Table 3.

Quality assessment of randomized trials of maintenance treatment in AML

	Reference	No. of patients*	Quality domains (risk of bias)
			No. of postremission courses	Sequence generation	Allocation concealment	Blinding	Incomplete outcome data	Selective outcome reporting	Other sources of bias
1	22	41	1	Unclear	Unclear	High	Unclear	Unclear	High
2	20	74	2-3	Unclear	Unclear	Unclear	Unclear	Unclear	High
3	11	145	1	Unclear	Unclear	Unclear	High	High	Unclear
4	13	32	1	Unclear	Low	Unclear	High	Unclear	High
5	16	45	6	Unclear	Unclear	Unclear	Low	Low	Low
6	15	76	1	Unclear	Unclear	Unclear	Unclear	Low	Unclear
7	18	70	2	Unclear	Low	Unclear	Unclear	Unclear	High
8	10	163	1	Unclear	Unclear	Unclear	High	Unclear	High
9	14	102	1	Unclear	Unclear	Unclear	Unclear	Low	Unclear
10	21	528	1	Unclear	Low	Unclear	Unclear	Low	Unclear
11	17	161	2	Unclear	Unclear	Unclear	Unclear	Low	Low
12	19	154	3	Unclear	Unclear	Unclear	Unclear	High	Low
13	12	226	1	Unclear	Unclear	Unclear	Unclear	Low	Unclear

Positive trials are shown in bold. Details of scoring are provided in Higgins et al.⁷ Low corresponds to low risk of bias; high corresponds to high risk of bias. The following score/domain correspondence was used: sequence generation (method to generate the allocation sequence [eg, random number table, computer random number generator, or minimization], allocation concealment (method used to conceal the allocation sequence [eg, central allocation, sequentially numbered drug containers of identical appearance]), binding (measures used to blind study participants and personnel from knowledge of which intervention a participant received [eg, no blinding, single blinded, or double blinded]), incomplete outcome data (completeness of outcome data for each main outcome, including attritions and exclusions and reasons for each [eg, no missing outcome data, imputation, reasons for missing outcome data unlikely to be related to true outcome, missing outcome data balanced in numbers across intervention groups with similar reasons for missing data across groups]), selective outcome reporting (the possibility of selective outcome reporting [eg, all prespecified outcomes have been reported in the prespecified way]), other sources of bias (any concerns for bias from other sources [eg, early termination as a result of some data-dependent process]).

^*Randomly assigned between maintenance and observation arms.