Demographic and Health Survey 2009-10 - Timor-Leste Ministry of ...

with their death, or they may live in a culture that discourages discussion of the dead. The possible
occurrence of these data problems in the 2009-10 TLDHS is discussed with reference to the data
quality tables in Appendix C. Underreporting of births and deaths is generally more common the
further back in time an event has occurred.
An unusual pattern in the distribution of births by calendar years is an indication of omission
of children or age displacement. In the TLDHS, the cutoff date for asking health questions was
January 2004. Table C.4 shows that the overall percentage of births for which a month and year of
birth was reported is almost 100 percent for both children who have died and children who are alive.
Table C.4 shows some age displacement across this boundary for both living and dead children. The
distribution of living children and the total number of children shows a deficit in 2004 and an excess
in 2003, as denoted by the calendar year ratios. The deficit in 2004 can be attributed to the
transference of births by interviewers out of the period for which health data were collected.
Transference is proportionately higher for dead children than for living children, and this
displacement may affect mortality rates. The transference of children, especially deceased children,
out of the five-year period preceding the survey is likely to result in an underestimate of the true level
of childhood mortality for that period.
Underreporting of deaths is usually assumed to be more common for deaths that occur very
early in infancy. Omission of deaths or misclassification of deaths as stillbirths may also be more
common among women who have had several children or in cases where death took place a long time
ago. To assess the impact of omission on measures of child mortality, two indicators are used: (1) the
percentage of deaths that occurred within seven days of birth to the number that occurred within one
month of birth and (2) the percentage of neonatal to infant deaths. It is hypothesized that omission
will be more prevalent for those who died immediately after birth than for those who lived longer, and
that it will be more common for events that took place in the distant past compared with those that
took place in the more recent past. Table C.5 shows data on age at death for early infant deaths.
Selective underreporting of early neonatal deaths would result in an abnormally low ratio of deaths
within the first seven days of life to all neonatal deaths. Early infant deaths have not been greatly
underreported in the TLDHS survey, as suggested by the high ratio of deaths in the first seven days of
life to all neonatal deaths (76 percent in the five years preceding the survey).
Heaping of the age at death on certain digits is another problem that is inherent in most
retrospective surveys. Misreporting of age at death biases age pattern estimates of mortality if the net
result is the transference of deaths between age segments for which the rates are calculated. For
example, child mortality may be overestimated relative to infant mortality if children who died in the
first year of life are reported as having died at age 1 or older. In an effort to minimize misreporting of
age at death, interviewers were instructed to record deaths at under 1 month in days and deaths at
under 2 years in months. In addition, they were trained to probe deaths reported at exactly 1 year or 12
months to ensure that they had actually occurred at 12 months. The distribution of deaths under 2
years during the 20 years prior to the survey by month of death shows that there is some heaping at 6,
9, and 12 months of age, with corresponding deficits in adjacent months (Table C.6). However,
heaping is less pronounced for deaths in the five years preceding the survey, for which the most recent
mortality rates are calculated.
8.3 LEVELS AND TRENDS IN INFANT AND CHILD MORTALITY
Neonatal, postneonatal, infant, child, and under-5 mortality rates are shown in Table 8.1 for
cohorts of children born in three consecutive five-year periods before the survey. Under-5 mortality
for the most recent period (0-4 years before the survey or, roughly, during the calendar years 2005-
2009) is 64 deaths per 1,000 live births. This means that 1 in 16 children born in Timor-Leste dies
before the fifth birthday. Seventy percent of deaths among children under age 5 occur during the first
year of life: infant mortality is 45 deaths per 1,000 live births. During infancy, the risk of neonatal
deaths and postneonatal deaths is 22 and 23 per 1,000 live births, respectvely. The proportion of child
deaths that occurs in the neonatal period (34 percent) in Timor-Leste is lower than the global
estimation of 38 percent (Lawn et al., 2005).
100 | Infant and Child Mortality