Table 5.

Generalised Estimating Equation models of associations between baseline and time-varying network characteristics with change from baseline in monthly injection frequency

Baseline network metric	Level	Adjusted β coefficient (95% CI) P-valuea
Degrees	Continuous	−0.28 (−0.95 to 0.38) 0.406
	1 + degrees	−25.50 (−35.34 to −15.65) <0.001
Eccentricity	Continuous	−1.67 (−2.31 to −1.04) <0.001
	1 +	−22.93 (−29.22 to −16.63) <0.001
	2 +	−6.01 (−9.66 to 2.36) 0.001
Closeness centrality	Continuous	5.41 (4.13–6.70) <0.001
Betweenness centrality	Continuous	0.05 (0.03–0.07) <0.001
Clustering coefficient	Continuous	−2.89 (−10.14 to 4.35) 0.433
	>0	8.05 (4.46–11.65) <0.001
Eigenvector centrality	Continuous	9.93 (2.15–17.70) 0.012
Time-varying network metric
Degrees	Continuous	2.42 (1.18–3.66) <0.001
	1 + degrees	−1.64 (−21.46 to 18.18) 0.871
Eccentricity	Continuous	1.42 (−0.39 to 3.24) 0.124
	1 +	−13.76 (−18.57 to −8.95) <0.001
	2 +	−5.61 (−9.54 to −1.68) 0.005
Closeness centrality	Continuous	−1.85 (−4.93 to 1.23) 0.239
Betweenness centrality	Continuous	1.30 (0.90–1.70) <0.001
Clustering coefficient	Continuous	4.52 (−4.33 to 13.36) 0.317
	>0	8.85 (4.50–13.19) <0.001
Eigenvector centrality	Continuous	15.35 (7.10–23.61) <0.001

Bold values are statistically significant (p-value >0.5).

^aEach network metric modelled separately adjusted for age, sex, main drug, OST and interview density