A new hybrid record linkage process to make epidemiological databases interoperable: application to the GEMO and GENEPSO studies involving BRCA1 and BRCA2 mutation carriers

Table 1 An example of a record pair comparison and its PRL likelihood score calculation

	CTR	NUMFAM	SUJID	GENDER	Yob	Mob	Dob	BRCA1	BRCA2	MUT_HGVS	PRL Score
Individual GEMO_{_5789}	1	17455	0001	2	1959	08	05	1	0	c.3403C > T	–
Individual GENEPSO_{_01082300001}	1	08230	0001	2	1958	08	05	1	0	c. 3481_3491del	–
Similarity s	1	0	1	1	0	1	1	1	1	0.7825	–
f	0.02272	0.00025	0.0018	0.5000	0.01098	0.07692	0.03125	0.3333	0.3333	0.0006	–
w	5.45	11.95	9.1	0.99	6.49	3.68	4.99	1.57	1.57	10.69	sum(w) = 56.48
ws*	5.45	0	9.1	0.99	0	3.68	4.99	1.57	1.57	8.36	sum(w * s) = 35.71
score S											0.6322

Ten matching variables were used to identify record pairs: BRCA1 mutational status (BRCA1), BRCA2 mutational status (BRCA2), mutation description using the HGVS nomenclature (MUT_HGVS), gender (GENDER), recruiting center number (CTR), family number (NUMFAM), individual number in the family (SUJID), year of birth (Yob), month of birth (Mob) and day of birth (Dob). BRCA1 and BRCA2 matching variable: 1: “carrier of a BRCA1/2 mutation”, 0: “non-carrier of a BRCA1/2 mutation”. GENDER matching variable: 1: male, 2: female. The similarity vector s in the third row is used as input in the machine learning approaches. The PRL score S is calculated from the weight w and the similarity s

ISSN: 1471-2288