TABLE 3.

Risk of fatal COVID-19 in various clusters

	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Cluster 5	Cluster 6
Subjects n (%)	51 (20.1)	21 (8.3)	35 (13.8)	15 (5.9)	82 (32.3)	50 (19.7)
Cluster 1
Cluster 2	1.94 (0.86–4.37), p=0.11
Cluster 3	3.36 (1.51–8.55), p=0.004	1.87 (0.66–5.24), p=0.24
Cluster 4	2.95 (0.88–9.93), p=0.08	1.51 (0.39–5.79), p=0.55	0.88 (0.23–3.41), p=0.89
Cluster 5	15.34 (4.55–51.72), p<0.0001	8.39 (2.17–32.51), p=0.002	5.24 (1.34–20.48), p=0.02	5.36 (1.08–26.67), p=0.04
Cluster 6	21.69 (2.89–163.61), p=0.003	10.97 (1.20–82.67), p=0.03	7.33 (0.89–60.24), p=0.067	7.41 (0.75–72.75), p=0.09	1.66 (0.17–16.43), p=0.67

Data presented as HR (95% CI), p-value unless otherwise indicated. Values represent hazard ratios for death from COVID-19, with 95% confidence intervals for the average effect of each cluster compared with other clusters. p-values assess whether the distribution of characteristics within a given cluster is significantly different from a characteristic in other clusters in a regression model adjusted for age and WHO severity ordinal scale [21] at the time of admission. Numbers highlighted in boldface represent clusters that had significantly (p<0.05) different characteristics from characteristics in other clusters. CI: confidence interval; HR: hazard ratio; WHO: World Health Organization.