Table 1.

Expected mortality differences between migrants and non-migrants for different combinations of selection and exposure effects.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)
Case of return migrants
Expected difference in mortality risk between return migrants and non-migrants:
Short exposure in	>	>	>	>	>	>	=	=	=	<	<	<	<	<	<
Long exposure in	=	>	=	>	=	>	=	=	=	=	<	=	<	=	<
Long exposure outside	=	=	<	<	>	>	=	<	>	=	=	<	<	>	>
Selection effect	neg	neg	neg	neg	neg	neg	none	none	none	pos	pos	pos	pos	pos	pos
Re-adaptation effect (i.e. convergence with non-migrants)	yes	no	yes	no	yes	no	n.t.	n.t.	n.t.	yes	no	yes	no	yes	no
Propagation effect (i.e. origin conditions persist at destination)	no	no	pos	pos	neg	neg	no	pos	neg	no	no	pos	pos	neg	neg
Case of new migrants
Assumed difference in health risks exposure before (B) and after (A) migration	B = A	B = A	B < A	B < A	B > A	B > A	B=A	B < A	B > A	B=A	B = A	B<A	B < A	B > A	B > A
Expected difference in mortality risk after migration between new migrants and non-migrants:
Short exposure in	>	>	=	=	≫	≫	=	<	>	<	<	≪	≪	=	=
Long exposure in	=	>	<	=	>	≫	=	<	>	=	<	<	≪	>	=
Selection effect	neg	neg	neg	neg	neg	neg	none	none	none	pos	pos	pos	pos	pos	pos
Inferred adaptation effect (i.e. convergence with non-migrants)	yes	no	yes	no	yes	no	n.t.	n.t.	n.t.	yes	no	yes	no	yes	no
Inferred socialisation effect (i.e. persistence of exposure B)	no	no	pos	pos	neg	neg	no	pos	neg	no	no	pos	pos	neg	neg

n.t.: not testable. pos: positive. neg: negative. In bold: assumption of no difference in health risks before and after migration in rural areas.