Pivot Madagascar
Impact score: 100
Location: Ifanadiana District, southeastern Madagascar
Focus: Preventable deaths from weak health systems
Pivot Madagascar - Fierce Philanthropy Research Report
Date: March 21, 2026 Methodology: Todd Manwaring's Social Impact Evaluation Framework Organization: Pivot (pivotworks.org)
PROMPT 1 -- Organization and Social Problem Summary
- Social Problem: Preventable deaths from weak health systems
- Population: Rural population of ~200,000 in a remote district
- Location: Ifanadiana District, southeastern Madagascar
PROMPT 2 -- Top 20 Negative Consequences of Preventable Deaths from Weak Health Systems Among Rural Populations in Southeastern Madagascar
| # | Negative Consequence |
|---|---|
| 1 | High under-5 child mortality from treatable diseases |
| 2 | High infant mortality from lack of postnatal care |
| 3 | High neonatal mortality from preventable causes |
| 4 | High maternal mortality from pregnancy and delivery complications |
| 5 | Low vaccination coverage leaving children vulnerable to outbreaks |
| 6 | Most deliveries occurring outside health facilities without skilled attendants |
| 7 | Lack of antenatal care leading to undetected pregnancy complications |
| 8 | Delayed care-seeking allowing treatable illnesses to become fatal |
| 9 | Geographic isolation -- populations living 15+ km from health facilities |
| 10 | Malaria epidemics causing high morbidity and mortality |
| 11 | Chronic malnutrition weakening children's survival capacity |
| 12 | Lack of community health workers to bridge facility gaps |
| 13 | Health facilities understaffed, undersupplied, and poorly maintained |
| 14 | Declining per capita health expenditure making care unaffordable |
| 15 | Severe health inequity -- poorest populations have worst outcomes |
| 16 | Low health literacy reducing care-seeking behavior |
| 17 | Infectious disease outbreaks (plague, measles, COVID-19) overwhelming fragile systems |
| 18 | Cyclone and climate damage destroying health infrastructure |
| 19 | Loss of household productivity from preventable illness and death |
| 20 | Premature adult mortality reducing family and community resilience |
PROMPT 3 -- Negative Consequences Classified as Intermediary or Ultimate Outcomes
| # | Negative Consequence | Outcome Type |
|---|---|---|
| 5 | Low vaccination coverage leaving children vulnerable | Intermediary |
| 6 | Deliveries outside facilities without skilled attendants | Intermediary |
| 7 | Lack of antenatal care leading to undetected complications | Intermediary |
| 8 | Delayed care-seeking allowing treatable illness to become fatal | Intermediary |
| 9 | Geographic isolation from health facilities | Intermediary |
| 12 | Lack of community health workers | Intermediary |
| 13 | Health facilities understaffed, undersupplied, poorly maintained | Intermediary |
| 14 | Declining per capita health expenditure | Intermediary |
| 15 | Severe health inequity -- poorest have worst outcomes | Intermediary |
| 16 | Low health literacy reducing care-seeking behavior | Intermediary |
| 1 | High under-5 child mortality from treatable diseases | Ultimate |
| 2 | High infant mortality from lack of postnatal care | Ultimate |
| 3 | High neonatal mortality from preventable causes | Ultimate |
| 4 | High maternal mortality from pregnancy/delivery complications | Ultimate |
| 10 | Malaria epidemics causing high morbidity and mortality | Ultimate |
| 11 | Chronic malnutrition weakening children's survival | Ultimate |
| 17 | Infectious disease outbreaks overwhelming fragile systems | Ultimate |
| 18 | Cyclone and climate damage destroying health infrastructure | Ultimate |
| 19 | Loss of household productivity from preventable illness | Ultimate |
| 20 | Premature adult mortality reducing family and community resilience | Ultimate |
PROMPT 4 -- Positive Results Shared by Pivot
| # | Negative Consequence | Outcome Type | Positive Results Shared by Organization |
|---|---|---|---|
| 5 | Low vaccination coverage | Intermediary | Yes. Vaccination coverage improved significantly: program-specific effect OR 1.96 (95% CI [1.14, 3.36]) for children 12-23 months receiving vaccinations. This was measured while national vaccination rates declined (from 62% to 49% nationally). |
| 6 | Deliveries outside facilities | Intermediary | Yes. Facility births increased with program-specific effect OR 2.14 (95% CI [1.17, 3.92]). This while national facility delivery rates barely changed (35% to 39% between 2009-2021 DHS surveys). |
| 7 | Lack of antenatal care | Intermediary | Yes. Antenatal care (1+ visit) improved with program-specific effect OR 2.61 (95% CI [1.46, 4.68]). |
| 8 | Delayed care-seeking | Intermediary | Yes. Care-seeking for child illness (under-5): facility-level effect OR 1.89 (95% CI [1.19, 3.00]). Care-seeking for all ages: OR 1.84 (95% CI [1.30, 2.59]). Nearly doubled odds of care at public providers. |
| 9 | Geographic isolation from health facilities | Intermediary | Yes. Community health workers deployed to serve populations beyond 15km from facilities. CHWs provide the majority of primary care consultations for children beyond 15km, achieving 1.5-2 consultations per capita-year. |
| 12 | Lack of community health workers | Intermediary | Yes. CHW network deployed across the district. For populations >15km from facilities, less than 15% of consultations came from health centers -- the remainder from CHWs. |
| 13 | Understaffed, undersupplied facilities | Intermediary | Yes. Health center consultations per capita increased with rate ratio 2.14 (95% CI [1.98, 2.32]) for all ages. Under-5 consultation rate ratio 1.48 (95% CI [1.37, 1.61]). 15 PHC2s and 6 PHC1s supported, plus 1 district hospital. |
| 14 | Declining per capita health expenditure | Intermediary | Partial. District health system operates at approximately $60 per capita (2018), compared to national per capita health expenditure declining from $20 to $16 (2009-2022). Service delivery accounts for nearly half of costs. |
| 15 | Severe health inequity | Intermediary | Yes. Relative concentration index declined for nearly all coverage indicators in initial catchment, indicating reduced relative inequalities. Mixed results for absolute inequalities (slope index of inequality). |
| 16 | Low health literacy | Intermediary | Partial. Care-seeking improvements (nearly doubled odds) suggest improved health literacy. No standalone health knowledge metrics reported. |
| 1 | High under-5 child mortality | Ultimate | Yes. Under-5 mortality decreased approximately 20-30 deaths per 1,000 live births from 2014-2023 in initial HSS catchment. Initial U5 mortality was over twice the national average. By study end, all child mortality rates converged with national averages. 30% decrease in under-5 deaths over the decade. |
| 2 | High infant mortality | Ultimate | Yes. Consistent decreases in infant mortality across the 10-year study period in HSS catchment. Comparison area saw increases through 2021, then declined after HSS expansion. |
| 3 | High neonatal mortality | Ultimate | Yes. Neonatal mortality decreased 20-30 deaths per 1,000 live births in HSS catchment. Rest of district saw increases. |
| 4 | High maternal mortality | Ultimate | Yes. Postnatal care within 48 hours improved: OR 2.08 (95% CI [1.14, 3.79]). Co-coverage index (5+ interventions): OR 2.23 (95% CI [1.11, 4.50]). Direct maternal mortality reduction data captured in overall mortality decline. |
| 10 | Malaria epidemics | Ultimate | Partial. Improved care-seeking and treatment capacity addresses malaria. National malaria incidence and mortality increased over 75% versus 2015, but Pivot's catchment showed improved outcomes. No malaria-specific outcome data isolated. |
| 11 | Chronic malnutrition | Ultimate | Partial. Improved healthcare access addresses malnutrition screening and treatment. No standalone malnutrition outcome data reported. |
| 17 | Infectious disease outbreaks | Ultimate | Yes. Mortality reductions sustained even through plague, measles, COVID-19, and cyclone damage, demonstrating health system resilience. |
| 18 | Cyclone and climate damage | Ultimate | Yes. Health system maintained function and continued mortality reductions through "one of the strongest cyclones ever recorded," demonstrating infrastructure resilience. |
| 19 | Loss of household productivity | Ultimate | No direct results shared on productivity impact. |
| 20 | Premature adult mortality | Ultimate | Partial. Overall mortality declined. No adult-specific mortality data isolated beyond maternal. |
PROMPT 5 -- Counterfactual Results Shared by Pivot
| # | Negative Consequence | Outcome Type | Positive Results Shared | Counterfactual Results Shared |
|---|---|---|---|---|
| 5 | Low vaccination coverage | Intermediary | Yes. OR 1.96 for program-specific effect. | Yes. Quasi-experimental design compares program catchment to rest of district and national trends. National vaccination rates declined (62% to 49%) while program areas improved. OR 1.96 (95% CI [1.14, 3.36]) represents effect relative to comparison area. |
| 6 | Deliveries outside facilities | Intermediary | Yes. OR 2.14 for program effect. | Yes. National facility delivery rates barely changed (35% to 39%, 2009-2021 DHS) while program areas showed OR 2.14 (95% CI [1.17, 3.92]) improvement relative to comparison. |
| 7 | Lack of antenatal care | Intermediary | Yes. OR 2.61 for program effect. | Yes. OR 2.61 (95% CI [1.46, 4.68]) represents program-specific effect relative to comparison area in quasi-experimental design. |
| 8 | Delayed care-seeking | Intermediary | Yes. OR 1.89 for child illness. | Yes. OR 1.89 (95% CI [1.19, 3.00]) for under-5 care-seeking and OR 1.84 (95% CI [1.30, 2.59]) for all ages represent facility-level effects relative to non-program areas. |
| 9 | Geographic isolation | Intermediary | Yes. CHWs serve populations >15km. | Partial. CHW impact documented through utilization data showing majority of consultations for remote populations come from CHWs. No randomized counterfactual. |
| 12 | Lack of CHWs | Intermediary | Yes. CHW network deployed. | Partial. Utilization data shows CHWs provide majority of care >15km from facilities. Quasi-experimental design captures CHW contribution within overall health system effects. |
| 13 | Understaffed facilities | Intermediary | Yes. Consultation rate ratio 2.14. | Yes. Rate ratio 2.14 (95% CI [1.98, 2.32]) for health center consultations per capita represents effect relative to pre-intervention baseline and comparison areas. |
| 14 | Declining health expenditure | Intermediary | Partial. $60 per capita district cost. | Partial. Comparison to declining national expenditure ($20 to $16 per capita) provides context but not a controlled counterfactual. |
| 15 | Severe health inequity | Intermediary | Yes. Reduced relative inequalities. | Yes. Relative concentration index comparisons over time within the program area show inequality reductions, using pre-post within the quasi-experimental framework. |
| 16 | Low health literacy | Intermediary | Partial. Improved care-seeking. | No standalone counterfactual on health literacy. |
| 1 | High under-5 child mortality | Ultimate | Yes. 30% decrease over 10 years. | Yes. Quasi-experimental longitudinal cohort (IHOPE): 1,522 households, ~8,000 individuals, five survey waves over 10 years (2014-2023), 94-96% follow-up rates. U5 mortality decreased in program catchment while increasing in comparison area (rest of district) and nationally. By study end, program area mortality converged with national averages from a starting point over twice the national rate. Published in PLOS Medicine (October 2025). Not a randomized trial but among the most rigorous quasi-experimental evaluations of health system strengthening in Africa. |
| 2 | High infant mortality | Ultimate | Yes. Consistent decreases. | Yes. IHOPE cohort shows consistent infant mortality decreases in program catchment while comparison area saw increases through 2021. Phased expansion allows dose-response analysis. |
| 3 | High neonatal mortality | Ultimate | Yes. 20-30/1,000 decrease. | Yes. Neonatal mortality decreased in program catchment while rest of district saw increases. Comparison provides quasi-experimental counterfactual. |
| 4 | High maternal mortality | Ultimate | Yes. Postnatal care OR 2.08. | Partial. Maternal health service uptake has quasi-experimental counterfactual (ORs with CIs), but direct maternal mortality counterfactual not isolated. |
| 10 | Malaria epidemics | Ultimate | Partial. Improved treatment capacity. | Partial. National malaria incidence increased 75%+ versus 2015, providing context for program area performance, but no malaria-specific counterfactual. |
| 11 | Chronic malnutrition | Ultimate | Partial. Improved access. | No counterfactual. |
| 17 | Infectious disease outbreaks | Ultimate | Yes. Maintained reductions through outbreaks. | Partial. Sustained mortality reductions during crises compared to national trends provides implicit counterfactual evidence of system resilience. |
| 18 | Cyclone and climate damage | Ultimate | Yes. System resilience maintained. | Partial. Performance during cyclone compared to broader context. |
| 19 | Loss of household productivity | Ultimate | No direct results. | No counterfactual. |
| 20 | Premature adult mortality | Ultimate | Partial. Overall mortality declined. | No adult-specific counterfactual. |
SUMMARY REPORT
Section 1 -- Our Recommendation
We strongly recommend Pivot for donor support. This organization demonstrates one of the most rigorously evaluated health system strengthening programs in sub-Saharan Africa. Working in partnership with Harvard Medical School researchers and Madagascar's Ministry of Public Health since 2014, Pivot has built a model health district in Ifanadiana, achieving a 30% reduction in under-5 mortality over 10 years -- during a period when national health indicators in Madagascar deteriorated significantly. The evidence base is exceptional: a decade-long longitudinal cohort study (IHOPE) with 1,522 households, five survey waves, and 94-96% follow-up rates, published in PLOS Medicine. While not a randomized controlled trial, the quasi-experimental design with within-district comparison areas and national trend data provides strong counterfactual evidence. The program shows statistically significant improvements across virtually every major health indicator: vaccination coverage, facility births, antenatal and postnatal care, care-seeking behavior, and health equity. Critically, these gains were sustained through devastating crises including one of the strongest cyclones ever recorded, plague and measles outbreaks, and COVID-19. The phased expansion model -- now scaling to the broader Vatovavy Region (~1 million people) at the government's request -- demonstrates both sustainability and government ownership. At approximately $60 per capita, the model provides a concrete proof-of-concept for comprehensive health system strengthening in the world's poorest settings.
Seven-Point Evaluation:
- a. Understands the Social Issue
- b. Has Intermediate Outcome Goals
- c. Has Ultimate Outcome Goals
- d. Measures these Intermediate & Ultimate Outcomes
- e. Measures Intermediate Counterfactuals
- f. Measures Ultimate Counterfactuals
- g. Evidence of Continual Learning & Adaptation
Section 2 -- The Social Problem
Madagascar is one of the poorest countries in the world, with health indicators that have been declining for over a decade. Per capita health expenditure fell from $20 to $16 between 2009 and 2022. National vaccination coverage for children 12-23 months dropped from 62% to 49%. Facility delivery rates barely increased (35% to 39%). Malaria incidence and mortality increased over 75% versus 2015 levels. In Ifanadiana District in southeastern Madagascar, the situation was even worse: under-5 mortality was over twice the national average when Pivot began work in 2014. The district's approximately 200,000 people are spread across remote, mountainous terrain where many communities are 15+ kilometers from the nearest health facility -- accessible only on foot. Health facilities were understaffed, undersupplied, and poorly maintained. Community health workers were absent. When children fell ill with malaria, pneumonia, or diarrhea, families had neither the knowledge nor the access to seek timely care. The result was devastating preventable mortality -- children dying from diseases that are entirely treatable with basic interventions, mothers dying from complications detectable through routine antenatal care, and communities trapped in a cycle of illness, poverty, and loss.
Section 3 -- The Solution
Pivot implements comprehensive health system strengthening (HSS) in partnership with Madagascar's Ministry of Public Health and researchers from the Blavatnik Institute at Harvard Medical School. The approach addresses the entire health system rather than targeting single diseases or interventions:
- Health Facility Strengthening: Pivot supports 15 PHC2s (primary health centers), 6 PHC1s (community health posts), and 1 district hospital with staffing, supplies, infrastructure, and quality improvement. The goal is to build functional, well-equipped facilities staffed by competent health workers.
- Community Health Worker Network: CHWs are deployed to reach populations living beyond 15km from health facilities. For these remote populations, CHWs provide the majority of primary care consultations, achieving 1.5-2 consultations per capita per year -- comparable to facility-based care rates.
- Demand-Side Interventions: Health education and community engagement to increase care-seeking behavior, vaccination uptake, facility delivery, and antenatal/postnatal care utilization.
- Research and Monitoring: The IHOPE longitudinal cohort (1,522 households, ~8,000 individuals) provides continuous population-level health monitoring, enabling real-time course correction and rigorous impact evaluation. This research partnership with Harvard Medical School ensures the evidence base meets the highest academic standards.
- Government Partnership and Scale: The model is designed for government ownership and scale. Following the district-level proof of concept, Pivot is now expanding to the broader Vatovavy Region (~1 million people, 3 districts) at the Ministry of Public Health's request.
Section 4 -- Key Outputs
- 200,000 people served in Ifanadiana District
- ~1 million people in expansion region (Vatovavy, 3 districts)
- 15 PHC2s, 6 PHC1s, 1 district hospital supported
- Community health workers deployed to remote communities (>15km from facilities)
- 10 years of continuous operation (2014-2024)
- ~$60 per capita health system cost (2018)
- IHOPE longitudinal cohort: 1,522 households, ~8,000 individuals, five survey waves, 94-96% follow-up rates
- 1,600 households targeted per survey wave
- 4,063-4,380 children under-5 analyzed per mortality wave
- 4,800 households surveyed for regional baseline (23,000 individuals, 2023)
- Published research: PLOS Medicine (October 2025) and multiple peer-reviewed publications
- Partnership: Harvard Medical School Blavatnik Institute, Madagascar Ministry of Public Health
- Phased expansion: Phase 1 (2014-2016, 4 communes), Phase 2 (2017-2020, 3 additional communes), Phase 3 (2021, all PHC2s)
Section 5 -- Key Intermediate Outcomes
Pivot demonstrates strong intermediate outcome measurement with quasi-experimental counterfactual evidence across multiple indicators:
- Care-seeking behavior (under-5): OR 1.89 (95% CI [1.19, 3.00]) for facility-level effect. Nearly doubled odds of seeking care at public providers. Counterfactual: measured relative to non-program areas within the quasi-experimental design.
- Care-seeking (all ages): OR 1.84 (95% CI [1.30, 2.59]). Counterfactual: measured against comparison areas.
- Health center consultations per capita: Rate ratio 2.14 (95% CI [1.98, 2.32]) for all ages. Under-5 rate ratio 1.48 (95% CI [1.37, 1.61]). Counterfactual: pre-post comparison with concurrent comparison area trends.
- Vaccination coverage (12-23 months): OR 1.96 (95% CI [1.14, 3.36]) for program-specific effect. Counterfactual: national vaccination rates declined from 62% to 49% during same period, providing strong contextual counterfactual.
- Antenatal care (1+ visit): OR 2.61 (95% CI [1.46, 4.68]). Counterfactual: quasi-experimental comparison.
- Facility births: OR 2.14 (95% CI [1.17, 3.92]). Counterfactual: national facility delivery rates barely changed (35% to 39%), while program areas showed significant improvement.
- Postnatal care within 48 hours: OR 2.08 (95% CI [1.14, 3.79]). Counterfactual: quasi-experimental comparison.
- Health equity: Relative concentration index declined for nearly all coverage indicators, indicating reduced relative inequality between rich and poor in access to health services.
Section 6 -- Key Ultimate Outcomes
Pivot demonstrates strong ultimate outcome measurement with quasi-experimental counterfactual evidence from a 10-year longitudinal cohort:
- Under-5 child mortality: 30% decrease over 10 years (2014-2023). Mortality decreased approximately 20-30 deaths per 1,000 live births in program catchment. Starting point was over twice the national average; by study end, rates converged with national averages. Counterfactual: U5 mortality increased in the comparison area (rest of district) and nationally during the same period. The divergence between program and non-program areas within the same district provides strong quasi-experimental evidence. Published in PLOS Medicine (October 2025).
- Infant mortality: Consistent decreases across the 10-year study period in program catchment. Counterfactual: comparison area saw increases through 2021, then decline after HSS expansion -- demonstrating a dose-response relationship between intervention timing and mortality reduction.
- Neonatal mortality: Decreased 20-30 deaths per 1,000 live births in program catchment. Counterfactual: rest of district saw increases, with further increase after 2021 when HSS expanded -- the new areas showing improvement after receiving the intervention reinforces causal inference.
- Health system resilience: Mortality reductions sustained through devastating cyclone damage, plague and measles outbreaks, COVID-19, and ongoing political turmoil. Counterfactual: national health indicators deteriorated during these crises while program area maintained gains, providing natural experiment evidence of system resilience.
- Co-coverage index (5+ interventions): OR 2.23 (95% CI [1.11, 4.50]) for program-specific effect, indicating comprehensive improvement across multiple health service dimensions simultaneously.
Section 7 -- Continual Learning & Adaptation
Pivot demonstrates exceptional commitment to continual learning, representing one of the strongest learning organizations in global health:
Embedded Research Partnership: The partnership with Harvard Medical School's Blavatnik Institute is not a one-time evaluation but an ongoing, embedded research program. The IHOPE longitudinal cohort has conducted five survey waves over 10 years with 94-96% follow-up rates, providing continuous population-level feedback on health outcomes.
Phased Implementation Based on Evidence: The three-phase expansion within Ifanadiana District (Phase 1: 4 communes 2014-2016, Phase 2: 3 additional 2017-2020, Phase 3: all PHC2s 2021) represents deliberate, evidence-informed scaling. Each phase expansion was informed by results from previous phases.
Dose-Response Learning: The phased design created a natural experiment: when Phase 3 expanded to previously non-program areas, those areas began showing mortality reductions -- confirming the causal link between the intervention and outcomes, and providing evidence that the model is transferable within the district.
Publication and Knowledge Sharing: Multiple peer-reviewed publications, including PLOS Medicine (2025), demonstrate commitment to contributing to the global evidence base for health system strengthening. This transparency enables external scrutiny and cross-context learning.
Government-Requested Scale-Up: The Madagascar Ministry of Public Health requested expansion to the entire Vatovavy Region (~1 million people, 3 districts), representing government validation of the model's effectiveness and sustainability. A population-representative baseline survey of 4,800 households (23,000 individuals) was conducted in 2023 to establish regional baselines.
Resilience Through Crises: The program maintained mortality reductions through multiple devastating crises (cyclone, plague, measles, COVID-19, political turmoil), demonstrating the system's robustness and the organization's adaptive capacity. The ability to sustain gains during crises provides evidence that the health system strengthening is structural, not dependent on stable conditions.
Science for Sustaining Health: Pivot explicitly frames its work as "advancing a science for sustaining health," suggesting a commitment to developing replicable, evidence-based approaches rather than just delivering services.
Report prepared using Todd Manwaring's Social Impact Evaluation Framework for Fierce Philanthropy. Sources: pivotworks.org, PLOS Medicine (October 2025), PMC/NIH, IRD le Mag', Cartier Philanthropy, Partners in Health, Medical Xpress.
Written by
