This document summarizes presentations given on the effectiveness of Care Groups and interpersonal approaches. It includes:
- An overview of findings from a paper analyzing the impact of typical CSHGP projects on child health coverage indicators and under-5 mortality reduction. Significant increases in coverage and estimated additional reductions in under-5 mortality were found.
- A presentation on an analysis of Care Group projects funded by CSHGP, finding they achieved greater improvements in coverage indicators and reductions in under-5 mortality than non-Care Group projects, as modeled in the Lives Saved Tool.
- A discussion of the implementation strategies commonly used in effective projects, including frequent interpersonal communication with caretakers through outreach and home visits.
2. PRESENTATIONS
Overview and introduction of presenters (5 mins)
Presentation on the findings and analysis from Jim Ricca’s
Health Policy and Planning paper” (15 mins)
Presentation on the findings from Perry and George's
review and analysis of CSHGP Care Group projects and the
evidence regarding the effectiveness of PLA groups (Perry,
25 mins)
Q&A (15 mins)
Present on changes and features of Care Groups that are
in the newly released FSN Network Care Groups manual
(DeCoster, 15 mins)
Discussion on mechanisms of CG effectiveness, wrap-up,
and next steps (15 mins)
4. WHAT ARE CARE GROUPS?
Developed by Dr. Pieter Ernst with
World Relief/ Mozambique, and
pioneered by FH and WR for the past
decade. Now used by at least 22
organization in 20 countries.
Focuses on building teams of
volunteer women who
represent, serve, and do health
promotion with blocks of <15
households each
A community-based strategy for
improving coverage and behavior
change
Different from typical mothers groups:
Each volunteer is chosen by her
peers, and is responsible for regularly
visiting 10-15 of her neighbors.
5. MAJOR PROGRAMMATIC INPUTS
One paid Promoter (~10th grade educ.) per 700-1,200 beneficiary
households, and one Supervisor (nurse) per 7-10 Promoters.
Initial 6 day training on the Care Group model.
4-5 day training on each module, 3-4 trainings/yr for first 2 years.
Health promotion materials (e.g. flipcharts) for Promoters and
CGVs, bicycles or motorcycles for Supervisors and
Promoters, vitamin A, deworming meds, other supplies.
One Program Manager, 0.33 FTE M&E staff, 0.25-0.5 FTE HQ.
Sometimes integrated into MOH structure.
Usually no food supplements provided and few “give-aways” aside
from deworming tablets and vitamin A.
See www.CareGroupInfo.org for more details.
6. FH/Mozambique Care Group Model
Promoter #6
Promoter #3
Promoter #7
12 Leader Mothers
12 families
12 families
12 families
12 families
12 families
12 families
12 families
12 families
Promoter #5
Promoter #4
Promoters
(Paid CHWs)
Each Health Promoter
educates and motivates 5 Care
Groups. Each Care Group has
12 Care Group Volunteers
(a.k.a., Leader Mothers)
12 families
12 families
Promoter #2
Promoter #1
12 families
12 families
12 Leader Mothers
12 Leader Mothers
Each Care Group Volunteer
educates and motivates
pregnant women and mothers
with children 0-23m of age in 12
households every two weeks.
Children in households with
children 24-59m are visited every
six months.
Care
Groups
With this model, one Health Promoter can cover 720 beneficiary households.
12 Leader Mothers
12 Leader Mothers
7. CSHGP Programming Can Help
Countries Significantly Accelerate
Progress Toward MDG4
May 8, 2014
Jim Ricca
Senior Learning Advisor
MCHIP Washington
8. Presentation Overview
Analysis of typical set of pre-OR CSHGP projects:
• What are coverage increases for child health
interventions?
• What is estimated additional impact on U5MR?
• What implementation strategies are
responsible?
• What are implications for donor priorities?
8
9. Acknowledgements
Co-authors: Nazo Kureshy, Karen LeBan,
Debra Prosnitz, Leo Ryan
Also Michel Pacque, Claire Boswell, Karen
Fogg helped with key pieces of analysis
Analysis wouldn’t have been possible without
well-done & well-documented projects
9
10. Methods
Inclusion criteria:
Final evaluation within 12 months of when
analysis of done (30 projects)
Had complete baseline & final KPC (3
excluded 27 projects)
DHS data within 3 years of baseline AND 3
years of final (15 excluded 12 projects)
Confirmed all coverage data, reviewed all
project documents, interviewed manager
10
11. Logic model: Project documentation (top), implementation (middle),
and analyses done in the publication (bottom)
INPUTS ACTIVITIES OUTPUTS OUTCOMES IMPACT
USAID + NGO match
funds
USAID technical
assistance to NGO
NGO partners with
health facilities &
district health system
Underlying
epidemiological
situation
NGO partners with
community / civil
society organizations
Project strategies to
increase service
quality
Project strategies to
increase access to
services
Project strategies to
improve health-
related behaviors of
mothers / caretakers
Non-project activities
that increase quality,
access, and healthy
behaviors
Increased quality of
services
Increased access to
services
(e.g., peripheralizatio
n of services, bicycle
ambulances, etc.
Improved
determinants of
mother / caretaker
behavior
(i.e., knowledge, attitu
des), resulting in
increased demand for
services
Increased demand for
utilization of health
services (e.g.
immunization, antibiotic
s for pneumonia, etc.)
Improved health
behaviors
(e.g., EBF, ORS
use, etc.)
Decreased child
morbidity and
mortality
Projects report
population based
outcomes through
KPC surveys
Analyzed population-
based outcomes
Analyzed through
review of documents
and interviews of
NGO staff
Funded through
established NGOs ,
with same material
& technical
resources
Analyzed project
inputs &contextual
factors like health
system strength
Projects report
annually on
progress against
plan
Projects design
activities with
standard strategies
& receive expert
technical review
Summary report
compiled, using
LiST to estimate
U5MR drop
Estimated through
LiST modeling
12. Coverage increases for all interventions
significantly better than trend
12
-10
0
10
20
30
40
50
60
70
80
90
100
ANC4
TT2
IFA
IPTp
SBA
EBF
CF
PPV
VitA
ITNM
easles
DPT3Handwash
Latrine
ORT
Abxpneum
oniaM
alTreat
Project
DHS
14. Implementation Strategies
Looked at six general strategies: facility
improvement, governance groups,
interpersonal BC, outreach, CHW treatment,
local media approaches
Frequent IPC (at least monthly with a majority
of caretakers) in 10 of 12 projects through
outreach, community meetings, or HH visits –
associated with better outcomes
14
15. Conclusions – CSHGP ahead of its time
11 of 12 better than trend (p = 0.003)
How much better?
On average, U5MR decrease = 5.8%
annually vs. 2.5% in comparison areas.
15
16. Conclusions (2)
Grantee strategies operationalized Alma
Ata in a way that no set of projects before
& very few since have done
CSHGP doing Implementation Science
before it was recognized as such
16
17. Implications
Very few countries will make MDG4 targets.
Many are calling for equity-focused strategies.
These approaches (community-based, heavily
focused on IPC) approximately doubled ARR for
U5MR (which is exactly what’s needed to reach
targets).
17
18. Thank you to all who have been
involved with CSHGP for a
generation, to grantees for
phenomenal passion, dedication,
and effective work, and to
partners and beneficiaries all
over the globe!
18
19. Lives Saved Tool (LiST) Analysis of
Care Group versus Non-Care Group
Child Survival Projects
Christine Marie George, PhD, International Health, JHSPH
Emilia Vignola, MSPH Candidate, International Health, JHSPH
Jim Ricca, MD, MPH, ICF Macro
Jamie Perin, PhD, International Health, JHSPH
Henry Perry, MD, PhD, MPH, International Health, JHSPH
20. Overview
• What are Care Groups?
• Rationale
• Methods
• Findings
• Discussion
• Conclusion and next steps
21. What are Care Groups?
“A Care Group is a group of 10-15 volunteer, community-
based health educators who regularly meet together with
project staff for training and supervision. They are different
from typical mother’s groups in that each volunteer is
responsible for regularly visiting 10-15 of her neighbors,
sharing what she has learned and facilitating behavior
change at the household level. Care Groups create a
multiplying effect to equitably reach every beneficiary
household with interpersonal behavior change
communication.
http://www.caregroupinfo.org/blog/criteria
23. Rationale
• There is widespread experience with Care Group project
implementation and enthusiasm is growing among
program managers
• 23 organizations implementing Care Group projects in
20 countries
• Published articles documenting the effectiveness of Care
Groups
• Edward et al. 2007
• Perry et al., 2011
• Davis et al., 2013
24. Edward et al. 2007
Examining the evidence of the under-five mortality reduction in a
community-based programme in Gaza, Mozambique
25. Perry et al., 2011
Source: Chapter in Essentials of Global Community Health, 2011
26. Davis et al., 2013
Source: Journal of Global Health: Science and Practice, 2013
27. Study Rationale
• Many evaluations of Care Group projects exist, but no
systematic assessment of them
• More evidence of effectiveness of Care Groups is
needed
• Participatory Learning and Action (PLA) Groups have
substantial evidence of effectiveness from multiple
randomized controlled trials and a meta-analysis of
these results (and almost all of these results have been
generated by the same research group)
28. Research Questions
• Do Care Group CSHGP projects achieve greater
improvement in high-impact child survival coverage
indicators than non-Care Group projects?
• Do Care Group projects achieve greater reductions
in the under-five mortality rate than non-Care Group
projects?
31. Differences in Participatory Learning and
Action (PLA) Groups and Care Groups
Care Groups PLA Groups
Type of
empowerment
At Care Group level among Care
Group volunteers (mostly)
At village level among
pregnant women
Method of
contact
One on one through home visits
(mostly), ensuring all pregnant
women or mothers of young
children are included
At group meetings where
all pregnant women are
invited to come (with no
strategy for recruiting all
eligible women)
Type of
interventions
Maternal, neonatal and child health Maternal and neonatal
health
Process for
education and
behavior
change
“Cascade” dissemination of one key
message per round, ensuring that
the complete repertoire of
messages is covered (and with
iteration presumably the
conveyance of messages becomes
more effective)
Facilitator shares health
messages gradually while
at the same time
facilitating process for
enabling women to reflect
on how to take action
33. High-impact coverage indicators
modelled in LiST
Coverage of 4 antenatal care
visits
Multiple micronutrient
consumption during pregnancy
Skilled birth attendance
Postnatal preventive care
Exclusive breastfeeding
Appropriate complementary
feeding
Handwashing
Presence of a latrine
Antibiotic treatment of
pneumonia
Oral rehydration therapy for
diarrhea
Insecticide-treated bed net
coverage; malaria treatment;
IPTp coverage
Measles, tetanus and full
immunization coverage
Vitamin A supplementation
34. Validation of LiST
• Several reports now have validated LiST as a
measurement tool for estimating mortality impact
• Ricca et al., BMC Public Health 2011
35. Care Group Eligibility Criteria
Selection criteria: Care Groups
• Care Group projects found at:
http://www.caregroupinfo.org/blog/implementors
• Project evaluations downloaded from the MCHIP website
• DHS or MICS available for the country where the Care Group
project was conducted within 3 years of both the project
baseline and endline
• A non-Care Group child survival project conducted in the
same country within 3 years of the Care Group project where
there was also a DHS and MICS survey available within 3
years of baseline and endline.
36. Non Care Group Eligibility Criteria
Selection criteria: Non-Care Group projects
• There must be a DHS or MICS survey available within 3
years of their baseline and endline survey
• A Care Group project in the same country meeting the
criteria for inclusion
37. Eligible Child Survival Programs
• Nine care group and 12 non-care group child survival
projects met these study eligibility criteria.
Care Group
projects
Non-Care Group
projects
Cambodia 3 3
Kenya 1 2
Malawi 2 1
Mozambique 3 1
Rwanda 1 2
38. Excluded Child Survival Programs
Care Group projects in three countries were
excluded
• Liberia (MTI), no matching non-Care Group project
• Guatemala (Curamericas), no recent DHS survey
available
• Zambia (SAWSO), no recent DHS survey available
Non-Care Group projects excluded in one country
• Malawi (PSI) – only nationally implemented
• Malawi (STC) – no true baseline or endline surveys
available
39. Projects included in the analysis
Country Region Organization Type Project Period
Target area
children 0-59
months
Cambodia Kampong Thum
Adventist
Development Relief
Agency Non-Care Group 2001-2006 17,477
Cambodia Battambang
Catholic Relief
Services Non-Care Group 2001-2006 24,896
Cambodia Kampong Chhnang
International Relief
and Development Non-Care Group 2006-2010 6,217
Cambodia Siem Reap Red Cross Care Group 2005-2008 43,610
Cambodia Kompong Cham World Relief Care Group 1998-2002 12,167a
Cambodia Kompong Cham World Relief Care Group 2003-2007 12,875
40. Projects included in the analysis
Country Region Organization Type Project Period
Target area
children 0-59
months
Kenya Western Province
African Medical
and Research
Foundation Non-Care Group 2005-2010 31,644
Kenya Rift Valley HealthRight Non-Care Group 2006-2010 48,844
Kenya Coast Plan Care Group 2004-2009 46,354
Malawi Southern Region
International Eye
Foundation Non-Care Group 2002-2006 42,500
Malawi Northern Region World Relief Care Group 2000-2004 36,732
Malawi Northern Region World Relief Care Group 2005-2009 32,025
Mozambique Sofala
Food for the
Hungry Care Group 2006-2010 60,666
Mozambique Sofala
Food for the
Hungry Care Group 2009-2010 83,778
Mozambique
Manica and Sofala
Provinces
Health Alliance
International Non-Care Group 2002-2007 97,200
Mozambique Gaza Province World Relief Care Group 2004-2009 33,451
41. Projects included in the analysis
Country Region Organization Type Project Period
Target area
children 0-59
months
Kenya Western Province
African Medical
and Research
Foundation Non-Care Group 2005-2010 31,644
Kenya Rift Valley HealthRight Non-Care Group 2006-2010 48,844
Kenya Coast Plan Care Group 2004-2009 46,354
Malawi Southern Region
International Eye
Foundation Non-Care Group 2002-2006 42,500
Malawi Northern Region World Relief Care Group 2000-2004 36,732
Malawi Northern Region World Relief Care Group 2005-2009 32,025
Mozambique Sofala
Food for the
Hungry Care Group 2006-2010 60,666
Mozambique Sofala
Food for the
Hungry Care Group 2009-2010 83,778
Mozambique
Manica and Sofala
Provinces
Health Alliance
International Non-Care Group 2002-2007 97,200
Mozambique Gaza Province World Relief Care Group 2004-2009 33,451
42. Projects included in the analysis
Country Region Organization Type Project Period
Target area
children 0-59
months
Rwanda Butare Province Concern Non-Care Group 2001-2006 24,494
Rwanda Kibungo
International
Rescue Committee Non-Care Group 2001-2005 109,700
Rwanda Cyangugu World Relief Care Group 2001-2006 24,021
43. Model Assumptions
• Beginning under-5 mortality rate for the project is
assumed to be the same as that for the region of the
project (based on DHS data)
• LiST estimates the under-5 mortality rate at the end of
the project according to changes in coverage of key
child survival indicators
• The average annual change in under-5 mortality is
calculated taking into account the length of the project
44. -20
0
20
40
60
80
100
ANC4 TT2 IFA IPTp SBA EBF Comp
Feed
PPV Vit A ITN Meas Full
Vacc
Hand
Wash
ORT Abx
Pneum
Mal
Treat
MeanChangeinCoverage
Coverage Indicators
Care Group Projects
Non-Care Group Projects
1
6
4
9
3
2
3
5
7
8
9
7
3
2
4
8
5
8
9
8
5
9 5
9
3
8
2
6
3
3
1
0
High Impact Child Survival
Indicator Coverage Changes
45. Coverage Results
• For all 15 high-impact indicators for which change in
coverage was calculated for Care Group and non-Care
Group projects, the mean change in coverage was
greater in the Care Group projects
• However, after controlling for country, the results are of
marginal statistical significance, p=0.07 (using the
Wilcoxon signed-ranked test)
46. Coverage Results
• The difference in coverage was significantly greater for
Care Group projects (p=0.014) (ignoring clustering
effects by country)
• The probability of this result occurring by chance
(assuming no clustering effects are present) is 0.0007.
47. Under Age 5 Mortality Rates (U5MR)
Country Care (N) Non-Care (N)
Cambodia -5.52% (3) -4.23% (3)
Kenya -3.78% (1) -3.21% (2)
Malawi -3.23% (2) -3.64% (1)
Mozambique -5.18% (3) -3.66% (1)
Rwanda -5.70% (1) -0.94% (2)
Average -4.68 -3.14
Estimated mean annual percent change in U5MR
48. Mean Annual Percent Reduction in
Under Age 5 Mortality (U5MR)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
Cambodia Kenya Malawi Mozambique Rwanda
Meanannualpercentreductionin
U5MR
Care Group Projects
Non-Care Group Projects
49. Summary findings
• Care Group projects have an estimated average annual
under-5 mortality decline that is 1.49 greater than the
non-Care Group projects
• The rate of decline of the estimate under-5 mortality
rate for Care Group projects is 49% greater than for
Care Group projects
• Malawi is an “outlier”
50. Discussion
• Care Group projects achieve greater changes in
coverage of key child survival interventions than non-
Care Group CSHGP projects after controlling for the
country in which the projects were implemented
51. Discussion
Is the effect due to the Care Group methodology?
• Not clear that any specific interventions achieve
higher coverage levels using Care Groups
compared to those using other approaches
• Not clear what specifically it is about the Care
Group methodology that makes it effective (or is
it the net combination of characteristics of the
methodology?)
52. Alternative explanations
• The organizations that implement Care Groups
are more effective than organizations
implementing non-Care Group projects
• The contexts in which Care Group projects are
implemented are more conducing to achieving
higher coverage levels (even after controlling
for the country of intervention)
53. Limitations
• Small number of projects included in the analysis
• Direct measures of mortality would be preferable, but
this is not feasible
54. Next steps
• Since there are increasing numbers of Care Group
projects with data for baseline and endline coverage, a
further similar analysis with larger number of projects
would be useful
• The growing evidence that Care Groups are effective
suggests that there is now a need for randomized
controlled trials involving Care Groups as one arm of an
intervention (perhaps head-to-head with PLA groups)
59. Adapted by TOPS & FSN Network
• Final draft projected for end of May 2014
• Members of Care Groups Forward Interest Group and FSN’s SBC
Task Force: Piloted sections with field
staff, reviewed, edited, added examples and additional material
• Piloted by TOPS:
• June 2013 in Arlington VA
• Liberia in July 2013
• January 2014 in Washington DC
• Uptake is promising
• PCI
• World Vision
• Counterpart International
• Oxfam
• CRS
60. What’s new/different in this version?
• Reflects experiences and examples from multiple
NGO’S
• Emphasis on Peer Support has been made explicit /
clarified
• New lessons:
• Using Formative Research to Strengthen Care
Groups
• Behavior Change and Care Groups
• What Happens in a Care Group Meeting?
• Conducting a Home Visit
• Planning for Sustainability
70. Interactive presentation on facilitation cues
Developed by Mitzi Hanold,
Food for the Hungry
http://www.caregroupinfo.org/vids/CGFacilitation/story.html
71. The TOPS Program was made
possible by the generous support of
the American people through the
United States Agency for
International Development (USAID)
Office of Food for Peace. The
contents of this presentation do not
necessarily reflect the views of
USAID or the United States
Government.
72. ATTN: COST EFFECTIVENESS
Cost per DALY averted in
FH/Mozambique CG Project: $15 (cost
per beneficiary/yr: $2.78)
Cost per DALY averted in Bangladesh
PLA Project: $220-$393 (Fottrell, 2013)
73. KEY RESULTS OF CONCERN WORLDWIDE’S
OR ON THE INTEGRATED CARE GROUP
MODEL
Tested traditional CG model with NGO workers as
Promoters vs. an “integrated” model where Burundi MOH
CHWs serve as Promoters. Clusters randomized to each
model.
Both models were successful in indicator improvement. No
significant differences between the integrated in traditional
model. 36 of 40 indicators were similar in results.
Met or surpassed all five CG operational indicators
(attendance, home visits reporting). Cost per beneficiary
was lowered $0.90/beneficiary.
Somewhat better sustainability trend (last 6m, no
Promoters) in the integrated model.
74. HOW DOES PEER EDUCATION WORK?
What are your theories on why CGs
work?
Theories of health behavior, learning and social
influence explain how peer education
approaches work. Three primary mechanisms:
Diffusion of new ideas
Changing social norms
Increasing self-efficacy / empowerment
Decreasing depression?
Empowerment/ Decreased GBV / Increased
respect?
75. WHY PEER EDUCATORS WORK:
CHANGING SOCIAL NORMS
Prominent Theorists: Albert Bandura, Robert
O’Connor
What those around us think is true is enormously
important to us in deciding what we ourselves
think is true.
One means we use to determine what is correct is
to find out what other people think is
correct, especially in terms of the way we decide
what constitutes correct behavior.
We view a behavior as more correct in a given
situation to the degree that we see others
performing it.
76. EXAMPLES OF CHANGING PERCEPTIONS
OF SOCIAL NORMS TO CHANGE
BEHAVIOR
EX: School-based antismoking program.
EX: Video for children terrified of dogs.
(Bandura, Grusec, Menlove, 1967)
EX: Video for severely withdrawn children. (Robert
O’Connor, 1972)
Catherine Genovese murder: Bystander inaction
Sign up on conserving water in the shower (“Navy
shower”) – 6% compliance. One modeler: 49% do
it. Two modelers: 67% compliance.
77. Which line is closer in length to the line on the left:
Line A, Line B, or Line C?
Click for Asch conformity experiment video
78. WRAP-UP AND NEXT STEPS
Additional questions on the model/
findings?
Are their models that you have seen that
are more effective than this in behavior
change in the same amount of time? Given
these results, should this become our
default health promotion model?
What steps do you think we should take in
further diffusing the model, especially given
that the CSHGP program has closed?
Editor's Notes
Jim Ricca works at MCHP’s HQ and has over 20 years of professional experience implementing and studying facility and community-based health programs in maternal, newborn and child health (MNCH), reproductive health, HIV/AIDS and infectious diseases.Henry Perry
All 17 interventions showed much larger average changes in coverage than concurrent trend from sub-national DHS data.
Conclusions more provisional in nature
(chapter in Essentials of Global Community Health, 2011)
2 projects that qualified were later excluded (1 was national in scope without adequate data and 1 had evaluation data that was not representative of the project area)2 Care Group projects implemented the activities in a staged fashion (expanding to a major portion of the project area after 2-3 years)
The number of projects implementing each intervention is displayed above the bars.
The volunteers gather with the Promoter, who introduces the session with the current objectives (today we will be talking about the important times to wash our hands). Then the group will play a game or sing a song as a warm-up activity (its ideal if this is fun, as well as reinforcing previous topics). Next is attendance and troubleshooting. Each volunteer has a chance to share how its going with her neighbor group, and where she may need advice or assistance. Then, the behavior change lesson through pictures…
How does peer education work? Theories of health behavior, learning and social influence explain howpeer education approaches work to effect change in people’s knowledge, attitudes, skills and behavior. Peer education usually works through the following three mechanisms: Diffusion of new ideas: Theories of diffusion suggest that change is initiated by a few key people in a group who are trusted, liked and whose views are valued by others. Others may copy their actions. Peer educators are often selected because they have this potential for influence. This is one reason that we recommend having the beneficiaries themselves choose the peer educator (as long as they meet certain, limited criteria). There’s no doubt that Care Groups exhibit their effect in part through this mechanism – just being able to get more information out to more people. But this is only part of the effect.Changing social norms: Another way that peer education works is by changing people’s social norms. A person’s social norms are their perceptions about what is seen as normal among the people whose opinions and views they respect. Peer education can expose people to role models and “make invisible behavior visible”, changing inaccurate perceptions that they may hold. There are many aspects of child care and family life that are not seen regularly by others and known by one’s peers. You can think that something done in your family is universal or even normative when it’s not. For example, many women may not know how many of their neighbors have their children fully immunized, and assume that few people do it. The peer educator can spread the word about how common the practice is (if that’s helpful) and can do other things to help make invisible behaviors visible (e.g., giving people a sticker to put on their house when they have committed to exclusive breastfeeding).Increasing self efficacy and empowerment: When training in new skills is accompanied by building an individual’s confidence in their ability to apply the skills, it is more likely that they will be more effective and empowered. Social Learning Theory explains this: When people become more confident (increasing their self-efficacy), they can take control of what happens to them and what they do. Peer education tries to increase people’s child- and self-care skills and their confidence to put those skills into practice. Part of this self-efficacy may come about through altering the person’s mood – increasing their happiness, their feelings of worth, their hope, or decreasing maternal depression.We have not measured how depression changes across the life of a project (in both the beneficiaries and the volunteers), but we need to do that. We hope to do a cluster randomized controlled trial soon where we treat the depressed women in half of a set of Care Groups for depression using a proven method -- Interpersonal Therapy for Groups – to see what effect that has improving behavior change amongst depressed women and at the population level. But there may be some improvements in mood and decreases in depression just from pulling women into groups and relationships with others (in the Neighbor Circle). We have also seen evidence of decreased gender-based violence in the volunteers and beneficiaries, and increased empowerment and respect for the volunteers but have not studied that well.We need to do a better job of measuring what changes occur in both volunteers and the women they serve aside from their knowledge, attitudes, and practices.