Is /testfor more efficient

You definitely don't need to be using testfor here. The only reason you might want to use testfor before conditionally running something like effect is if you wanted to test for something that can't be tested in a selector, i.e. NBT data, and even then, that would be such a narrow case that you're better off putting that test in a scoreboard. Since you're applying the effect against a selector already, the testfor is basically doing nothing for you; either the testfor succeeds and you run 2 commands, or it fails and you run only 1, whereas without it, you run 1 command, or half a command. If you don't have a lot of commands, you're not going to see a performance difference though.

But this belies another issue, which is testfor is only marginally useful right now, but will be going away in 1.13, replaced with a new and improved execute command.

In the end though, my rule of thumb is to only use testfor if it results in the fewest command blocks, and almost never if chained conditional command blocks target players (or entities in general) instead of the world.