Description of studies included in analysis

Figure A

Figure B

Figure C

Figure D

Description of studies included in analysis

Variable	% (No)
Disease:
Acute lymphoblastic leukaemia	43 (65)
Acute myeloblastic leukaemia	8 (12)
Hodgkin’s disease	3 (5)
Lymphoma	10 (15)
Rhabdomyosarcoma	6 (10)
Wilm’s tumour	8 (12)
Others	22 (33)
Type of treatment:
Neo- adjuvant	3 (4)
Adjuvant	24 (36)
Induction	7 (11)
Consolidation	7 (10)
Maintenance	39 (60)
Supportive	2 (3)
Definitive*	17 (26)
Other	1 (2)
Type of intervention:
Chemotherapy	59 (89)
Radiation therapy	2 (3)
Chemotherapy and radiation therapy	31 (48)
Others	8 (12)
Study design:
Parallel	97 (147)
Cross over	3 (5)
No masking:	100 (152)
Expected difference in outcomes, and sample size prespecified:
Yes	78 (119)
No	9 (14)
Missing data	13 (19)
a and b error prespecified:
Yes	76 (116)
No	11 (17)
Missing data	13 (19)
Type of comparison:
Active	91 (138)
Placebo or no therapy	9 (14)
Primary endpoint:
Survival	43 (65)
Event free survival	39 (60)
Others	18 (27)
Adequate allocation concealment:
Yes	93 (142)
No	7 (10)
Intention to treat analysis:
Yes	93 (142)
No or unclear	7 (10)
Description of drop outs:
Yes	93 (141)
No	7 (11)

*With intent to prolong survival or achieve cure.

Fig A Summary analyses according to publication status

Figure A [As supplied by authors]

  

 

 

Fig B Distribution of outcomes according to investigator’s judgment and achievement of statistical significance (126 trials involved 152 comparisons; there were missing data for 7% of studies regarding statistical significance. However, survival data from these studies were available for meta-analysis).

Figure B [As supplied by authors] 

Fig C Children’s Oncology Group - Forest Plot of all trials included in the analysis – Overall Survival

Fig C [As supplied by authors] Children’s Oncology Group - Forest Plot of all trials included in the analysis – Overall Survival

 

Random-effect analysis: Meta-Analysis

 

 

 

  

 

 

 

 

Fig D Summary analyses according to the methodological quality of trials and the choice of comparator. Notation same as in fig 2. The results were not affected by any of these methodological factors. ‡ Diff=difference; Calc=calculation.

Figure D [As supplied by authors]