Influence of definition based versus pragmatic birth registration on international comparisons of perinatal and infant mortality: population based retrospective study

We thank Zeitlin et al [1] and Gupta et al [2,3] for their comments related to our paper. We agree that international variations in birth registration are of concern and applaud efforts by the Euro-Peristat group to achieve common standards. We also agree that perinatal and infant mortality indices are best compared among live births ≥28 weeks gestation (or ≥1000 g birth weight) in less industrialized countries.

Unlike Zeitlin et al [1] and Gupta et al [3], however, we are less sanguine about birth registration in industrialized countries. We are well aware that approximately 45-50% of infant deaths in Canada occur among infants with birth weights <1000 g [4]; care of such infants represents an enormous cost to the Canadian health care system. Nevertheless, we followed the World Health Organization’s recommendation for using birth weight-specific mortality >=1000 g for international comparisons [5]. This index was used as a heuristic to attenuate the bias in international comparisons caused by variable birth registration at very early gestation. In fact, there are numerous other impediments to international comparisons including misclassification of live births as stillbirths and incomplete registration of fetuses and infants with serious congenital malformations and those delivered following prenatal diagnosis and pregnancy termination. Variation in the modality of gestational age ascertainment is another factor that precludes simple comparisons of gestational age-specific mortality [6]. Comparisons of gestational age-specific infant mortality rates in the United States (where gestational age in national statistics is typically based on menstrual dating) versus those from other countries (which tend to rely on ultrasound based estimates of gestational age) are a case in point [7].

Although rates of live births and stillbirths <500 g or <22 weeks are substantially higher in Canada and the United States than in Europe, there is evidence of regional variability in the completeness of reporting even within these 2 countries [4,8,9]. Our analyses of data from Europe suggest similar regional variability. The apparent lack of a relationship between the proportion of live births <500 g and crude neonatal mortality rates [1] is due to the influence of countries that use gestational age, birth weight or other criteria to limit the registration of extremely early gestation births. A recalculation carried out with the 17 European countries that have ‘no limit’ birth registration (including Norway [10]) showed a healthy association between live births <500 g and crude neonatal mortality rates (Figure 1). Random intercept Poisson regression showed that the crude neonatal mortality rate increased by 3.1% (95% confidence interval 2.9 to 3.4%, P<0.0001) for a unit increase in the proportion of live births <500 g. In other analyses, rankings of European countries by crude neonatal mortality rates and neonatal mortality rates ≥1000 g birth weight also showed differences; only 12 of 23 countries retained the same ranks (+/- 2 ranks) for both indices. This discrepancy was also highlighted in contrasts of neonatal deaths ≥1000 g birth weight when expressed as a proportion of all neonatal deaths; countries such as Latvia (75.9%, 95% confidence interval 67.0-83.3) and Sweden (74.3%, 95% CI 67.8-80.1) had substantially higher proportions of neonatal deaths ≥1000 g compared with others such as England and Wales (51.6%, 95% CI 49.5-53.7) and Germany (50.8%, 95% CI 48.5-53.1).

References
1. Zeitlin J, Blondel B, Mohangoo AD, Cuttini M, Macfarlane A, Gissler M, Alexander S, Szamotulska K on behalf of the Euro-Peristat Scientific Committee. Re: Influence of definition based versus pragmatic birth registration on international comparisons of perinatal and infant mortality: population based retrospective study. BMJ 22 February 2012.
2. Gupta N, Jani KK, Kishore J. Saving the babies of lower birth weights and gestational ages is an index of better neonatal services. BMJ 19 February 2012.
3. Gupta N, Singh S. Two different cut-offs can be used separately for developed (20, 22 or 24 weeks) and developing (28 weeks of gestation) countries. 23 February 2012.
4. Public Health Agency of Canada. Canadian Perinatal Health Report, 2008 Edition. Ottawa, 2008.
5. World Health Organization. International statistical classification of diseases and related health problems. 10th revision. Vol 2. Instruction manual. WHO, 1993:129-34.
6. Joseph KS, Huang L, Liu S, Ananth CV, Allen AC, Sauve R, Kramer MS. Reconciling the high rates of preterm and postterm birth in the United States. Obstet Gynecol 2007;109:813-22.
7. MacDorman MF, Mathews TJ. Behind international rankings of infant mortality: how the United States compares with Europe. National Center for Health Statistics, 2009. (NCHS data brief, no 23.)
8. Ehrenthal DB, Wingate MS, Kirby RS. Variation by state in outcome classification for deliveries less than 500 g in the United States. Matern Child Health J 2011;15:42-8.
9. Joseph KS, Kramer MS, Allen AC, Sauve R. Infant mortality in Alberta and all of Canada. CMAJ 2005;172:856-7.
10. Euro-Peristat Project in collaboration with SCPE, Eurocat and Euroneostat. European perinatal health report: data for 2004. Eurocat and Euroneostat, 2008.