There are several approaches. You can skip those spaces while parsing resulting in empty fields:

my $csv = Text::CSV_XS->new ({
    binary           => 1,
    sep_char         => "\t",
    auto_diag        => 1,
    allow_whitespace => 1,
    });
[download]

but as you are dealing with tab separated data, I'd personally would coose to do it inside the loop

while (my $row = $csv->getline ($fh)) {

    # Check if the 5th field contains data
    if ($row->[4] =~ m/\S/) { # more than just whitespace
        $csv_o->print ($fhv, $row);
        }
    else { # sorry, this is not filled: invalid    
        $csv_o->print ($fhi, $row);
        }
[download]