From the look of your data it appears that you have a hardware problem which cannot be completely resolved by software. The first 2-3 seconds of data show a saturation effect. The data is not really valid until after about 5 seconds and your DC transient does not end until about 15 seconds.
You either need to get a different signal conditioner or decide not to use the data which occurs before the transient dies out.
The true wandering baseline issue is what is going on at 15-16 seconds, 25 seconds, and 58-65+ seconds. Are you trying to analyze the waveform for detecting subtle arrhythmias where you need good waveform fidelity or are you just looking at heart rate?
Lynn