Exploring the complexity of commonly held attitudes and beliefs of low back pain—a network analysis

Objectives The current study used a network analysis approach to explore the complexity of attitudes and beliefs held in people with and without low back pain (LBP). The study aimed to (1) quantify the adjusted associations between individual items of the Back Pain Attitudes Questionnaire (Back-PAQ), and (2) identify the items with the strongest connectivity within the network. Methods This is a secondary data analysis of a previously published survey using the Back-PAQ (n = 602). A nonparametric Spearman’s rank correlation matrix was used as input to the network analysis. We estimated an unregularised graphical Gaussian model (GGM). Edges were added or removed in a stepwise manner until the extended Bayesian information criterion (EBIC) did not improve. We assessed three measures of centrality measures of betweenness, closeness, and strength. Results The two pairwise associations with the greatest magnitude of correlation were between Q30–Q31 [0.54 (95% CI 0.44 to 0.60)] and Q15–Q16 [0.52 (95% CI 0.43 to 0.61)]. These two relationships related to the association between items exploring the influence of attentional focus and expectations (Q30–Q31), and feelings and stress (Q15–Q16). The three items with the greatest average centrality values, were Q22, Q25, and Q10. These items reflect beliefs about damaging the back, exercise, and activity avoidance, respectively. Conclusion Beliefs about back damage, exercise, and activity avoidance are factors most connected to all other beliefs within the network. These three factors may represent candidate targets that clinicians can focus their counseling efforts on to manage unhelpful attitudes and beliefs in people experiencing LBP.


Introduction
Low back pain (LBP) is a highly prevalent and costly musculoskeletal pain disorder, which occurs in all countries and affects individuals across the lifespan (1).In 2019, LBP accounted for over half a billion prevalent cases and over 60 million years lived with disability (YLDs) (1).In 2016, LBP was ranked first in healthcare expenditure of US$134.5 billion in the United States (2).Two-thirds of those with an acute painful episode of LBP recover within the first three months (3), although relapses and remissions are common (4).Approximately 20% of LBP sufferers go on to experience severe, high-impact chronic pain (5), which contributes most to healthcare costs.The attitudes and beliefs of both patients and clinicians are thought to be important contributors to the development of LBP, its recovery, and how the condition is managed.Harboring negative attitudes and beliefs regarding LBP may elevate catastrophic thoughts and avoidant behaviors (6), which cascade into greater disuse and depression (7), leading to delayed recovery and functional return (8,9).The beliefs of healthcare practitioners have been reported to explain as much as 20% of the variance in their recommendations to patients suffering from LBP (10).The importance of the relationship between beliefs and clinical management has also been reported across many healthcare professions and in different countries (11).To measure the attitudes and beliefs regarding LBP, the 34-item Back Pain Attitudes Questionnaire (Back-PAQ) was developed (12).Further work on the Back-PAQ resulted in the development of an abbreviated 20-item (13) and also a 10-item version (12).
Regardless of the versions of the Back-PAQ, a summative total score is determined by aggregating the values across all items (12).For example, the original Back-PAQ has a total score ranging from 34 to 170, with higher scores indicating more negative beliefs regarding LBP (12).This aggregate score has been used in clinical trials to determine the effects of different interventions on changes in the patient's attitudes and beliefs (14).Analyzing only the aggregate score of the Back-PAQ does not fully maximize the use of the information.This is because two individuals can have similar aggregate Back-PAQ scores, but have very different scores on individual items.Determining the most important Back-PAQ items could improve its utility for clinical decision-making.
There are potentially many approaches in seeking to understand the most important facets underlying an individual's beliefs about LBP.In a previous study of the Back-PAQ in the general population, items relating to posture (Q8), muscle strength (Q7), and lifting technique (Q5) were the most negatively scored (6); it may be that items with the worse score are considered the most important items.Patients and clinicians are commonly thought to hold negative beliefs about the safety of certain lifting postures (15, 16), and the appropriateness of physical activity during an episode of LBP (17).The importance of the beliefs about posture and activity resumption is evidenced by the development of therapeutic interventions seeking to target these specific beliefs (18,19).Alternatively, facets of an individual's belief system with strong prognostic value may be deemed as important, including expectations about recovery (20), and selfefficacy (21).
Another approach to determining the importance of items is via network analysis (22).Network analysis focuses on quantifying the multivariate relationships between individual items (23,24).The importance of any item in network analysis, also termed centrality, is typically defined by the magnitude of association, and the closeness of associations to all other items (22).An item with a very high score may not be central, if it is connected to very few items.In a hypothetical scenario of negative beliefs about posture, a low centrality would mean that this specific belief does not affect the beliefs on other items.From a treatment perspective, targeting a low central item would not be the most efficient approach.
The current study explored the multivariate relationship of the items within the Back-PAQ.The main aims of the study were to (1) describe the network and identify the item pairs with the strongest adjusted associations, and (2) identify a reduced set of items with the strongest connectivity within the network.Given that network analysis is a data-driven approach, and that this is the first study to apply such techniques on the Back-PAQ, there were no priori hypotheses made about what item pairs would be the most correlated, or which items would have the greatest centrality measures.

Study design
Secondary data analysis using the methodology of network analysis.

Participants
This study used data from a previously published survey of the New Zealand population that used the Back-PAQ (6).One thousand people who were 18 years and older were randomly selected from the New Zealand Electoral Roll and invited by mail to complete the survey.The survey was completed by 602 participants (female = 331, male = 271).Participant characteristics are described in detail in the original publication ( 6), but briefly, 76 participants self-reported never having experienced a back pain history, 361 reported a past experience of a back pain history, 164 reported a current experience a back pain history, and one participant did not self-report.

Approach to network analysis 2.4.1 Software and packages
The dataset was analyzed with R statistical software (version 4.2.2).Several packages were used to perform the analyzes, including qgraph (26) for network estimation and plotting, and bootnet (27) for stability analysis.Since the Back-PAQ items are ordinal, a nonparametric Spearman's rank correlation matrix was used as input to the network analysis.We estimated an unregularised Graphical Gaussian model (GGM), using the ggmModselect algorithm with the following parameters (28): tuning = 0.25, stepwise = TRUE, consider PerStep = "subet, " and missing = "pairwise." From a graphical lasso network model, edges were iteratively added and removed until the extended Bayesian information criterion (EBIC) did not improve (28).This is similar to performing stepwise selection in regression models using Akaike information criterion.Presently, we assessed three measures of centrality: betweenness (how often one node lies on the shortest path between other nodes), closeness (shortest edges to other nodes), and Strength (magnitude of all the node's immediate edges) (29).Clinically, a node high in Strength can directly influence many adjacent nodes, without the influence of other nodes (29).A node high in Closeness can be interpreted as the speed of influence a change in one node has on all other nodes in the network (29).Lastly, if a node high on Betweenness were to be removed, the relationship between all other nodes become more indirect (29).
We assessed the variability of the edge weights using bootstrapping (B = 1,000) (27), to estimate the 95% confidence interval of the estimated edge weights (i.e., the partial correlations).To gain an estimate of the variability of the found centrality indices (CS-coefficient)-meaning if the order of centrality indices remains the same after re-estimating the network with fewer participants, we applied the participant-dropping subset bootstrap (B = 1,000) (27).This procedure drops a percentage of participants, re-estimates the network, and re-calculates the three centrality indices.The percentage of participants dropped ranged from 5% to 75%, across 10 sampling levels.The CS-coefficient reflects the maximum proportion of participants that can be dropped, such that with 95% probability the correlation (of the centrality value of the bootstrapped sample vs. that of the original) would reach a certain value (0.7 in the current study, CS cor = 0.7 ).It is suggested that the CS cor = 0.7 should be >0.25 and is better if it is >0.5 (27).

Discussion
Attitudes and beliefs are thought to be important contributors to the development, recovery, and management of LBP.The present study aimed to understand the complex relationship between the individual items of the Back-PAQ to better understand how different beliefs interact with each other.The top two most correlated edges were between focus and expectations (Q30-Q31), and feelings and stress (Q15-Q16).In addition, the three items with the greatest average centrality values across betweenness, closeness, and strength, were Q22, Q25 and Q10.These items reflect beliefs about damaging the back, exercise and activity avoidance, respectively.
A recent systematic review have reported that recovery expectations is a prognostic factor of return to work and recovery outcomes (20), and that expectation of symptom change modulates changes in pain and impairment (30,31).From our network analysis, a more positive belief about recovery expectations was associated with more positive beliefs about the benefits of focusing on things other  Network analysis of the association between the 34 items of the Back-PAQ.Each edge in the network represents either positive regularized adjusted associations (blue edges) or negative regularized adjusted associations (red edges).The thickness and color saturation of an edge denotes its weight (the strength of the association between two nodes).than the back (Q30), staying active (Q27), and acknowledging the role of thoughts and feelings in LBP (Q15).These associations may represent candidate mechanisms by which recovery expectations influence LBP outcomes.From the literature, it is thought that recovery expectations might affect LBP outcomes by modifying coping, healthcare-seeking, and withdrawal behaviors (20).It may be that with a more positive belief about focusing on other things and staying active, patients have greater self-efficacy in pursuing activities, despite the presence of pain, which ultimately benefits the recovery of LBP.
Even though beliefs about good posture (Q8) were not most correlated with having strong muscles (Q7), our findings still support their direct association.The present finding supports prior research which reported that patients frequently viewed correct lifting techniques, posture, and having strong muscles as collective strategies for protecting the back (32).Interestingly, beliefs about bending (Q3), sitting (Q4), and lifting (Q5), were not directly associated with each other (Figure 1).Some of these beliefs have been thought to have their roots in communication with clinicians (32) and mass media campaigns (33).If beliefs about bending, sitting, and lifting had a common cause, it would be likely that they are directly associated with each other.Findings from the present study suggest that each of these three beliefs may not be as closely associated as previously thought (34), and may have different antecedent causes.Clinically, this suggests that if educational efforts were to be directed toward altering the beliefs of these activities, they will have to be done so individually, rather than with the expectation that changing the beliefs on one task will influence another.
Items on the Back-PAQ with the worse score may not always be the most connected items within the network.For example, items relating to posture (Q8), muscle strength (Q7), and lifting technique (Q5) were the most negatively scored ( 6), but represented some of the least central items (Figure 4).In other words, these aforementioned items are relatively isolated from all other items.The most central items relate to beliefs about causing back damage (Q22), the benefits of avoiding exercise (Q25), and activity avoidance (Q10).These three beliefs have close relations with prior reported perceived myths about LBP, particularly on the role of tissue damage in LBP, and the importance of stopping exercise and activity when LBP occurs (34).Not surprisingly, these unhelpful beliefs about exercise are also held by clinicians [e.g., Q9 in (35)], reinforcing the importance of the enduring influence of clinical opinions on the beliefs of LBP on lay people (32).Prior qualitative research has reported that negative beliefs about low back tissue damage results in high pain-related fear (36), while quantitative longitudinal research have reported that fear is a prognostic indicator of persistent LBP symptoms (9).Our findings also support prior research which identified that LBP individuals with high pain-related fear have two predominant beliefs-the potentially damaging effects of physical activity and that performing an activity with pain will increase suffering (37).
The network visualization is clinically very intuitive, enabling rapid and unique clinical insights which may be used to efficiently guide patient counseling.For example, our findings showed that the belief about the ease of injury (Q6) is directly associated with the belief about the safety of lifting (Q5), and not sitting (Q4).This means that for clinicians desiring to alter a patient's beliefs about sitting safety, educational efforts to modify the patient's beliefs about the vulnerability of the spine to injury may not be the most efficient Bootstrapped 95% confidence interval of the estimated edge weights of the network.Only edges where 100% of the bootstrapped estimated correlation was non-zero retained for plotting.(34).The present finding supplements prior clinical recommendation reports (34), providing evidence for the most efficient approach to navigating these beliefs with patients.This study has several limitations.First, no attempt was made to distinguish the network dynamics of the Back-PAQ among people with and without LBP.Future investigations on understanding the differences in belief systems among different LBP subgroups may be useful for personalizing education efforts in managing and preventing LBP.Second, the longitudinal relationship between individual items of the Back-PAQ and clinical outcomes was not investigated.Including both the items of the Back-PAQ and measures related to clinical outcomes (e.g., pain intensity and impairment at follow-up) in a prospective study, may help to identify specific beliefs driving clinical outcomes.Third, the original study recruited participants with and without LBP randomly selected from an Electoral Roll.Information concerning the duration of current LBP and whether LBP had a specific cause (e.g., spondyloarthropathy), was not collected.A previous study reported that individuals with axial spondyloarthropathy reported lesser LBP intensity and better health related quality of life, than those with chronic non-specific LBP (38).Whether similar attitudes and beliefs are held in people with specific and non-specific LBP remains to be investigated.

Conclusion
Network analysis of the Back-PAQ revealed unique insights into the beliefs about LBP.Beliefs about back damage, exercise, and activity avoidance are factors most connected to all other beliefs within the network.This suggests that these three factors represent candidate targets that clinicians can focus their patient counseling efforts on.

FIGURE 1
FIGURE 1Mean and error bar as one standard deviation of the cohort's individual item score of the Back-PAQ.

FIGURE 4
FIGURE 4 Standardized (0-1) centrality scores of the 34 items of Back-PAQ.Items shaded in green indicate the top three most central items and those in red indicate the least central items.

TABLE 1
Individual items and their questions of the Back-PAQ.
33There is a high chance that an episode of back pain will not resolve 34 Once you have a back problem, there is not a lot you can do about it 10.3389/fmed.2024.1327791Frontiers in Medicine 04 frontiersin.org