Derivation and validation of a novel method to subgroup patients with functional dyspepsia: beyond upper gastrointestinal symptoms

Background: Conventionally, patients with functional dyspepsia are subgrouped based on upper gastrointestinal symptoms, according to the Rome criteria. However, psychological co-morbidity and extraintestinal symptoms are also relevant to functional gastrointestinal disorders. Aim: To investigate whether it is possible to subgroup people with functional dyspepsia using factors beyond upper gastrointestinal symptoms. Methods: We collected demographic, symptom and psychological health data from adult subjects meeting the Rome III criteria for functional dyspepsia in two secondary care cross-sectional surveys in Canada and the UK. We performed latent class analysis, a method of model-based clustering, to identify specific subgroups (clusters). For each cluster, we drew a radar plot, and compared these by visual inspection, describing cluster characteristics. Results: In total, 400 individuals met Rome III criteria for functional dyspepsia in the Canadian cohort, and 262 the UK cohort. A four-cluster model was the optimum solution and the characteristics of the clusters were almost identical between the two cohorts. The clusters were defined by a pattern of gastrointestinal symptoms and were further differentiated by the extent of extraintestinal and psychological co-morbidity. Cluster 1 (mean age 46.7 years, 66.7% female) consisted of epigastric pain and nausea with high psychological burden, cluster 2 (mean age 41.5 years, 77.7% female) high overall gastrointestinal symptom severity with high psychological burden, cluster 3 (45.8 years, 67.2% female) oesophageal symptoms and early satiety with low psychological burden, and cluster 4 (mean age 40.4 years, 71.5% female) postprandial fullness with low psychological burden. We validated the model derived using the Canadian study population externally by applying it to the UK dataset. We demonstrated reproducibility; it would perform similarly when applied to a different dataset. Conclusions: Latent class analysis identified four distinct functional dyspepsia subgroups characterised by varying degrees of gastrointestinal symptoms, extraintestinal symptoms and psychological co-morbidity. Further research is needed to assess whether they might be used to direct treatment.