Psychological and behavioral differences between low back pain populations: a comparative analysis of chiropractic, primary and secondary care patients

A cross-sectional study design.

The data materials for this study came from four samples of individuals experiencing LBP and/or NP. The first was an ongoing randomized controlled trial and the other three were previously conducted trials.

The data materials were chosen because they were thought to represent distinctly different patient populations with regards to the extent of LBP interference with the subjects’ lives. Possibly, these populations would also display differences in behavioral and psychological profiles indicating clinically meaningful differences across samples. In line with previous research [ 26, 27], it was hypothesized that the chiropractic primary care population would show a favorable psychological (less affective distress, lower pain severity, less interference and higher level of life control) and behavioral (lower frequency of behaviors by significant others as a response to displays of pain and suffering) profile compared to the secondary care populations but similar or better off than the sample from gainfully employed industry workers (primary care).

The first sample was an ongoing randomized controlled trial [ 32] investigating a population of patients with LBP from chiropractic primary care clinics in Sweden. The trial started in April 2012 and the inclusion period is expected to take 2 years with a follow-up period of 1 year. The purpose of the RCT was to investigate the effectiveness and cost-effectiveness of preventive manual care (chiropractic maintenance care) for recurrent and persistent LBP. The patients sought care when experiencing acute LBP. Most patients consult their chiropractor directly and pay for the treatment themselves; however a minority is referred and/or may have reimbursement from 3d party payers. Having agreed to participate, subjects were asked to fill in a screening questionnaire. The RCT has been described in detail in a published study protocol [ 32]. Table  1 describes the eligibility criteria. This population will be referred to as “sample A (primary care, Chiropractic)”.

The second sample came from a large intervention study entitled “Work and Health

in the Processing and Engineering Industries” (abbreviated AHA in Swedish) conducted at four companies in Sweden between 2000 and 2003, and is described in detail elsewhere [ 33, 34]. The purpose of the study was to evaluate an extensive risk assessment tool and an evidence based work place intervention to improve workers’ health. Subjects considered at high risk of developing chronic disabling NP and/or LBP and long term sick leave based on the responses in the screening assessment, were included in this study. This population will be referred to as “sample B (primary care, sick listed with high risk of chronicity)”.

The third and fourth samples came from the HUR project (Health-economic Evaluation and Rehabilitation) that commenced in 1994 with the purpose of evaluating multidisciplinary rehabilitation interventions (at specialized secondary care units) with regards to their effect on sick leave and health related quality of life as well as their cost-effectiveness (herein described as the secondary care populations). The part of the HUR study that focused on NP/LBP was designed as two separate prospective trials, one controlled observational outcome study consisting of subjects with intermittent sickness absence (herein described as “sample C (secondary care, multimodal single-center)”) [ 35– 37] and one randomized controlled trial consisting of subjects with ongoing sickness absence (herein described as “sample D (secondary care, multimodal multi-center)”) [ 38, 39]. Data were collected as part of the baseline assessment at the initial visit to the clinics. The projects have been described in detail elsewhere [ 35, 37– 39], and Table  1 describes the eligibility criteria.

The West Haven-Yale Multidimensional Pain Inventory (MPI) has been used to assess patients with a wide variety of chronic pain conditions such as neck pain (NP) and LBP [ 34, 36, 40], tempero-mandibular disorders [ 41], headaches [ 42], fibromyalgia [ 43] and cancer pain [ 44] and has been used cross culturally with translations into several languages [ 45– 47]. All four studies used the Swedish version of the MPI (MPI-S) to investigate the psychological and behavioral characteristics of the study populations. The MPI-S is described in previous publications and has been shown to have acceptable reliability and validity [ 48– 50]. Sample D (secondary care, multimodal multi-center) [ 38, 39] was used in the validation process of the Swedish version of MPI and has been included as a reference sample in this study to ensure reliable estimates.

In short, MPI-S is a 34-item, 8 scales inventory divided into two parts. Part one consists of five scales and is designed to measure important dimensions of psychological factors of pain; pain severity, interference, life control, affective distress and support. Part two consists of 3 scales and is designed to measure behavioral factors of pain; punishing responses, solicitous responses and distracting responses. See Table  2 for a detailed description of the scales in which the text is modified from the original article by Kerns et al. [ 29].

Further, three different subgroups have been empirically derived from the scales through a cluster analytical strategy [ 50– 52] to differentiate patients based on their psychological and behavioral characteristics. The subgroups are named adaptive copers (AC), interpersonally distressed (ID) and dysfunctional (DYS), have been replicated in several studies [ 31] and are described in additional Table  3. Some authors have added hybrid clusters to adjust for subjects that do not fit perfectly into any of the three suggested subgroups. It was decided not to include these hybrids to allow for better comparison with the reference population [ 50], and the hybrid subjects were therefore inserted into the closest and most representative cluster.

Both the scales and the subgroups have been used to quantify aspects of the pain experience. The MPI-subgroups are thought to constitute clinically meaningful patient groups. Information about subgroup assignment could be useful for treatment planning, descriptive or evaluative purposes. The patient groups have been investigated with regards to treatment outcome [ 53– 55] and sick leave [ 34, 36] in LBP patients and have been found to have predictive value and clinical relevance. What represents a clinically important difference in the scales on the MPI instrument is not well established and probably differs between populations, diseases and scales [ 56]. However, based on normative data, for pain and functioning, a difference of 0.6 on the interference scale represents a clinically relevant difference between populations [ 56].

The data were analyzed using SPSS (Statistical Package for the Social Sciences) v20 [ 57]. Differences in the MPI scales between the samples were analyzed with one-way analysis of variance (ANOVA) starting with an overall test for differences between the samples followed by post-hoc tests where the study population A was contrasted with each one of the study populations B, C and D according to Dunnett’s t-test (to avoid significance due to multiple testing). The comparison of the distributions of the MPI-S subgroups in each of the study populations started with a derivation phase where one study population, sample D (secondary care, multimodal multi-center) [ 38, 39], acted as the reference and the subjects of the other study populations were classified relative to this reference. The method used in this phase was a non-hierarchical cluster procedure (K-Means algorithm). The computations started with a standardization of the MPI-S scales using the mean values and standard deviations of the reference group to form Z-scores and then T-scores. For the reference group a complete cluster analysis was then done to create the centroid vectors (mean values of the MPI-S scales) for the three MPI subgroups AC, ID and DYS. Using these centroid vectors the subjects of the remaining study populations A, B and C, were classified into the MPI-S subgroups but now using only the first initial step of the cluster algorithm, that is, only one iteration in a straight-forward classification, a classify sub-procedure in the K-Means algorithm. Thus the structure of the MPI-S subgrouping from the reference group was applied to the remaining study populations and distributional differences and similarities between all four study populations could be examined. Figure  1 gives a visual description of the processes.

A replication process using a somewhat different statistical approach followed the derivation phase. To this end a discriminant analysis was used with sample D (secondary care, multimodal multi-center) [ 38, 39], and its MPI-S subgroup structure acting as reference. The subjects of the other study populations were classified in the MPI-S subgroups in accordance with the discriminant function formed from the reference population. In the end the classifications from the cluster approach and the discriminant approach were compared by cross-tabulations and weighted kappa calculated as a measure of agreement.

Finally the distributions of the MPI-S subgroups for the four study populations were compared with a chi-square test for distributional differences.

The current study was approved by the local Ethical Committee at the Karolinska Institutet: 2015/1483-32.

The ongoing RCT from which sample A (primary care, Chiropractic) was collected was approved by the local Ethical Committee at the Karolinska Institutet: 2007/1458-31/4, and was registered at Clinical trials.gov; NCT01539863. Written informed consent was obtained from all subjects. The study from which sample B (primary care, sick listed with high risk of chronicity) was collected was approved by the local Ethical Committee at the Karolinska Institutet: (AHA; Dnr 00–012). Written informed consent of each of the employees was obtained.

The studies from which sample C (secondary care, multimodal single-center) and D (secondary care, multimodal multi-center) were collected were approved by the local Ethical Committee at the Karolinska Institutet (Dnr 94:340) and written informed consent was obtained from all study subjects.