Table 3.

Cox proportional hazards model: baseline and time-varying network metrics as predictors of injecting frequency progression

		First progression event (events = 162)	Six-month sustained progression (events = 75)
Baseline network metric	Level	Adjusted HR (95% CI) P-valuea	Adjusted HR (95% CI) P-valuea
Degrees	Continuous	1.00 (0.94–1.07) 0.966	0.90 (0.81–0.99) 0.044
	1 + degrees	1.08 (1.02–1.18) 0.032	0.54 (0.22–1.33) 0.181
Eccentricity	Continuous	0.95 (0.89–1.02) 0.130	1.02 (0.93–1.11) 0.666
	1+	0.49 (0.28, 0.85) 0.012	0.80 (0.36, 1.77) 0.588
	2+	0.64 (0.46, 0.89) 0.009	0.72 (0.45, 1.17) 0.184
Closeness centrality	Continuous	0.88 (0.74–0.98) 0.039	1.04 (0.88–1.23) 0.670
Betweenness centrality	Continuous	1.00 (1.00–1.01) 0.415	1.00 (0.99–1.00) 0.671
Clustering coefficient	Continuous	0.83 (0.46–1.47) 0.514	0.47 (0.17–1.29) 0.144
Eigenvector centrality	Continuous	0.59 (0.28–1.26) 0.172	0.36 (0.09–1.41) 0.142
Time-varying network metric	Level	Adjusted HR (95% CI) P-valueb	Adjusted HR (95% CI) P-valueb
Degrees	Continuous	1.04 (0.99–1.10) 0.151	1.04 (0.78–1.28) 0.828
	1 + degrees	1.41 (1.21–1.64) <0.001	1.05 (0.86–1.20) 0.479
Eccentricity	Continuous	0.99 (0.75–1.30) 0.923	0.78 (0.56–0.99) 0.048
	1+	0.95 (0.80, 1.14) 0.600	0.73 (0.57, 0.95) 0.017
	2+	0.47 (0.39, 0.56) <0.001	0.84 (0.65, 1.09) 0.190
Closeness centrality	Continuous	1.32 (1.19–1.46) <0.001	1.25 (1.01–1.54) 0.043
Betweenness centrality	Continuous	1.01 (0.98–1.03) 0.560	0.99 (0.96–1.03) 0.617
Clustering coefficient	Continuous	0.78 (0.55–1.10) 0.158	0.67 (0.39–1.15) 0.143
Eigenvector centrality	Continuous	1.40 (1.20–1.63) <0.001	1.25 (1.02–1.55) 0.039

^aEach network metric modelled separately adjusted for sex, baseline injection frequency and interview density.

^bEach network metric modelled separately adjusted for age, sex, baseline injection frequency and interview density.