. When collecting data from the Microcensus surveys (1973,1983,1991,1999), participants were asked if they suffered from back pain at the time of the survey. If they have suffered from the disorder in the past 12 months. Surveys have used different definitions to identify back pain. In the last survey, the 12-month prevalence of back pain was collected, while in the first four Microcensus surveys, the point prevalence was measured. Despite the different collection methods of Microcensus and AT-HIS, it has been reported that back pain is often chronic and that there is not much difference in the prevalence of back pain at the time of investigation or rather within 12 months. The back pain data from the AT-HIS 2006-07 did not appear obvious and were similar and roughly matched the data from the Microcensus surveys. Therefore, the effect of the different definitions is probably minimal and reflects only a slight change in the prevalence of back pain over the study period. Guido et al, 2020

Europe -17 different countries
Depending on the country, the definition varies: Pain is included in the state of health and functional limitations. It was measured in 14 of the 17 studies, with different approaches. Some studies, for example, Collaborative Research on Aging in Europe and Longitudinal Study on Health and Retirement in China have approached it in terms of pain severity (e.g. none, mild, moderate, severe, extreme or other similar formats). Other studies, e.g. the Australian Longitudinal Study on Aging and the European Health, Aging and Retirement Survey dichotomously addressed the presence of pain (e.g. yes or no), sometimes addressing the idea that pain is 'often felt', as in the Irish Longitudinal Study on Aging. The harmonization procedure aims to generate inferentially equivalent content across studies to make the content of the variables collected in different studies uniform. For the pain variable, the content was "self-reported pain experienced at the time of the interview", and the modality of the variable was dichotomous. Holte et al, 2003 (6) Norway Rheumatoid arthritis (RA): is a form of arthritis that causes pain, swelling, stiffness, and loss of function in the joints. It can affect any joint, but it is common in the wrist and fingers. More women than men suffer from rheumatoid arthritis. It often begins between the ages of 25 and 55. You may only have the disease for a short time, or the symptoms may come and go. The severe form can last a lifetime. Rheumatoid arthritis is different from osteoarthritis, the common arthritis that often occurs with age. RA can affect parts of the body in addition to the joints, such as your eyes, mouth, and lungs. RA is an autoimmune disease, which means arthritis results from your immune system attacking your body's own tissues.

Sweden
The questionnaires included a question on low back pain, modified from the Standardized Nordic Questionnaire, with a recall period of 12 months (1990 and 1994) or 6 months (1998, 2002 and 2006): the last six (twelve) months? '' The questions had identical alternative answers in each of the five surveys: "No, never", "Yes, a few days in the last six (twelve) months", "Yes, a few days a month", "Yes, a few days a week" and "Yes, every day." From a clinical point of view and for the purposes of this study, low back pain was defined as pain a few days a week or every day.
Martin et al, 2014 (9) Finland Questionnaire on 9 physical functions, pain that induces difficulty in: bending, bending, or kneeling, standing for 2 hours, pushing or pulling a large object; walk a quarter of a mile; climb ten steps; seated 2 hours; lift and carry ten pounds; reach above the head; and grab small items.
Paloneva et al, 2015 Belgium Osteoarthritis of the knee: Criteria: Either imaging with a characteristic appearance; either a joint disorder that has progressed for at least three months, without constitutional symptoms comprising three or more of the following three signs: intermittent swelling, crepitus, stiffness or limitation of movement, speed of sedimentation, rheumatoid tests, normal uric acid; over 40 years. Includes: knee osteoarthritis secondary to dysplasia; knee osteoarthritis secondary to trauma. See the link below to view the source they used for this definition: https://www.hetop.eu/hetop/3CGP/en/?rr=CIP_D_L91&q=CIP_D_L91#rr=CIP_D_L90&q=CIP_D_L90 Swain et al, 2020 (15) United Kingdom The osteoarthritis incident was defined as the first diagnosis of osteoarthritis in each year of study. Prevalent osteoarthritis was defined as having a diagnosis of osteoarthritis on July 1 of each year of study. Read codes were used: a medical coding system for clinical terms used by the National Health Services (NHS), United Kingdom. The Read code list available (www.keele.ac.uk/mrr) to identify people with osteoarthritis diagnosed by general practitioners (GPs) has been adapted according to the inclusion and exclusion criteria of the study. Two types of osteoarthritis were excluded (acromioclavicular and sternoclavicular joints), due to the possible low precision of the diagnosis at the level of these joints and the expected incidence is very low. The codes obtained from the given website were previously mapped to the ICD-10 codes. Yu et al, 2017 (16) United Kingdom Two definitions of osteoarthritis: 1) Defined cases having had at least 1 consultation with a recorded diagnosis of osteoarthritis or, at least 1 consultation with a recorded peripheral joint pain symptom affecting the knee, hip and hand/wrist likely to reflect osteoarthritis (clinical osteoarthritis); 2) Cases defined more narrowly as having at least 1 consultation with a recorded diagnosis of osteoarthritis.

Supplementary material 4: Detail of bias risk and quality in statistical analysis
An important step in systematic literature review methods is to assess the risk of bias of individual studies. Therefore, we adapted the methodology of two tools usually used : the RoB -SPEO tool (17) and the Navigation Guide's evidence quality assessment tool inspired by the article by Alexis Descatha et al, 2018 (appendix H) (18).
From the RoB-SPEO criteria (17) we used bias: 1) in selection of participants into the study, 2) due to misclassification of MSD, 3) due to conflict of interest, and 4) other bias. Thus, we considered 4 biases among 8 since the other bias was not relevant because of their specificity to exposures and measures of risk factors and in our study we only assessed the time trends in MSDs (reminder of the biases that have not been retained because they are not relevant here: bias due to a lack of blinding of study personnel, bias due to incomplete exposure data, bias due to selective reporting of exposures and bias due to differences in numerator and denominator).
In the Navigator guide described in the Alexis Descatha et al, 2018 (appendix H) article (18), most of the criteria were close to RoB-SPEO but we added the relative criterion on confounding factors which was consistent with the needs of our study (bias due to poor consideration of confounding factors).
Classifications of each bias for the selected articles are provided below.
S4.a: Selection bias: potential bias resulting from study groups not adequately representing the population of interest.
• Low: the target population represented the whole working population or general population.
• Probably low: there was insufficient information about participant selection to permit a judgment of low risk of bias, but the worker population used in the study was specific to one category of worker (eg: farmer, worker, executives, etc.).
• Probably high there was insufficient information about participant selection to permit a judgment of high risk of bias, but there is indirect evidence which suggests that inclusion/exclusion criteria, recruitment and enrollment procedures, and participation and follow-up rates were inconsistent across groups, as described by the criteria for a judgment of high risk of bias.
• High: participant selection was based on voluntary participation, or there were indications from descriptions of the source population that risk of selection effects were substantial, whether due to issues in inclusion/exclusion criteria, recruitment and enrolment procedures, participation and follow-up rates, or data on the distribution of relevant study sample and population characteristics. • Low: the diagnosis was made by a medical exam • Probably low: the diagnosis was made by and/or via an interview or auto-questionnaire.
• Probably high: information on the diagnosis was insufficient to judge a high risk of bias, but there was indirect evidence suggest that the symptom identification method criteria, were inconsistent between groups, the bias was considered to be probably high. For example, when the questionnaires distributed according to the groups were not the same each year.
• High: If it was proven that the diagnosis was not an element of the study design capable of introducing a risk of bias into the study, the bias was considered to be high. S4.c: Bias due to incorrectly taking confounding factors into account: Regarding bias due to a poor consideration of confounding factors, this criterion had to be adapted to our study. In our study, we sought to review the articles which deal with the temporal evolution of MSDs by taking age into account to see whether it is possible to distinguish the effect of age from that of time in the occurrence of these pathologies. Therefore, since in our selection procedure, we only selected articles dealing with temporal trends in MSDs by age (and not only raw trends) we do not risk having selected articles for which there will be high or probably high risk of bias (absence of the age factor). We therefore defined 2 categories of evaluations: • Low: when, in addition to the age factor, other confounding factors, the bias was considered to be low. These factors provided a more detailed analysis of temporal trends • Probably low: when only the age factor has been considered, the bias was defined as probably low. The age factor allows to take into account the effect of demographics in the temporal trends of the occurrence of MSDs. S4.d: Bias due to potential conflict of interest: Bias due to potential conflict of interest resulting from support from a company, a study author or other entity with a financial interest were also assessed: • Low: no conflict in the paper was mentioned • Probably low: conflict of interest was not mentioned, but the laboratories that published the papers were affiliated with public research agencies or non-profit scientific institutions • Probably high: there was insufficient information to allow a judgment of high risk of bias, but there was circumstantial evidence which suggests that the study was not free of support from a company, study author or another entity with a financial interest in the outcome of the study, as described by the criteria for a judgment of high risk of bias • High: there was circumstantial evidence to suggest that the study was not exempt from support from a company, study author, or another entity with a financial interest in the outcome of the study