(1995-2000) 33 41 Kazakhstan 40 DHS 1995 (1991-1995) 27 13 Kazakhstan 19 62 DHS 1999 (1994-1999) 24 38 Azerbaijan 79 MICS 2000 (1996) 20 59 Azerbaijan 18 74 RHS 2001 (1996-2000) 17 57 Romania 18 32 RHS 1999 (1995-1999) 21 11 Uzbekistan 49 DHS 1996 (1992-1996) 30 19 Uzbekistan 18 52 MICS 2000 Not reported Moldova 16 Armenia 15 36 DHS 2000 (1995-2000) 15 21 Russia 15 Bulgaria 14 FR Yugoslavia 13 Albania 12 28 MICS 2000 Not reported Georgia 12 43 RHS 1999 (1995-1999) 16 27 FYR Macedonia 12 Ukraine 11 16 RHS 1999 (1995-1999) 14 2 Latvia 11 Belarus 9 Estonia 9 Hungary 8 Lithuania 8 Poland 8 Croatia 8 BosniaHerzegovina 8 Slovakia 6 Slovenia 4 Czech Republic 4 Source: Official rates - UNICEF (2003), Statistical Annex Table 3.1; survey data - see Appendix I. Serbia and Montenegro (MICS 2000), did not collect information on child and infant mortality, suggesting perhaps that official data were seen as adequate in these countries. 5 There are several exogenous reasons why we should be wary of official infant mortality rates in many countries in the region (see UNICEF, 2003). In this paper, we set aside these exogenous factors, and concentrate on the differences between official and survey data, and inconsistencies in the official data themselves. The method is as follows: 1. In nine countries, independent surveys suggest that official infant mortality rates may understate the true situation. 2. Evidence within the official data, for example unexpected patterns, may support our claim that official rates in these nine countries appear low. 3. If we find similar evidence in official infant mortality data for other countries in the CEE/CIS region for which we do not have survey data, then we may be able to suggest that the level of infant mortality in these countries is higher than the official data suggest. In undertaking this analysis, it is worth defining explicitly what we wish to capture. The true infant mortality rate m can be conceptualised as follows: •1000 + + r u r u N N M M m = where Nr and Nu are registered and unregistered births, respectively, and Mr and Mu are registered and unregistered deaths. Therefore, the official infant mortality rate is calculated from Mr and Nr . The underlying assumption with official administrative data is that Nr and Mr represent 100 per cent of all births and deaths, respectively. If Mr and Nr do not capture all births and deaths, then the extent to which the official infant mortality rate is a good proxy for m can be called into question. It is a question which underlies the discussion in Sections 2, 3 and 4 below. The degree to which m can be captured from survey data is analogous. If Nr and Mr are defined as births and deaths recorded in the surveys, then survey based estimates of infant mortality will be representative of m only to the extent that sampling and non-sampling error allow. These issues are discussed in Section 5. 2. LIVE BIRTHS, STILLBIRTHS AND MISCARRIAGES The first reason why estimates of infant mortality from official sources, and those calculated from surveys, may differ is related to definitions. If an infant is not considered to be born alive, then he or she cannot be considered to have died. Thus, the definition of ‘live birth’ is one crucial determinant of the infant mortality rate. The World Health Organization developed the following definition of a ‘live birth’ in 1950: 6 “The complete expulsion or extraction from its mother of a product of conception, irrespective of the duration of the pregnancy, which, after such separation, breathes or shows any other evidence of life, such as beating of heart, pulsation of the umbilical cord, or definite movement of voluntary muscles, whether or not the umbilical cord has been cut or the placenta is attached.” (WHO, 1992, definition 3.1) In other words, any birth with any signs of life, however insignificant, and which is then followed by death, however quickly, should be defined as a live birth and infant death. This definition is officially recognised in most countries, both developing and developed. Partly, perhaps, as a result of Cold War differences, the Soviet Union maintained an alternative and less rigorous definition of live birth. This was also adopted in some communist countries of Central and Eastern Europe. The differences between the WHO and Soviet definitions are highlighted on Table 2. The Soviet definition only counts breathing as a sign of life, and presumes infants who are born before the end of 28 weeks of gestation, or who weigh less than 1,000 grams at birth (there is considerable overlap between these two groups) to be non-viable – they are not counted as live births until they have survived a full seven days (or 168 hours). If they survive for less than this time, they are considered as miscarriages, and not counted at all. Table 2: Soviet and WHO definitions of live birth Infant born after the end of the 28th week of pregnancy No signs of life No breath but other signs of life Died during the first 7 days Survived the first 7 days USSR Stillbirth Live birth WHO Stillbirth Live birth Infant born before the end of the 28th week of pregnancy, or with weight under 1,000 gr. or length under 35 cm No sign of life No breath but other signs of life Died during the first 7 days Survived the first 7 days USSR Miscarriage Live birth WHO Stillbirth Live birth Source: Anderson and Silver (1986). 7 In some Central and East European