Title: Physical activity and chronic back disorders in Canadian adults
Chronic back disorders [CBD] are a major public health concern and an expensive problem, not only for direct costs due to the use of health services, but also indirect costs of productivity losses, reduced quality of life, and lost wages. CBD include a broad range of recurring and persistent pathologies and symptoms. The prevalence of CBD among Canadian adults is approximately 20%. In this context of high national and worldwide public health and economic burden, the identification of potentially modifiable risk factors, such as physical activity [PA], could be useful in decreasing the magnitude of this prevalent and expensive health condition. Although PA may be a potential factor in managing or preventing CBD, the evidence supporting this hypothesis is inconclusive. This dissertation aimed to address this gap in evidence through three distinct objectives.
The first objective of this dissertation was to describe overall time trends in CBD prevalence and in CBD prevalence by gender, age, geographical location, and PA levels among Canadian adults. Data from Canadians aged 18-65 years included in the 2007 to 2014 cycles of the Canadian Community Health Survey [CCHS] were analyzed. PA was evaluated by self-report using Canada’s PA Monitor Questionnaire [CPAMQ]. Based on transportation and leisure-time PA energy expenditure, participants were classified into three categories: active [>3.0 kcal/kg/day], moderately active [1,5-3.0 kcal/kg/day], and inactive [<1.5 kcal/kg/day]. Joinpoint regressions were used to evaluate the trends. Analysis of the CCHS showed that CBD prevalence was stable from 2007 to 2014 [≈20%]. A slight, non-statistically significant decrease in CBD prevalence was found over the 7-year time span of the study. CBD prevalence was consistently higher in people classified as inactive, and trends in CBD prevalence decreased faster among active samples than inactive samples over the eight cross-sectional survey cycles.
The second objective of this dissertation was to investigate the association between: PA and CBD within the same survey cycle [concurrent]; and PA within the previous survey cycle and subsequent CBD adjusted for previous CBD and other covariates [time-lag]. Canadians aged >=18 years from the National Population Health Survey [NPHS] [1994/1995 to 2010/2011] were included. Using the CPAMQ, three PA variables were derived: leisure-time PA levels [active, moderately active, inactive]; utilitarian walking or cycling [none, <1 hour/week, 1-5 hours/week, and >5 hours/week]. Marginal models [concurrent] and transition models [time-lag] were applied through generalized estimating equations. NPHS analyses in the time-lagged approach showed that being active during leisure time and reporting CBD in the previous survey cycle were associated with reduced CBD prevalence in the subsequent cycle [active versus inactive OR=0.89, 95% CI 0.8, 0.95]. Females engaged in >5hours/week of utilitarian cycling had a higher likelihood of CBD compared to males with no utilitarian cycling. Conversely, overweight respondents engaged in >0 to 1 hour/week or >5 hours/week of utilitarian walking, and pregnant women engaged in >0 to 1 hour/week of utilitarian walking, were less likely to report CBD compared to those who were not engaged in any utilitarian walking and were underweight or normal weight.
The third objective of this dissertation was to explore the relationship between PA and CBD in Canadian adults using either accelerometer-based or self-reported PA measures. Canadians aged 18-79 years from the 2014/2015 and 2016/2017 Canadian Health Measures Surveys [CHMS] were included in the analysis. PA was assessed using the PA Adult Questionnaire and an accelerometer-based measure. Two binary logistic regression models were fitted: one containing self-reported PA, and another containing accelerometer-based PA; each model was adjusted for several covariates. The beta coefficients obtained were compared using the Z-test for equality of coefficients and a newly developed Bias Attributable to PA Measurement [BAPAM] indicator. Analysis of the CHMS showed that the percentage of participants meeting the Canadian PA guidelines was higher when assessed with a questionnaire [57.3%] than accelerometry [17.0%]. Meeting the Canadian PA guidelines was not associated with CBD, whether PA was measured by accelerometer or by questionnaire; the Z-test [p=0.962], and the BAPAM indicator = 0.038, 95% CI -2.42, 2.49; p=0.443 showed that the coefficients of the association under study were not significantly different by PA assessment.
In conclusion, findings showed a stable, relatively high prevalence of CBD. The serial cross-sectional studies showed a higher CBD prevalence in the inactive samples during free time and transportation compared with the active samples. CBD was related with specific domains of PA rather than total PA. Prior LTPA and CBD were associated with reporting CBD two years later.
Keywords: Accelerometry; Back Pain; Joinpoint; Longitudinal Studies; Physical Activity; Prevalence