The objective of this study was to examine the trends in the overall prevalence of CBD and specific CBD prevalence by age, gender, residence (rural/urban and province/territory), and physical activity levels among Canadian adults aged 18–65 years. We found that the overall and specific trends in CBD prevalence have been stable from 2007 to 2014 in Canadians aged 18 to 65 years. These findings contrast with the increase in CBD prevalence reported in other high income countries [27,28,29,30,31,32,33]. CBD prevalence was consistently higher in women, older age groups, rural populations, and people classified as inactive within each cycle 2007–2014. Trends (Fig. 2e) and maps (Fig. 1) consistently showed that the prevalence of CBD remained relatively higher in Nova Scotia and relatively lower in Quebec in comparison with the remainder of the Canadian provinces and territories.
Overall trend prevalence
Evidence about the population prevalence of CBD over time in North America is scarce. We found 11 articles studying secular trends in back disorders in Finland [33,34,35, 37], Britain [27, 28], USA , Germany , Sweden , and Austria [30, 31]. All studies included information collected before 2007, and only one study focused on chronic problems . Seven of the studies conducted in Finland [34, 35, 37], Austria [30, 31], Britain , and Germany  reported prevalence estimates of back disorders greater than 20% for at least one study period. While the remaining four studies from Finland , Britain , Sweden , and USA  showed a prevalence lower or equal to 20%. The lower CBD prevalence found in the present study could be partially explained by the definition of the outcome, which includes back problems that have lasted at least six months, while the studies with greater prevalence did not explicitly define the duration of back pain. Thus, the back pain definition in the previous studies may include those who have acute back problems lasting less than six months duration, leading to higher prevalence rates.
Stable chronic pain prevalence rates were also found by Reitsma et al.  using two Canadian surveys, the National Population Health Survey (cycles 1994/5, 1996/7, 1998/9), and the CCHS (from 2000/1 to 2007/8). The former survey included people aged 25 years and older, and the latter included people aged 20 years and older. Chronic pain was defined as a negative response to the question: “Are you usually free of pain or discomfort?”, which would likely include people with CBD . Reitsma et al. results indicated that the overall temporal trends were not significant. The initial prevalence was 18.9%, the final prevalence was 18.5%, and the minimum prevalence was 15.1% in 1996/97 . They also affirmed that studies using a specific timeframe to define chronic pain such as three or six months were more likely to find greater chronic pain prevalence compared to using an unspecified timeframe like “usual” pain .
Our results also concur with those reported by Leino et al. . In their study of back problems from 1979 to 1992 in Germany, Leino et al. reported back-pain prevalence close to 30%, but did not find changes between 1985 and 1992. Leino et al. stated that the stable prevalence could be explained by changes in the societal judgment of good health and functional capacity, and attitudes towards pain. Thus, although people reported more acute musculoskeletal problems, the improvement of the health care system has prevented the development of more severe chronic conditions after the onset of back pain. While this trade-off phenomenon might have been at work in 1980s and 1990s Germany, a similar tradeoff may be in effect during the later period investigated in this current work, where a stable CBD prevalence could be the result of following the evolving guidelines for back pain management . The guidelines recommended staying active instead of bed rest in order to reduce fear-avoidance beliefs and other negative psychological consequences. Following these guidelines could reduce the number of people progressing from acute to chronic back pain, thus those suffering during at least six months remain stable under the assumption of a stable incidence .
Another possible explanation of fairly stable CBD prevalence is that in our study CBD was measured using consistent, comparable questions with similar CCHS methods. The five [27, 30,31,32, 35] out of the eight population-based studies showing an increased trend in CBD [27,28,29,30,31,32,33, 35] reported methodological changes over time involving variable: response options ; mode of data collection ; sampling process ; and questions asked to ascertain CBD [30, 31]. Modifications in the questions and survey methods may mask real trends in CBD due to the inability to determine whether observed trends are due to real changes in population morbidity or due to changes in research methodology.
Specific trend prevalence
Gender and age
Our findings relating to a higher CBD prevalence in women and older people concurred with other studies. Among studies evaluating trends in CDB, Leijon et al. , Harkness et al. , and Freburger et al.  found that CBD prevalence rates were significantly greater in women than men across time. In general, trends in CBD prevalence in both men and women followed the overall trend except in two studies [32, 34]. Leijon et al.  found that the increasing trend was statistically significant among women but not among men, and Heistaro et al.  revealed that after controlling for age, the declining trend was statistically significant among men but not among women. On the other hand, Heistaro et al. , Palmer et al. , Harkness et al.  and Großschadl et al. [30, 31] found greater CBD prevalence in older (mainly over 55 years) people. In our study, we found that the slope of the younger group was greater in the 18–34 years group compared to the 50–65 years group, meaning that the CBD prevalence declined more slowly in the 50–65 years group. A similar pattern was found by Heistaro et al. , whose time graphs showed that the significant decreasing trend was more noticeable in younger people (30–39 years and 40–49 years) than older people (50–59 years).
In this study, rural dwellers had a higher CBD prevalence over time in comparison to urban dwellers. In addition, trends and maps consistently showed that the prevalence of CBD remained relatively high in the province of Nova Scotia and relatively low in Quebec. The high CBD prevalence in the province of Nova Scotia could be due partly to the greater percentage (34%) of rural population in that province, which is higher than the proportion of rural population in Quebec (19%) . The higher CBD prevalence in the Yukon, Northwest and Nunavut Territories and rural settings may be related to challenges in accessing health care services. People living in rural or remote places have more difficulties to get health care services compared with those living in urban settings . It is probable that in remote and rural places, people with an acute back problem are less likely to receive adequate or any health care advice or treatment, which may lead to a chronic condition. In addition, people living in rural and remote places are involved in different industries and working conditions (e.g. agriculture, mining), and thus they could be exposed to different risk factors for CBD . Furthermore, there is a high proportion of Indigenous people living in the Canadian Northern territories  and Indigenous people have a higher reported prevalence of CBD in comparison to non-Indigenous populations .
We found that people classified as inactive had a higher CBD prevalence over the eight cycles. In addition, we found a slope difference between inactive and active people, indicating that the CBD prevalence decreased faster in active people than their physically inactive counterparts. Differences in CBD prevalence trends by the level of leisure time physical activity were found by Heistaro et al. . Their time graphs showed that the significant decreasing trend was more noticeable in men engaged in high physical activity than men engaged in moderate or low physical activity. In contrast, women engaged in high physical activity exhibited a rising trend in back pain prevalence. These findings, however, are not completely comparable with our results because the questionnaire used was different, and we included both transportation and leisure physical activity in the analysis.
There are some proposed mechanisms that explain how physical activity could decrease the likelihood of having CBD. First, being physically active may strengthen back muscles and increase trunk flexibility that can provide the stability and range of motion needed in functional activities, and thus the risk of back injury will be reduced . Second, physical activity may increase circulation of the blood to the back muscles, joints, and intervertebral fibrocartilage, reducing damage and stiffness that can result in back pain . Third, people who regularly exercise can complete daily tasks with less effort, thus decreasing fatigue, and maintain muscular strength and endurance; which may ultimately reduce the likelihood of overload from daily tasks later in life [61, 62]. Moreover, regular physical activity reduce serotonin transporter expression, increases serotonin levels, and increases endogenous opioids in central inhibitory pathways; decreasing the perception of back pain .
Strengths and limitations
One strength of our study was the use of eight fully comparable CCHS cycles. The same CBD question employed in this study allowed an exploration of time trends in a large and representative sample of Canadians aged 18 to 65 years. In addition, we used a joinpoint regression analysis that has never been utilized in CBD trends research; this flexible statistical method is useful for the assessment of changes of the outcome (CBD prevalence) as a function of the independent variable (year, from 2007 to 2014) in general and by groups. Joinpoint regression analysis was initially used to determine the change points and trend pattern of cancer rates . However, to our knowledge, this is the first time it is being used in CBD trends research.
There are some limitations in this study to consider when interpreting our findings. The first limitation is the CBD classification. As there was not a body diagram alongside the CBD question, the CBD category might contain those who have only upper back pain, and those with concomitant neck, upper and lower back pain. Similarly, although the CCHS’s CBD question asked for back problems excluding fibromyalgia and arthritis, people with concomitant fibromyalgia and arthritis may also have CBD . The prevalence of arthritis and fibromyalgia was around 16%  and 2%  in Canada (2014), respectively, this implies an underestimation of the CBD prevalence up to 16% on an 18% prevalence of CBD in the same year. Also, interpreting the CCHS data relies on the assumption that respondents have correctly interpreted the questions. For example, some vocabulary such as fibromyalgia/arthritis might be unfamiliar, though we anticipate that those who received a diagnosis will have heard the terms from their care providers. Furthermore, self-report questionnaires introduce possible biases, including recall error . In addition, the method of interview could influence the participants’ response. For example, St-Pierre found that in the 2003 CCHS, people interviewed in-person reported being more inactive (42.3%) than people interviewed by telephone (34.4%) . A further limitation is that the generalizability of the results to the complete Canadian population or the European population may be restricted due to contextual factors and the exclusion of people living on reserves and other aboriginal settlements, residents of institutional facilities, members of the Canadian Forces, and residents of certain remote regions.
Future work, implications and applications
The great variety of questions used to determine CBD prevalence makes comparisons among studies difficult. Thus, it is important for researchers to report the features of the CBD questions used as well as disclose any modifications in the methodology to allow for the appropriate interpretation of their results. Cross-sectional surveys do not allow identification of causal relationships. Consequently, next steps would incorporate analysis of longitudinal data that include repeated measurements of CBD prevalence. This study contributes to the understanding of the epidemiology of CBD using relatively recent data. In addition, we believe that ongoing monitoring of CBD over time could be used to evaluate the effectiveness of population-based prevention programs and management approaches among adults suffering from back disorders.