Table 4.

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

		First reduction event (events = 218)	Six-month sustained reduction (n = 43 events)
	Level	Adjusted HR (95% CI) P-valuea	Adjusted HR (95% CI) P-valueb
Baseline network metric
Degrees	Continuous	1.07 (1.01–1.14) 0.043	1.10 (0.95–1.27) 0.191
	1 + degrees	0.95 (0.47–1.95) 0.898	1.50 (0.20–11.08) 0.688
Eccentricity	Continuous	0.97 (0.91–1.02) 0.219	0.91 (0.80–1.04) 0.162
	1+	0.96 (0.56, 1.64) 0.877	0.89 (0.31, 2.51) 0.891
	2+	0.84 (0.62, 1.14) 0.268	0.70 (0.38, 1.31) 0.267
Closeness centrality	Continuous	0.94 (0.85–1.05) 0.258	0.87 (0.69–1.10) 0.232
Betweenness centrality	Continuous	1.00 (0.99–1.00) 0.635	1.00 (0.99–1.01) 0.936
Clustering coefficient	Continuous	0.99 (0.59–1.68) 0.997	1.20 (0.40–3.56) 0.742
Eigenvector centrality	Continuous	1.53 (0.72–3.25) 0.269	0.93 (0.14–6.37) 0.941
	>0	1.47 (1.01–2.24) 0.041	1.85 (0.77–4.41) 0.168
Time-varying network metric
Degrees	Continuous	1.10 (1.05–1.16) <0.001	1.14 (1.00–1.29) 0.048
	1 + degrees	1.47 (0.85–2.54) 0.173	1.93 (0.48–7.81) 0.355
Eccentricity	Continuous	1.21 (1.05–1.69) 0.001	1.25 (1.11–1.41) <0.001
	1+	2.41 (2.02, 2.86) <0.001	3.34 (2.27, 4.92) <0.001
	2+	1.21 (1.01, 1.44) 0.036	1.28 (0.87, 1.90) 0.211
Closeness centrality	Continuous	0.89 (0.58–1.36) 0.591	1.19 (0.52–2.73) 0.683
Betweenness centrality	Continuous	1.00 (0.98–1.02) 0.842	1.01 (0.97–1.06) 0.542
Clustering coefficient	Continuous	1.06 (0.75–1.49) 0.733	1.06 (0.47–2.38) 0.882
Eigenvector centrality	Continuous	1.28 (0.85–1.91) 0.240	0.79 (0.25–2.53) 0.693
	>0	0.92 (0.71–1.20) 0.557	1.16 (0.72–1.85) 0.542

^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.