Docstoc

Low birth weight and birth asphyxia

Document Sample
Low birth weight and birth asphyxia Powered By Docstoc
					                                    Chapter 7

            Low birth weight and birth asphyxia


7.1 Introduction
The previous chapter, Chapter 6, estimated the relative importance of congenital anomalies in
the EME region, at both the individual level and the level of the population. The present
chapter does the same for the other selected adverse birth and pregnancy outcomes, i.e. low
birth weight, preterm birth, intrauterine growth retardation/small-for-gestational-age
(IUGR/SGA), and birth asphyxia (see also Sections 3.4.2 and 3.4.3).
        The relative importance of the selected risk factors at the individual level is assessed
on the basis of incidence of loss/death and the relative risk of loss/death. The relative
importance at the population level is based on prevalence of the risk factor in the population,
the attributable risk among the exposed, and the etiologic fraction (see Chapter 4). The
relative risk (RR) indicates how many times more likely persons with the risk factor will
experience the outcome (i.e. loss or death) as compared to persons without the risk factor. The
attributable risk among the exposed, AR(E), indicates the proportion of cases of the outcome
(i.e. loss or death) in the exposed population which is attributable to, or due to, the exposure
or risk factor of interest. Lastly, the etiologic fraction (EF) is the proportion of all cases of the
outcome in the total population that can be attributed to exposure to the risk factor.
        The values of all these measures come from secondary sources or are estimated by
combining assumptions derived from these sources with the hypothetical cohort. For a more
detailed discussion of the methods and equations used, please refer to Section 4.2.4. Overall,
it is important to note that the results represent educated guesses and not absolute, ‘true’
figures. Indeed, they frequently consist of ranges of estimates.
        The sections in the present chapter deal with low birth weight (Section 7.2), preterm
birth (Section 7.3), IUGR/SGA (Section 7.4), and birth asphyxia (Section 7.5). Each of these
sections presents and discusses birth data (i.e. about stillbirth, live birth and all births) and
neonatal data (i.e. about neonatal death). In addition, Section 7.4 on IUGR/SGA presents
antepartum data (i.e. about induced abortion and selective abortion, spontaneous loss, and
prevalence in utero). Section 7.6 gives attention to associations between the adverse
pregnancy and birth outcomes, and the combination of multiple outcomes or risk factors in
one person. Finally, the results of the chapter are summarised and discussed in Section 7.7.
Throughout the chapter, ‘SA’ refers to spontaneous abortion, ‘SB’ to stillbirth, ‘LB’ to live
birth, and ‘B’ to all births (live and stillbirths combined). All antepartum figures refer to, or
are assumed to refer to, the situation in which induced abortion (IA) is absent.
EARLY LIFE CHANGES

7.2 Low birth weight

7.2.1 BIRTH DATA

Stillbirth
Table 7.1 presents prevalence figures for low birth weight in stillbirths, for several weight
categories. The figures are population-based. According to the table, low birth weight below
2,500 g (LBW) is observed in about 58 to 76% of stillbirths. The prevalences of birth weight
< 2,000 g and < 1,500 g appear to range around 48-57% and 39-62% respectively. The lowest
weight categories of < 1,000 g and < 500 g occur in 13-53% and 10-30% of stillbirths
respectively.
         The results show a wide range. Part of this large variation can be explained by
differences in the definition of stillbirth, including lower limits for gestational age and birth
weight. According to WHO guidelines for national statistics, births below 500 g should be
excluded (WHO 1993). Nevertheless, births (both live and stillbirths) with weights less than
500 g are increasingly reported, such as in Alabama, USA (Phelan et al. 1998). This is
probably related to a decreasing gestational age of viability. The choice of either including or
excluding of cases < 500 g is likely to affect prevalence figures and mortality rates. As an
example, Table 7.1 includes prevalence figures from Cartlidge and Stewart (1995) based on
different criteria. For the weight category < 1,000 g, exclusion of those < 500 g yields a
prevalence of only 12.5% while inclusion results in a prevalence of 32.1%. Similar, though
less extreme, differences are observed for the other weight categories. Apart from the
differences in definitions, geographic location also seems to be related to low weight
prevalence. In Table 7.1, the highest prevalence proportions come from studies in the USA.
         The Consultive Council on Obstetric and Paediatric Mortality and Morbidity
(CCOPMM 2000) in Victoria, Australia, divided birth weight data on stillbirths according to
the timing of death, i.e. either before the onset of labour or during labour. Table 7.2 presents
birth weight distribution of the stillbirths by time of foetal death. For comparison purposes,
the birth weight distribution of all births (live and stillbirths) with known birth weight in
Victoria in 1998 has been included in the table. In total, 65.7% of stillbirths weighed below
2,500 g compared to only 6.5% of all births. Of those stillbirths in which death occurred
before labour, 68.7% were LBW (< 2,500 g) while of intrapartum stillbirths 52.0% had a birth
weight of less than 2,500 g.
         The incidence of stillbirth in LBW cases is an inappropriate term: the proportion of
stillbirths in LBW cases is a measure of prevalence rather than incidence. Intrinsic to its
definition, birth weight is not to be assessed before birth and therefore, the term ‘LBW cases’
refers to LBW newborns (also see Chapter 4). Only if the risk set or the population at risk of
experiencing stillbirth, i.e. foetuses in utero, were the denominator, would the measure be an
incidence. Therefore, logically, it is also not possible to calculate measures based on
incidence comparisons such as the RR, AR(E) and EF of stillbirth in LBW cases.




196
Table 7.1: Prevalence of low birth weight in stillbirths, various weight categories

Weight (g) Source                               Country                Period   Population                                %     Remarks
 < 500 g Goldenberg et al. 1987                 USA (Alabama)          1983     stillbirths >= 20 wks                     23 incl. some SB after IA (< 5%)
         Cartlidge and Stewart 1995             UK (Wales)             1993     221 stillbirths >= 24 wks                9.5 -
         Phelan et al. 1998                     USA (Alabama)          1994     585 stillbirths                          29.9 -
< 1,000 g Goldenberg et al. 1987                USA (Alabama) 1974-1983 stillbirths >= 20 wks                             40    incl. some SB after IA (< 5%)
          Tzoumaka-Bakoula et al. 1990          Greece               1983 115 stillbirths >= 500 g                       28.7   singletons
          Cartlidge and Stewart 1995            UK (Wales)           1993 221 stillbirths >= 24 wks                      32.1   -
          Cartlidge and Stewart 1995            UK (Wales)           1993 200 stillbirths >= 24 wks and >= 500 g         12.5   -
          Phelan et al. 1998                    USA (Alabama)        1994 585 stillbirths                                53.2   -
          CCOPMM 2000                           Australia (Victoria) 1998 290 stillbirths >= 500 g or >= 22 wks          27.2   all stillbirths

< 1,500 g Tzoumaka-Bakoula et al. 1990          Greece                 1983     115 stillbirths >= 500 g                 48.7   singletons
          Cartlidge and Stewart 1995            UK (Wales)             1993     221 stillbirths >= 24 wks                44.3   -
          Cartlidge and Stewart 1995            UK (Wales)             1993     200 stillbirths >= 24 wks and >= 500 g   38.5   -
          Phelan et al. 1998                    USA (Alabama)          1994     585 stillbirths                          62.3   -
          CCOPMM 2000                           Australia (Victoria)   1998     290 stillbirths >= 500 g or >= 22 wks    42.8   all stillbirths

< 2,000 g Tzoumaka-Bakoula et al. 1990 Greece                          1983     115 stillbirths >= 500 g               56.5 singletons
          Cartlidge and Stewart 1995   UK (Wales)                      1993     200 stillbirths >= 24 wks and >= 500 g 48.0 -
          CCOPMM 2000                  Australia (Victoria)            1998     290 stillbirths >= 500 g or >= 22 wks 54.8 all stillbirths

< 2,500 g Goldenberg et al. 1987                USA (Alabama) 1974-1983 stillbirths >= 20 wks                             69    incl. some SB after IA (< 5%)
          Goldenberg et al. 1987                USA (Alabama)         1983 stillbirths >= 20 wks                          76    incl. some SB after IA (< 5%)
          Tzoumaka-Bakoula et al. 1990          Greece                1983 115 stillbirths >= 500 g                      66.1   singletons
          Sipilä et al. 1994                    Finland (North)     1985-1986 44 stillbirths >= 500 g or >= 28 wks       68.2   singletons; based on table I, p. 614
          Cartlidge and Stewart 1995            UK (Wales)            1993 221 stillbirths >= 24 wks                     61.5   -
          Cartlidge and Stewart 1995            UK (Wales)            1993 200 stillbirths >= 24 wks and >= 500 g        57.5   -
          Waldhoer et al. 1996                  Austria             1984-1993 "stillbirths", not specified                62    <= 2,500 g
          CCOPMM 2000                           Australia (Victoria) 1998 290 stillbirths >= 500 g or >= 22 wks          65.5   all stillbirths
Notes: SB - stillbirth; IA - induced abortion
EARLY LIFE CHANGES


            Table 7.2: Birth weight distribution of births and stillbirths, by time of
            foetal death, Victoria, Australia, 1998

                           Stillbirths                                            Births
              Birth weight During*            Before** Unknown            Total     All
                   (g)         (%)              (%)      (%)               (%)     (%)
                500-999             40.0          23.8             41.7    27.3      0.5
              1,000-1,499            6.0          17.6             16.7    15.6      0.7
              1,500-1,999            2.0          14.5              8.3    12.1      1.3
              2,000-2,499            4.0          12.8              0.0    10.7      4.0
              2,500-2,999           18.0          14.1              0.0    14.2     15.4
              3,000-3,499           14.0          11.5             16.7    12.1     35.7
              3,500-3,999            8.0           3.1              8.3     4.2     30.5
              >= 4,000 g             8.0           2.6              8.3     3.8     11.9

               Total (%)          100.0          100.0         100.0      100.0    100.0
                  N                 50            227            12        289    61,914
            Notes: *During labour; **Before the onset of labour.
            Based on CCOPMM 2000, table 15, p. 23.


       Nevertheless, the prevalence proportions of stillbirth in LBW cases do provide useful
information. In northern Finland, the prevalence of stillbirth among singleton LBW births (<
2,500 g) was 10.5% during the period July 1985 through June 1986 (Sipilä et al. 1994). Table
7.3 presents further prevalence proportions by birth weight category. Unsurprisingly,
prevalence increases with decreasing birth weight. Of births below 500 g, 56-71% are born
dead compared to only 2 to 3% of births weighing 2,000-2,499 g. With regard to even greater
weights, Wolf (1991) found that only 0.9% of births weighing 2,500-2,999 g were stillbirths
and only 0.3% of births weighing ≥ 3,000 g. Similarly, Goldenberg et al. (1987) found
frequencies of 0.3% in births of 2,500-3,999 g and 0.2% in births ≥ 4,000 g. In Finland, in
1994, only 1.3 out of 1,000 births weighing ≥ 2,500 g was a stillbirth (Forssas et al. 1998).
       However, the frequency of stillbirth increases in the highest birth weight categories. In
Austria, Waldhoer et al. (1996) concluded that among births during 1984-1993, the stillbirth
rate was lowest in the birth category of 3,300-3,899 g. Wilcox and Russell (1983b) have
modelled the pattern of weight-specific perinatal mortality, and in general, perinatal mortality
is very high at the lowest birth weights, drops to a minimum within the range of the most
frequent birth weights, but rises again for the heaviest birth weights. A similar pattern was
observed among foetal deaths (stillbirths), using data for 1968 from the USA (Wilcox and
Russell 1983b).

Live birth and all births
According to literature, the prevalence of LBW (≤ 2,500 g) ranges from 3 to 10% of births
(see Table 7.4) and from 3 to 15% of live births (see Table 7.5). In addition to the weight limit
of 2,500 g, Tables 7.4 and 7.5 also present prevalence data on other low birth weight



198
Table 7.3: Prevalence of stillbirth in low birth weight births, by weight category

     Weight          Source                           Country                       Period   Definition SB           % SB Remarks
< 500 g              Goldenberg et al. 1987           USA (Alabama)                  1983    >= 20 wks               71.1   incl. some SB after IA
                     Phelan et al. 1998               USA (Alabama)                  1994    does not include BW      56    -

500-749 g            Wolf 1991                        NL (Amsterdam)             1983-1989 >= 500 g and >= 24 wks     43    -

500-999 g            Goldenberg et al. 1987           USA (Alabama)                  1983    >= 20 wks               28.5   incl. some SB after IA
750-999 g            Wolf 1991                        NL (Amsterdam)             1983-1989 >= 500 g and >= 24 wks     13    -
< 1,000 g            Forssas et al. 1998              Finland                        1994    >= 500 g or >= 22 wks   30.9   -

1,000-1,249 g        Wolf 1991                        NL (Amsterdam)             1983-1989 >= 500 g and >= 24 wks     7     -
1,000-1,499 g        Goldenberg et al. 1987           USA (Alabama)                  1983    >= 20 wks               13.5   incl. some SB after IA
                     Forssas et al. 1998              Finland                        1994    >= 500 g or >= 22 wks    9.6   -
                     Phelan et al. 1998               USA (Alabama)                  1994    does not include BW       9    -
1,250-1,499 g        Wolf 1991                        NL (Amsterdam)             1983-1989 >= 500 g and >= 24 wks     7     -
1,500-1,999 g        Goldenberg et al. 1987           USA (Alabama)                1983    >= 20 wks                 6.70   incl. some SB after IA
                     Wolf 1991                        NL (Amsterdam)             1983-1989 >= 500 g and >= 24 wks      4    -
                     Forssas et al. 1998              Finland                      1994    >= 500 g or >= 22 wks      4.6   -

2,000-2,499 g        Goldenberg et al. 1987           USA (Alabama)                1983    >= 20 wks                 2.04   incl. some SB after IA
                     Wolf 1991                        NL (Amsterdam)             1983-1989 >= 500 g and >= 24 wks      3    -
                     Forssas et al. 1998              Finland                      1994    >= 500 g or >= 22 wks      1.8   -

< 2,500 g            Sipilä et al. 1994               Finland (north)            1985-1986 >= 500 g or >= 28 wks     10.5   -
Notes: SB - stillbirth; IA - induced abortion; NL - the Netherlands; BW - birth weight.
Data by Wolf (1991) are hospital-based.
Table 7.4: Prevalence of low birth weight in births, various weight categories
 Weight    Source                            Region                   Period    Population                                           %     Remarks
 < 500 g Cartlidge and Stewart 1995          UK (Wales)               1993      stillbirths (>= 24 wks) + live births               0.07      -
          Phelan et al. 1998                 USA (Alabama)            1994      live + stillbirths                                  0.51      -
          Riley and Halliday 1999            Australia (Victoria)     1998      births >= 20 wks or >= 400 g                         0.3      -
< 1,000 g Tzoumaka-Bakoula et al. 1990       Greece                   1983      live + stillbirths >= 500 g                         0.5       -
          Cartlidge and Stewart 1995         UK (Wales)               1993      stillbirths (>= 24 wks) + live births                0.5      -
          Forssas et al. 1998                Finland                1983-1994   stillbirths (>= 500 g or >= 22 wks) + live births   0.44      -
          Phelan et al. 1998                 USA (Alabama)            1994      live + stillbirths                                  1.41      -
          Riley and Halliday 1999            Australia (Victoria)     1998      births >= 20 wks or >= 400 g                         0.8      -
< 1,500 g Tzoumaka-Bakoula et al. 1990       Greece                   1983      live + stillbirths >= 500 g                         1.2       -
          Working Group VLBWI 1990           Denmark (Odense)          NS       births >= 500 g                                     1.13      -
          Working Group VLBWI 1990           Italy                     NS       births >= 500 g                                     1.10      -
          Working Group VLBWI 1990           UK (Grampian)             NS       births >= 500 g                                     1.38      -
          Working Group VLBWI 1990           UK (Oxford)               NS       births >= 500 g                                     1.01      -
          Cartlidge and Stewart 1995         UK (Wales)               1993      stillbirths (>= 24 wks) + live births                1.2      -
          Forssas et al. 1998                Finland                  1994      stillbirths (>= 500 g or >= 22 wks) + live births   0.84      -
          Phelan et al. 1998                 USA (Alabama)            1994      live + stillbirths                                  2.42      -
          Riley and Halliday 1999            Australia (Victoria)     1998      births >= 20 wks or >= 400 g                         1.5      -
< 2,000 g Tzoumaka-Bakoula et al. 1990       Greece                   1983      live + stillbirths >= 500 g                           2       -
          Cartlidge and Stewart 1995         UK (Wales)               1993      stillbirths (>= 24 wks) + live births                2.5      -
          Riley and Halliday 1999            Australia (Victoria)     1998      births >= 20 wks or >= 400 g                         2.8      -
< 2,500 g Newton 1989                        -                          -       "general population"                                  8       I
          Tzoumaka-Bakoula et al. 1990       Greece                   1983      live + stillbirths >= 500 g                          4.5      -
          Sipilä et al. 1994                 Finland (north)        1985-1986   live births; stillbirths >= 500 g or >= 28 wks      3.1      II
          Cartlidge and Stewart 1995         UK (Wales)               1993      stillbirths (>= 24 wks) + live births                6.5      -
          De Onis et al. 1998                developed                  -       NS                                                   6.2      I
          Forssas et al. 1998                Finland                  1994      stillbirths (>= 500 g or >= 22 wks) + live births    4.1      -
          Kalter et al. 1998                 USA (NY City)          1990-1991   "births", incl. < 500 g                              9.5      -
          Riley and Halliday 1999            Australia (Victoria)     1998      births >= 20 wks or >= 400 g                         6.8      -
          UNICEF 2000                        developed              1990-1997   "infants"                                           4-7      III
Notes: Working Group VLBWI - Working Group on the Very Low Birthweight Infant; NS - not stated.
Remarks: I - general estimate; II - singletons; III - estimates for various countries.
Table 7.5: Prevalence of low birth weight in live births, various weight categories

 Weight Source                              Region               Period      %        Remarks
 < 500 g Cartlidge and Stewart 1995         UK (Wales)           1993       0.01      -
         NCHS 1995                          USA (California)     1988       0.08      singletons; known birth weight
         NCHS 1995                          USA                  1988       0.13      known birth weight

< 1,000 g Cartlidge and Stewart 1995        UK (Wales)           1993        0.3      -
          NCHS 1995                         USA (California)     1988        0.4      singletons; known birth weight
          NCHS 1995                         USA                  1988        0.6      known birth weight
< 1,500 g Verloove-Vanhorick et al. 1988    the Netherlands      1983       0.68      -
          Cartlidge and Stewart 1995        UK (Wales)           1993       0.9       -
          NCHS 1995                         USA (California)     1988       0.9       singletons; known birth weight
          NCHS 1995                         USA                  1988       1.2       known birth weight

< 2,000 g Cartlidge and Stewart 1995        UK (Wales)           1993        2.2      -
          NCHS 1995                         USA (California)     1988        1.8      singletons; known birth weight
          NCHS 1995                         USA                  1988        2.6      documentation table 3; known birth weight
< 2,500 g Goldenberg et al. 1987            USA (Alabama) 1974-1983       7           -
          Kline et al. 1989                 -                    -      5-15          estimated crude frequency in unexposed population
          Buekens et al. 1995               Belgium          1986-1987   4.9          -
          Buekens et al. 1995               USA              1986-1987   5.9          -
          Cartlidge and Stewart 1995        UK (Wales)         1993      6.2          -
          NCHS 1995                         USA (California)   1988      5.1          singletons; known birth weight
          NCHS 1995                         USA                1988      6.9          documentation table 3; known birth weight
          Shibuya and Murray 1998a          EME                1990    3.3-7.0        estimates for various countries
EARLY LIFE CHANGES

categories. Unsurprisingly, the proportion decreases steadily for the reduced weight
categories. Ultimately, the weight category of < 500 g constitutes only a small proportion of
all births and live births: less than 0.6%. Again, part of the variation in the table can probably
be explained by differences in definitions.

7.2.2 NEONATAL DATA

Neonatal death
The prevalence of low birth weight among neonatal deaths is high. According to the data in
Table 7.6, about 54 to 75% of neonatal deaths have a birth weight below 2,500 g. The table
also presents prevalences for the birth weight categories of < 2,000 g, < 1,500 g, < 1,000, and
< 500 g. Over these groups, prevalence among neonatal deaths declines steadily with


      Table 7.6: Prevalence of low birth weight in neonatal deaths, various weight categories

        Weight     Source                           Country                     Period          %     Remarks
       < 500 g Cartlidge and Stewart 1995           UK (Wales)                  1993            2.3      -
               NCHS 1995                            USA (California)            1988           16.1      I
               Phelan et al. 1998                   USA (Alabama)               1974            3.0      -
               Phelan et al. 1998                   USA (Alabama)               1994           32.4      -

      < 1,000 g Goldenberg et al. 1983              USA (Alabama)        1978-1980             37.0     II
                Cartlidge and Stewart 1995          UK (Wales)             1993                36.7      -
                NCHS 1995                           USA (California)       1988                48.5      I
                Phelan et al. 1998                  USA (Alabama)          1974                29.2      -
                Phelan et al. 1998                  USA (Alabama)          1994                65.3      -
                CCOPMM 2000                         Australia (Victoria)   1998                36.6     III
      < 1,500 g Goldenberg et al. 1983              USA (Alabama)        1978-1980             46.4     II
                Cartlidge and Stewart 1995          UK (Wales)             1993                49.2      -
                NCHS 1995                           USA (California)       1988                58.3      I
                Phelan et al. 1998                  USA (Alabama)          1974                51.4      -
                Phelan et al. 1998                  USA (Alabama)          1994                71.1      -
                CCOPMM 2000                         Australia (Victoria)   1998                49.4     III
      < 2,000 g Goldenberg et al. 1983              USA (Alabama)        1978-1980             56.6     II
                Cartlidge and Stewart 1995          UK (Wales)             1993                56.3      -
                NCHS 1995                           USA (California)       1988                66.5      I
                CCOPMM 2000                         Australia (Victoria)   1998                57.3     III

      < 2,500 g Goldenberg et al. 1983              USA (Alabama)        1978-1980             65.4     II
                Sipilä et al. 1994                  Finland (north)      1985-1986             54.3      -
                Cartlidge and Stewart 1995          UK (Wales)             1993                65.6      -
                NCHS 1995                           USA (California)       1988                74.7      I
                CCOPMM 2000                         Australia (Victoria)   1998                65.2     III
      Remarks: I - singletons; II - excl. < 500 g; III - known birth weight & excl. < 500 g.



202
                                                 CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA
  Table 7.7: Prevalence of low birth weight in early neonatal deaths, various weight categories

    Weight     Source                                Country                    Period      %     Remarks
   < 500 g Cartlidge and Stewart 1995                UK (Wales)                  1993       3.3      -
           NCHS 1995                                 USA (California)            1988      20.0      I
  < 1,000 g Tzoumaka-Bakoula et al. 1990             Greece                      1983      20.3     II
            Cartlidge and Stewart 1995               UK (Wales)                  1993      38.0      -
            NCHS 1995                                USA (California)            1988      54.9      I
            CCOPMM 2000                              Australia (Victoria)        1998      40.9     III

  < 1,500 g Tzoumaka-Bakoula et al. 1990             Greece                      1983      40.6     II
            Cartlidge and Stewart 1995               UK (Wales)                  1993      46.7      -
            NCHS 1995                                USA (California)            1988      64.0      I
            CCOPMM 2000                              Australia (Victoria)        1998      51.3     III
  < 2,000 g Tzoumaka-Bakoula et al. 1990             Greece                      1983      49.9     II
            Cartlidge and Stewart 1995               UK (Wales)                  1993      53.3      -
            NCHS 1995                                USA (California)            1988      72.5      I
            CCOPMM 2000                              Australia (Victoria)        1998      59.1     III

  < 2,500 g Tzoumaka-Bakoula et al. 1990             Greece                      1983      59.2     II
            Cartlidge and Stewart 1995               UK (Wales)                  1993      63.0      -
            NCHS 1995                                USA (California)            1988      79.6      I
            CCOPMM 2000                              Australia (Victoria)        1998      67.0     III
  Remarks: I - singletons; II - excl. < 500 g; III - known birth weight & excl. < 500 g.



declining birth weight: 56-67% weigh < 2,000 g, 46-71% < 1,500 g, 29-65% < 1,000 g, and
finally, 2-32% weigh less than 500 g.
        In Table 7.6, many of the findings (based on various registration systems) agree
reasonably well. The exceptions are the more recent figures for the USA from NCHS (1995)
and Phelan et al. (1998) that include large proportions of births below 500 g. As noted earlier,
WHO (1993) recommends the exclusion from national statistics of foetuses and infants
(whether alive or dead) with birth weights below 500 g. However, survival chances have
improved over the years and weights below 500 g are increasingly reported. For example,
nursery survival of infants admitted to the NICU of the University of Washington (USA) with
birth weights less than 800 g increased from 20% between 1977 and 1980 to 49% between
1986 and 1990 (La Pine et al. 1995). In Alabama (USA), only 0.20% of births in 1974
weighed less than 500 g compared to 0.51% in 1994 (Phelan et al. 1998). Among neonatal
deaths, 3.0% weighed below 500 g in 1974 compared to as many as 32.4% in 1994 (see Table
7.6). Table 7.6 suggests that increased reporting of the lowest birth weight categories mainly
takes place in the USA. Indeed, figures from Cartlidge and Stewart (1995) for Wales that also
include births less than 500 g are much lower compared to those from NCHS (1995) and
Phelan et al. (1998). This may be related to policies and procedures and to decisions by
individual doctors whether or not to report the baby as a birth.



                                                                                                          203
EARLY LIFE CHANGES


  Table 7.8: Incidence of neonatal death in low birth-weight neonates, by weight category

      Weight       Source                             Country                  Period      % NND Remarks
  < 500 g          Cartlidge and Stewart 1995 UK (Wales)                        1993         75.0          -
                   NCHS 1995                  USA (California)                  1988         88.4          I
  500-599 g        Phelps et al. 1991                 USA                    1986-1987       70.0         II
  501-750 g        The Investigators 1993 [a]         NS                         NS           43          III
                   NCHS 1995 [b]                      USA                       1988          64           -
  500-999 g        Goldenberg et al. 1983             USA (Alabama)          1978-1980       59.5          -
                   Kalter et al. 1998                 USA (NY City)          1990-1991       42.15         -
  751-1,000 g      The Investigators 1993 [a]         NS                         NS           18          III
                   NCHS 1995 [b]                      USA                       1988          29           -
  800-899 g        Phelps et al. 1991                 USA                    1986-1987       23.0         II
  < 1,000 g        Goldenberg et al. 1983             USA (Alabama) 1978-1980                59.5         IV
                   Nishida 1993                       Japan            1989                   25           -
                   Cartlidge and Stewart 1995         UK (Wales)       1993                  43.1          -
                   NCHS 1995                          USA (California) 1988                  51.5          I
  1,001-1,250 g The Investigators 1993 [a]            NS                         NS           7           III
                NCHS 1995 [b]                         USA                       1988          11           -
  1,000-1,499 g Goldenberg et al. 1983                USA (Alabama)          1978-1980       10.1          -
  1,200-1,250 g Phelps et al. 1991                    USA                    1986-1987        8.7         II
  1,251-1,500 g The Investigators 1993 [a]            NS                         NS            4          III
                NCHS 1995 [b]                         USA                       1988           7           -
  < 1,500 g        Goldenberg et al. 1983             USA (Alabama) 1978-1980                29.8         IV
                   Cartlidge and Stewart 1995         UK (Wales)       1993                  19.2          -
                   NCHS 1995                          USA (California) 1988                  28.8          I
  1,500-1,999      Goldenberg et al. 1983             USA (Alabama) 1978-1980                 4.8          -
  < 2,000          Goldenberg et al. 1983     USA (Alabama)                  1978-1980       15.5         IV
                   Cartlidge and Stewart 1995 UK (Wales)                       1993           9.0          -
                   NCHS 1995                  USA (California)                 1988          16.0          I
  2,000-2,499      Goldenberg et al. 1983             USA (Alabama)          1978-1980        1.4          -
  < 2,500 g        Goldenberg et al. 1983             USA (Alabama)          1978-1980        6.4         IV
                   Cartlidge and Stewart 1995         UK (Wales)               1993           3.7          -
                   NCHS 1995                          USA (California)         1988           6.2          I
                   Kalter et al. 1998                 USA (NY City)          1990-1991       6.44         V
                   Kalter et al. 1998                 USA (NY City)          1990-1991       6.76         VI
  Notes: NS - not stated; NND - neonatal death.
  [a] Investigators of the Vermont-Oxford Trials Network Databased Project cited by Van der Veen 2001;
  [b] calculated by Van der Veen 2001 from NCHS data set
  Remarks: I - singletons; II - inborns of 23 hospitals; III - hospital-based, incl. outborns; IV - excl. < 500 g;
  V - Caucasian population; VI - Afro-American population.




204
                                             CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

        In early neonatal deaths, the proportions with low birth weights are similar to those
among all neonatal deaths although, perhaps, slightly higher. According to the figures in
Table 7.7, 59 to 80% of early neonatal deaths weigh less than 2,500 g at birth. Of course, the
figure declines with decreased weight limit: 50-73% weigh < 2,000 g, 41-64% < 1,500 g, 20-
55% < 1,000 g, and 3-20% weigh less than 500 g. The pattern of weight-specific early
neonatal mortality is comparable to the patterns for foetal death and perinatal mortality in
general. Wilcox and Russell (1983b) presented this pattern for early neonatal deaths in the
USA in 1960. Mortality is very high at the lowest birth weights, declines to a minimum within
the range of the most common birth weights, and then rises again for the heaviest birth
weights.
        With regard to the incidences of neonatal death and early neonatal death in liveborn,
low birth weight infants, the survival probabilities have improved considerably over the years.
Tables 7.8 and 7.9 try to be mainly representative of the most recent data available, but also
show a decline in incidence since periods somewhat longer ago. In Table 7.8, the incidence
proportion of neonatal death is 4-7% in LBW births (< 2,500 g) and 9-16% in births below
2,000 g. In the lowest weight categories, risks are considerably higher. Among births
weighing less than 500 g, the vast majority (75-88%) end in neonatal death. In comparison,
figures on the incidence of neonatal death in live births ≥ 2,500 g are much lower: 0.30%
(Goldenberg et al. 1983), 0.13% (Cartlidge and Stewart 1995), 0.11% (NCHS 1995), 0.11%
(Caucasian population, Kalter et al. 1998), and 0.20% (Afro-American population, Kalter et
al. 1998). Table 7.9 presents incidence figures of early neonatal death among liveborn, low
birth weight infants. Of course, these are lower than the figures for all neonatal death, albeit
only slightly.
        In general, the birth weight of male newborns is heavier than that of females. Despite
this, females have an advantage over males in terms of perinatal survival. According to Kline
et al. (1989), this advantage holds at all gestational ages although some of the female

Table 7.9: Incidence of early neonatal death in low birth-weight neonates, by weight category

   Weight       Source                         Country            Period   % ENND Remarks
   < 500 g      Cartlidge and Stewart 1995     UK (Wales)          1993      75.0 -
                NCHS 1995                      USA (California)    1988      86.9 singletons

  < 1,000 g     Cartlidge and Stewart 1995     UK (Wales)         1993       32.1   -
                NCHS 1995                      USA (California)   1988       46.2   singletons

  < 1,500 g     Cartlidge and Stewart 1995     UK (Wales)         1993       13.1   -
                NCHS 1995                      USA (California)   1988       25.1   singletons

   < 2,000      Cartlidge and Stewart 1995     UK (Wales)         1993       6.1    -
                NCHS 1995                      USA (California)   1988       13.9   singletons

  < 2,500 g     Cartlidge and Stewart 1995     UK (Wales)         1993       2.6    -
                NCHS 1995                      USA (California)   1988       5.3    singletons
Notes: ENND - early neonatal death




                                                                                            205
EARLY LIFE CHANGES

advantage “can be attributed to the fact that, given the same birth weight, the average girl
must be older and more mature than the average boy” (p. 222). Phelps et al. (1991), who
studied neonates born in 1986-1987 in the USA weighing 1,250 g or less at birth, also found a
female advantage in sex-specific neonatal survival by birth weight. The difference between
the sexes diminished when the outcome was examined by gestational age though a small but
significant residual difference remained. Lastly, Afro-American infants have been found to
have a survival advantage over Caucasian infants of the same weight (Van der Veen 2001).

Relative risk and attributable risk: neonatal death
Table 7.10 presents estimates of the relative risk based on data from the studies that were
discussed previously. In the process of restructuring a continuous variable, such as weight,
into a dichotomous variable, information is lost. The RR is affected by the weight distribution
within the population. In particular, the distribution within the low weight population, with its
huge internal variations in incidence of death, is likely to affect the results. Moreover, the
weight limit of 2,500 g to distinguish between ‘high-risk’ and ‘low-risk’ or ‘unhealthy’ and
‘healthy’ has been questioned, in particular as applied to developing countries (Shibuya and
Murray 1998a). Therefore, Table 7.10 also compares risks on the basis of 1,500 g and 2,000 g
as defining limits. Infants weighing 2,000-2,499 g or 1,500-2,499 g are the most common
within low weight populations, but their risks of mortality are small compared to neonates
with lower birth weights.
        In Table 7.10, the RR of neonatal death for newborns below 1,500 g, in comparison to
those ≥ 1,500 g, ranges from 69 to 160. Using weight limits of 2,000 g and 2,500 g, the RR
varies around 44-107 and 21-59 respectively. The lowest figures are from data discussed by
Goldenberg et al. (1983) and can be explained by the exclusion of weights below 500 g.
However, the differences between the results for Wales (UK) based on Cartlidge and Stewart
(1995), and the US data from NCHS (1995) and Kalter et al. (1998) are also considerable.
About two decades ago, Shapiro et al. (1980 cited by McCormick 1985) estimated that LBW
infants are around 40 times as likely to die during the neonatal period as normal birth-weight
infants. For infants with a very low birth weight (< 1,500 g), the relative risk of neonatal death
was believed to be as high as 200.
        Table 7.10 also presents attributable risks among the exposed (AR(E)) and etiologic
fractions (EF). Similar to the RR, the AR(E) declines as the weight limit is increased.
However, this is not the case for the EF, which is also affected by the prevalence of the low
birth weights within the total population. The EF increases with increasing weight limit: 45.5
to 57.8% in infants < 1,500 g, 54.8 to 65.2% in those < 2,000 g, and 61.8 to 84.5% in those <
2,500 g. In other words, about half or even more of neonatal deaths in EME populations can
be attributed to low birth weight. However, it is important to note that these figures do not
reflect LBW as the cause of death. Moreover, a relatively large proportion of LBW newborns
are preterm or are affected by congenital anomalies (see Section 7.6).
        With regard to early neonatal death, the results from Cartlidge and Stewart (1995) vary
widely from those based on NCHS (1995) (see Table 7.10). Moreover, on the basis of the
Welsh data from Cartlidge and Stewart (1995), the RR, AR(E) and EF of early neonatal death



206
Table 7.10: Prevalence of low birth weight in live births and incidence, RR, AR(E), and EFof neonatal death, various weight categories

                                                                                  Incidence Incidence Prevalence
Weight Source                                Country                  Period       in lower in higher  of lower
                                                                                   weights weights weights in LB     RR AR(E)       EF     Remarks
< 1,500 g vs. >= 1,500 g
NND Goldenberg et al. 1983                   USA (Alabama) 1978-1980                0.298     0.004    0.012        69.3   0.986 0.455 excl. < 500 g
        Cartlidge and Stewart 1995           UK (Wales)       1993                  0.192     0.002    0.009       106.7   0.991 0.487 -
        NCHS 1995                            USA (California) 1988                  0.288     0.002    0.009       160.0   0.994 0.578 singletons
ENND Cartlidge and Stewart 1995              UK (Wales)       1993                  0.131     0.001    0.009        93.6   0.989 0.454 -
        NCHS 1995                            USA (California) 1988                  0.251     0.001    0.009       209.2   0.995 0.642 singletons

< 2,000 g vs. >= 2,000 g
NND Goldenberg et al. 1983                   USA (Alabama) 1978-1980                0.155     0.004    0.029        44.3   0.977   0.554   excl. < 500 g
        Cartlidge and Stewart 1995           UK (Wales)       1993                  0.090     0.002    0.022        56.3   0.982   0.548   -
        NCHS 1995                            USA (California) 1988                  0.160     0.002    0.018       106.7   0.991   0.652   singletons
ENND Cartlidge and Stewart 1995              UK (Wales)       1993                  0.061     0.001    0.022        50.8   0.980   0.522   -
        NCHS 1995                            USA (California) 1988                  0.139     0.001    0.018       154.4   0.994   0.731   singletons
< 2,500 g vs. >= 2,500 g
NND Goldenberg et al. 1983                   USA (Alabama) 1978-1980                0.064     0.003    0.080        21.3   0.953 0.618 excl. < 500 g
        Cartlidge and Stewart 1995           UK (Wales)       1993                  0.037     0.001    0.062        28.5   0.965 0.630 -
        NCHS 1995                            USA (California) 1988                  0.062     0.001    0.051        56.4   0.982 0.738 singletons
        Kalter et al. 1998                   USA (NY City) 1990-1991                0.064     0.001    0.095        58.5   0.983 0.845 Caucasian*
        Kalter et al. 1998                   USA (NY City) 1990-1991                0.068     0.002    0.095        33.8   0.970 0.756 Afro-American*
ENND Cartlidge and Stewart 1995              UK (Wales)       1993                  0.026     0.001    0.062        26.0   0.962 0.608 -
        NCHS 1995                            USA (California) 1988                  0.053     0.001    0.051        75.7   0.987 0.792 singletons
Notes: LB - live birth; NND - neonatal death; ENND - early neonatal death.
*Prevalence figure in live births for both Caucasian and Afro-American population combined.
EARLY LIFE CHANGES

are higher compared to overall neonatal death whereas data from the USA (NCHS 1995)
indicate the opposite. This implies that, in the US data set, LBW infants who die during the
neonatal period are more prone to die within one week of birth than normal weight neonatal
deaths. The opposite holds true for the Welsh data.
         The hypothetical cohort that was constructed in Chapter 5 can also provide a basis for
further estimations of RR, AR(E) and EF. In the hypothetical cohort, 88,410 live births take
place of whom 299 die during the early neonatal period and a total of 383 during the neonatal
period as a whole. On the basis of the data discussed above, assumptions can be made for (1)
the prevalence of low birth weight in live births (pB1), (2) the incidence of neonatal death in
low-weight live births (I1), and (3) the prevalence of low birth weight in neonatal deaths (pD1).
Table 7.11 presents the results using these assumptions for three weight categories and the
numbers in the hypothetical cohort. We have assumed the following for birth weights below
1,500 g: (1) pB1 is 0.9%, (2) I1 is 25%, and (3) pD1 is 50%. For weights < 2,000 g and < 2,500
g, it is assumed that: (1) pB1 is 2 and 6%, (2) I1 is 13 and 5.5%, and (3) pD1 is 60 and 65%
respectively.
         The resultant RRs range from 110 to 119 for weights less than 1,500 g and from 29 to
50 for weights less than 2,500 g. For weights below 2,000 g, the RR is about 73.5. As with the
RR, the AR(E) decreases with increasing birth weight. However, due to the increase in
prevalence in live births (pB1), the EF increases as the weight limit increases. Overall, about
half of neonatal deaths in the entire population can be attributed to a birth weight below 1,500
g (the EF is 49.5-51.5%). For births < 2,000 g, this is about 59%, while 63 to 75% of neonatal
deaths can be attributed to birth weights below 2,500 g. The final column in Table 7.11
contains whichever recalculated measure was missing from the input data, i.e. pB1, I1, or pD1
(based on equations (4.22), (4.23), and (4.24)). This serves as a check as to whether the

Table 7.11: Estimated RR, AR(E), and EF of neonatal death in low weight live births,
hypothetical cohort, various weight categories

         Input                                                          Results
  Weight    B                 D         pB1          I1        pD1        RR         AR(E)         EF          check*
 < 1,500 g 88,410           383        0.009       0.250        -         119.0       0.992      0.515      pD1 = 0.519
           88,410           383        0.009         -        0.500       110.1       0.991      0.495        I1 = 0.241
           88,410           383          -         0.250      0.500       114.4       0.991      0.496      pB1 = 0.009
 < 2,000 g 88,410           383        0.020       0.130        -          73.6       0.986      0.592      pD1 = 0.600
           88,410           383        0.020         -        0.600        73.5       0.986      0.592        I1 = 0.130
           88,410           383          -         0.130      0.600        73.5       0.986      0.592      pB1 = 0.020

 < 2,500 g 88,410           383        0.060       0.055        -          50.1       0.980      0.747      pD1 = 0.762
           88,410           383        0.060         -        0.650        29.1       0.966      0.628        I1 = 0.047
           88,410           383          -         0.055      0.650        34.4       0.971      0.631      pB1 = 0.051
Notes: B - total no. of live births; D - total no. of neonatal deaths; pB1 - prevalence of low weight among live births;
I1 - incidence of neonatal deaths among low weight live births; pD1 - prevalence of low birth weight in neonatal deaths.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.




208
                                                     CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

specific combination of input data and its resultant estimates are plausible. In this case, the
values are more or less equal to the assumptions, which implies that the series of estimates of
RR, AR(E) and EF are credible. Only the figures for birth weights below 2,500 g show
deviations.
        Table 7.12 was constructed in the same manner as Table 7.11 but presents figures on
early neonatal death. The results are almost equal to those for all neonatal deaths. However,
the estimates for the weight category < 2,500 g seem more credible.

Relative risk and attributable risk, polytomous weight categories
In the literature, data on the relative risk of death has been presented for polytomous weight
categories. Ashworth (1998) quantified the risks of mortality and morbidity associated with
intrauterine growth retardation. In this study, the author includes relative risks of neonatal
death among infants of different birth weights, as reviewed by Ashworth and Feachem (1985).
Table 7.13 presents the results where birth weight of 2,500-2,999 g was taken as the reference
category. Clearly, risk is highest in the lowest birth weight groups while being lowest in the
category 3,500-4,500 g. On the basis of the 10 studies included in Table 7.13 plus three
additional studies – two data sets from India and one from Guatemala – Ashworth and
Feachem (1985 cited by Ashworth 1998) calculated median relative risks of neonatal death.
The results were as follows: an RR of 50.0 for birth weight category 1,000-1,499 g, an RR of
18.0 for 1,500-1,999 g, an RR of 4.0 for 2,000-2,499 g, an RR of 0.5 for 3,000-3,499 g, and
RRs of 0.4 for 3,500-3,999 g and for ≥ 4,000 g. Again, birth weight of 2,500-2,999 was taken
as the reference category. Since most of the studies that were included in the analysis are
somewhat dated, while neonatal care has improved over the years, it is likely that the relative
risks are nowadays reduced.

Table 7.12: Estimated RR, AR(E) and EF of early neonatal mortality in low weight live births,
hypothetical cohort, various weight categories

              Input                                                     Results
  Weight         B            D         pB1          I1        pD1        RR         AR(E)         EF          check*
 < 1,500 g 88,410           299        0.009       0.190        -         112.6       0.991      0.501      pD1 = 0.506
           88,410           299        0.009         -        0.500       110.1       0.991      0.495        I1 = 0.188
           88,410           299          -         0.190      0.500       111.4       0.991      0.496      pB1 = 0.009
 < 2,000 g 88,410           299        0.020       0.100        -          70.9       0.986      0.583      pD1 = 0.591
           88,410           299        0.020         -        0.600        73.5       0.986      0.592        I1 = 0.101
           88,410           299          -         0.100      0.600        72.4       0.986      0.592      pB1 = 0.020

< 2,500 g      88,410       299        0.060       0.040        -          38.3       0.974      0.691      pD1 = 0.710
               88,410       299        0.060         -        0.700        36.6       0.973      0.681        I1 = 0.039
               88,410       299          -         0.040      0.700        37.1       0.973      0.681      pB1 = 0.059
Notes: B - total no. of live births; D - total no. of neonatal deaths; pB1 - prevalence of low weight among live births;
I1 - incidence of neonatal deaths among low weight live births; pD1 - prevalence of low birth weight in neonatal deaths.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.




                                                                                                                       209
Table 7.13: Relative risks of neonatal death for polytomous weight categories with weight 2,500-2,999 g as reference group

                                  UK          Belgium       USA         USA               USA          USA             USA            Norway            USA             USA
Birth weight                     1993        1986-1987      1977      1986-1987           1980       1974-1980       1974-1975       1967-1968        1959-1966         1960
  1,000-1,499 g                 [I]   99.6       49.6          52.7          34.0            46.7          32.8        [I]   99.6          41.0             47.5          55.0
  1,500-1,999 g                       10.4       15.9          13.5          12.9            13.5           9.4              17.5          16.6             20.6          19.6
  2,000-2,499 g                        2.8        3.3           3.0           3.7             4.0           2.1               3.8           4.5              4.4           4.4
  2,500-2,999 g                        1.0        1.0           1.0           1.0             1.0           1.0               1.0           1.0              1.0           1.0
  3,000-3,499 g                        0.5        0.4           0.3           0.4             0.5           0.3               0.5           0.4              0.4           0.4
  3,500-3,999 g                        0.4        0.4           0.3           0.3             0.3           0.3               0.3           0.2              0.3           0.3
  4,000-4,499 g                        0.4        0.4           0.3           0.3             0.4           0.0               0.4           0.2              0.2           0.4
  >= 4,500 g                           0.4        0.4           0.4           0.9             0.9           0.0               0.4           0.4              0.3           0.8
Neonatal mortality rate*       4.1                 4.8         5.1       5.6         7.3     17.5                           8.2             8.5            13.8          16.9
% LBW                     [II] 6.5                 4.9     [III]
                                                               7.0       5.9         6.0     15.2                           6.1            5.2              5.2           6.8
Number of births         673,000              229,000     290,000 7.4 million 3.5 million 14,400                       234,000         700,000          40,000      3.6 million
                            all                single      single    single     single,     all                         single           all            single,        white
                                                                               all races                                                                 white
Reference                         [a]           [b]          [c]           [d]              [e]           [f]             [g]             [h]             [i]            [j]
Notes: *Neonatal mortality rate: deaths per 1,000 live births except for Williams and Chen 1982 (per 1,000 vaginal births) and Erickson and Bjerkedal 1982 (per 1,000 total births).
[I] includes all births < 1,500 g; [II] 1992 figure; [III] estimated from Erickson and Bjerkedal 1982.
References in the table: [a] Platt and Pharoah 1995; [b] Buekens et al. 1995; [c] Williams and Chen 1982; [d] Buekens et al. 1995; [e] Buehler et al. 1987; [f] Koops et al. 1982
[g] Starfield et al. 1982; [h] Erickson and Bjerkedal 1982; [i] Behrman et al. 1971; [j] Chase 1969.
Source: Ashworth 1998, tables 1 and 2.
                                           CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA


7.3 Preterm birth

7.3.1 BIRTH DATA

Stillbirth
Some authors have presented results that indicate the prevalence of preterm birth in stillbirths.
Cartlidge and Stewart (1995) studied the 1993 birth cohort in Wales, UK, which included 221
stillbirths from 24 gestational weeks onwards. In this cohort, 39.4% of stillbirths had been
born before 32 gestational weeks and as many as 71.5% before 38 weeks. Data by Hilder et
al. (1998) for a London (UK) sample from 1989-1991 suggested that 55.5% of stillbirths aged
28 through 45 gestational weeks are preterm (< 37 weeks) and that 19.9% are below 32
weeks.
         Both these publications also provide data that can be used to estimate the prevalence of
stillbirth in preterm births. These figures also indicate an association between preterm birth
and stillbirth. In the Welsh cohort from 1993, the prevalence of SB was 3.7% in births aged
24 through 37 gestational weeks compared to only 0.2% in those aged 38 weeks or over.
Similarly, 18.3% of births aged 24 through 31 weeks, and 0.4% of births aged 32 weeks or
over were stillbirths (based on data from Cartlidge and Stewart 1995). In the sample by Hilder
et al. (1998), 3.9% of preterm births (≥ 28 weeks, but < 37 weeks) were stillbirths compared
to only 0.2% of births aged 37 weeks and over. Of preterm births below 32 weeks (but ≥ 28
weeks), 10.1% were born dead compared to 0.4% of births over 32 weeks. Besides these two
studies, additional results are available from Wolf (1991) who analysed 10,273 births of at
least 24 gestational weeks born between 1983 and 1989 in a hospital in Amsterdam, the
Netherlands. In preterm births, the prevalence of stillbirth ranged between 2.7% for those
aged 34-36 weeks and 20% for those aged 24-25 weeks. In comparison, only 0.4% of births
aged 37-41 weeks and 0.7% aged ≥ 42 weeks were born dead.
         The hypothetical cohort in Table 5.6, that was based on Californian data from
Goldhaber and Fireman (1991), provides the following results on preterm birth in relation to
stillbirth. Of the 434 stillbirths occurring from 28 gestational weeks onwards, 60% are
preterm (< 37 weeks) and 32% are < 32 weeks. Naturally, these proportions increase if the
gestational age limit that distinguishes stillbirths from spontaneous abortions is lowered from
≥ 28 weeks to ≥ 24 weeks or ≥ 22 weeks. For example, 73% of stillbirths ≥ 24 weeks and
77% of stillbirths ≥ 22 weeks are preterm (< 37 weeks). The prevalence of stillbirth among
preterm births aged 28 through 36 gestational weeks is relatively high at 5%, compared to
only 0.2% in births of 37 weeks and over. In preterm births ≥ 24 weeks and ≥ 22 weeks, the
percenages born dead are as high as 8 and 10% respectively. Among births aged 28 through
31 gestational weeks, 23% are stillborn compared to only 0.3% of births ≥ 32 weeks. Lastly,
for foetuses still in utero, the conditional weekly risk of live birth exceeds the risk of stillbirth
from gestational week 28 onwards (see Table 5.6).
         The causal relationship between stillbirth and preterm birth is in both directions. On
the one hand, intrauterine death before gestational week 37 is likely to bring about preterm


                                                                                                211
EARLY LIFE CHANGES

expulsion or birth. On the other hand, the preterm birth of a foetus who is still alive at the
onset of labour may result in complications for that baby – including hypoxia and asphyxia –
and finally, in death during labour. Within a population or cohort, both pathways or
mechanisms probably play a role. Therefore, a distinction between death prior to labour and
death during labour, i.e. antepartum and intrapartum death, is valid.
         Gardosi et al. (1998) obtained data for 1992 on 149 singleton stillbirths of at least 24
weeks, excluding those with congenital anomalies, in Trent, UK. The authors only included
antepartum deaths in their sample and found that 56% were preterm births (< 37 gestational
weeks). Bakketeig et al. (1978) made a distinction in their life table between deaths prior to
labour and deaths during labour. The data were on 440,452 single births of known gestational
age that occurred in Norway during 1967-1973. Overall, 52.9% of antepartum deaths ≥ 28
weeks were preterm, while only 28.3% of intrapartum deaths ≥ 28 weeks were preterm. When
the defining age limit of stillbirth is moved to ≥ 24 weeks, ≥ 22 weeks, and ≥ 16 weeks, the
prevalence proportions among deaths prior to labour become 59.2%, 61.9%, and 65.1%
respectively. For deaths during labour, the results are 36.0%, 41.2%, and 43.4% respectively.
From the same Norwegian data set, the prevalence of antepartum stillbirth among preterm
births (stillbirths ≥ 28 weeks and all live births) < 37 weeks is 7.9%. The percentage is only
1.0% of intrapartum foetal deaths. The figures increase if a lower age limit is used to define
stillbirth. When stillbirth is defined as ≥ 16 weeks, 12.3% of preterm births are foetal deaths
prior to labour, and 1.8% are deaths during labour (based on data from Bakketeig et al. 1978).

Live birth and all births
Table 7.14 presents some of the more recent figures on the proportion of newborns who are
preterm. In general, as Shibuya and Murray (1998a) concluded, the percentage seems to lie
between 5 and 7%. In comparison to most other Western and European countries, the
prevalence of preterm birth is higher in the USA (Paneth 1995) at about 10% (see Table 7.14).
Within the United States, large differences exist between racial/ethnic groups. For example,
Afro-American infants are about twice as likely as Caucasian infants to have been born
preterm (Paneth 1995; CDC 1999). In the 1988 birth cohort, 8.5% of Caucasian newborns
were preterm compared to 18.3% of Afro-American newborns (NCHS 1995). Despite this,
several studies have shown that when infants of the same birth weight categories are
compared, mortality figures in the USA are equal to or even lower than those in European
countries such as Sweden, Belgium, and Norway (Erickson et al. 1984 cited by Paneth 1995;
Buekens et al. 1995; Wilcox et al. 1995).
        In the hypothetical cohort, based on data from California (USA), 5.8% (5,167 out of
88,410) of live births are preterm. This figure is closer to prevalence figures observed in
European and other Western countries than it is to figures from the USA (see Table 7.14). The
neonatal life table for the hypothetical cohort was based on the 1988 birth cohort of singleton
live births in California (see Section 5.4). In this US cohort, 9.1% were born preterm (based
on data from NCHS 1995). Thus, although the two base samples, i.e. the data from Goldhaber
and Fireman (1991) and NCHS (1995), come from the same region, they do not seem to
match in terms of the prevalence of preterm birth.



212
                                                     CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

        In general, the majority of preterm live births take place close to term. The prevalence
of births below 32 gestational weeks is only about 0.8 to 2% of live births (NCHS 1995;
Magowan et al. 1998; Riley and Halliday 1999). In the hypothetical cohort, only 0.8% of live
births were below 32 weeks. In the 1988 birth cohort of Californian singletons, preterm live
births are distributed by gestational age as follows: 35.1% at 36 wks, 36.5% at 34-35 wks,
12.5% at 32-33 wks, 6.5% at 30-31 wks, 3.5% at 28-29 wks, 2.3% at 26-27 wks, and 3.6% ≤
25 wks (based on data from NCHS 1995).
        Several studies have shown an increased proportion of males among preterm births
(McGregor et al. 1992; Cooperstock and Campbell 1996; Astolfi and Zonta 1999; James
2000). For example, Cooperstock and Campbell (1996) studied birth records from 1977-1988
in New England, USA, and found an excess of males among singleton preterm births. Among
Caucasian births aged 20-36 weeks’ gestation, 54.9% were male and 45.1% female while at
term, 51.3% of singleton births were male and 48.7% female. This pattern was observed in
Caucasian foetal deaths as well as in Caucasian live births. However, the excess of males in
preterm births was significantly less pronounced within the Afro-American population
(Cooperstock and Campbell 1996). In Italy, Astolfi and Zonta (1999) studied live births of
first and second birth order in 1990-1994. The prevalence of preterm birth was 4.4% among
males compared with 4.0% among females. At term, only 51.5% of newborns were male
compared with 54.1% among preterm newborns. Similar differences between males and
females can be observed in the Californian birth cohort of 1988 (based on data from NCHS
1995). The prevalence of preterm birth was 9.6% among male live births with known
gestational age compared to 8.6% among females. In addition, the percentage of males among
preterm live births (< 37 weeks) was 54.1% compared to only 50.9% among live births ≥ 37
weeks.


Table 7.14: Prevalence of preterm birth in births

Source                                         Country                       Period         %      of            Remarks
Clausson et al. 1998                           Finland (Turku)            1970-1981 3.0-3.5 all births                 I
Newton 1989                                    -                              -         6   NS                         I
Bennebroek Gravenhorst et al. 1995             Netherlands                    -       6-7 all births                   I
NCHS 1995                                      USA                          1988      10.2 live births                II
Ohde et al. 1995                               -                              -       6-9 all births                   I
Clausson et al. 1998                           Sweden                     1992-1993 6.4 live births                  III
Joseph et al. 1998                             Canada                     1992-1994 6.8 live births                    -
Magowan et al. 1998                            UK (Scotland)              1985-1994 5.4 live births                  IV
Shibuya and Murray 1998a                       -                              -       5-7 NS                           I
CDC 1999                                       USA                          1996      9.7 live births                IV
Riley and Halliday 1999                        Australia (Victoria)         1998       6.7 "births"                    -
Notes: NS - not stated.
Remarks: I - general estimation; II - excl. births with unknown gestational age; III - singletons in nulliparous women;
IV - singletons.




                                                                                                                           213
EARLY LIFE CHANGES

7.3.2 NEONATAL DATA

Neonatal death
In recent decades, mortality for high-risk newborns (i.e. LBW, preterm and/or SGA) has
declined sharply. A large part of this decline can probably be attributed to the introduction of
the NICU (Neonatal Intensive Care Unit) and its technology (Van der Veen 2001). As a
result, findings of older studies on survival of preterm neonates may no longer be relevant
(Bryce 1991).
        Magowan et al. (1998) studied 33,912 singleton births that took place in Scotland
(UK) between 1985-1994. In total, 4.1% of live births aged 24 through 36 gestational weeks
and 19.4% aged 24 through 31 weeks did not survive the neonatal period. Friedman et al.
(1995), while studying the effects of pre-eclampsia, found a similar incidence of neonatal
death in preterm newborns, of 4.5%, in both the study and the control groups. Risks are
somewhat lower in the 1988 birth cohort from California (USA) with the incidence of
neonatal death being 3.0% in preterm singletons and 14.7% in single births below 32 weeks
(based on data from NCHS 1995). Copper et al. (1993) found incidence of neonatal death
among preterm births (< 37 weeks) to be as high as 10.3% for the period 1982-1986 in the
USA. Their study was based on 3,386 preterm singleton live births ≥ 20 weeks that were born
in university hospitals, and is therefore likely to include many high risk cases.
        With regard to the prevalence of preterm birth among neonatal deaths, many
publications report preterm birth in their data as a cause of neonatal death. Straightforward
prevalence proportions of preterm birth among neonatal deaths are often missing. Cartlidge
and Stewart (1995) found that out of the 128 neonatal deaths in the 1993 birth cohort in
Wales, 73.4% had been born below 38 gestational weeks. With preterm birth defined as less
than 37 weeks, the prevalence must have been somewhat lower. Furthermore, 53.1% of
neonatal deaths were very preterm, i.e. born at less than 32 weeks. In the Californian birth
cohort of 1988, 69.7% of neonatal deaths among singletons with known gestational age were
preterm, and 54.6% < 32 weeks (based on data from NCHS 1995).
        Earlier studies have produced prevalence figures that are slightly lower. In the data set
that Bakketeig et al. (1978) used to construct their foeto-infant life table, 63.5% of neonatal
deaths were born preterm and 42.2% before 32 weeks. The incidence of neonatal death in
preterm births was 11.0% but as high as 44.2% in very preterm births (< 32 weeks). These
Norwegian data were based on 440,452 single births occurring between 1967 and 1973.
Hardy et al. (1979) studied data from an even earlier survey. They studied the outcomes of
pregnancies in university-affiliated medical centres during the period 1959 through 1965 in
the USA. In Caucasian infants, 60.7% of neonatal deaths were preterm and the incidence of
neonatal death in preterm births was 10.2%. For Afro-American infants the corresponding
figures were 66.4 and 6.5% respectively.
        Detailed data by gestational age are available for the 1988 birth cohort of Californian
singletons. For this sample, neonatal deaths are distributed over the various gestational age
categories as follows (for those of known gestational age): 30.3% ≥ 37 weeks, 4.1% 36 wks,
5.8% 34-35 wks, 5.1% 32-33 wks, 4.8% 30-31 wks, 5.7% 28-29 wks, 9.7% 26-27 wks, and



214
                                                      CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

34.5% ≤ 25 wks (based on data from NCHS 1995). Over the years, as mortality has declined,
research interests have shifted to the most compromised infants: those born alive in the lowest
gestational age categories. Unsurprisingly, the risk of neonatal death is highest in these
categories. Table 7.15 presents incidence proportions by gestational age category based on
some relatively recent data sources. In the table, the risk of neonatal death is indeed highest
for the lowest gestational age categories. Compared with the other sources, the NICU-based
figures by Philip (1995) are more favourable in the lower gestational age groups while being
poorer for those born at more advanced gestational ages. Perhaps, the explanation for this is
found in the severe health problems being present in nearly term newborns who are admitted
to the NICU, while very preterm infants are almost always admitted to hospital.
       In the 1993 Welsh cohort reported by Cartlidge and Stewart (1995) mentioned earlier,
51.5% of early neonatal deaths were born at less than 32 gestational weeks and 71.7% were
below 38 weeks. The incidence of early neonatal death was 11.9% among live births aged 20
through 31 weeks, but only 1.6% among those aged 20 through 37 weeks. In Italy, Bottino et
al. (1991) found that early neonatal mortality is significantly lower starting from the 32nd
week of pregnancy (3.2% versus 35.4%). The authors studied preterm deliveries in a
university hospital between 1982 and 1988. Overall, the incidence of early neonatal death in
preterm newborns was 10.6%, and 8.3% after the exclusion of malformed cases. Early
neonatal death was lower for females than for males: 8.4% compared to 11.1% (Bottino et al.
1991).

Table 7.15: Incidence of neonatal death in preterm live births, by gestational age
Studies included
                                                                        Gestational         Incidence
Source                      Country                     Period             age              NND (%)     Remarks
Copper et al. 1993          USA                       1982-1986         20-36 wks             10.3      hospital-based
NCHS 1995                   USA (California)            1988             < 37 wks              3.0      singletons
Philip 1995                 USA (Maine)               1990-1991         = < 35 wks            7.4       NICU-based
Magowan et al. 1998         UK (Scotland)             1985-1994         24-36 wks              4.1      singletons

Neonatal mortality rate, by gestational age
               Incidence of neonatal death (%)
Gestational age Copper et al.       NCHS                             Philip         Magowan et al.
  (in weeks)        1993*            1995                            1995               1998
      = < 25           84.5 - 100.0              41.4                 30.4                  63.0
      26-27            33.0 - 45.3               17.6                 4.5                   34.5
      28-29            14.8 - 22.6                7.0                 3.3                   14.8
      30-31             5.8 - 9.4                3.1                  8.0                   7.3
      32-33             2.1 - 3.5                1.8                  2.9                   3.3
      34-35             0.8 - 1.3                0.7                  3.6                   1.2
        36                 0.5                   0.5                  NS                     0.9
Notes: NS - not stated; NND - neonatal death.
*Predictions for survival based on linear regression; data presented by gestational week.



                                                                                                                   215
EARLY LIFE CHANGES

        Advantage in survival of preterm females over preterm males has been reported in
several studies (Copper et al. 1993; Holtrop et al. 1994; Synnes et al. 1994). For example,
Copper et al. (1993) studied 3,386 preterm live births (≥ 20 weeks, but < 37 weeks of
gestation) during 1982-1986 that took place in five university hospitals in the United States.
Preterm males were found to have twice the neonatal mortality rate as females before 29
weeks. However, the difference in mortality between the sexes was not statistically different
after 29 weeks of gestation.

Relative risk and attributable risk: neonatal death
The hypothetical cohort and the data on live births and neonatal deaths that have been
discussed in the previous subsections, provide a basis for the estimation of the RR, AR(E) and
EF for preterm neonates. Table 7.16 presents three series of assumptions for those born < 37
weeks: for the USA, for other EME countries in general, and as based on the 1988 birth
cohort of Californian singletons (NCHS data). In the case of the United States, it is assumed
that (1) the prevalence of preterm birth in all live births (pB1) is 10%, (2) the incidence of
neonatal death in preterm infants (I1) is 3%, and (3) the prevalence of preterm birth among
neonatal deaths (pD1) is 65%. For EME countries other than the USA, the corresponding
figures are assumed to be: (1) 7%, (2) 4.3%, and (3) 70%. In the final series, according to
NCHS data for the 1988 Californian birth cohort, (1) pB1 is 9.1%, (2) I1 is 3.0%, and (3) pD1 is
69.7%. The resultant RRs of neonatal death, estimated using equations (4.18) and (4.19), are
highest in non-USA EME countries (30.2-31.0) and lower in the United States (16.7-20.3)

Table 7.16: Estimated RR, AR(E) and EF of neonatal death in preterm live births, hypothetical
cohort

                     Input                                                     Results
                        B            D         pB1          I1        pD1        RR        AR(E)         EF          check*
USA                   88,410        383       0.100      0.030         -         20.3       0.951      0.658      pD1 = 0.683
                      88,410        383       0.100        -         0.650       16.7       0.940      0.611        I1 = 0.028
                      88,410        383         -        0.030       0.650       17.9       0.944      0.614      pB1 = 0.094

Other EME             88,410        383       0.070      0.043         -         30.2       0.967      0.672      pD1 = 0.695
                      88,410        383       0.070        -         0.700       31.0       0.968      0.677        I1 = 0.043
                      88,410        383         -        0.043       0.700       30.8       0.967      0.677      pB1 = 0.071

California, 1988 88,410             383       0.091      0.030         -         17.0       0.941      0.593      pD1 = 0.630
                 88,410             383       0.091        -         0.697       23.0       0.956      0.667        I1 = 0.033
                 88,410             383         -        0.030       0.697       20.6       0.951      0.663      pB1 = 0.101

< 32 wks              88,410        383       0.014      0.170         -         85.9       0.988      0.543      pD1 = 0.549
                      88,410        383       0.014        -         0.540       82.7       0.988      0.533        I1 = 0.167
                      88,410        383         -        0.170       0.540       84.1       0.988      0.534      pB1 = 0.014
Notes: B - total no. of live births; D - total no. of neonatal deaths; pB1 - prevalence of preterm birth among live births;
I1 - incidence of neonatal deaths among preterm live births; pD1 - prevalence of preterm birth in neonatal deaths.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions
See Chapter 4 for equations. Based on the hypothetical cohort.




216
                                                       CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

and 1988 California (17.0-23.0). Variations in AR(E) and EF between the three regions are
less pronounced: the AR(E) varies between 94.0 and 96.8% and the EF ranges from 59.3 to
67.7% in all three data sets (see Table 7.16). For all three series, the combination of input data
and the resultant estimates appear to be plausible based on the check calculation in the last
column of the table.
        In addition to the three series of assumptions that were discussed above, Table 7.16
also estimates the RR, AR(E) and EF of neonatal death for neonates born at less than 32
gestational weeks (i.e. very preterm). If we assume for these infants that (1) pB1 is 1.4%, (2) I1
is 17%, and (3) pD1 is 54%, the subsequent RR of neonatal death is quite high. It appears that
the risk of neonatal death in infants < 32 weeks is about 83 to 86 times the risk among
newborns ≥ 32 gestational weeks. Moreover, as many as 99% of neonatal deaths among very
preterm cases can be attributed to the low gestational age at birth of less than 32 weeks
(AR(E) = 0.988). Within the total population, 53.3 to 54.3% of neonatal deaths are
attributable to a gestational age of less than 32 weeks (EF). On the basis of the re-calculated
checks in the final column of Table 7.16, these assumptions and results all appear to be
credible.
        Table 7.17 estimates the relative risks and attributable risks of early neonatal death.
The assumptions for this estimation are mainly derived from the data by Cartlidge and
Stewart (1995) and from the previous assumptions in Table 7.16 for EME countries other than
the USA. The incidence of early neonatal death (I1) is assumed to be around 1.5% in preterm
neonates and 11.9% in very preterm infants. In addition, 51.5% of early neonatal deaths are
believed to be very preterm (pD1). After the application of these assumptions to the
hypothetical cohort, the RR is 6 to 31 for neonates < 37 gestational weeks, and 68 to 75 for
those born < 32 weeks. The AR(E) for these two groups is 83-97% and 99% respectively.
Finally, the EF lies between 26 and 68% for preterm birth, and between 49 and 51% for very
preterm birth. However, the final column of Table 7.17 indicates that the estimates for
preterm birth < 37 weeks should be used with caution.

Table 7.17: Estimated RR, AR(E) and EF of early neonatal death in preterm live births,
hypothetical cohort

             Input                                                        Results
                B             D          pB1          I1         pD1        RR          AR(E)         EF           check*
 < 37 wks 88,410             299        0.070       0.015         -           6.0       0.833       0.259       pD1 = 0.310
          88,410             299        0.070         -         0.700        31.0       0.968       0.677         I1 = 0.034
          88,410             299          -         0.015       0.700        12.5       0.920       0.644       pB1 = 0.158

 < 32 wks 88,410             299        0.014       0.119         -          68.4       0.985       0.485       pD1 = 0.493
          88,410             299        0.014         -         0.515        74.8       0.987       0.508         I1 = 0.124
          88,410             299          -         0.119       0.515        71.5       0.986       0.508       pB1 = 0.015
Notes: B - total no. of live births; D - total no. of neonatal deaths; pB1 - prevalence of preterm birth among live births;
I1 - incidence of neonatal deaths among preterm live births; pD1 - prevalence of preterm birth in neonatal deaths.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.




                                                                                                                              217
EARLY LIFE CHANGES
Relative risk and attributable risk, polytomous age categories
Data from the 1988 birth cohort of singletons in California provide the opportunity to
calculate RR, as well as AR(E) and EF, by gestational age. Table 7.18 presents the results for
the polytomous gestational age categories. Each preterm category is compared to newborns ≥
37 weeks. As an example, newborns ≤ 25 weeks are 314 times as likely as newborns ≥ 37
gestational weeks to die during the neonatal period. Unsurprisingly, Table 7.18 shows that the
relative risk of neonatal death drops sharply with increasing gestational age at birth. The same
holds true for the AR(E) and the EF. However, the concept of AR(E) by gestational age is not
easy to interpret. It refers to the proportion of neonatal deaths in a specified preterm category
that can be attributed to this preterm gestational age. Nevertheless, the trend in AR(E) over
the various gestational age categories is worth noting. The proportion reduces with increasing
gestational age. The etiologic fractions for polytomous categories in Table 7.18 were
calculated on the basis of equation (4.12), following Kramer et al. (2000) and Miettinen
(1974). Despite the small prevalence of the lowest gestational ages, their contribution to
neonatal death within the total population is considerable. Overall, 66.7% of neonatal deaths
in the population can be attributed to preterm birth (< 37 weeks) and 54.2% to very preterm
birth (< 32 weeks).
        Kramer et al. (2000) used polytomous preterm categories and divided mortality into
early and late neonatal mortality. The authors analysed linked live birth-infant death cohort
files of singletons from 1995 for the USA and from 1992-1994 for Canada (excluding
Ontario). The risks of late neonatal death were based on early neonatal survivors and the
various preterm groups were compared to infants born ≥ 37 weeks. Table 7.19 presents the
results for the RR and the EF which are, in general, more pronounced for early neonatal
mortality than for late neonatal mortality. As in Table 7.18, the RRs in very preterm infants
are much higher than those in near-term neonates. In addition, the categories below 28
gestational weeks contribute a relatively large proportion of all the neonatal deaths within the

   Table 7.18: Estimated RR, AR(E) and EF of neonatal death for polytomous age
   categories with non-preterm as reference group, California, 1988

   GA (wks)        % of LB              LB       % of NND         NND            RR      AR(E)    EF
      < = 25            0.3       1,723.6            34.5        777.3           313.5   0.997   0.344
       26-27            0.2       1,096.8             9.7        218.5           138.5   0.993   0.096
       28-29            0.3       1,671.3             5.7        128.4            53.4   0.981   0.056
       30-31            0.6       3,133.7             4.8        108.1            24.0   0.958   0.046
       32-33            1.1       5,954.1             5.1        114.9            13.4   0.925   0.047
       34-35            3.3      17,392.2             5.8        130.7             5.2   0.809   0.047
         36             3.2      16,713.2             4.1         92.4             3.8   0.740   0.030
      All < 37          9.1      47,684.9            69.7      1,570.3            22.9   0.956   0.667

      > = 37          90.9      474,603.1            30.3        682.7             1.0     -       -
   Notes: LB - live birth; NND - neonatal death
   Each group of preterm neonates compared to birth > = 37 weeks of gestation.
   Based on data from NCHS 1995 for 1988 birth cohort, singletons, California.




218
                                                    CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA
    Table 7.19: RRs and EFs of neonatal death for polytomous age categories with non-
    preterm as reference group, United States, 1995, and Canada*, 1992-1994
                     USA, 1995                                            Canada*, 1992-1994
    Gestational           RR                    EF (%)                        RR                     EF (%)
    age (wks)         ENND LNND               ENND LNND                   ENND LNND                ENND LNND
       < 28            465.5       129.7         59.9       27.7            496.2      185.6          44.6       19.3
       28-31            40.1        20.2          8.6       10.0             70.2       26.6          10.0        7.1
       32-33            14.6         6.3          3.6        3.4             33.0       13.1           6.2        4.6
       34-36             5.2         2.8          6.3        6.8              7.9        3.6           9.0        6.9
    Notes: ENND - early neonatal death; LNND - late neonatal death; *Canadian data exclude Ontario.
    US data adjusted for maternal age, parity, race, and education; Canadian data adjusted for maternal age and parity.
    EFs based on multiple logistic regression analyses..
    Among singleton live births. Each group of preterm neonates compared to birth > = 37 weeks of gestation.
    Source: Kramer et al. 2000


total population. The results have led Kramer et al. (2000) to conclude that “infants born at 27
gestational weeks or earlier account for a very large proportion of neonatal deaths despite
their rare occurrence (0.7% in the United States and 0.4% in Canada)” (p.847).

7.4 Intrauterine growth retardation (IUGR) and small-for
gestational-age (SGA)

7.4.1 ANTEPARTUM AND BIRTH DATA

Incidence of IUGR
In theory, growth retardation can arise at any time during gestation although delayed growth
early in gestation is more likely to manifest itself in the development of organs (e.g.
congenital anomalies) while retarded growth during the third trimester mainly affects the size
of the foetus. Vik et al. (1997) studied singleton foetuses born in Norway and Sweden in
1986-1988 and found that growth retardation – defined as birth weight < 10th percentile based
on a Norwegian reference standard (Bakketeig et al. 1993) – started in the second trimester.
From gestational week 25 onwards differences could be observed in measurements of
midabdominal diameter, femur length, and biparietal diameter between SGA and non-SGA
infants. Generally, symmetric growth retardation is assumed to start as early as the first
trimester of gestation whereas asymmetric growth retardation may start as late as the third
trimester. However, the study by Vik et al. (1997) did not confirm this hypothesis and found
similar prenatal growth patterns in both groups of SGA infants.

Induced abortion and selective abortion
No data are available on intrauterine growth retardation in relation to induced abortion. At
first sight, it seems plausible to assume that the prevalence in induced abortuses of a certain
gestational age would equal the prevalence among all embryos/foetuses in utero of the same


                                                                                                                          219
EARLY LIFE CHANGES

age. However, intrauterine growth is assumed to be, indirectly, affected by a very young
maternal age (Kramer 1987a) and the youngest, as well as the most advanced, age groups are
overrepresented among women who opt for an induced abortion. Moreover, retardation of
growth appears to be related to the development and presence of congenital anomalies (see
Section 7.6), which suggests that there is at least some influence by selective abortion.

Spontaneous loss and stillbirth
Several studies indicate that the risk of spontaneous loss and stillbirth is increased in growth-
retarded embryos and foetuses. Bromley et al. (1991) found that 94% of first-trimester small
sacs (but normal cardiac activity) had spontaneous abortions compared to only 8% of embryos
with normal sac size (i.e. RR = 94 / 8 = 11.8). Data for later gestational ages come from
Kramer et al. (1990) who studied singleton births without congenital anomalies or congenital
infections, born in a hospital in Montreal, Canada, between 1980 and 1986. They defined
IUGR/SGA on the basis of foetal growth ratio (FGR, see Chapter 3) and classified growth
retardation by severity. The incidence of stillbirth was 0.3% in the group without IUGR
compared to: 1.2% in the group of mild IUGR, 0.5% in moderate IUGR, and 7.1% in severe
IUGR (p < 0.0001) (Kramer et al. 1990).
        Figures on the prevalence of IUGR/SGA in spontaneous abortuses come from Fantel et
al. (1980) who collected data in the USA between 1971 and 1979. The group of embryos and
foetuses defined as growth retarded included “all those specimens which would be described
as stunted, nodular, cylindrical, etc.” (p.74). After excluding anomalous specimens and cases
whose normality could not be determined, the prevalence of growth retardation was found to
be 32.5%. Fantel et al. (1980) compared their findings to the results of several other studies
that are presented in Table 7.20. Overall, the prevalence appears to range between 10 and
36%. Part of the variation is probably explained by differences among the studies. For
example, differences in study design, inclusion criteria, definitions and operationalisation,
gestational age, and sample size.
        Concerning the prevalence of IUGR/SGA in stillbirths, several studies have reported
that nearly 50% of stillbirths are growth retarded (Knutzen and Sher 1982 cited by
Goldenberg et al. 1987). Similarly, in the 1992 Trent Regional Epidemiological Survey of
Perinatal Mortality (UK), 41% of 149 singleton stillbirths ≥ 24 weeks (excluding congenital
anomalies and intrapartum deaths) were SGA as defined by < 10th percentile (Gardosi et al.
1998).

Live birth and all births
The prevalence of IUGR/SGA in births strongly depends on the ‘standard’ curve chosen
(often sex-specific) and its cut-off point. Intrinsic to a definition based on either the 3rd, 5th, or
10th percentile of a birth weight distribution, the prevalence of SGA in births is approximately
3, 5, or 10% respectively. This will clearly be the case with weight-for-gestational age curves
that are internally derived, i.e. based on data from the population under study. For good
reason Van der Veen (2001) writes: “A major weakness of using the 10th percentile in
combination with birth outcome data is that, by definition, the birth prevalence of IUGR in
the population remains 10%, irrespective of possible changes” (p.185).


220
                                                CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA
         Table 7.20: Prevalence of growth retardation in spontaneous losses

         Source                              Sample size           %              Remarks
         Mall and Meyer 1921                      749               24            -
         McMahon et al. 1954                      637             24.6            -
         Fujikura et al. 1966                     130             10.0            -
         Poland 1968                               91             22.0            -
         Stratford 1970                           119             36.1            -
         Nelson et al. 1971                       107             17.8            USA
         Fantel et al. 1980                       317             32.5            USA, 1971-1979
         Note: Recalculated proportions after exclusion of anomalous specimens and cases whose
         normality could not be determined.
         Source: Fantel et al. 1980


        Apart from the percentiles, another cut-off point that is frequently chosen to
distinguish SGA newborns from non-SGA newborns is 2 standard deviations below mean
weight. For example, Clausson et al. (1998) used the Swedish antenatal growth curve by
Marsál et al. (1996) and placed the cut-off point at 2 standard deviations. In their sample of
singleton live births to nulliparous women in Sweden, in 1992-1993, 3.8% were ultimately
classified as SGA.
        Further, the concept of foetal growth ratio (FGR) has also been applied to define SGA.
FGR is the ratio of the observed birth weight to the mean birth weight for gestational age of a
growth standard (Kramer et al. 1990; Frisbie et al. 1996). The cut-off point is usually set at
0.85, with an FGR below 0.85 indicating growth retardation. Kramer et al. (1990) expanded
this distinction by including cut-off points for severity: mild IUGR when FGR was 0.80-0.84,
moderate IUGR when FGR was 0.75-0.79, and severe IUGR when FGR was less than 0.75.
In a Canadian cohort of live births between 1980 and 1986, these cut-off points corresponded
to the 9.0, 4.3, and 1.7 percentiles respectively (Kramer et al. 1990).

Prevalence in utero
Some indications of the prevalence of IUGR among the population of embryos and/or
foetuses in utero come from studies based on ultrasound examinations. Drugan et al. (1992b)
obtained crown-rump lengths for first trimester embryos/foetuses in women referred for
chorionic villus sampling between 1988 and 1991. They found that 8.7% had a foetal length
that was smaller than expected by at least 7 days. At later gestational ages, Yoshida et al.
(2000) studied singleton pregnancies without apparent foetal malformations that later resulted
in term deliveries in the University of Tokyo Hospital between 1987 and 1992. Foetal body
weight was estimated on the basis of ultrasound examinations performed at least every four
weeks from 20 weeks gestation onwards, and growth retardation was diagnosed where the
estimated weight was below the standard birth weight by more than 1.5 standard deviations.
Overall, 10.2% were diagnosed as IUGR, and in another 7.6% of cases IUGR was presumed.
Langhoff-Roos and Lindmark (1997) examined 1,435 singleton foetuses that went on to term
live birth in Norway and Sweden. Foetal weight was estimated ultrasonographically and
IUGR was diagnosed where the weight deviated by more than 2 standard deviations from the


                                                                                                   221
EARLY LIFE CHANGES

mean for length of gestation. At 33 weeks of gestation, only 0.8% were found to be growth
retarded while at week 37 the figure was 3.0%.
       On the basis of the data discussed above, it is also possible to make an estimation for
the hypothetical cohort that was constructed in Chapter 5. The 100,000 singleton pregnancies
that are present at the beginning of gestational week 5 in this cohort have the following
outcomes: 11,157 spontaneous losses (SA) before 28 weeks of gestation, 434 stillbirths (SB)
≥ 28 weeks, and 88,410 live births (LB). Let us suppose, on the basis of the data in the
previous subsections that the prevalence of IUGR/SGA among SA and SB are 24% and 45%
respectively. In addition, the prevalence among live births is assumed to be 10%, as based on
the 10th percentile. Although this does not seem to the best percentile to distinguish high-risk
cases from cases with normal risk (McIntire et al. 1999), the figure of 45% used for stillbirths
was also based on the 10th percentile (by Gardosi et al. 1998). Subsequently, the total number
of growth-retarded cases is 11,714 (2,678 SA + 195 SB + 8,841 LB). This implies that 11.7%
of hypothetical embryos will experience growth retardation. However, it is not known from
what gestational week onwards their growth will lag behind.

Relative risk and attributable risk: foetal loss
The results from the studies by Bromley et al. (1991) and Kramer et al. (1990) on the
incidence of spontaneous abortion and stillbirth, provide the opportunity to estimate the
relative risk. Based on Bromley et al. (1991), the risk of SA for growth-retarded
embryos/foetuses is about 12 times the risk for normal ones since RR is 94 divided by 8
(11.8). Consequently, the attributable risk (exposed), AR(E), is:
    (11.8 – 1) / 11.8 = 0.915
which means that 91.5% of spontaneous losses among growth-retarded embryos/foetuses can
be attributed to IUGR. In addition, the etiologic fraction (EF) can be calculated as:
    [0.087 * (11.8 – 1)] / [0.087 * (11.8 – 1) + 1] = 0.484
using the assumption that, in total, 8.7% of embryos/foetuses are growth retarded (cf. Drugan
et al. 1992b). This implies that about 48% of all spontaneous abortions within the population
are attributable to IUGR. On the basis of the data by Kramer et al. (1990), the RR of stillbirth
can be estimated by severity of growth retardation: 1.2 divided by 0.3 is 4.0 in the case of
mild IUGR, 0.5 / 0.3 = 1.7 for moderate IUGR, and 7.1 / 0.3 = 23.7 for severe IUGR (see data
above). However, these authors also calculated RR and presented their results in a figure.
According to these findings, the RR values are about 4.8, 1.8, and 27.5 respectively.
        RR, AR(E), and EF can also be estimated on the basis of the hypothetical cohort.
Continuing the estimations in the subsection on prevalence in utero, where the prevalence
proportions of IUGR/SGA in SA, SB and LB were assumed to be 24%, 45%, and 10%
respectively, this implies that about 25% of growth-retarded embryos end in spontaneous loss
or stillbirth. For normally grown embryos, the resultant incidence of SA/SB is about 10%. As
a result, the RR is estimated to be 2.5. Consequently, the AR(E) is 1.5 divided by 2.5 or about
60% and the EF is:
    [0.117 * (2.5 – 1)] / [0.117 * (2.5 – 1) + 1] = 0.149
or about 15%. Table 7.21 summarises these assumptions and the resultant estimates. In
addition, the table extends the previous assumptions and results to obtain separate estimates


222
                                                    CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

for spontaneous loss (SA) and stillbirth (SB). This requires an additional assumption on the
prevalence of IUGR among foetuses still in utero at the end of the second trimester. In the
hypothetical cohort, 88,614 foetuses are still in utero at the beginning of gestational week 28
(see Table 5.6, Chapter 5) and Table 7.21 assumes that 10% of these are growth retarded. A
combination of prevalence proportions with total numbers then yields incidences of SA and
SB, for both the growth-retarded and the normally grown population (see Table 7.21).
Ultimately, the RR, AR(E) and EF of spontaneous abortion are 2.4, 58.0%, and 13.9%
respectively. In the case of stillbirth, the effect of IUGR/SGA is stronger: the RR is 7.4, the
AR(E) is 86.4%, and the EF is 38.9%. It should be noted that these estimates are based on the
10th percentile.
        An additional method of obtaining estimates is to combine data from the literature with
the hypothetical cohort through equations (4.18) and (4.19). In the case of spontaneous
abortion, the assumptions are that: (1) the prevalence of IUGR in embryos/foetuses (pB1) is
8.7% (cf. Drugan et al. 1992b), (2) the incidence of SA in growth-retarded embryos/foetuses
(I1) is 94% (cf. Bromley et al. 1991), and/or (3) the prevalence of IUGR in SA (pD1) is 24%
(based on Table 7.21). For stillbirth, the assumptions are as follows: (1) pB1 is 10.2% (cf.
Yoshida et al. 2000 and based on 10th percentile), (2) I1 is 4% (based on Kramer et al. 1990),
and/or (3) pD1 is 45% (cf. Knutzen and Sher 1982 cited by Goldenberg et al. 1987 and by
Gardosi et al. 1998). Table 7.22 presents the resultant RRs, AR(E)s and EFs. For spontaneous
loss, the RR ranges from 3 to 29, the AR(E) from 70 to 97%, and the EF from 17 to 71%.
Once again the effect of IUGR on stillbirth appears to be stronger: the RR varies between 7
and 44, the AR(E) between 86 and 98%, and the EF between 39 and 81%. The final column
of Table 7.22 checks the consistency of the data and assumptions by recalculating the one
measure – either pB1, I1, or pD1 – that was missing from the input data. Unfortunately, the
differences between these re-estimated measures and the input assumptions are large.
Therefore, the credibility of the estimates leaves much to be desired.


Table 7.21: Estimated RR, AR(E) and EF of SA and SB in growth-retarded embryos/foetuses,
hypothetical cohort

              B        pB1        SA        SB       pSA1      pSB1        I1        I0        RR      AR(E)       EF
SA/SB 100,000 0.117 11,157                  434      0.240     0.450     0.246     0.099       2.5      0.598     0.148
 SA 100,000 0.117 11,157                     -       0.240       -       0.229     0.096       2.4      0.580     0.139
 SB   88,614 0.100     -                    434        -       0.450     0.022     0.003       7.4      0.864     0.389
Notes: B - total no. of embryos or foetuses in utero at the beginning of the interval; pB1 - prevalence of IUGR among B;
SA - total no. of SA; SB - total no. of SB; pSA1 - prevalence of IUGR among SA; pSB1 - prevalence of IUGR among SB;
I1 - incidence of SA and/or SB among growth-retarded embryos/foetuses; I0 - incidence of SA and/or SB among normally
grown embryos/foetuses.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.




                                                                                                                     223
EARLY LIFE CHANGES
Table 7.22: Estimated RR, AR(E), and EF of SA/SB in growth-retarded embryos/foetuses,
hypothetical cohort

      Input                                                        Results
          B          D           pB1          I1          pD1         RR          AR(E)          EF          check*
SA 100,000        11,157        0.087       0.940          -          28.8         0.965       0.708      pD1 = 0.733
   100,000        11,157        0.087         -          0.240         3.3         0.698       0.168        I1 = 0.308
   100,000        11,157          -         0.940        0.240        10.8         0.907       0.218      pB1 = 0.028
SB 88,614           434         0.102       0.040          -          43.9         0.977       0.814      pD1 = 0.833
   88,614           434         0.102         -          0.450         7.2         0.861       0.388        I1 = 0.022
   88,614           434           -         0.040        0.450        14.0         0.929       0.418      pB1 = 0.055
Notes: B - total no. of embryos or foetuses in utero at the beginning of the interval; pB1 - prevalence of IUGR among B;
SA - total no. of SA; SB - total no. of SB; pSA1 - prevalence of IUGR among SA; pSB1 - prevalence of IUGR among SB;
I1 - incidence of SA and/or SB among growth retarded embryos/foetuses; I0 - incidence of SA and/or SB among normally
grown embryos/foetuses.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.


7.4.2 NEONATAL DATA

Neonatal death
Results from studies suggest that the incidences of neonatal death and early neonatal death are
indeed higher among SGA infants than non-SGA infants. Clausson et al. (1998) studied
singleton live births to nulliparous women in Sweden in 1992-1993 and defined SGA as 2
standard deviations below mean birth weight. The neonatal mortality rate among SGA infants
was 17.6 per 1,000 births (i.e. 1.8%) compared to only 2.0 per 1,000 births among non-SGA
infants. McIntire et al. (1999) derived smoothed birth-weight curves and reapplied them to the
sample population of singleton live births in a hospital in Dallas (USA) between 1988 and
1996. Out of 3,562 infants ≤ 3rd percentile, 1.0% died during the neonatal period and the
prevalence of SGA in neonatal deaths was 12.8%. Using the 10th percentile as a cut-off point,
the corresponding figures were 0.6% and 26.7% respectively. In comparison, only 0.3% of the
group of AGA (appropriate-for-gestational-age) infants in the 26th-75th percentile band died
during the neonatal period.
        The incidence of neonatal death in SGA infants and the prevalence of SGA in neonatal
deaths can also be estimated for the Californian birth cohort of 1988 (NCHS 1995). However,
first, a growth curve needs to be applied to the cohort to distinguish SGA from non-SGA
newborns. Alexander et al. (1996) developed a sex-specific growth curve on the basis of
singleton live births to US resident mothers in 1991 (based on data from the NCHS) which
appears to be an appropriate choice here. Applying the 10th percentile to the Californian
singletons of known sex, known gestational age, and known birth weight yields a birth
prevalence of 8.6%. Ultimately, 1.4% of these SGA newborns died during the neonatal
period; 1.5% of male SGA neonates and 1.3% of female SGA neonates. The prevalence of
SGA among those neonatal deaths of known sex, known gestational age, and known birth
weight was as high as 30.6% (based on NCHS 1995).


224
                                                       CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

        Results by severity of growth retardation from Kramer et al. (1990) are not about
neonatal mortality but about in-hospital mortality. As noted earlier, the authors defined IUGR
on the basis of foetal growth ratio (FGR) and classified growth retardation as mild, moderate,
or severe. In their sample of singleton births in Montreal (Canada) between 1980 and 1986,
in-hospital mortality was 0.7% in the group with mild IUGR, 1.3% with moderate IUGR, and
1.4% with severe IUGR compared to only 0.3% in non-IUGR births. The risk of neonatal
mortality has also been said to differ between stunted (i.e. proportional) and wasted (i.e.
disproportional) IUGR/SGA infants. In Italy, Cuttini et al. (1991 cited by Ashworth 1998)
found a five-fold increase in neonatal death for proportionate IUGR infants (< 5th percentile,
Italian reference population) at term compared to those who were disproportionate.

Relative risk and attributable risk: neonatal death
The results above on the incidence of neonatal death from the studies by Clausson et al.
(1998), McIntire et al. (1999), and Kramer et al. (1990) can be easily transformed into relative
risks. The data by Clausson et al. (1998) suggest that the RR of neonatal death for SGA
newborns (weighing more than 2 S.D. below the mean) is 17.5 divided by 2.0, or 8.8.
Subsequently, AR(E) is 7.8 divided by 8.8, or 88.6%. Since the total prevalence of SGA
among live births was 3.8% within the sample, the EF is calculated as:
    [0.038 * (8.8 – 1)] / [0.038 * (8.8 – 1) + 1] = 0.229
or 22.9%. McIntire et al. (1999) compared SGA neonates (≤ 3rd and ≤ 10th percentile) to AGA
neonates in the 26th-75th percentiles. The relative risk of neonatal death is 1.0 divided by 0.3
or 3.3 for SGA infants ≤ 3rd percentile, and 0.6 divided by 0.3 or 2.0 for SGA infants ≤ 10th
percentile. On the basis of Kramer et al. (1990), the RRs of in-hospital mortality by severity
of growth retardation are 2.3 in cases of mild IUGR (0.7 divided by 0.3), 4.3 with moderate

Table 7.23: Estimated RR, AR(E) and EF of neonatal death in growth-retarded live births,
hypothetical cohort

                           Input                                                Results
                              B           D        pB1        I1       pD1        RR AR(E)              EF         check*
< 10th percentile [a]       88,410       383      0.086     0.014        -         4.1      0.756     0.210 pD1 = 0.278
                            88,410       383      0.086       -        0.306       4.7      0.787     0.241   I1 = 0.015
                            88,410       383        -       0.014      0.306       4.2      0.763     0.233 pB1 = 0.095

=< 10th percentile [b] 88,410            383      0.100     0.006        -         1.4      0.309     0.043 pD1 = 0.139
                       88,410            383      0.100       -        0.267       3.3      0.695     0.186   I1 = 0.012
                       88,410            383        -       0.006      0.267       1.5      0.344     0.092 pB1 = 0.193

=< 3rd percentile [b] 88,410             383      0.030     0.010        -         2.4      0.584     0.040 pD1 = 0.069
                      88,410             383      0.030       -        0.128       4.7      0.789     0.101   I1 = 0.018
                      88,410             383        -       0.010      0.128       2.5      0.600     0.077 pB1 = 0.055
Notes: B - total no. of live births; D - total no. of neonatal deaths; pB1 - prevalence of SGA among live births; I1 - incidence
of neonatal deaths among SGA live births; pD1 - prevalence of SGA in neonatal deaths.
[a] based on NCHS 1995; [b] based on McIntire et al. 1999.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.



                                                                                                                            225
EARLY LIFE CHANGES

IUGR (1.3 divided by 0.3), and 4.7 with severe IUGR (1.4 divided by 0.3); all compared to
non-IUGR infants.
        Combining data from NCHS (1995) or McIntire et al. (1999) with the hypothetical
cohort also results in estimates of the RR. For example, using the data from the 1988
Californian birth cohort: (1) the prevalence of IUGR/SGA in live births (pB1) is 8.6%, (2) the
incidence of neonatal death in growth-retarded live births (I1) is 1.4%, and/or (3) the
prevalence of IUGR/SGA in neonatal deaths (pD1) is 30.6%. Subsequently, the RR, AR(E),
and EF can be estimated using equations (4.18) and (4.19). Table 7.23 presents the results,
based both on input data from NCHS (1995) and from McIntire et al. (1999). If IUGR/SGA is
defined on the basis of the 10th percentile, the RR varies between 1.4 and 4.7, the AR(E)
between 31 and 79%, and the EF between 4 and 24%. If IUGR/SGA is defined on the basis of
the 3rd percentile, the RR ranges from 2.4 to 4.7, the AR(E) from 58 to 79%, and the EF from
4 to 10%. However, the last check column in Table 7.23 suggests that the results need to be
treated with caution, especially those based on McIntire et al. (1999).

7.5 Birth asphyxia (BA)

7.5.1 BIRTH DATA

Stillbirth
The most common cause of stillbirth is generally believed to be chronic foetal hypoxia or
asphyxia (Konchak et al. 1989). Hovatta et al. (1983) examined all stillbirths (≥ 26 weeks) at
the Helsinki University Central Hospital, Finland during the years 1974-1979. In total, there
were 243 stillbirths of whom 200 had died before labour. The predominant causes of stillbirth
were intrauterine asphyxia and major malformations. According to autopsy findings, asphyxia
accounted for 38% of foetal deaths. However, no diagnosis could be made as a result of foetal
maceration in another 42% of cases. More recently, Bastian et al. (1998) studied 7,002
planned home births in Australia between 1985-1990. Of the 31 foetal deaths (> 500 g, both
antepartum and intrapartum) in their sample, 51.6% (16 cases) were associated with
intrapartum asphyxia, including two cases of birth asphyxia due to shoulder dystocia. In
another 25.8% of cases death was unexplained.
        In intrapartum stillbirths, the proportion of asphyxiated cases appears to be very high.
In Finland, Erkkola et al. (1984) studied intrapartum foetal deaths born at the University
Hospital in Turku during the years 1970-1981. In about 90% of cases, asphyxia was the
ultimate cause.

Live birth and all births
Table 7.24 presents results of several studies on the proportion of birth asphyxia in births and
live births. Many authors refer to this proportion as incidence. However, in the present study,
BA is regarded as a characteristic of the newborn, and the proportion of BA cases among
births or live births as a prevalence (see Chapter 4). Nevertheless, the present book frequently
refers to this measure using more neutral expressions such as ‘proportion’ and ‘frequency’.


226
Table 7.24: Frequency of birth asphyxia in births
In births
Source                                Country                    Period     per 1,000 Definition                                      Remarks
Ergander et al. 1983 [a]              Sweden                  1973-1979        2.6      low Apgar score (< 3 at 5 min.)               -
Cry et al. 1984 [a]                   Canada                  1960-1962        15.4     HIE                                           -
Cry et al. 1984 [a]                   Canada                  1978-1980        10.4     HIE                                           -
Low et al. 1990b [a]                  USA                     1987-1989         27      umbillical blood                              -
Shibuya and Murray 1998b              EME                         -            2-9      HIE                                           general estimation, p.435
Heinonen and Saarikoski 2001          Finland (Kuopio)        1990-1998         25      umbilical artery base deficit > 12 mmol/L     in structurally normal
                                                                                                                                        singletons

In live births
Source                                Country                    Period     per 1,000 Definition                                      Remarks
Levene et al. 1985 [b]                UK (Leicester)       NS                    6      HIE                                           in term infants
Hull and Dodd 1992                    UK (Derby)        1976-1980               7.7     HIE                                           in term infants
Hull and Dodd 1992                    UK (Derby)        1984-1988               4.6     HIE                                           in term infants
Palme-Kilander 1992 [b]               Sweden               NS                   17      Apgar =< 3 at 1 min. and/or =< 6 at 5 min.    incl. preterm infants
Thornberg et al. 1995                 Sweden (Göteborg) 1985-1991               6.9     term + Apgar score < 7 at 5 min.              -
Thornberg et al. 1995                 Sweden (Göteborg) 1985-1991               5.4     term + Apgar score < 7 at 5 min.*             -
Thornberg et al. 1995                 Sweden (Göteborg) 1985-1991               1.8     term + Apgar score < 7 at 5 min. + HIE*       -
Badawi et al. 1998                    Australia (Perth) 1993-1995              3.80     moderate or severe "newborn encephalopathy"   in term infants
Shibuya and Murray 1998b              EME                   -                  5-10     HIE                                           general estimation, p. 440
Smith et al. 2000                     UK (Derby)        1992-1996               1.9     HIE                                           in term infants
Notes: NS - not stated; HIE - hypoxic-ischaemic encephalopathy.
*Excluding infants with congenital malformations, infections and opioid-induced respiratory depression.
[a] cited by Shibuya and Murray 1998b; [b] cited by Thornberg et al. 1995
EARLY LIFE CHANGES

        Many of the studies in Table 7.24 are hospital-based. The frequency of BA among all
births appears to range from 2 to 27 per 1,000 births (0.2 to 2.7%) whereas estimates of
prevalence in live births vary from 1.8 to 17 per 1,000 live births (0.18 to 1.7%). The large
variation in results between studies can be partly explained by the study period and, to a large
extent, to differences in definition of birth asphyxia (including severity).
        Over the years, the frequency of birth asphyxia has seen a declining trend (Shibuya and
Murray 1998b). Such a trend was, for example, observed in the Derby study covering the
periods 1976-1980, 1984-1988, and 1992-1996 (Hull and Dodd 1992; Smith et al. 2000).
Further, the Swedish study by Thornberg et al. (1995) in Table 7.24 demonstrates how the
definition of BA can affect its frequency. The frequency of BA was 6.9 per 1,000 live-born
infants when all term newborns with Apgar scores below 7 at 5 minutes were included in the
birth asphyxia group. The authors, however, decided to exclude infants with congenital
malformations, infections, and opioid-induced1 respiratory depression from the ‘birth
asphyxia group’, and the frequency then declined to 5.4 per 1,000 live-born infants. The
figure dropped even further to 1.8 per 1,000 live-born infants when only birth asphyxia cases
with HIE were included. Levene et al. (1985 cited by Hull and Dodd 1992, and by Smith et al.
2000) have graded HIE (hypoxic-ischaemic encephalopathy) into mild (grade I), moderate
(grade II), and severe (grade III). In the Derby study, the overall frequency of HIE in term
infants was 1.9 per 1,000 live births in 1992-1996 (Smith et al. 2000). When only moderately
and severely affected infants (grades II and III) were included, the figure dropped to 1.2 per
1,000 live births. Even stronger declines were observed in Derby in 1976-1980 and in 1984-
1988, when 2.6 and 1.8 per 1,000 live births respectively, were labelled as moderate to severe
HIE (Hull and Dodd 1992).

Relative risk and attributable risk: intrapartum stillbirth
Based on the Finnish study by Erkkola et al. (1984), asphyxia at birth appears to be a very
important contributor to intrapartum stillbirth: in about 90% of cases, asphyxia is the ultimate
cause. Nevertheless, causes underlying asphyxia may be diverse, including for example
placental abruption, placenta praevia, cord entanglement, cord prolapse, uterine rupture, and
pre-eclampsia (Erkkola et al. 1984).
         Table 7.25 combines estimates based on the previous subsections with results for the
hypothetical cohort to calculate the RR, AR(E), and EF of intrapartum stillbirth. The table
presents no less than eight versions, or series, of assumptions. Overall, 15 to 18% of stillbirths
are assumed to occur during labour (see Table 5.3). In addition, 90 to 95% of intrapartum
stillbirths and 0.3 to 1.0% of live births are assumed to have experienced birth asphyxia. The
resultant estimates show that the RR is very sensitive to changes in the assumptions whereas
the AR(E) and EF are more consistent over the eight series. The EF is mainly affected by the
proportion of intrapartum stillbirths that are assumed to be asphyxiated. Overall, the risk of
intrapartum stillbirth in asphyxiated newborns is over 800 times the risk that non-asphyxiated

1
    Opioid-induced: induced by opiates and substances resembling opiates.




228
Table 7.25: Estimated RR, AR(E) and EF of intrapartum stillbirth in asphyxiated newborns, hypothetical cohort

total no. SB (>= 28 weeks)                          434
total no. LB                                       88,410

                                                     (1)           (2)           (3)             (4)       (5)       (6)      (7)       (8)
% of SB intrapartum                                   15            18            15             18        15        15       15        15
total no. IPSB                                        65            78            65             78        65        65       65        65
% of IPSB asphyxiated                                 90            90            95             95        90        95       90        95
total no. BA in IPSB                                  59            70            62             74        59        62       59        62

% of LB asphyxiated                                  0.5           0.5           0.5            0.5       0.3       0.3       1.0      1.0
total no. BA in LB                                   442           442           442            442       265       265       884      884
total no. IPSB + LB                                88,475        88,488        88,475          88,488    88,475    88,475    88,475   88,475
total no. of BA in IPSB + LB                        501           512           504             516       324       327       943      946
total % BA in IPSB + LB                             0.6           0.6           0.6             0.6        0.4       0.4       1.1      1.1
% of BA cases ending in IPSB                        11.7          13.7           12.3           14.4      18.1      18.9      6.2      6.5
% of non-BA cases ending in IPSB                    0.01          0.01           0.00           0.00      0.01      0.00      0.01     0.00
RR                                                1,581.5       1,545.4        3,317.1         3,237.6   2,450.0   5,120.6   835.7    1,758.1
AR(E)                                              0.999         0.999          1.000           1.000     1.000     1.000    0.999     0.999
EF                                                 0.899         0.899          0.950           0.950     0.900     0.950    0.899     0.949
Notes: SB - stillbirth; LB - live birth; IPSB - intrapartum stillbirth; BA - birth asphyxia.
Based on the hypothetical cohort.
EARLY LIFE CHANGES

newborns run (i.e. RR > 800). Among asphyxiated newborns, the percentage of intrapartum
stillbirths that are attributable to the birth asphyxia is virtually 100% (i.e. AR(E) = 0.999-
1.000). Within the total population, the percentage of intrapartum stillbirths that can be
attributed to birth asphyxia is 90 to 95% (i.e. EF = 0.899-0.950).

7.5.2 NEONATAL DATA

Neonatal death
Figures on the relationship between birth asphyxia and neonatal mortality are scarce. Badawi
et al. (1998) studied term infants with newborn encephalopathy2 in Perth, Western Australia,
between 1993-1995. Neonatal case fatality was 9.1%. In an article on neonatology,
Vermeylen et al. (1998) concluded that mortality had decreased over the previous decade.
According to them, mortality related to asphyxia had diminished from 21 to 12%.
Furthermore, mortality from respiratory failure of prematurity had decreased from 22 to 12%.
However, it is unclear whether these figures refer to neonatal or in-hospital mortality.
        Additional data are available from developing countries. In a university hospital in
Libya between 1983-1984, Mir et al. (1989) studied neonates with birth asphyxia, defined as a
need for positive pressure ventilation for > 1 minute before sustained respiration occurred or
an Apgar score < 5 at 1 minute. They found that 28.2% of neonates with BA died compared to
only 1.1% in the non-asphyxiated group (i.e. RR = 28.2 / 1.1 = 26). However, it should be
noted that these figures are somewhat dated and moreover, that the incidence of neonatal
death is, in general, higher in developing countries than in EME countries.
        Information on the total proportion of neonatal deaths that suffered from BA is also
scarce but some figures are available on BA as a cause of death. Table 2.6a has already
presented data on the proportion of neonatal deaths that are caused by hypoxia and birth
asphyxia, as classified by the ICD-9, for several countries in the EME region. These data were
derived from publications by the WHO (World Health Statistics Annual, various years).
According to these figures, hypoxia and birth asphyxia cause 20.6 to 28.4% of neonatal deaths
in the EME region, but 46.4% in Italy. In early neonatal deaths, the corresponding
percentages are somewhat higher. Additional data from the same source show that hypoxia
and birth asphyxia cause 21.1% of early neonatal deaths in the USA, 22.4% in Spain, 24.3%
Sweden, 29.9% in the Netherlands, 33.3% in Japan, and 50.7% in Italy. It is uncertain why
the proportions in Italy are so high as compared to other EME countries, but it may be related
to differences in diagnosis and choice for underlying cause of death. In developing countries,
birth asphyxia is believed to account for 20-50% of neonatal deaths (Shibuya and Murray
1998b).
        Finally, Finan et al. (1999) applied Wigglesworth’s classification of causes to neonatal
deaths among Irish live births (> 500 g) between 1991 and 1996. Using this classification,
only 4% were due to asphyxia. Yet, the proportion due to congenital anomalies was
comparatively high: 51% (Finan et al. 1999).


2
    Encephalopathy: any disorder of the brain (Stedman’s Medical Dictionary 1995).


230
                                                      CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA


Table 7.26: Estimated RR, AR(E) and EF of neonatal death in asphyxiated live births,
hypothetical cohort

    Input                                                             Results
        B             D           pB1            I1         pD1          RR          AR(E)           EF           check*
(1) 88,410           383         0.003        0.100           -           24.7        0.960        0.066       pD1 = 0.069
    88,410           383         0.003          -           0.300        142.4        0.993        0.298         I1 = 0.433
    88,410           383           -          0.100         0.300         32.5        0.969        0.291       pB1 = 0.013
(2) 88,410           383         0.005        0.100           -           26.0        0.961        0.111       pD1 = 0.115
    88,410           383         0.005          -           0.300         85.3        0.988        0.296         I1 = 0.260
    88,410           383           -          0.100         0.300         32.5        0.969        0.291       pB1 = 0.013

(3) 88,410           383         0.010        0.100           -           29.7        0.966        0.223       pD1 = 0.231
    88,410           383         0.010          -           0.300         42.4        0.976        0.293         I1 = 0.130
    88,410           383           -          0.100         0.300         32.5        0.969        0.291       pB1 = 0.013
(4) 88,410           383         0.010        0.120           -           37.9        0.974        0.270       pD1 = 0.277
    88,410           383         0.010          -           0.240         31.3        0.968        0.232         I1 = 0.104
    88,410           383           -          0.120         0.240         36.1        0.972        0.233       pB1 = 0.009
Notes: B - total no. of live births; D - total no. of neonatal deaths; pB1 - prevalence/inc. of BA among live births;
I1 - incidence of neonatal deaths among asphyxiated live births; pD1 - prevalence of BA in neonatal deaths.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.



Relative risk and attributable risk: neonatal death
In the hypothetical cohort, 88,410 live births take place of whom 299 die during the early
neonatal period and in total, 383 during the entire neonatal period. Table 7.26 combines the
hypothetical cohort with four series of assumptions on (1) the prevalence of birth asphyxia in
live births (pB1), (2) the incidence of neonatal death in BA-affected live births (I1), and/or (3)
the prevalence of birth asphyxia in neonatal deaths (pD1). If we assume that the values range
(1) from 0.3 to 1.0%, (2) from 10 to 12%, and (3) from 24 to 30%, then the resultant RRs
range between 25 and 142, with the lowest estimate comparable to that found in Libya by Mir
et al. (1989) (i.e. RR = 26). The percentage of deaths in affected neonates that can be
attributed to the birth asphyxia, AR(E), varies between 96.0 and 99.3%. The estimated EFs of
neonatal death range between 6.6 and 29.8%. However, in the final column of Table 7.26,
recalculation of the measures missing from the input data indicates inconsistencies within the
series of assumptions. Only the third and fourth series of assumptions appear to be plausible.
        Table 7.27 presents assumptions and estimates for early neonatal death. In this table,
the assumptions are as follows: (1) pB1 is 0.5 to 1.0%, (2) I1 is 9%, and/or (3) pD1 is 30%. The
estimated RRs of early neonatal death range between 31 and 85, the AR(E) is between 96.7
and 98.8%, and the estimates of EF range from 12.9 to 29.6%. As in Table 7.26, the series of
assumptions in which the percentage of asphyxiated cases in live births is set at 1% is the
most plausible.


                                                                                                                           231
EARLY LIFE CHANGES
Table 7.27: Estimated RR, AR(E) and EF of early neonatal death in asphyxiated live births,
hypothetical cohort

      Input                                                           Results
          B           D           pB1            I1         pD1          RR          AR(E)           EF           check*
 (1) 88,410          299         0.005        0.090           -           30.5        0.967        0.129      pD1 = 0.133
     88,410          299         0.005          -           0.300         85.3        0.988        0.296        I1 = 0.203
     88,410          299           -          0.090         0.300         37.6        0.973        0.292      pB1 = 0.011
 (2) 88,410          299         0.010        0.090           -           35.9        0.972        0.259      pD1 = 0.266
     88,410          299         0.010          -           0.300         42.4        0.976        0.293        I1 = 0.101
     88,410          299           -          0.090         0.300         37.6        0.973        0.292      pB1 = 0.011
Notes: B - total no. of live births; D - total no. of early neonatal deaths; pB1 - prevalence/inc? of BA among live births;
I1 - incidence of early neonatal deaths among asphyxiated live births; pD1 - prevalence of BA in early neonatal deaths.
*Check: calculation of either pB1, I1, or pD1 on the basis of the input assumptions.
See Chapter 4 for equations. Based on the hypothetical cohort.




7.6 Overlap and associations
Clearly, the category ‘LBW’ overlaps with both ‘preterm’ and ‘IUGR/SGA’. However, the
distribution of IUGR/SGA and preterm birth within the group of low birth weight newborns
differs by region. In regions where the percentage of LBW births is less than 10%, such as
most countries in the EME region, preterm birth makes up the major component. When the
percentage of LBW births is over 10%, this is almost exclusively due to an increase in
IUGR/SGA infants whereas the proportion of preterm births remains almost unchanged at 5-
7% (Villar and Belizan 1982).
        The following section, Section 7.6.1, discusses observed associations between LBW,
preterm birth, and IUGR/SGA. Subsequently, Sections 7.6.2 and 7.6.3 relate these three
pregnancy/birth outcomes to birth asphyxia and congenital anomalies respectively.

7.6.1 LBW,        PRETERM BIRTH, AND IUGR/SGA


LBW and preterm birth
Among singleton births in Sweden in 1973-1981, 57.9% of LBW cases were preterm (< 37
weeks) and 39.5% of preterm cases were LBW (< 2,500 g) (Källén 1988). Similar findings
occur with the 1988 US birth cohort: 54.8% of LBW live births were preterm and 37.8% of
preterm live births were LBW (based on data from NCHS 1995). In California, these figures
were 57.3 and 30.9% respectively. However, more recent data from 1985-1994 on live births
in Scotland suggest a much higher proportion of LBW among preterm cases, with 75.2% of
preterm deliveries (24 through 36 weeks) found to be LBW (Magowan et al. 1998). This may
be due to changes in distribution over the lower birth weight categories. Goldenberg et al.
(1985) studied 12,818 live births between 1979 and 1981 at hospitals in Birmingham,
Alabama (USA). They found that below 1,750 g nearly all infants were preterm, while in the
higher birth weight groups, the proportion of preterm infants decreased with rising birth


232
                                         CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

weight. In general, the percentage of births or live births that are both LBW and preterm
appears to be around 2 to 4%. The corresponding figures were 2.9% in Sweden in 1973-1981
(Källén 1988), 2.8% in California in 1988 (based on data from NCHS 1995), and 4.1% in
Scotland in 1985-1994 (Magowan et al. 1998). Moreover, Villar and Belizan (1982) reviewed
11 populations, from the 1960s and 1970s, in ten developed countries (including Hungary)
and found that, on average, the percentage of preterm LBW was 3.3%.
       According to US data from 1959-1965, neonatal survival is lowest in the group of
infants who are both LBW and preterm, and highest among infants with a birth weight ≥
2,500 g and a gestational age ≥ 37 weeks. Further, preterm neonates with a birth weight of
more than 2,500 g seem to have a survival advantage over non-preterm LBW neonates (based
on data from Hardy et al. 1979). According to results by Copper et al. (1993), before 29
weeks, gestational age is a stronger predictor of neonatal mortality than birth weight.
However, from 29 to 36 weeks, birth weight is a highly significant predictor of mortality
while the effect of gestational age diminishes. Copper et al. (1993) collected their data in
1982-1986 from several university hospitals in the USA. In addition to the results above, they
found for each gestational age group that heavier infants suffered less neonatal mortality.
Furthermore, within each birth weight category, mortality decreased with advancing
gestational age.

LBW and IUGR/SGA
        In terms of IUGR/SGA, it has been stated that about one-third of live births weighing
less than 2,500 g are growth retarded (Mann et al. 1981 cited by Goldenberg et al. 1985). In
Alabama, 20 to 30% of live births in weight categories 500-2,000 g were SGA as based on the
10th percentile by Brenner et al. (1976) (Goldenberg et al. 1985). Among live births weighing
2,001 to 2,750 g, almost 50% were classified as growth retarded. Survival rates until
discharge were higher among growth-retarded infants than with non-growth-retarded infants,
in all LBW groups (Goldenberg et al. 1985). Based on a review of studies, Ashworth (1998)
estimated the relative risks of neonatal death for IUGR infants by birth weight group. The
estimates were based on the assumptions that birthweight-specific data for countries with a
high prevalence of LBW and birthweight-specific data for term infants both reflect risks in
IUGR births. The weight group of 2,500-2,999 g was taken as the reference category. The
estimated RRs of neonatal death for IUGR infants were as follows: 18.0 for those weighing
1,500-1,999 at birth, 4.0 for 2,000-2,499 g, 0.4 for 3,000-3,499 g, and 0.3 for those ≥ 3,500 g
(Ashworth 1998). Within the group of IUGR/SGA neonates, survival is thus affected by birth
weight.

IUGR/SGA and preterm birth
IUGR and SGA have been associated with preterm delivery in both live and stillbirths
(Hartikainen-Sorri and Sorri 1989; Ott 1995; Chen et al. 1996; Sherer et al. 1997; Clausson et
al. 1998; Gardosi et al. 1998; Smith et al. 1998). Clausson et al. (1998) studied singleton live
births to nulliparous women in Sweden in 1992-1993. Among SGA infants – defined as birth
weight > 2 standard deviations below the Swedish birthweight curve by Marsál et al. (1996) –
8.1% were ≤ 32 weeks gestation and another 11.8% were 33 to 36 gestational weeks


                                                                                            233
EARLY LIFE CHANGES

(Clausson et al. 1998). The corresponding figures in all singleton live births were only 1.2 and
5.2% respectively. Conversely, 24.7% of very preterm births (defined as ≤ 32 weeks) and
8.7% of moderately preterm births (33-36 weeks) were classified as SGA. Among term births,
only 3.3% were small-for-gestational-age (Clausson et al. 1998). In Western Australia, 31%
of very preterm live births (defined as < 33 weeks gestation) during 1990 to 1991 were
classified as SGA on the basis of the 10th percentile using standard curves by Blair and
Stanley (1985) (Hagan et al. 1996). In Finland, Hartikainen-Sorri and Sorri (1989) compared
singleton preterm births with single full-term newborns of the same sex. Among the preterm
newborns, 26% had a low birth weight for gestational age – based on Finnish reference values
by Rantakallio (1973) – compared to only 7% of the full-term controls. However, the
difference between preterm and term singleton infants with regard to weight for gestational
age was only minor in the study by McIntire et al. (1999). The researchers studied singleton
live births at a hospital in Dallas (USA) between 1988 and 1996. On the basis of the 3rd
percentile, 4.1% of preterm and 3.9% of term live births were SGA. The study was hospital-
based and did not include cases with congenital malformations.
        According to Piper et al. (1996), “a common belief of obstetricians and neonatologists
is that growth-retarded foetuses are ‘stressed’ by their intrauterine environment and that this
stress accelerates their maturation so that growth-retarded preterm infants have fewer
complications of prematurity than their appropriately grown peers” (p.169). However, the
authors had found no survival advantage in the SGA preterm group over the AGA preterm
group in a study of singleton preterm live and stillbirths delivered during 1970 to 1985 in a
teaching hospital in San Antonia (USA). Pregnancies complicated by diabetes were excluded
from the sample. The perinatal mortality rate, foetal death rate, and neonatal mortality rate
were significantly higher for the SGA group than the AGA newborns (Piper et al. 1996).
Findings from other researchers (Koops et al. 1982; Palo and Erkkola 1993; Clausson et al.
1998; Chard et al. 2001) are similar. Nevertheless, in the data used by McIntire et al. (1999),
neonatal mortality in preterm SGA infants is only slightly higher than mortality among those
in the 4th through 75th birth-weight percentiles (6.9% versus 6.3%). Further, in Western
Australia, Hagan et al. (1996) did find that mortality in very preterm infants (defined as < 33
weeks gestation) was higher in the AGA/LGA group than the SGA group (13% versus 12%).

7.6.2 LBW,    PRETERM BIRTH, IUGR/SGA AND BA

Birth asphyxia has been associated with low birth weight, IUGR/SGA, and preterm birth, but
also with LGA and postterm birth (Mir et al. 1989; Xu 1992; Shibuya and Murray 1998b;
Heinonen and Saarikoski 2001). In addition, Badawi et al. (1998) conclude that risk factors
for newborn encephalopathy include intrauterine growth restriction and postmaturity.
However, Thornberg et al. (1995) found in their study in Sweden that SGA infants were
overrepresented in the birth asphyxia group (i.e. term, Apgar score < 7 at 5 min., excluding
congenital malformations, infections, and opioid-induced respiratory depression) but not in
the birth asphyxia-HIE group. Results from a study by Low et al. (1990 cited by Shibuya and
Murray 1998b) in the United States show that birth asphyxia is associated with low birth




234
                                        CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

weight. In 1978-1986, the frequency of BA was only 22 per 1,000 births in infants weighing
over 2,000 g at birth compared to 61 per 1,000 in infants below 2,000 g.
        Comprehensive information is available from a hospital-based study in Benghazi,
Libya (not part of the EME region) between 1983-1984, by Mir et al. (1989). The researchers
defined asphyxia as a need for positive pressure ventilation for > 1 minute before sustained
respiration occurred or an Apgar score < 5 at 1 minute. Overall, asphyxia occurred in 2.8% of
live births. A strong association was found between asphyxia and birth weight with asphyxia
occurring in 68% of live births below 1,000 g but only in 1.2% of births between 3,000 to
4,000 g. The frequency of asphyxia in SGA live births was 12.8% and in LGA live births
18.7%. Both figures were significantly higher (p < 0.0001) than for AGA infants (2.4%).
Likewise, the frequency of asphyxia was significantly higher (p < 0.0001) among preterm and
postterm live births (24.5% and 16.7% respectively) when compared to term infants (1.7%)
(Mir et al. 1989).
        With regard to survival, Mir et al. (1989) compared the asphyxiated cases to infants
without any evidence of birth asphyxia (i.e. the control group). For both groups, the authors
presented data on neonatal mortality (presumably, early neonatal mortality) by birth weight
and gestational age. The mortality risk in asphyxiated cases exceeded the risk among the
controls from birth weight category 1,001-1,500 g upwards. As birth weight increased, the
relative risk of death increased from 1.9 among the 1,001-1,500 g group to 16.3 among those
over 4,000 g (based on data from Mir et al. 1989). In live births below 31 weeks of gestation,
the mortality risk in asphyxiated infants more or less equalled the risk among the control
group. Subsequently, the RR increased with advancing gestational age up to 27.3 for those of
38-41 gestational weeks. However, for live-born infants of 42 weeks and over, the RR
declined back to 8.3. In addition, Mir et al. (1989) found that SGA was significantly
associated with an increased risk of death among asphyxiated newborns.

7.6.3 LBW,    PRETERM BIRTH, IUGR/SGA AND CONGENITAL ANOMALIES

The presence of congenital anomalies has been associated with lower birth weight, with
preterm birth and lower gestational age, and with IUGR/SGA (Yerushalmy 1970; Tafari
1980; Khoury et al. 1988; Hartikainen-Sorri and Sorri 1989; Ott 1995; Chen et al. 1996; Piper
et al. 1996; Kalaidzhieva et al. 1997; Milerad et al. 1997; Bhat and Babu 1998; Clausson et al.
1999; Linhart et al. 2000). Below, are some examples illustrating the various findings.
        In a hospital in Sofia, Bulgaria (not part of the EME region), between 1991-1995, the
frequency of congenital malformations among live-born VLBW neonates was about six times
higher than among newborns weighing over 1,500 g (Kalaidzhieva et al. 1997). In San
Antonia (USA), Piper et al. (1996) found congenital anomalies in as many as 13-14% of
singleton, preterm deliveries to nondiabetic mothers. Similarly, Hartikainen-Sorri and Sorri
(1989) found 14% of preterm newborns in university hospitals in Finland, in 1982, to be
anomalous versus 4% of term newborns. Linhart et al. (2000) studied preterm deliveries in a
university medical centre in Israel (not part of the EME region) between 1990 and 1995. The
proportion of congenital anomalies in preterm newborns was significantly higher than in those
born at term (8.7% vs. 2.1%). In addition, the authors found that even among preterm births,



                                                                                           235
EARLY LIFE CHANGES

the gestational age at delivery was significantly lower in the congenital anomaly group than in
those without anomalies. Ott (1995) examined data from a series of high-risk obstetric
patients referred to a perinatal centre during 1990 and 1991. Among newborns classified as
SGA – based on 10th percentile from combined curves of Hadlock et al. (1991) and Ott (1993)
– 9.4% had significant anomalies compared to 4.2% of non-SGA newborns (p < 0.001). In
addition, 35.3% of newborns with anomalies, compared to only 3.5% of non-anomalous
infants, were classified as SGA (p < 0.0001) (Ott 1995)
       IUGR has been frequently reported among infants with chromosomal abnormalities,
neural tube defects, certain types of congenital heart disease, and many dysmorphic
syndromes. Khoury et al. (1988) analysed data from the Metropolitan Atlanta Congenital
Defects Program on infants with major malformations born in 1970-1984. Overall, 22.3% of
anomalous infants were SGA as defined by the 10th percentile of a birth weight for gestational
age curve. The researchers concluded that almost all types of major congenital anomalies are
associated with SGA, although the highest proportions of SGA in anomalous infants were
seen for trisomy 18 (83.7%) and anencephaly (73.3%). Other anomalies that were strongly
associated with SGA included spina bifida (32.0%), microcephaly (52.4%), reduction of the
limbs (36.9%), and several chromosomal abnormalities (28.2-66.7%) (Khoury et al. 1988).
Clearly, the absence of certain organs or limbs is likely to reduce birth weight. However, there
may also be other mechanisms that govern the relationship between IUGR/SGA and the
presence of congenital anomalies. First, IUGR may be the result of, or a reaction to, the
presence of an anomaly. Conversely, IUGR may play a role in the origin of abnormalities or
may predispose the foetus to anomalies. Finally, it is also possible that IUGR and anomalies
are affected by the same underlying factors and that they are the result of a common
mechanism (Källén 1988; Khoury et al. 1988).
       As a cause of death, congenital anomalies are less significant among lower birth
weights and at lower gestational ages than with non-LBW and non-preterm neonates
(Goldenberg et al. 1983; Philip 1995; Magowan et al. 1998; CCOPMM 2000). Nevertheless,
the presence of congenital anomalies has been found to increase mortality among preterm and
SGA infants (Kramer et al. 1990; Bottino et al. 1991; Piper et al. 1996; Clausson et al. 1999;
Linhart et al. 2000). Kalter (1991) reviewed hospital-based studies in Europe and North
America (USA and Canada) published from 1950 onwards. He concluded that birth weight is
associated with congenital anomalies but with opposite trends for stillbirth and early neonatal
death. In stillbirths, the prevalence of congenital anomalies was significantly greater among
those below 2,500 g while in early neonatal deaths, higher prevalence was found among those
≥ 2,500 g (Kalter 1991).

7.7 Summary and discussion
The present chapter has assessed the relative importance of low birth weight, intra-uterine
growth retardation/small-for-gestational-age (IUGR/SGA), preterm birth, and birth asphyxia
as risk factors for foetal loss and neonatal death in the EME region, at both the individual and
the population level. In order to obtain the desired estimates, assumptions concerning the
prevalence and incidence were derived from literature and combined with the hypothetical


236
                                                   CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

cohort that was constructed in Chapter 5. Hence, the chapter has presented a thorough and
elaborate overview of previously published study results and estimates on low birth weight,
IUGR/SGA, preterm birth, and birth asphyxia in the EME region. The results of the various
studies were tabulated and briefly discussed but comparison of results was complicated by
differences between the studies, e.g. in study design and year of study.
        The assumptions that were derived from the literature, as well as the newly calculated
estimates of relative importance, are summarised and discussed in the sections below. The
first part deals with the effects of the risk factors at the individual level and the following
section discusses their influence on the population. As in Chapter 6, it should be noted that the
results represent educated guesses and are estimates.

7.7.1 INDIVIDUAL LEVEL
The relative importance at the level of the individual has been established in terms of risk (cf.
incidence) of loss or death for affected individuals, and in the relative risk of loss or death
(see also Chapter 4). Table 7.28 summarises the assumed and estimated incidence proportions
of loss (if applicable) and death by risk factor. The figures for spontaneous abortion in
growth-retarded foetuses are inconclusive due to the wide range (23-94%). The incidence of
stillbirth in this group (2-4%) is comparable to the incidence proportions found among
foetuses with congenital anomalies (see Table 6.25). The incidence of neonatal death is 4-7%
in LBW babies (< 2,500 g), 3-4% in preterm infants (< 37 weeks), 0.6-1.4% in IUGR/SGA
cases, and 10-12% in asphyxiated newborns. Within the five main risk categories (i.e.
anomalies, LBW, IUGR/SGA, preterm, and birth asphyxia), the ranking by highest incidence
of neonatal death is: (1) birth asphyxia, (2) LBW and congenital anomalies, (3) preterm birth,
and (4) IUGR/SGA. However, the majority of live-born infants who are affected by one of the
risk factors in Tables 6.25 and 7.28 do survive the neonatal period, with neural tube defects
being the only exception. During gestation and birth, survival chances are much less
favourable for anomalous and growth-retarded foetuses.

     Table 7.28: Estimated or assumed incidence of spontaneous abortion, stillbirth,
     and neonatal death, by risk factor, EME region
                                           Incidence proportion (%)
                                                  SA                SB             ENND               NND
     Birth weight < 2,500 g                       -                 -                3-5               4-7
     Birth weight < 2,000 g                       -                 -               6-14              9-16
     Birth weight < 1,500 g                       -                 -               13-25             19-30
     Preterm birth (< 37 wks)                     -                 -                 2                3-4
     Preterm birth (< 32 wks)                     -                 -                12                17
     IUGR/SGA                                   23-94              2-4               NA              0.6-1.4
     Birth asphyxia                               -              6-19 [a]             9              10-12
     Notes: SA - spontaneous abortion; SB - stillbirth (>= 28 wks); ENND - early neonatal death; NND - neonatal
     death; NA - not available.
     [a] Intrapartum stillbirth
     Based on studies discussed in the literature and/or assumptions in combination with the hypothetical cohort.



                                                                                                                    237
EARLY LIFE CHANGES
      Table 7.29: Estimated relative risks of spontaneous abortion, stillbirth, and neonatal
      death, by risk factor, EME region
                                            Relative risk (RR)
                                                   SA                SB             ENND               NND
      Birth weight < 2,500 g                        -                -               26-76             21-59
      Birth weight < 2,000 g                        -                -              51-154            44-107
      Birth weight < 1,500 g                        -                -              94-209            69-160
      Preterm birth (< 37 wks)                      -                -                6-31             17-31
      Preterm birth (< 32 wks)                      -                -               68-75             83-86
      IUGR/SGA                                    2-29             7-44               NA               1-18
      Birth asphyxia                                -            > 800 [a]          31-85             25-142
      Notes: SA - spontaneous abortion; SB - stillbirth (>= 28 wks); ENND - early neonatal death; NND - neonatal
      death; NA - not available.
      [a] Intrapartum stillbirth
      Based on studies discussed in the literature and/or assumptions in combination with the hypothetical cohort.


        How do the incidence proportions in Table 7.28 compare to incidence proportions of
loss and death among foetuses and neonates without these risk factors? Table 7.29 restates the
estimates of relative risk. The relative risk (RR) compares the incidence in affected
individuals to the incidence in individuals who are not affected. It indicates how many times
more likely persons with the risk factor will experience a particular outcome as compared to
persons without the risk factor. For example, preterm infants are 17 to 31 times as likely as
non-preterm infants to experience neonatal death. Most of the relative risks in Table 7.29 have
a value (well) over 1.0, indicating that foetuses and neonates affected by these risk factors
indeed experience elevated risks of loss and death as compared to their non-affected
counterparts. The figures for spontaneous abortion and stillbirth in growth-retarded foetuses
seem meaningless because of their wide ranges (2-29 and 7-44 respectively), but it is clear
that the relative risk of intrapartum stillbirth in children affected by birth asphyxia is
extremely high. The relative risk of neonatal death is 21-59 for LBW babies, 17-31 for
preterm born children, 1-18 for IUGR/SGA neonates, and 25-142 for asphyxiated newborns.
Within the five main categories (including anomalies, see Table 6.26), the ranking in terms of
highest relative risk of neonatal death is: (1) birth asphyxia, (2) LBW, (3) preterm birth and
congenital anomalies, and (4) IUGR/SGA. In comparison to the previous ranking related to
simple incidence proportions of death, congenital anomalies have dropped to a lower position.
This implies that the incidence of death among non-anomalous neonates is comparatively
high. Furthermore, it is interesting to note that low birth weight is a stronger risk factor for
neonatal death than preterm birth or IUGR/SGA.
        In conclusion, the relative risks in Tables 6.26 and 7.29 do indicate elevated risks of
loss/death for foetuses and neonates affected by one of the risk factors as compared to those
non-affected. Birth asphyxia is an important risk factor, in particular in relation to stillbirth. In
addition, low birth weight, and especially the lowest birth weights, are significant risk factors
for neonatal death. The least important is IUGR/SGA. However, its significance as a risk
factor is in part dependent on its definition and on the choice of a reference growth curve.


238
                                                    CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA

7.7.2 POPULATION LEVEL
The relative importance at the level of the population has been established in terms of
frequency (cf. prevalence) of the risk factor within the total population, and in the proportion
of losses or deaths in a population that can be attributed to the risk factor (see also Chapter 4).
Table 7.30 lists the prevalences of the selected pregnancy/birth outcomes among losses,
deaths, and births. Overall, 5-10% of live births are LBW, 3-10% are IUGR/SGA, 6-10% are
preterm, and only 0.3-1.0% are asphyxiated at birth. All the adverse pregnancy/birth
outcomes are more frequently observed among stillbirths and neonatal deaths. Among
stillbirths, the prevalence proportions exceed 30%, but the causal relationships are not
straightforward and occur in both directions. Antepartum death affects weight and gestational
age at birth, while low weight at birth, preterm birth, and IUGR/SGA may be risk factors for
intrapartum stillbirth. Birth asphyxia is observed in as many as 90-95% of intrapartum
stillbirths. Among early neonatal deaths and all neonatal deaths, the prevalence of the risk
factors generally exceeds 20%. Including anomalies (see Table 6.27), the ranking of the main
risk categories by highest prevalence among neonatal deaths is: (1) preterm birth and LBW,
(2) congenital anomalies, and (3) birth asphyxia and IUGR/SGA.
         Table 7.31 summarises the estimates of attributable risk, both among the exposed and
in the total population, by risk factor. The attributable risk among the exposed, AR(E),
indicates the proportion of losses/deaths in affected foetuses and neonates that can be
attributed to the risk factor. For example, 94 to 97% of neonatal deaths among preterm infants
are attributable to the preterm birth. The table shows that the vast majority of losses and
deaths in the population affected by one of the risk factors can be attributed to that risk factor.
Indeed, 95% or more of early neonatal deaths and all neonatal deaths in the affected
populations can be attributed to the risk factors in question. Chapter 6 showed that this was
also the case for congenital anomalies (see Table 6.28). However, an exception is IUGR/SGA
which contributes to only 31 to 89% of neonatal deaths in the SGA population. Nevertheless,

Table 7.30: Estimated or assumed prevalence of risk factors in spontaneous abortions,
stillbirths, live births, all births, and neonatal deaths, by type, EME region
                                Prevalence proportion (%)
                                     SA             SB             LB          all births       ENND            NND
Birth weight < 2,500 g                -          66 [a]           5-10           5-10             70            65
Birth weight < 2,000 g                -         53-55 [a]          2-3            2-3             60            60
Birth weight < 1,500 g                -         44-47 [a]        0.9-1.2        0.9-1.2           50            50
Preterm birth (< 37 wks)              -            60             6-10           6-10             70           65-70
Preterm birth (< 32 wks)              -            32            0.8-1.4        0.8-1.4           52            54
IUGR/SGA                             24            45             3-10           3-10             NA           13-31
Birth asphyxia                        -         90-95 [b]        0.3-1.0        0.3-1.0           30           24-30
Notes: SA - spontaneous abortion; SB - stillbirth (>= 28 wks); LB - live birth; ENND - early neonatal death; NND - neonatal
death; NA - not available.
[a] Based on mean and median in Table 7.1; [b] Intrapartum stillbirth.
Based on studies discussed in the literature and/or assumptions in combination with the hypothetical cohort.




                                                                                                                      239
EARLY LIFE CHANGES

the majority of spontaneous abortions (58-97%) and stillbirths (86-98%) in the IUGR/SGA
population seem to be due to growth retardation.
        A more interesting measure from a population point of view is the etiologic fraction
(EF). The EF indicates the proportion of losses/deaths in the total population that can be
attributed to the risk factor and is calculated on the basis of prevalence in the population (see
Table 7.30) and relative risk of loss/death (see Table 7.29). For example, 59-68% of neonatal
deaths in the total population can be attributed to preterm birth. With regard to spontaneous
abortion and stillbirth, the figures calculated for IUGR/SGA have wide ranges (14-71% and
39-81% respectively). Nevertheless, in comparison to congenital anomalies (see Table 6.28),
a relatively large proportion of stillbirths are attributable to IUGR/SGA. This could be a result
of data problems related to missed abortion and delayed expulsion after death. Birth asphyxia
is clearly extremely important in relation to intrapartum stillbirth in the population. Turning
to neonatal death, 62-85% can be attributed to LBW, 59-68% to preterm birth, 4-24% to
IUGR/SGA, and 7-30% to birth asphyxia.
        Table 7.31 shows that the most important contributors to neonatal death in the EME
population are LBW (< 2,500 g), preterm birth (< 37 weeks), and birth weight < 2,000 g. The
high rankings of LBW and preterm birth are explained by a high prevalence among live births
whereas the strong population impact of birth weight < 2,000 g is due to the combination of
both a high relative risk and a relatively high prevalence at live birth. Overall, low weight at
birth (< 2,500 g, < 2,000 g, and < 1,500 g) and preterm birth (< 37 and < 32) contribute to
over half of neonatal deaths in the EME population. Including congenital anomalies (see
Table 6.28), the ranking of the five main risk categories by highest etiologic fraction of
neonatal death is: (1) LBW, (2) preterm birth, (3) congenital anomalies, and (4) birth asphyxia
and IUGR/SGA. Although the prevalences of LBW and preterm birth in live births are almost
equal, their difference in EF ranking can be explained by a higher relative risk of death among
LBW neonates. The low rankings of IUGR/SGA and birth asphyxia are explained by the low
relative risk and the low prevalence in the population of live births respectively.
        In conclusion, the etiologic fractions in Tables 6.28 and 7.31 demonstrate that large
proportions of neonatal deaths, spontaneous abortions, and possibly stillbirths in the EME
population can be attributed to the selected risk factors.




240
                                              CHAPTER 7: LOW BIRTH WEIGHT AND BIRTH ASPHYXIA



Table 7.31: Estimated attributable risks of spontaneous abortion, stillbirth, and
neonatal death, by risk factor, EME region
AR(E): Attributable risk among the 'exposed' (%)
                                             SA                SB              ENND              NND
Birth weight < 2,500 g                       -                 -               96-99             95-98
Birth weight < 2,000 g                       -                 -               98-99              99
Birth weight < 1,500 g                       -                 -              99-100              99
Preterm birth (< 37 wks)                     -                 -               83-97             94-97
Preterm birth (< 32 wks)                     -                 -                99                99
IUGR/SGA                                   58-97             86-98              NA               31-89
Birth asphyxia                               -              100 [a]           97-99              96-99


EF: Etiologic fraction (%), in total population
                                             SA                SB              ENND              NND
Birth weight < 2,500 g                       -                 -               61-79             62-85
Birth weight < 2,000 g                       -                 -               52-73             55-65
Birth weight < 1,500 g                       -                 -               45-64             46-58
Preterm birth (< 37 wks)                     -                 -               26-68             59-68
Preterm birth (< 32 wks)                     -                 -               49-51             53-54
IUGR/SGA                                   14-71            39-81               NA               4-24
Birth asphyxia                               -             90-95 [a]           13-30             7-30
Notes: SA - spontaneous abortion; SB - stillbirth (>= 28 wks); ENND - early neonatal death; NND - neonatal
death; NA - not available.
[a] Intrapartum stillbirth.
Based on studies discussed in the literature and/or assumptions in combination with the hypothetical cohort.




                                                                                                               241

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:49
posted:6/15/2011
language:English
pages:48