Page 51 - THE PERCEPT STUDY Illness Perceptions in Physiotherapy Edwin de Raaij

Study characteristics
The study characteristics of the 26 included papers are presented in Table 3. The number of participating patients in studies varied from 1117 to 211325. Twelve different musculoskeletal conditions were identified:
1) rheumatoid arthritis18,45,48,50,51,56, 2) LBP2,7,12,16,37,49, 3) chronic pain14,32,42,44, 4) chronic headache4, 5) fibromyalgia55,57, 6) systemic lupus erythematosus17, 7) hand problems26, 8) chronic repetitive strain injuries54, 9) acute injury8, 10) chronic orofacial pain13 , 11) gout9 and 12) osteoarthritis of the knee27. For the research question concerning associations of illness perceptions with pain intensity, authors found 11 cross-sectional studies (Appendix A), and for illness perceptions with physical function, 11 cross-sectional studies (Appendix B). The study of Groarke et al18 has a longitudinal design, but cross-sectional associations were also reported and were used to answer the questions of illness perceptions’ association with pain intensity or physical function. For the prognosis of pain intensity by illness perception, the researchers found 4 longitudinal studies (Appendix C), and for the prognosis of physical function by illness perception, 10 longitudinal studies (Appendix D).
Risk of bias
All studies but 1 had a high risk of bias (Table 4). The study by Foster et al12 was scored as low risk on all 6 domains. There was considerable variance between studies in percentages of items scored “yes,” ranging from 29%47 to 87%2,12,13,28,45, with an average of 66%. The QUIPS domain “study confounding” was most frequently scored as high risk of bias, and “study participation” was scored most frequently as low risk of bias. After initial assessment, there was 82% agreement on the risk-of-bias assessment of the 6 QUIPS domains between the 2 independent reviewers (E.D.R. and H.W.). Differences were resolved between the 2 assessors without the need to consult a third assessor. Funnel plots were processed and showed risk of publication bias for all illness perception dimensions.
Results of individual studies
The data extraction for all 26 studies is presented in tables comprising Appendices A- D. The authors found a total of 321 different variables for illness perceptions’ association with or prognosis for pain intensity and physical function. These variables ranged from univariate, multivariate, beta, and odds ratio to relative risk. For the prognostic value of illness perceptions for pain, the researchers found only 4 studies with short- or medium-term results16,18,27,51. They found no studies with long-term results of more than 1 year.
A systematic literature review
49
3