Conclusions and Discussion

Document Sample
Conclusions and Discussion Powered By Docstoc
					                                 Chapter 10

                   Conclusions and discussion

10.1 Introduction
Risk factors and processes during prenatal life, birth, and the neonatal period are becoming
increasingly important predictors of morbidity and mortality patterns in populations. In order
to improve population forecasting, in particular in terms of infant mortality, information is
thus required about these risk factors and processes, and their effects on health and survival.
The present study has identified risk factors and mechanisms and gained insights into their
demographic consequences, thereby contributing to the improvement of infant mortality
forecasting. The focus of the study was on factors that are causally related to the outcomes of
interest, as opposed to statistical associations and ‘explanations’ based on demographic and
socioeconomic characteristics. Knowledge and understanding of causal processes are essential
if one wants to understand and predict events, and if one wants to intervene and change the
outcomes, including at the population level. According to Manton et al. (1993), “current
forecasting procedures (…) do not directly reflect physiological processes at the individual
level or the mixture of individuals in a cohort” (p.25). However, if models do not involve a
structure that is isomorphic to the biological processes in question, their results are limited
even when they appear to be statistically sophisticated (Manton et al. 1993).
        A shift of focus to causal factors and mechanisms of morbidity and mortality requires
additional data sources to the ones that are traditionally used in demography. Therefore, the
present study aimed to expand and complement demographic data sources by using secondary
data from the health sciences and hospital data.
        The risk factors and outcomes were studied within the framework of an epidemiologic
transition. The original theory of the epidemiologic transition by Omran (1971), as well as
most extensions to it, bear no or hardly any reference to pregnancy, gestation, birth, and the
newborn. Only recently has Van der Veen (2001) reviewed contemporary changes in infant
mortality during the transition in low-mortality countries. The present study set out to gain
insights into changes during earlier stages of the transition, and focused on a case region that
has not yet reached the later stages of the epidemiologic transition as observed in most
developed countries.
        The research question in the present study was divided into two parts (see Chapter 1).
The first question was directed at determining the most important causal factors of loss and
death in the developed world, i.e. regions in the later stages of the epidemiologic transition.
The second research question focused on the significance of these factors within the
framework of an epidemiologic transition, as based on the current situation in regions in
transition that are approaching the later stages of the epidemiologic transition.

        Both parts of the study aimed to assess the relative importance of several selected risk
factors at the individual and population levels. The significance at the level of the individual
was established in terms of risk (cf. incidence) of adverse outcome (e.g. death) for persons
affected by the risk factor, and in the relative risk of adverse outcome. The relative risk (RR)
indicates how many times more likely persons with a certain risk factor will experience a
particular outcome as compared to persons without that risk factor. The relative importance at
the level of the population was established in terms of frequency (cf. prevalence) of the risk
factor within the total population, and in attributable risks. The attributable risk among the
exposed, AR(E), indicates the proportion of adverse outcomes in the exposed population that
are due to, or can be attributed to, the risk factor whereas the etiologic fraction (EF) refers to
the total population, both those exposed and those non-exposed to the risk factor. Making a
distinction between an individual level and a population level implies two types of
interventions in the underlying processes: (1) improving the survival chances of individuals
who are affected by the risk factor, and (2) lowering the prevalence of the risk factor in the

10.2 Data and methods
The study was divided into two parts that were each based on a different approach. Data were
obtained from secondary sources, mostly from literature in fields such as obstetrics and
gynaecology, neonatology, paediatrics, genetics, cardiology, epidemiology, and public health.
To illustrate the situation in regions in transition that are approaching the later stages of the
transition, a hospital survey was conducted in a case region.


The situation in the developed world was described and analysed on the basis of secondary
data, mainly obtained from literature in the health sciences. The term ‘developed world’ refers
to regions in the later stages of the epidemiologic transition, i.e. low-mortality countries. The
developed world was operationalised as the EME region, the established market economies,
(cf. Global Burden of Disease Study; see Chapter 4).
       Despite the huge amount of publications on foetus and neonate, and on pregnancy and
birth outcome in relation to survival, the desired information was not readily available. Data
turned out to be limited and fragmented (see Chapter 4). Consequently, the study adopted a
method that combined the fragmented data and assumptions with mortality figures obtained
using a hypothetical cohort. Checks and recalculations, however, indicated that certain
combinations of input data were not plausible. A further improvement of the results and an
extension to model-building would require calibration and fine-tuning.


The second part of the study focused on the situation in regions that are approaching the later
stages of the epidemiologic transition. For most less developed countries, data on health and
survival are sparse, if available at all. During recent decades, the main sources of information
on child survival in developing countries have been large-scale demographic surveys, such as

                                                  CHAPTER 10: CONCLUSIONS AND DISCUSSION

the Demographic Health Survey (DHS) and the National Family Health Survey (NFHS) in
India. However, these surveys have been found to provide insufficient information to explain
child survival levels and the mechanisms underlying child health and survival (see Boerma
1996 and Padmadas 2000). The present study tried to overcome the shortfall in data by
implementing a hospital survey in a case region. The survey data were compared to, and
complemented by, secondary data. The hospital survey was conducted at the end of 2000 in
Sri Avittom Thirunal (SAT) Hospital in Thiruvananthapuram, Kerala, South India.
       The study demonstrated the added value of hospital data and hospital surveys as a
supplementary data source in less developed regions. Information was available that could not
be obtained from the standard sources used by demographers. For some other variables, such
as birth weight and gestational age, the reliability of the data was improved. Nevertheless,
some difficulties and challenges were encountered in relation to the collection of data in a
hospital. These included: selection bias; the availability, reliability, and validity of hospital
records; differences between the disciplines of medicine and demography/statistics;
administrative practices in the hospital; and ethical concerns (see Chapter 4).
       The magnitude and direction of selection bias will depend on the type of hospital, its
patient population, and the segment of the total population of pregnant women that comes to
hospital to have their baby delivered. In the present study, the hospital in Kerala, where the
survey was conducted, was a large referral hospital in the public sector. Consequently, high-
risk and complicated cases as well as women from lower socioeconomic classes are likely to
have been overrepresented in the survey. Nevertheless, when compared to population-based
data for Kerala, no difference was observed in terms of mortality rates. The mortality rates in
our hospital survey generally fell between the lower rates from the Sample Registration
System (SRS) and the higher rates from the National Family Health Survey (NFHS-1 and/or
NFHS-2). Other studies based on hospital data from Kerala have shown mortality rates that
are generally higher than our survey found. The prevalence of low birth weight (< 2,500 g) in
the SAT Hospital (SATH) survey was comparable to other findings for Kerala.
       In the survey, data on maternal infections during pregnancy are suspected to have been
incomplete, while data on monthly income were clearly unreliable. Information that was not
available in the hospital records included height of the mother and socioeconomic and
sociodemographic factors such as education, employment, and religion. There, thus, seems to
be a gap between the traditional data sources and hospital data. While surveys such as the
NFHS and DHS have no or only limited information on health status and medical risk factors,
hospital records contain no, or unreliable, data on socioeconomic and sociodemographic
factors. The gap in available data, and between disciplines, suggests that studies in recent
decades have failed to address and explain the relationship between socioeconomic and
sociodemographic factors on the one hand and adverse health outcome, medical risk factors,
and mortality on the other.
       In the present survey, some of the information that was not in the hospital records was
obtained by posing supplementary questions directly to the women. This included the follow-
up of newborns through correspondence, to see whether they had survived the neonatal
period. Nevertheless, about one-third of infants were lost to follow-up before the end of the
neonatal period. To assure a more complete and reliable follow-up, and to avoid problems of


missing data, a future hospital survey could be supplemented by home visits. These home
visits should include the presence of medically-trained staff in order to answer questions by
the mothers and to offer a proper medical check-up to the child.
        Statistical analysis of our hospital data was complicated by the small numbers
involved, especially when considering specific health problems and complications. However,
in Kerala, similar problems also exist with the SRS (see Irudaya Rajan and Mohanachandran
1999a, 1999b) and possibly also with the NFHS data. The ideal would be a comprehensive
registration system that covers all births and deaths within the region. Without such extensive
databases, and with declining infant and neonatal mortality and a growing use of health
services, hospital records and surveys may become increasingly important data sources for

10.3 Current situation
This section describes the situation in the EME region and in Kerala in terms of their position
in the transition, stillbirth and neonatal mortality, and the relative importance of several
selected risk factors for foetal loss and/or neonatal death. The results on relative importance of
the selected risk factors are also presented in Table 10.1 (see Section 10.4). Later, Section
10.4 will compare the results for the two regions and describe the observed developments
during the transition.

10.3.1 THE EME        REGION

The EME region is already strongly progressing through what is called the fourth stage of the
epidemiologic transition (see Chapter 2). Mortality is low in this region. During the first half
of the 1990s, the stillbirth ratio was around 3 to 5.5 deaths per 1,000 live births, early neonatal
mortality around 2 to 4 deaths per 1,000 live births, and neonatal mortality around 2 to 5
deaths per 1,000 live births. The probability of death is relatively high on the first day of life,
but declines thereafter. During the first half of the 1990s, about 36 to 44% of neonatal deaths
occurred on the first day of life, with a figure of 60% in the USA as the outlier. The
percentage of neonatal deaths during the first week ranged between 71 and 82%.
        The present study assessed the relative importance of the following risk or causal
factors of spontaneous abortion and stillbirth: congenital anomalies (including the
subcategories of chromosomal aberrations, neural tube defects, and congenital heart disease),
intrauterine growth retardation/small-for-gestational-age (IUGR/SGA), and birth asphyxia
(see Chapters 6 and 7). Foetuses with congenital anomalies and/or IUGR/SGA do have
elevated risks of spontaneous abortion and stillbirth as compared to foetuses without these
risk factors (i.e. a relative risk > 1). Overall, around 35% (although possibly as high as 80%)
of spontaneous abortions and 11% of stillbirths in the EME population can be attributed to
anomalies (based on the etiologic fraction). Figures for IUGR/SGA show wide bands: with
14-71% of spontaneous abortions and 38-81% of stillbirths being attributable to this factor.
The relative risk of intrapartum stillbirth in children affected by birth asphyxia is extremely
high (> 800). As a consequence, as many as 90 to 95% of intrapartum stillbirths in the EME
population can be ascribed to birth asphyxia.

                                                  CHAPTER 10: CONCLUSIONS AND DISCUSSION

       In relation to neonatal death, the relative importance of the following risk or causal
factors was assessed: congenital anomalies (including chromosomal aberrations, neural tube
defects, and congenital heart disease), low birth weight, preterm birth, IUGR/SGA, and birth
asphyxia (see Chapters 6 and 7). In general, 6-10% of live births in the EME region are
preterm (< 37 weeks), 5-10% weigh below 2,500 g (LBW), 3-10% are IUGR/SGA, 3% are
anomalous, and only 0.3 to 1.0% suffer from birth asphyxia (see Table 10.1). The majority of
live-born infants affected by one of the risk factors survive the neonatal period (neural tube
defects being the only exception). However, these children do experience elevated risks of
death as compared to their non-affected counterparts. The relative risk of neonatal death is 25-
142 for asphyxiated newborns, 21-59 for LBW babies (< 2,500 g), 17-31 for preterm-born
children (< 37 weeks), 20-23 for anomalous infants, and 1-18 for IUGR/SGA neonates.
Within the total EME population, 62-85% of neonatal deaths can be attributed to LBW (<
2,500 g), 59-68% to preterm birth (< 37 weeks), 36-40% to congenital anomalies (highest for
congenital heart disease), 7-30% to birth asphyxia, and 4-24% to IUGR/SGA.
       To sum up, low birth weight (< 2,500 g) is a relatively important risk factor for
neonatal death in the EME region and, at the level of the individual, an even stronger
predictor than preterm birth (< 37 weeks). This is explained by a comparatively high relative
risk of death in combination with a relatively high prevalence of low weight at live birth.
Birth asphyxia is uncommon, but asphyxiated newborns run quite high risks of neonatal
death. The least important of the risk factors for neonatal death considered is IUGR/SGA.


Kerala, a state in South India, appears to be at the end of the third stage of the epidemiologic
transition (see Chapter 2). The state is more advanced than India as a whole but seems to be
experiencing an ‘overlap of eras’ (cf. Frenk et al. 1989a, 1989b; Bobadilla et al. 1993).
Important causes of general mortality and of infant mortality include leading causes of death
in the EME region as well as causes found in less developed regions. During the 1990s, the
estimated stillbirth rate ranged from 6 to 19 deaths per 1,000 births, early neonatal mortality
was around 9 to 14 deaths per 1,000 live births, and neonatal mortality around 11 to 16 deaths
per 1,000 live births. These estimates are based on population samples (SRS, NFHS).
        Our survey in SAT Hospital, in 2000, resulted in the following mortality figures (see
Chapter 8). The stillbirth ratio was 13 per 1,000 live births, early neonatal mortality was 11
per 1,000 live births, and neonatal mortality was 14 per 1,000 live births. Only 21% of
neonatal deaths occurred on the first day of life whereas as many as 79% occurred during the
first week.
        On the basis of the hospital survey, the present study assessed the relative importance
of the following risk or causal factors in stillbirth: congenital anomalies and IUGR/SGA (see
Chapter 8). Anomalous babies do indeed experience an elevated risk of stillbirth as compared
to non-anomalous children (based on the relative risk) but for growth-retarded or small
babies, the risk of stillbirth was not significantly increased for certain definitions of
IUGR/SGA. Overall, 14% of stillbirths in the population of the SATH survey could be
attributed to anomalies and between 1 and 55% (depending on the definition) to IUGR/SGA


(based on the etiologic fraction). None of the stillborn infants were diagnosed as having
suffered from birth asphyxia, but this finding is implausible and is likely explained by the
small number of stillbirths or by problems with the data.
        In relation to neonatal death, the relative importance of the following risk or causal
factors was assessed: congenital anomalies, low birth weight, preterm birth, IUGR/SGA, and
birth asphyxia (see Chapter 8). In the hospital survey, 19% of liveborn babies weighed less
than 2,500 g at birth (LBW), 9 to 34% (depending on definition) were IUGR/SGA, 6% were
born at less than 37 weeks (i.e. preterm), 2% had congenital anomalies, and 2% suffered from
birth asphyxia (see Table 10.1). Most LBW babies weighed between 2,000 and 2,499 g. The
majority of live-born infants affected by one of the risk factors survived the neonatal period.
However, most of these neonates did experience elevated risks of death as compared to their
non-affected counterparts. The relative risk of neonatal death was 38 for asphyxiated
newborns, 30 for anomalous infants, 17 for preterm-born children (< 37 weeks), 11 for LBW
babies (< 2,500 g), and 2-11 for IUGR/SGA neonates. Moreover, further analysis showed that
65% of all neonatal deaths in the population of the SATH survey can be attributed to LBW (<
2,500 g), 47% to preterm birth (< 37 weeks), 41% to birth asphyxia, 34% to congenital
anomalies, and 20 to 70% to IUGR/SGA.
        Overall, low birth weight (< 2,500 g) is an important contributor to neonatal death in
the population (as a result of its high prevalence) although preterm birth (< 37 weeks) is a
better indicator of risk of neonatal death at the individual level. The risk of neonatal death for
individuals is highest for those affected by birth asphyxia. The least significant risk factor for
neonatal death is IUGR/SGA, although the figures cover a wide band depending on the
reference growth curve and the definition of IUGR/SGA (see Chapter 8).

Underlying risk factors and pathways
In addition to the above analysis, the data from the hospital survey were also analysed to
assess the importance of several selected underlying risk factors (see Chapter 9). The
following risk factors were included in the analysis: hypertensive disorders (pre-existing and
pregnancy-related), diabetes (pre-existing and pregnancy-related), infections during
pregnancy, smoking and alcohol consumption, anaemia, advanced maternal age, antepartum
haemorrhage (abruptio placentae and placenta praevia), and prolonged and obstructed labour.
These are risk factors for adverse pregnancy and birth outcome (e.g. low birth weight, preterm
birth), including stillbirth.
        The most frequently observed risk factors among pregnant women and births ≥ 28
gestational weeks in the hospital survey were: haemoglobin (Hb) level < 11 g/dl (39%),
instrumental or operative delivery (29%), pregnancy-related hypertensive disorders (15%),
maternal age ≥ 30 years (14%), obstructed labour (13%), Hb level < 10 g/dl (8%), and
malpresentation or malposition (5%). The percentage of caesarean sections was quite high at
26%, with cephalo-pelvic disproportion (42%) and previous caesarean section (40%) as the
most common indications. Population-based data from NFHS-2 suggest an even higher figure
for caesarean sections in Kerala as a whole (29% of total births). The least prevalent risk
factors in the hospital survey were: eclampsia (0.1%), pre-existing hypertension (0.2%),
abruptio placentae (0.5%), and anaemia according to hospital diagnosis (0.6%). However,

                                                         CHAPTER 10: CONCLUSIONS AND DISCUSSION

pre-existing hypertension may not always be recognised and the true number of cases may be
higher (see Chapter 9). No information was available on smoking and alcohol consumption
but other sources indicate that these habits are rare among women of reproductive ages in
Kerala. In terms of infections, 2.9% of pregnant women were tested positive for hepatitis B,
0.9% of deliveries were complicated by maternal fever, and 0.9% of women suffered from
other infectious diseases during pregnancy. However, it is believed that, in Kerala, many
infections during pregnancy remain unreported and/or unregistered.
        At the level of the individual, pathways of significantly increased risk1 involved
hypertensive disorders, diabetes, infections, advanced maternal age (≥ 30 yrs, ≥ 35 yrs),
antepartum haemorrhage, and abruptio placentae. It should be noted that the number of cases
in the SATH survey was often small and that the results were not adjusted for potential
confounding factors. Children born of mothers with hypertensive disorders ran elevated risks
of LBW, preterm birth, SGA, and stillbirth compared to children born to non-affected
mothers. Diabetes also increased the risk of stillbirth, but this risk factor could not be related
to other significantly increased, or reduced, risks of adverse outcomes. Nevertheless, pre-
existing diabetes was the only risk factor that could be related to an elevated risk of congenital
anomalies. Antepartum haemorrhage and abruptio placentae resulted in elevated risks of
LBW, preterm birth, and stillbirth. Infections during pregnancy or labour increased the risks
of birth asphyxia and stillbirth. Advanced maternal age also increased the risk of birth
asphyxia. It is important to note that the observed associations between advanced maternal
age and adverse outcome could well have been confounded by other risk factors, such as
hypertensive disorders, diabetes, abruptio placentae, hepatitis B, and abnormalities of the
reproductive tract. Anaemia and low haemoglobin level were not found to be related to any
adverse outcome, and nor were complications during labour.
        Important contributors to the total number of stillbirths in the survey population (i.e. an
EF > 10%) were hypertensive disorders, diabetes, antepartum haemorrhage, and abruptio
placentae. In addition, some risk factors contributed to neonatal mortality. Over 10% of LBW,
preterm, and SGA births in the population could be attributed to hypertensive disorders, while
hepatitis B and maternal age ≥ 30 years contributed to birth asphyxia. It can thus be concluded
that the presence of hypertensive disorders (mostly pregnancy-related) is a very important risk
factor. Relatively large proportions of preterm births (31%) and stillbirths (46%) in the survey
population could be ascribed to hypertensive disorders. Furthermore, hypertensive disorders
contributed to LBW (14%) and SGA (4-11%) in the population.

10.4 The transition
Having assessed the current situation in both regions under consideration in Section 10.3, the
present section compares the results for the regions and describes the observed developments
during the transition. Section 10.4.1 focuses on the general developments with regard to
perinatal and infant mortality, and adverse pregnancy and birth outcome (see Chapters 2 and
8). Section 10.4.2 discusses the health transition in Kerala (see Chapters 8 and 9).

    A relative risk greater than 1.0, with the 95% confidence interval excluding the value of 1.0.


A comparison of results and data for Kerala, the EME region, and all-India and other Indian
regions leads to the following conclusions about an epidemiologic or, more general, health
transition (see also Table 10.1).

Perinatal and infant mortality
Data demonstrate a clear decline in infant mortality during the epidemiologic transition.
Similarly, perinatal, early neonatal, and neonatal mortality decline during the transition. The
decline is less distinct for stillbirth (as officially registered), but this may be a result of
registration procedures and differing criteria or definitions.
        The assumed shift in timing of infant death from the postneonatal to the neonatal
period, and even to the early neonatal period, cannot be clearly observed. The ratio of
neonatal to postneonatal deaths does not seem to be a good indicator of a region’s position in
the epidemiologic transition. This may be explained by a general lack of data for the least
developed countries. In addition, the postponement of infant death during the later stages of
the transition may add to the confusion. In general, in most countries where data are available,
the majority of infant deaths take place during the neonatal period.
        With regard to the causes of infant death, the proportions due to congenital anomalies
and perinatal conditions (i.e. endogenous causes) increase as a country advances in the
transition, whereas the relative contributions of infectious and parasitic diseases and diseases
of the respiratory system (i.e. exogenous causes) diminish.

Adverse pregnancy and birth outcome
The frequencies of the selected adverse pregnancy and birth outcomes, i.e. congenital
anomalies, low birth weight, preterm birth, IUGR/SGA, and birth asphyxia, change during the
transition (see Table 10.1). In general, the prevalence proportions among births decline. A
comparison of data indicates that the frequencies among live births of the following adverse
outcomes decline during the transition: birth weights < 2,500 g (LBW) and < 2,000 g,
gestational ages < 37 weeks (preterm) and < 32 weeks (very preterm), IUGR/SGA, and birth
asphyxia. The proportion of live births weighing less than 1,500 g does not appear to change
during the later stages of the transition. The observed decline in the frequency of birth
asphyxia is plausible since good-quality antenatal care and care during delivery are able to
prevent severe birth asphyxia. For congenital anomalies, no clear trend was observed.
Changes may be masked by differences in rates of detection, available diagnostic methods,
financial resources available for diagnosis, and length of follow-up.
        Findings from the present study in terms of LBW, preterm birth, and IUGR/SGA
suggest that, during the transition, the proportion of preterm births declines earlier and/or
more rapidly than the proportion of IUGR/SGA births. Possible explanations include earlier
or more rapid improvements in the underlying causes of preterm birth, and a stronger
intergenerational influence on IUGR/SGA. However, the finding could also be specific to the

Table 10.1: The relative importance of the selected risk factors for stillbirth and neonatal death in the EME region and in
Kerala*, and suspected changes as countries advance through the epidemiologic and health transition
                           Prevalence in births                        RR                                          EF (%)
                            Kerala*      EME             change         Kerala*         EME          change         Kerala*         EME          change
congenital anomalies             2             3           ?**               9             5             –              14           11             –
IUGR/SGA                       9-34          3-10            –              1-5          7-44            +             1-55         39-81          –/+

Neonatal death
                           Prevalence in live births                   RR                                          EF (%)
                            Kerala*      EME         change             Kerala*         EME          change         Kerala*         EME          change
congenital anomalies             2            3            ?**              30          20-23            –             34           36-40           +
< 2,500 g                       18          5-10             –              11          21-59           +              65           62-85          –/+
< 2,000 g                        4           2-3             –              24         44-107           +              48           55-65           +
< 1,500 g                        1         0.9-1.2           =              15         69-160           +              13           46-58           +
< 37 wks                         6          6-10            –**             17         17-31           =/+             47           59-68           +
< 32 wks                         2         0.8-1.4           –              19          83-86           +              27           53-54           +
IUGR/SGA                       9-34         3-10             –             2-11          1-18          =/+            20-70          4-24          –/=
birth asphyxia                   2         0.3-1.0           –              38         25-142          –/+             41            7-30           –
Notes: *Based on the survey in SAT Hospital, Trivandrum, October 2000; **Also based on data from Kerala, all-India, and regions in India other than Kerala.

situation in Kerala (see Chapter 8). The decrease in the prevalence of preterm birth during the
epidemiologic transition may be related to declines in causes and risk factors such as
infections, uterine abnormalities, and IUGR. In addition, improvements in the nutritional
status of the mothers might play a role. Further research is necessary to clarify the findings.
        Within the high-risk categories of LBW (< 2,500 g) and preterm birth (< 37 weeks),
distributions by weight and age change during the transition. A shift takes place, during the
later stages, towards the extremely low-weight and extremely preterm categories. In regions
in transition, such as Kerala, the vast majority of LBW births can be expected to weigh
between 2,000 g and 2,499 g. The increased reporting and registration as births (as opposed to
abortions) of children who are born very preterm and/or with extremely low weight, is
probably the result of lower viability limits and more aggressive life-saving efforts at lower
weights and younger ages (see Chapter 8). In addition, many women in regions that are less
advanced in the health transition may not be able to reach high-level medical care in time in
the event of sudden and rapidly progressing complications, or unexpected and extremely
preterm labour.
        By comparing the results for the EME region and for the SATH survey in Kerala, the
following developments were observed in the relative risks of death (see Table 10.1). For
anomalous infants, the relative risks of stillbirth and neonatal death appear to decline during
the later stages of the epidemiologic transition. However, this finding is likely to be affected
by differences in the detection of congenital anomalies after birth, with detected and reported
anomalies in less developed regions probably consisting of those that are easiest to observe,
such as those most severe and/or most lethal. For preterm-born children (< 37 weeks),
IUGR/SGA neonates, and asphyxiated newborns, the relative risks of neonatal death remain
around the same level during the later stages of the transition. However, the relative risk of
neonatal death increases for infants with low birth weight (< 2,500 g, < 2,000 g, or < 1,500 g)
and for very preterm infants (< 32 weeks) in the later stages of the transition. At first glance,
this is not what one would expect in terms of a health transition. Explanations may be sought
in differences in natural antenatal selection, in quality of care for non-compromised infants, in
underlying causes, and in racial/genetic variations in birth weight and/or gestational duration.
However, the most plausible explanation seems to be in the changing distribution of birth
weights and gestational ages within the low-weight and preterm categories, with a trend
towards the extremely low-weight and preterm categories (as discussed above). Indeed, the
inclusion of these highest-risk groups in the category of births, instead of abortions, will
almost certainly negatively affect the relative risk of neonatal death.
        Turning to attributable risk (i.e. EF), a comparison of the results from the SATH
survey with the estimates for the EME region suggest the following (see Table 10.1). Within
the total population, the proportion of stillbirths that can be attributed to congenital anomalies
and the proportion of neonatal deaths that can be attributed to birth asphyxia decline during
the later stages of the transition. Conversely, the proportions of neonatal deaths that are
attributable to congenital anomalies, low weight at birth, and preterm birth increase. Despite
these changes, it is clear that the five selected pregnancy/birth outcomes (i.e. congenital
anomalies, low birth weight, preterm birth, IUGR/SGA, and birth asphyxia) account for a
large share of neonatal deaths both in regions that are approaching the later stages of the

                                                  CHAPTER 10: CONCLUSIONS AND DISCUSSION

transition as well as in regions that are more advanced. In both populations, over 60% of
neonatal deaths can be ascribed to a birth weight below 2,500 g. In addition, more than 40%
of deaths can be ascribed to preterm birth.


Risk transition?
Smoking, the consumption of alcohol, and advanced maternal ages (≥ 30 years and ≥ 35
years) are clearly less frequent in the SATH survey and Kerala than in the EME region or the
developed world. These results do not show any indication of a risk transition, i.e. changing
patterns of consumption and of lifestyles (cf. WHO 2002, see Chapter 2). In addition, low
haemoglobin levels (< 11 g/dl and < 10 g/dl) are still somewhat more common in Kerala than
in the EME region, although no longer as frequent as in some other Indian regions. What,
however, is important to note is that the group of women with delayed childbirth may differ in
composition between the two regions. In Kerala, delayed childbirth may possibly be more
frequently related to health and reproductive problems than in the EME region, where it may
be a more active decision. In general, women in Kerala give birth at a comparatively young
age and they prefer small family size. It would be informative to know why the older women
in the hospital survey had delayed childbirth.
        The frequency of complicating factors such as hypertensive disorders and diabetes in
the SATH survey is equal to, or even exceeds, figures in the EME region. This could be
related to SAT Hospital’s status as a referral hospital, but it may also indicate changes during
the transition. The prevalence of chronic hypertension can be expected to increase during the
transition, and frequency of eclampsia to decline. In our hospital survey, a lower proportion of
eclamptic cases was observed in comparison to that found in other developing countries. On
the other hand, the prevalence of pre-existing hypertension still seems low in the SATH
population. However, it is possible that chronic hypertension is not always recognised, and
the true number of cases of pre-existing hypertension in the SATH survey may be higher (as
discussed earlier). Furthermore, many of the known risk factors for pregnancy-induced
hypertension and pre-eclampsia can be expected to increase during the transition, such as
diabetes, obesity, and factors that predispose to cardiovascular disease. The comparatively
high prevalence of hypertensive disorders (pre-existing and pregnancy-related) in the SATH
survey may thus indicate changes in Kerala among the lower socioeconomic groups in
nutrition patterns and towards more sedentary lifestyles (cf. the nutrition transition; Popkin
1993, 1994). The observed importance of hypertensive disorders in pregnancy, as risk factors
of adverse outcome in Kerala, could thus be an indication of a risk transition taking place.

Epidemiologic transition
In terms of perinatal and neonatal health and survival, and maternal health, Kerala still shows
some of the features of developing countries that are not as advanced in the epidemiologic
transition as the low-mortality countries, namely: higher mortality rates and higher frequency
of IUGR/LBW, maternal anaemia (Hb < 11 g/dl), obstructed labour, birth asphyxia, and
possibly maternal infections. However, most of these figures are lower than figures for India


as a whole and, as a result, Kerala can be seen as occupying a middle position. The transition
can also be observed in declines in eclampsia and severe anaemia to similar low levels as
those found in the low-mortality countries. Furthermore, obstructed labour and maternal
anaemia do not seem to increase the risk of adverse outcome for the child within the region.
In practice, the majority of cases diagnosed as obstructed labour are managed by operative
delivery, i.e. caesarean section, and care during delivery thus appears to be safe for the child.
IUGR/SGA has only a small effect on the risk of neonatal death, and the category of low birth
weight mainly includes newborns who weigh between 2,000 and 2,499 g (76% of LBW
births). The possibly high proportions of hypertension and diabetes (pre-existing and
pregnancy-related) may indicate that further advances in the transition are to be expected in
the near future.

Health care transition: medicalisation
Although the present study focused on an epidemiologic transition, it also revealed a health
care transition. A major component of the epidemiologic and health transitions in Kerala
seems to be the comparatively good availability, accessibility, and quality of the health care
services. The health care transition seems to have progressed with a trend towards the
medicalisation of pregnancy and birth. The observed global ‘epidemic’ of caesarean sections
is part of this trend. In Kerala, increases in the reported numbers of obstructed labour (mainly
cephalo-pelvic disproportion) and caesarean sections indicate a trend of medicalisation (see
Chapter 9). Other indicators of medicalisation in the region are the large proportions of
women undergoing an ultrasound scan during pregnancy or an episiotomy during labour.
        A paradox seems to be that a lack of resources, in combination with higher demands
set by the population, is resulting in medicalisation. The present study indicated that, in
Kerala, public hospitals are crowded and limited in facilities, space, time, and financial
resources, and that this is likely to put pressure on both medical staff and the women in
labour. In order to avoid potentially dangerous situations and to alleviate pressure, the public
health sector reaches out for interventions, sometimes unnecessarily (see Chapter 9). At the
same time, according to Thankappan (2001), the poor situation in the public health sector in
Kerala has opened doors for private hospitals and these are more likely to seek refuge in
interventions and technology for commercial reasons. Other authors (Padmadas et al. 2000;
Thankappan 2001) have already expressed their concerns about the large proportion of
caesarean sections in Kerala and the accompanying increase in health care costs. It is
important to note that medicalisation may possibly also have negative effects on the health of

The findings that were discussed in the sections above have indicated that during the
transition mortality declines as do the frequencies of several adverse pregnancy and birth
outcomes. Nevertheless, not all changes during the transition are improvements. Increases in
hypertensive disorders and diabetes of course are not, and increases in smoking and alcohol
consumption would also not be considered as a positive development. Further, certain
developments may hide possible improvements. These include:

                                                   CHAPTER 10: CONCLUSIONS AND DISCUSSION

•   changes in the composition of the group of women who use health care services (cf.
    selection bias),
•   changes in the distribution within risk categories towards the highest-risk groups as a
    result of improved survival and more aggressive life-saving efforts (as observed among
    LBW and preterm births),
•   improved detection and diagnosis of health problems, and
•   medicalisation, i.e. more cases indicated as complicated/problematic.

10.5 The future: expectations and recommendations
Globalisation has been affecting the course of the health transition, including the
epidemiologic transition, in both developing and developed countries. For example, the re-
emergence of infectious and parasitic diseases in developed countries, defined as the so-called
fifth stage of the epidemiologic transition, is in part explained by migration, urbanisation, and
the growth in international travel movements. Globalisation seems to have been affecting the
way countries and regions are passing through the stages, as defined in the models. For
developing countries, the stages in the transition will be different than they were for
developed countries. The addition of the fifth stage, and perhaps even the fourth stage, needs
to be reconsidered for developing countries. In fact, their completion of the perceived
epidemiologic transition seems to have been overtaken by time and globalisation processes.
        Health care systems in developing countries face an increasing ‘double demand’. Basic
requirements for care, infrastructure, availability, and accessibility should be fulfilled for the
whole population. At the same time, the health care system is facing strong demands for the
latest technology and sometimes unnecessary interventions whether or not enforced by fear of
litigation and/or commercial motives. Policymakers in regions in transition working within
the public health sector need to focus on expected changes and developments. Anticipation of
developments is of vital importance in order to respond to them in a timely and effective
manner. In addition, the existing healthy behaviour and lifestyles of the population should be
appreciated and encouraged. The public health care sector needs to try to keep up with
increasing demands for health care and has to keep investing. This is likely to be complicated
by bureaucratic and slow procedures such as in India and Kerala, where a government
hospital such as SAT Hospital is still using antenatal charts from the 1960s. Furthermore, it is
necessary to look at the efficiency of the health care system and the use of resources (staff,
supplies, beds), and to avoid over-medicalisation and medically unnecessary interventions.
Medical staff and policymakers should critically review intervention procedures and the
reasons for interventions as well as the pros and cons of some routine procedures.

Future research
The present study has put forward several points for further research. With respect to content,
these include further research on:
• the shift towards extremely low birth weights and extremely preterm births during the
• the observed decline in the prevalence of preterm births during the transition,


•   hypertensive disorders in pregnancy,
       including: What factors underlie hypertensive disorders in pregnancy? How are
       hypertension and nutrition related? Does the frequency of pregnancy-related
       hypertensive disorders change during the transition and, if so, in what direction?
•   other regions in transition and other states in India,
       in order to find out whether the findings of the present study are specific to the
       situation in Kerala.

In relation to data and methods, the following points are of interest:
• the definition and measurement of intrauterine growth retardation, small-for-gestational-
    age, and birth asphyxia,
        In order to distinguish high-risk infants from healthy, low-risk infants, and to simplify
        reliable international comparison. In addition, authors need to be more explicit about
        their definitions and measurements than has been the case in the past.
• the gap in available data between data sources and disciplines,
        In order to bridge the findings from studies in different disciplines, and to identify the
        causal mechanisms that underlie the observed associations between socioeconomic and
        sociodemographic factors on the one hand and adverse health outcomes, medical risk
        factors, and mortality on the other.
• the further improvement of population forecasting.
        Forecasting procedures should reflect the underlying biological processes and
        mechanisms. By understanding the nature and sequencing of processes and
        mechanisms, forecasting can be improved, and simulation models can be designed to
        estimate the demographic effects of developments and interventions in public health.
        As such, the results of the present study can be used for microsimulation.

To conclude, it is clear that life starts before birth. It is time for demographers to realise that
early life, even before birth, is a key indicator of the health and survival of populations.