tye has shown you one way. However, let me question the reasons for wanting what you are asking

Hashes are data structures where you can access elements by name rather than by position, as you do with arrays. Therefore a sorted hash is a contradiction. (Although Perl Tie mechanism can change something)

Now, I see two reasons for wanting the result from a query in a hash:

• When you want to access your columns by name
• When you want to process (e.g. display) your column names with their values.

In the first case, you don't really need a sorted hash, since you are using the hash names hardcoded in your application, so that the application itself will set the order. For example

while (my $rec = $sth->fetchrow_hashref()) {
    print $rec->{xy}, $rec->{xz}, $rec->{yz};
}
[download]

If you need to process your columns according to a list of column names, here is a slight variation on tye's answer:

while (my $row = $sth->fetchrow_hashref()) {
    print map( { "$_ => $row->{$_}\t"} @{$sth->{NAME}}),  "\n";
}

#or 

while (my $row = $sth->fetchrow_hashref()) {
    for (@{$sth->{NAME}}) {
        print "$_ => $row->{$_}\t";
    }
    print "\n";
}
[download]

If you just want to use the column names together with their values, then you can use two different methods, according to your needs:

while (my $row = $sth->fetchrow_arrayref()) { # arrayref, not hash
    my $index = 0;
    print "$_ => ", $row->[$index++], "\t"  for (@{$sth->{NAME}});
    print "\n";
}

# or the "C like" way

while (my $row = $sth->fetchrow_arrayref()) { # arrayref, not hash
    for (my $index = 0; $index < $sth->{NUM_OF_FIELDS}; $index++) {
        print "$sth->{NAME}->[$index] => ", $row->[$index], "\t"
    }
    print "\n";
}
[download]

For more ideas, tips, and caveats on the same subject, see DBI Recipes.

Update
If, as you say, you need an array of hash references, then it would be much easier this idiom:

my $aref = $sth->fetchall_arrayref({});
[download]

Notice the empty hashref passed as argument.

 _  _ _  _  
(_|| | |(_|><
 _|