Methods

1
How many child deaths can we prevent? Best
estimates for 2003. Saul S Morris, Robert E
Black, Kenji Shibuya, Simon N Cousens, Jennifer
Bryce, Gary L. Darmstadt
Abstract In 2003, the Bellagio study group
estimated the numbers of under-five deaths that
could be prevented by making efficacious child
survival interventions universally available.
Subsequent changes in world population, mortality
rates, and intervention coverage, as well as the
publication of a new series of estimates of the
by-cause distribution of deaths in each country
and a recent review of interventions for the
neonatal period, make it appropriate to update
these estimates. Out of a total of 9.9 million
deaths taking place in 60 high risk countries in
2003, 63 could have been averted using currently
available technology. The interventions that have
the greatest life-saving potential if implemented
in isolation are much as previously identified.
Background The 2003 Lancet Child Survival Series
estimated, for the year 2000, the distribution of
under-five deaths by cause and the proportion
that could be prevented by making universally
available all child survival interventions of
demonstrated efficacy. New information has since
become available on under-five mortality rates,
world population, the coverage of life-saving
interventions, and the epidemiology of neonatal
deaths (Lancet 2005, Vol.365, pp.977-88). The
estimates published in 2003 may therefore no
longer reflect current circumstances. This study
aims to update cause-specific estimates of
under-five deaths and the proportion of deaths
preventable using existing interventions, using
2003 as the new reference year.

• Methods
• This study used the same methods as originally
  presented by the Bellagio group in Lancet, 2003,
  vol.362, pp.65-71. For each country and for each
  life-saving intervention, we estimated the
  increase in coverage necessary to achieve
  universal coverage and combined this with
  estimates of intervention efficacy to estimate
  the potential number of deaths that could be thus
  prevented. These deaths averted were removed from
  the total current number of deaths, and the
  exercise was repeated for an additional
  intervention, until all interventions of
  demonstrated efficacy were exhausted.
• The model differs from that presented in 2003 in
  the following major respects
• The current coverage of efficacious
  interventions was updated to reflect the
  situation in 2003, as reported in State of the
  Worlds Children 2005.
• Population estimates were updated to reflect the
  situation in 2003, as reported by the UN
  Population Division.
• The baseline proportional distribution of
  child deaths by cause in country was that
  estimated by WHO and reported (at an aggregate
  level) in the World Health Report 2005.
• Interventions for the neonatal period were taken
  from Lancet 2005, Vol.365, pp.977-88, using the
  mid-point efficacy.
• The list of countries was modified to include
  those currently having under-five mortality rates
  greater than or equal to 90 per 1000 live births
  or total number of under-five deaths greater than
  or equal to 50,000 (the 60 UNICEF priority
  countries).
• Universal coverage was defined as 99 for all
  interventions, with no exceptions.

Results Of the 10.6 million under-five deaths in
2003, 9.9 million (93) took place in one of the
60 priority countries. 6.3 million deaths (63)
could have been averted by scaling up 26
efficacious interventions so that they reached
all those in need. Table 1 shows deaths averted
by cause. The Figures show the proportion of
deaths that could be averted by scaling up each
intervention in turn, leaving the coverage rates
of all other interventions unchanged.
Conclusions and Recommendations Approximately
two-thirds of child deaths can be averted with
existing interventions. In the future, this model
needs updating to include a range of
interventions for which new efficacy data are now
available. These include pneumococcal and
rotavirus vaccines, handwashing, chlorhexidine
cleansing of the cord, and others.