Why are IDLE connections using cpu according to TOP.

I have an overloaded DB and I see several IDLE connections that are using
significant CPU.. (Not Idle in transaction)
Why would an idle process be eating so much cpu? Or is it not actually idle?

Here is an example from pg_top:

last pid: 11821; load avg: 6.11, 6.32, 7.64; up 1+21:05:31
50 processes: 3 running, 42 sleeping, 5 uninterruptable
CPU states: 21.7% user, 0.0% nice, 7.8% system, 46.9% idle, 23.6% iowait
Memory: 29G used, 149M free, 13M buffers, 27G cached
Swap:

PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND
4779 postgres 20 0 4383M 573M disk 3:16 4.79% 39.42% postgres:
gpup gpup 10.202.99.5(46391)
UPDATE

11591 postgres 20 0 4383M 108M sleep 0:12 2.08% 19.61% postgres:
gpup gpup 10.202.99.6(52459)
idle

4191 postgres 20 0 4384M 709M sleep 4:33 2.50% 19.41% postgres:
gpup gpup 10.202.99.6(42288)
idle

10942 postgres 20 0 4383M 242M sleep 0:42 5.08% 16.86% postgres:
gpup gpup 10.202.99.5(58373)
idle

10930 postgres 20 0 4390M 281M sleep 0:43 1.62% 15.30% postgres:
gpup gpup 10.202.99.6(52273)
idle

11571 postgres 20 0 4390M 210M run 0:25 4.32% 14.51% postgres:
gpup gpup 10.202.99.6(52455)
SELECT

11533 postgres 20 0 4383M 109M run 0:14 2.31% 12.75% postgres:
gpup gpup 10.202.99.6(52453)
SELECT

7494 postgres 20 0 4384M 1611M disk 2:31 2.44% 12.35% postgres:
gpup gpup 10.202.99.6(53620) SELECT