“…In intervention studies, random sequence generation was adequate in 4 studies, 16,19,20,27,33 unclear in 4, 15,21,26,29 and inadequate in the remainder (where participants were recruited using nonrandom methods under the direction of facility or research staff, 11,18,23,24,28,31,32 using screening tests 10,22 or resident lists 13,30 ; Figure 2). None clearly demonstrated adequate allocation concealment, although 7 were judged "unclear" 15,19e21,26,29,33 (3 of these were low risk for random sequence generation 19,20,33 ).…”