Coggon et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
The CUPID (Cultural and Psychosocial Influences on Disability) Study: Methods of Data Collection and Characteristics of Study Sample
1Medical Research Council Lifecourse Epidemiology Unit, University of Southampton, Southampton, UK
2School of Nursing, University of São Paulo, São Paulo, Brazil
3Corporación para el Desarrollo de la Producción y el Medio Ambiente Laboral – IFA (Institute for the Development of Production and the Work Environment), Quito, Ecuador
4School of Engineering, Pontificia Universidad Javeriana, Bogotá, Colombia
5Southwest Center for Occupational and Environmental Health, The University of Texas Health Science Center at Houston School of Public Health, Houston, Texas, United States of America
6Center for Disease Control and Prevention/National Institute for Occupational Safety and Health, Atlanta, Georgia, United States of America
7Medical Research Council Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Kings College, London, UK
8Center for Research in Occupational Health (CiSAL), Pompeu Fabra University, Barcelona, Spain
9Carlos III Health Institute: Biomedical Research Networking Center of Epidemiology and Public Health, Granada, Spain
10Occupational Health Department, Parc de Salut MAR, Barcelona, Spain
11Epidemiology and Preventive Medicine Research Center, University of Insubria, Varese, Italy
12Department of Social Medicine, Medical School, University of Crete, Heraklion, Greece
13Department of Public health, University of Tartu, Tartu, Estonia
14Department of Environmental Health, Faculty of Health Sciences, American University of Beirut, Beirut, Lebanon
15Department of Occupational Health, Faculty of Health, Shahroud University of Medical Sciences, Shahroud, Iran
16Department of Community Health Sciences, Aga Khan University, Karachi, Pakistan
17Department of Medical Education and Health Sciences, Faculty of Medical Sciences, University of Sri Jayewardenepura, Gangodawila, Nugegoda, Sri Lanka
18Clinical Research Centre for Occupational Musculoskeletal Disorders, Kanto Rosai Hospital, Kawasaki, Japan
19National Institute for Occupational Health, National Health Laboratory Service, Johannesburg, South Africa
20Faculty of Health Sciences, University of Witwatersrand, Johannesburg, South Africa
21Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne, Victoria, Australia
22Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
23School of Nursing of Ribeirão Preto, University of São Paulo, São Paulo, Brazil
24Federal University of Paraná, Curitiba-PR, Brazil
25Institute for Studies on Toxic Substances (IRET), National University of Costa Rica, Heredia, Costa Rica
26Department of Occupational and Environmental Health, Università degli Studi di Milano, Milan, Italy
27Fondazione Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy
28Department of Psychiatry, Medical School, University of Crete, Heraklion, Greece
29Centre for Research in Environmental Epidemiology (CREAL), Barcelona, Spain
30IMIM (Hospital del Mar Research Institute), Barcelona, Spain
31Consorcio de Investigación Biomédica de Epidemiología y Salud Pública (CIBERESP), Barcelona, Spain
32National School of Public Health, Athens, Greece
33North Estonia Medical Centre, Tallinn, Estonia
34Põlva Hospital, Põlva, Estonia
35Klinikum Leverkusen, Leverkusen, Germany
36Department of Physiology, Faculty of Medical Sciences, University of Sri Jayewardenepura, Gangodawila, Nugegoda, Sri Lanka
37Section of Epidemiology and Biostatistics, School of Population Health, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand
38Department of Epidemiology, School of Public Health, University of Alabama at Birmingham, Birmingham, Alabama, United States of America
39Faculty of Medicine, University of Kalaniya, Kelaniya, Sri Lanka
40Department of Joint Disease Research, University of Tokyo, Tokyo, Japan
41Centre for Occupational and Environmental Health, Department of Social and Preventive Medicine, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
42Injury Prevention Research Unit, Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand
Antony Bayer, Editor
Cardiff University, United Kingdom
Received April 10, 2012; Accepted May 28, 2012.
The CUPID study has generated substantial information which will be the subject of multiple reports. A particular strength is its use of standardised questions to collect information from participants in many different countries and cultural settings. This should provide valuable insights into the determinants of common musculoskeletal illness and associated disability, and particularly the extent of differences between countries.
The occupational groups were chosen for study with the aim that the prevalence of relevant physical tasks should differ between the three broad categories (nurses, office workers and “other workers”), but that within each of these categories, it should be broadly similar across countries. For nurses and office workers this objective was fairly well achieved, although inevitably there was some heterogeneity. For example, in some countries, nurses routinely lift and move patients, whereas in others such tasks may normally be undertaken by care assistants or patients’ family members. For “other workers”, there was more variation in occupational activities, reflecting the greater diversity of groups selected for study. Nevertheless, the mix of activities tended to differ from that of nurses and office workers, with a relatively high prevalence of work with the arms elevated; and apart from sales personnel in Japan, all groups of “other workers” had a high prevalence of work involving prolonged repetitive movement of the wrists or hands.
The international analysis of data is restricted to subjects aged 20–59 years at baseline, who had held their current job for at least 12 months. These restrictions were set when the CUPID study was first planned, the latter because some outcomes of interest from the baseline survey, such as sickness absence in the past 12 months, would otherwise be difficult to interpret.
The questions used in the baseline and follow-up surveys were for the most part well-established, having been used successfully in previous studies. In particular, the items on mental health and somatising tendency were taken from validated instruments, and have previously demonstrated predictive validity for the incidence and persistence of musculoskeletal symptoms 
. Similarly, the questions on fear avoidance beliefs were based on a validated questionnaire 
, and have shown predictive validity in a longitudinal study 
. The questions on occupational physical activities have been successfully used in earlier studies 
, and the consistency of answers with expectation (e.g. the high prevalence of prolonged keyboard use in office workers) supports their validity. There is no reliable standard against which to assess the accuracy with which subjective symptoms such as pain are reported, but the questions about pain and disability had again been used successfully in earlier studies. Moreover, the style of our questions about symptoms was similar to that of the Nordic questionnaire, which has been shown to have acceptable reliability 
Ensuring the accuracy with which the questionnaire was translated into local languages was a challenge. Care was taken to check the accuracy of translation by independent back-translation to English, and this revealed a number of problems. One was the distinction between “stairs” and “flights of stairs”, and despite attempts to resolve this problem, it is not certain that the term “30 flights of stairs” was always interpreted correctly. Therefore, this question will be ignored in future analyses based on the full dataset. Another difficulty arose with questions of the form “Do you expect that your back pain will be a problem in 12 months time”. In some languages this became “Do you expect your back pain will be a problem over the next 12 months”. Attempts were made to correct this misunderstanding, but it is possible that they were not fully successful.
In addition, terms such as “pain” may be understood differently in different languages even though translated as closely as possible. For this reason, when comparing countries, differences in the relative frequency of pain at different anatomical sites may be particularly revealing – there should have been little ambiguity in the understanding of anatomical sites since they were depicted clearly in diagrams. Interpretation should also be assisted by the questions that were asked about associated difficulty with tasks of daily living, since these were probably understood more uniformly.
Another difficulty that had not been expected was in the use of dates. It emerged that some participants in Iran and Japan used different numbering for calendar years, and where this occurred, corrections had to be made.
Some local investigators opted to include extra questions in addition to the core questions prescribed by CUPID. However, these additions were relatively minor and generally followed after the core questions. Thus, it seems unlikely that they will have influenced answers to the core questions importantly.
Ideally, all questionnaires would have been completed in the same way (interview or self-administration) by all participants. However, this proved impractical. Some occupational groups (especially manual workers in developing countries) would have had great difficulty in answering a written questionnaire, while some employers were unwilling to release their staff for interviews. Moreover, in New Zealand, where nurses and office workers were recruited from across the country, interviews would have been prohibitively expensive.
To explore whether the two methods of answering the questionnaire might lead to systematic differences in answers, we therefore elected to interview a random subset of UK participants while collecting data from the remainder by self-administration. Comparison of responses using the two approaches () suggests that no major bias will have occurred as a consequence using both interviews and self-administration. However, if appropriate, method of data collection can be taken into account in statistical analyses.
Participation rates among subjects eligible for study were mostly high, but were less than 50% in five occupational groups (). We have no reason to expect that those who elected to take part were importantly unrepresentative in the prevalence of pain and its associations with risk factors. However, in future work it may be appropriate to carry out sensitivity analyses, excluding the occupational groups with the lowest response rates. The incomplete response to the baseline questionnaire will be less of a concern in longitudinal analyses based on the follow-up questionnaire.
The numbers of participants by occupational group that were suitable for analysis ranged from 92 to 1018 with a mean of 264. At the outset, our aim was to recruit at least 200 subjects in each group, and this was for the most part achieved (only 7 groups provided fewer than 150 subjects). Furthermore, the occupational groups studied varied substantially in their employment conditions (), access to healthcare (), and prevalence of psychosocial risk factors (, , and ). When exploring possible reasons for differences in the prevalence of pain and disability between occupational groups, it will be important to investigate these group-level characteristics as well as individual-level risk factors such as mental health and somatising tendency. The heterogeneity in their distribution should enhance statistical power to address their impact.
As might be expected, the demographic constitution of occupational groups also varied. In particular, many of the samples of nurses were largely or completely female, whereas some groups of “other workers” were all men. This reflects the nature of the occupations of interest. However, it should not be a major problem in interpretation of comparisons since there were an adequate number of occupational groups with a fairly even distribution of sex and age. Moreover, the occurrence of common musculoskeletal complaints appears not to vary greatly between men and women or between older and younger adults of working age 
In summary, the CUPID study is a major resource for the investigation of cultural and psychological determinants of common musculoskeletal disorders and associated disability. Although the data collected have inevitable limitations, the large differences in psychosocial risk factors (including knowledge and beliefs about MSDs) between occupational groups carrying out similar physical tasks in different countries should allow the study hypothesis to be addressed effectively. It will also allow exploration of differences in patterns of musculoskeletal complaint between the three categories of occupation examined, and the consistency of these differences across countries.