Identifying subgroups of patients using latent class analysis: should we use a single-stage or a two-stage approach? A methodological study using a cohort of patients with low back pain

Abstract

BackgroundHeterogeneity in patients with low back pain (LBP) is well recognised and different approaches to subgrouping have been proposed. Latent Class Analysis (LCA) is a statistical technique that is increasingly being used to identify subgroups based on patient characteristics. However, as LBP is a complex multi-domain condition, the optimal approach when using LCA is unknown. Therefore, this paper describes the exploration of two approaches to LCA that may help improve the identification of clinically relevant and interpretable LBP subgroups.MethodsFrom 928 LBP patients consulting a chiropractor, baseline data were used as input to the statistical subgrouping. In a single-stage LCA, all variables were modelled simultaneously to identify patient subgroups. In a two-stage LCA, we used the latent class membership from our previously published LCA within each of six domains of health (activity, contextual factors, pain, participation, physical impairment and psychology) (first stage) as the variables entered into the second stage of the two-stage LCA to identify patient subgroups. The description of the results of the single-stage and two-stage LCA was based on a combination of statistical performance measures, qualitative evaluation of clinical interpretability (face validity) and a subgroup membership comparison.ResultsFor the single-stage LCA, a model solution with seven patient subgroups was preferred, and for the two-stage LCA, a nine patient subgroup model. Both approaches identified similar, but not identical, patient subgroups characterised by (i) mild intermittent LBP, (ii) recent severe LBP and activity limitations, (iii) very recent severe LBP with both activity and participation limitations, (iv) work-related LBP, (v) LBP and several negative consequences and (vi) LBP with nerve root involvement.ConclusionsBoth approaches identified clinically interpretable patient subgroups. The potential importance of these subgroups needs to be investigated by exploring whether they can be identified in other cohorts and by examining their possible association with patient outcomes. This may inform the selection of a preferred LCA approach.Electronic supplementary materialThe online version of this article (doi:10.1186/s12891-017-1411-x) contains supplementary material, which is available to authorized users.

Novel approach towards musculoskeletal phenotypes

“…It has thus been argued that research should advance towards clustering of patients who share similar features and prognostic factors across the biopsychosocial domains, and tune management to the characteristics of the different clusters (Hill et al, ). Studies using cluster analyses in the MSK patient population have either been confined to specific pain diagnosis (Barons et al, ; Carlesso, Raja Rampersaud, & Davis, ; de Luca, Parkinson, Downie, Blyth, & Byles, ; Murphy, Lyden, Phillips, Clauw, & Williams, ; Nielsen, Kent, Hestbaek, Vach, & Kongsted, ; Rabey, Smith, Beales, Slater, & O'Sullivan, ; Reme et al, ; Stynes, Konstantinou, Ogollah, Hay, & Dunn, ), a single subgrouping variable (i.e. pain sites; Hartvigsen, Davidsen, Hestbaek, Sogaard, & Roos, ; Lacey et al, ) or factors related to one specific dimension (i.e.…”

Latent class analysis derived subgroups of low back pain patients – do they have prognostic capacity?

“…The main focus in this study was the subgroup membership in the two subgroupings previously identified by way of two separately conducted LCAs [ 30 ], one using a traditional single stage approach and another using a two-stage approach [ 33 , 34 ]. No imputations were performed during the subgrouping analyses, as LCA uses a likelihood approach which accommodates the inclusion of patients with missing values.…”

“…This resulted in seven single-stage subgroups and nine two-stage subgroups. The selection of the preferred subgroup solutions for each LCA approach had been informed by both statistical performance measures and a qualitative evaluation of clinical interpretability (face validity) [ 30 ]. The qualitative evaluation emphasised differences between the subgroups that were not only on a continuum of severity, but distinct differences in scoring patterns across the baseline characteristic.…”

“…Using this approach, potentially novel relationships between the baseline characteristics were more likely to be identified. A brief description of each subgroup is presented in Table 1 and a thorough description was previously published [ 30 ]. Within the analyses in this paper, each subgroup solution was used as a categorical variable.…”

“…A recent study [ 30 ] attempted to address these issues using Latent Class Analysis (LCA), as LCA allows exploration of how multidimensional baseline variables might form distinct patient subgroups when these are not assumed to be related to each other in the same way in all patients [ 31 ]. In that study, 112 baseline characteristics covering aspects of activity, contextual factors (such as sex, age, educational level and comorbidity), pain, participation, physical impairment and psychology were used in a LCA to identify subgroups of patients.…”