Just to add a bit to targaman's thoughts, one can probably also factor in the number of aerodromes where the take-off path is benign. Then, comparing that result to the number of aerodromes where the take-off path is not so benign, you'd have to reckon that there's a reasonable chance of getting away with a (somewhat) poorly executed OEI continued take-off.
Can't add anything meaningful to what j_t has said about near and far obstacles. However, for the benign take-off path scenario, take-off performance is based on the Type A chart, which should take care of the problem. In the not-so-benign scenario, performance engineers will crunch the numbers for OEI procedures, for client airlines, thereby also (hopefully) also eliminating the problem of having to define near and far obstacles.