File
Carvalho F.A.1, Morelhao P.K.1, Franco M.R.1, Maher C.G.2, Smeets R.J.E.M.3, Oliveira C.B.1, Pinto R.Z.1
1Sao Paulo State University (UNESP), Physiotherapy Department, Presidente Prudente, Brazil, 2The George Institute for Global health, University of Sydney, Musculoskeletal, Sydney, Australia, 3Maastricht University and Libra Rehabilitation and Audiology, Department of Rehabilitation Medicine, Research School Caphri, Eindhoven/Weert, Netherlands
Background: Measuring physical activity (PA) levels in chronic low back pain (LBP) is thought to have important clinical implications. Although it is still debatable whether patients with chronic LBP decrease their PA level, physical activity-based interventions are effective in managing this condition and being physically active has shown to be a predictor for good prognosis. Most evidence regarding PA in this area has been generated from self-reported assessment methods. However, objective PA assessment, such as accelerometers, have become one of the most commonly used methods to assess PA in free-living activities, due to their small size, low participant burden and relatively low cost.
Purpose: The objective of this study was to determine the test-retest reliability of two multidimensional self-reported PA measures, the International Physical Activity Questionnaire (IPAQ - long-version) and Baecke Physical Activity Questionnaire (BPAQ), and to assess the criterion-related validity of these measures against objective PA measured derived from accelerometers in patients with chronic LBP.
Methods: Patients aged between 18 and 60 years old with non-specific LBP reporting pain for more than 3 months were considered eligible. Each participant attended the clinic twice (one week interval). At the first session, information about demographic and anthropometric data; duration and severity of LBP (i.e. 0-10 scale); disability (i.e. Roland Morris Disability Questionnaire); and self-report PA were collected. Patients were asked to wear an accelerometer during waking hours 7 days (at least 5 days and 10 hours/day). At the second session participants returned the accelerometer and completed the self-reported PA measures for the second time. Intraclass Correlation Coefficients (ICC) and Bland and Altman method were used to determine reliability and spearman rho correlation were used for criterion-related validity.
Results: A sample of 73 patients predominantly men (68%) with mean age of 37.2 (12.2), median symptoms duration of 12 [30.0], and mean pain and disability of 6.9 (1.5) and 36.9 (20.6) respectively, was included. The results revealed that BPAQ and its subscales has excellent reliability (ICC > 0.75), whereas IPAQ and most IAPQ domains shows only poor reliability (ICC > 0.40). The mean difference between the two days (limits of agreements) for the IPAQ Total PA was 408.8 (-3027.6 to 3845.3) min/week, indicating significant discrepancies between the two days. In contrast, the results for BPAQ total PA was -0.05 (-1.78 to 1.81) with no observable systematic discrepancies.
For the validity analysis, the IPAQ-total PA and BPAQ-total PA showed fair significant correlation with counts/min and steps/day (rho coefficients ranging from 0.25 to 0.33). The time spent in moderate-to-vigorous PA measured by the IPAQ and the accelerometer showed no significant correlation.
Conclusion(s): The BPAQ showed better reliability than the IPAQ long-version in patients with chronic LBP. Both self-reported measures, however, did not demonstrate acceptable validity against accelerometer data. Fair correlation was found only between objective PA measures and disability.
Implications: These findings suggest that self-report and objective PA measures should not be used interchangeably in this population, once the results may be overestimated.
Funding acknowledgements: Supported by the São Paulo Research Foundation (grant nos. 2015/17093-7, 2015/07704-9, 2015/02744-2, and 2014/14077-8).
Topic: Musculoskeletal
Ethics approval: This study was approved by the ethics research committee of the Sao Paulo State University (CAAE36332514.0.0000.5402).
All authors, affiliations and abstracts have been published as submitted.
