Table 1 |
|||||||
| Details of the data sources | |||||||
| Voight, 2010 | Olsen, 2010 | Kramer, 2001 | Roberts, 1999 | Bonellie, 2008 | Bertino, 2010 | WHO, 2006 | |
| Data source | German Perinatal Survey | Pediatrix Medical Group hospitals | Canadian national file | Australian National Perinatal Statistics Unit | Scottish maternity data collection | Italian Neonatal Study | WHO multicentre growth reference study |
| Sample size | 2,300,000 | 130,111 | 676,605 | 734,145 | 100,133 | 45,462 | 882 |
| n < 30 weeks | 14146 | 11377 | 3247 | 3193 | 2053 | 623 | N/A |
| Lowest gestational age | 22 | 23 | 22 | 20 | 24 | 23 | term |
| Dates | 1995 to 2000 | 1998 to 2006 | 1994 to 1996 | 1991 to 1994 | 1998 to 2003 | 2005 to 2007 | 1997-2003 |
| Data | Weight | Weight, head, length | Weight | Weight | Weight | Weight, head, length | Weight, head, length |
| Exclusion criteria | None stated, included both live and stillborn | Multiple births, congenital anomalies, death before discharge, outlier measures (> 2 x interquartile range below the first and 3rd quartile). | Ontario province was excluded due to problems with data quality. | Omitted multiple and still births (births < 400 grams did not need to be recorded) | Multiple births, lethal anomalies, weights < 250 grams, and outlier measures (> 2 x interquartile range outside the first and 3rd quartile). | Multiple births, stillbirths, major congenital anomalies, and fetal hydrops | Maternal smoking, not breastfeeding, solids before 4 months. Screened for environmental or economic constraints. |
| Method to assess gestational age | Ultrasound assessment 8–14 weeks and Naegle’s rule. | Neonatologist assessment | “early ultrasound has increasingly been the basis for gestational age assessments in recent years” | Dates, prenatal, or postnatal assessment | Clinician assessment based on ultrasound, maternal dates, and clinical estimates | Ultrasound assessment first trimester | Not stated |
| Outliers/smoothing method | Cubic regression, LOESS smoothing, LMS parameter smoothing | LMS methods, with the skew set to one and further manual smoothing | Assumed a log normal distribution of birthweight at each gestational age and compared the probabilities of accurate versus misclassification of infant’s gestational age | Omitted outlier measures (> 2 x interquartile range below the first and 3rd quartile). | Cubic spline fitting | Generalized logistic functions | Omitted outliers > 3 SD, LMS parameter smoothing, skew set to one for weight, cubic spline fitting. |
Fenton and Kim
Fenton and Kim BMC Pediatrics 2013 13:59 doi:10.1186/1471-2431-13-59
