Preterm birth rates are stratified into three categories on the PeriStats web site: very preterm (live birth prior to 32 completed weeks); late preterm (live birth delivered between 34 and 36 weeks); and preterm (live birth prior to 37 completed weeks, including both late and very preterm births). Live births at or beyond the 37th week of pregnancy, include term (37-42 weeks) and post-term (42+ weeks) births. Calculations are based on the number of live births to infants in a specific gestational age category divided by all live births excluding those missing data on gestational age, multiplied by 100. Singleton-specific rates are calculated the same way.
The March of Dimes Perinatal Data Center stratifies data by the gestational age categories listed above, however, the individual level algorithm for calculating gestational age in weeks was developed by the National Center for Health Statistics (NCHS). All individual level gestational age calculations are completed by NCHS prior to receiving electronic data. The following provides a detailed description of the NCHS algorithm, and uses data from 2001 as an example:
The primary measure used by NCHS to determine the weeks of completed gestation of the newborn is the interval between the first day of the mother's last normal menstrual period (LMP) and the infant's date of birth. In 2001, the gestational age for 95.1% of live births was calculated using LMP. LMP is reported as the day, month and year.
In cases where the day of LMP is missing (but the month and year are complete), NCHS assigns that record the weeks of gestation of the previous completed record in the file with a similar race and birthweight and the same computed months of gestation. For these records, the gestational age is said to be 'imputed'. The number of months of gestation needed for this process is computed for both missing-day records and completely reported records by subtracting the reported month of LMP from the reported month of birth.
In cases where the month, year, or entire LMP is missing, or when the calculated or assigned (imputed) gestational age appears to be inconsistent with birthweight, the clinical estimate of gestation is used (4.9% of live births in 2001).
In cases where the reported birthweight is inconsistent with both the LMP-computed gestational age and the clinical estimate, the LMP-computed gestational age is used and birthweight is reclassified as "not stated" (less than 0.01% of live births in 2001).
In cases where both the LMP and the clinical estimate are missing, gestational age is set to 'missing.' In 2001, 0.1% of live births were missing gestational age.
As with other indicators, preterm birth rates on PeriStats may differ from those provided directly by state health departments. This may be due to differences in the way the health department and NCHS calculate the gestational age of the infant or in the handling of missing data.
Additional detail on the NCHS algorithm for calculating gestational age can be found in:
- National Center for Health Statistics, S. Taffel, D. Johnson, and R. Heuser: A method of imputing length of gestation on birth certificates. Vital and Health Statistics. Series 2, No. 93. DHHS. Pub. No. (PHS) 82-1367. Public Health Service. Washington. U.S. Government Printing Office, May 1982.
- Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM, Sutton PD. Births: Final data for 2001. National vital statistics reports; vol 51 no. 2. Hyattsville, Maryland: National Center for Health Statistics. 2002.