"They did the right things," Jennifer Bryce, evaluator and researcher at Johns Hopkins Bloomberg School of Public Health, told IRIN. "They did what is feasible to do quickly and outside of health systems. Where there was disappointment and missed opportunity was that they did not do more."
Focusing on Benin, Ghana and Mali, the evaluation analysed the impact of UNICEF's US$27 million child survival programme implemented between 2001 and 2005.
Findings showed mortality in under-fives decreased by 13 percent in Benin, 24 percent in Mali and 20 percent in Ghana, but that the decreases in the former two were no greater than in non-project areas, while no comparison data was available for Ghana.
One reason for this is that UNICEF pushed Health Ministries, donors and aid agencies in the countries involved to adopt similar approaches to child survival, which led mortality rates to drop elsewhere too, Mickey Chopra, UNICEF's head of health, told IRIN.
"In most instances there was a dramatic decline in child mortality. These are very poor countries and this was not a time of economic prosperity or declining poverty, so the interventions were the things that really made the difference," Chopra told IRIN.
UNICEF implemented the project in some of the hardest-to-reach and worst-off areas of targeted countries, Chopra pointed out.
Other criticisms of UNICEF's approach include an insufficient emphasis on under-nutrition, and interventions to address neonatal deaths (mortality of babies aged 28 days and under).
UNICEF did not push Health Ministries hard enough to change their policy on allowing community volunteers to treat children suffering from pneumonia, diarrhoea and malaria - the three biggest child killers in West Africa. Changing this could drastically reduce child mortality rates, said evaluator Bryce.
"Children cannot wait for child health days to be treated. They need a trained person right then and there, when they are sick."
Proven interventions include giving children antibiotics for pneumonia, zinc and oral rehydration salts for diarrhoea, and artemisinin-based combination therapy and bed nets for malaria.
UNICEF has listened and responded to these lessons, said Bryce. "They really responded to the results, even up to the highest level."
The agency has changed tack, working with others to lobby ministries to shift their policy on community health worker responsibilities to administer treatment, said UNICEF's Chopra, and policy change is under way in 46 African countries, including Mali, Malawi and Ethiopia.
Neo-natal mortality is now more central to child survival interventions, and UNICEF is among many donors to push nutrition up the agenda.
The most exciting change to emerge out of child survival analysis in recent years, said Bryce, is the development of computer models that can accurately analyse causes of child mortality and the effects of different interventions. This allows experts to input different response scenarios - upping mosquito-net provision while scaling back on HIV intervention for instance - to ascertain which would save the most lives.
"With this technology, you don't have to have donors driving the agenda. You can have the causes of death driving it," said Bryce.
The publication of the evaluation's frank findings of the weaknesses of UNICEF's programme have worried some that UNICEF and other aid agencies will refrain from publishing less-than-glowing evaluation results in the future. "After the past few days why would anyone ever want to publish an evaluation?" asked Bryce.
But that would be a mistake, said Chopra. Aid agencies and donors have promised to be more transparent about their aid efforts in a bid to make the often-murky sector more accountable.
"We commissioned this evaluation and we didn't block publication. We have a responsibility to share these lessons. We're committed to continue to learn how to improve the work we do. and are committed to putting our findings in the public domain."