The voltage/current spec suggests the load's neutral connected to mid supply or connected through a DC blocking capacitor. So the problem is identical to a bipolar +/- 6V supply configuration.
If you are motivated, you can calculate the exact power disspation by integral calculus. If I understand right, alfa = 90° (reactive load) is worst case. I guess, it's sufficient to use a simplified upper bound, I suggest 6V * mean(|iout|), or about 0.38 W. OP bias current adds some losses, of course.
For an amp with real load, there's a simple upper bound with a rectangular output waveform of +/- 3V. You get 3V * 100 mA = 0.3 W maximum power disspitation, for the load and the amplifier as well. When increasing the output voltage, the amplifier dissipation decreases below this value. But the behaviour is different with a reactive or complex load.