Table 4:

Meta-regression of potential sources of heterogeneity in the pooled all-cause crude mortality rate (CMR) and standardised mortality ratio (SMR)

	Crude mortality rate				Standardised mortality ratio
	N studies	Coefficient (SE)	Adj. R2	P	N studies	Coefficient (SE)	Adj. R2	P
Sample characteristics at baseline
% Women	9	0.194 (0.543)	−9.76%	0.576	9	0.038 (0.051)	39.96%	0.045
% Injecting	8	5.871 (3.272)	68.62%	0.019	8	3.091 (2.472)	17.35%	0.208
Type of cocaine use			−1.54%				3.19%
Cocaine/Cocaine and crack cocaine	16	ref			16	ref
Speedball/Cocaine and Heroin	6	1.383 (0.546)		0.422	2	2.113 (1.400)		0.276
Geographic region			1.20%				−7.77%
% Western Europe	8	ref			8	ref
% High-Income North America	7	1.689 (0.755)		0.262	7	−0.207 (0.494)		0.682
% Tropical Latin America	1	2.447 (2.199)		0.337	1	0.844 (0.995)		0.412
Study characteristics
Year of final follow-up	16	0.939 (0.034)	12.54%	0.103	16	0.925 (0.035)	16.88%	0.056
Sample size	15	1.000 (< 0.001)	−7.82%	0.797	15	1.000 (< 0.001)	0.29%	0.345
Person years	16	1.014 (0.038)	−6.45%	0.713	16	0.973 (0.038)	−4.64%	0.502
Recruitment setting			−14.62%				20.51%
Treatment clinics and other health services	6	ref			6	ref
Hospital	3	0.750 (0.492)		0.664	3	0.247 (0.146)		0.034
Other	7	0.961 (0.489)		0.939	7	0.587 (0.268)		0.264
Sampling frame			34.92%				24.65%
National/sub-national	9	ref			9	ref
City	7	2.776 (0.985)		0.012	7	2.579 (1.042)		0.034

Note. Regions are defined as per the Global Burden of Disease (GBD) project. No studies were found for the following GBD regions: Central Asia, Central Europe, Eastern Europe, Australasia, High-Income Asia Pacific, Southern Latin America, Andean Latin America, Caribbean, Central Latin America, North Africa & Middle East, South Asia, East Asia, Oceania, Southeast Asia, Central Sub-Saharan Africa, Eastern Sub-Saharan Africa, Southern Sub-Saharan Africa or Western Sub-Saharan Africa. Ref: reference category; SE: standard error