The randomisation was prepared by writing the two treatment options on papers and then placing the papers in an order taken from a table of random numbers generated from a computer. The papers were folded and put in sealed, numbered envelopes. It was not possible to reveal the choice of treatment without opening the envelopes.