We found that physical disability, physical activity levels, depression, kinesiophobia and catastrophising are the main modifiable biopsychosocial predictors of SM and its change in patients with CLBP. Further, we also found that age, pain intensity and pain duration do not predict SM and its change. Pain intensity and duration not predicting SM agrees with previous research [26]. However, our finding that age is not a predictor of SM contrasts with a cross-sectional study [26], where age correlates with SM negatively.
In our study, perceived physical disability negatively predicted three out of the eight SM constructs. However, physical disability measured using the Oswestry Disability Index was not a significant predictor of SM measured using the PAM in 230 patients with CLBP [26]. This difference in the findings could be due to the populations and different scales to measure SM and disability. For example, Kawi measured SM using PAM, which measures only patients’ activation and engagement from primary care and specialist pain centre in the USA.
We found that baseline depression had a significant negative predictive association for five out of eight SM constructs, suggesting that lower mood (i.e., symptoms of depression) was associated with poorer self-management outcomes. Although depression has not been examined as an explanatory variable in previous research investigating predictors of SM in patients with CLBP [26], depression is common in patients with other long-term conditions and has been found to impact negatively on SM. For example, depression is common in diabetes mellitus [39] and is an established negative predictor of diabetes SM in children [40] and adults [41, 42]. Depression has also been identified as a predictor of SM in older adults (n = 3292) in the UK, albeit using a different outcome measure [43]. Depression significantly predicted SM, measured using the Skill and Techniques Acquisition (STA) subscale of the German version of the heiQ, in patients with chronic conditions (n = 580), including rheumatism, asthma, orthopaedic disorders and inflammatory bowel disease [44].
Therefore, our results suggest that depression is a key predictor of certain constructs of SM in patients with CLBP, which agrees with broader research in patients with diverse long-term conditions.
Kinesiophobia and catastrophising have not previously been investigated as predictors of SM in patients with CLBP. However, distress and/or anxiety were investigated as a predictor of SM in patients with diabetes [45, 46]. An earlier study by Albright et al. [46] found stress had a significant negative predictive association with exercise and diet SM in 392 type II diabetes patients. Similarly, Schinkus et al. [45] found distress (measured using Diabetes Distress Scale) and anxiety (measured using the State-Trait Anxiety Inventory) were significant predictors of overall diabetes SM (measured using the Diabetes Self-Management Questionnaire) in 146 patients with type-I and type-II and gestational diabetes. These studies highlight the importance of measuring distress or anxiety or related variables as an explanatory variable in SM predictor studies.
In the present study, healthcare use, measured using the self-reported number of sessions attended at the general practitioner, physiotherapist, specialist, and other practitioners for CLBP, significantly predicted the SMI construct of SM. Further, education, income, living arrangements, being employed, being married, high annual income (> £30,000) and white ethnicity had significant associations in univariate GLM analysis. These results agree with the previous cross-sectional study [26], where age, education and income were significant predictors of SM in patients with CLBP. However, no significant predictive association was found in the multivariate GLM analysis for demographic and socioeconomic factors, suggesting that change in SM does not depend on age, education, and income.
Changes in depression, kinesiophobia, catastrophising, and physical activity level similarly predicted SM changes (R2 10% and 32%). Change in catastrophising predicted change in five out of eight SM constructs (HDA, PAEL, CAA, SIS and HSN). Catastrophising is a negative predictor for patients with CLBP and might contribute to the delayed recovery [47]. Patients with CLBP who had higher catastrophising were associated with a significantly higher disability using Roland Morris Disability Questionnaire in a UK population at a 12-month follow-up [48]. Further, patients with CLBP reported fluctuating negative pain-related thoughts affecting their coping and pain-related meta-cognition in a recent qualitative study [49], which could potentially influence HDA, PAEL, CAA, SIS and HSN. Change in depression predicted change in PAEL and CAA. Similarly, change in depression predicted SM in patients with diabetes [42] and long-term conditions [44].
Theoretical underpinning
According to the Social Cognitive Theory, one of the critical theories underpinning SM, cognitive factors and psychological states modify self-judgement and the SM [50, 51]. Therefore, depression, excessive negative pain-related emotions or catastrophising and fear related to pain or re-injury or kinesiophobia may influence one's SM ability. Similarly, physiological states, including depression, kinesiophobia and catastrophising, influence self-efficacy and SM [51]. Therefore, along with promoting healthy living and physical activity [52], psychological and behavioural factors should be targeted to enhance SM in patients with CLBP. From a behaviourist point of view, capability, opportunity, and motivation interact to generate behaviour, in which capability includes one's physical and psychological abilities to engage in (SM) activity. So, SM programmes can utilise the Behaviour Change Wheel to create opportunities using interventions and policies to motivate individuals to change their capability [53].
Strength and limitations
To our knowledge, this is the first prospective multi-centre longitudinal cohort study investigating predictors of SM in patients with CLBP. A strength of the study is the use of a multi-construct SM scale which provides a comprehensive assessment of SM constructs, and multi-component measures have not been used in previous studies of CLBP or other chronic conditions. The study has some limitations. The attrition rate was relatively high, with 117 patients lost to follow-up (43.33% attrition, compared with an anticipated 30%). However, there was no difference in baseline disease-related and SM outcomes between patients that completed follow-up and those who did not. The study had a poor representation of the Asian and male gender. However, it has been found that women are more likely to participate in survey research [54]. The lack of ethnic diversity may be due to excluding patients without good English language abilities due to a lack of funding for interpreters. However, the figures for ethnicity are comparable to the UK 2011 Census data [55] proportion of White and working citizens. Future studies may focus on ways to improve SM in low-income, non-White populations with CLBP, particularly individuals with poor health literacy.