Article Text

Download PDFPDF
P54 Can multi-morbid phenotypes be described in patients with advanced COPD using cluster analysis?
  1. BD James1,
  2. NJ Greening1,
  3. N Toms2,
  4. G Woltmann2,
  5. RC Free2,
  6. P Haldar1,
  7. MC Steiner2,
  8. RA Evans2
  1. 1University of Leicester, Leicester, UK
  2. 2Institute for Lung Health, Respiratory BRU, University Hospitals NHS Trust, Leicester, UK


Introduction and objectives Comorbidities have a negative effect upon outcomes in patients with COPD, and ‘phenotypes’ of comorbidity have been described.1 International guidelines recommend that comorbidities “should be looked for routinely”.2 We aimed to objectively assess comorbidities, and investigate whether comorbidity phenotypes could be described using cluster analysis, in a cohort of patients with advanced COPD.

Methods Patients with advanced COPD were prospectively recruited to undergo a ‘Comprehensive Respiratory Assessment’ (CRA), as previously described.3 13 comorbidities were objectively assessed using validated definitions and their prevalence determined. K-means cluster analysis was applied with the objective measurements, and FEV1%predicted. The clusters formed were compared with respect to demographic features, measures of health status, self-reported exacerbation frequency, and future cardiovascular risk.

Results Between June 2013 and December 2015, 246 patients with advanced COPD underwent a CRA: 61.0% male, mean (SD) age 66.0 (9.1) yrs, FEV1%predicted 31.1% (10.4%). 98.4% of participants were in GOLD combined assessment group D. The prevalence of the 13 comorbidities ranged from 68.8% (muscle wasting) to 7.7% (renal impairment). 93.9% of participants had at least two of the assessed comorbidities. Cluster analysis was applied to a subsample of 203 participants with sufficient data: five multimorbid clusters were identified according to a significantly higher prevalence of certain comorbidities: Cluster 1 – psychological disease, Cluster 2 – left ventricular systolic dysfunction and anaemia, Cluster 3 – features of cachexia, Cluster 5 – features of metabolic syndrome and vitamin D deficiency. Cluster 4 had a significantly lower prevalence of comorbidities. Table 1 shows the differences between the five clusters in demographics, airflow limitation, health status, hospital admissions, use of antibiotics and steroids, and future cardiovascular risk.

Conclusions In this cohort of patients with advanced COPD, five multi-morbid phenotypes were identified. The phenotypes differed significantly in comorbidity prevalence, airflow limitation, health status and future coronary heart disease risk, however the number of hospital admissions in the past year was similar.


  1. Vanfleteren LE, et al. Clusters of comorbidities based on validated objective measurements and systemic inflammation in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013;187(7):728–35.

  2. GOLD Global Strategy updated2016.

  3. Steiner MC, et al. Comprehensive respiratory assessment in advanced COPD: a 'campus to clinic' translational framework.Thorax 2015;70(8):805–8.

Abstract P54 Table 1

Features of multi-morbid clusters

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.