Elsevier Editorial System(tm) for The Lancet Manuscript Draft Manuscript Number: THELANCET-D-10-04958R2 Title: Maternal, neonatal and

child health: Now (more than ever) for Southeast Asia Article Type: Invited Series Corresponding Author: Dr. Cecilia Santos Acuin, Corresponding Author's Institution: University of the Philippines National Institutes of Health First Author: Cecilia S Acuin Order of Authors: Cecilia S Acuin; Geok Lin Khor; Endang L Achadi; Tippawan Liabsuetrakul; Thein Thein Htay; Rebecca Firestone; Zulfiqar A Bhutta, PhD Abstract: While maternal and child mortality are on the decline in Southeast Asia, the region is witness to major disparities across and within its ten member countries, and greater equity will be key to achieving the MDGs. We used comparable cross-national data sources to document mortality trends from 1990 in a standardized approach and to assess major causes of maternal and child deaths. We present inequalities in intervention coverage by two common measures of wealth quintiles and rural/urban status. Case studies of mortality reduction in Thailand and Indonesia present the mixed picture of success within this diverse region and point to some of the factors that may be capitalized on to accelerate progress in the future. We developed a Lives Saved Tool (LiST) analysis for the region as a whole and for country subgroups defined by mortality trends to estimated deaths averted by cause and intervention at varying coverage levels. We found three major patterns of maternal and child mortality reduction: 1) early, rapid downward trends- Brunei, Malaysia, Singapore, Thailand; 2) initially high declines sustained by Vietnam but faltering in Philippines, Indonesia; 3) high initial rates with a downward trend that require more focus to accelerate - Cambodia, Myanmar, Laos. High achievers in the region point to early declines before rapid economic growth, suggesting that economic development provides an important context that must be coupled with broader health system interventions. Increasing coverage will have a significant impact on maternal deaths by unsafe abortion, hypertensive diseases and postpartum hemorrhage , neonatal deaths caused by pneumonia/ sepsis and birth asphyxia and child deaths caused by infectious diseases. These actions will require consideration of health system contexts, and a regional push through ASEAN may provide greater policy support to achieve MNCH goals. Funding The Rockefeller Foundation and China Medical Board provided funds that allowed the authors of this paper to meet and to access research assistance.

Manuscript

Title: Maternal, neonatal and child health: Now (more than ever) for Southeast Asia
Order of Authors: Cecilia S Acuin; Geok Lin Khor; Tippawan Liabsuetrakul; Endang L Achadi; Thein Thein Htay; Rebecca Firestone; Zulfiqar A Bhutta,

Key messages Southeast Asia has sustained substantial reductions in maternal, neonatal and child mortality since 1990, but this progress has been uneven. Mortality reductions in some countries have been the result of trajectories of rapid decline begun well before the MDG period. Others have succeeded in acerbating progress since 1990s, but some countries continue to struggle. Causes of death suggest a mortality transition in maternal deaths in the region. Child deaths are primarily due to the persistence of neonatal causes along with key preventable factors in the postneonatal period Disparities in intervention coverage are most acute in the countries with the lowest intervention coverage overall Despite the mixed picture, some countries stand out as success stories. Suggested key factors include the ability to link MNCH interventions to broader health system investments and to target access to rural and disadvantaged populations Increasing coverage to 60% will have a significant impact on maternal deaths caused by unsafe abortion, hypertensive diseases and postpartum hemorrhage and neonatal deaths caused by pneumonia/sepsis and birth asphyxia. While MNCH may have no quick solutions in the region, coordinated expansion of proven effective interventions can contributed to greater mortality reductions There is a need for stronger regional cooperation through ASEAN to support countries that need to accelerate progress to meet the MDGs
Abstract: While maternal and child mortality are on the decline in Southeast Asia, the region is witness to major disparities across and within its ten member countries, and greater equity will be key to achieving the MDGs. We used comparable cross-national data sources to document mortality trends from 1990 in a standardized approach and to assess major causes of maternal and child deaths. We present inequalities in intervention coverage by two common measures of wealth quintiles and rural/urban status. Case studies of mortality reduction in Thailand and Indonesia present the mixed picture of success within this diverse region and point to some of the factors that may be capitalized on to accelerate progress in the future. We developed a Lives Saved Tool (LiST) analysis for the region as a whole and for country subgroups defined by mortality trends to estimated deaths averted by cause and intervention at varying coverage levels. We found three major patterns of maternal and child mortality reduction: 1) early, rapid downward trends- Brunei, Malaysia, Singapore, Thailand; 2) initially high declines sustained by Vietnam but faltering in Philippines, Indonesia; 3) high initial rates with a downward trend that require more focus to accelerate - Cambodia, Myanmar, Laos. High achievers in the region point to early declines before rapid economic growth, suggesting that economic development provides an important context that must be coupled with broader health system interventions. Increasing coverage will have a significant impact on maternal deaths by unsafe abortion, hypertensive diseases and postpartum hemorrhage , neonatal deaths caused by pneumonia/ sepsis and birth asphyxia and child deaths caused by infectious diseases. These actions will require consideration of health system contexts, and a regional push through ASEAN may provide greater policy support to achieve MNCH goals. Funding The Rockefeller Foundation and China Medical Board provided funds that allowed the authors of this paper to meet and to access research assistance.

Introduction Southeast Asia has achieved substantial reductions in child and maternal mortality within a relatively short period of time, but these achievements are unevenly distributed among the countries in the region. Of the ten countries in the political coalition known as the Association of Southeast Asian Nations (ASEAN), only three have infant and child mortality rates below 10/1000 live births, namely Brunei, Singapore and Malaysia. Infant and under-five mortality in Thailand and Vietnam have declined dramatically to below 15, but the Philippines and Indonesia have seen a leveling off of rates in the 30s and 40s. Myanmar, Cambodia and Laos still see mortality levels comparable to their neighbors more than two decades ago and rank among the highest for Asia (1). The United Nations estimates that each year approximately 350,000 women die as a result of pregnancy or childbirth (2), as do nearly 9 million children below 5 years (3). Southeast Asia contributed approximately 18,000 maternal (2) and 400,000 child (3) deaths to this global burden in 2008. Two of the seven countries with the highest maternal mortality ratios outside of sub-Saharan Africa are Laos and Cambodia, while Indonesia is among the countries accounting for 65% of all maternal deaths worldwide. While Southeast Asia, as a region, may achieve the child mortality reductions set by the United Nations Millenium Development Goal 4 (MDG4), Cambodia and Myanmar have been rated as showing "insufficient progress" (4). Declines in Laos, Indonesia and the Philippines are also faltering. Similarly, while all countries demonstrate declines in maternal deaths, the rates of decline for Indonesia, Myanmar and the Philippines have slowed considerably. Effective and affordable technology to reduce the majority of maternal, newborn and child deaths is at hand (5, 6, 7, 8), so why has progress been so uneven? This paper focuses on a region, collectively the 9th largest economy in the world, whose performance and achievements are often hidden by larger states like India or China and UN agency regional groupings that do not take into account historical and geo-political ties within ASEAN (28). Southeast Asia as a region has received markedly little attention in recent efforts to revitalize and strengthen the MNCH policy agenda, despite the complexity of national trends, including the significant burden of morbidity and mortality in several countries and the existence of documented successes (31). As Southeast Asia becomes more integrated economically, there is a growing need to take stock of this major unfinished agenda and identify policy options for sustaining if not accelerating the pace of mortality reduction. This paper aims to critically review the regions achievements in maternal and child mortality reduction and point to key factors that explain success and challenges in reaching these goals amidst competing global, regional and national health problems. We first report on patterns of mortality reduction within Southeast Asia as well as major causes of maternal and child deaths in the context of MDGs 4 and 5. We investigated two country cases to illustrate the significant variations in mortality reduction. Finally, we used an analysis of the deaths to be averted through expanded coverage to identify more effective approaches for pursuing maternal, neonatal and child health in Southeast Asia.

Methods Data sources For the ten countries considered here, we reviewed estimates from national data sources including Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS), as well as global datasources from UNICEF, WHO, and the Institute for Health Metrics and Evaluation (IHME) (webTable 1). We present country-specific estimates on maternal, infant, and under-five mortality rates from recent UN MDG reports (2, 3), as these estimates enable cross-country comparisons on trends in mortality using replicable estimation methods that seek to reduce sources of non-sampling error. These estimates tend to be more conservative in the rate of decline than IHME estimates. Based on growing awareness of the burden of neonatal mortality, we sought comparable estimates of trends disaggregating neonatal (death within the first 28 days following birth) and postneonatal (death between 28 days and one year following birth) mortality. As UN models do not have neonatal time trends for all countries of the region, we present estimates from IHME (10). We present estimates on causes of neonatal and child deaths based on standardized methods for estimating the distribution of causes of child deaths (11). We have compiled estimates of causes of maternal deaths from Countdown 2015 country reports (4) and WHO (12) and we used annual State of the Worlds Children reports (13) to review standard assessment of nutrition trends (web Appendix Fig 2). Data from DHS (14) and MICS (15, 16) were evaluated to assess existing intervention coverage within the region, with these data sources providing the ability to disaggregate coverage estimates by wealth quintile and rural/urban status (17-20) to unpack country average and asses program coverage among disadvantaged populations. Regional estimates were calculated using country-level data from the specific source cited, unless otherwise indicated. Analysis Case studies were developed to test the contribution of health sector inputs to mortality reductions. Thailand and Indonesia were selected as cases, as high and lower achievers, and we focused on maternal and neonatal mortality as outcomes sensitive to health system development. National data were used for case studies in order to extend the analysis to the period prior to the MDG baseline year (38, 42). For the case studies, data was fitted to a quadratic equation (log10MMR or log10NMR = Intercept + linear effect of year+ quadratic effect of year) as well as a linear equation (log10 MMR or log10NMR= Intercept + linear effect of year) to determine whether declines in maternal mortality could be attributed to program changes or temporal trends. We calculated potential deaths to be averted through increasing population coverage of the interventions proven to be effective in reducing maternal, neonatal and child mortality using the Lives Saved Tool (LiST) (21,49). LiST operates within the Spectrum modelling platform, initially developed to project demographic change and complemented by modules to model the impact of family planning and HIV/AIDS interventions (50). The model yields estimates of deaths averted by cause and intervention for user-specified intervention coverage levels, based on inputs of demographic projections, numbers of maternal and child deaths, data on the distribution of deaths by cause, intervention effectiveness, and data on local health status (6, 22, 23, 24). The tool has been previously used for analysis of impact of intervention packages on maternal and child survival in South Africa (25) and sub-Saharan Africa, but this is a first application in Southeast Asia (26). For this analysis, we assessed all of the MNCH interventions included in LiST (51). The interventions and their 3

effectiveness estimates are shown in webtable 2. Values for the effectiveness of interventions were developed through a standardized review process using established criteria to determine which interventions to include based on levels of evidence (51). The analysis was conducted for all ten countries and then for three subgroups of countries based on observed patterns of mortality reduction: subgroup 1 (Brunei, Malaysia, Singapore, Thailand); subgroup 2 (Indonesia, Philippines, Vietnam); subgroup 3 (Cambodia, Laos, Myanmar). We assessed potential lives saved at three hypothetical coverage levels: 60%, 90% and 99%. Results Patterns of Mortality Reduction Reduction in maternal, infant and child mortality (Fig1a-Fig 1c, data not shown for Brunei and Singapore) in Southeast Asia reflect the diversity of this region, presenting three divergent patterns (27, 28). The first pattern reflects countries achieving low mortality rates between 1990, the MDG baseline year, and 2008 in Brunei, Singapore, Malaysia and Thailand. In 1990, maternal mortality ratios in these countries were well below 100/100,000 live births, and infant and under-five mortality rates were already at or below 20/1000 live births. These most economically advanced countries in the region have also invested in their health systems over time. A second, less distinct, pattern, seen in the Philippines, Indonesia and Vietnam, starts with relatively high mortality rates and ratios in 1990, fairly large initial reductions (except for Indonesia's maternal mortality ratio) that somewhat faltered after 2000 in Indonesia and the Philippines. In contrast, Vietnam witnessed accelerated mortality reductions during this period, with mortality rates and ratios beginning to come close to those of Thailand. The third pattern, observed in Laos, Cambodia and Myanmar has very high levels at the beginning of 1990, followed by sustained reductions from 1990 to 2005 with the exception of Cambodia's maternal mortality ratio. These three countries, which are on the UN list of least developed countries, continue to experience high maternal, infant and child mortality. Plotting maternal mortality reductions against Gross National Income (GNI) per capita (Web Appendix Fig 1a) indicates that while countries with high maternal mortality achieved reductions in mortality as their GNI per capita increased, some of the most dramatic declines in mortality took place earlier than the rapid rise in GNI. Thailand's rapid maternal mortality reductions occurred pre-1990. As maternal mortality declined to levels around 100, smaller reductions take place even as GNI continues to improve. Similar patterns are observed for infant and under 5 mortality vs GNI per capita plots (Web Appendix Fig 1 b and c). Neonatal and Postneonatal Mortality Reductions Disaggregating infant mortality reduction between 1990 and 2010 into neonatal and postneonatal (Fig 2) indicates that the largest declines in infant mortality over time were mainly due to substantial postneonatal mortality reductions, as seen in Vietnam, Malaysia and Thailand. The Philippines and Indonesia had neonatal and postneonatal reductions comparable to Cambodia, Laos and Myanmar. Although starting with comparably lower baseline mortality levels in 1990, rates of decline in these two countries were not sufficiently accelerated during the MDG period. Reductions in infant mortality in

Brunei and Singapore stem from larger proportions of decline in neonatal deaths, a pattern similar to other high income countries (29). The slower rates of decline for neonatal mortality for 8 of 10 ASEAN countries is a cause for concern. Interventions for reducing neonatal mortality are more closely linked to maternal interventions in terms of policy and program implementation and may not be as noticeably tracked towards their impact on under-5 mortality. The Philippines, which is considered to be "on track" in achieving child mortality reductions (4), has the lowest reduction in neonatal mortality in the region, lower than Cambodia or Myanmar, which have been identified as showing "insufficient progress" towards achieving MDG 4. Causes of Mortality The distribution of maternal mortality causes (Fig 3a) reflects the significant variations in health status and health system development seen within the region. Haemorrhage is a leading cause of death, likely indicative of delays in attaining emergency obstetric care. Hypertensive disorders contribute to about one in every six maternal deaths in Southeast Asia and suggest a different causal pathway more comparable to developed country settings. The proportion of other indirect causes may indicate the still significant burden of infectious disease within the region and the impact of malaria and HIV on maternal health (30). Unsafe abortion is a factor in almost one-tenth of maternal deaths. These patterns reflect a causal transition in maternal mortality as the overall risk of maternal death declines, and they will influence the extent to which interventions, both as single modalities or included in a package, can be predicted to avert deaths. (12). Differential rates of child mortality reduction can be attributed in part to variations in causes of death (Fig 3b). Neonatal conditions contributed approximately 40% of child mortality, accounting for the single largest proportion of preventable deaths, even as a number of the ASEAN countries are successfully reducing their postneonatal and child mortality burdens. Infectious diseases including pneumonia, diarrhea and others, still account for almost half of child deaths, reflecting considerable scope for continued reductions in child mortality (11).

Within Country Disparities in Intervention Coverage Inequalities are substantial across countries of the region, but also within, as shown by the current scope of intervention coverage by income and rural/urban sub-groups (Fig 4), considering antenatal care coverage, use of skilled birth attendance (SBA), DPT and measles vaccination along with use of oral rehydration therapy, all key to the development of a continuum of care (4, 7). Laos remains substantially lower than other countries on overall program coverage and far from a 60% coverage level even for the most well-off groups. Antenatal care coverage is most widespread, being close to or above 90%, in countries other than Laos and Cambodia, for the most well-off and urban areas, suggesting that there is scope to effectively scale up prenatal interventions that can avert maternal deaths. Vaccination coverage varies widely, although several countries in the region are GAVI-eligible and have received substantial financial and policy support likely to lead to increases in vaccination coverage over time. Laos and Cambodia present the greatest disparities in program coverage. Vaccination levels for the 5

wealthiest quintile are comparable to those of the other SEA countries, but they are well below 50% amongst the poorest households. The countries represented in Fig 4 demonstrate relatively low coverage of skilled birth attendance (except Thailand) with inequality particularly acute in Laos, Cambodia and the Philippines. Skilled birth attendance may be viewed as one indicator of broader health system development, and the generally low coverage coupled with a high degree of inequality illuminates the need for more comprehensive and coordinated health system strengthening in the region overall (31). These patterns also point to the necessity of targeting the most vulnerable populations and maintaining attention to equity while scaling up program coverage (32). Case studies: Thailand and Indonesia We look in more depth at two countries with diverging experiences of mortality reduction to understand potential determinants. Thailand's maternal mortality reduction began in the 1960s (Fig 5) at a time when skilled birth attendants, primarily midwives, were systematically trained and deployed to community hospitals (35-37). At the time of Alma Ata, Thailand's maternal mortality ratio was already below 200 and continued to drop even further in the 1980s as the Thai economy took off and a health care insurance program for low-income populations was introduced along with specific safe motherhood interventions. Another round of health system reforms and MNCH interventions were introduced in the early 2000s, including universal health coverage. Coordinated health policy support through successive national plans provided a context and investments to stimulate structural, financial and social capacities to deliver services, particularly in the district health system (38-40). Mandatory rural service for medical graduates has provided a stable human resource base within community hospitals (35). Using a log linear model, no single program could explain the decline in maternal mortality between 1960 and 1995, suggesting that the accelerated decline may be due to multiple developments. However, model fit after the 1997 economic crisis was not as good compared to earlier time periods. There was an increase in maternal mortality from 1997 to 2000 followed by a steady decline. This was in parallel with economic recovery and the introduction of universal coverage for health insurance and the MCH broad & healthy Thailand and Saiyairak programs (38-40). For this relatively short period, it is difficult to evaluate the effect of any intervention programs. The systematic deployment of community-based health personnel took place in Indonesia (41) about a decade later than in Thailand in the 1970s. Major, targeted safe motherhood initiatives were introduced in the late 1980s, but by that time Indonesia's maternal mortality ratio was about nine times higher than that of Thailand (Fig 6) (14, 42). A village midwife program was put in place between 1989 - 1996, but the comparatively rapid training and deployment of 54,000 village midwives may have compromised quality of care (41). Access to care in Indonesia varies by geography, rural/urban, poverty and education. Unlike in Thailand where the provision of skilled birth attendants was followed by increasing facility and referral level capacities, in Indonesia not all health centers can provide basic obstetric care. About 40% of district hospitals do not have an obstetrician (41), indicating limited provision of the 24-hour continuum of care necessary for dealing with emergency situations. A fragmented and devolved health system has challenged the capacity to sustain a comprehensive and concerted focus on maternal and child health. 6

Reductions in neonatal mortality (Fig 7) for the two countries mirror reductions in maternal mortality. Interventions to reduce neonatal mortality require more from health systems than either a maternal or child program alone (29). In Thailand, neonatal interventions have been linked with maternal programs (35, 39, 40), but this has not been documented in Indonesia. Maternal and neonatal mortality reductions in Thailand and Indonesia occurred in the context of rapid economic growth in both settings along with widespread increases in education levels and in gender equity (43). Although these factors may have influenced levels of success, other determinants of mortality have been at play. Policy implementation in Thailand has been multi-sectoral, involving Thai royalty and different ministries, including finance. Investments in primary health care in the 1970s have reaped dividends in the long term (31). However, geographic and demographic context likely also play a role. At the time of its rapid maternal mortality reduction in the 1970s, Thailand had a smaller, more circumscribed population compared to the larger and more dispersed Indonesian population and this difference may have critically determined physical access, a bottom line requirement for program coverage. Lives Saved Tool (LiST) Analysis We investigated the potential impacts of expanding program coverage through a LiST analysis of the ASEAN region as a whole, and for subgroups based on the mortality reduction patterns described earlier. ASEAN regional averages are closer to those of subgroups 2 and 3, where the bulk of the population and the higher mortality rates are (Fig 8a). While differences in maternal deaths averted across the groups are substantial, common trends across the region highlight critical gaps. Expanding coverage of interventions for hypertensive disease of pregnancy and safe management of abortions, for example, will reduce maternal deaths substantially throughout the region, while addressing postpartum hemorrhage causes will markedly reduce deaths for sub-groups 2 and 3 but not 1. Counterpart calculations were made for neonatal and child mortality (Fig 8 b and c). Interventions for birth asphyxia are more likely to avert deaths in subgroups 2 and 3. The high proportion of neonatal and child deaths averted through interventions for infectious diseases in all subgroups is indicative that infectious disease remains a challenge for the region as a whole (11). To focus on universal coverage, now receiving greater attention in other health policy circles (60), we present the deaths averted at 99% program coverage. At the regional level, universal basic obstetric care coverage will save about 1in 5 mothers (Table 1), but with universal comprehensive obstetric care coverage more than half of maternal deaths would be averted. Almost all of these lives saved would be in sub-groups 2 and 3 where current levels of coverage for these services are low (Fig 4). On the other hand, basic post-abortion case management will save a higher proportion of mothers in sub-group 1 but more deaths will actually be averted in sub-groups 2 and 3. The less lives saved in sub-group 1 with comprehensive as compared to basic abortion care may be due to the already high access to comprehensive obstetric care that could be utilized for abortion management in this group. Universal basic obstetric care will avert about 1 in 5 neonatal deaths in subgroup 3 while comprehensive obstetric care will save almost twice as many lives as basic care will for the region as a 7

whole but particularly for subgroups 2 and 3. Yet even at 99% coverage the maximum proportion of neonates that could be saved with either of these interventions does not go beyond a quarter of deaths, indicating the need for other interventions such as those addressing prematurity through antenatal steroids and providing Kangaroo care. Interventions directed towards infectious diseases such as diarrhea and pneumonia will, likewise, affect postneonatal and child deaths mostly in subgroups 2 and 3. A small but noticeable effect on death in subgroup 1 may also be seen when coverage increases to 99%. Preventive measures such as improving access to safe water can contribute substantially to mortality reduction, averting more deaths than Pneumococcal vaccination will in all the sub-groups. Discussion Despite significant improvements in maternal, neonatal and child health since 1990, most notably in Thailand, Malaysia and Vietnam, high mortality, poor coverage and high inequity continue to challenge other countries in the region, such as Cambodia, Laos and Myanmar. Improvements in the first three countries appear to be due to socioeconomic progress in part as well as to a consistent policy focus on maternal and child health programs and coordinated health systems components (33,44,45), notably a stable and strategically deployed health workforce coupled with supportive finance mechanisms in Thailand and Malaysia. The importance of favorable health systems is highlighted by the Thai case study which shows that 1) mortality reductions have taken place at modest levels of economic growth, and 2) no single factor or intervention could account for these reductions. Indonesia shows how similar interventions applied to a setting with differing system capacities and geopolitical features can result in differing outcomes. Thus, while the LiST analysis can estimate the potential impact of interventions given maximum levels of coverage, the case studies caution us that improving health outcomes is not just about increasing the amount of money spent in health. Rather, targeted and sustained interventions to reduce the barriers that prevent the most vulnerable population groups from accessing the interventions they critically need must be ensured in the long haul. For example, the LiST analysis indicates that providing basic and comprehensive emergency obstetric care has the potential to avert half of all maternal deaths and about 1 in 6 neonatal deaths in Cambodia, Laos and Myanmar. However, this may be possible only if care coverage is rapidly expanded among lower income and rural populations, possibly through sustained donor investments, given low levels of domestic health spending. In the case of the Philippines and Indonesia, the two most populous countries in the region, regaining momentum in mortality reduction in the face of considerable geographic and cultural access challenges may mean correcting inequitable access to services through health financing (60). Despite the varying agendas that countries of the region may need to adopt to complete the unfinished MNCH agenda, our findings indicate critical areas of common ground. Many key interventions to reduce child deaths have been rolled out at the community level throughout Southeast Asia, but necessary health service investments that will enable the countries of the region to gain traction on maternal and neonatal deaths have yet to be fully considered. Access to safe abortion services and management of hypertensive disorders during pregnancy will prevent maternal deaths of these causes from Singapore to Cambodia. Similarly, all countries in the region see space for expanding coverage of critical neonatal interventions to prevent pre-term births and neonatal deaths from infection.

Our study has several limitations. We have used estimates of mortality reduction from UN agencies that may not match national estimates, which use different methodologies; however these estimates are preferred for cross-country analysis. These estimates for Southeast Asia are still primarily derived from household surveys and subject to potential error from underreporting and misclassification of deaths. Only five of the ten countries considered have vital registration systems and not all of these systems have valid registration of causes of death (2). The need for better data is acute in Laos and Myanmar, particularly on maternal mortality, but Malaysias pioneering work in maternal death audits provides a potential model for the region (52). The selection of data sources was largely affected by the availability of comparable, reliable data across all the ASEAN countries. For example, while we used the more inclusive GNI/capita for Web Appendix Fig. 1, this measurement was not consistently available for the case studies, hence the use of GDP for the latter. We were not able to disaggregate LiST estimates into relevant national sub-groups by wealth or rural/urban status. This is important, since the poorest populations are likely to have higher mortality rates as well as lower levels of intervention coverage, and this could influence estimates (53). Data limitations restricted our ability to develop this analysis, even as a test case. We have also not analyzed the costs involved in extending coverage, which we hope to develop in a future study. A Challenge to ASEAN Since its formation in 1967, ASEAN has positioned itself as an important hub for economic and sociocultural cooperation. Infectious diseases have thus far commanded much of ASEAN's attention in health matters. Recently, regional focus has begun to shift to other health issues. The ASEAN Strategic Framework on Health Development 2010-2015 (54), which focuses on access to health care services besides communicable diseases and pandemic preparedness, has also gained regional support. Given the economic vigour of ASEAN, regional cooperation in health promises to be the key for less developed members to break free from systems inertia and address the low-lying fruits of maternal and child health and nutrition. The pivotal role played by ASEAN in stimulating and channeling international financial aid to tsunami-devastated Indonesia in 2004 and cyclone-stricken Myanmar in 2008 testifies to the power of this promise. But how should this aid be used? ASEAN experience, as distilled in this paper, suggests that effective interventions to curb maternal and child mortality must be deployed to actively seek out the disadvantaged populations who are most ravaged by unsafe abortion, hypertensive diseases, postpartum hemorrhage, pneumonia, sepsis and birth asphyxia. Far from expecting coverage of these programs to passively diffuse to the very poor, governments must innovatively combine health interventions with non-health programs such micro-finance schemes and conditional cash transfer mechanisms that have proven successful in other settings (55, 56). Achievement of the MDGs at the global level will not happen without individual country efforts. As the donor community focuses its attention on the burdens of Africa and South Asia, ASEAN countries must support each other.

References: 1. WHO. World Health Statistics 2010. Geneva: World Health Organization. 2. WHO. Trends in maternal mortality: 1990-2008.Geneva: World Health Organization, 2010. 3. UNICEF. Levels and trends in child mortality, Report 2010. Estimates developed by the UN Inter-agency Group for Child Mortality Estimation. 4. Countdown 2015. 2010 Country profiles for Cambodia, Indonesia, Lao PDR, Myanmar, and Philippines. http://www.countdown2015mnch.org/reports-publications/2010-country-profiles (accessed May 12, 2010) 5. Campbell OMR and Graham WJ, on behalf of The Lancet Maternal Survival Series steering group. Strategies for reducing maternal mortality: getting on with what works The Lancet - 7 October 2006 (Vol. 368, Issue 9543, Pages 1284 - 1299) 6. Darmstadt GL, Bhutta ZA , Cousens S, Adam T, Walker N, de Bernis L, for the Lancet Neonatal Survival Steering Team . Evidence-based, cost-effective interventions: how many newborn babies can we save? The Lancet - 12 March 2005 (Vol. 365, Issue 9463, Pages 977 - 988) 7. Kerber KJ, de Graft-Johnson JE, Bhutta ZA , Okong P, Starrs A ,LawnJE. Continuum of care for maternal, newborn, and child health: from slogan to service delivery. The Lancet - 13 October 2007 (Vol. 370, Issue 9595, Pages 1358 - 1369) 8. Bhutta ZA, Ali S, Cousens S, Ali TM, Haider BA, Rizvi A,Okong P, Bhutta SZ, Black RE. Interventions to address maternal, newborn, and child survival: what difference can integrated primary health care strategies make? The Lancet - 13 September 2008 (Vol. 372, Issue 9642, Pages 972 - 989) 9. Hogan M, Foreman K, Naghavi M, Ahn S, Wang M, Makela S, et al. Maternal mortality for 181 countries, 1980-2008: a systematic analysis of progress towards Millennium development Goal 5. Lancet. 2010; 375(9726):1609-1623. 10. Rajaratnam J, Marcus J, Flaxman A, Wang H, Levin-Rector A, Dwyer L, Costa M, Lopez A, Murray C. Neonatal, postnatal, childhood, and under-5-mortality for 187 countries, 1970-2010: A systematic analysis of progress towards Millennium Development Goal 4. The Lancet 2010(375): 1988-2007. 11. Black R, Cousens S, Johnson H, Lawn J, Rudan I, Bassani D, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet 2010; 375(9730):19691987 12. Khan KS, Wojdyla D, Say L, Glmezoglu AM, Van Look PFA. WHO analysis of causes of maternal death: a systematic review. The Lancet 2006; 367: 106674 13. UNICEF. State of the World's Children Reports, 1996-2010. New York: United Nations Children's Fund 14. Measure DHS. Demographic and Health Survey data for Cambodia 2000 and 2005, Indonesia 1987, 1991, 1994, 2002-03 and 2007, Philippines 1993, 1998, 2003 and 2008, and Vietnam 1997 and 2002. http://www.measuredhs.com (accessed Apr 27, 2010) 15. UNICEF. Multiple Indicator Cluster Surveys. http://www.childinfo.org/mics.html (accessed Jul 4,2010) 16. National Statistics Office. Monitoring the Situation of Children and Women Thailand Multiple Indicator Cluster Survey December 2005 February 2006 Final Report. http://www.unicef.org/thailand/MICS_Final_Report2.pdf (Accessed September 20, 2010) 17. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. SocioEconomic Differences in Health, Nutrition, and Population in Cambodia (Washington, D.C.: The World Bank, 2007). 18. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. SocioEconomic Differences in Health, Nutrition, and Population in Indonesia (Washington, D.C.: The World Bank, 2007). 19. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. Socio-

10

Economic Differences in Health, Nutrition, and Population in the Philippines (Washington, D.C.: The World Bank, 2007). 20. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. SocioEconomic Differences in Health, Nutrition, and Population in Vietnam (Washington, D.C.: The World Bank, 2007). 21. Plosky WD, Stover J, and Winfrey B. The Lives Saved Tool , A Computer Program for Making Child and Maternal Survival Projections. December 2009. 22. Jones G, Steketee RW, Black RE, Bhutta ZA, Morris SS. The Bellagio Child Survival Study Group. How many child deaths can we prevent this year? Lancet 2003; 362: 65-71. 23. Bhutta ZA, Ahmad T, Black RE, et al for the Maternal and Child Undernutrition Study Group. What works? Interventions for maternal and child undernutrition and survival. Lancet 2008; 371: 417-440. 24. Boschi-Pinto C, Young M, Black RE. The child health epidemiology reference group reviews of the effectiveness of interventions to reduce maternal, neonatal and child mortality. Int. J. Epidemiol. (2010) 39(suppl 1): i3-i6. 25. Chopra M, Daviaud E, Pattinson B, Fonn S, Lawn JE. Saving the lives of South Africa's mothers, babies, and children: can the health system deliver? Lancet 2009; 374: 835-846 26. Friberg IK, Kinney MV, Lawn JE, Kerber KJ, Odubanjo MO, et al. (2010) Sub-Saharan Africa's Mothers, Newborns, and Children: How Many Lives Could Be Saved with Targeted Health Interventions? PLoS Med 7(6): e1000295. doi:10.1371/journal.pmed.1000295 27. World Bank. World Bank database. http://siteresources.worldbank.org/DATASTATISTICS/ Resources/GNIPC.pdf (Accessed 19 Sept 2010 ) 28. Virasakdi C, KH Phua, MT Yap, NS Pocock, J Hashim, R Chhem, SA Wilopo and AD Lopez, Health in Southeast Asia: Diversity and Transitions, Lancet (overview paper, accepted for publication) 29. Lawn JE, Cousens S, Zupan J, and Lancet Neonatal Survival Steering Team. 4 million neonatal deaths: When? Where? Why? Lancet. 2005 Mar 5-11;365(9462):891-900. 30. Coker RJ, Hunter B, Rudge J, Liverani M, Hanvoravongchai P. The emergence of and response to infectious diseases in South East Asia. Lancet 2011 31. Rohde J, Cousens S, Chopra M, Tangcharoensathien V, Black R, Bhutta ZA, Lawn JE. 30 years after Alma-Ata: has primary health care worked in countries? Lancet 2008:372:950-61. 32. Countdown 2008 Equity Analysis Group, Boerma JT, Bryce J, Kinfu Y, Axelson H, Victora CG. Mind the gap: equity and trends in coverage of maternal, newborn and child health services in 54 Countdown countries. Lancet. 2008 Apr 12;371(9620):1259-67. 33. Limwattananon S, Tangcharoensathien V, Prakongsai P. Equity in maternal and child health in Thailand. Bull World Health Organ. 2010 Jun;88(6):420-7. Epub 2009 Dec 8. 34. Vapattanawong P, Hogan MC, Hanvoravongchai P, Gakidou E, Vos T, Lopez AD, Lim SS. Reductions in child mortality levels and inequalities in Thailand: analysis of two censuses. Lancet. 2007 Mar 10;369(9564):850-5 35. Wibulpolprasert S. Community financing: Thailand experience. Health Policy Plan 1991; 6: 354-360. 36. Van Lerberghe W, De Brouwere V. Of blind alleys and things that have worked: historys lessons on reducing maternal mortality. Studies Health Serve Organ Policy 2001; 17: 733. 37. Bureau of Policy and Strategy, Ministry of Public health. Health Policy in Thailand. Nonthaburi: Ministry of Public Health, 2009. 38. Wibulpolprasert S, editor. Thailand Health Profile 2005-2007. Bangkok: The War Veterans Organization of Thailand Printing Press; 2007. 39. Pramualratana P, Wibulpolprasert S. Health insurance systems in Thailand. Nonthaburi: Health System Research Institute, 2002. 40. WHO SEARO. Improving Maternal, Newborn and Child Health in the South-East Asia Region. Data source: Basic Indicators: Health Situation in South-East Asia, World Health Organization, South-East Asia Region, 2004.

11

41. Azwar A. Evolution Of Safe-Motherhood Policies In Indonesia. Presented in: IMMPACT Country Co-ordinating Group & Technical Partners Meeting, Aberdeen, January 2004. 42. Ministry of Health, Indonesia. National Health Surveys 1980 - 2008. 43. Gakidou E, Cowling K, Lozano R, Murray CJL. Increased educational attainment and its effects on child mortality in 175 countries between 1970 and 2009: a systematic analysis, Lancet 2010; 376: 95974. 44. UN Country Team Malaysia. Malaysia, Successes and challenges, Kuala Lumpur: UNDP, 2005. 45. Health Policy and Strategy Institute. Measuring Health Systems Performance in Vietnam: Results from Eight Provincial Health Systems Assessments. http://www.healthsystems2020.org/content/resource/detail/2515/ (accessed September 25, 2010) 46. UNESCAP. Development of Health Systems in the Context of Enhancing Economic Growth towards Achieving the Millennium Development Goals in Asia and the Pacific. Bangkok: ESCAP; 2007. 47. Jordan S, Lim L, Seubsman SA, Bain C, Sleigh A; the Thai Cohort Study Team. Secular changes and predictors of adult height for 86 105 male and female members of the Thai Cohort Study born between 1940 and 1990. J Epidemiol Community Health. 2010 Aug 30. [Epub ahead of print] 48. ASEAN Charter of the Association of Southeast Asian Nations. http://www.aseansec.org/21069.pdf (accessed September 22, 2010). 49. Institute for International Programs. LiST: The Lives Saved Tool An evidence-based tool for estimating intervention impact. http://www.jhsph.edu/dept/ih/IIP/list/index.html (accessed October 30, 2010 ) 50. Stover J, McKinnon R, Winfrey B. Spectrum: A model platform for linking impact of maternal and child survival intervention with AIDS, family planning and demographic projections. Int. J. Epidemiol. (2010) 39(suppl 1): i7-i10. 51. Walker N, Fischer-Walker C, Bryce J, Bahl R, Cousens S and writing for the CHERG Review Groups on Intervention Effects. Standards for CHERG reviews of intervention effects on child survival. Int. J. Epidemiol. (2010) 39(suppl 1): i21-i31 doi:10.1093/ije/dyq036. 52. Abu Bakar S, Mathews A, Jegasothy R, Ali R, Kandiah N. Strategy for reducing maternal mortality Bulletin of the WHO 1999; 77: 190-193. 53. Victora CG. Commentary: LiST: using epidemiology to guide child survival policymaking and programming. Int. J. Epidemiol. (2010) 39(suppl 1): i1-i2 doi:10.1093/ije/dyq044. 54. Association of Southeast Asian Nations. http://www.aseansec.org/24938.htm (accessed October 30, 2010). 55. Health Microinsurance: A Comparative Study of Three Examples in Bangladesh, by Mosieh Ahmed, Syed Khairul Islam, Md. Abul Quashem and Nabil Ahmed, CGAP working Group on Microinsurance, Case Study No. 13: September 2005. 56. Fiszbein A, Schady N. et al. Conditional Cash Transfers: Reducing Present and Future Poverty. (Washington, D.C.: The World Bank, 2009). 57. Estimation Methods Used by the UN Inter-agency Group for Child Mortality Estimation. http://www.childmortality.org/stock/documents/Methods%20for%20Estimating%20Child%20Mortality_2010.pd f (accessed October 30, 2010). 58. Wilmoth J, Zureick S, Mizoguchi N, Inoue M, Oestergaard M. Levels and Trends of Maternal Mortality in the World: The Development of New Estimates by the United Nations. http://www.who.int/reproductivehealth/publications/monitoring/MMR_technical_report.pdf (accessed October 30, 2010). 59. Caulfield LE, de Onis M, Blossner M, Black RE. Undernutrition as an underlying cause of child deaths associated with diarrhea, pneumonia, malaria and measles. Am. J. Clin. Nutr. 2004. 80:193-8. 60. Tangcharoensathien V, Patcharanarumol W, Ir P, Aljunid SM, Ghufron Mukti A, Akkhavong K, Banzon E, Huong DB, Thabrany H, Mills A, Health Financing Reforms in South East Asia: challenges in achieving universal coverage. Lancet 2010

12

Table 1. Percentage of deaths averted with 99% coverage of selected interventions for ASEAN region and sub-groups as defined by mortality reduction patterns

Maternal deaths averted (%) Interventions Basic emergency obstetric care (clinic) Comprehensive emergency obstetric care Post abortion case management, basic level Post abortion case management, comprehensive level Basic emergency obstetric care (clinic) Comprehensive emergency obstetric care Antenatal corticosteroids for preterm labor Kangaroo mother care Use of water connection in the home Pneumococcal vaccine Pneumonia case management (oral antibiotics) Zinc for diarrhea treatment Sub-group 1 * 1.6 9.2 4.4 Sub-group 2 19.4 55.1 6.1 10.3 Sub-group 3 21.5 55.6 6.1 10.3 ASEAN region 19.0 52.9 5.8 6.5

Neonatal deaths averted (%) * 0.1 7.3 20.3 3.1 2.9 5.7 1.2 11.7 22.8 20.1 20.6 13.4 9.8 19.0 5.0 17.2 31.1 17.2 16.8 11.3 7.6 15.9 3.9 12.8 24.1 18.5 19.4 12.3 8.7 17.4 4.4

Child (including postneonatal) deaths averted (%)

13

Fig 1a Trends in maternal mortality in Southeast Asia, 1990-2008 (data from WHO Maternal Mortality September 2010)

14

Fig. 1b Trends in infant mortality in Southeast Asia, 1990-2008 (Data from Unicef Child Mortality September 2010)

15

Fig. 1c Trends in under-five mortality in Southeast Asia, 1990-2008 (Data from Unicef Child Mortality September 2010)

16

Fig 2 Neonatal and Postneonatal Mortality Rate Reductions between 1990 and 2008 (Data from IHME, Rajaratnam et al, 2007)

17

Fig 3a. Causes of maternal deaths in Southeast Asia (Data from UN MDG Southeast Asia, 2010, includes 10 ASEAN countries + Timor Leste)

Fig 3b Causes of child deaths in Southeast Asia (Data from Black et al, 2010)

18

Fig 4. Inequities in MNCH intervention coverage in selected SEA countries Data from Gwatkin et al, 2007; National Statistics Office, www.childinfo.org)

19

20

21

Fig. 5. Trends in maternal mortality and MNCH programs in Thailand, by GDP 1960 - 2008

(Data from Thailand Ministry of Health)

22

Fig. 6. Trends in maternal mortality and safe motherhood programs in Indonesia, by GDP 1960 - 2008

(Data from Indonesian Ministry of Health)

Fig. 7. Neonatal Mortality reductions in Thailand (red) and Indonesia (blue) by GDP 1960-2008 (Data from Thailand and Indonesian Ministries of Health)

23

Fig 8a. Maternal deaths averted at 60, 90 and 99% coverage of interventions for the ASEAN region (red, orange and purple bars) and regional sub-groups as defined by mortality reduction patterns (blue, green and yellow bars)

Fig 8b. Neonatal deaths averted at 60, 90 and 99% coverage of interventions for the ASEAN region (red, orange and purple bars) and regional sub-groups as defined by mortality reduction patterns (blue, green and yellow bars)

24

Fig 8 c. Child deaths averted at 60, 90 and 99% coverage of interventions for the ASEAN region (red, orange and purple bars) and regional sub-groups as defined by mortality reduction patterns (blue, green and yellow bars)

25

Web Table 1 Data sources

Data Source Country Coverage Indicators Estimation methods Direct sisterhood method for maternal mortality estimates; direct method for estimation of child mortality rates; estimates of intervention coverage are weighted proportions Indirect estimation of infant and underfive mortality rates; estimates of intervention coverage are weighted proportions Limitations Only 4/10 countries covered, time periods per country variable; direct sisterhood method may underestimate maternal deaths; birth transference may underestimate infant mortality

DHS

Cambodia, Indonesia, Philippines, Vietnam

Time trends for neonatal, infant, under-five mortality, maternal mortality; intervention coverage (vaccination, IMCI, skilled birth attendance, delivery care)

MICS

Laos, Myanmar, Thailand, Vietnam

Estimation of infant, under-five mortality rates; intervention coverage (vaccination, IMCI, skilled birth attendance, delivery care)

Inter-agency Group for Child Mortality Estimation (UNICEF and WHO)

All 10 countries

Standardized estimation of time trends in infant, under-five mortality rates, 1990-2009 for 196 countries

Linear spline regression for countries without high HIV/AIDS prevalence, Loess regression for countries with high HIV/AIDS prevalence; weights applied for country data sources Gaussian process regression of under-five mortality accounting for nonsampling error; infant and neonatal mortality modeled from under-five mortality using multilevel regression Direct estimation from civil registration sources; multilevel regression model for other types of data sources

IHME: Rajaratnam et al. Maternal Mortality Estimation Inter-Agency Group (WHO)

All 10 countries

Standardized estimation of time trends in neonatal, infant, underfive mortality rates for 187 countries from 1970-2009 countries

All 10 countries

Standardized estimation of time trends in maternal mortality ration 1990-2008 for 172 countries

Earlier rounds of MICS subject to Only 3/10 countries covered Do not address potential mortality shocks other than HIV/AIDS epidemic; Loess forecasts of mortality decline may be conservative; vital registration available for 5/10 Southeast Asia countries Yields lower estimates of mortality than other methods; similar modeling approach for countries with and without high HIV prevalence; vital registration available for 5/10 Southeast Asian countries Underreporting/m isclassification of maternal deaths in household survey sources; did not include subnational data sources

26

IHME: Hogan et al.

All 10 countries

Standardized estimation of time trends in maternal mortality ratio for 181 countries, 1980-2008

CHERG:

All 10 countries

Cause of death estimation for neonatal, infant, under-fives

Two-step spatiotemporal regression Direct estimation with ICD-10 codes for complete vital registration data; for countries without complete vital registration, separate multicause multinomial regression models for countries with low and high child mortality using verbal autopsy sources; separation estimation of neonatal tetanus, malaria, measles, pertussis, HIV/AIDS

Underreporting/m isclassification of maternal deaths in household survey sources; idid not extract incidental pregnancyrelated deaths; mpact of HIV epidemic modeled directly; vital registration available for 5/10 Southeast Asian countries; Myanmar and Vietnam no national data available

Highest mortality countries of Southeast Asia do not have vital registration, estimates are modeled

27

Web Table 2 Effect sizes LiST References: Institute for International Programs. LiST: The Lives Saved Tool An evidence-based tool for estimating
intervention impact. http://www.jhsph.edu/dept/ih/IIP/list/index.html; Plosky WD, Stover J, and Winfrey B. The Lives Saved Tool , A Computer Program for Making Child and Maternal Survival Projections. December 2009.

28

29

30

Web Fig. 1 Mortality Reductions with GNI per capita Mortality data from UN MDG Reports September 2010; GNI per capita from World Bank Fig 1a Scatterplot of maternal mortality against GNI per capita in 1990 and 2008

31

Fig 1b Scatterplot of infant mortality against GNI per capita in 1990 and 2008

32

Fig 1c Scatterplot of under-five mortality against GNI per capita in 1990 and 2008

33

*Reply to Reviewers Comments

Reviewer #2: OVERALL This paper has improved considerably since the last review in July. The big picture is clearer and the authors have made considerable efforts to respond including to specific issues such as which countries are included. The shift from MCH to MNCH is appropriate especially in this region given the epidemiology. The trend analysis for maternal is improved and they have added a MMR/GNI plot although this is hard to interpret easily.. The LiST analysis adds novel data inputs, although is not clear which interventions are included and seems primarily maternal focused plus some neonatal and apparently no child. The program priorities are clearly stated for maternal (unsafe abortion, PPH and HDP) but are not at all clear for child and neonatal. Generally the paper remains much clearer for the maternal but is still weak for neonatal and child. It still requires more work to be a Lancet level paper with clear inputs, discussion of the limitations and yet clearer outputs especially for the child/neonatal parts.

1. Data used and sources cited Table 1 makes the data sources much clearer, although the column on estimation methods is so non-specific it does not add much and several cells are empty. This column could be deleted or else needs more detail to be of standard for Lancet. More information on estimation methods and on limitations has been added to Table 1 which has now been moved to the Web Appendix There is still some confusion in the text regarding sources especially of estimates for cause of child and neonatal deaths. These estimates are referred to as CHERG (note is Child Health Epidemiology Reference Group nor research group), Lancet paper by Black and colleagues, Countdown, and WHO. They are all the same. CHERG does the estimates with WHO, they were published by Black et al and then Countdown uses these estimates as do all of UN system. We have included the published source, reference 11, rather than the original source for easier tracking on the part of the ordinary Lancet reader Some remaining issues with the data a. Under five or infant mortality or neonatal? Given the MDG 4 target and the epi the appropriate measure are under five mortality and neonatal mortality. Infant mortality just parallels under five and the neonatal diffrs, links more to maternal and requires differing policy/program solutions and so is the outcome being tracked with under five by Countdown and most data for action movements. Fig 1 has under five and infant. Fig 3 has neonatal and postneonatal mortality. Postneonatal is only to one year of age and

MDG 4 is for under five mortality. Is this really postneonatal just up to infant or is it under five? There are several times throughout the paper where the outcome of interest is not clear especially in the child health sections. We have clarified our definitions for neonatal, postneonatal and child mortality in the text of the methods section on p 2 and made distinctions when writing about these different outcomes in the relevant sections on pages 3-4. b. Neonatal percentage of under five deaths: The table states that the data source for child and neonatal deaths is Black et al Lancet (CHERG/WHO) but gives the neonatal as "more than a quarter of child mortality" (page 4) and 26/7% in fig 4b. The SEA countries included in this paper are from WHO regions WPRO and SEARO but these regions have neonatal as 54% or 52% of under five deaths. Where does the data in fig 4c come from? Seems implausible to me although I have not added up the specific countries. I wonder if the authors have taken the neonatal % of infant instead of under five? Or some other explaination? This neonatal % of under five would be lower than Africa which has malaria and HIV in the postnatal and child proportion which pushed down the neonatal %. Does not seem at all likely that SEA has a lower neonatal % of under five deaths than Africa. Important to check this data carefully. The ASEAN countries are split between WPRO and SEARO, and one reason for singling out this region is to assess its performance outside the influence of larger states like India (which dominates SEARO) and China (which does likewise in WPRO). We have rechecked our data for Fig 4b (now renamed Fig 3b) and re-made the figure, which is discussed in the third full paragraph on p 4. c. Causes of death in the neonatal period: Neonatal is a time period not a cause and so should not be left as a slice in the pie - the UN estimates give the specific causes in the neonatal period and if this paper is to move the agenda forward then these causes should be listed and some actual interventions mentioned somewhere in the paper. Fig 4b is renamed Fig 3b and now includes the neonatal causes of mortality 2. LiST analysis This analysis adds novel data inputs. However the input and methods for LiST are not at all clear. The LiST methods discussion has been revised to add more detail beginning in the last paragraph on p 2. a. Interventions: it is not clear which interventions are included apart from BmoC and CEmOC (table 3). The BEmOC seems to save more maternal lives than the CEmoC which is surprising. Might this have been transposed? The LiST analysis seems primarily maternal focused with none of the specific neonatal high impact interventions after

childbirth care, many of which are highly appropriate for this region eg Kangaroo Mother Care. There seem to be no child interventions or nutrition interventions included?? Additional interventions are included and discussed for maternal, neonatal and child LiST analysis b. Coverage targets seem to be 60, 90 and 99% However many SEA countries will have baseline coverage above 60%? Not clear what was done. While there are ASEAN countries whose program coverages are beyond 60%, Fig 4 shows disparities in coverage within these countries, highlighting countries like Laos where coverage is much lower even among the well-off. Fig 4 is discussed in the second to last paragraph on p 4. Both overall coverage levels plus critical disparities provided a rationale for the selection the coverage levels used in the LiST analysis. c. Modelling methods: The text refers to two previous uses of LiST (Lancet South Africa series and PLoS MNCH series) but gives no details at all of the LiST model, even the link to the software and details or the more detailed papers describing the model (IJE March 2010). Other papers using LiST have given more details on the model, much more details on the specific analysis - eg which interventions, which coverage targets, which years. Also most such papers have a webtable of which interventions are included in the analysis and the key input assumptions eg effectiveness of the interventions, etc. This seems a minimum to give for a Lancet paper as readers should be able to understand the basis of the analysis. The LiST discussion has been revised in the discussion section non p 2 and the suggested references included. We have provided webtable 2 to show the levels of intervention effectiveness used in the analysis d. Results: see comment above about table 3. Also fig 9 and 10 are not easy to follow. For neonatal as diarrhea is about 2% of neonatal deaths could be left off the figure or at least not put first. Not clear why only maternal and neonatal and no child or nutrition esp as nutrition is highlighted as a problem Child interventions have been added in the LiST analysis through fig 8c and discussed on the first full paragraph on p7, but the nutrition discussion has been removed from the discussion of causes of child death as it was not contributing in a substantial way to the flow of the paper Case studies (Thailand and Indonesia) Fig 7 is a nice contribution with a detailed time and event mapping for Thailand's MMR reduction 1960 -2008 but the paragraph of text is still to very more analytical.

Additional analysis has been made and presented for the case studies. We discuss analytical methods and data used in more detail in the methods section on p 2 and have extended the rationale and discussion on p5. Then the text box is also on MMR reduction but is Malaysia - for a paper on MNCH it seems that having 2 case studies on MMR reduction is not very balanced. What about child, and/or TFR and/or adolescents? Also note some of the text in the text box is missing. We have not addressed this comment as we believe it was referring to a previous draft of the manuscript. The Indonesia fig (8) is a nice eg but is not very analytical and may be hard to follow for those not familiar with Indonesia. Both examples are wholly maternal which does not fit with the title of MNCH and the global shift to a linked MNCH agenda. This portion has been re-written on p 5 and a graph and discussion on neonatal mortality reduction in Indonesia and Thailand added, fig 7.

Messages from the data to inform policy and programme action There is some synthesis of the data trends - eg high inequity, increasing % of U5M that is neonatal, high prevalence of underweight children, yet rapid progress in almost all SEA countries (fig 1) etc. Hiwever there is still very limited focus on what this means for programmes - if high inequity is a major issue what can policy makers do? Targeting? Other strategies? Seems little point in describing them and not acting. The Brazil success for reducing disparities in mortality, under five stunting etc as per the Victora et al analysis in WHO Bulletin may give useful insights. Major trends in the region such as urbanization and models of care to reach the urban poor are not covered. Also as this series is in Lancet not a national or regional journal then other countries will be interested in lessons learned that may apply to other regions, and this remains weak. Some suggestions have been for policy and program action, although the aim is really for stronger solidarity within the ASEAN region for MNCH rather than to point out specific actions. In summary, this paper is much stronger than it was but still does reach the standards expected for a Lancet paper. The data inputs especially for child and neonatal have some technical questions outstanding. The LiST analysis is a useful addition but requires more description, discussion of limitations and application. There is limited critical analysis of differences between countries so reads more like a UN piece than a Lancet paper. Why have some countries progressed? If some are tailing off why is this? The implications and actions are not very clear for policy/accountability. The gaps are more notable for child/neonatal nutrition than for maternal.

Reviewer #3: The revised version of the manuscript is improved over the past document. However, there are still many important issues that need to be addressed. The issue of the selection of the countries as a viable or actual organization or institution still needs to be clarified or explained better.Other commeents will be in the dicussions below.

1. There is an improved attempt to defend the choice of the ASEAN countries as the focus for the study. The rationale to use this geographic grouping as countries needs to be clear and improved upon, as the issue of India, Pakistan and Bangladesh and even China are important as well. It is just that most regional analyses usually look at Asia as a bigger area and involve not only these countries in the paper, but also India, Pakistan, Bangladesh, and China. Why ASEAN? Why choose this aggrupation of countries? Would there be something special in this that would facilitate or enhance progress towards meeting the MDGs? Although the reply to the request to clarify the selection of this region or group of countries was mentioned, the reply was not convincing and clear enough. We included additional justifications for ASEAN on p 1, but also refer the reviewer to the Overview paper for this series on why this group of countries has been selected
Virasakdi C, KH Phua, MT Yap, NS Pocock, J Hashim, R Chhem, SA Wilopo and AD Lopez, Health in Southeast Asia: Diversity and Transitions, Lancet (overview paper, accepted for publication)

2. In the first paragraph of the Introduction, may I suggest changing "plateauing" to "leveling off"?. Plateau gives an indication of an flattened surface that is elevated, when the intent of the authors seem to suggest a levelling off at a declining surface. We changed the word "plateauing", as recommended on p 1 3. Some of the citations, such as the reference for the WHO mortality estimates have been published since and may need to be updated. We used the most recent reports from WHO and UNICEF, September 2010 as our key references, reference numbers 2 and 3. 4. The statement of the objectives in the last part of the introduction is not very clear. There also is not enough discussion on the significance of this paper, other than those issues that are already known to these countries. As suggested by the past reviews, there is still a lack of novelty in the paper, as the information presented would have been

known already to the supposed target audience. What is new to this paper, and what is the important message that needs to be mentioned with regards to the findings? We have highlighted the following key messages: major disparities, significant success stories, general progress on child health, more to be done on maternal and neonatal health, we provide recommendations on major interventions of which to expand coverage and recommend integrated, coordinated approaches within the health system to increase coverage. We have strengthened the rationale for the paper on p 1 of the introduction. We note in the methods section on p 2 that this type of LiST analysis undertaken in the paper has not been done for this region previously. 5. Since a lot of acronyms are mentioned, it may be helpful to spell out some of these in the first instance they appear in the paper, or in a glossary somewhere. We have spelled out any acronyms when first mentioned in the paper. 6. While the authors attempted to use many sources of information, there are inconsistencies between the methodologies, sampling, analysis and results of these various sources that the issue of comparability and using them for trends analysis may be brought up. There may be a need to expand the limitations column in the table of the data sources to give a deeper analytical view on these methodological variations that affects the data. It can be stated that the variations may be present, yet may triangulate each other to a certain pattern or trend. The Table on data sources has been expanded and moved to the Web Appendix as table 1. Limitations of the data sources are discussed beginning on p 8. 7. I am not sure if a reorganization of the paper to focus first on the rates and trends of a particular statistic, then its causes and factors, effects of interventions, and possible recommendations may be a better way. There still seems to be a lack of flow from one topic to the next. Regrouping the data together thematically may improve this flow. We considered the reviewers suggestion but were concerned that it would require extensive reworking of text. We have attempted to improve the links between the different parts of the results and discussion and hope this addresses the reviewer's concern 8. The case studies of Thailand and Indonesia focused only on maternal mortality reduction when the whole paper intends to address also newborn and child mortality. A graph, fig 7and discussion on neonatal reduction for Thailand and Indonesia has been added on p 6, first paragraph. We have also clarified the rationale for these case studies on p 5 at the beginning of the case study section in that we are not focusing specifically on maternal mortality reductions per se, but seeking to understand what factors account for Thailands more rapid pace of mortality reduction. The statistical model underlying figs 5 and 6 has been added in the methods section on p2 to clarify this.

9. On page 6, on the issue of Lives Saved (which is not a good heading for a section by the way), aside from the numbers in the table, there seems to be no further explanation or evidence on the effect of providing comprehensive care in certain subgroups. This analysis seems superficial or shallow, and may need to be further explained if the target audience of the paper are policy makers and programme planners. This portion has been re-written and the heading revised on p6. The discussion of the LiST for this paper is not intended to point out specific interventions to implement, but rather to illustrate the effect of contextual differences between and among the sub-groups on deaths averted, keeping coverage constant. This approach was taken so that policymakers may appreciate the interaction between coverage and the health system and sub-group context in setting expectations about the extent to which mortality can be reduced. 10. ASEAN is a political aggrupation of countries nearby each other within a geographically defined region. But aside from this, how much is Health in their agenda? Is there some coordinating mechanism to support any health measures? Have there been any past health issue that has been raised in ASEAN and has been addressed properly? Is there a way to include some description of the "health track record" of ASEAN? This concern has been addressed at the end of the paper on p 8. 11. Table 1 would probably need some reference citation or superscript citation. As mentioned previously, the limitations column should be expanded to give a more critical description of the use of the data. Table 1 has been moved to the Web Appendix and expanded as suggested 12. Why is Table 2 being presented? The table shows that there are differences between sources of information. This was not properly discussed in this paper, nor is a reference to this dicussion possibly in other sources is not found. The title denotes a comparison, yet there is no actual comparing in the paper. Table 2 has been deleted as not contributing substantially to the key messages of the paper. 13. Table 3 shows the subgroups in terms of levels of obstetric care. Are these subgroups referring to what? Are these the countries mentioned previously? Also the total in the last column under ASEAN does not add up from the previous columns. Where did this number come from? Is there a reference for this table, as it looks adapted from another source. If it is an originally produced table, the methodology in getting that information would b needed. Table 3 is now Table 1 and has been revised. This is an originally produced table and the methods section has been re-written to reflect this

14. Figure 1a, 1b and 1c seem to be copied and pasted. They are not very clear. Fig 1 a, b and c have been revised; Brunei and Singapore (which are high income countries) were removed for better clarity 15. Figure 2 states at the bottom that the plots for IMR and U5MR are similar to MMR: If such is the case for the pattern in putting figures, it would not be proper not to put the other figures, as people may want to use this paper as a reference for information, if it gets published. Fig 2 has been renmaed Fig 1a, b and c for the Web Appendix and includes the IMR and U5MR plots against GNI per capita 16. Figure 3 is vague. What is the value of providing the ifnromation on rate reduction of neonatal and post neonatal mortality, without a proper discussion. And why is such a table presented when the rate of maternal mortality reduction may also be important, being oneof the main objectives of this paper. Fig 3 has been renamed Fig 2 and its related discussion improved on p3. We use fig 3 to show that reductions in neonatal mortality have not been consistent across countries of the region, nor have they kept pace with postneonatal mortality reductions. This then links to the LiST analysis on the need to expand coverage of targeted neonatal mortality interventions. The rate of maternal reduction has been extensively written about in relation to Fig 1a and the case studies. We made a graph on the rate of maternal reduction in response to this comment, but we find that it does not provide anything additional to the paper so we have decided not to present it. 17. Figure 5. The information of underweight children seems to be out of place, as most of the paper is discussing mortality. This Fig and the corresponding discussion on underweight has been deleted 18. Figure 6. The bar diagrams continue to be small, and the information provided is vague and not useful. It may be helpful if the authors insist on including these figures to explain each trend or finding separately, and to describe the changes within in each country and across countries more clearly. Fig 6 has been renamed Fig 4 and enlarged. Additional explanations re: this figure have been made in section beginning on p 4 19. Figure 7 and 8 would need references. Fig 7 and 8 have been renamed Fig 5 and 6; references have been included in the text and the figures

20. Figure 9 and 10 would need references, and some clarification if the subgroups mentioned here are the same as in other parts of the paper. Fig 9 and 10 have been renamed Fig 8 a, b and c (a graph on interventions to reduce child mortality has been added to address another reviewer's comment) and these figures come from primary data so references are not needed; clarification regarding the subgroups has been added to the text on p6 21. In the last page of the main body of text, there is a statement on "ASEAN neighbors may support each other in the spirit of South to South cooperation, ......". What do the authors exactly mean by this recommendation, which reads like a motherhood statment, without clarifying specific programmatic implications. This statement has been deleted and this paragraph re-written

Reviewer #4:

A very interesting investigation, though at the moment the figures really let the manuscript down. 1. Due to the different patterns of the countries some countries are almost impossible to see on the Figures because they were already low on 1990 and therefore are hidden from view. Figures 1 a, b and c have been revised, removing Brunei and Singapore (which are considered high income countries) for clarity 2. The x-axis on Figure 1b is not ideal. Time should be evenly spaced.

This has been addressed in the revised graphs 3. The percentages expressed and the details in table 1 are not clear. The total does not seem to be the sum of the constituent parts. The table with the LiST interventions being referred to has been revised 4. The graph in Figure 2 is poorly showing the data as most of the points are too close to the point of origin, and the two different time points are not easily distinguishable. Fig 2 has been renamed Fig 1 in the Web Appendix and has been revised for better clarity; Brunei and Singapore have been removed from this Figure as well

5.

What are the additional lines on Figure 8, they obscure the text.

The revised Fig 8 is now Fig 6 and these extraneous lines have been deleted 6. The authors have not really discussed the differences between the surveys; these are fundamental differences, though the patterns over time are more similar. This leads to a wider point that the limitations of the data sources are not really fully detailed. The discussion does not go into the limitations in the required depth at all. Additional explanations including details of data sources have been added in the methods section to address this, including extension of what is now webtable 1. A section on limitations of the study has been added in the discussion beginning on p8. 7. On a related point the impact of the health changes is not really discussed in enough detail to understand whether the changes mentioned will really have the impact as the data with the estimate of the impact is not provided and the limitations not discussed at all. We are not sure whether the reviewer is referring to the case studies with this comment or to the LiST results - improvements have been in the discussion of both these sections which we hope addresses this concern. Limitations of data sources and analysis have been added to the discussion section on p 8. 8. A LiST analysis does not easily have a confidence interval in the normal sense of the word, but it is usual to discuss scenarios for providing some form of a range on the estimates provided. This is a separate issue to the coverage ranges which have the same error.
LiST is based on point estimates and currently does not have the ability to provide uncertainty bounds in a meaningful way. One could do individual countries and then do median effect sizes and range / 95% CI but would not be appropriate given that we are now presenting pooled results and regional bands. The number of countries in each group is also small. So the simpler approach is to restrict the model to estimates at varying levels of coverage and see where the gains are.

*Manuscript with revisions highlighted

Title: Maternal, neonatal and child health: Now (more than ever) for Southeast Asia
Order of Authors: Cecilia S Acuin; Geok Lin Khor; Tippawan Liabsuetrakul; Endang L Achadi; Thein Thein Htay; Rebecca Firestone; Zulfiqar A Bhutta,

Key messages Southeast Asia has sustained substantial reductions in maternal, neonatal and child mortality since 1990, but this progress has been uneven. Mortality reductions in some countries have been the result of trajectories of rapid decline begun well before the MDG period. Others have succeeded in acerbating progress since 1990s, but some countries continue to struggle. Causes of death suggest a mortality transition in maternal deaths in the region. Child deaths are primarily due to the persistence of neonatal causes along with key preventable factors in the postneonatal period Disparities in intervention coverage are most acute in the countries with the lowest intervention coverage overall Despite the mixed picture, some countries stand out as success stories. Suggested key factors include the ability to link MNCH interventions to broader health system investments and to target access to rural and disadvantaged populations Increasing coverage to 60% will have a significant impact on maternal deaths caused by unsafe abortion, hypertensive diseases and postpartum hemorrhage and neonatal deaths caused by pneumonia/sepsis and birth asphyxia. While MNCH may have no quick solutions in the region, coordinated expansion of proven effective interventions can contributed to greater mortality reductions There is a need for stronger regional cooperation through ASEAN to support countries that need to accelerate progress to meet the MDGs
Abstract: While maternal and child mortality are on the decline in Southeast Asia, the region is witness to major disparities across and within its ten member countries, and greater equity will be key to achieving the MDGs. We used comparable cross-national data sources to document mortality trends from 1990 in a standardized approach and to assess major causes of maternal and child deaths. We present inequalities in intervention coverage by two common measures of wealth quintiles and rural/urban status. Case studies of mortality reduction in Thailand and Indonesia present the mixed picture of success within this diverse region and point to some of the factors that may be capitalized on to accelerate progress in the future. We developed a Lives Saved Tool (LiST) analysis for the region as a whole and for country subgroups defined by mortality trends to estimated deaths averted by cause and intervention at varying coverage levels. We found three major patterns of maternal and child mortality reduction: 1) early, rapid downward trends- Brunei, Malaysia, Singapore, Thailand; 2) initially high declines sustained by Vietnam but faltering in Philippines, Indonesia; 3) high initial rates with a downward trend that require more focus to accelerate - Cambodia, Myanmar, Laos. High achievers in the region point to early declines before rapid economic growth, suggesting that economic development provides an important context that must be coupled with broader health system interventions. Increasing coverage will have a significant impact on maternal deaths by unsafe abortion, hypertensive diseases and postpartum hemorrhage , neonatal deaths caused by pneumonia/ sepsis and birth asphyxia and child deaths caused by infectious diseases. These actions will require consideration of health system contexts, and a regional push through ASEAN may provide greater policy support to achieve MNCH goals. Funding The Rockefeller Foundation and China Medical Board provided funds that allowed the authors of this paper to meet and to access research assistance.

Introduction Southeast Asia has achieved substantial reductions in child and maternal mortality within a relatively short period of time, but these achievements are unevenly distributed among the countries in the region. Of the ten countries in the political coalition known as the Association of Southeast Asian Nations (ASEAN), only three have infant and child mortality rates below 10/1000 live births, namely Brunei, Singapore and Malaysia. Infant and under-five mortality in Thailand and Vietnam have declined dramatically to below 15, but the Philippines and Indonesia have seen a leveling off of rates in the 30s and 40s. Myanmar, Cambodia and Laos still see mortality levels comparable to their neighbors more than two decades ago and rank among the highest for Asia (1). The United Nations estimates that each year approximately 350,000 women die as a result of pregnancy or childbirth (2), as do nearly 9 million children below 5 years (3). Southeast Asia contributed approximately 18,000 maternal (2) and 400,000 child (3) deaths to this global burden in 2008. Two of the seven countries with the highest maternal mortality ratios outside of sub-Saharan Africa are Laos and Cambodia, while Indonesia is among the countries accounting for 65% of all maternal deaths worldwide. While Southeast Asia, as a region, may achieve the child mortality reductions set by the United Nations Millenium Development Goal 4 (MDG4), Cambodia and Myanmar have been rated as showing "insufficient progress" (4). Declines in Laos, Indonesia and the Philippines are also faltering. Similarly, while all countries demonstrate declines in maternal deaths, the rates of decline for Indonesia, Myanmar and the Philippines have slowed considerably. Effective and affordable technology to reduce the majority of maternal, newborn and child deaths is at hand (5, 6, 7, 8), so why has progress been so uneven? This paper focuses on a region, collectively the 9th largest economy in the world, whose performance and achievements are often hidden by larger states like India or China and UN agency regional groupings that do not take into account historical and geo-political ties within ASEAN (28). Southeast Asia as a region has received markedly little attention in recent efforts to revitalize and strengthen the MNCH policy agenda, despite the complexity of national trends, including the significant burden of morbidity and mortality in several countries and the existence of documented successes (31). As Southeast Asia becomes more integrated economically, there is a growing need to take stock of this major unfinished agenda and identify policy options for sustaining if not accelerating the pace of mortality reduction. This paper aims to critically review the regions achievements in maternal and child mortality reduction and point to key factors that explain success and challenges in reaching these goals amidst competing global, regional and national health problems. We first report on patterns of mortality reduction within Southeast Asia as well as major causes of maternal and child deaths in the context of MDGs 4 and 5. We investigated two country cases to illustrate the significant variations in mortality reduction. Finally, we used an analysis of the deaths to be averted through expanded coverage to identify more effective approaches for pursuing maternal, neonatal and child health in Southeast Asia.

Comment [RF1]: Wording revised per reviewer 3

Comment [RF2]: References checked and updated per reviewer 3

Comment [RF3]: Justification of ASEAN focus extended and rationale strengthened per reviewer 3

Methods Data sources For the ten countries considered here, we reviewed estimates from national data sources including Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS), as well as global datasources from UNICEF, WHO, and the Institute for Health Metrics and Evaluation (IHME) (webTable 1). We present country-specific estimates on maternal, infant, and under-five mortality rates from recent UN MDG reports (2, 3), as these estimates enable cross-country comparisons on trends in mortality using replicable estimation methods that seek to reduce sources of non-sampling error. These estimates tend to be more conservative in the rate of decline than IHME estimates. Based on growing awareness of the burden of neonatal mortality, we sought comparable estimates of trends disaggregating neonatal (death within the first 28 days following birth) and postneonatal (death between 28 days and one year following birth) mortality. As UN models do not have neonatal time trends for all countries of the region, we present estimates from IHME (10). We present estimates on causes of neonatal and child deaths based on standardized methods for estimating the distribution of causes of child deaths (11). We have compiled estimates of causes of maternal deaths from Countdown 2015 country reports (4) and WHO (12) and we used annual State of the Worlds Children reports (13) to review standard assessment of nutrition trends (web Appendix Fig 2). Data from DHS (14) and MICS (15, 16) were evaluated to assess existing intervention coverage within the region, with these data sources providing the ability to disaggregate coverage estimates by wealth quintile and rural/urban status (17-20) to unpack country average and asses program coverage among disadvantaged populations. Regional estimates were calculated using country-level data from the specific source cited, unless otherwise indicated. Analysis Case studies were developed to test the contribution of health sector inputs to mortality reductions. Thailand and Indonesia were selected as cases, as high and lower achievers, and we focused on maternal and neonatal mortality as outcomes sensitive to health system development. National data were used for case studies in order to extend the analysis to the period prior to the MDG baseline year (38, 42). For the case studies, data was fitted to a quadratic equation (log10MMR or log10NMR = Intercept + linear effect of year+ quadratic effect of year) as well as a linear equation (log10 MMR or log10NMR= Intercept + linear effect of year) to determine whether declines in maternal mortality could be attributed to program changes or temporal trends. We calculated potential deaths to be averted through increasing population coverage of the interventions proven to be effective in reducing maternal, neonatal and child mortality using the Lives Saved Tool (LiST) (21,49). LiST operates within the Spectrum modelling platform, initially developed to project demographic change and complemented by modules to model the impact of family planning and HIV/AIDS interventions (50). The model yields estimates of deaths averted by cause and intervention for user-specified intervention coverage levels, based on inputs of demographic projections, numbers of maternal and child deaths, data on the distribution of deaths by cause, intervention effectiveness, and data on local health status (6, 22, 23, 24). The tool has been previously used for analysis of impact of intervention packages on maternal and child survival in South Africa (25) and sub-Saharan Africa, but this is a first application in Southeast Asia (26). For this analysis, we assessed all of the MNCH interventions included in LiST (51). The interventions and their 3

Comment [RF4]: Time period of outcomes of interest clarified per reviewer 2; rationale for showing neonatal and postneonatal rate reductions extended per reviewer 3 Comment [RF5]: References to CHERG in text removed per reviewer 2

Comment [RF6]: References to data sources added per reviewer 3

Comment [RF7]: Information on data and analysis undertaken for case studies added per reviewer 2

Comment [RF8]: Significance of analysis added per reviewer 3

effectiveness estimates are shown in webtable 2. Values for the effectiveness of interventions were developed through a standardized review process using established criteria to determine which interventions to include based on levels of evidence (51). The analysis was conducted for all ten countries and then for three subgroups of countries based on observed patterns of mortality reduction: subgroup 1 (Brunei, Malaysia, Singapore, Thailand); subgroup 2 (Indonesia, Philippines, Vietnam); subgroup 3 (Cambodia, Laos, Myanmar). We assessed potential lives saved at three hypothetical coverage levels: 60%, 90% and 99%. Results Patterns of Mortality Reduction Reduction in maternal, infant and child mortality (Fig1a-Fig 1c, data not shown for Brunei and Singapore) in Southeast Asia reflect the diversity of this region, presenting three divergent patterns (27, 28). The first pattern reflects countries achieving low mortality rates between 1990, the MDG baseline year, and 2008 in Brunei, Singapore, Malaysia and Thailand. In 1990, maternal mortality ratios in these countries were well below 100/100,000 live births, and infant and under-five mortality rates were already at or below 20/1000 live births. These most economically advanced countries in the region have also invested in their health systems over time. A second, less distinct, pattern, seen in the Philippines, Indonesia and Vietnam, starts with relatively high mortality rates and ratios in 1990, fairly large initial reductions (except for Indonesia's maternal mortality ratio) that somewhat faltered after 2000 in Indonesia and the Philippines. In contrast, Vietnam witnessed accelerated mortality reductions during this period, with mortality rates and ratios beginning to come close to those of Thailand. The third pattern, observed in Laos, Cambodia and Myanmar has very high levels at the beginning of 1990, followed by sustained reductions from 1990 to 2005 with the exception of Cambodia's maternal mortality ratio. These three countries, which are on the UN list of least developed countries, continue to experience high maternal, infant and child mortality. Plotting maternal mortality reductions against Gross National Income (GNI) per capita (Web Appendix Fig 1a) indicates that while countries with high maternal mortality achieved reductions in mortality as their GNI per capita increased, some of the most dramatic declines in mortality took place earlier than the rapid rise in GNI. Thailand's rapid maternal mortality reductions occurred pre-1990. As maternal mortality declined to levels around 100, smaller reductions take place even as GNI continues to improve. Similar patterns are observed for infant and under 5 mortality vs GNI per capita plots (Web Appendix Fig 1 b and c). Neonatal and Postneonatal Mortality Reductions Disaggregating infant mortality reduction between 1990 and 2010 into neonatal and postneonatal (Fig 2) indicates that the largest declines in infant mortality over time were mainly due to substantial postneonatal mortality reductions, as seen in Vietnam, Malaysia and Thailand. The Philippines and Indonesia had neonatal and postneonatal reductions comparable to Cambodia, Laos and Myanmar. Although starting with comparably lower baseline mortality levels in 1990, rates of decline in these two countries were not sufficiently accelerated during the MDG period. Reductions in infant mortality in

Comment [RF9]: More information on modeling methods, intervention effectiveness, references added per reviewer 2

Comment [RF10]: Time periods of outcomes of interest clarified per reviewer 2

Brunei and Singapore stem from larger proportions of decline in neonatal deaths, a pattern similar to other high income countries (29). The slower rates of decline for neonatal mortality for 8 of 10 ASEAN countries is a cause for concern. Interventions for reducing neonatal mortality are more closely linked to maternal interventions in terms of policy and program implementation and may not be as noticeably tracked towards their impact on under-5 mortality. The Philippines, which is considered to be "on track" in achieving child mortality reductions (4), has the lowest reduction in neonatal mortality in the region, lower than Cambodia or Myanmar, which have been identified as showing "insufficient progress" towards achieving MDG 4. Causes of Mortality The distribution of maternal mortality causes (Fig 3a) reflects the significant variations in health status and health system development seen within the region. Haemorrhage is a leading cause of death, likely indicative of delays in attaining emergency obstetric care. Hypertensive disorders contribute to about one in every six maternal deaths in Southeast Asia and suggest a different causal pathway more comparable to developed country settings. The proportion of other indirect causes may indicate the still significant burden of infectious disease within the region and the impact of malaria and HIV on maternal health (30). Unsafe abortion is a factor in almost one-tenth of maternal deaths. These patterns reflect a causal transition in maternal mortality as the overall risk of maternal death declines, and they will influence the extent to which interventions, both as single modalities or included in a package, can be predicted to avert deaths. (12). Differential rates of child mortality reduction can be attributed in part to variations in causes of death (Fig 3b). Neonatal conditions contributed approximately 40% of child mortality, accounting for the single largest proportion of preventable deaths, even as a number of the ASEAN countries are successfully reducing their postneonatal and child mortality burdens. Infectious diseases including pneumonia, diarrhea and others, still account for almost half of child deaths, reflecting considerable scope for continued reductions in child mortality (11).

Comment [RF11]: Calculations of estimate updated per reviewer 2

Within Country Disparities in Intervention Coverage Inequalities are substantial across countries of the region, but also within, as shown by the current scope of intervention coverage by income and rural/urban sub-groups (Fig 4), considering antenatal care coverage, use of skilled birth attendance (SBA), DPT and measles vaccination along with use of oral rehydration therapy, all key to the development of a continuum of care (4, 7). Laos remains substantially lower than other countries on overall program coverage and far from a 60% coverage level even for the most well-off groups. Antenatal care coverage is most widespread, being close to or above 90%, in countries other than Laos and Cambodia, for the most well-off and urban areas, suggesting that there is scope to effectively scale up prenatal interventions that can avert maternal deaths. Vaccination coverage varies widely, although several countries in the region are GAVI-eligible and have received substantial financial and policy support likely to lead to increases in vaccination coverage over time. Laos and Cambodia present the greatest disparities in program coverage. Vaccination levels for the 5

Comment [RF12]: Text of this section revised to clarify key national trends per reviewer 3

Comment [RF13]: Overview of overall coverage levels added per reviewer 2

wealthiest quintile are comparable to those of the other SEA countries, but they are well below 50% amongst the poorest households. The countries represented in Fig 4 demonstrate relatively low coverage of skilled birth attendance (except Thailand) with inequality particularly acute in Laos, Cambodia and the Philippines. Skilled birth attendance may be viewed as one indicator of broader health system development, and the generally low coverage coupled with a high degree of inequality illuminates the need for more comprehensive and coordinated health system strengthening in the region overall (31). These patterns also point to the necessity of targeting the most vulnerable populations and maintaining attention to equity while scaling up program coverage (32). Case studies: Thailand and Indonesia We look in more depth at two countries with diverging experiences of mortality reduction to understand potential determinants. Thailand's maternal mortality reduction began in the 1960s (Fig 5) at a time when skilled birth attendants, primarily midwives, were systematically trained and deployed to community hospitals (35-37). At the time of Alma Ata, Thailand's maternal mortality ratio was already below 200 and continued to drop even further in the 1980s as the Thai economy took off and a health care insurance program for low-income populations was introduced along with specific safe motherhood interventions. Another round of health system reforms and MNCH interventions were introduced in the early 2000s, including universal health coverage. Coordinated health policy support through successive national plans provided a context and investments to stimulate structural, financial and social capacities to deliver services, particularly in the district health system (38-40). Mandatory rural service for medical graduates has provided a stable human resource base within community hospitals (35). Using a log linear model, no single program could explain the decline in maternal mortality between 1960 and 1995, suggesting that the accelerated decline may be due to multiple developments. However, model fit after the 1997 economic crisis was not as good compared to earlier time periods. There was an increase in maternal mortality from 1997 to 2000 followed by a steady decline. This was in parallel with economic recovery and the introduction of universal coverage for health insurance and the MCH broad & healthy Thailand and Saiyairak programs (38-40). For this relatively short period, it is difficult to evaluate the effect of any intervention programs. The systematic deployment of community-based health personnel took place in Indonesia (41) about a decade later than in Thailand in the 1970s. Major, targeted safe motherhood initiatives were introduced in the late 1980s, but by that time Indonesia's maternal mortality ratio was about nine times higher than that of Thailand (Fig 6) (14, 42). A village midwife program was put in place between 1989 - 1996, but the comparatively rapid training and deployment of 54,000 village midwives may have compromised quality of care (41). Access to care in Indonesia varies by geography, rural/urban, poverty and education. Unlike in Thailand where the provision of skilled birth attendants was followed by increasing facility and referral level capacities, in Indonesia not all health centers can provide basic obstetric care. About 40% of district hospitals do not have an obstetrician (41), indicating limited provision of the 24-hour continuum of care necessary for dealing with emergency situations. A fragmented and devolved health system has challenged the capacity to sustain a comprehensive and concerted focus on maternal and child health. 6

Comment [RF14]: Rationale for case studies extended per reviewer 3

Reductions in neonatal mortality (Fig 7) for the two countries mirror reductions in maternal mortality. Interventions to reduce neonatal mortality require more from health systems than either a maternal or child program alone (29). In Thailand, neonatal interventions have been linked with maternal programs (35, 39, 40), but this has not been documented in Indonesia. Maternal and neonatal mortality reductions in Thailand and Indonesia occurred in the context of rapid economic growth in both settings along with widespread increases in education levels and in gender equity (43). Although these factors may have influenced levels of success, other determinants of mortality have been at play. Policy implementation in Thailand has been multi-sectoral, involving Thai royalty and different ministries, including finance. Investments in primary health care in the 1970s have reaped dividends in the long term (31). However, geographic and demographic context likely also play a role. At the time of its rapid maternal mortality reduction in the 1970s, Thailand had a smaller, more circumscribed population compared to the larger and more dispersed Indonesian population and this difference may have critically determined physical access, a bottom line requirement for program coverage. Lives Saved Tool (LiST) Analysis We investigated the potential impacts of expanding program coverage through a LiST analysis of the ASEAN region as a whole, and for subgroups based on the mortality reduction patterns described earlier. ASEAN regional averages are closer to those of subgroups 2 and 3, where the bulk of the population and the higher mortality rates are (Fig 8a). While differences in maternal deaths averted across the groups are substantial, common trends across the region highlight critical gaps. Expanding coverage of interventions for hypertensive disease of pregnancy and safe management of abortions, for example, will reduce maternal deaths substantially throughout the region, while addressing postpartum hemorrhage causes will markedly reduce deaths for sub-groups 2 and 3 but not 1. Counterpart calculations were made for neonatal and child mortality (Fig 8 b and c). Interventions for birth asphyxia are more likely to avert deaths in subgroups 2 and 3. The high proportion of neonatal and child deaths averted through interventions for infectious diseases in all subgroups is indicative that infectious disease remains a challenge for the region as a whole (11). To focus on universal coverage, now receiving greater attention in other health policy circles (60), we present the deaths averted at 99% program coverage. At the regional level, universal basic obstetric care coverage will save about 1in 5 mothers (Table 1), but with universal comprehensive obstetric care coverage more than half of maternal deaths would be averted. Almost all of these lives saved would be in sub-groups 2 and 3 where current levels of coverage for these services are low (Fig 4). On the other hand, basic post-abortion case management will save a higher proportion of mothers in sub-group 1 but more deaths will actually be averted in sub-groups 2 and 3. The less lives saved in sub-group 1 with comprehensive as compared to basic abortion care may be due to the already high access to comprehensive obstetric care that could be utilized for abortion management in this group. Universal basic obstetric care will avert about 1 in 5 neonatal deaths in subgroup 3 while comprehensive obstetric care will save almost twice as many lives as basic care will for the region as a 7

Comment [RF16]: NMR trends added to shift focus of case studies away for solely focusing on MMR, per reviewers 2 and 3

Comment [RF17]: New heading per reviewer 3

Comment [RF18]: Estimates of BemOC and CemOC checked per reviewer 2

whole but particularly for subgroups 2 and 3. Yet even at 99% coverage the maximum proportion of neonates that could be saved with either of these interventions does not go beyond a quarter of deaths, indicating the need for other interventions such as those addressing prematurity through antenatal steroids and providing Kangaroo care. Interventions directed towards infectious diseases such as diarrhea and pneumonia will, likewise, affect postneonatal and child deaths mostly in subgroups 2 and 3. A small but noticeable effect on death in subgroup 1 may also be seen when coverage increases to 99%. Preventive measures such as improving access to safe water can contribute substantially to mortality reduction, averting more deaths than Pneumococcal vaccination will in all the sub-groups. Discussion Despite significant improvements in maternal, neonatal and child health since 1990, most notably in Thailand, Malaysia and Vietnam, high mortality, poor coverage and high inequity continue to challenge other countries in the region, such as Cambodia, Laos and Myanmar. Improvements in the first three countries appear to be due to socioeconomic progress in part as well as to a consistent policy focus on maternal and child health programs and coordinated health systems components (33,44,45), notably a stable and strategically deployed health workforce coupled with supportive finance mechanisms in Thailand and Malaysia. The importance of favorable health systems is highlighted by the Thai case study which shows that 1) mortality reductions have taken place at modest levels of economic growth, and 2) no single factor or intervention could account for these reductions. Indonesia shows how similar interventions applied to a setting with differing system capacities and geopolitical features can result in differing outcomes. Thus, while the LiST analysis can estimate the potential impact of interventions given maximum levels of coverage, the case studies caution us that improving health outcomes is not just about increasing the amount of money spent in health. Rather, targeted and sustained interventions to reduce the barriers that prevent the most vulnerable population groups from accessing the interventions they critically need must be ensured in the long haul. For example, the LiST analysis indicates that providing basic and comprehensive emergency obstetric care has the potential to avert half of all maternal deaths and about 1 in 6 neonatal deaths in Cambodia, Laos and Myanmar. However, this may be possible only if care coverage is rapidly expanded among lower income and rural populations, possibly through sustained donor investments, given low levels of domestic health spending. In the case of the Philippines and Indonesia, the two most populous countries in the region, regaining momentum in mortality reduction in the face of considerable geographic and cultural access challenges may mean correcting inequitable access to services through health financing (60). Despite the varying agendas that countries of the region may need to adopt to complete the unfinished MNCH agenda, our findings indicate critical areas of common ground. Many key interventions to reduce child deaths have been rolled out at the community level throughout Southeast Asia, but necessary health service investments that will enable the countries of the region to gain traction on maternal and neonatal deaths have yet to be fully considered. Access to safe abortion services and management of hypertensive disorders during pregnancy will prevent maternal deaths of these causes from Singapore to Cambodia. Similarly, all countries in the region see space for expanding coverage of critical neonatal interventions to prevent pre-term births and neonatal deaths from infection.

Comment [RF19]: Discussion of child deaths added per reviewer 2

Comment [RF20]: Cross-cutting programmatic recommendations added per reviewer 2

Our study has several limitations. We have used estimates of mortality reduction from UN agencies that may not match national estimates, which use different methodologies; however these estimates are preferred for cross-country analysis. These estimates for Southeast Asia are still primarily derived from household surveys and subject to potential error from underreporting and misclassification of deaths. Only five of the ten countries considered have vital registration systems and not all of these systems have valid registration of causes of death (2). The need for better data is acute in Laos and Myanmar, particularly on maternal mortality, but Malaysias pioneering work in maternal death audits provides a potential model for the region (52). The selection of data sources was largely affected by the availability of comparable, reliable data across all the ASEAN countries. For example, while we used the more inclusive GNI/capita for Web Appendix Fig. 1, this measurement was not consistently available for the case studies, hence the use of GDP for the latter. We were not able to disaggregate LiST estimates into relevant national sub-groups by wealth or rural/urban status. This is important, since the poorest populations are likely to have higher mortality rates as well as lower levels of intervention coverage, and this could influence estimates (53). Data limitations restricted our ability to develop this analysis, even as a test case. We have also not analyzed the costs involved in extending coverage, which we hope to develop in a future study. A Challenge to ASEAN Since its formation in 1967, ASEAN has positioned itself as an important hub for economic and sociocultural cooperation. Infectious diseases have thus far commanded much of ASEAN's attention in health matters. Recently, regional focus has begun to shift to other health issues. The ASEAN Strategic Framework on Health Development 2010-2015 (54), which focuses on access to health care services besides communicable diseases and pandemic preparedness, has also gained regional support. Given the economic vigour of ASEAN, regional cooperation in health promises to be the key for less developed members to break free from systems inertia and address the low-lying fruits of maternal and child health and nutrition. The pivotal role played by ASEAN in stimulating and channeling international financial aid to tsunami-devastated Indonesia in 2004 and cyclone-stricken Myanmar in 2008 testifies to the power of this promise. But how should this aid be used? ASEAN experience, as distilled in this paper, suggests that effective interventions to curb maternal and child mortality must be deployed to actively seek out the disadvantaged populations who are most ravaged by unsafe abortion, hypertensive diseases, postpartum hemorrhage, pneumonia, sepsis and birth asphyxia. Far from expecting coverage of these programs to passively diffuse to the very poor, governments must innovatively combine health interventions with non-health programs such micro-finance schemes and conditional cash transfer mechanisms that have proven successful in other settings (55, 56). Achievement of the MDGs at the global level will not happen without individual country efforts. As the donor community focuses its attention on the burdens of Africa and South Asia, ASEAN countries must support each other.
Comment [RF22]: Discussion of ASEANs potential role revised and clarified per reviewer 3

Comment [RF21]: Discussion of limitations added per reviewers 3 and 4

References: 1. WHO. World Health Statistics 2010. Geneva: World Health Organization. 2. WHO. Trends in maternal mortality: 1990-2008.Geneva: World Health Organization, 2010. 3. UNICEF. Levels and trends in child mortality, Report 2010. Estimates developed by the UN Inter-agency Group for Child Mortality Estimation. 4. Countdown 2015. 2010 Country profiles for Cambodia, Indonesia, Lao PDR, Myanmar, and Philippines. http://www.countdown2015mnch.org/reports-publications/2010-country-profiles (accessed May 12, 2010) 5. Campbell OMR and Graham WJ, on behalf of The Lancet Maternal Survival Series steering group. Strategies for reducing maternal mortality: getting on with what works The Lancet - 7 October 2006 (Vol. 368, Issue 9543, Pages 1284 - 1299) 6. Darmstadt GL, Bhutta ZA , Cousens S, Adam T, Walker N, de Bernis L, for the Lancet Neonatal Survival Steering Team . Evidence-based, cost-effective interventions: how many newborn babies can we save? The Lancet - 12 March 2005 (Vol. 365, Issue 9463, Pages 977 - 988) 7. Kerber KJ, de Graft-Johnson JE, Bhutta ZA , Okong P, Starrs A ,LawnJE. Continuum of care for maternal, newborn, and child health: from slogan to service delivery. The Lancet - 13 October 2007 (Vol. 370, Issue 9595, Pages 1358 - 1369) 8. Bhutta ZA, Ali S, Cousens S, Ali TM, Haider BA, Rizvi A,Okong P, Bhutta SZ, Black RE. Interventions to address maternal, newborn, and child survival: what difference can integrated primary health care strategies make? The Lancet - 13 September 2008 (Vol. 372, Issue 9642, Pages 972 - 989) 9. Hogan M, Foreman K, Naghavi M, Ahn S, Wang M, Makela S, et al. Maternal mortality for 181 countries, 1980-2008: a systematic analysis of progress towards Millennium development Goal 5. Lancet. 2010; 375(9726):1609-1623. 10. Rajaratnam J, Marcus J, Flaxman A, Wang H, Levin-Rector A, Dwyer L, Costa M, Lopez A, Murray C. Neonatal, postnatal, childhood, and under-5-mortality for 187 countries, 1970-2010: A systematic analysis of progress towards Millennium Development Goal 4. The Lancet 2010(375): 1988-2007. 11. Black R, Cousens S, Johnson H, Lawn J, Rudan I, Bassani D, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet 2010; 375(9730):19691987 12. Khan KS, Wojdyla D, Say L, Glmezoglu AM, Van Look PFA. WHO analysis of causes of maternal death: a systematic review. The Lancet 2006; 367: 106674 13. UNICEF. State of the World's Children Reports, 1996-2010. New York: United Nations Children's Fund 14. Measure DHS. Demographic and Health Survey data for Cambodia 2000 and 2005, Indonesia 1987, 1991, 1994, 2002-03 and 2007, Philippines 1993, 1998, 2003 and 2008, and Vietnam 1997 and 2002. http://www.measuredhs.com (accessed Apr 27, 2010) 15. UNICEF. Multiple Indicator Cluster Surveys. http://www.childinfo.org/mics.html (accessed Jul 4,2010) 16. National Statistics Office. Monitoring the Situation of Children and Women Thailand Multiple Indicator Cluster Survey December 2005 February 2006 Final Report. http://www.unicef.org/thailand/MICS_Final_Report2.pdf (Accessed September 20, 2010) 17. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. SocioEconomic Differences in Health, Nutrition, and Population in Cambodia (Washington, D.C.: The World Bank, 2007). 18. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. SocioEconomic Differences in Health, Nutrition, and Population in Indonesia (Washington, D.C.: The World Bank, 2007). 19. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. Socio-

10

Economic Differences in Health, Nutrition, and Population in the Philippines (Washington, D.C.: The World Bank, 2007). 20. Gwatkin D, Rutstein S, Johnson K, E, Suliman E, A, Wagstaff A, and Amouzou A. SocioEconomic Differences in Health, Nutrition, and Population in Vietnam (Washington, D.C.: The World Bank, 2007). 21. Plosky WD, Stover J, and Winfrey B. The Lives Saved Tool , A Computer Program for Making Child and Maternal Survival Projections. December 2009. 22. Jones G, Steketee RW, Black RE, Bhutta ZA, Morris SS. The Bellagio Child Survival Study Group. How many child deaths can we prevent this year? Lancet 2003; 362: 65-71. 23. Bhutta ZA, Ahmad T, Black RE, et al for the Maternal and Child Undernutrition Study Group. What works? Interventions for maternal and child undernutrition and survival. Lancet 2008; 371: 417-440. 24. Boschi-Pinto C, Young M, Black RE. The child health epidemiology reference group reviews of the effectiveness of interventions to reduce maternal, neonatal and child mortality. Int. J. Epidemiol. (2010) 39(suppl 1): i3-i6. 25. Chopra M, Daviaud E, Pattinson B, Fonn S, Lawn JE. Saving the lives of South Africa's mothers, babies, and children: can the health system deliver? Lancet 2009; 374: 835-846 26. Friberg IK, Kinney MV, Lawn JE, Kerber KJ, Odubanjo MO, et al. (2010) Sub-Saharan Africa's Mothers, Newborns, and Children: How Many Lives Could Be Saved with Targeted Health Interventions? PLoS Med 7(6): e1000295. doi:10.1371/journal.pmed.1000295 27. World Bank. World Bank database. http://siteresources.worldbank.org/DATASTATISTICS/ Resources/GNIPC.pdf (Accessed 19 Sept 2010 ) 28. Virasakdi C, KH Phua, MT Yap, NS Pocock, J Hashim, R Chhem, SA Wilopo and AD Lopez, Health in Southeast Asia: Diversity and Transitions, Lancet (overview paper, accepted for publication) 29. Lawn JE, Cousens S, Zupan J, and Lancet Neonatal Survival Steering Team. 4 million neonatal deaths: When? Where? Why? Lancet. 2005 Mar 5-11;365(9462):891-900. 30. Coker RJ, Hunter B, Rudge J, Liverani M, Hanvoravongchai P. The emergence of and response to infectious diseases in South East Asia. Lancet 2011 31. Rohde J, Cousens S, Chopra M, Tangcharoensathien V, Black R, Bhutta ZA, Lawn JE. 30 years after Alma-Ata: has primary health care worked in countries? Lancet 2008:372:950-61. 32. Countdown 2008 Equity Analysis Group, Boerma JT, Bryce J, Kinfu Y, Axelson H, Victora CG. Mind the gap: equity and trends in coverage of maternal, newborn and child health services in 54 Countdown countries. Lancet. 2008 Apr 12;371(9620):1259-67. 33. Limwattananon S, Tangcharoensathien V, Prakongsai P. Equity in maternal and child health in Thailand. Bull World Health Organ. 2010 Jun;88(6):420-7. Epub 2009 Dec 8. 34. Vapattanawong P, Hogan MC, Hanvoravongchai P, Gakidou E, Vos T, Lopez AD, Lim SS. Reductions in child mortality levels and inequalities in Thailand: analysis of two censuses. Lancet. 2007 Mar 10;369(9564):850-5 35. Wibulpolprasert S. Community financing: Thailand experience. Health Policy Plan 1991; 6: 354-360. 36. Van Lerberghe W, De Brouwere V. Of blind alleys and things that have worked: historys lessons on reducing maternal mortality. Studies Health Serve Organ Policy 2001; 17: 733. 37. Bureau of Policy and Strategy, Ministry of Public health. Health Policy in Thailand. Nonthaburi: Ministry of Public Health, 2009. 38. Wibulpolprasert S, editor. Thailand Health Profile 2005-2007. Bangkok: The War Veterans Organization of Thailand Printing Press; 2007. 39. Pramualratana P, Wibulpolprasert S. Health insurance systems in Thailand. Nonthaburi: Health System Research Institute, 2002. 40. WHO SEARO. Improving Maternal, Newborn and Child Health in the South-East Asia Region. Data source: Basic Indicators: Health Situation in South-East Asia, World Health Organization, South-East Asia Region, 2004.

11

41. Azwar A. Evolution Of Safe-Motherhood Policies In Indonesia. Presented in: IMMPACT Country Co-ordinating Group & Technical Partners Meeting, Aberdeen, January 2004. 42. Ministry of Health, Indonesia. National Health Surveys 1980 - 2008. 43. Gakidou E, Cowling K, Lozano R, Murray CJL. Increased educational attainment and its effects on child mortality in 175 countries between 1970 and 2009: a systematic analysis, Lancet 2010; 376: 95974. 44. UN Country Team Malaysia. Malaysia, Successes and challenges, Kuala Lumpur: UNDP, 2005. 45. Health Policy and Strategy Institute. Measuring Health Systems Performance in Vietnam: Results from Eight Provincial Health Systems Assessments. http://www.healthsystems2020.org/content/resource/detail/2515/ (accessed September 25, 2010) 46. UNESCAP. Development of Health Systems in the Context of Enhancing Economic Growth towards Achieving the Millennium Development Goals in Asia and the Pacific. Bangkok: ESCAP; 2007. 47. Jordan S, Lim L, Seubsman SA, Bain C, Sleigh A; the Thai Cohort Study Team. Secular changes and predictors of adult height for 86 105 male and female members of the Thai Cohort Study born between 1940 and 1990. J Epidemiol Community Health. 2010 Aug 30. [Epub ahead of print] 48. ASEAN Charter of the Association of Southeast Asian Nations. http://www.aseansec.org/21069.pdf (accessed September 22, 2010). 49. Institute for International Programs. LiST: The Lives Saved Tool An evidence-based tool for estimating intervention impact. http://www.jhsph.edu/dept/ih/IIP/list/index.html (accessed October 30, 2010 ) 50. Stover J, McKinnon R, Winfrey B. Spectrum: A model platform for linking impact of maternal and child survival intervention with AIDS, family planning and demographic projections. Int. J. Epidemiol. (2010) 39(suppl 1): i7-i10. 51. Walker N, Fischer-Walker C, Bryce J, Bahl R, Cousens S and writing for the CHERG Review Groups on Intervention Effects. Standards for CHERG reviews of intervention effects on child survival. Int. J. Epidemiol. (2010) 39(suppl 1): i21-i31 doi:10.1093/ije/dyq036. 52. Abu Bakar S, Mathews A, Jegasothy R, Ali R, Kandiah N. Strategy for reducing maternal mortality Bulletin of the WHO 1999; 77: 190-193. 53. Victora CG. Commentary: LiST: using epidemiology to guide child survival policymaking and programming. Int. J. Epidemiol. (2010) 39(suppl 1): i1-i2 doi:10.1093/ije/dyq044. 54. Association of Southeast Asian Nations. http://www.aseansec.org/24938.htm (accessed October 30, 2010). 55. Health Microinsurance: A Comparative Study of Three Examples in Bangladesh, by Mosieh Ahmed, Syed Khairul Islam, Md. Abul Quashem and Nabil Ahmed, CGAP working Group on Microinsurance, Case Study No. 13: September 2005. 56. Fiszbein A, Schady N. et al. Conditional Cash Transfers: Reducing Present and Future Poverty. (Washington, D.C.: The World Bank, 2009). 57. Estimation Methods Used by the UN Inter-agency Group for Child Mortality Estimation. http://www.childmortality.org/stock/documents/Methods%20for%20Estimating%20Child%20Mortality_2010.pd f (accessed October 30, 2010). 58. Wilmoth J, Zureick S, Mizoguchi N, Inoue M, Oestergaard M. Levels and Trends of Maternal Mortality in the World: The Development of New Estimates by the United Nations. http://www.who.int/reproductivehealth/publications/monitoring/MMR_technical_report.pdf (accessed October 30, 2010). 59. Caulfield LE, de Onis M, Blossner M, Black RE. Undernutrition as an underlying cause of child deaths associated with diarrhea, pneumonia, malaria and measles. Am. J. Clin. Nutr. 2004. 80:193-8. 60. Tangcharoensathien V, Patcharanarumol W, Ir P, Aljunid SM, Ghufron Mukti A, Akkhavong K, Banzon E, Huong DB, Thabrany H, Mills A, Health Financing Reforms in South East Asia: challenges in achieving universal coverage. Lancet 2010

12

Table 1. Percentage of deaths averted with 99% coverage of selected interventions for ASEAN region and sub-groups as defined by mortality reduction patterns

Comment [RF23]: Previously table 3; absolute estimates of deaths removed to enhance readability

Maternal deaths averted (%) Interventions Basic emergency obstetric care (clinic) Comprehensive emergency obstetric care Post abortion case management, basic level Post abortion case management, comprehensive level Basic emergency obstetric care (clinic) Comprehensive emergency obstetric care Antenatal corticosteroids for preterm labor Kangaroo mother care Use of water connection in the home Pneumococcal vaccine Pneumonia case management (oral antibiotics) Zinc for diarrhea treatment Sub-group 1 * 1.6 9.2 4.4 Sub-group 2 19.4 55.1 6.1 10.3 Sub-group 3 21.5 55.6 6.1 10.3 ASEAN region 19.0 52.9 5.8 6.5

Neonatal deaths averted (%) * 0.1 7.3 20.3 3.1 2.9 5.7 1.2 11.7 22.8 20.1 20.6 13.4 9.8 19.0 5.0 17.2 31.1 17.2 16.8 11.3 7.6 15.9 3.9 12.8 24.1 18.5 19.4 12.3 8.7 17.4 4.4

Child (including postneonatal) deaths averted (%)

13

Fig 1a Trends in maternal mortality in Southeast Asia, 1990-2008 (data from WHO Maternal Mortality September 2010)

Comment [RF24]: Brunei and Singapore removed to enhance readability, y-axis adjusted

14

Fig. 1b Trends in infant mortality in Southeast Asia, 1990-2008 (Data from Unicef Child Mortality September 2010)

Comment [RF25]: Brunei and Singapore removed to enhance readability, y-axis adjusted

15

Fig. 1c Trends in under-five mortality in Southeast Asia, 1990-2008 (Data from Unicef Child Mortality September 2010)

Comment [RF26]: Brunei and Singapore removed to enhance readability, y-axis adjusted

16

Fig 2 Neonatal and Postneonatal Mortality Rate Reductions between 1990 and 2008 (Data from IHME, Rajaratnam et al, 2007)

17

Fig 3a. Causes of maternal deaths in Southeast Asia (Data from UN MDG Southeast Asia, 2010, includes 10 ASEAN countries + Timor Leste)

Fig 3b Causes of child deaths in Southeast Asia (Data from Black et al, 2010)

Comment [RF27]: Causes of neonatal deaths added

18

Fig 4. Inequities in MNCH intervention coverage in selected SEA countries Data from Gwatkin et al, 2007; National Statistics Office, www.childinfo.org)

19

20

21

Fig. 5. Trends in maternal mortality and MNCH programs in Thailand, by GDP 1960 - 2008

(Data from Thailand Ministry of Health)

22

Fig. 6. Trends in maternal mortality and safe motherhood programs in Indonesia, by GDP 1960 - 2008

Comment [RF28]: Text of blue lines edited to enhance readability

(Data from Indonesian Ministry of Health)

Fig. 7. Neonatal Mortality reductions in Thailand (red) and Indonesia (blue) by GDP 1960-2008 (Data from Thailand and Indonesian Ministries of Health)

Comment [RF29]: New figure added

23

Fig 8a. Maternal deaths averted at 60, 90 and 99% coverage of interventions for the ASEAN region (red, orange and purple bars) and regional sub-groups as defined by mortality reduction patterns (blue, green and yellow bars)

Comment [RF30]: Title of this and figs 8b and 6 updated to clarify derivation of subgroups

Fig 8b. Neonatal deaths averted at 60, 90 and 99% coverage of interventions for the ASEAN region (red, orange and purple bars) and regional sub-groups as defined by mortality reduction patterns (blue, green and yellow bars)

24

Fig 8 c. Child deaths averted at 60, 90 and 99% coverage of interventions for the ASEAN region (red, orange and purple bars) and regional sub-groups as defined by mortality reduction patterns (blue, green and yellow bars)

Comment [RF31]: New figure added to show LiST results for causes of child deaths

25

Web Table 1 Data sources

Data Source Country Coverage Indicators Estimation methods Direct sisterhood method for maternal mortality estimates; direct method for estimation of child mortality rates; estimates of intervention coverage are weighted proportions Indirect estimation of infant and underfive mortality rates; estimates of intervention coverage are weighted proportions Limitations Only 4/10 countries covered, time periods per country variable; direct sisterhood method may underestimate maternal deaths; birth transference may underestimate infant mortality

Comment [RF32]: Now a webappendix; additional information on estimation methods and data limitations added to table, comments from reviewers 3 and 4

DHS

Cambodia, Indonesia, Philippines, Vietnam

Time trends for neonatal, infant, under-five mortality, maternal mortality; intervention coverage (vaccination, IMCI, skilled birth attendance, delivery care)

MICS

Laos, Myanmar, Thailand, Vietnam

Estimation of infant, under-five mortality rates; intervention coverage (vaccination, IMCI, skilled birth attendance, delivery care)

Inter-agency Group for Child Mortality Estimation (UNICEF and WHO)

All 10 countries

Standardized estimation of time trends in infant, under-five mortality rates, 1990-2009 for 196 countries

Linear spline regression for countries without high HIV/AIDS prevalence, Loess regression for countries with high HIV/AIDS prevalence; weights applied for country data sources Gaussian process regression of under-five mortality accounting for nonsampling error; infant and neonatal mortality modeled from under-five mortality using multilevel regression Direct estimation from civil registration sources; multilevel regression model for other types of data sources

IHME: Rajaratnam et al. Maternal Mortality Estimation Inter-Agency Group (WHO)

All 10 countries

Standardized estimation of time trends in neonatal, infant, underfive mortality rates for 187 countries from 1970-2009 countries

All 10 countries

Standardized estimation of time trends in maternal mortality ration 1990-2008 for 172 countries

Earlier rounds of MICS subject to Only 3/10 countries covered Do not address potential mortality shocks other than HIV/AIDS epidemic; Loess forecasts of mortality decline may be conservative; vital registration available for 5/10 Southeast Asia countries Yields lower estimates of mortality than other methods; similar modeling approach for countries with and without high HIV prevalence; vital registration available for 5/10 Southeast Asian countries Underreporting/m isclassification of maternal deaths in household survey sources; did not include subnational data sources

26

IHME: Hogan et al.

All 10 countries

Standardized estimation of time trends in maternal mortality ratio for 181 countries, 1980-2008

CHERG:

All 10 countries

Cause of death estimation for neonatal, infant, under-fives

Two-step spatiotemporal regression Direct estimation with ICD-10 codes for complete vital registration data; for countries without complete vital registration, separate multicause multinomial regression models for countries with low and high child mortality using verbal autopsy sources; separation estimation of neonatal tetanus, malaria, measles, pertussis, HIV/AIDS

Underreporting/m isclassification of maternal deaths in household survey sources; idid not extract incidental pregnancyrelated deaths; mpact of HIV epidemic modeled directly; vital registration available for 5/10 Southeast Asian countries; Myanmar and Vietnam no national data available

Highest mortality countries of Southeast Asia do not have vital registration, estimates are modeled

27

Web Table 2 Effect sizes LiST References: Institute for International Programs. LiST: The Lives Saved Tool An evidence-based tool for estimating
intervention impact. http://www.jhsph.edu/dept/ih/IIP/list/index.html; Plosky WD, Stover J, and Winfrey B. The Lives Saved Tool , A Computer Program for Making Child and Maternal Survival Projections. December 2009.

Comment [RF33]: New table showing interevention effectiveness estimates used in LiST modeling

28

29

30

Web Fig. 1 Mortality Reductions with GNI per capita Mortality data from UN MDG Reports September 2010; GNI per capita from World Bank Fig 1a Scatterplot of maternal mortality against GNI per capita in 1990 and 2008

Comment [RF34]: Now a webappendix; y-axis adjusted to

31

Fig 1b Scatterplot of infant mortality against GNI per capita in 1990 and 2008

Comment [RF35]: New addition to webappendix

32

Fig 1c Scatterplot of under-five mortality against GNI per capita in 1990 and 2008

Comment [RF36]: New addition to webappendix

33