Clustering of patients with chronic low back pain in terms of physical and psychological factors: A cross-sectional study based on the STarT Back Screening Tool

1. Introduction

Low back pain (LBP) is one of the most prevalent musculoskeletal disorders and is strongly associated with disability, low quality of life, emotional stress and absenteeism [1, 2, 3]. Most LBP episodes resolve quickly, but about 50–75% of the cases become chronic [4, 5]. There is an increasing acceptance that psychosocial factors play a crucial role in a transition from acute to a chronic LBP [6]. Early screening of psychological factors is proposed for improving the quality of care by identifying patients who are at risk of developing chronic LBP [7, 8, 9, 10, 11]. There are two types of psychological screening tools; uni-dimensional and multidimensional questionnaires [12, 13]. While uni-dimensional questionnaires include several items that are all related to one psychological factor, multidimensional questionnaires include several items that are related to one specific psychological factor [14]. The STarT Back Screening Tool (SBT) is a multi-dimensional questionnaire which includes some physical factors in addition to psychological factors [15]. Based on SBT, patients are categorized to be at a low, medium, or high risk of developing chronic LBP [15]. Psychometric properties of SBT were evaluated and confirmed in several studies [1, 2, 12, 14, 16, 17, 18, 19, 20, 21, 22]. Previous studies used uni-dimensional psychological and disability questionnaires as reference standards to validate the SBT [1, 12, 13, 15, 19]. For instance, Beneciuk et al. clustered the patients based on uni-dimensional psychological and disability questionnaires, and found 2 subgroups of patients with high and low intensity of clinical and psychological profile [12]. Based on this finding, Beneciuk et al. concluded that SBT cutoff scores in a physical therapy setting may need to be reevaluated [12].

Physical impairment measures are described as an important part of clinical examination in patients with LBP [23] and there are valid and reliable assessment tools for physical examination of patients with LBP, e.g. the Physical Impairment Index (PII) [24, 25]. Studies revealed an association between physical impairments, psychological factors, and disability in chronic LBP [26, 27, 28]. However, physical examination measures were not included in the SBT development [15], and there is no study that investigated the relationship between SBT-based subgrouping and physical impairment measures. Therefore, the primary purpose of the current study was to evaluate the relationship between SBT-based categorization and physical examination measures. The second purpose was to evaluate the association between SBT-based subgrouping of patients with LBP and clustering using PII, clinical and psychological measurement tools. We assumed that adding physical impairment examination to the cluster analysis, in addition to a uni-dimensional psychological and disability questionnaires used in the study by Beneciuk et al. [12], may result in a more accurate clustering of patients with LBP. The results of this study can be used by physiotherapists for better screening of patients with LBP.

2. Methods

This was an analytic cross-sectional study. Data was collected between November 2016 and January 2018 from four private and three educational outpatient Physical Therapy (PT) clinics of the university, located in Ahvaz city.

3. Results

During the data collection, 369 patients with LBP were referred to the examiner. 68 were excluded due to acute symptoms, and 144 were excluded because their age was above 65 ( n = 54), they were unable to read and write ( n = 43), or for other reasons ( n = 47). Finally, 157 patients participated in the study. Obtained data from this population and differences between SBT-based subgroups are summarized in Table 1.

Visual inspection of the dendrogram and the elbow method indicated that 3 cluster solutions are optimal. The differences of demographic, clinical, and psychological measures between the clusters are summarized in Table 2. The 3 identified clusters were labeled based on their characteristics. As the level of NPRS, TSK, BDI, FABQ, PCS, ODI and PII measures were relatively high, moderate and low in cluster 1 ( n = 61), 2 ( n = 54) and 3 ( n = 42), respectively (Table 2) we named them severe, moderate and mild physical-psychological distress clusters.

The results of the Pairwise comparisons of the clusters for NPRS, RMDQ, ODI, TSK, BDI, PCS, FABQ-PA, FABQ-W, and PII are summarized in Table 3.

Based on the SBT score, 36% of the patients in the severe physical-psychological distress cluster, 81% of the patients in moderate physical-psychological distress cluster, and 71% of the patients in mild physical-psychological distress cluster were categorized as medium risk (MR). The percentage of high risk (HR) patients in severe, moderate, and mild physical-psychological distress clusters were 63%, 14%, and 0%, and low risk (LR) patients with severe, moderate, and mild physical-psychological distress clusters were 0%, 3% and 29%, respectively (Fig. 1) ( χ 2 = 76.18, 0.01).

OPEN IN VIEWER

The aim of our study was to evaluate the relationship between the SBT-based categorization and physical examination measures and to make a comparison between the clustering of patients with LBP based on uni-dimensional psychological, physical and clinical assessment tools to SBT-based subgrouping in physical therapy settings. Our findings showed that MR patients based on the SBT classification might be categorized in different subgroups when clustered based on uni-dimensional psychological, clinical and physical factors.

Based on SBT most of the patients in our sample were categorized as MR (above 60%), only 9% as LR, and 30% as HR subgroups (Table 1). SBT risk-dependent relationship (LR MR HR) was seen for all uni-dimensional psychological measures (TSK, BDI, PCS, FABQ-PA, and FABQ-W), clinical measures (NPRS, ODI, and RMDQ), and for PII. These were consistent with previous reports [12, 14, 15, 41] and may confirm that clinicians can use SBT as first-line screening questionnaire, instead of using multiple full-length uni-dimensional questionnaires [12, 16, 17]. This SBT risk-dependent relationship for PII may indicate that SBT also can be used as a substitution of a detailed physical examination. However, one should consider that SBT provides overall information about the physical and psychological factors and patients may need further evaluation.

In our study, cluster analysis using uni-dimensional questionnaires identified 3 clusters. This is consistent with the results of Beneciuk et al. who found 3 clusters in patients with acute or sub-acute LBP [37] and inconsistent with the results of another study by the same group in which 2 clusters of patients with chronic LBP have been identified [12]. These conflicts regarding the number of clusters in patients with LBP may be due to the difference in the studied populations or the variables that were used for cluster analysis.

A high proportion of the patients in moderate and mild physical-psychological distress clusters were categorized as MR (approximately 81% and 71%, respectively) and in severe physical-psychological distress cluster as HR (63%) based on SBT classification. Despite significant differences in almost all uni-dimensional psychological, clinical and physical factors, the mean difference of the physical factors, i.e. PII, was relatively higher than those of the psychological measures (TSK, BDI, FABQ, and PCS) when comparing moderate and mild physical-psychological distress clusters. These findings suggest that patients who are categorized as MR, based on SBT subgrouping, may show a different profile when including physical factors in cluster analysis. It may reveal that the prognostic factors included in SBT cannot sufficiently differentiate among patients with chronic LBP at physical therapy settings and including additional physical factors in the SBT questionnaire may results in better differentiating. This may be due to the fact that SBT was initially developed for a primary care clinic setting and its application by physical therapists may require some modification. Including some additional physical construct in the SBT may be required.

In general, the patients who are categorized as LR based on SBT were minimal in each physical-psychological cluster. This may be due to the unequal distribution of the patients in the 3 SBT-based subgroups in our sample.

Some limitations of our study should be considered. First, in cluster analysis, each uni-dimensional questionnaire and the PII were considered as one score, although they consist of several items. Previous research revealed that considering each item of the questionnaires as a single item resulted in more distinguishing characteristics [42]. We suggest that future research considers each item of the assessment tools as a single item. The other limitation was related to the objective nature of the dendrogram and elbow method. Other types of cluster analysis, i.e., a two-step cluster analysis, which determines the number of clusters automatically, with a larger sample of patients can be used in future studies [43].

5. Conclusion

The results of our study indicate that SBT-based subgrouping of patients with LBP into 3 clusters is consistent with clustering based on psychological and physical factors. However, it seems that adding physical factors to the cluster analysis may differentiate among the patients in the MR subgroup based on SBT categorization. Further studies are required to evaluate this finding in populations with an equal number of patients in the 3 SBT-based subgroups.