Kangaroo Mother Care

Published: February 2021

What is the problem?

Over 20 million infants are born each year with low birth weight (weighing less than 2.5kg), and low- and middle-income countries account for 96% of low birth weight (LBW) infants.1

Conventional care for low birth weight neonates, which requires expensive medical equipment, trained personnel, and permanent logistical support, is generally expensive to implement.2 Many health facilities in low- and middle-income countries lack resources to adequately treat all their patients, and a lower cost alternative to conventional neonatal intensive care may free up resources to be used elsewhere.3

What is the program?

Kangaroo mother care (KMC) is intended as a low-cost alternative to conventional neonatal intensive care for low birth weight infants during their stay at a healthcare facility and primarily involves keeping infants warm through skin-to-skin contact with their mothers.4 KMC can also include exclusive breastfeeding and relaxing the criteria for discharge from the facility in order to reduce length of hospitalization, though these components of KMC are not implemented consistently in studies on KMC.5 In contrast, conventional neonatal intensive care for low birth weight infants consists of keeping infants warm in an incubator or radiant warmer with minimal contact from their mothers.6

Standard KMC is implemented in healthcare facilities, but KMC has also been implemented in community settings,7 where community health workers visit mothers with babies at home within 72 hours of birth8 and counsel mothers of low birth weight infants on giving skin-to-skin contact and exclusive breastfeeding to their newborns.9

WHO recommends KMC implemented in healthcare facilities as routine care for LBW infants.10

There is some evidence that KMC may lead to reductions in neonatal mortality and morbidity for LBW infants.11 Some hypothesized mechanisms driving KMC’s impact on neonatal outcomes are:

• Skin-to-skin contact with the mother regulates the infant’s body temperature, reducing hypothermia.12
• Skin-to-skin contact and exclusive breastfeeding improve mother-infant attachment and lead to increased milk production.13
• Exclusive breastfeeding provides benefits to the infant's immune system, reducing infections.14

We have not investigated the plausibility of these hypotheses.

Does the program have strong evidence of effectiveness?

In this section, we discuss evidence for the impact of KMC on the following outcomes:

• Neonatal mortality: Overall, we believe there is strong evidence that KMC reduces neonatal mortality in low birth weight infants compared to conventional neonatal intensive care. According to a 2016 Cochrane meta-analysis consisting of 9 studies on standard KMC and one large and reasonably strong RCT on community-initiated KMC, KMC appears to reduce neonatal mortality in low birth weight infants by around 30%,15 which is corroborated by reductions in morbidity16 and increases in early-life growth.17 We conducted a medium-depth literature search and identified several RCTs published after the 2016 Cochrane review that found results consistent with the Cochrane review. Our main uncertainties around the findings from the Cochrane review come from whether mortality at latest follow-up is a reasonable proxy for 28-day neonatal mortality, variation in how KMC was implemented across the underlying studies, use of a fixed-effects model for a few of the meta-analyses on key outcomes of interest, and the Cochrane review rating the individual studies as having moderate to high risks of bias.
• Cost savings to hospitals: Overall, we think the evidence is compelling that KMC saves hospital costs, based on academic cost-effectiveness studies which appear to consistently find that KMC costs less for a hospital to implement than conventional neonatal intensive care. In addition, the Cochrane review finds that KMC reduces the length of hospital stays by 1.6 days, though this effect is not statistically significant. We are uncertain about the degree to which KMC results in shorter hospital stays and the resulting costs saved to hospitals.
• Developmental benefits (i.e., increases in adult income from early-life health improvements): In other early-life health interventions implemented by our top charities, we have seen evidence that effects on childhood cognitive development, growth, and morbidity may be associated with increases in productivity and income later in life. Based on the evidence for KMC’s effects on early-life morbidity and growth, our guess is that KMC may lead to developmental effects in a similar range to other early-life health interventions we’ve reviewed. However, we have a moderate level of uncertainty about whether KMC’s effect on early-life growth translates to adult earnings potential, given the lack of direct evidence that we are aware of on KMC’s effect on outcomes in adulthood, as well as some evidence that KMC has no effect on infant neurodevelopmental outcomes.
• Morbidity: Based on the 2016 Cochrane meta-analysis and smaller-scale RCTs published since 2016, we have a high level of confidence that KMC leads to substantial reductions in morbidity.

In addition to uncertainties described above, we’re also uncertain about the extent to which findings from the Cochrane review will generalize to future implementation settings.

Benefits of KMC that we model in our cost-effectiveness analysis include effects on mortality, cost savings to hospitals, developmental benefits, and reductions in morbidity. We exclude several potential additional benefits, as well as potential offsetting impacts, which we guess are small.

Main findings of Cochrane meta-analysis on mortality and morbidity

Overall, we believe there is strong evidence that KMC reduces neonatal mortality in low birth weight infants compared to conventional neonatal intensive care.

We found three meta-analyses (one published in 2010 and two in 2016) that estimate the effect of KMC on mortality.18 We believe the most credible evidence for the impact of KMC comes from a 2016 Cochrane Collaboration meta-analysis by Conde-Agudelo and Díaz-Rossello 2016,19 because it is relatively recent and only includes RCTs covering all components of standard KMC,20 and because we believe Cochrane reviews are generally of high quality.21 The other two systematic reviews on KMC find larger average effects on neonatal mortality compared to the Cochrane review.22

In addition to observational studies, the Cochrane review excludes a number of RCTs that are included in the other two systematic reviews, for the following reasons:

• The only RCT included in Lawn et al. 2010 but excluded from the Cochrane review is Sloan et al. 2008. Conde-Agudelo and Díaz-Rossello 2016 excludes this study because a substantial portion of newborns were not weighed at birth, and this portion differed significantly between the treatment and control group, making it impossible to estimate KMC’s effect on low birth weight infants.23
• Boundy et al. 2016 includes 55 RCTs,24 while the Cochrane review only includes 21 RCTs.25 Reasons for excluding studies from Conde-Agudelo and Díaz-Rossello 2016 include the trials being quasi-randomized, only evaluating KMC’s effects on healthy full-term infants, and reporting only physiological outcomes.26

The 21 RCTs included in Conde-Agudelo and Díaz-Rossello 201627 compared standard KMC to conventional neonatal intensive care in low birth weight infants.28 Conventional care mostly consisted of incubators or radiant warmers in the neonatal intensive care unit of a health facility, but in some studies the mothers in the control group held their infants through normal clothing and were encouraged to breastfeed. Otherwise, infants in the control group did not receive any enhanced care from participating in the study.29 Seven of the nine studies on mortality were conducted in low- and middle-income countries.30

The meta-analysis pools the effect of KMC on mortality at latest follow-up in nine trials, consisting of a total of 2,29331 participants and encompassing studies from 1988-2008.32 Based on these nine trials, the meta-analysis estimates that KMC reduces mortality at latest follow-up33 by 33% (95% CI = 5%-52%).34

Moreover, KMC’s effect on neonatal mortality is corroborated by beneficial effects on morbidity and exclusive breastfeeding:

• 72% reduction in hypothermia at discharge or at 40 or 41 weeks’ postmenstrual age.35
• 50% reduction in severe infection/sepsis at latest follow-up.36
• 16% increase in exclusive breastfeeding at discharge or at 40 to 41 weeks' postmenstrual age.37

Overall, we have a relatively high level of confidence that KMC reduces neonatal mortality for the following reasons:

• As noted above, we believe Cochrane reviews are generally of high quality.38
• There is low heterogeneity in mortality effects across studies.39
• The reduction in mortality is corroborated by large reductions in sepsis and hypothermia and increases in weight and length gain.40
• The Cochrane review is corroborated by more recently published RCTs. In particular, the Cochrane review finds a similar effect size on neonatal mortality (33% reduction)41 as a large-scale and high-quality RCT on community-initiated KMC (29% reduction).42 See the section on supporting evidence on mortality and morbidity.

However, a few uncertainties we have about the evidence from the Cochrane meta-analysis somewhat limit our confidence. These include:

• Mortality at latest follow-up as a proxy for neonatal mortality rate. We use mortality at latest follow-up as a proxy for 28-day neonatal mortality, where the latest follow-up in six of the nine studies was at discharge or at 40 or 41 weeks’ postmenstrual age.43 Because mortality is measured earlier than 28 days in most of the underlying studies, the reported percentage reduction in mortality from KMC could be either an overestimate or underestimate of the actual percentage reduction in mortality at 28 days of age compared to conventional neonatal care. However, there is some evidence that the large majority of neonatal mortality occurs in the first week of life,44 and average hospital stay was longer than a week for all but one study in the Cochrane review,45 so we would guess that using mortality at latest follow-up will not greatly bias our estimate of KMC’s effect on neonatal mortality.
• Variation in how KMC was implemented. The studies varied in the duration of KMC each day, breastfeeding patterns, and length of hospitalization.46 We believe the subgroup analyses mostly do not show compelling differences in effect size from these differences in KMC implementation.47 See the external validity section for more detail regarding subgroup differences.
• Use of fixed-effects models. The authors decided on whether to use a fixed- or random-effects model for each analysis based on how much heterogeneity was detected across the studies,48 and the meta-analyses on neonatal mortality,49 severe infection/sepsis,50 and exclusive breastfeeding51 use a fixed-effects model. We are uncertain whether a fixed-effects model is the appropriate choice52 and find it concerning that the authors did not decide which model to use a priori.53 In the future we might re-run these meta-analyses using random-effects models, but because of the low heterogeneity across studies for these outcomes,54 we would guess that using a random-effects model is unlikely to substantially change the pooled effect size.55
• Evidence rated as having moderate risk of bias. The authors of Conde-Agudelo and Díaz-Rossello 2016 rate most of the evidence in the meta-analysis as having a moderate risk of bias, with some studies having a high risk of bias.56 The main risks of bias identified are from lack of blinding of participants, personnel, and outcomes assessors, and from lack of details on the methods of treatment allocation concealment.57 We have not vetted these judgments. Excluding the studies with high risk of bias makes no difference to the meta-analysis results.58 Moreover, based on a light review, we believe that the placebo effect in non-blinded participants is likely weaker for objective outcomes such as neonatal mortality compared to subjective outcomes.59

Additional evidence on mortality and morbidity

We identified several other RCTs that estimate the impact of KMC on mortality or morbidity that have been published since the Cochrane review was published in 2016. We haven’t vetted any of these individual studies or incorporated them into our best guess for KMC’s effect, but they generally corroborate the results from the Cochrane review and increase our confidence in KMC’s effectiveness in reducing neonatal mortality and morbidity.

Community-initiated KMC

Mazumder et al. 2019 report on a large-scale RCT (8,402 newborns) conducted in two districts of Haryana, India from 2015-2018. They find that low birth weight newborns randomly assigned to receive community-initiated KMC had a 29% (95% CI = 3%-48%) lower likelihood of mortality at 28 days and a 24% (95% CI = 5%-40%) lower likelihood of mortality at 180 days, compared to low birth weight newborns in communities that were randomized to receive routine care (referral to hospital or standard home care).60 KMC’s effect on 28-day mortality is similar to that from the Cochrane meta-analysis (33% reduction in mortality at latest follow-up, 95% CI = 5%-52%),61 which increases our confidence in our estimate of KMC’s effect on mortality. Although we have not vetted this study in depth, it appears to be high quality, based on the large sample size,62 pre-registration of the trial and key outcome,63 balanced characteristics at baseline,64 adequate random allocation concealment,65 and low attrition in both the treatment and control groups.66

Because community-initiated KMC is implemented differently from standard KMC delivered in a health facility, we focus this report on standard KMC and do not incorporate this RCT’s results quantitatively into our estimates of KMC’s effect on mortality. In our cost-effectiveness model, we also focus on costs of the intervention as delivered in hospitals. However, we think that KMC implemented in community settings may also be promising, and we plan to model the cost-effectiveness of community-initiated KMC separately if we find a compelling opportunity to direct funding to it.

Standard KMC implemented in health facilities

We identified the following studies on hospital-based KMC published after the 2016 Cochrane review:

• Gavhane, Eklare, and Mohammad 2016 report on a small RCT (140 infants at baseline, 91 at follow-up) conducted in a tertiary care hospital in India in 2011 that compared malnutrition and growth in very low birth weight infants.67 The study did not find significant differences in malnutrition, stunting, wasting, or incidence of a small head between treatment and control infants at 6 or 12 months of age.68 Because of the small sample size and high loss to follow-up,69 we don’t place much weight on this study.
• Rahman et al. 2017 report on a small RCT (80 infants) conducted in a tertiary care hospital in Bangladesh from 2014-2015.70 They find that low birth weight newborns who receive KMC see an average weight gain of 18.1 grams/day, compared to an average of 13.0 grams/day among low birth weight newborns who receive conventional neonatal intensive care (p < 0.001).71 KMC also leads to a significant increase in exclusive breastfeeding at discharge (90% in the treatment group vs. 60% in the control group).72 The study also finds that KMC is associated with decreased apnea,73 hypothermia,74 sepsis,75 length of hospitalization,76 and mortality,77 though none of these effects are statistically significant. We put limited weight on this trial because of its small sample size, but we find its consistent and sizable effects across various outcomes to be compelling.
• Ramani et al. 2018 report on an RCT conducted in a tertiary delivery center in Zambia in 201478 that compared hypothermia in 20379 term infants who were randomized to receive some KMC with standard thermoregulation care vs. only standard thermoregulation care.80 They find no significant differences in rates of moderate or severe hypothermia at 1 hour after birth (RR=0.93, 95% CI = 0.59-1.4)81 or at discharge (RR=2.8, CI% 0.6 to 13.9)82 between treatment and control groups. However, duration of KMC was positively correlated with infants' temperature at discharge, and hypothermia was not found among infants who received KMC for at least 80% of their hospital stay.83 We find this evidence consistent with KMC reducing the likelihood of hypothermia when implemented for a long enough duration.
• Habib et al. 2019 report on an RCT in a tertiary care hospital in Pakistan from 2014-2015 that randomized 1,542 infants to three arms:84 (1) KMC, Essential Neonatal Care (ENC), and chlorhexidine on the umbilical cord, (2) ENC and chlorhexidine, and (3) ENC only.85 Neonatal mortality was a secondary outcome, and the study found negative but insignificant effects of KMC on neonatal mortality compared to the other treatment arm and control group.86 KMC also led to statistically significant reductions in neonatal infection87 and increased weight gain88 compared to both other groups.

Overall, we believe these studies corroborate the Cochrane review on KMC’s effect on neonatal morbidity and mortality.

Evidence on developmental effects

In other early-life health interventions implemented by our top charities, we have seen evidence that effects on childhood cognitive development, growth, and morbidity may be associated with increases in productivity and income later in life. Based on the evidence for KMC’s effects on early-life morbidity and growth, our guess is that KMC may lead to developmental effects in a similar range to those of vitamin A supplementation, seasonal malaria chemoprevention, and long-lasting insecticide-treated nets. However, we are moderately uncertain about KMC’s long-term effects, given the lack of direct evidence that we are aware of on KMC’s effect on outcomes in adulthood, as well as some evidence that KMC has no effect on infant neurodevelopmental outcomes.

Some evidence that factors into our guess that KMC may lead to developmental effects:

• Early-life morbidity. As described above, Conde-Agudelo and Díaz-Rossello 2016 finds that KMC leads to a 72% reduction in hypothermia at discharge and a 50% reduction in severe infection/sepsis at latest follow-up.89 In other programs implemented by our top charities, childhood morbidity seems to be associated with long-term increases in productivity and income in adulthood,90 so we would guess that KMC may also be associated with developmental effects from reducing infant morbidity.
• Early-life growth.
  • We believe there is compelling evidence that KMC increases infant growth. The Cochrane review finds positive and significant effects on daily weight gain, length gain, and head circumference gain at follow-up.91 It reports insignificant effects on weight and length at discharge or at 40 or 41 weeks’ postmenstrual age,92 but the confidence intervals around those outcomes are wide.93 Mazumder et al. 2019 also find that community-initiated KMC led to increased weight gain.94
  • In our earlier work on early-life growth interventions, studies of differences in birth weight between twins showed that an increase in birth weight leads to an increase in future consumption.95 Extrapolating from the long-term developmental effects of weight at birth, we guess that weight and length gain in the first year of life from KMC also leads to developmental benefits that increase adult earnings potential, though we are highly uncertain about the magnitude of this effect.
• Cognitive effects.
  • Taneja et al. 2020 followed up with the first 552 infants enrolled in the trial in Mazumder et al. 2019 (see Supporting evidence on mortality and morbidity) to evaluate the effect of community-initiated KMC on neurodevelopment.96 The study did not find significant differences between the treatment and control groups on any of nine cognitive, language, motor, and socio-emotional outcomes measured on the Bayley Scale of Infant Development (BSID-III) at 6 or 12 months.97 On the BSID scale, 15 points is equivalent to one standard deviation in neurodevelopment.98 The difference in means between the KMC and control groups on the composite scores at 6 months of age are as follows, with similar results at 12 months of age:99
    • Cognitive score: 0.98 (95% CI = -1.30 to 3.26)100
    • Language score: -0.20 (95% CI = -1.99 to 1.58)101
    • Motor score: 0.83 (95% CI = -1.91 to 3.57)102
    • Infant temperament score: -2.01 (95% CI = -5.07 to 1.06)103
  • Given the relatively precise null effects across the nine underlying neurodevelopmental outcomes, we find this to be compelling evidence that community-initiated KMC may not have strong effects on early-life cognitive development. However, the null effects on neurodevelopment may also reflect the difficulty in measuring differences in cognitive functioning in young infants.104
• Exclusive breastfeeding. The Cochrane review also finds that KMC leads to a 16% (95% CI = 7% to 25%) increase in exclusive breastfeeding at discharge.105 While statistically significant, this increase in exclusive breastfeeding is much smaller than what we’ve seen from other programs focused exclusively on promoting breastfeeding. We guess that exclusive breastfeeding may have long-term developmental benefits, but we have not vetted the evidence on this.106 Given the relatively small effects that KMC has on exclusive breastfeeding, we would guess that any contribution exclusive breastfeeding has to KMC’s developmental effects is minor.

Given evidence that KMC is associated with decreased morbidity, increased weight and length gain, and minor increases in exclusive breastfeeding, we think it's plausible that KMC leads to long-term developmental benefits for infants that increase adult earnings potential. Our rough guess is that 13% of KMC's benefits come from developmental effects, but we are highly uncertain about this and plan to investigate further in the future.

Evidence on cost savings for hospitals

Overall, we think the evidence is compelling that this intervention saves hospital costs, though we are uncertain about the amount it saves. We have not generally accounted for indirect costs or savings to health systems in the cost-effectiveness analyses of our top charities, but because the savings to hospitals are a direct and intended effect of KMC, we have chosen to incorporate the value of saving hospital spending into our cost-effectiveness analysis for KMC. Our guess is that costs saved to a hospital, through reductions in staff time required per neonate, length of hospitalization, and use of other facility resources, may free up resources to be directed toward providing higher quality health services or reaching more patients.

We found six academic cost-effectiveness studies on KMC, which appear to consistently find that KMC costs less for a hospital to implement than conventional neonatal intensive care, though we have not vetted any of these cost-effectiveness analyses.107 We use Broughton et al. 2013 to inform our estimate of the costs saved to a hospital from implementing KMC instead of conventional care, since the other studies don’t include initial fixed costs of implementation. Their cost estimates are based on a particularly high volume maternal hospital in Nicaragua,108 and costs to a hospital may look different in contexts where we would consider funding KMC. Broughton et al. 2013 estimates that the total average cost per neonate receiving KMC was $1,808, compared to $2,322 per neonate receiving conventional neonatal intensive care.109 Note that this is the total cost to care for a neonate receiving KMC, including hospital costs that are not directly related to providing KMC, and is distinct from the costs of providing technical assistance to support the scale-up of KMC. In actual implementation settings, a portion of the costs included in this estimate, such as the fixed costs of training, may be incurred by an implementer supporting scale-up of KMC rather than the hospital itself. In that case, the savings to a hospital from replacing conventional neonatal care with KMC may be even greater than suggested by the estimates from Broughton et al. 2013. Because hospital costs to care for a neonate may vary widely by settings, we don’t directly use these cost figures in our cost-effectiveness analysis, but rather use them to inform a subjective adjustment for benefits coming from costs saved to a hospital.110

Moreover, the Cochrane review finds that KMC reduces the length of a hospital stay by 1.6 days (95% CI = -0.2 to 3.4 days), though this effect is not statistically significant.111 We use this estimate of reduction in the length of hospital stays, combined with the estimates of costs per day of hospitalization from Broughton et al. 2013, to inform our estimate of the proportion of benefits coming from costs saved to the average hospital by implementing KMC.

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.

All the underlying studies in Conde-Agudelo and Díaz-Rossello 2016 implemented the standard form of KMC recommended by the World Health Organization. Most of the studies were implemented in hospitals with neonatal intensive care units (NICUs),112 and 16 of the 21 studies were implemented in low- or middle-income countries, including countries in South Asia and Sub-Saharan Africa.113 We guess that if we were to fund implementation of KMC, it would likely be the standard form of KMC in hospital NICUs in low- and middle-income countries, similar to most of the studies included in the Cochrane review.

However, the quality of implementation in a small-scale RCT at one hospital may be higher than when providing technical assistance to multiple hospitals to implement KMC. We adjust the effect size from the meta-analysis down in our cost-effectiveness model to account for this possibility.

We are unsure about whether the studies differ from future implementation settings in terms of population or intervention characteristics that influence responsiveness to intervention. The Cochrane review authors model how KMC’s effect on neonatal mortality varies according to the characteristics of intervention implementation, including providing KMC intermittently or continuously throughout the day, providing KMC for a shorter or longer duration each day, and whether infants were entered into a trial before or after stabilization.114 The effect size did not differ in a compelling way across subgroups for most of these analyses.115

The intervention’s impact on neonatal mortality showed statistically significant and practically meaningful differences in the following subgroup analyses:

• Infant age at initiation of KMC. Studies for which the median or mean infant age at initiation of KMC was at most 10 days have a pooled effect size of 44% reduction in mortality (95% CI = 15%-63%),116 while those in which the median or mean infant age was greater than 10 days detect no effect on mortality (3% increase, 95% CI = 47% decrease - 100% increase).117 Our impression is that the studies where the average infant age at initiation of KMC was higher reflect a difference in quality of implementation and adherence rather than a difference in intervention design.118 Because we might expect a portion of infants to be initiated on KMC after 10 days of age in future implementation settings, we don’t adjust our estimated mortality effect based on the results of this subgroup analysis.
• Low/middle-income countries vs. high-income countries. Seven of the nine studies reporting on KMC’s mortality effects were implemented in low- or middle-income countries, and the pooled mortality effect for this subgroup (35% reduction in mortality)119 is similar to the effect size across all nine studies (33% reduction).120 We expect that if we were to fund implementation of KMC, it would be in low- or middle-income country settings. Because the overall effect size is similar to that of the low- and middle-income countries subgroup, we don’t adjust our estimated mortality effect based on the results of this subgroup analysis.
• KMC duration of less than 2 hours/day. KMC had no effect on mortality (18% increase, 95% CI = 68% decrease - 330% increase) in the subgroup of three studies in which the average duration of KMC was less than 2 hours/day.121 Since WHO recommends implementing KMC with continuous and prolonged skin-to-skin contact,122 KMC would likely be implemented for more than 2 hours/day for most infants in KMC programs we would consider funding, so we don’t find the null effect in this subgroup concerning. However, we might expect KMC to be implemented for less than 2 hours/day in a portion of infants due to imperfect compliance, so we don’t adjust our estimated mortality effect to exclude this subgroup either.

There could be other factors that influence generalizability of the evidence base to other contexts, and this is an area for future investigation. For example, we are uncertain about what routine care LBW infants delivered in hospitals would receive in the settings where we would consider funding KMC. In addition, many of the studies in the Cochrane review are implemented in tertiary care level hospitals, which may provide higher quality care and serve a different population than hospitals where we might fund implementation.

Potential additional benefits

KMC might have a number of other benefits outside of mortality, morbidity, and developmental effects. A few examples of potential additional benefits are:

• Mother-infant attachment. Because KMC requires mothers to maintain skin-to-skin contact with their infant rather than keeping their infant in an incubator, KMC could improve bonding between caregivers and infants. The Cochrane review provides suggestive evidence that mother-infant attachment was greater in the KMC group than the control group,123 but we have not vetted this evidence. We also have not examined the importance of mother-infant attachment, but we speculate that improved bonding could have positive effects on the infant’s development and long-term emotional well-being, as well as the quality of caregivers’ interactions with their infants beyond their time at the hospital.
• Maternal mental health. The Cochrane review provides some suggestive evidence that KMC increases mothers’ sense of competence and reduces feelings of worry and stress.124 It seems plausible that KMC may have a positive impact on maternal mental health, but we have not vetted the evidence on this.
• Home environment. One trial included in the Cochrane review found that KMC improved the home environment for infants.125 This may lead to improved quality of care for the infant beyond their time at a hospital.
• Decreased re-admission to health facilities. We speculate that through reducing infant morbidity, KMC may result in lower rates of re-admission to health facilities, which may save additional costs to those facilities.
• Benefits associated with breastfeeding. The Cochrane review finds that KMC leads to a 16% (95% CI = 7% to 25%) increase in exclusive breastfeeding at discharge,126 and exclusive breastfeeding is associated with reduced diarrhea morbidity and potentially other benefits that are not captured in our cost-effectiveness analysis of KMC.127
• Savings to families. Access to KMC as an alternative to conventional neonatal intensive care could reduce costs to families, including reduced medical costs for NICU stays, reduced medical costs for sepsis cases averted, more income for both parents due to less work missed, and reduced funeral costs. We plan to investigate this further in the future.

We do not currently incorporate these potential benefits in our assessment of the cost-effectiveness of KMC.

Potential offsetting/negative effects

We have neither investigated nor seen evidence of potential offsetting or negative effects of replacing conventional neonatal intensive care with KMC, but this is an area for further investigation in the future. We speculate that potential negative impacts of KMC could include:

• Caregiver time. Providing KMC requires caregivers to be present at the facility for extended periods of time, which could lead caregivers to lose income or be burdened with long, costly commutes to the hospital when staying overnight at the facility is not possible.128
• Mental and physical distress on caregivers. Caregivers may develop fatigue or pain from holding their infants for long durations.
• Potential negative impacts of promoting exclusive breastfeeding. Our breastfeeding promotion programs intervention report lists a few problems we’ve speculated might arise from exclusive breastfeeding, including risk of malnourishment in a minority of infants and risk of HIV transmission to infants via breastfeeding in HIV-positive mothers without access to antiretroviral drug regimes.129
• Offsetting costs to hospitals. Facilities may incur other costs from implementing KMC that offset the savings from reducing length of hospitalization and incubator use. Needing to train caregivers to provide skin-to-skin contact to their infants could increase the workload of healthcare providers, which may be harmful in settings where there is already a shortage of staff.130 Moreover, KMC requires caregivers to be present at the facility for extended periods of time, which may contribute to crowding in facilities where there is already insufficient space.131

How cost-effective is the program?

Based on our preliminary cost-effectiveness analysis, our best guess is that KMC is within the range of cost-effectiveness of the opportunities that we expect to direct marginal donations to (about 10x cash or higher, as of 2021).132 This cost-effectiveness analysis is at an early stage, and we think it’s likely that our bottom-line cost-effectiveness estimate will change substantially if we engage with a specific giving opportunity with further review.

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. This cost-effectiveness analysis is in an early stage and therefore is not directly comparable to the cost-effectiveness analyses of our top charities. As a general rule, our estimates of a given program's cost-effectiveness tend to go down as we gain more information.

We model KMC as highly cost-effective primarily because KMC is associated with a substantial reduction in neonatal mortality among low birth weight infants, and many low- and middle-income countries, particularly in Africa and South Asia, have high baseline rates of neonatal mortality among this population.133

A sketch of the cost-effectiveness model is below:134

• Neonatal mortality rate among low birth weight infants. We are interested in funding opportunities to implement KMC in settings with high baseline rates of neonatal mortality. Based on a study in Bangladesh, a country with high neonatal mortality, we estimate the neonatal mortality rate among low birth weight infants to be around 13% in settings where we would consider funding KMC in the future.135
• Effect of KMC on neonatal mortality. We use KMC’s effect on the mortality rate at latest follow-up from the Cochrane review as a proxy for KMC’s effect on 28-day neonatal mortality rate. Our best guess is that 65% of KMC’s benefits come from averting neonatal mortality.136 We add further speculative adjustments for the internal and external validity of the evidence.137 These are highly subjective estimates which we use for the purposes of comparing the cost-effectiveness of different interventions as directly as possible to each other.
• Effect on costs for hospitals. There is some evidence that KMC is cheaper to implement for a health facility than conventional neonatal intensive care, and we assume that those costs saved to a facility are then used elsewhere. Our best guess is that 16% of the benefit comes from cost savings.138
• Additional benefits. Based on the evidence that KMC increases early-life growth and reduces infant morbidity, we assume that 13% of KMC’s benefits come from developmental effects that increase adult earnings potential and that 6% of KMC’s benefits come from direct morbidity effects.139
• Initial costs of implementing KMC. Based on an academic cost-effectiveness study that took place in a teaching hospital in Nicaragua, our best guess is that it costs $8,850 to implement KMC in one health facility, with the bulk of the cost coming from training.140 We have limited confidence in this estimate.
• Cost-effectiveness. Using an estimate of $53 per child treated,141 our best guess, as of February 2021, is that KMC is 26.2 times as cost-effective as cash transfers.142 This is in the range of cost-effectiveness of the opportunities that we expect to direct marginal donations to as of February 2021 (about 10x cash or higher).

We have high uncertainty about costs of initial implementation of KMC, baseline neonatal mortality rates among low birth weight infants, KMC’s developmental and morbidity effects, costs saved to hospitals, and the counterfactual value of hospital spending. Because changes to these parameters could significantly affect our cost-effectiveness analysis, our current best guess that KMC is within the range of programs we would recommend funding is subject to change based on new information about these parameters.

• Costs. We used estimates from an academic cost-effectiveness analysis. The costs of providing technical assistance to implement KMC will depend on the specific activities that make up the technical assistance, as well as the volume of LBW infants that are delivered at the health facility. While we have limited confidence in these estimates, we believe they are the best ones available at this stage of our investigation. We expect to be able to refine our estimates if we investigate a specific giving opportunity.
• Baseline mortality among low birth weight infants. We are interested in funding opportunities to implement KMC in settings with high baseline rates of neonatal mortality. Based on a study in Bangladesh, a country with high neonatal mortality, we estimate the neonatal mortality rate among low birth weight infants to be around 13% in settings in which we would consider funding KMC in the future.143 We did not vet this study, nor did we do a comprehensive search for literature estimating the neonatal mortality rate among LBW infants. We are highly uncertain about this estimate, and our cost-effectiveness estimate is highly sensitive to this input.
• Costs saved to hospitals and counterfactual value of hospital spending. We have not generally accounted for indirect costs or savings to health systems in the cost-effectiveness analyses of our top charities, but because the savings to hospitals are a direct and intended effect of KMC, we have chosen to incorporate the value of saving hospital spending into our cost-effectiveness model for KMC. We have a high level of uncertainty about the amount of costs saved to hospitals by switching from conventional neonatal intensive care to KMC, as well as the value of a hospital’s counterfactual spending. Under our current assumptions on these parameters, including and excluding an adjustment for costs saved to hospitals brings KMC’s overall cost-effectiveness to around 26.2 times and 21.9 times as cost-effective as cash transfers, respectively.144
• Additional benefits:
  • Morbidity effects. KMC is associated with substantial reductions in neonatal sepsis and hypothermia. Our sense is that sepsis is a severe condition145 and that hypothermia is associated with general morbidity in infants.146 As a placeholder, we’ve included a 10% adjustment to account for morbidity effects, but this is highly uncertain and an area for further investigation in the future.147
  • Developmental effects. Given evidence that KMC is associated with decreased morbidity, increased weight and length gain, and minor increases in exclusive breastfeeding, we think KMC likely leads to developmental benefits for infants that increase adult earnings potential. As a placeholder, we’ve included a 20% adjustment to account for benefits coming from long-term developmental effects, but we expect to investigate this further in the future if we identify specific KMC programs to evaluate.148
• Subjective adjustments. Our estimates of the internal and external validity of the Cochrane meta-analysis point estimates are highly subjective. We expect to investigate this further in the future, especially as we look into specific opportunities to fund KMC implementation. Moreover, we are highly uncertain about the value to apply to averting neonatal mortality relative to other outcomes that our top charities avert, and our cost-effectiveness estimate relies heavily on this parameter.

How feasible is implementation of the program?

Based on a shallow review of the feasibility of KMC implementation, our impression is that KMC would be difficult to implement with high quality on a large scale. The program requires sustained behavior change from both health providers and caregivers, and qualitative research suggests that there are major barriers to adoption, particularly in under-resourced facilities.

Barriers to adoption

KMC appears to be a relatively complex intervention with major potential barriers to scale-up.

We identified two systematic reviews of barriers and enablers to KMC implementation (Chan et al. 2016a, Chan et al. 2017). These reviews identify the following barriers to adoption:

• Burden on caregivers: KMC required caregivers to stay at the facility for extended periods of time or make long and costly commutes to the facility.149
• Crowding and lack of privacy in facilities: Many facilities lack the necessary space and equipment for caregivers to stay for extended periods of time to conduct KMC. Moreover, facilities that lacked private areas made it less likely for caregivers to feel comfortable conducting KMC.150
• Staff shortages and high staff turnover: Staff shortages and a high rate of staff turnover hindered sustained practice of KMC.151
• Staff buy-in: Many health providers did not see neonatal care in general as a high priority, and KMC in particular was perceived as a sub-standard, cheaper method of care compared to routine neonatal care.152
• Staff time: Health providers commonly believed that KMC increased their workload and limited the time they could spend attending to other newborns in the NICU.153

Based on our review of the literature on the feasibility of KMC implementation, we expect that in settings where we would be interested in supporting KMC, lack of equipment and human resources will be major barriers to sustained implementation of KMC at the facility level. However, we have a moderate level of uncertainty about the tractability of addressing these barriers to implementation, and our beliefs may change if we see evidence of high adherence to KMC in implementation settings similar to ones where we would consider recommending funding.

Slow scale-up across facilities

Moreover, there appear to be major barriers to implementing KMC on a large scale.

We reviewed 3 case studies that examined large-scale KMC implementation in low- and middle-income countries:

• Vesel et al. 2015, an analysis of barriers to scale-up in 12 countries in Africa and Asia,154 finds that countrywide adoption of KMC has been hindered by a number of major health system bottlenecks, including a lack of national scale-up plans and commitment to KMC, poor procurement and supply chain systems for KMC supplies, and logistical constraints in health system delivery.155
• Bergh et al. 2016, a case study of KMC implementation in 3 countries in Asia,156 finds that KMC implementation has been limited to a small number of facilities in each country despite efforts to encourage more widespread adoption of KMC.157 Factors that may have contributed to slow uptake include having a decentralized health system,158 lack of national scale-up plans,159 lack of buy-in,160 no systematic reporting systems for KMC,161 inadequate training,162 and competing priorities.163
• Bergh et al. 2014, a case study of KMC implementation in 4 countries in Africa164 that have received financial and/or technical support from external partners to scale-up KMC,165 provides evidence of poor sustainability of KMC services in spite of external support. In a non-random sample of facilities that reportedly provided KMC services,166 only a third of facilities showed evidence of integration of KMC into routine services, and only 5% had achieved sustainable practice of KMC.167

In light of increased interest in the acceleration of KMC implementation in recent years,168 we have a moderately high level of uncertainty about whether the feasibility of systematic scale-up of KMC has changed since the time of the case studies we reviewed. However, our impression from talking with an external KMC expert169 is that major obstacles to more systematic scale-up of KMC within national health systems persist.

Given the evidence on major bottlenecks to scale-up, as well as the substantial gaps in countrywide adoption of KMC, we think it would be challenging to implement KMC with high quality on a large scale.

Lack of data on key indicators

In addition, data on KMC coverage does not seem to be systematically collected,170 and we have seen limited evidence on the adherence to KMC in settings where it has been implemented. The lack of a global monitoring system and publicly available data on key indicators for KMC make it difficult to understand current KMC coverage levels. Given the lack of existing monitoring systems for KMC, we anticipate difficulties in conducting routine monitoring and evaluation that will tell us if the program is operating as expected on a large scale.

Is there room for more funding?

We expect there is likely room for more funding for KMC. Despite the World Health Organization recommending KMC as routine care for low birth weight infants,171 it is our impression that uptake of KMC globally remains low.172

Based on our preliminary conversations with a few organizations implementing this program, it appears that these organizations could productively absorb more funding to expand KMC.

Key questions for further investigation

• What are the costs of providing technical assistance to implement KMC, specifically in the contexts where we might be interested in funding KMC? We have limited confidence in the academic cost estimates we rely on. We would guess that the costs of implementing KMC through technical assistance vary widely depending on context and the specific activities involved in the technical assistance, and we expect to be able to refine our cost estimates once we investigate a specific giving opportunity.
• What is the magnitude of KMC’s effects on infant morbidity and adult earnings potential?
• What are the costs saved to a hospital from implementing KMC over conventional neonatal intensive care, and what is the counterfactual value of hospital spending?
• Would our best guess of KMC’s effect on mortality change if we conduct a more thorough review of the literature beyond the studies included in the 2016 Cochrane review?
• What is our best guess for the effect on 28-day neonatal mortality given that the reported effect size in the 2016 Cochrane review is on mortality at latest follow-up rather than 28-day neonatal mortality?
• How might the large tertiary care level hospitals in many of the RCTs differ from settings where we would fund KMC implementation, in terms of the targeted population, quality of healthcare, and feasibility of KMC?
• What counterfactual care would infants receive in the absence of KMC in settings where we would consider funding KMC? What proportion of these infants would receive conventional neonatal intensive care?

Our process

• Literature review of KMC’s impact. To find literature on KMC, we briefly reviewed the systematic reviews referenced on the World Health Organization page for KMC. The highest quality and most relevant review we found was a Cochrane Collaboration meta-analysis from 2016. We searched in Google Scholar and along the citation trails of these systematic reviews to find trials of KMC conducted after the Cochrane review was completed. We reviewed the Cochrane review and briefly skimmed the RCTs published since 2016 that were not included in the meta-analysis.
• Cost estimates. To obtain cost estimates, we found six academic cost-effectiveness analyses of KMC on the WHO’s page on KMC and from searching in Google Scholar. Based on a brief review, the most relevant academic cost estimates seemed to be from Broughton et al. 2013, as it included training and initial implementation costs. We did not review the other academic cost-effectiveness analyses.
• Conversations with potential implementing partners. We talked to a few organizations that implement KMC to understand what their programs consist of and where there is potential to expand KMC.

Sources

Document	Source
Andrews 2018	Source (archive)
Bergh et al. 2014	Source (archive)
Bergh et al. 2016	Source (archive)
Borenstein et al. 2009	Source
Boundy et al. 2016	Source
Broughton et al. 2013	Source (archive)
Cattaneo et al. 2007	Source
Chan et al. 2016a	Source (archive)
Chan et al. 2016b	Source (archive)
Chan et al. 2017	Source (archive)
Conde-Agudelo and Díaz-Rossello 2016	Source
Deeks et al. 2020	Source (archive)
Engmann et al. 2013	Source
Gavhane, Eklare, and Mohammad 2016	Source
GiveWell, "Criteria for evaluating programs - 2009-2011"	Source
GiveWell, Cost-effectiveness analysis: Early life growth interventions	Source
GiveWell, Intervention report: "Breastfeeding promotion programs," 2018	Source
GiveWell, Intervention report: "Mass distribution of long-lasting insecticide-treated nets (LLINs)," 2018	Source
GiveWell, The impact of early life growth on adult economic status	Source
Guenther et al. 2017	Source (archive)
Habib et al. 2019	Source (archive)
Horta and Victora 2013	Source (archive)
Kumar et al. 2009	Source
Lawn et al. 2010	Source (archive)
Mazumder et al. 2019	Source
Rahman et al. 2017	Source (archive)
Ramani et al. 2018	Source
Ruiz et al. 2017	Source
Sloan et al. 2008	Source
Taneja et al. 2020	Source (archive)
UNICEF, "Neonatal mortality," 2020	Source (archive)
Vahidi et al. 2014	Source (archive)
Vesel et al. 2015	Source (archive)
Victora et al. 2016	Source
Wartolowska et al. 2016	Source (archive)
WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019	Source (archive)
WHO, "Newborns: improving survival and well-being," 2020	Source (archive)
WHO, "Sepsis"	Source (archive)
WHO, Kangaroo mother care: a practical guide, 2003	Source (archive)
Yasmin et al. 2001	Source (archive)

• 1

  “Every year, more than 20 million infants are born weighing less than 2.5kg – over 96% of them in developing countries. These low-birth-weight (LBW) infants are at increased risk of early growth retardation, infectious disease, developmental delay and death during infancy and childhood.” WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019

• 2“Conventional neonatal care of LBW infants is expensive and needs both highly skilled personnel and permanent logistic support.” WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019
• 3“In low‐ and middle‐income countries, financial and human resources for neonatal care are limited, and hospital wards for LBW infants are often overcrowded.” Conde-Agudelo and Díaz-Rossello 2016, Pg 6.
• 4“Conventional neonatal care of LBW infants is expensive and needs both highly skilled personnel and permanent logistic support. Evidence suggests that kangaroo mother care is a safe and effective alternative to conventional neonatal care, especially in under-resourced settings and may reduce morbidity and mortality in LBW infants as well as increase breastfeeding. Kangaroo mother care involves:
  • early, continuous and prolonged skin-to-skin contact between a mother and her newborn
  • frequent and exclusive breastfeeding
  • early discharge from hospital.

  WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019

• 5“We screened 1035 articles and reports; 299 contained data on KMC and neonatal outcomes or qualitative information on KMC implementation. Eighty–eight of the studies (29%) did not define KMC. Two hundred and eleven studies (71%) included skin–to–skin contact (SSC) in their KMC definition, 49 (16%) included exclusive or nearly exclusive breastfeeding, 22 (7%) included early discharge criteria, and 36 (12%) included follow–up after discharge.” Chan et al. 2016b, Pg 1.
• 6See Conde-Agudelo and Díaz-Rossello 2016, Characteristics of studies, Pg. 32-62, for descriptions of the neonatal care given to control group infants.
• 7“Community-initiated kangaroo mother care substantially improves newborn baby and infant survival. In low-income and middle-income countries, incorporation of kangaroo mother care for all infants with low birthweight, irrespective of place of birth, could substantially reduce neonatal and infant mortality.” Mazumder et al. 2019, Pg 1
• 8“Newborn babies were identified using a pregnancy surveillance and early birth identification system and visited at home. If they weighed 1500–2250 g at home, they were enrolled as soon as possible after birth but not later than 72 h after birth.” Mazumder et al. 2019, Pg 3.
• 9“The intervention comprised promotion and support of skin-to-skin contact and exclusive breastfeeding by intervention workers and supervisors. They counselled that skin-to-skin contact be done for as long as possible during the day and night, preferably for 24 h a day, with the assistance of other family members. During home visits, the worker observed the mother practising kangaroo mother care, enquired about skin-to-skin contact and breastfeeding in the preceding 24-h period, and supported the mother and family to solve any problems or overcome barriers to effective kangaroo mother care.” Mazumder et al. 2019, Pg 4.
• 10“Kangaroo mother care is recommended for the routine care of newborns weighing 2000 g or less at birth, and should be initiated in health-care facilities as soon as the newborns are clinically stable. Newborns weighing 2000 g or less at birth should be provided as close to continuous Kangaroo mother care as possible. Intermittent Kangaroo mother care, rather than conventional care, is recommended for newborns weighing 2000 g or less at birth, if continuous Kangaroo mother care is not possible.” WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019
• 11“Compared with conventional neonatal care, KMC was found to reduce mortality at discharge or at 40 to 41 weeks' postmenstrual age and at latest follow‐up, severe infection/sepsis, nosocomial infection/sepsis, hypothermia, severe illness, and lower respiratory tract disease.” Conde-Agudelo and Díaz-Rossello 2016, Pg. 2
• 12“Studies carried out in low-income countries show that prolonged skin-to-skin contact
  between the mother and her preterm/LBW infant, as in KMC, provides effective thermal
  control and may be associated with a reduced risk of hypothermia.” WHO, Kangaroo mother care: a practical guide, 2003, Pg. 8.
• 13“SSC would allow that an infant´s demands for care may trigger neuropsychobiological paths that increase maternal behavior and immediate response to infant needs, as well as increased lactogenesis (Diaz‐Rossello 2008). In addition, KMC would empower the mother (parents or caregivers) by gradually transferring the skills and responsibility for becoming the child’s primary caregiver and meeting every physical and emotional need (Nyqvist 2010).” Conde-Agudelo and Díaz-Rossello 2016, Pg. 6.
• 14“Several mechanisms for a possible protective effect of breastfeeding against gastrointestinal infections have been proposed, including the presence in breastmilk of substances with antimicrobial or immunological properties, avoidance of contamination (as in non-human milk or baby bottles), and the general nutritional status of breastfed infants.” Horta and Victora 2013, Pg. 12
• 15
  • ”Kangaroo mother care was associated with a statistically significant reduction in risk of mortality at discharge or at 40 to 41 weeks’ postmenstrual age (3.2% vs 5.3%; RR 0.60, 95% CI 0.39 to 0.92; I2 = 0%; NNTB = 47, 95% CI 31 to 236; eight trials, 1736 infants) (Analysis 1.1), and at latest follow‐up (4.0% vs 6.0%; RR 0.67, 95% CI 0.48 to 0.95; I2 = 0%; NNTB = 50, 95% CI 32 to 331; 12 trials, 2293 infants; moderate‐quality evidence) (Analysis 1.4) (Figure 3).” Conde-Agudelo and Díaz-Rossello 2016, Pg. 16.
  • ”Between enrolment and 28 days, 73 infants died in 4423 periods of 28 days in the intervention group and 90 deaths in 3859 periods of 28 days in the control group (hazard ratio [HR] 0·70, 95% CI 0·51–0·96; p=0·027). Between enrolment and 180 days, 158 infants died in 3965 periods of 180 days in the intervention group and 184 infants died in 3514 periods of 180 days in the control group (HR 0·75, 0·60–0·93; p=0·010). The risk ratios for death were almost the same as the HRs (28-day mortality 0·71, 95% CI 0·52– 0·97; p=0·032; 180-day mortality 0·76, 0·60–0·95; p=0·017).” Mazumder et al. 2019, Pg. 1.
  • For standard KMC at latest follow-up, 1 minus RR 0.67 equals 0.33 (i.e., a reduction of 33%). For community-based KMC at 180-day follow-up, 1 minus RR 0.76 equals 0.24 (i.e., a reduction of 24%).

• 16"In stabilized LBW infants, KMC was associated with a statistically significant reduction in severe infection/sepsis at latest follow‐up (6.6% vs 13.1%; RR 0.50, 95% CI 0.36 to 0.69; I² = 24%; NNTB = 15, 95% CI 12 to 25; eight trials, 1463 infants; moderate‐quality evidence) (Analysis 1.5) (Figure 4), severe illness at six months' follow‐up (5.3% vs 17.8%; RR 0.30, 95% CI 0.14 to 0.67; NNTB = 8, 95% CI 7 to 17; one trial, 283 infants) (Analysis 1.6), nosocomial infection/sepsis at discharge or at 40 to 41 weeks’ postmenstrual age (4.0% vs 11.4%; RR 0.35, 95% CI 0.22 to 0.54; I² = 0%; NNTB = 14, 95% CI 11 to 19; five trials, 1239 infants) (Analysis 1.7), lower respiratory tract disease at six months' follow‐up (4.6% vs 12.5%; RR 0.37, 95% CI 0.15 to 0.89; NNTB = 13, 95% CI 9 to 73; one trial, 283 infants) (Analysis 1.9), and hypothermia at discharge or at 40 to 41 weeks' postmenstrual age (7.6% vs 27.1%; RR 0.28, 95% CI 0.16 to 0.49; I² = 52%; NNTB = 5, 95% CI 4 to 7; nine trials, 989 infants; moderate‐quality evidence) (Analysis 1.11)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 19.
• 17"Infants given kangaroo mother care gained more weight per day (MD 4.1 g, 95% CI 2.3 to 5.9; 11 trials, 1198 infants; moderate‐quality evidence) (Analysis 1.18) (Figure 5) and had greater increases in length (MD 0.21 cm, 95% CI 0.03 to 0.38; three trials, 377 infants) (Analysis 1.22) and head circumference (MD 0.14 cm, 95% CI 0.06 to 0.22; four trials, 495 infants) (Analysis 1.26) per week than controls."Conde-Agudelo and Díaz-Rossello 2016, Pg. 24.
• 18The three meta-analyses are:
  • Conde-Agudelo and Díaz-Rossello 2016: Kangaroo mother care to reduce morbidity and mortality in low birthweight infants (Review). Cochrane Database of Systematic Reviews.
  • Boundy et al. 2016: Kangaroo mother care and neonatal outcomes: a meta-analysis. Pediatrics.
  • Lawn et al. 2010: ‘Kangaroo mother care’ to prevent neonatal deaths due to preterm birth complications. International Journal of Epidemiology.

• 19
  • We looked at two other systematic reviews, but one included mostly observational and lower-quality studies, and the other was published in 2010 and was missing many more recent studies. We also found two additional systematic reviews on skin-to-skin contact, but these did not include the other components of KMC.
  • From Conde-Agudelo and Díaz-Rossello 2016, Pg 25:
    • "Lawn 2010 performed a systematic review and meta-analysis to estimate the effect of KMC on neonatal mortality due to direct complications of preterm birth. This review included observational studies and excluded RCTs that initiated KMC after the first week of life. In the meta-analysis of RCTs, which included three studies (Charpak 1997; Suman 2008; Worku 2005) that provided data on neonatal specific mortality, KMC was associated with a reduction in neonatal death among infants < 2000 g (risk ratio [RR] 0.49, 95% CI 0.29 to 0.82; I² = 0%; 988 infants). In the meta-analysis of three observational studies, KMC was associated with decreased risk of neonatal death in infants < 2000 g (RR 0.68, 95% CI 0.58 to 0.79; I² = 54%; 8151 infants). Another meta-analysis, which included five RCTs, showed that KMC reduced significantly the risk of severe morbidity (RR 0.34, 95% CI 0.17 to 0.65; I² = 70%; 1520 infants)."
    • "Boundy 2016 conducted a systematic review and meta-analysis of RCTs and observational studies to assess the effect of KMC on neonatal outcomes among infants of any birthweight or gestational age. Studies with fewer than 10 participants, lack of a comparison group without KMC, and not reporting a quantitative association were excluded. Among LBW infants < 2000 g, KMC was associated with a significant decrease in the risk of mortality at latest follow-up (RR 0.64, 95% CI 0.46 to 0.89; I² = 72%; 15 studies [9 RCTs and 6 observational studies])."

• 20"This review covered all randomized controlled trials (RCTs) of KMC with all its components, irrespective of duration of intervention, breastfeeding patterns, and time to discharge from hospital." Conde-Agudelo and Díaz-Rossello 2016, Pg. 6.
• 21"The Cochrane Library publishes reviews of the evidence for healthcare interventions, focusing on high-quality "micro" evaluations (as defined above). We have found that its reports generally review a large number of studies and are very clear about the findings, strengths and weaknesses of these studies. For health programs, when there are often many high-quality evaluations available, we therefore use Cochrane as our main source of information on "micro" evidence when possible." Givewell, "Criteria for Evaluating Programs - 2009-2011".
• 22"Among LBW newborns <2000 g, KMC decreased mortality at latest follow-up time by 36% (n = 15; 95% CI, 0.46 to 0.89; I² = 72%)." Boundy et al. 2016, Pg. 4.
• 23"One randomized controlled cluster trial (Sloan 2008) assessed the effect of community‐based KMC on overall neonatal mortality, infant mortality, and LBW neonatal mortality; investigators assigned 4165 infants in rural Bangladesh to community‐based KMC or control without KMC. Unfortunately, we did not include this study in the review because 40% overall and 65% of newborns who died were not weighed at birth, and missing birthweight was differential for study group." Conde-Agudelo and Díaz-Rossello 2016, Pg. 24.
• 24"Of the 124 included studies, 110 (89%) were published between 2000 and 2014 (Table 1). Seventy-six studies (61%) had 100 participants. Fifty-five (44%) were RCTs, 8 (6%) were randomized crossover trials, and 61 (49%) were observational or nonrandomized intervention studies." Boundy et al. 2016, Pg 3.
• 25"We identified 21 randomized controlled trials (3042 infants) for inclusion in this review by searching medical databases in June 2016." Conde-Agudelo and Díaz-Rossello 2016, Pg. 2.
• 26"We excluded 50 studies: 21 because they were non‐randomized trials (Ahn 2010; Arandia 1993; Bera 2014; Bergman 1994; Broughton 2013; Charpak 1994; Dala Sierra 1994; de Almeida 2010; de Macedo 2007; Feldman 2002; Gregson 2011; Ibe 2004; Kashaninia 2015; Kristoffersen 2016; Lamy Filho 2008; Legault 1995; Lincetto 2000; Lizarazo‐Medina 2012; Ohgi 2002; Silva 2016; Tallandini 2006), 10 because they included infants with birthweight ≥ 2500 g and did not report results separately for the subgroup of infants with birthweight < 2500 g (Anderson 2003; Chiu 2009; Chwo 2002; Hake Brooks 2008; Huang 2006; Karimi 2014; Lai 2006; Mörelius 2015; Samra 2015; Sloan 2008), seven because they reported only physiological outcomes (Bergman 2004; Dehghani 2015; Ludington‐Hoe 1991; Ludington‐Hoe 2000; Ludington‐Hoe 2004; Ludington‐Hoe 2006; Mitchell 2013), three because the method of generation of allocation to treatment was quasi‐randomized (Kambarami 1998; Miltersteiner 2005; Swarnkar 2016), three because allocation was performed by a cross‐over design (Legault 1993; Lyngstad 2014; Miles 2006), two because KMC was part of a preventive package of interventions for essential newborn care (Darmstadt 2006; Kumar 2008), one because it evaluated only KMC for rewarming hypothermic infants (Christensson 1998), one because it assessed only the effect of KMC on the mental health of mothers (Badiee 2014), one because it evaluated only the effect of KMC on colonization status of newborns’ nostrils (Lamy Filho 2015), and one because it was published as an abstract only, and our attempts to locate full publications or to contact study authors were unsuccessful (Udani 2008)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 13.
• 27"Twenty‐one studies, including 3042 infants, fulfilled inclusion criteria. Nineteen studies evaluated KMC in LBW infants after stabilization, one evaluated KMC in LBW infants before stabilization, and one compared early‐onset KMC with late‐onset KMC in relatively stable LBW infants. Sixteen studies evaluated intermittent KMC, and five evaluated continuous KMC." Conde-Agudelo and Díaz-Rossello 2016, Pg. 1.
• 28"Objectives: To determine whether evidence is available to support the use of KMC in LBW infants as an alternative to conventional neonatal care before or after the initial period of stabilization with conventional care, and to assess beneficial and adverse effects." Conde-Agudelo and Díaz-Rossello 2016, Pg. 1.
• 29See Conde-Agudelo and Díaz-Rossello 2016, Pg. 32-58, Characteristics of included studies for descriptions of counterfactual care.
• 30See Figure 3, Analysis 1.4.9, Conde-Agudelo and Díaz-Rossello 2016, Pg. 18.
• 31 See Summary of Findings, Conde-Agudelo and Díaz-Rossello 2016, Pg. 3.
• 32See Figure 3, Analysis 1.4.1, Conde-Agudelo and Díaz-Rossello 2016, Pg. 17. Although 12 trials report on mortality effects, 3 observe 0 deaths in both the treatment and control groups and are excluded from the pooled estimate.
• 33We used Analysis 1.1, Analysis 1.2, and Analysis 1.3 to determine the timing of latest follow up for each of the nine studies that reported mortality at latest follow up in Analysis 1.4. It appears that latest follow-up occurred at discharge or at 40 or 41 weeks’ postmenstrual age for 6 studies, at 6 months for 2 studies, and at 12 months for one study. See Conde-Agudelo and Díaz-Rossello 2016, Pg. 73-39.
• 34"At latest follow-up, KMC was associated with a significantly decreased risk of mortality (RR 0.67, 95% CI 0.48 to 0.95; 12 trials, 2293 infants; moderate-quality evidence)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 6. Although 12 trials report on mortality effects, 3 observe 0 deaths in both the treatment and control groups and are excluded from the pooled estimate.
• 35
  • "In stabilized LBW infants, KMC was associated with a statistically significant reduction in… hypothermia at discharge or at 40 to 41 weeks' postmenstrual age (7.6% vs 27.1%; RR 0.28, 95% CI 0.16 to 0.49; I² = 52%; NNTB = 5, 95% CI 4 to 7; nine trials, 989 infants; moderate‐quality evidence) (Analysis 1.11)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 19.
  • 1 minus RR 0.28 equals 0.72 (i.e., a reduction of 72%).

• 36"In stabilized LBW infants, KMC was associated with a statistically significant reduction in severe infection/sepsis at latest follow‐up (6.6% vs 13.1%; RR 0.50, 95% CI 0.36 to 0.69; I² = 24%; NNTB = 15, 95% CI 12 to 25; eight trials, 1463 infants; moderate‐quality evidence) (Analysis 1.5)..." Conde-Agudelo and Díaz-Rossello 2016, Pg. 19.
• 37"Compared with conventional care, KMC was associated with an increase in the likelihood of exclusive breastfeeding at discharge or at 40 to 41 weeks' postmenstrual age (66.3% vs 56.3%; RR 1.16, 95% CI 1.07 to 1.25; I² = 39%; NNTB = 11, 95% CI 7 to 25; six studies, 1453 mothers) (Analysis 1.31), and at one to three months' follow‐up (86.9% vs 76.5%; RR 1.20, 95% CI 1.01 to 1.43; I² = 76%; NNTB = 7, 95% CI 3 to 131; five studies, 600 mothers) (Analysis 1.32)..." Conde-Agudelo and Díaz-Rossello 2016, Pg. 21.
• 38"The Cochrane Library publishes reviews of the evidence for healthcare interventions, focusing on high-quality "micro" evaluations (as defined above). We have found that its reports generally review a large number of studies and are very clear about the findings, strengths and weaknesses of these studies. For health programs, when there are often many high-quality evaluations available, we therefore use Cochrane as our main source of information on "micro" evidence when possible." Givewell, "Criteria for Evaluating Programs - 2009-2011"
• 39"Kangaroo mother care was associated with a statistically significant reduction in risk of mortality at discharge or at 40 to 41 weeks’ postmenstrual age (3.2% vs 5.3%; RR 0.60, 95% CI 0.39 to 0.92; I² = 0%; NNTB = 47, 95% CI 31 to 236; eight trials, 1736 infants) (Analysis 1.1), and at latest follow‐up (4.0% vs 6.0%; RR 0.67, 95% CI 0.48 to 0.95; I² = 0%; NNTB = 50, 95% CI 32 to 331; 12 trials, 2293 infants; moderate‐quality evidence) (Analysis 1.4) (Figure 3)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 16.
• 40"At latest follow‐up, KMC was associated with a significantly decreased risk of mortality (RR 0.67, 95% CI 0.48 to 0.95; 12 trials, 2293 infants; moderate‐quality evidence) and severe infection/sepsis (RR 0.50, 95% CI 0.36 to 0.69; eight trials, 1463 infants; moderate‐quality evidence). Moreover, KMC was found to increase weight gain (mean difference [MD] 4.1 g/d, 95% CI 2.3 to 5.9; 11 trials, 1198 infants; moderate‐quality evidence), length gain (MD 0.21 cm/week, 95% CI 0.03 to 0.38; three trials, 377 infants) and head circumference gain (MD 0.14 cm/week, 95% CI 0.06 to 0.22; four trials, 495 infants) at latest follow‐up…" Conde-Agudelo and Díaz-Rossello 2016, Pg. 2.
• 41"At latest follow-up, KMC was associated with a significantly decreased risk of mortality (RR 0.67, 95% CI 0.48 to 0.95; 12 trials, 2293 infants; moderate-quality evidence)" Conde-Agudelo and Díaz-Rossello 2016, Pg. 2.
• 42"Between enrolment and 28 days, 73 infants died in 4423 periods of 28 days in the intervention group and 90 deaths in 3859 periods of 28 days in the control group (hazard ratio [HR] 0.70, 95% CI 0.51–0.96; p=0.027). Between enrolment and 180 days,
  158 infants died in 3965 periods of 180 days in the intervention group and 184 infants died in 3514 periods of 180 days in the control group (HR 0.75, 0.60–0.93; p=0.010). The risk ratios for death were almost the same as the HRs (28-day mortality 0.71, 95% CI 0.52– 0.97; p=0.032; 180-day mortality 0.76, 0.60–0.95; p=0.017)." Mazumder et al. 2019, Pg 1.
• 43Of the other three studies, two had the latest follow-up at 6 months, and one at 12 months. See Analysis 1.2 and Analysis 1.3, Conde-Agudelo and Díaz-Rossello 2016, Pg. 75-76.
• 44"The majority of all neonatal deaths (75%) occurs during the first week of life, and about 1 million newborns die within the first 24 hours." WHO, "Newborns: improving survival and well-being," 2020
• 45Conde-Agudelo and Díaz-Rossello 2016, Pg. 84, Analysis 1.13
• 46See Conde-Agudelo and Díaz-Rossello 2016, Pg. 32-58, Characteristics of included studies.
• 47See Pg. 17-19, Figure 3 of Conde-Agudelo and Díaz-Rossello 2016 for forest plots of subgroup analyses of KMC's effect on neonatal mortality.
• 48"We planned to pool data across studies using the fixed‐effect model if substantial statistical heterogeneity was not present. If we noted substantial heterogeneity (I² values ≥ 50%), we used a random‐effects model to pool data and made an attempt to identify potential sources of heterogeneity based on subgroup analysis by type of KMC, infant age at initiation of KMC, setting in which the trial was conducted, and risk of bias of trials." Conde-Agudelo and Díaz-Rossello 2016, Pg. 10.
• 49Conde-Agudelo and Díaz-Rossello 2016, Pg. 73-75, Analysis 1.1.
• 50Conde-Agudelo and Díaz-Rossello 2016, Pg. 79-80, Analysis 1.5.
• 51Conde-Agudelo and Díaz-Rossello 2016, Pg. 93, Analysis 1.31.
• 52Fixed effects models are generally used when all the underlying studies in a meta-analysis are implementing the same intervention on similar populations. Given the variation in how KMC was implemented and potential differences in the population targeted in each study, we believe a random effects model is the more appropriate choice for this analysis.
• 53"Some have adopted the practice of starting with a fixed-effect model and then switching to a random-effects model if the test of homogeneity is statistically significant. This practice should be strongly discouraged because the decision to use the random-effects model should be based on our understanding of whether or not all studies share a common effect size, and not on the outcome of a statistical test (especially since the test for heterogeneity often suffers from low power)." Borenstein et al. 2009, Pg 84.
• 54Heterogeneity of the outcomes for which a fixed-effects model was used:
  • Mortality at discharge: I² = 0.0% (Analysis 1.1, Pg. 73-75)
  • Severe infection/sepsis at latest follow-up: I² = 24% (Analysis 1.5, Pg. 79-80)
  • Exclusive breastfeeding at discharge: I² = 39% (Analysis 1.31, Pg. 93)

  Conde-Agudelo and Díaz-Rossello 2016

• 55"The random-effects method and the fixed-effect method will give identical results when there is no heterogeneity among the studies. When heterogeneity is present, a confidence interval around the random-effects summary estimate is wider than a confidence interval around a fixed-effect summary estimate. This will happen whenever the I² statistic is greater than zero, even if the heterogeneity is not detected by the Chi² test for heterogeneity (see Section 10.10.2)." Deeks et al. 2020.
• 56A summary of their judgments can be found in Figure 2 of Conde-Agudelo and Díaz-Rossello 2016, Pg. 14-15.
• 57"We judged that no study adequately addressed all seven domains. We judged only two studies to adequately address six domains. The methodological quality of the included trials was mixed, and we carried out a sensitivity analysis to examine the impact of excluding trials at high risk of bias. See Sensitivity analysis. The main threats to validity were performance bias (by lack of blinding of participants, personnel, and outcomes assessors) and selection bias (by lack of information on methods used for concealment of treatment allocation)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 13.
• 58"The sensitivity analysis limited to studies with adequate concealment of allocation revealed a similar reduction in mortality at discharge or at 40 to 41 weeks’ postmenstrual age, and at latest follow‐up, although this was not statistically significant (mortality at discharge or at 40 to 41 weeks’ postmenstrual age: RR 0.59, 95% CI 0.19 to 1.81; I² = 0%; five trials; mortality at latest follow‐up: RR 0.68, 95% CI 0.26 to 1.77; I² = 0%; six trials). Similar results were obtained when we excluded studies with high risk of attrition bias (mortality at discharge or at 40 to 41 weeks’ postmenstrual age: RR 0.64, 95% CI 0.41 to 1.00; I² = 0%; six studies; mortality at latest follow‐up: RR 0.71, 95% CI 0.49 to 1.01; I² = 0%; 10 studies)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 19.
• 59"The magnitude of [placebo] response depended primarily on the outcome type, i.e. whether the outcome was subjective or objective. For trials with subjective outcomes, the pooled effect size of placebo response was large and it persisted for the duration of the blinded follow-up; i.e. the follow-up values did not return to baseline. However, for trials with objective outcomes, the effect was small or not significant across all time points." Wartolowska et al. 2016, Pg. 8.
• 60"In this randomised controlled, superiority trial, undertaken in Haryana, India, we enrolled babies weighing 1500-2250 g at home within 72 h of birth, if not already initiated in kangaroo mother care, irrespective of place of birth (ie, home or health facility) and who were stable and feeding...Between July 30, 2015, and Oct 31, 2018, 8402 babies were enrolled, of whom 4480 were assigned to the intervention group and 3922 to the control group...Between enrolment and 28 days, 73 infants died in 4423 periods of 28 days in the intervention group and 90 deaths in 3859 periods of 28 days in the control group (hazard ratio [HR] 0·70, 95% CI 0·51–0·96; p=0·027). Between enrolment and 180 days, 158 infants died in 3965 periods of 180 days in the intervention group and 184 infants died in 3514 periods of 180 days in the control group (HR 0·75, 0·60–0·93; p=0·010). The risk ratios for death were almost the same as the HRs (28-day mortality 0·71, 95% CI 0·52– 0·97; p=0·032; 180-day mortality 0·76, 0·60–0·95; p=0·017)." Mazumder et al. 2019, Pg. 1.
• 61"At latest follow-up, KMC was associated with a significantly decreased risk of mortality (RR 0.67, 95% CI 0.48 to 0.95; 12 trials, 2293 infants; moderate-quality evidence)" Conde-Agudelo and Díaz-Rossello 2016, Pg. 2.
• 62"... 8402 babies were enrolled, of whom 4480 were assigned to the intervention group and 3922 to the control group." Mazumder et al. 2019, Pg 1.
• 63"This study is registered with ClinicalTrials.gov, NCT02653534 and NCT02631343." Mazumder et al. 2019, Pg 7.
• 64"Table 1 shows the baseline characteristics of the newborn babies, their mothers, and their families. Sociodemographic characteristics between study groups were very similar." Mazumder et al. 2019, Pg 7.
• 65"The randomisation sequence was prepared by a WHO statistician who was not involved with other trial activities using random permuted blocks of variable size. The WHO statistician concealed the allocation in serially numbered opaque sealed envelopes that were shipped to the off-site randomisation coordinator in India." Mazumder et al. 2019, Pg 3.
• 66
  • "These newborn babies were randomly assigned to either the intervention (n=4480) or control (n=3922) group (figure 1)... Vital status was known for 4470 infants in the intervention group and 3914 in the control group at age 28 days, and for 3653 in the intervention group and 3331 in the control group at age 180 days (figure 1)." Mazumder et al. 2019, Pg 7.
  • This is equivalent to around a 0.2% attrition rate in both the treatment and control groups before age 28 days.

• 67"This randomized controlled trial was done at a Level III Neonatal Intensive Care Unit (NICU) of a teaching institution in southern India. One hundred and forty neonates with birth weight <1500gm were enrolled. Inborn singleton, VLBW (birth weight <1500gm) infants, tolerating spoon feeds of 150mL/kg/day and haemodynamically stable (not on oxygen or respiratory support, no apnea for 72 hours, not on any intravenous fluids) were eligible. Infants with major malformation were excluded. Babies were randomized to KMC group or CMC [Conventional Medical Care] group. At 6 to 12 months corrected age, the assessment included the measurement of growth parameters in terms of malnutrition, wasting, stunting and having small head." Gavhane, Eklare, and Mohammad 2016, Pg. 13.
• 68Gavhane, Eklare, and Mohammad 2016, Table 3, Pg. 14.
• 69"After the exclusions, 140 infants were enrolled and randomized either to KMC (n=71) and CMC (n= 69) groups [Table/Fig-1]. Over all there was 65% follow-up till the corrected age of 6 to 12 months. Twenty three infants (32%) in the KMC group and 22 (32%) infants from the CMC group were lost to follow-up because the family got relocated and could not be contacted by telephone. Four infants (5.63 %) from the KMC group died before the first follow-up. Two of them died at 3 months of age with septicaemia, one at 4 months with pneumonia and another one at 2 months of age with intra-cranial bleed. None of the infants from the CMC group died after discharge. Adequate data for analysis was available for 44 (62%) and 47 (68%) infants from KMC and CMC group respectively." Gavhane, Eklare, and Mohammad 2016, Pg. 14.
• 70"This randomized controlled trial study was conducted in neonatal ward of Dhaka Shishu (Children) Hospital, Dhaka, Bangladesh from November, 2014 to October 2015. All preterm, neonates with birth weight 1250gm to 1800gm., gestational age < 30 weeks to > 35 weeks (Gestational age determined by New Ballard Score), haemodynamically stable (criteria of stable baby was 1) normal heart rate 100-160 per minutes, 2) respiratory rate 30-59 per minute with breaths comfortably, no sign of respiratory distress pink in room air or with 40% oxygen, no prolonged or frequent apnea) were included in this study… Forty neonates (40) were in KMC group and forty neonates (40) were in control group." Rahman et al. 2017, Pg. 2.
• 71"In this study, comparison of 40 KMC babies with 40 control showed that rate of weight gain per day in KMC group was 18.1±7.7 gms and in control group, it was 13.0±4.5 gms (p< 0.001)." Rahman et al. 2017, Pg. 3.
• 72"Ninety percent of KMC and 60% of control group babies were discharged with exclusive breast feeding (p=0.002)." Rahman et al. 2017, Pg. 3.
• 73"Apnea occurred less in KMC group (8% vs. 15%) (p=0.23)," Rahman et al. 2017, Pg. 3.
• 74"...episodes of hypothermia was also recorded less in KMC group (10% vs 18%)) (p=0.21)." Rahman et al. 2017, Pg. 3.
• 75"Culture positive sepsis was 15% in KMC group and 20% in control group (p=0.56)." Rahman et al. 2017, Pg. 3.
• 76"In this study mean hospital stay was 15.6 ± 10.6 days vs 18.2 ± 4.5 days, in KMC and control group respectively that was statistically insignificant, p=0.15." Rahman et al. 2017, Pg. 3.
• 77"The mortality was 2(5%) in KMC group and those was 6(15%) in control group and that was also statistically insignificant (P=0.14) (Tables 1-4)." Rahman et al. 2017, Pg. 3.
• 78"This single-centred randomised controlled trial was conducted at the University Teaching Hospital in Lusaka, Zambia, from August 2014 to September 2014. The University Teaching Hospital at Lusaka is the only tertiary care teaching hospital in Zambia and has approximately 12000 annual births." Ramani et al. 2018, Pg. 493
• 79"Randomised for Phase 1 at Birth (n=203)," Ramani et al. 2018, Figure 1, Pg. 493.
• 80"Term neonates born at a tertiary delivery centre in Zambia were randomised in two phases (phase 1: birth to 1 hour, phase 2: 1 hour to discharge) to either as much KMC as possible in combination with standard thermoregulation care (KMC group) or to standard thermoregulation care (control group)." Ramani et al. 2018, Pg. 493.
• 81"The proportion of neonates with moderate or severe hypothermia did not differ between the KMC and control groups at 1hour after birth (26/101 (25%) vs 28/102 (27%), relative risk (RR)=0.93, 95% CI 0.79 to 1.48, P=0.78) (table 3)." Ramani et al. 2018, Pg. 494.
• 82"The proportion of neonates with moderate or severe hypothermia did not differ between the KMC and control groups at discharge (6/84 (7%) vs 2/81 (2%), RR=2.8, 95% CI 0.6 to 13.9, P=0.16) (table 4)." Ramani et al. 2018, Pg. 495.
• 83"In phase 2 of the trial, the duration of KMC had a positive correlation to the infants’ temperature at discharge (r2 =0.47, P<0.001, figure 3). Hypothermia (mild, moderate or severe) was not found among the infants who had at least 9hours of KMC (r2 =0.47, P<0.0001, figure 3) or had KMC at least 80% of the time during their hospital stay (r2 =0.24, P<0.03, figure 4)." Ramani et al. 2018, Pg. 495.
• 84"We conducted this trial in the district of Dadu of Pakistan which is a rural district between October 2014 and July 2015. We enrolled mothers and neonates from the maternal unit of a tertiary care hospital." Habib et al. 2019, Pg. 3.
• 85"In this three arm Randomized Controlled Trial; we compared intervention A (ENC, application of chlorhexidine on umbilical cord and KMC) with intervention B (ENC and application of chlorhexidine on umbilical cord) and control arm C (ENC only)." Habib et al. 2019, Pg. 1.
• 86"We also calculated the neonatal mortality rates (NMR) per 1000 live births (Table 5) and found that the NMR was 27.3, 33.0 and 36.8 (per 1000 live births) in intervention arm A, intervention arm B and control arm respectively. Similar to the exclusive breastfeeding data, although we observed a difference it was not statistically significant." Habib et al. 2019, Pg. 10.
• 87"...we found that neonates who received intervention A were at reduced risk of having neonatal infection compared to the neonates in the control arm (RR 0.36, 95% CI 0.27- 0.63), and intervention A was superior to intervention B (RR 0.59, 95% CI 0.44-0.63)." Habib et al. 2019, Pg. 9.
• 88"We examined the weight data and found that the neonates in either intervention arm weighed more on day 28 compared to the control, intervention A (RR 2.27, 95% CI 1.54-3.46) and intervention B (RR 1.47, 95% CI 1.04-1.74, Table 3)." Habib et al. 2019, Pg. 9.
• 89"At discharge or 40 to 41 weeks' postmenstrual age, KMC was associated with a statistically significant reduction in the risk of...hypothermia (RR 0.28, 95% CI 0.16 to 0.49; nine trials, 989 infants; moderate-quality evidence. At latest follow‐up, KMC was associated with a significantly decreased risk of...severe infection/sepsis (RR 0.50, 95% CI 0.36 to 0.69; eight trials, 1463 infants; moderate‐quality evidence)..." Conde-Agudelo and Díaz-Rossello 2016, Pg. 2.
• 90"Bleakley 2010 makes a case that reducing the burden of malaria may have a lasting impact on children's development, and thus on their ability to be productive and successful throughout life." GiveWell, Intervention report: "Mass Distribution of long-lasting insecticide-treated nets (LLINs)," 2018.
• 91"Moreover, KMC was found to increase weight gain (mean difference [MD] 4.1 g/d, 95% CI 2.3 to 5.9; 11 trials, 1198 infants; moderate-quality evidence), length gain (MD 0.21 cm/week, 95% CI 0.03 to 0.38; three trials, 377 infants) and head circumference gain (MD 0.14 cm/week, 95% CI 0.06 to 0.22; four trials, 495 infants) at latest follow-up," Conde-Agudelo and Díaz-Rossello 2016, Pg. 2.
• 92"Investigators observed no differences in weight, length, or head circumference at discharge or at 40 to 41 weeks' postmenstrual age (Analysis 1.15; Analysis 1.19; Analysis 1.23) or at 12 months' corrected age (Analysis 1.17; Analysis 1.21; Analysis 1.25), or in weight or length at 6 months’ corrected age (Analysis 1.16; Analysis 1.20)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 20.
• 93See Pg. 85, Analysis 1.15, Pg. 87-88, Analysis 1.19, and Pg. 89, Analysis 1.23 of Conde-Agudelo and Díaz-Rossello 2016.
• 94"Infants in the intervention group had higher weight-for-age Z scores than the control group at age 28 days (adjusted mean difference 0·12, 95% CI 0·08–0·16; p0·0001) and age 90 days (0·07, 0·02–0·12; p=0·006). Similarly, weight-for-length Z scores were higher in the intervention group than in the control group at age 28 days (adjusted mean difference 0·14, 0·09–0·19; p0·0001) and at age 90 days (0·06, 0·01–0·12; p=0·025)." Mazumder et al. 2019, Pg. 8.
• 95This is based on our cost-effectiveness analysis of early-life growth interventions, which is preliminary and has not been vetted as deeply as many of our other public cost-effectiveness analyses. More background on the estimate can be found in the accompanying report, which has also not been vetted deeply.
• 96"Recently, a large community-based randomized controlled trial in North India showed a substantial improvement in neonatal survival and infant survival up to 6 months age in stable LBW infants, majorly late preterm or term small for gestational age infants, without any problems at birth and weighing 1500–2250 g at birth, as an effect of ciKMC [21]. However, the effects of ciKMC on neurodevelopment outcomes in LBW infants are still unclear. Therefore, in the first 552 of the infants enrolled in the above trial, our objective was to assess the impact of ciKMC on neurodevelopmental outcomes at 6 and 12 months of age and on maternal depressive symptoms; maternal sense of competence; mother-infant attachment and home environment." Taneja et al. 2020, Pg. 2.
• 97Taneja et al. 2020, Table 2, Pg. 8.
• 98"The lower (ΔL) and upper (ΔU) equivalence limits were set as − 3.0 and + 3.0 points respectively (1 SD =15 points for BSID; 3.0 points equate to 0.20 SD)." Taneja et al. 2020, Pg. 4.
• 99"The results were similar at 12 months of infant-age where no significant effect of ciKMC was found on cognitive, language, motor, socio-emotional and infant temperament scores using multivariable or ordinal regression analysis (Table 2 and Supplementary Table 1)." Taneja et al. 2020, Pg. 5.
• 100"At 6 months of infant-age, findings from univariable and multivariable linear regression were not suggestive of any significant effect of the ciKMC on composite cognitive scores (difference-in-means 0.98; 95% CI − 1.30 to 3.26)," Taneja et al. 2020, Pg. 5.
• 101"At 6 months of infant-age, findings from univariable and multivariable linear regression were not suggestive of any significant effect of the ciKMC on composite… language scores (difference-in-means-0.20; 95% CI − 1.99 to 1.58)," Taneja et al. 2020, Pg. 5.
• 102"At 6 months of infant-age, findings from univariable and multivariable linear regression were not suggestive of any significant effect of the ciKMC on composite... motor scores (difference-in-means 0.83; 95% CI − 1.91 to 3.57)," Taneja et al. 2020, Pg. 5.
• 103"At 6 months of infant-age, findings from univariable and multivariable linear regression were not suggestive of any significant effect of the ciKMC on composite... infant temperament scores (difference-in-means − 2.01; 95% CI − 5.07 to 1.06) (Table 2)" Taneja et al. 2020, Pg. 5.
• 104"It is important to recognize that measurement of developmental outcomes during infancy is challenging. Although BSID-III scale has been used in young infants to document the effects of interventions on developmental outcomes and has been shown as a reliable tool for assessment, yet it may not be able to detect small differences in certain aspects of brain functioning by virtue of it being a global development assessment scale. For example, in a randomized trial of docosahexaenoic acid (DHA) supplementation in infancy, no differences between intervention groups were found on Bayley scores at age 18 months. However, differences were found in sustained attention using a visual habituation task at four, six, and nine months, indicating enhanced attention in infants who received higher doses of DHA [34]. Interestingly, a follow-up study of the same DHA trial found differences between intervention groups in several cognitive tasks at age five years [35]. This suggests that to evaluate effect of interventions, examination of individual cognitive systems is needed." Taneja et al. 2020, Pg. 10.
• 105"Compared with conventional care, KMC was associated with an increase in the likelihood of exclusive breastfeeding at discharge or at 40 to 41 weeks' postmenstrual age (66.3% vs 56.3%; RR 1.16, 95% CI 1.07 to 1.25; I² = 39%; NNTB = 11, 95% CI 7 to 25; six studies, 1453 mothers) (Analysis 1.31)," Conde-Agudelo and Díaz-Rossello 2016, Pg. 21.
• 106"Our meta-analyses indicate protection against child infections and malocclusion, increases in intelligence, and probable reductions in overweight and diabetes. We did not find associations with allergic disorders such as asthma or with blood pressure or cholesterol, and we noted an increase in tooth decay with longer periods of breastfeeding. For nursing women, breastfeeding gave protection against breast cancer and it improved birth spacing, and it might also protect against ovarian cancer and type 2 diabetes. The scaling up of breastfeeding to a near universal level could prevent 823 000 annual deaths in children younger than 5 years and 20 000 annual deaths from breast cancer. Recent epidemiological and biological findings from during the past decade expand on the known benefits of breastfeeding for women and children, whether they are rich or poor." Victora et al. 2016, Summary.
• 107Broughton et al. 2013, Ruiz et al. 2017, Cattaneo et al. 2007, Vahidi et al. 2014, and Andrews 2018.
• 108"The maternal and neonatal hospital in Nicaragua participating in this program is a 270-bed national referral and teaching hospital in Managua that provides medical and surgical care in obstetrics, gynecology, neonatology, and adolescent health. Just under 11 000 deliveries are performed there annually, of which 18% are premature deliveries and 14% are low-birth-weight (LBM)." Broughton et al. 2013, P. 177.
• 109"The total average cost for care for a neonate was US$ 2 322 in the pre-KMC period and US$ 1 808 in the KMC period. The largest part of these costs was training the three health professionals at the regional KMC training center for 15 days." Broughton et al. 2013, P. 178.
• 110See this spreadsheet, sheet "KMC," row "Adjustment for costs saved to hospital."
• 111"Intermittent KMC decreased length of hospital stay by 1.6 days, although this difference was not statistically significant (95% CI ‐0.2 to 3.4; P value = 0.08; 11 studies, 1057 infants) (Analysis 1.13)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 20.
• 112See Conde-Agudelo and Díaz-Rossello 2016, Pg. 32-58, Characteristics of included studies, for descriptions of the implementation settings.
• 113"Sixteen studies were conducted in low‐ or middle‐income countries (India (Ali 2009; Gathwala 2008; Ghavane 2012; Kadam 2005; Kumbhojkar 2016; Nimbalkar 2014; Ramanathan 2001; Suman 2008); Ethiopia (Cattaneo 1998; Worku 2005); Malaysia (Boo 2007); Madagascar (Nagai 2010); Indonesia (Cattaneo 1998; Eka Pratiwi 2009); Nepal (Acharya 2014); Ecuador (Sloan 1994); Colombia (Charpak 1997); and Mexico (Cattaneo 1998)), and five in high‐income countries (United States (Blaymore Bier 1996; Neu 2010; Rojas 2003); United Kingdom (Whitelaw 1988); and Australia (Roberts 2000)). Conde-Agudelo and Díaz-Rossello 2016, Pg. 12."
• 114"We performed subgroup analyses for the primary outcome of mortality at discharge or at 40 to 41 weeks' postmenstrual age and at latest follow‐up according to type of KMC (intermittent vs continuous), infant age at initiation of KMC (≤ 10 days vs >10 days), setting in which the trial was conducted (low/middle‐income countries vs high‐income countries), and infant stabilization (before vs after). For all outcomes in stabilized LBW infants, we performed subgroup analyses according to type of KMC (intermittent vs continuous). In addition, we included RCTs that compared early‐onset (starting within 24 hours after birth) versus late‐onset (starting after 24 hours after birth) KMC." Conde-Agudelo and Díaz-Rossello 2016, Pg. 10.
• 115See Pg. 76-79, Analysis 1.4 of Conde-Agudelo and Díaz-Rossello 2016 for subgroup analysis results for mortality at latest follow-up.
• 116"RR 0.56, 95% CI 0.37 to 0.85; I^2 = 0%; six trials, 1489 infants" Conde-Agudelo and Díaz-Rossello 2016, Pg. 16
• 117RR 1.03, 95% CI 0.53 to 2.00; I^2 = 0%; five trials, 678 infants. Conde-Agudelo and Díaz-Rossello 2016, Analysis 1.3, Pg. 76
• 118See Conde-Agudelo and Díaz-Rossello 2016, Pg. 32-58, Characteristics of included studies, for descriptions of implementation across studies.
• 119Conde-Agudelo and Díaz-Rossello 2016, Analysis 1.4.9, Pg. 78.
• 120Conde-Agudelo and Díaz-Rossello 2016, Analysis 1.4.1, Pg. 76.
• 121Conde-Agudelo and Díaz-Rossello 2016, Analysis 1.4.4, Pg. 77.
• 122"Newborns weighing 2000 g or less at birth should be provided as close to continuous Kangaroo mother care as possible." WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019
• 123"Gathwala 2008 evaluated mother‐infant attachment at three months' follow‐up through a structured maternal interview that used attachment questions scored in such a manner that a higher score indicated greater attachment. The total attachment score for the KMC group (24.46 ± 1.64) was significantly higher than that obtained for the control group (18.22 ± 1.79) (Analysis 1.45)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 22.
• 124"Overall, scores on six comparisons (mother's sense of competence [interval between birth and start of intervention of one to two days], mother's sense of competence [infant admitted to NICU], mother's sense of competence [infant not admitted to NICU], mother’s feelings of worry and stress [interval between birth and start of intervention of one to two days], mother’s sensitivity [interval between birth and start of intervention > 14 days], and infant responsiveness [interval between birth and start of intervention > 14 days]) were significantly higher in the KMC group than in the control group. Scores on two comparisons (mother’s perceptions of social support [interval between birth and start of intervention > 14 days, and infant not admitted to NICU]) were significantly lower in the KMC group than in the control group." Conde-Agudelo and Díaz-Rossello 2016, Pg. 22.
• 125"One trial (Charpak 1997) evaluated home environment and father involvement at 12 months' corrected age through a structured interview administered to parents during a home visit. The total Home Observation for Measurement of the Environment (HOME) score was significantly higher among kangaroo families (0.28 ± 0.24) than in conventional care families (‐0.51 ± 0.26) (Analysis 1.50). Scores on father involvement were not reported, but study authors claimed that KMC increased father involvement (the father's sense of responsibility and competence)." Conde-Agudelo and Díaz-Rossello 2016, Pg. 23.
• 126"Compared with conventional care, KMC was associated with an increase in the likelihood of exclusive breastfeeding at discharge or at 40 to 41 weeks' postmenstrual age (66.3% vs 56.3%; RR 1.16, 95% CI 1.07 to 1.25; I² = 39%; NNTB = 11, 95% CI 7 to 25; six studies, 1453 mothers) (Analysis 1.31)," Conde-Agudelo and Díaz-Rossello 2016, Pg. 21.
• 127"There is additional evidence that increasing breastfeeding reduces diarrhea morbidity, which likely leads to reductions in childhood mortality from diarrhea. Breastfeeding may also cause additional benefits that we have not yet vetted." GiveWell, Intervention report: "Breastfeeding promotion programs," 2018.
• 128"The time needed to provide kangaroo mother care was a potential barrier for mothers, fathers and families, due to responsibilities at home and work and time needed for commuting, preventing them from devoting the time needed for continuous and extended kangaroo mother care." Chan et al. 2016b, Pg. 134.
• 129"This section proposes possible negative impacts of the intervention for further investigation in the future. We have not examined these negative impacts carefully or seen compelling evidence that they would offset potential benefits of the intervention, but we speculate that breastfeeding promotion interventions could pose several problems:
  • A risk of accidentally malnourishing a minority of infants if advice to breastfeed exclusively is given and taken too rigidly. Some mothers may struggle to breastfeed or face physical constraints including infection and undersupply.
  • A risk of causing mental distress in mothers who feel social pressure to breastfeed but are unable to do so.
  • A risk of compounding malnutrition in mothers in low-resource environments.
  • A risk that compliance with the breastfeeding guidelines may be an obstacle to maternal employment and financial stability.
  • A risk of HIV transmission to infants via breastfeeding in HIV-positive mothers without access to recommended antiretroviral drug regimens."

  GiveWell, Intervention report: "Breastfeeding promotion programs," 2018

• 130"There was a common belief among nurses that training the mothers to do SSC would take more time than they had available, and nurses were concerned about not having time to attend to their other newborn patients in the NICU (Charpak and Ruiz-Pelaez 2006; Engler et al. 2002). In certain facilities, when there was an overabundance of patients, KMC mothers were the lowest priority for HCWs (Bergh et al. 2012a)." Chan et al. 2017, Pg. 1,469.
• 131"Structurally, crowding and insufficient space in facilities presented a barrier to KMC…" Chan et al. 2017, Pg. 1,471.
• 132See our cost-effectiveness analysis of KMC, “KMC” sheet, “Cost-effectiveness in multiples of cash” row.
• 133"Regionally, neonatal mortality was highest in sub-Saharan Africa and South Asia, with the neonatal mortality rate estimated at 27 and 25 deaths per 1,000 live births, respectively, in 2019. A child born in sub-Saharan Africa was 10 times more likely to die in the first month than a child born in a high-income country, while a child born in South Asia was nine times more likely to die." UNICEF, "Neonatal mortality," 2020
• 134See here for calculations.
• 135
  • "The neonatal mortality rate (NMR) for these infants was 133 per 1000 live births (95% confidence interval: 110-159)." Yasmin et al. 2001, Pg. 1.
  • See this spreadsheet, sheet “KMC,” row “Baseline neonatal mortality rate among low birth weight infants.”

• 136See this spreadsheet, sheet “KMC,” row “Proportion of KMC's benefits coming from mortality.”
• 137See this spreadsheet, sheet “KMC,” section “Internal and external validity.”
• 138See this spreadsheet, sheet “KMC,” row “Proportion of KMC's benefits coming from cost savings to hospitals.”
• 139See this spreadsheet, sheet “KMC,” rows “Proportion of KMC's benefits coming from developmental effects” and “Proportion of KMC's benefits coming from morbidity.”
• 140
  • Broughton et al. 2013, Table 2, Pg. 179.
  • We divide the subtotal for 12 hospitals ($106,200) by 12 to arrive at $8,850.

• 141See this spreadsheet, sheet “Costs,” row “Total per-patient cost of KMC.”
• 142See this spreadsheet, sheet “KMC,” section “Accounting for cost savings to hospital,” row “Cost-effectiveness in multiples of cash.”
• 143
  • "The neonatal mortality rate (NMR) for these infants was 133 per 1000 live births (95% confidence interval: 110-159)." Yasmin et al. 2001, Pg. 1.
  • See this spreadsheet, sheet “KMC,” row “Baseline neonatal mortality rate among low birth weight infants.”

• 144See this spreadsheet for calculations.
• 145"Sepsis is a life-threatening condition that arises when the body’s response to infection causes injury to its own tissues and organs. It is frequently a final common pathway to death for many infectious diseases worldwide. It involves organ dysfunction caused by a dysregulated host response to infection and if not recognized early and managed promptly, it can lead to septic shock, multiple organ failure and death." WHO, "Sepsis"
• 146"Hypothermia during the newborn period is widely regarded as a major contributory cause of significant morbidity in developing countries and, at its extreme, mortality.⁵ High prevalence of hypothermia has been reported from countries with the highest burden of neonatal mortality, where hypothermia is increasingly gaining attention and significance as a critical intervention for newborn survival." Kumar et al. 2009, Pg. 401.
• 147See this spreadsheet, sheet "KMC," row "Adjustment for morbidity effects."
• 148See this spreadsheet sheet "KMC," row "Adjustment for developmental effects."
• 149

  “The time needed to provide kangaroo mother care was a potential barrier for mothers, fathers and families, due to responsibilities at home and work and time needed for commuting, preventing them from devoting the time needed for continuous and extended kangaroo mother care… While parents believed that kangaroo mother care was less costly than incubator care, lack of money for transportation and the distance to hospital were often reported as the biggest challenges as were low resources for newborn-care services.” Chan et al. 2016a, Pg. 134.

• 150
  • “Structurally, crowding and insufficient space in facilities presented a barrier to KMC, hastening discharge (one study reported a discharge in under two hours) or restricting visitation policies because of staffing shortages and space concerns (Bergh et al. 2012d). If mothers are unable to visit their babies due to restricted visitation policies, KMC cannot be performed. Mothers are also less likely to stay in the hospital to practice KMC if their families cannot visit them. In addition, insufficient privacy due to space concerns and lack of privacy screens coupled with discomfort with being undressed in the presence of strangers acted as a KMC barrier (Blomqvist et al. 2013; Nahidi et al. 2014)” Chan et al. 2017, Pg. 1471.
  • “Lack of private space for mothers to perform kangaroo mother care and to remain in the hospital with the newborn hindered its uptake, as did allocation of resources intended for kangaroo mother care to other programmes. Uptake improved with transportation for mothers not staying at the hospital, wrappers to hold the baby, furniture/ beds where mothers could conduct kangaroo mother care, rooms where mothers could spend the night with the baby, private spaces and dedicated resources.” Chan et al. 2016a, Pg. 134.

• 151

  “Staff shortages, high turnover of staff, and an insufficient number of trained HCWs inhibited buy-in of KMC at health facilities. When a particular HCW has served as the primary promoter of SSC in a facility, if they departed the hospital, it increased the difficulty of educating other staff members on the practice of SSC (Lee et al. 2012). At some institutions, researchers heard reports of practicing ‘intermittent KMC in our hospital’, but observed no individuals practicing KMC during surprise visits with explanations of ‘staff shortages’ or lack of presence of KMC supporters (Bergh and Pattinson 2003). Even with a presence of KMC supporters, an insufficient number of staff inhibited KMC: ‘We’re still kind of stumbling a little bit because of our lack of manpower to move forward with a lot of our things. I think the intent and the will is there, just we require more team members.’ (Health facility staff) (Lee et al., 2012)” Chan et al. 2017, Pg. 1468.

• 152

  “Several factors affected HCW buy-in of KMC. Many healthcare staff believed newborn care was not a high priority at their facility.(Bergh et al. 2007) Furthermore, KMC was perceived as a disadvantage by nurses because the mother needed to remain in the hospital (Solomons and Rosant 2012). In some cases, KMC was cited as the ‘poor man’s alternative’ for developing countries and considered to be a sub-standard method of care (Charpak and RuizPelaez 2006).” Chan et al. 2017, Pg. 1468.

• 153

  “There was a common belief among nurses that training the mothers to do SSC would take more time than they had available, and nurses were concerned about not having time to attend to their other newborn patients in the NICU (Charpak and Ruiz-Pelaez 2006; Engler et al. 2002). In certain facilities, when there was an overabundance of patients, KMC mothers were the lowest priority for HCWs (Bergh et al. 2012a). In some cases, staff did not have time to learn new KMC protocols (Engler et al. 2002). “KC takes too much work, too much time” and “[I am] not willing to take extra time with the family that KC requires.” (Healthcare Worker) (Engler et al. 2002).” Chan et al. 2017, Pg. 1469.

• 154

  “The bottleneck analysis tool was applied in 12 countries in Africa and Asia as part of the Every Newborn Action Plan process. Country workshops involved technical experts to complete the survey tool, which is designed to synthesise and grade health system "bottlenecks", factors that hinder the scale-up, of maternal-newborn intervention packages. We used quantitative and qualitative methods to analyse the bottleneck data, combined with literature review, to present priority bottlenecks and actions relevant to different health system building blocks for KMC.” Vesel et al. 2015, Abstract, Pg. 1.

• 155
  • “The most significant or very major bottlenecks were reported for: health financing (10 countries), community ownership and partnership (10 countries), health service delivery (10 countries), leadership and governance (9 countries) and health workforce (9 countries).“ Vesel et al. 2015, Pg. 8.
  • See Vesel et al. 2015, Table 1, Pg. 7 for a summary of bottlenecks found.

• 156

  “The aim of this case study was to understand the institutionalisation processes of facility-based KMC services in three Asian countries (India, Indonesia and the Philippines) and the reasons for the slow uptake of KMC in these countries.” Bergh et al. 2016, Abstract, Pg. 1.

• 157

  “Despite awareness and orientation programs in KMC and fairly wide coverage of training in some countries, with individual hospitals starting implementation, the expansion of KMC services has taken off slowly in all countries. There has been a gap between implementation of KMC by the first hospital(s) and the further expansion of KMC services to other health facilities. In most cases there has been no significant spread without systematic donor or NGO input.” Bergh et al. 2016, Pg. 6.

• 158

  “The decentralised nature of the health system in all three countries may also have contributed to pockets of inertia in KMC scale-up. In Indonesia, for example, support for KMC from the central Ministry of Health did not guarantee commitment at regional or district level [33].” Bergh et al. 2016, Pg. 10.

• 159

  “The reason why these hospitals could not achieve large-scale expansion to other facilities may be that national scale-up plans or coordination mechanisms were not in place by 2013.” Bergh et al. 2016, Pg. 7.

• 160

  “Professional resistance and the lack of political priority for newborn and LBW care within the political and healthcare structures at the time when early adopters introduced KMC into their health facilities explain why the vast majority of institutions were left behind.” Bergh et al. 2016, Pg. 11.

• 161

  “None of the countries had any systematic national reporting mechanisms for the progress of KMC implementation or for reporting on indicators related to the provision of KMC services (e.g. coverage: number of total of LBW babies receiving KMC; impact: mortality and morbidity figures for preterm and LBW babies receiving KMC compared to those not receiving it).” Bergh et al. 2016, Pg. 7.

• 162

  “The initial training packages tended to focus on training in the practice of KMC without sufficient focus on preparation for implementing KMC (including institutional and structural support needed) and on supportive supervision for sustainability. This is typical of training that takes place in project mode [24]. Where KMC was integrated into newborn and obstetric care training packages, the training may have been more theoretical and limited in scope [63] without providing a sense of agency of how individuals could influence and change practice that improves patient care.” Bergh et al. 2016, Pg. 10.

• 163

  “In other countries KMC also ‘got lost’ [64] because too many newborn interventions and improvements were required at the same time.” Bergh et al. 2016, Pg. 10.

• 164

  “The aim of this study was to systematically evaluate the implementation status of facility-based kangaroo mother care services in four African countries: Malawi, Mali, Rwanda and Uganda.” Bergh et al. 2014, Abstract, Pg. 1.

• 165

  “All four countries had received financial and/or technical support from the Saving Newborn Lives program and/or the Maternal and Child Health Integrated Program (MCHIP, and its predecessor ACCESS) for the implementation of KMC on its own or as part of essential newborn and/or obstetric care initiatives. In all countries different forms of support were also provided by other agencies or non-governmental organizations at different levels of the health system, either as part of an intervention or on an ad hoc basis.” Bergh et al. 2014, Pg. 2.

• 166

  “A convenience sample of facilities to receive on-site visits was identified (total n = 39). Logistic and cost constraints did not allow for the use of probability sampling. It was, however, important to ensure that the assessed facilities included all levels of care, where applicable, and also to allow for sufficient geographic spread. Table 2 gives a summary of the types of facilities visited per country. Across all countries combined, more than one in every five facilities reportedly providing KMC services received a site visit, ranging from 100% of facilities in Mali with only seven facilities providing KMC services to 12% in Malawi, where 121 facilities were reported to provide these services.” Bergh et al. 2014, Pg. 2.

• 167
  • “Thirteen facilities (33%) had a score on the fifth stage of change (evidence of routine and integration) (Malawi 6; Rwanda 4; Uganda 3), whereas two facilities (5%) had reached the level of sustainable practice (Rwanda 1; Mali 1).” Bergh et al. 2014, Pg. 4.
  • See Table 3, p. 4 for the scoring rubric for stages of implementation at a facility.

• 168

  “On Oct 21–22, 2013, stakeholders in newborn health convened in Istanbul, Turkey, to discuss how to accelerate the implementation of kangaroo mother care (KMC) globally. Focused attention on newborn deaths, which now account for 44% of under-5 mortality, is required to accelerate progress toward Millennium Development Goal 4 (to reduce child mortality by two-thirds) and beyond… We reached consensus, based on the available evidence, that KMC should be adopted and accelerated as standard of care as an essential intervention for preterm newborns. We defined success as augmented and sustained global and national level action to achieve 50% coverage of KMC among preterm newborns by the year 2020 as part of an integrated RMNCH package.” Engmann et al. 2013, Pg. e26.

• 169

  Details of this conversation have not been published.

• 170
  • “Despite the strong evidence base for KMC, progress in taking KMC implementation to scale has been slow [5,11,12]... One of the crosscutting challenges underpinning these barriers was effective information systems and data on KMC coverage and quality [5,12].” Guenther et al. 2017, Pg. 2.
  • In case study of 4 countries in Africa: “No facility recorded intermittent and continuous KMC data separately and in two countries some facilities conflated KMC data with the general LBW data that was routinely collected. Where the introduction of KMC was part of a project’s data collection, processing seemed to deteriorate after the end of the project. Staff did not know what to do with their data or did not see a reason for continuing with the effort. In one country facilities were supposed to send regular reports with more detailed information to a central office at the ministry of health, but it was unclear if and how these reports were used to monitor KMC quality or scale up. In the other three countries only seven facilities regularly reported on KMC-specific activities and statistics to a higher level of in-facility management.” Bergh et al. 2014, Pg. 6.
  • In case study of 3 countries in Asia: “None of the countries had any systematic national reporting mechanisms for the progress of KMC implementation or for reporting on indicators related to the provision of KMC services (e.g. coverage: number of total of LBW babies receiving KMC; impact: mortality and morbidity figures for preterm and LBW babies receiving KMC compared to those not receiving it).” Bergh et al. 2016, Pg. 7.

• 171"Kangaroo mother care is recommended for the routine care of newborns weighing 2000 g or less at birth, and should be initiated in health-care facilities as soon as the newborns are clinically stable." WHO, "Kangaroo mother care to reduce morbidity and mortality in low-birth-weight infants," 2019
• 172"KMC has been proven to reduce newborn mortality, but only a very small proportion of newborns who could benefit from KMC receive it." Engmann et al. 2013