Ethnic- and gender-specific differences in the prevalence of HIV among patients in opioid maintenance treatment—a case register analysis

Table 2 Model estimates of predictors of HIV-positive status according to GEE analysis

Independent variables	Complete dataset			Imputed dataset
Independent variables	Estimate	SE	P value	Estimate	SE	P value
Intercept	-0.593	0.050	<0.001	-0.631	0.048	<0.001
Calendar year	-0.277	0.038	<0.001	-0.254	0.036	<0.001
Year of birth	-0.039	0.009	<0.001	-0.043	0.009	<0.001
Year of birth^a	-0.054	0.012	<0.001	-0.042	0.010	<0.001
Year of birth^b	0.019	0.005	<0.001	0.013	0.005	0.012
Age	0.043	0.009	<0.001	0.035	0.008	<0.001
Age^c	-0.048	0.006	<0.001	-0.043	0.006	<0.001
Age^d	0.012	0.002	<0.001	0.011	0.002	<0.001
Female, non-Swiss, non-injector	-1.760	0.492	<0.001	-1.435	0.543	0.011
Female, non-Swiss, ever injector	0.497	0.158	0.002	0.457	0.150	0.002
Female, Swiss, non-injector	-1.344	0.197	<0.001	-1.372	0.201	<0.001
Female, Swiss, ever injector	0.186	0.069	0.007	0.167	0.065	0.011
Male, non-Swiss, non-injector	-1.520	0.299	<0.001	-1.422	0.261	<0.001
Male, non-Swiss, ever injector	-0.074	0.109	0.496	-0.087	0.107	0.417
Male, Swiss, non-injector	-1.547	0.149	<0.001	-1.567	0.143	<0.001

The time variables were rescaled to fit the GEE model as follows: calendar year = logarithm of year - 1990, year of birth = year - 1960, age = age - 30. ^aYear of birth = Year of birth × Year of birth / 10. ^bYear of birth = Year of birth × Year of birth × Year of birth / 100. ^cAge = Age × Age / 10, ^dAge = Age × Age × Age / 100. The group ‘Male, Swiss, ever injector’ was reference category and is therefore omitted from the independent variable list. SE, standard error.

ISSN: 1477-7517