Table 1.

Examples of why certain allocation methods are not random and how they may break concealment.

Example of nonrandom allocation by:	How it may break randomization	How it may break concealment
Allocation of participants first to only one treatment group until desired sample size, then randomization of the rest among treatment groups (e.g., [12])	Nonrandomized participants do not have a known probability of being in the other group(s)	The researcher knows the assignments of the participants enrolled without randomization
Alternating, such as allocating every other individual (e.g., [158, 159])	Participants may enroll in groups in nonrandom ways, and with small numbers of groups this can create imbalances	The researcher knows the next group assignment
Day of the week or time of day of enrollment [160]	Participants with certain characteristics may be more likely to be available for enrollment based on the day of the week or time of day	The researcher knows the group assignment
Patient chart number [161]	Chart numbers may be associated with known or unknown patient characteristics	If chart numbers are not randomly assigned, the researcher may be able to predict the next assignment
Participant characteristics (such as in the Lanarkshire Milk study [5, 6], or matching [162, 163])	Characteristics may not be evenly distributed (i.e., confounding may occur)	The researcher may be able to predict assignments based on characteristics