While I agree with Peter about filtering in the data file, a possible
alternative method would be to have four different merge documents each
accessing the same data file.

In this case the label would have

Case 1
{IF {Mergefield Country} = "Canada" "{Mergefield LastName} {Mergefield
Initial}
{Mergefield Address1} {Mergefield Address2}
{Mergefield City}, {Mergefield Prov}
{Mergefield PostalCode}
{Mergefield Country}"}

Case 2
{IF {Mergefield Country} = "Canada" "{Mergefield LastName} {Mergefield
Initial}
{Mergefield Address1} {Mergefield Address2}
{Mergefield City}, {Mergefield Prov}
{Mergefield PostalCode}
{Mergefield Country}"}

Case 3
{ IF { Mergefield Country } = "Canada" "{ IF { Mergefield Email } = "" "{
Mergefield LastName } { Mergefield Initial }

{ Mergefield Address1 } { Mergefield Address2 }

{ Mergefield City }, { Mergefield Prov }

{ Mergefield PostalCode }

{ Mergefield Country }" ""}" "" }

Case 4

{ IF { Mergefield Country } <> "Canada" "{ IF { Mergefield Email } = "" "{
Mergefield LastName } { Mergefield Initial }

{ Mergefield Address1 } { Mergefield Address2 }

{ Mergefield City }, { Mergefield Prov }

{ Mergefield PostalCode }

{ Mergefield Country }" ""}" "" }

This would test each record to see that it fits the criteria

You could even save each of the four blocks of fields as autotext entries
and add them to the first label of a blank merge label document and
propagate then merge. Note that the fields are entered from the keyboard
using CTRL+F9 for each pair of brackets.

For other examples see http://www.gmayor.com/formatting_word_fields.htm