use strict;
use warnings;

my $s1 = '<citref idrefs="cit">5c,5d,5Es</citref>';
my $s2 = '<qwerty alrefs="xad">5c,5d,5Es</qwerty>';

print parse($s1)."\n";
print parse($s2)."\n";

sub parse {
    my $orig = shift;
    my ($cit, $idrefs, $values, @values);

    $orig    =~ m/.*?>(.*)<\//;
    $values  = $1;
    @values  = split ',', $values;
    $orig    =~ m/.*?"(.*)">/;
    $cit     = $1;
    $idrefs .= "$cit$_ " for @values;
    chop $idrefs;

    $orig =~ s/(.*?=").*(">.*)/$1$idrefs$2/;

    @values = map{ s/([a-z]+)/<it>$1<\/it>/i; $_;} @values;
    $values = join ',', @values;
    $orig   =~ s/>.*?</'>'.$values.'<'/es;

    return $orig;
}
[download]

And the output

<citref idrefs="cit5c cit5d cit5Es">5<it>c</it>,5<it>d</it>,5<it>Es</i
+t></citref>
<qwerty alrefs="xad5c xad5d xad5Es">5<it>c</it>,5<it>d</it>,5<it>Es</i
+t></qwerty>
[download]