Intuitively the standard practice of using only 1 antibiotic for a standard (non-resistant) infection poses the highest risk of developing resistance (because the bacterial population only has to beat the antibiotic it faces).
What if every antibiotic prescription had to be a mix of 2 or more antibiotics with different modes of action?
Assuming all the bacteria in a given population have to "develop" (mutate) antibiotic resistance on their own (could be as a plasmid or part of the bacterial genome, no cheating allowed via plasmid or gene transfer from bacteria outside the population), then resistance mutations would occur randomly with a low but significant chance per replication. If the bacterial population has to face 2 fundamentally different antibiotics at the same time, then bacteria that want to survive have to develop resistance to both (at the same time) which would be another order of difficulty.
Mathematically, if the chance per replication of developing resistance to each of the 2 antibiotics can be written as 1/A and 1/B respectively (where A and B are large numbers), then the chance of a bacterial cell replicating a mutant that survives the 2-antibiotic mix is (1/A) * (1/B). For a 3-antibiotic mix the chance becomes (1/A) * (1/B) * (1/C). As more are added to the mix, the survival chances quickly get astronomically stacked against the bacteria, with none of them surviving or getting a chance to spread resistance. Importantly, this formula assumes that all antibiotics in the mix use different enough modes of action that resistance to 1 of them doesn't significantly confer resistance to any other in the mix.