Table 2.

Quality of commonly used birth and neonatal factors

Variable	Variable definition	Missingness by decade	Validity
Variables used to estimate gestational age a
-Date of birth	Year-month-date	None	Excellent
-Estimated date by ultrasound	Year-month-date	1990’s 22.5% 2000’s 12.7% 2010’s 6.9%	Excellent
-First day of last menstrual period (LMP)	Year-month-date	1990’s 5.8% 2000’s 8.7% 2010’s 7.7%	Very good
-Estimated date of delivery by LMP	Year-month-date	1990’s 12.1% 2000’s 10.1% 2010’s 9.9%	Very good
-Postnatal assessment of gestational age in weeks	Completed weeks	1990’s 2.3% 2000’s 0.6% 2010’s 0%	Information on underlying method is missing
-Postnatal assessment of gestational age in days in addition to weeks (range 0–6)	Completed days	1990’s 0.5% 2000’s 1.6% 2010’s 0.4%	-“-
Other birth and neonatal factors
Birth weightb	In grams	1990’s 0.3% 2000’s 0.3% 2010’s 0.1%	Very good
Sex of infant	Check-box 1 = boy; 2 = girl	None	Excellent
Single or multiple birth	Check-box 1 = single; 2 = multiple	None	Excellent
Stillbirthc	Check box recorded at birth	Cannot be calculated	Excellent
Neonatal mortalityd	In completed (range 0–27) days	Cannot be calculated	Excellent
Apgar scores at 5 minutese	Check-box. Values from 0 to 10 are accepted.	1990’s 1.4% 2000’s 0.4% 2010’s 0.6%	See footnote
Infant diagnosesf	ICD codes (ICD-10 since 1997)	Cannot be calculated	See footnote
Infant surgery and proceduresg	Swedish Classification of Operations and Major Procedures	Cannot be calculated	See footnote

^aGestational age in the MBR can be calculated by combining information from date of birth and estimated date of birth by ultrasound, the last menstrual period (LMP), or a postnatal assessment of gestational age in weeks and days. The NBHW has also constructed an algorithm, which, in principle is based on the following hierarchy: gestational age by (a) ultrasound; (b) LMP; (c) as noted in the neonatal record. ^bFor live births, birth weights < 270 and > 6999 g are coded as missing by the MBR. For stillbirths, all values are kept. ^cFrom July 2008, stillbirths from 22 completed weeks are included (from 1973 through June 2008, only stillbirths from 28 completed weeks were included). ^dAll births recorded in the MBR are individually matched to the Total Population Register, kept at Statistics Sweden (including information on live births and date of death). Neonatal deaths are recorded in the Cause of Death Register, which includes information of all deaths in Sweden. ^eApgar score is also recorded at 1 and 10 min (see Table S1 in web supplement). A cautious approach (to enhance validity) is that, in infants with recorded very low Apgar scores at 5 or 10 min (i.e. “0” or “1”), also check information on other Apgar score values, infant diagnoses, and care at the neonatal ward. Some infants with a recorded Apgar score value at 10 min = 0 (i.e. no sign of life), had full Apgar score (i.e., 10) at 5 min. This likely error was noted during some years, but no such discrepancies in Apgar score values were noted for those with Apgar scores at 1 or 5 min = 0. Reasons for missing information on Apgar score at 10 min could be full Apgar score at 5 min (i.e., additional reporting at 10 min seems to be of little clinical importance). ^fValidity of infant diagnoses in the MBR should be similar to corresponding information in the Patient Register(5). ^gIn a study of moderately preterm infants, infant surgery and procedure codes were underreported in the MBR, compared with similar information from the Swedish Neonatal Quality Register (see thesis by Altman [64])