The cost and cost outcomes of school feeding programs in

Download Report

Transcript The cost and cost outcomes of school feeding programs in

THE REVIEW TEAM











Elizabeth Kristjansson
Vivian Welch
Mark Petticrew
Barbara MacDonald
Julia Krasevec
Trish Greenhalgh
Laura Janzen
Jessie McGowen
George Wells
Bev Shea
Peter Tugwell
THE COST OUTCOME
TEAM






Rae Galloway
Elizabeth Kristjansson
Aulo Gelli
Francisco Espejo
Ute Meir
Donald Bundy
•
•
Dearth of evidence on the effectiveness, cost
and cost effectiveness of school feeding
Some costing
•
Del Rosso (1999)
• Cost standardized over 365 days and 1000 kilocalories
(kcal), ranged from US$ 19.35 to 208.59
•
WFP (Gelli, et al., 2006)
• Costs (standardized over 200 days and 700 kcal), US$ 34
per child per year in 2001 and US$ 20 per child per year
in 2006
•
Little information on cost breakdown
•
WFP only estimated its own costs
•
•
No costs to government or civil society
NO Cost-outcome analysis before
And
•
New systematic review by Kristjansson et al
(2007) provided solid evidence on outcomes of
school feeding
•
•
•
To provide an up-todate, realistic estimate
of the costs of school
feeding
To combine these
estimates with results of
the Kristjansson (2007)
review to calculate cost
per outcome
Focus is on low and
middle income
countries, and in
particular, on Africa.
1.
Determine benefit of
school feeding for health
of poorer children and
youth
2.
Compare effectiveness for
socio-economically
disadvantaged children
and advantaged children.
3.
Understand factors that
impact on effectiveness

Inclusion criteria

Participants: elementary or high school children
and youth aged 5 to 19

LMIC and higher income countries

Meals, snacks, or milk.

Studies must be targeted at children who are
economically disadvantaged or must compare
economically disadvantaged to advantaged





Formal search of databases broad, included grey
literature Focused on intervention. Added
search terms for SES, disadvantaged. To
November 2007
Hand-searched, reference lists from articles and
reviews, asked experts
2 researchers independently reviewed search
results
2 reviewers decided on inclusion/exclusion (with
advice from others)





Data extraction (4 people)
Tables of included and excluded studies
Wrote to authors for missing info
Where possible, performed meta-analysis
Adjusted for the fact that some programs were
assigned by school or class
 This means that children in these groups share some
characteristics and that treatment is more likely to affect them
in the same way

Analyses results to see if there were differences by
nutritional status, gender, age




Process evaluation performed: intensity of
intervention, adherence, substitution, setting.
Realist review performed
Tabulated results by various study
characteristics
Nutritionists reviewed all studies




Adequacy of energy intake (Low= 15% of RDA for energy)
Physical outcomes
Cognitive measures and outcomes:
neuropsychologist
Two reviewers independently assessed study
quality

LMIC: 9 studies
•
•


5 RCTS 4 CBAS.
2 RCTs and 1 CBA less than 15% RDA for energy
In 3 studies that assessed substitution,
children only had a net gain of half of the
calories given
Higher Income countries: 9 studies
•
•
•
•
2 RCTs
5 CBAs
2 ITS.
1 RCT and 1 CBA less than 15% RDA for energy
•
Kenya, Malawi, Lesotho (highlands and
lowlands), and The Gambia for the analysis
•
•
Cooked meals
Aimed to get at all costs
•
WFP costs
•
Government costs
•
Community costs
•
WFP costs
•
Commodities
•
Transport
•
Storage and handling
•
WFP staff time in-country
•
Overhead
•
•
Costs
•
Salaries
•
Commodities
•
Transportation
•
Supplies
•
Cash and in-kind contributions made by teachers and
communities
•
Subtracted cost of take-home rations
Estimated by Field visits, Interviews with Ministry
of Education staff and teachers
•
•
To compare:
standardized for 200
days and 700 kcal.
Needed to adjust for
actual days fed
•
Gambia
•
•
Knew actual days fed
Kenya, Lesotho,
Malawi
•
•
•
Planned versus actual
tonnage
To account for pipeline
breaks.
Scaled the cost per
beneficiary by planned
versus actual tonnage
•
•
For each study, gains
per month calculated
by dividing total
effect by number of
months feeding given
Standardized effects
to 10 months or 200
days of feeding
•
In each region
•
Divided cost for 200 days by outcome (e.g. IQ
points gain, weight gain) averaged over 200
days

School feeding resulted in small increases in
weight
Mixed evidence for height
 Improved attendance
 Improved math performance
 Improvements in intelligence type scores
 Increases in attention, on-task behaviour

No effects on:



Spelling
Reading
Vocabulary
Attendance
Weight Gain
Height Gain
RCT
CBA
RCT
CBA
RCT
CBA
No. of Studies
2
1
3
3
3
3
Duration of
feeding
1 and 8 mos
3 mos
Avg. 15.3 kg
Avg. 9.7 kg
Avg. 15.3 cm
Avg. 9.7 cm
Per month effect
.40-.60 days
1.65 days
.025 kg
.073 kg
.025 cm
.147 cm
Effect size
(Hedges adjusted
.18
.19
.25
.75
.17
1.12
5 - 7 days **
16.5
days
.25 kg
.73 kg
.25 cm
2.7 cm
(5 to 6 years old)
1.29 cm
(6 to 8 years old)
1.47 cm (overall)
g.)
Per school year
effect*
Intelligence Type Tests
Math Performance
RCT
CBA
RCT
CBA
No. of studies
1
1
2
2
Duration of feeding
20 mos
18 mos
8 and 10 mos
Avg: 6 mos
Points per school year
.34 on Raven’s
Progressive Matrices
2.2 IQ
.89*
.18 on Math subtest
of WISC
1.2*
Significance
P < .05
Significant
Not Significant
Effect size: Hedges
adjusted g
.11
.15
P < .05
Significant
.12 WRAT Math
.11 WISC Math
Significant
.31
Cost per extra day of
attendance
Range: RCTS
Average: RCTS
Range: CBAS
Average: CBAS
Cost per extra
centimeter of height
gain per child
Cost per extra
kilogram of weight
$ 5 6 to $12.3
$112.4 to $245.8
$112.4 to $245.8
$ 9.7
$195
$200
$ 1.7 to $ 3.72 4
$ 19.4 to $41.8
$38.5 to $84.2
$2.95
$33
$66.
Cost per point on
Cost per IQ point
Raven’s
Progressive
Matrices
Range: RCTS
$ 82.6 to $180.7
Average: RCTS
Range: CBAS
Average: CBAS
$152.6
Cost per point on
math
achievement/aptitu
de
$31.6 to $69
(WRAT).
$156.1 to $341.3
(Math subtest
WISC).
$54.6
$ 270.1
(WISC Math)
$12.8 to $27.9
$23.1 to $51.2
$22.1
$40.5
Thorough and realistic estimate
Only study besides Carr-Hill in Lethoso to
calculate full costs
Based on field interviews
BUT
Assumed linear gains
We believe that effects in Kristjansson (2007)
review are an underestimate of possible effects
•
•
•
•
•
•
•
Substitution, inadequate energy provision in some
studies
•
Reduce substitution effect
•
•
•
•
•
•
•
•
Take home rations
Timing: give meals as snack rather than breakfast
Educational messages to parents
Give biscuits rather than cooked meal
Fortify meals with micronutrients
Add omega-3-polyunstaurated fatty acids
Locally produced foods
Possibly target better (Seems to be more
effective in underweight children)