Table 2:

Comparison of a research cohort (Pediatric Sarcoma as training data) and a clinical cohort (CCG as validation data) on the extent of missing ages at last contact and missing ages at cancer diagnoses at both family and individual levels. Summary statistics for the number of individuals per family are reported to contrast the depth of data collection procedures in research and clinical cohorts as they happen in the unit of families

	RPB training data	CCB validation data
Number of families
All family members
Complete data	189 (100%)	10 (8%)
Missing ages at last contact only	0 (0%)	46 (37%)
Missing ages at cancer diagnosis only	0 (0%)	0 (0%)
Missing both ages at last contact and ages at cancer diagnosis	0 (0%)	68 (55%)
Total	189	124
Chi-square test	P < 0.001
First-degree relatives and spouse only
Complete data	189 (100%)	68 (55%)
Missing ages at last contact only	0 (0%)	41 (33%)
Missing ages at cancer diagnosis only	0 (0%)	10 (8%)
Missing both ages at last contact and ages at cancer diagnosis	0 (0%)	5 (4%)
Total	189	124
Chi-square test	P < 0.001
Number of individuals
All family members
Complete data	3,706 (100%)	1,748 (53%)
Missing ages at last contact only	0 (0%)	1,339 (41%)
Missing ages at cancer diagnosis only	0 (0%)	138 (4%)
Missing both ages at last contact and ages at cancer diagnosis	0 (0%)	72 (2%)
Total	3,706	3,297
Chi-square test	P < 0.001
First-degree relatives and spouse only
Complete data	1,126 (100%)	487 (79%)
Missing ages at last contact only	0 (0%)	105 (17%)
Missing ages at cancer diagnosis only	0 (0%)	19 (3%)
Missing both ages at last contact and ages at cancer diagnosis	0 (0%)	2 (0.3%)
Total	1,126	613
Chi-square test	P < 0.001
Number of individuals per family
Min	3.00	1.00
5% percentile	4.00	1.00
10% percentile	4.80	4.00
25% percentile	6.00	16.00
Median	7.00	26.50
Mean	19.69	26.59
75% percentile	10.00	36.00
90% percentile	14.80	48.00
95% percentile	72.20	53.85
Max	719.00	75.00