every_nth = [2, 3]
index = pd.Series(range(len(df.index)))  # or just df.index if it's already the 0-indexed indices
mask = reduce(np.logical_or, [index % i == 0 for i in every_nth])
df[mask.astype(bool).values]