The only statistics that are valid across storage engine implementations are:

• insert
• query
  *update
• delete
• getmore
• commands
• netIn
• netOut
• conn
• time

And in replsets with --discover

• hostname
• set name
• repl (type of node: primary, secondary, mongos)

One approach is to auto-detect the type of storage engine and display the right stats. This breaks down in the --discover mode because you can have mixed sets. This could be overcome, by re-writing the header line every time the node type "changes"