Participatory Learning and Action – Maternal and Neonatal Health

This is an interim intervention report. We have spent limited time to form an initial view of this program and, at this point, our views are preliminary.

Summary

Published: November 2022

What is the problem?

The World Bank estimates the following mortality rates for low-income countries:

• Child mortality rate of 66 per 1,000 live births (2020)1
• Neonatal mortality rate of 26.4 per 1,000 live births (2020)2
• Maternal mortality ratio of 453 per 100,000 live births (2017)3

Some researchers believe that a low uptake of essential newborn care practices such as drying and wrapping after birth and skin-to-skin contact contributes significantly to the neonatal mortality burden, but we did not evaluate this claim.4 Some researchers also believe that a lack of timely care-seeking contributes significantly to neonatal mortality.5

What is the program?

PLA-MNH (Participatory Learning and Action – Maternal and Neonatal Health) consists of running facilitated meetings of groups of women of reproductive age, especially pregnant women, to develop localized strategies to increase appropriate care-seeking and improve uptake of prevention practices aiming to improve maternal and newborn health.6

Within one trial, for example, every group met monthly for a total of 20 meetings, facilitated by a community-appointed local woman.7 Facilitators used localized case studies, picture-card games, role-play, and stories to help group members identify and prioritize maternal and newborn health problems in the community.8 Members were encouraged to recognize the causes and effects of typical problems in mothers and infants, and to devise strategies for prevention, homecare support, and consultations.9 For instance, members discussed clean delivery practices and care-seeking behavior.10 More details can be found here.

Does the program have strong evidence of effectiveness?

The evidence for PLA-MNH's effectiveness in reducing neonatal and maternal mortality is strong, based on a meta-analysis (Prost et al. 2013) including seven trials and approximately 119,000 births11 and one additional trial including approximately 7,200 births published after the meta-analysis (Tripathy et al. 2016).12 In studies with participation rates above 30%, the meta-analysis finds that PLA-MNH reduces neonatal mortality by 33% and maternal mortality by 49%.13 Results from Tripathy 2016 largely confirm the results of the meta-analysis.14 In addition, one follow-up study in Nepal finds non-significant reductions in post-neonatal child deaths and disability (Heys et al. 2018).15

It appears likely that the rate of participation of pregnant women is a key driver in overall effect sizes (see below). Since we believe we may be able to selectively fund higher-participation programs, we base our cost-effectiveness estimates from trials with moderate-to-high participation.16

We have key uncertainties around external validity and post-neonatal health effects. We do not feel we have a good understanding of which mechanisms are driving mortality reductions and think it is likely that mechanisms of impact vary by context. In addition, we rely primarily on trial data from South Asia.17 Therefore, it is unclear to what degree our current cost-effectiveness estimates apply to sub-Saharan Africa, which we otherwise think is plausibly more cost-effective than South Asia (due to higher initial burden).18 We are also concerned about the evidence for post-neonatal health effects; the evidence is drawn from a single study with some concerns about bias.19 Because of these concerns, we use internal and external validity adjustments.20

We have not reviewed individual studies in detail; however the authors of the meta-analysis find an overall low risk of bias for included studies.21

Mechanisms

PLA-MNH likely affects neonatal and maternal mortality by changing the behavior of pregnant people and/or birth attendants.22 A 2017 analysis of data from some of the above trials23 (Seward et al. 2017) found that PLA-MNH changed home care and home delivery practices, improving use of safe delivery kits, sterile blades for cutting the umbilical cord, washing of hands by birth attendants, delayed bathing of newborns, and wrapping of newborns within 10 minutes of delivery.24

Among trials with data on mechanisms, increases in the above behaviors ranged from 0 to 35 percentage points and there was substantial heterogeneity among trials in terms of which health behaviors showed improvements.25

The analysis did not find any pooled effects on uptake of antenatal care, facility delivery, initiating breastfeeding within one hour, or exclusive breastfeeding for six weeks after delivery, although individual trials found improvements in some of these practices.26

There is some evidence consistent with spillover effects of the PLA-MNH program on pregnant people living in intervention areas who did not attend meetings. Specifically, changes in behaviors occurred in both attendees and non-attendees in the intervention arm, making such large mortality effects more plausible.27 This may be a result of attendees sharing healthcare information with non-attendees, although this is speculative.

We do not feel we have a good understanding of which mechanisms are driving mortality reductions and think it is likely that mechanisms of impact vary by context. It is unclear to us how we would monitor program data to predict mortality effects in implementation settings. This is a priority for further research.

Participation rate

Participation rate appears to be a key driver of program impact. Since we believe we may be able to selectively fund programs with participation rates over 30%, we use meta-analysis results from trials with a participation rate above 30%.

The PLA-MNH trials attained participation rates among pregnant people ranging from 2% to 66%.28 It seems intuitive that groups with lower rates of participation would cause smaller reductions in mortality, given that they would encourage behavior change in a smaller portion of the population. This intuition is supported by meta-regressions showing associations between participation rates and neonatal and maternal mortality reductions.29

Further evidence that participation likely drives effect size comes from two trials in the same context which found larger mortality reduction in the trial with higher participation. Fottrell et al. 2013 is a follow-up of the Azad et al. 2010 trial in which the population per group was reduced to increase participation.30 This trial found a reduction in neonatal mortality of 38% versus 10% in Azad et al. 2010, indicating that higher participation may drive stronger effects.31

We believe we may be able to predict participation rates prior to implementing a PLA-MNH program by reducing group sizes, similar to the Fottrell et al. trial above. We are interested in funding monitoring and evaluation to confirm and improve our ability to predict participation rates prior to implementation.

To account for the possibility that we cannot predict participation rates and that participation rates don’t drive higher effect sizes, we incorporate an internal validity adjustment in our cost-effectiveness estimates.32

Evidence on neonatal mortality

Prost et al. 2013 finds that PLA-MNH reduces neonatal mortality by 33% (OR 0.67 95% CI 0.60-0.75) among trials with a participation rate above 30%. Within that set of trials, the reduction ranges from 29% to 41%.33

Evidence on maternal mortality

Prost et al. 2013 finds that PLA-MNH reduces maternal mortality by 49% (OR 0.51 95% CI 0.29-0.89) among trials with a participation rate above 30%. Within that set of trials, the reduction ranges from 26% to 80%.34

Evidence on post-neonatal mortality

Overall, we interpret the evidence for post-neonatal mortality to be weak. The evidence is drawn from a single study with some concerns about bias.35 Because of these concerns, we use substantial internal and external validity adjustments.36

The study (Heys et al. 2018) is a follow-up of the randomized controlled trial (RCT) of PLA-MNH groups in Nepal from 2001 to 2003.37 In follow-up interviews "a mean 11.5 years later" with 73% of original trial participants, it found non-statistically significant reductions in child death (after the neonatal period), child disability, and maternal deaths.38 While the study authors point to a plausible mechanism by which PLA-MNH might reduce post-neonatal child mortality,39 we have substantial uncertainty about the size of this benefit.

Does the program work at scale?

A "pragmatic cluster non-randomised controlled trial of women’s groups practising PLA scaled up by government front-line workers" in India finds a 24% reduction in neonatal mortality,40 as well as an incremental cost-effectiveness ratio of $1,272 per neonatal death averted.41 At face value, this increases our confidence in the feasibility of implementing PLA-MNH at scale, but we have not evaluated this study in detail.

External validity

We have a moderate level of uncertainty about the extent to which findings from trials will generalize to future implementation settings, though we haven’t explored this issue in depth yet.

Given that the exact strategies implemented in trials are localized, their effects may vary considerably across settings. Given our lack of understanding of which mechanisms drive mortality reductions, we are uncertain about how mortality effects will vary across settings.

Some relevant variables we expect to vary across settings include:

• Baseline mortality rates
• Neonatal mortality rates attributable to inadequate home care practices and care-seeking, which may be moderated by e.g.:
  • The percent of births taking place in facilities
  • The availability of high-quality antenatal care
  • Baseline prevalence of proper practices such as safe delivery kits, sterile blades for cutting the umbilical cord, washing of hands by birth attendants, delayed bathing of newborns, and wrapping of newborns within 10 minutes of delivery
• Availability of trained and motivated meeting facilitators

We have data from several different program settings across Nepal, Bangladesh, India, and Malawi.42 This increases our confidence that the program’s effects will reproduce in novel settings. However, given that we expect to primarily consider funding programs in sub-Saharan Africa and most of the trial evidence for this intervention comes from South Asia, we have some remaining reservations on replicability.

Specifically, the trials in sub-Saharan Africa appear to have a higher per-person cost than the trials in South Asia.43 It is unclear whether this implies programs will be more expensive in sub-Saharan Africa, but is possibly suggestive.

To account for these uncertainties around replicability and cost, we include an external validity adjustment of 0.85 for the neonatal/maternal mortality reductions we model.44

How cost-effective is the program?

We conducted a preliminary cost-effectiveness analysis. We found that PLA-MNH’s cost-effectiveness may be within the range of programs we would recommend funding.

Note that our cost-effectiveness analyses are simplified models that do not take into account a number of factors. There are limitations to this kind of cost-effectiveness analysis, and we believe that cost-effectiveness estimates such as these should not be taken literally due to the significant uncertainty around them. We provide these estimates (a) for comparative purposes and (b) because working on them helps us ensure that we are thinking through as many of the relevant issues as possible.

PLA-MNH relies on behavior change through large community meetings. Given a large mortality impact and a low cost per individual, the intervention is relatively cost-effective.

A sketch of the cost-effectiveness model for Nigeria is below:

• Prevalence of the problem. We estimate that approximately 3.6% of neonates, approximately 1% of new mothers, and approximately 8% of children under 5 who survive the neonatal period die each year in Nigeria.45
• Effect of the intervention on reducing mortality. We estimate that, after adjusting for internal and external validity concerns, PLA-MNH reduces neonatal mortality by 29%,46 maternal mortality by 43%,47 and post-neonatal child mortality by 6%.48
• Cost of the program. We estimate that the program would cost approximately $2.50 (USD) per person in the intervention area (including non-pregnant people), per year,49 or approximately $64 per birth.50
• Cost-effectiveness. Our best guess is that PLA-MNH is approximately 11 times as cost-effective in Nigeria as cash transfers from GiveDirectly.51 This is within the range of programs we would recommend funding.

We are uncertain about the following aspects of our model:

• Mortality reduction in post-neonatal children. As noted above, our estimate of the reduction in post-neonatal mortality is drawn from a single study with limitations. We have attempted to adjust for the quality of evidence in our model, but the proper magnitude of this adjustment is unclear.
• Costs. Our cost estimates are based on the academic trials of PLA-MNH’s effectiveness, which showed significant variation in per person costs.52 Our estimates are highly uncertain and we expect to refine them with specific implementation proposals. It is unclear to what extent these costs would replicate in non-trial settings.

Is there room for more funding?

Though PLA-MNH has likely been implemented at scale in at least one country,53 we are not aware of other scaled implementations.

We spoke with a non-governmental organization that provides technical assistance to implement PLA-MNH programs, Women and Children First (WCF). WCF claims the main bottleneck to community health worker (CHW) delivery is the government's logistical capacity to add the intervention to existing CHW platforms.54 However, WCF is optimistic about the government's logistical capacity to implement the program in Malawi and Nigeria.55

Key questions for further investigation

• Where and how can PLA-MNH be feasibly implemented?
• What further data can we collect to demonstrate the primary mechanisms driving PLA-MNH’s reduction in neonatal/maternal mortality?
• How can we predict participation rates prior to funding a program?
• What are PLA-MNH’s effects on post-neonatal children?
• What monitoring and evaluation data predicts program impact?

Our process

• We identified the Prost et al. 2013 meta-analysis and did a light review of related literature.
• We adjusted costs from trials to estimate program costs.
• We spoke with WCF directly about implementation feasibility.

Sources

Document	Source
Akter, Dawson, and Sibbritt 2016	Source
Azad et al. 2010	Source (archive)
Fottrell et al. 2013	Source
GiveWell, GiveWell cost-effectiveness analysis of PLA-MNH (Participatory Learning & Action - Maternal and Neonatal Health), 2022	Source
GiveWell's non-verbatim summary of a conversation with Dr. Mikey Rosato and Joanna Drazdzewska, January 6, 2021	Source
Herbert et al. 2012	Source (archive)
Heys et al. 2018	Source (archive)
Nair et al. 2021	Source (archive)
Prost et al. 2013	Source (archive)
Pulkki-Brännström et al. 2020	Source (archive)
Seward et al. 2017	Source (archive)
Tripathy et al. 2010	Source (archive)
Tripathy et al. 2016	Source (archive)
Women and Children First (UK), "Participatory learning and action for maternal and newborn health technical assistance package"	Source (archive)
World Bank DataBank, Maternal mortality ratio (modeled estimate, per 100,000 live births) - low income	Source (archive)
World Bank DataBank, Maternal mortality ratio (modeled estimated, per 100,000 live births) – Nigeria	Source (archive)
World Bank DataBank, Mortality rate, neonatal (per 1,000 live births) - low income	Source (archive)
World Bank DataBank, Mortality rate, neonatal (per 1,000 live births) – Nigeria	Source (archive)
World Bank DataBank, Mortality rate, under-5 (per 1,000 live births) - low income	Source (archive)

• 1

  World Bank DataBank, Mortality rate, under-5 (per 1,000 live births) - low income (accessed July 11, 2022).
  

• 2

  World Bank DataBank, Mortality rate, neonatal (per 1,000 live births) - low income (accessed July 11, 2022).

• 3

  World Bank DataBank, Maternal mortality ratio (modeled estimate, per 100,000 live births) - low income (accessed July 11, 2022).

• 4

  "Essential Newborn Care (ENC) practices can contribute to decreasing the incidence of neonatal morbidity and mortality. Such ENC practices, as recommended by the World Health Organization, include drying (wiping) and wrapping the newborn immediately after birth, initiating skin-to-skin (STS) contact, dry cord care (not applying any potentially harmful substance to the umbilical cord), immediate initiation of breastfeeding and delayed bathing (for at least 6 hours)." Akter, Dawson, and Sibbritt 2016, pg. 3.

• 5

  "In LMIC settings most babies are born at home so inappropriate and delayed care seeking can contribute substantially to neonatal mortality." Herbert et al. 2012, p. 15.
  

• 6

  "A third approach involved women's groups in a four-phase participatory learning and action cycle. Phase 1 was to identify and prioritise problems during pregnancy, delivery, and post partum; phase 2 was to plan and phase 3 implement locally feasible strategies to address the priority problems; phase 4 was to assess their activities. Women's groups aimed to increase appropriate care-seeking (including antenatal care and institutional delivery) and appropriate home prevention and care practices for mothers and newborns." Prost et al. 2013, p. 1736.

• 7

  "Every group met monthly for a total of 20 meetings, and a local woman, selected on the basis of criteria (including speaking the local language and having the ability to travel to meetings) identified by the community, facilitated the meetings. After a 7-day residential training course to review the cycle’s contents, and to practice participatory communication techniques, facilitators were given support through fortnightly meetings with district coordinators. Facilitators coordinated an average of 13 meetings every month with as many groups." Tripathy et al. 2010, p. 1185.

• 8
  • "Information about clean delivery practices and care-seeking behaviour was shared through stories and games, rather than presented as key messages. By discussion of case studies imparted through contextually appropriate stories, group members identified and prioritised maternal and newborn health problems in the community, collectively selected relevant strategies to address these problems, implemented the strategies, and assessed the results." Tripathy et al. 2010, p. 1185.
  • "Groups used methods such as picture-card games, role play, and story-telling to help discussions about the causes and effects of typical problems in mothers and infants, and devised strategies for prevention, homecare support, and consultations." Tripathy et al. 2010, p. 1185.

• 9

  "Groups used methods such as picture-card games, role play, and story-telling to help discussions about the causes and effects of typical problems in mothers and infants, and devised strategies for prevention, homecare support, and consultations." Tripathy et al. 2010, pg. 1185.

• 10

  "Information about clean delivery practices and care-seeking behaviour was shared through stories and games, rather than presented as key messages." Tripathy et al. 2010, p. 1185.

• 11

  "Seven trials (119 428 births) met the inclusion criteria." Prost et al. 2013, p. 1736.

• 12
  • "During the follow-up period (Jan 1, 2011, to Dec 31, 2012), we identified 3700 births in the intervention group and 3519 in the control group." Tripathy et al. 2016, pg. e119.
  • 3,700 + 3,519 = 7,219

• 13

  "Since the proportion of pregnant women participating in groups was a key predictor of mortality reduction, for our subgroup analyses we separated the trials into categories of high (≥30% of pregnant women participating in women's groups) and low coverage (<30% participating). Figure 4 shows that in high-coverage studies (48 333 livebirths), exposure to women's groups was associated with a 49% reduction in maternal mortality (figure 4A) and a 33% reduction in neonatal mortality (figure 4B)." Prost et al. 2013, p. 1740.

• 14

  "The neonatal mortality rate during this period was 30 per 1000 livebirths in the intervention group and 44 per 1000 livebirths in the control group (odds ratio [OR] 0.69, 95% CI 0.53–0.89)." Tripathy et al. 2016, pg. e119. This corresponds to a neonatal mortality reduction of 31%.

• 15
  • "There was a non-significant reduction in risk of reliable and estimated child and maternal death and childhood disability. For child deaths, considering deaths per 1000 person-years and weighting according to person-years of follow-up gave similar results (RR=0·65 (0·37 to 1·14), p=0·16)." Heys et al. 2018, pg. 5.
  • See also Heys et al. 2018, pg. 8, table 3, "Relative risks (RRs) weighted according to population size within clusters."

• 16

  See Prost et al. 2013, pg. 1743, Figure 4, "Subgroup analysis of the effect of women’s groups on maternal mortality (A) and neonatal mortality (B), by percentage of pregnant women participating in groups."
  

• 17

  Data used in the Prost et al. 2013 meta-analysis with ≥30% of pregnant women participating in groups are from Bangladesh, India, Nepal, and Malawi. See pg. 1743, figure 4 for a detailed list.

• 18

  See a comparison of sub-Saharan Africa (excluding high-income) and South Asia for maternal mortality here and neonatal mortality here.

• 19
  • Heys et al. 2018, pg. 8 finds a high intraclass correlation coefficient: "In fact, the ICCs for survival and disability outcomes were considerably higher than the original study and higher than we predicted, suggesting substantial intercluster variability." This can lead to a "relative lack of statistical power."
  • The study also has a risk of confounding since it is a follow-up to a randomized controlled trial that took place over a decade prior. "...it is possible that residual confounding is a factor although randomisation should have reduced this likelihood; for example, if there were an exposure that was related to the outcome such as quality of water supply, for which data were not collected and which by chance was not equally distributed between the intervention and control arm." Heys et al. 2018, pg. 10.

• 20

  See our internal and external validity adjustments in our cost-effectiveness analysis here.
  

• 21

  "The studies were of good quality and had low risk of bias, according to the standards of the CONSORT statement and Cochrane Collaboration’s tool for assessing risk of bias in randomised trials, for all items except masking of participants, personnel, and outcome assessment." Prost et al. 2013, p. 1740.
  

• 22

  "This meta-analysis suggests that women’s groups practising PLA improve key behaviours on the pathway to neonatal mortality, with the strongest evidence for home care behaviours and practices during home deliveries." Seward et al. 2017, pg. 2.

• 23

  Seward et al. 2017 analyzes all except one of the same trials as Prost et al. 2013.

• 24

  Overall, women’s groups practising PLA improved behaviours during and after home deliveries, including the use of safe delivery kits (odds ratio [OR] 2.92, 95% CI 2.02–4.22; I²= 63.7%, 95% CI 4.4%–86.2%), use of a sterile blade to cut the umbilical cord (1.88, 1.25–2.82; 67.6%, 16.1%–87.5%), birth attendant washing hands prior to delivery (1.87, 1.19–2.95; 79%, 53.8%–90.4%), delayed bathing of the newborn for at least 24 hours (1.47, 1.09–1.99; 68.0%, 29.2%–85.6%), and wrapping the newborn within 10 minutes of delivery (1.27, 1.02–1.60; 0.0%, 0%–79.2%). Seward et al. 2017, p. 2.
  

• 25

  See more details on changes in our cost-effectiveness analysis here.

• 26
  • "We did not find evidence of effects on uptake of antenatal care (OR 1.03, 95% CI 0.77–1.38; I2 = 86.3%, 95% CI 73.8%–92.8%), facility delivery (1.02, 0.93–1.12; 21.4%, 0%–65.8%), initiating breastfeeding within 1 hour (1.08, 0.85–1.39; 76.6%, 50.9%–88.8%), or exclusive breastfeeding for 6 weeks after delivery (1.18, 0.93–1.48; 72.9%, 37.8%–88.2%)." Seward et al. 2017, pg. 2.
  • See also Seward et al. 2017, pgs. 10 and 15-16, figures 1, 2, 8, and 9.

• 27

  Table 4 of the mechanism meta-analysis in Seward et al. 2017, pg. 17 shows that the odds of behavior change are generally similarly high in non-attendees in the intervention arms of trials vs the control arm.

• 28

  These ranged from ≤10% (three trials, two of which had participation rates below 3%), between 30% and 40% (three trials), and >50% (two trials). See Prost et al. 2013, pg. 1743, figure 4, "Subgroup analysis of the effect of women’s groups on maternal mortality (A) and neonatal mortality (B), by percentage of pregnant women participating in groups." See Tripathy et al. 2016, pg. e124, “The population coverage of pregnant women was high: 66% (2357/3539) of women who had given birth to a singleton infant during the assessment period in the intervention group reported ever attending a women’s group meeting.”

• 29
  • "In all but one study, the coverage of pregnant women in groups was calculated as the proportion of women who had delivered between 28 days and 8 weeks before the interview and reported ever attending a women's group, irrespective of the number of meetings attended. Results of meta-regression analyses indicated that the proportion of pregnant women participating in groups was linearly associated with reduction of both maternal and neonatal mortality (odds ratio −0·027, 95% CI −0·047 to −0·007, p=0·019; −0·011, −0·018 to −0·004, p=0·009, respectively; figure 3). We found no evidence of associations between intervention effects and the size of the population covered by a women's group, background mortality, or institutional delivery rates (appendix p 17)." Prost et al. 2013, p. 1740.
  • In our own unpublished meta-regression including Tripathy et al. 2016, we also found participation predicted ORs for neonatal and maternal mortality at conventional levels of statistical significance (p~=.001 and .003), but not for stillbirths (p =.15).

• 30
  • "We used a cluster randomized controlled trial to evaluate the effect of the participatory learning and action cycle with women’s groups when delivered at higher population coverage in 18 ‘unions’ in 3 districts (Bogra, Molavibazar, and Faridpur) of rural Bangladesh. [...] To fully assess the effect of scale-up of the women’s groups, the same intervention and control unions were used as for the earlier 2005-2007 trial. [...] In intervention areas, 648 new groups were formed by newly recruited facilitators in addition to the 162 women’s groups set up in the intervention areas as part of the 2005-2007 trial." Fottrell et al. 2013, p. 817.
  • "Total population coverage was 1 group per 309 population and 386 for the new groups, an approximately 4- to 5-fold increase in coverage relative to pre–scale-up levels. There was 1 women's group per 57 ever-married women of reproductive age, compared with 1 per 283 prior to scale-up, and 23% of the 45,330 ever-married women living in the intervention areas and of reproductive age were women's group members." Fottrell et al. 2013, p. 818.

• 31
  • "Intervention coverage at this scale, combined with active engagement with this target group of women, led to reductions in neonatal mortality of up to 38%." Fottrell et al. 2013, p. 820.
  • See “Adjusted risk ratio† (95% CI)”, Azad et al. 2010, pg. 1198, table 3, "Comparison of mortality rates in intervention and control clusters (2005–07)."

• 32

  See the internal validity adjustment in our cost-effectiveness analysis here.
  

• 33

  "A subgroup analysis of the four studies in which at least 30% of pregnant women participated in groups showed a 49% reduction in maternal mortality (0·51, 0·29–0·89) and a 33% reduction in neonatal mortality (0·67, 0·60–0·75)." Prost et al. 2013, p. 1736. See figure 4b, p. 1743 for the range of neonatal mortality reductions within the trials.
  

• 34

  "A subgroup analysis of the four studies in which at least 30% of pregnant women participated in groups showed a 49% reduction in maternal mortality (0·51, 0·29–0·89) and a 33% reduction in neonatal mortality (0·67, 0·60–0·75)." Prost et al. 2013, pg. 1736. See figure 4b, p. 1743 for the range of neonatal mortality reductions within the trials. See figure 4a, p. 1743 for the range of maternal mortality reductions within the trials.
  

• 35
  • Heys et al. 2018, p. 8 finds a high intraclass correlation coefficient: "In fact, the ICCs for survival and disability outcomes were considerably higher than the original study and higher than we predicted, suggesting substantial intercluster variability." This can lead to a "relative lack of statistical power."
  • The study also has a risk of confounding since it is a follow-up to a randomized controlled trial that took place over a decade prior.
    • "In 2001–2003, in Makwanpur, Nepal, a cluster randomised controlled trial (RCT) of community-based women’s groups practising participatory learning and action (PLA) reported improvements in newborn and maternal survival." Heys et al. 2018, p. 2.
    • "...it is possible that residual confounding is a factor although randomisation should have reduced this likelihood; for example, if there were an exposure that was related to the outcome such as quality of water supply, for which data were not collected and which by chance was not equally distributed between the intervention and control arm." Heys et al. 2018, p. 10.

• 36

  See our internal and external validity adjustments for post-neonatal child mortality in our cost-effectiveness analysis here.

• 37
  • "In 2001–2003, in Makwanpur, Nepal, a cluster randomised controlled trial (RCT) of community-based women’s groups practising participatory learning and action (PLA) reported improvements in newborn and maternal survival." Heys et al. 2018, pg. 2.
  • "Our study was a follow-up observational survey of the trial cohort." Heys et al. 2018, pg. 2.

• 38
  • "From 6075 children and 6117 mothers alive at 4 weeks post partum, 4419 children (73%) were available for interview a mean 11.5 years later." Heys et al. 2018, pg. 1.
  • See also Heys et al. 2018, pg. 8, table 3, "Relative risks (RRs) weighted according to population size within clusters."

• 39

  "Perinatal infection and obstructed delivery are risk factors for childhood disability and morbidity, and therefore potentially subsequent mortality beyond the perinatal period." Heys et al. 2018, p. 2.

• 40
  • "We did a pragmatic cluster non-randomised controlled trial of women’s groups practising PLA scaled up by government front-line workers in Jharkhand, eastern India." Nair et al. 2021, p. 1.
  • "During the 24-month evaluation period, the neonatal mortality rate was 29.1 per 1000 live births in the early arm and 39.2 in the delayed arm (table 2). This corresponded to a 24% reduction in neonatal mortality (adjusted OR (AOR) 0.76, 95%CI 0.59 to 0.98) (table 3)." Nair et al. 2021, p. 6.

• 41

  "The total and annual programme implementation costs in 20 districts were INR 318 488 929 (INT$15 017 396) and INR 94 018 573 (INT$4 433 165), respectively. Cost per person covered and per livebirths were INT$0.6 and INT$9.4, respectively. ICERs were INT$1272 (range: INT$732–INT$14 632) per neonatal death averted or INT$41 (range: INT$24–INT$475) per life year saved. The full economic evaluation will be reported elsewhere." Nair et al. 2021, p. 7.

• 42

  See Prost et al. 2013, p. 1743, figure 4 in for a list of studies in each program setting.

• 43

  See the average per-person cost of trials in South Asia and sub-Saharan Africa in our cost-effectiveness analysis here.

• 44

  See the external validity adjustment in our cost-effectiveness analysis here.
  

• 45

  See data for Nigeria in our cost-effectiveness analysis here.

• 46

  See our cost-effectiveness analysis here.

• 47

  See our cost-effectiveness analysis here.

• 48

  See our cost-effectiveness analysis here.
  

• 49

  See a full cost breakdown in our cost-effectiveness analysis here.

• 50

  ~$2.50 * 100,000 people in a cohort, divided by ~4,000 births. See our cost-effectiveness analysis here.

• 51

  See our cost-effectiveness estimate in multiples of cash transfers here.

• 52

  See our review of costs here. A rough sketch of how we are currently calculating costs:

  • We found a paper that collects and harmonizes costs data from some of the major trials included in the meta-analysis.
  • We separate the costs into start-up costs and scale-up costs.
  • We model the program over 10 years, marking the first 3 years as including startup costs, and the remaining years only including costs at scale. The startup costs are discounted by 50% and the scale-up costs are discounted by 10% to account for the much larger scale of the programs we’d fund, and the standardization achieved by scaled programs. The overall average costs of the program are driven almost entirely by scaled costs.
  • Costs among trials with participation above 30% over the 10 year modeled period are then averaged to arrive at the intervention cost of ~$2 per person in the intervention area per year.

• 53

  See Nair et al. 2021 for a trial of a scaled program in India.

• 54

  "Many countries that could otherwise benefit from PLA-MNH have overburdened health systems and cannot easily train their CHWs to conduct PLA-MNH, which can take ten to twelve days. Despite CHW strategies almost universally including some form of community engagement by CHWs, CHWs may also be unable to spare the time required to conduct the monthly group meetings that PLA-MNH involves, in addition to the other health care services they provide." GiveWell's non-verbatim summary of a conversation with Dr. Mikey Rosato and Joanna Drazdzewska, January 6, 2021.

• 55

  "The need for improvements in maternal and neonatal health in Nigeria is significant, and WCF believes its existing CHW platform is likely to be strong enough to support PLA-MNH, given the overall level of philanthropic support that has been directed toward healthcare improvements in Nigeria."

  "At a PLA-MNH webinar hosted in November, he told WCF that the government in Malawi would be interested in implementing PLA-MNH if it had the funding for it." GiveWell's non-verbatim summary of a conversation with Dr. Mikey Rosato and Joanna Drazdzewska, January 6, 2021.