Photo Credit: Stefanamer
Objective assessment of comorbidities shows potential to improve prognostication and deploy interventions to reduce modifiable risk.
Cluster analysis has been applied to various disease states to provide an exploratory multivariate method for determining the prognostic value of comorbidity patterns. Although this approach has been applied to the study of COPD, the prognostic value of cluster analysis in severe COPD cases remains unclear.
To explore this knowledge gap, Thomas J.C. Ward, MD, and colleagues developed a study that assessed COPD comorbidities objectively, which can be a challenge given that most studies rely on self-reported comorbid conditions. Dr. Ward and colleagues published the findings of the study in Respiratory Medicine. “We hypothesized that cluster analysis of clinically relevant measures of comorbidity would identify groups of individuals with different patterns of comorbidity,” said Dr. Ward. “Furthermore, we hypothesized that comorbidity phenotype would predict mortality and hospital attendance better than comorbidities considered in insolation.”
Comorbidity Clusters Identified
Of the 253 patients receiving advanced COPD care and who underwent a comprehensive respiratory assessment (CRA), 203 participants (age 66 ±9 years) were identified for participation in the study (Table). Upon analysis, the study team identified four clusters: Cluster A, cardiometabolic and anemia (n=56); Cluster B, malnourished and low mood (n=84); Cluster C, obese, metabolic, and mood disturbance (n=32); and Cluster D, less comorbid (n=31). FEV1% did not differ significantly between the identified clusters.
The specific comorbidities distributed among these clusters were anemia, determined by hemoglobin (g/L); symptomatic; anxiety, determined by the Hospital Anxiety and Depression Scale (HADS); symptomatic depression, determined by HADS; cardiac dysfunction, determined by B-type natriuretic peptide (BNP); uncontrolled dyslipidemia, determined by total cholesterol (TC) and HDL; uncontrolled hypertension, determined by blood pressure; muscle wasting, determined by Skeletal Muscle Index (SMI); obesity, determined by BMI; underweight status, determined by BMI; uncontrolled prediabetes and diabetes mellitus, determined by hemoglobin A1C (HbA1C); osteoporosis, determined by T-score; renal impairment, determined by estimated glomerular filtration rate (eGFR); and vitamin D deficiency, determined by 25-OH vitamin D.
Less Predictive Value
When examined individually, the comorbidities did not predict a patient’s time to hospitalization risk. A lower risk of mortality was observed in patients with uncontrolled dyslipidemia (adjusted HR 0.50 [95 % CI 0.26–0.97]) and uncontrolled hypertension (adjusted HR 0.57 [95 % CI 0.34–0.96]).
When compared, the cluster groups had a distinctive difference in mortality (P=0.001). Cluster A, for example, (HR 3.73 [95%CI 1.09–12.82] P=0.036) and Cluster B (HR 3.91 [95%CI 1.17–13.14] P=0.027), when compared with Cluster D, had a much higher risk of mortality during the time preceding study follow-up. Among the clusters, Cluster A had the highest time to hospital admission (HR 2.01 [95%CI 1.11–3.63] P=0.02). Compared to all the other clusters, Cluster C was not associated with an increased risk of death or hospitalization.
The researchers determined that patients in Cluster A were at greater risk of being hospitalized compared to Cluster D (adjusted HR 2.01 [95 % CI 1.11–3.63] P=0.02). Compared to Cluster D, Clusters B and C had no significant difference in time to hospitalization.
In terms of the number of days spent in the hospital, Cluster A had a significantly greater number of days in the hospital compared to Cluster D (adjusted IRR 2.84 [95 % CI 1.15–7.05] P=0.024). Cluster D had no significant difference for days in the hospital compared with Clusters B and C (Table).
Dr. Ward and colleagues concluded, “clustering of patients with advanced COPD into comorbidity phenotypes using widely available and established diagnostic measures of comorbidity provides better prognostic prediction than assessment of comorbidities in isolation.” In terms of applicability to practice, they wrote, “In severe disease, objective assessment of comorbidities has the potential to improve prognostication and thereby deploy interventions to reduce modifiable risk.”
