Interview versus questionnaire for assessing physical loads in the population-based MUSIC-Norrtalje study

Background MUSIC-Norrtalje study is a case-referent study, the aim of which is to find risk and health factors for low back and neck/shoulder disorder s. In this part of the study, the interview technique and the self-administ ered questionnaire used for assessment of physical loads are described and the inter-method reliability of parts of the self-administered questionnair e is estimated. The distribution of exposure levels in a general population is also described. Methods The study period was three years, from November 1993 to November 19 96 and the study subjects totaled 2,480 persons (813 female and 610 male re ferents, 380 female and 315 male low back cases, 252 female and 106 male ne ck/shoulder cases). The interview concerned "a typical working day" during the preceding 12 months and comprised assessment of energy expenditure, wor k postures, and manual materials handling for work and leisure rime, includ ing regular sport activities. The self-administered questionnaire comprised 18 questions, each covering 5 different points of time: right now, 5, 10, 15, 20 years ago. The answers to eight of the questions about current condi tions were compared to corresponding interview responses. The interview was considered as the "gold standard." Results Ninety-eight percent of the subjects completed the interview withou t any great difficulties. According to the interview: the distributions of different exposure levels were generally positively skewed, i.e., the frequ ency of highly exposed subjects was low in the study base. The correlation between interview and questionnaire responses among the referents was high for time spent "sitting at work" (r = 082), "VDU work" (r = 0.87), and work -related "motor vehicle driving" (r = 0.80). The correlation was moderate f or work-related "hands above shoulder level" (r(s) = 0.63), and "hands belo w knee level (trunk flexion)" (r(s) = 0.66). The correlation was lower for leisure time activities such as "domestic work"(r = 0.55), "time for own ac tivities" (r = 0.39), and "sitting during leisure time" (r = 0.38). Subject s seeking care for low back or neck/shoulder disorder estimated equally cor rectly or not, as had the referents. However; non-differential misclassific ation was present in all questions, which will attenuate observed estimates of the relative risk. Conclusions Even though interview data are preferable, questionnaire data m ay be useful for assessing well-defined work tasks and for "sitting at work . " Am. J. Ind. Med. 35:441-455, 1999. (C) 1999 Wiley-Liss, Inc.