By the way; I always prefer accessing my database rows using hashrefs for the records, to ensure I'm accessing the right fields... If you were to use this method, a partial hash comparison would suffise;

my @keylist=qw(field1 field2 field3); 

for $row (@$list) {
   $id = hashcmp($row,$vals,@keylist) ?
           $row->{id} :
           insert_row($vals)
   ;
}

sub hashcmp {
   my ($a,$b,@keylist) = @_;
   for(@keylist) {
      next unless $a->{$_} || $b->{$_};     # both undef
      return 0 unless                     
          String::Approx::amatch($a->{$_},$b->{$_});
   }
   return 1;
}
[download]

This assumes the named fields are always present in both the $vals and $row hashref...