To calculate replication delay in pgpool, the value is obtained from the view pg_stat_replication.replay_lag.

However, sometimes when executing show pool_nodes, it displays a significantly larger number.

In the pg_stat_replication view, the replay_lag is represented as values like 00:00:00.778134.

How repliation_delay calculated at pgpool? I suspect it is converted into microseconds.
I am looking for expert opinion.

node_id  |   hostname   | port | status | lb_weight |  role   | select_cnt  | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change
---------+--------------+------+--------+-----------+---------+-------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       |      db1     | 5678 | up     | 0.200000  | primary | 12401123776 | false             | 0                 |                   |                        | 2023-05-28 00:45:31
 1       |      db2     | 5678 | up     | 0.200000  | standby | 11909003735 | false             | 0                 | streaming         | async                  | 2023-05-28 00:45:31
 2       |      db3     | 5678 | up     | 0.200000  | standby | 11930293446 | true              | 61768             | streaming         | async                  | 2023-05-28 00:45:31
 3       |      db4     | 5678 | up     | 0.200000  | standby | 11829890748 | false             | 776136            | streaming         | async                  | 2023-05-28 00:45:31
 4       |      db5     | 5678 | up     | 0.200000  | standby | 11898267765 | false             | 124128            | streaming         | async                  | 2023-05-28 00:45:31