The utility of three national registers-the malformation, birth, and h
ospital in-patient discharge registers-in identification of malformati
ons among 60,255 children born in 1987 in Finland was compared. Inform
ation in the malformation register is collected by specific reporting
of physicians; information on the malformed children in the other two
registers in 1987 through 1990 was routinely collected and identified
by ICD 9-codes (740-759, 7886A). From October 1990 on, the ICD-9 codes
were omitted from the birth register and malformation data were asked
only in a yes-no question. In 1987, the malformation register include
d 1,032, the birth register 3,084 and the inpatient register 2,003 mal
formed infants identified up to the age of 1 year. There was underrepr
esentation of almost all malformation diagnoses in the malformation re
gister. Individual linkage of the three registers showed that the malf
ormation register revealed very few cases not recorded in the birth an
d the in-patient registers. With the adoption of the new recording met
hod in the birth register beginning in October 1990, reports of malfor
mation declined by 75%. Our analyses showed that data sources based on
diagnoses collected routinely and not requiring an explicit decision
on whether or not a problem is a malformation, can be more useful for
routine surveillance of occurrence of malformations. (C) 1993 Wiley-Li
ss, Inc.