## Abstract

Introduction The burden of disease resulting from neonatal conditions is substantial in developing countries. From 2003 to 2005, the Projahnmo I programme delivered community-based interventions for maternal and newborn health in Sylhet, Bangladesh. This analysis quantifies burden of disability and incorporates non-fatal outcomes into cost-effectiveness analysis of interventions delivered in the Projahnmo I programme.

Methods A decision tree model was created to predict disability resulting from preterm birth, neonatal meningitis and intrapartum-related hypoxia (‘birth asphyxia’). Outcomes were defined as the years lost to disability (YLD) component of disability-adjusted life years (DALYs). Calculations were based on data from the Projahnmo I trial, supplemented with values from published literature and expert opinion where data were absent.

Results 195 YLD per 1000 neonates [95% confidence interval (CI): 157–241] were predicted in the main calculation, sensitive to different DALY assumptions, disability weights and alternative model structures. The Projahnmo I home care intervention may have averted 2.0 (1.3–2.8) YLD per 1000 neonates. Compared with calculations based on reductions in mortality alone, the cost-effectiveness ratio decreased by only 0.6% from $105.23 to$104.62 ($65.15–$266.60) when YLD were included, with 0.6% more DALYs averted [total 338/1000 (95% CI: 131–542)].

Discussion A significant burden of disability results from neonatal conditions in Sylhet, Bangladesh. Adding YLD has very little impact on recommendations based on cost-effectiveness, even at the margin of programme adoption. This model provides guidance for collecting data on disabilities in new settings.

KEY MESSAGES

• Based on 2005 programme data, parameterizations from literature and expert opinion; home-based neonatal care saves newborn lives with a small but significant decrease in population morbidity. The intervention is cost-effective regardless of whether morbidity is included in the calculation.

• Projahnmo I saved 12 neonatal lives per 1000 based on mortality statistics alone, and an additional 2 years lost from disability (YLD) when morbidity is considered.

• Including disability represented by YLD in neonatal burden of disease estimates increased disability-adjusted life year (DALY) estimates by an additional 14% over estimates based only on mortality. Our analysis found a burden of morbidity equivalent to approximately seven deaths per 1000 neonates.

• Preterm birth contributed more than meningitis and birth asphyxia to the burden of disability; the proportional burden of disability is preterm (73%), ‘birth asphyxia’ (24%) and meningitis (3%).

## Introduction

The World Health Organization (WHO) estimated that perinatal conditions were the 6th leading cause of disability in the developing world in 2001 (Mathers et al. 2006a). About 3.575 million deaths worldwide occur annually during the neonatal period (Black et al. 2010); however, evidence is mixed on whether life-saving interventions increase population morbidity (Durkin et al. 2006; Winter et al. 2002; Boyle et al. 2011), or if morbidity patterns will remain constant or improve with time (Platt et al. 2007; D’Amore et al. 2011; Jang et al. 2011; van Haastert et al. 2011). Quantifying summary measures for population health for disabling conditions arising in the neonatal period is a key information gap in international public health (Azra Haider and Bhutta 2006; Maulik and Darmstadt 2007), and more evidence is needed about their impact on the cost-effectiveness of community-based interventions (Lawn et al. 2006; Maulik and Darmstadt 2009).

Three key categories of neonatal conditions that can cause disability include preterm birth, infections and intrapartum-related hypoxia (‘birth asphyxia’). Congenital anomalies, certainly associated with disability, have not yet been figured into our estimates due to methodological uncertainties given available data. Thirteen million or 9.6% of all births in low- and middle-income countries (LMICs) were preterm in 2005 (Beck et al. 2010). Data from four high-mortality countries showed that 71/1000 live births had culture-confirmed infection; however, the proportion of neonates with community-acquired infection that were brought to health centres may have been affected by selection bias (WHO Young Infants Study Group 1999). Clinical diagnosis suggests that 56–105/1000 newborns in South Asia have suspected serious infection depending on the algorithm used (Bang et al. 2005a; Baqui et al. 2009a). A review of studies from LMICs reported neonatal infection incidence rates of 0.8–6.1/1000 for meningitis, 30–156/1000 for diarrhoea, 142/1000 for severe pneumonia and 201/1000 for acute respiratory infection (Thaver and Zaidi 2009). Statistics on the incidence of viral infections are needed. Ten million neonates do not breathe immediately at birth, of which 6 million require resuscitation (Wall et al. 2009), and 1–1.2 million infants develop severe sequelae from ‘birth asphyxia’ (Saving Newborn Lives 2001; WHO 2005). In South Asia, 14–20% of newborns in community settings do not spontaneously initiate breathing (Bang et al. 2005b; Lee et al. 2011a).

Perinatal conditions lead to a variety of long-term neurological sequelae (Table 1a). A community-based test of the Ten Questions screening survey among 2- to 9-year-old children in Bangladesh found that 69/1000 had mild or serious disability in cognition, hearing, motor skills, vision, or seizures (Khan and Durkin 1995). Recent evidence (2005–06) from the Multiple Indicator Cluster Survey (MICS), a cross-sectional survey using the Ten Question Screen, estimated an incidence of disability in children 2–9 years old of 21% in Bangladesh (Gottlieb et al. 2009), although screening tools tend to overestimate compared with proper diagnostic instruments. As neonatal survival continues to improve in Bangladesh, the prevalence of learning and developmental disability may also be affected (Durkin et al. 2006).

Table 1a

Disability weights: Sequelae and disability weights from the Australian Burden of Disease (ABD) study

Sequelae Disability weight Condition Notes
Motor deficit 0.170 a, c ABD study
Visual deficit 0.170 a, c ABD study: Dutch weight for moderate vision loss
Speech delay 0.110 Mild Permanent Disability (low birth weight) in ABD study
Seizures 0.110 a, b, c ABD study
Behaviour 0.110 a, c* Mild Permanent Disability (low birth weight) in ABD study
Hearing loss 0.230 Deafness (asphyxia) in ABD study
Hearing loss 0.370 a, c Severe hearing loss (low birth weight) and deafness (infection) in ABD study
Cerebral palsy without intellectual deficit 0.170 ABD study
Mild intellectual disability 0.290 a, b, c ABD study
Moderate intellectual disability 0.430 a, b, c ABD study
Severe intellectual disability 0.820 a, b, c ABD study
Profound intellectual disability 0.760 a, b, c ABD study
Sequelae Disability weight Condition Notes
Motor deficit 0.170 a, c ABD study
Visual deficit 0.170 a, c ABD study: Dutch weight for moderate vision loss
Speech delay 0.110 Mild Permanent Disability (low birth weight) in ABD study
Seizures 0.110 a, b, c ABD study
Behaviour 0.110 a, c* Mild Permanent Disability (low birth weight) in ABD study
Hearing loss 0.230 Deafness (asphyxia) in ABD study
Hearing loss 0.370 a, c Severe hearing loss (low birth weight) and deafness (infection) in ABD study
Cerebral palsy without intellectual deficit 0.170 ABD study
Mild intellectual disability 0.290 a, b, c ABD study
Moderate intellectual disability 0.430 a, b, c ABD study
Severe intellectual disability 0.820 a, b, c ABD study
Profound intellectual disability 0.760 a, b, c ABD study

Notes: a = preterm birth; b = intrapartum-related neonatal encephalopathy; c = meningitis. *Not included in ABD for infections, but behavioral problems were included in Edmond et al. 2010.

These figures are necessarily rough estimates. More than half of births are unattended in high mortality settings and between 80% to >90% of births take place at home in rural Bangladesh. Most do not receive postnatal care, restricting capacity to diagnose conditions promptly after birth (Darmstadt et al. 2009a). In South Asia, 74% of newborn infants are not weighed (WHO 2004), and a smaller proportion is of known gestational age. Only clinical diagnosis is available for infections in community-based settings, and incidence estimates are substantially higher than those based on laboratory-confirmed diagnosis (Darmstadt et al. 2009b). Consensus has been difficult on definitions of ‘birth asphyxia’ (Nelson and Leviton 1991; Lawn et al. 2007; Lee et al. 2008; Halloran et al. 2009), as highlighted in a recent article to clarify terminology for intrapartum-related outcomes (Lawn et al. 2009). Data on long-term morbidity and impairment from LMICs is scarce and inconsistent.

Disability-adjusted life years (DALYs) are currently the standard summary measure of population health used in LMICs (Murray and Lopez 1996), and are commonly used in burden of disease and economic evaluation studies in international health. DALYs are composed of years of life lost (YLLs) due to premature mortality and years lost to disability (YLD) due to disease states of less-than-perfect health. To date, cost-effectiveness analyses of neonatal health interventions in LMICs have only included YLLs for neonates (Borghi et al. 2005; Darmstadt et al. 2005; Lefevre et al. 2010). The only mention of neonatal morbidity in the second edition of the Disease Control Priorities Project (DCP2) indicates that data are scarce (Lawn et al. 2006), and it is not clear how YLDs were calculated in WHO-CHOICE estimates (Adam et al. 2005).

Data are becoming available to facilitate calculation of YLD to enhance the comparability of neonatal cost-effectiveness analyses to the spectrum of health care interventions relevant to LMICs. Long-term outcome studies are being published from Dhaka Shishu hospital (Khan et al. 2006; Saha et al. 2009), Global Burden of Disease (GBD) reviews have been conducted, and pooled estimates for input parameters are becoming available (O’Brien et al. 2009; Watt et al. 2009; Edmond et al. 2010). This article presents a model framework for calculating YLD resulting from neonatal conditions that can be easily applied to cost-effectiveness analysis of relevant interventions, with an example using data from the Projahnmo I trial from rural Sylhet, Bangladesh. Outputs are presented as YLD associated with a modelled comparison arm, YLD averted due to intervention, DALYs averted and incremental cost-effectiveness.

## Methods

### Analytical overview

Burden of YLD was modelled using a tree structure, with differences in YLD expected between study arms calculated in a decision tree framework (Figure 1). Costs and YLL were factored in to calculate incremental cost-effectiveness. Relevant neonatal conditions and disabling sequelae were included; ‘birth asphyxia’ values from the Australian Burden of Disease (ABD) study (Mathers et al. 2001), meningitis values from meta-analysis (Edmond et al. 2010), and preterm values from a relevant Bangladeshi study (Khan et al. 2006). Low birth-weight (LBW) was replaced by preterm birth since preterm is a more specific predictor of disability and is consistent with current clinical paradigms and cause-of-death classification (Lawn et al. 2010).

Figure 1

Component trees describing the full model. Probabilities correspond to those for male neonates (female probabilities are similar). Terminal nodes are followed by disability weights / male life span / female life span (A) Root tree. (B) Preterm sequelae. In the reference case, these ten branches emanate from Nodes 1 & 2 (Khan et al. 2006). In sensitivity analysis, weights for 15% treated (0.256) and 85% untreated (0.291) survivors are tested. Further sensitivity analysis tests a simple disability weight (0.110). (C) IPR-NE sequelae. In the reference case, the outcomes structure below emanates from Node 3 of the root tree. In sensitivity analysis, a simple disability weight emanates (0.374). (D) Meningitis sequelae emanating from Nodes 4–7 derived from (Edmond et al. 2010). Probabilities reflect HiB / Pneumococcus / Meningococcus / Other infection. Deaths incur years of life lost (YLLs) but no years lost to disability (YLD). The tree portion of the cost-effectiveness model calculates only YLD, with YLLs added at the final calculation.

Figure 1

Component trees describing the full model. Probabilities correspond to those for male neonates (female probabilities are similar). Terminal nodes are followed by disability weights / male life span / female life span (A) Root tree. (B) Preterm sequelae. In the reference case, these ten branches emanate from Nodes 1 & 2 (Khan et al. 2006). In sensitivity analysis, weights for 15% treated (0.256) and 85% untreated (0.291) survivors are tested. Further sensitivity analysis tests a simple disability weight (0.110). (C) IPR-NE sequelae. In the reference case, the outcomes structure below emanates from Node 3 of the root tree. In sensitivity analysis, a simple disability weight emanates (0.374). (D) Meningitis sequelae emanating from Nodes 4–7 derived from (Edmond et al. 2010). Probabilities reflect HiB / Pneumococcus / Meningococcus / Other infection. Deaths incur years of life lost (YLLs) but no years lost to disability (YLD). The tree portion of the cost-effectiveness model calculates only YLD, with YLLs added at the final calculation.

Data were checked for consistency and quality, and results were generated according to probabilities of different outcomes emanating from different health states according to nodes and branches within the tree framework (Figure 1). To the extent possible, the intervention arm was parameterized using primary data from the home care arm of the Projahnmo I trial, using year 2005 to represent the programme at maturity (Table 2a) (Baqui et al. 2008; Baqui et al. 2009a). In the absence of data from the comparison arm of the trial (community health workers did not make home visits in that arm), the comparator arm was modelled assuming that all mortality reduction was due to averted infections. Variation around incremental survival and sequelae from infections was assumed to be 25%, similar to what is often observed in meta-analysis of very large trials (Table 2b). Information gaps were modelled with data from published literature.

### Study context

Intervention components for the Projahnmo I home care arm are described in a previous paper (Baqui et al. 2008). Perinatal prevention and treatment interventions for newborn health were delivered through a partnership between a local non-governmental organization (Shimantik) and Johns Hopkins University in Sylhet, Bangladesh. An enhanced version of the WHO Essential Newborn Care training and package of interventions was provided, including pregnancy surveillance, birth preparedness, algorithm-based recognition of maternal and newborn illness by community health workers (CHWs) or families, referral and/or home treatment for sepsis, and continued monitoring and advice (if not severe). While resuscitation training was provided, bag and mask were not included as part of the intervention, and only 5% of deliveries were attended by CHWs. The study population included all births in Sylhet district, a setting with the highest neonatal mortality rate (48 deaths per 1000 live births at baseline survey) (Baqui et al. 2008) and the highest total fertility rate (4.2) in the country at the start of the study (NIPORT et al. 2005). The district had poor access to health services and low utilization of skilled attendants at birth. To optimize the impact of the intervention, community meetings and household visits by CHWs were introduced to provide behaviour change communications, demonstrations to promote newborn care preparedness, and detection and management of serious illnesses. Attempts to measure morbidity were not undertaken in order to minimize the presence of CHWs in the community and their potential for impact and confounding of study results.

### Disease categories and definitions

Preterm birth is defined as <37 weeks gestation. It is a component of LBW, defined as newborns weighing <2500 g who are either preterm and/or have intrauterine growth restriction. Very preterm was defined in this analysis as <33 weeks gestation, and late preterm as 33 to <37 weeks. Complications of preterm birth may comprise the greatest proportion of the direct morbidity and mortality of LBW infants, so disability weights for LBW were used as a proxy due to the current lack of disability weights for preterm. Current Child Health Epidemiology Reference Group (CHERG) and Global Burden of Disease (GBD) research is underway to estimate the burden of disease due to preterm. Studies for disability parameterization excluded those limited to cohorts of <28 weeks since these neonates are unlikely to survive in community-based settings in developing countries (Kuti and Owa 2003), as confirmed in our data. Incidence and classification of preterm were derived from Projahnmo I data, with outcomes defined according to Khan et al. (2006).

Suspected clinical infections in Projahnmo I were identified according to a simplified version of the Integrated Management of Childhood Illness (IMCI) algorithm based on selection of seven clinical signs (Young Infants Clinical Signs Study Group 2008). Relevant infections include neonatal sepsis, bacterial meningitis, pneumonia and viral infections; which are difficult to differentiate in community settings. All of these infections may result in chronic impairment depending on the pathogen, disease severity and its management; however, data were only available to consider meningitis (O’Brien et al. 2009; Watt et al. 2009; Edmond et al. 2010). Other conditions such as kernicterus were not included as jaundice outcomes research is only emerging (Gordon et al. 2005; Olusanya and Somefun 2009). Diagnoses in the Projahnmo I trial were categorized as very severe disease (presence of certain single critical clinical signs), possible very severe disease (presence of multiple signs individually indicative of less severe disease) and possible very severe disease with only a single sign (Baqui et al. 2009a). The algorithm used has since been validated (Darmstadt et al. 2011). Our definition of the proportion of suspected community-acquired infection included all three of these categories. The proportion of suspected infections that was meningitis was quantified across four high mortality countries (WHO Young Infants Study Group 1999), and survival and proportion and types of sequelae were defined according to a pooled estimate (O’Brien et al. 2009; Watt et al. 2009; Edmond et al. 2010).

Birth asphyxia was defined in Projahnmo I as not crying/breathing, gasping and breathing slowly (<30/minute after birth). However, perinatal respiratory depression does not accurately predict long-term disability outcome and there has been advocacy for highlighting intrapartum events epidemiologically using more precise terminology (Lawn et al. 2005). Intrapartum-related neonatal encephalopathy (IPR-NE) is a strong marker of those who will develop chronic disability and is used in this calculation (Lawn et al. 2009; Lawn et al. 2010; Lee et al. 2011a). Incidence and sequelae parameterizations are based on updated GBD modelling and a previously published survival estimate from this work (Lawn et al. 2009).

### Model structure

Potential disease outcomes were calculated through a tree framework (Figure 1) excluding neonates with congenital anomalies or no condition, and not considering co-morbidity. This conceptual framework followed the pattern of incidence, post-neonatal survival and presence of sequelae; following the precedent of illness, death and disability used by the original national burden of disease exercise (Ghana Health Assessment Project Team 1981). Classification of sequelae was done according to proportions of each outcome used in the ABD study and two articles (Khan et al. 2006; Edmond et al. 2010). Branches were divided by sex to account for minor differences in incidence and life expectancy. Methods for calculating probabilities at tree nodes with three branches ensured that all randomly generated sets summed to one (Sendi and Clemen 1999). A time horizon including the lifespan of the neonate was assumed. Costs from a societal perspective and YLL components were derived from previous calculations.

### YLD calculation

YLD were calculated based on the standard DALY component equation (WHO 2009). We used Bangladesh-specific life expectancies in the reference case scenario assuming constant disability from each condition for the lifetime of the neonate. Reductions in Bangladeshi life expectancy due to each sequela were set proportional to reductions used in the ABD, and preterm life expectancies were tested according to a scenario based on published assumptions (Profit et al. 2010). The discount rate was set at 3% according to the WHO standard for economic evaluation of health interventions in LMICs (Tan-Torres Edejer et al. 2003). We excluded age weighting from the reference case as several arguments suggest that this level of complexity is not appropriate or necessary (Mathers et al. 2006b), it makes little difference to results (Barendregt et al. 1996) and DCP2 practice is to exclude it (Musgrove and Fox-Rushby 2006). We used Microsoft Excel to structure the calculation with a Visual Basic macro to perform Monte Carlo Simulations (1000 iterations), and evaluated model outputs with STATA 12 to generate regression coefficients for a tornado diagram (generated with 30 000 iterations to increase stability of results).

### Disability weights

Disability weights from the ABD study based on Dutch Public Health Status and Forecast study values were used in the main calculation, with weights from the GBD study tested in sensitivity analysis (Murray and Lopez 1996). Dutch weights were derived using a person trade-off ranking of disease stages with interpolation of others, and are the same construct as GBD weights (Mathers et al. 2006a). GBD weights for LBW (applied to preterm in this analysis) and IPR-NE are listed in Table 1b with ranges of uncertainty (Mathers et al. 2006a) and different values for treated and untreated forms of LBW (proxy for preterm) (Shibuya and Murray 1998a). Derivation of Dutch weights has been described (Stouthard et al. 2000), and the high degree of correlation between the Dutch and GBD studies has been demonstrated (Mathers et al. 2001).

Table 1b

Disability weights: Global Burden of Disease (GBD) estimates

Category Disability weight Range Untreated Treated Source
Birth asphyxia and birth trauma
SEARD specific estimate 0.374a     (Mathers et al. 2004
GBD2 estimate (World) 0.372 0.343 0.379 0.381a 0.334 (Mathers et al. 2006a
Health Dimensions in Sex and Reproduction 0.471   0.4793 0.4213 (Shibuya and Murray 1998a
Low birth weightb
SEARD specific estimate 0.106a     (Mathers et al. 2004
GBD2 estimate (World) 0.106     (Mathers et al. 2006a
Health Dimensions in Sex and Reproduction    0.291a 0.256a (Shibuya and Murray 1998b
Category Disability weight Range Untreated Treated Source
Birth asphyxia and birth trauma
SEARD specific estimate 0.374a     (Mathers et al. 2004
GBD2 estimate (World) 0.372 0.343 0.379 0.381a 0.334 (Mathers et al. 2006a
Health Dimensions in Sex and Reproduction 0.471   0.4793 0.4213 (Shibuya and Murray 1998a
Low birth weightb
SEARD specific estimate 0.106a     (Mathers et al. 2004
GBD2 estimate (World) 0.106     (Mathers et al. 2006a
Health Dimensions in Sex and Reproduction    0.291a 0.256a (Shibuya and Murray 1998b

Notes: aIncluded in analysis. bWeights for preterm birth will replace low birth weight when released. SEARD = countries in the WHO South East Asian sub-region with high adult and child mortality.

### Proportional differences between arms

Estimates of the difference between model arms were based on a variety of sources in the absence of specific data on disabilities across study arms or trial phases in Projahnmo I (Table 2b). Consistent with an odds ratio of 0.51 pooled from four studies on the effectiveness of hand washing (Blencowe et al. 2011), and a decline in incidence of infections by 48.2% from 1995–96 to 1997–98 in Gadchiroli India (CHW home-based management of neonatal infections) (Bang et al. 2005c), 48.2% was used for incremental reduction in infection incidence. We assumed the overall mortality reduction from Projahnmo I to be mostly due to home-based neonatal sepsis recognition and antibiotic treatment; infection mortality was 28% lower in the intervention arm relative to the comparison arm in 2005 (Baqui et al. 2008). This figure was applied to model reduction in mortality due to infection. Incidence of IPR-NE was reduced by only 1% since only 5% of Projahnmo I births were attended and 20% of IPR-NE can be prevented in community settings (Darmstadt et al. 2009c; Lee et al. 2011b). Bag-and-mask training was not included in Projahnmo I, although immediate stimulation through wiping and drying and mouth-to-mouth resuscitation were included. With low birth attendance in the community, and 12% hospital delivery in the home care arm (Baqui et al. 2009b), we assumed any change in IPR-NE survival to be negligible. No reduction of preterm incidence was expected as preventive interventions for preterm were beyond the scope of Projahnmo I. Cause-specific mortality data from Projahnmo I suggest that the difference in preterm survival was not significant. Interventions to prevent and treat infections, preterm birth and intrapartum events are likely to reduce incidence and severity of long-term sequelae (IOM 2003; Anon 2009; Barros et al. 2010; Furyk et al. 2011), but longitudinal studies that coincide with our defined outcomes in the Bangladeshi context are not currently available.

Our study modifies ABD calculations by incorporating Bangladesh-specific parameter definitions (Table 2a), a tree structure and sensitivity analysis around each model probability. Beta distributions were used to represent parameter uncertainty, with upper and lower limits defined according to 95% confidence intervals (CIs), interquartile ranges or ranges derived from literature. Alpha and beta moments were derived from the mean and standard deviation of triangular distributions; defined according to best, low and high values from our data and literature search. Disability weights from the original GBD were tested in scenario analysis (Shibuya and Murray 1998a; Shibuya and Murray 1998b), and the fraction of preterm patients treated was assumed to be 15%, consistent with the WHO assumption for treatment of LBW infants (Figure 1) (Murray and Lopez 1996). The discount rate was tested at 0% and 6%, and age weighting was tested according to GBD precedent. Regional life expectancies were tested to be consistent with DCP2 evaluations. Global standard life expectancy was tested according to the West Life Table Level 26 (80 males, 82.5 females) to enhance equity and make results comparable with GBD 2000 estimates (Coale and Demeney 1983; Lopez et al. 2006). Sources for uncertainty around other parameters were drawn from those reviewed by the GBD project (Shibuya and Murray 1998a; Shibuya and Murray 1998b), PubMed and internet searches, and review of bibliographies (Table 2a and b). Costs were determined through primary data collection and are presented as 2010 US$, inflated with consumer price indices from the International Monetary Fund (IMF 2012). Table 2a Input parameters Parameter Most likely Low value High value α β Source Sex malea 51.04% 49.74% 52.34% 4522.60 4338.33 Projahnmo I Liveborn (male)a 94.96% 94.16% 95.76% 4294.14 227.98 Projahnmo I Liveborn (female)a 95.79% 95.04% 96.54% 4155.12 182.69 Projahnmo I PT incidence (male)a 18.04% 16.60% 19.47% 774.50 3519.59 Projahnmo I PT incidence (female)a 16.80% 15.38% 18.22% 697.98 3457.10 Projahnmo I LPT proportion (male)a 83.67% 80.42% 86.92% 647.33 126.35 Projahnmo I LPT proportion (female)a 83.45% 80.00% 86.89% 581.74 115.41 Projahnmo I LPT survive (male)a 95.18% 93.12% 97.24% 615.98 31.19 Projahnmo I LPT survive (female)a 94.10% 91.71% 96.49% 547.27 34.30 Projahnmo I LPT sequelaeb 5.00% 4.23% 10.00% 14.45 274.92 (Shibuya and Murray 1998b; Moster et al. 2008; Chyi et al. 2008; Huddy et al. 2001 VPT proportion (male)a 16.33% 13.08% 19.58% 126.35 647.33 Projahnmo I VPT proportion (female)a 16.55% 13.11% 20.00% 115.41 581.74 Projahnmo I VPT survive (male)a 77.78% 68.72% 86.83% 97.62 27.89 Projahnmo I VPT survive (female)a 86.49% 78.70% 94.28% 99.09 15.48 Projahnmo I VPT sequelaeb 68.00% 58.00% 78.00% 89.59 42.16 (Khan et al. 2006 Suspected infection incidence (male)a 15.89% 14.52% 17.26% 682.37 3611.72 Projahnmo I Suspected infection incidence (female)a 13.83% 12.52% 15.14% 574.62 3580.46 Projahnmo I Percent meningitisa 3.42% 2.70% 4.14% 131.16 3699.06 (WHO Young Infants Study Group 1999 Percent Haemophilus influenzae type bc 31.10% 20.00% 72.00% 5.02 11.12 (Edmond et al. 2010 Percent Streptococcus pneumoniaec 23.60% 14.10% 40.20% 14.39 46.60 (Edmond et al. 2010 Percent Neisseria meningitidisc 9.40% 5.10% 32.10% 2.19 21.09 (Edmond et al. 2010 Percent other infectionc 13.40% 8.20% 28.10% 8.62 55.74 (Edmond et al. 2010 Haemophilus type b survivec 56.00% 38.00% 83.00% 15.58 12.24 (Watt et al. 2009 Streptococcus pneumoniae survivec 43.00% 29.00% 56.00% 34.25 45.40 (O’Brien et al. 2009 Neisseria meningitidis survivec 56.00% 38.00% 83.00% 15.58 12.24 Assumption based on (Watt et al. 2009 Other infection survivec 56.00% 38.00% 83.00% 15.58 12.24 Assumption based on (Watt et al. 2009 Haemophilus influenzae type b sequelaec 14.50% 0.01% 27.10% 5.72 33.75 (Edmond et al. 2010 Streptococcus pneumoniae sequelaec 34.70% 28.20% 45.10% 64.57 121.51 (Edmond et al. 2010 Neisseria meningitidis sequelaec 9.50% 5.10% 15.10% 19.41 184.94 (Edmond et al. 2010 Other infection sequelaec 19.90% 12.10% 35.20% 13.58 54.68 (Edmond et al. 2010 IPR-NE incidence overallb 1.34% 0.40% 2.70% 7.94 584.26 (Lawn et al. 2009), GBD estimate IPR-NE survive overallb 69.00% 67.00% 80.00% 180.03 80.88 (Lawn et al. 2009 IPR-NE sequelae mildb 40.00% 23.00% 56.00% 20.75 31.13 GBD estimate IPR-NE sequelae moderate/severeb 30.00% 13.00% 62.00% 5.81 13.55 GBD estimate Parameter Most likely Low value High value α β Source Sex malea 51.04% 49.74% 52.34% 4522.60 4338.33 Projahnmo I Liveborn (male)a 94.96% 94.16% 95.76% 4294.14 227.98 Projahnmo I Liveborn (female)a 95.79% 95.04% 96.54% 4155.12 182.69 Projahnmo I PT incidence (male)a 18.04% 16.60% 19.47% 774.50 3519.59 Projahnmo I PT incidence (female)a 16.80% 15.38% 18.22% 697.98 3457.10 Projahnmo I LPT proportion (male)a 83.67% 80.42% 86.92% 647.33 126.35 Projahnmo I LPT proportion (female)a 83.45% 80.00% 86.89% 581.74 115.41 Projahnmo I LPT survive (male)a 95.18% 93.12% 97.24% 615.98 31.19 Projahnmo I LPT survive (female)a 94.10% 91.71% 96.49% 547.27 34.30 Projahnmo I LPT sequelaeb 5.00% 4.23% 10.00% 14.45 274.92 (Shibuya and Murray 1998b; Moster et al. 2008; Chyi et al. 2008; Huddy et al. 2001 VPT proportion (male)a 16.33% 13.08% 19.58% 126.35 647.33 Projahnmo I VPT proportion (female)a 16.55% 13.11% 20.00% 115.41 581.74 Projahnmo I VPT survive (male)a 77.78% 68.72% 86.83% 97.62 27.89 Projahnmo I VPT survive (female)a 86.49% 78.70% 94.28% 99.09 15.48 Projahnmo I VPT sequelaeb 68.00% 58.00% 78.00% 89.59 42.16 (Khan et al. 2006 Suspected infection incidence (male)a 15.89% 14.52% 17.26% 682.37 3611.72 Projahnmo I Suspected infection incidence (female)a 13.83% 12.52% 15.14% 574.62 3580.46 Projahnmo I Percent meningitisa 3.42% 2.70% 4.14% 131.16 3699.06 (WHO Young Infants Study Group 1999 Percent Haemophilus influenzae type bc 31.10% 20.00% 72.00% 5.02 11.12 (Edmond et al. 2010 Percent Streptococcus pneumoniaec 23.60% 14.10% 40.20% 14.39 46.60 (Edmond et al. 2010 Percent Neisseria meningitidisc 9.40% 5.10% 32.10% 2.19 21.09 (Edmond et al. 2010 Percent other infectionc 13.40% 8.20% 28.10% 8.62 55.74 (Edmond et al. 2010 Haemophilus type b survivec 56.00% 38.00% 83.00% 15.58 12.24 (Watt et al. 2009 Streptococcus pneumoniae survivec 43.00% 29.00% 56.00% 34.25 45.40 (O’Brien et al. 2009 Neisseria meningitidis survivec 56.00% 38.00% 83.00% 15.58 12.24 Assumption based on (Watt et al. 2009 Other infection survivec 56.00% 38.00% 83.00% 15.58 12.24 Assumption based on (Watt et al. 2009 Haemophilus influenzae type b sequelaec 14.50% 0.01% 27.10% 5.72 33.75 (Edmond et al. 2010 Streptococcus pneumoniae sequelaec 34.70% 28.20% 45.10% 64.57 121.51 (Edmond et al. 2010 Neisseria meningitidis sequelaec 9.50% 5.10% 15.10% 19.41 184.94 (Edmond et al. 2010 Other infection sequelaec 19.90% 12.10% 35.20% 13.58 54.68 (Edmond et al. 2010 IPR-NE incidence overallb 1.34% 0.40% 2.70% 7.94 584.26 (Lawn et al. 2009), GBD estimate IPR-NE survive overallb 69.00% 67.00% 80.00% 180.03 80.88 (Lawn et al. 2009 IPR-NE sequelae mildb 40.00% 23.00% 56.00% 20.75 31.13 GBD estimate IPR-NE sequelae moderate/severeb 30.00% 13.00% 62.00% 5.81 13.55 GBD estimate Notes: a95% confidence interval; bLow/high range; cInter-quartile range. IPR-NE = intrapartum-related neonatal encephalopathy; PT = preterm birth; LPT = late preterm birth; VPT = very preterm birth; GBD = Global Burden of Disease. Table 2b Input parameters: Proportional differences between model arms Incremental Value Source Notes Infection incidence 48.24% (Bang et al. 2005c; Blencowe et al. 2011Estimate based on Gadchiroli data from 1995–96 to 1997–98. Reported incidence in Projahnmo I increased from 11% to 14% from 2004 to 2005 in the home care arm, and also increased in Gadchiroli up to 2003, probably reflecting improvements in reporting. Blencowe et al. (2011) found an odds ratio of 0.51 for infection incidence combining four studies for the impact of hand washing on cord infection. Infection survival 72.4% (Baqui et al. 2008Improvement in survival in Projahnmo I is assumed to be from treatment of infections. Mortality reduced in the comparison arm by 74.1% the amount in the intervention arm. IPR-NE incidence 1% Assumption 1% derives from 5% birth attendance times 20% of IPR-NE that can be avoided in community settings with appropriate intervention (1% = 5% × 20%). Projahnmo I used wiping, drying, and mouth-to-mouth resuscitation, but not bag and mask resuscitation. IPR-NE survival Not affected Assumption In Projahnmo I, there was little attendance at birth, to either reduce intrapartum injury via obstetric care or to provide immediate neonatal resuscitation. Preterm incidence Not affected (Arifeen et al. 2000Levels of preterm birth are similar between this study and a study from Dhaka slums (16.8%). Since progesterone administration and smoking cessation were not part of Projahnmo I, reduced incidence is unlikely. Preterm survival Not affected Projahnmo I No significant difference found in Projahnmo I cause-specific mortality. Incremental Value Source Notes Infection incidence 48.24% (Bang et al. 2005c; Blencowe et al. 2011Estimate based on Gadchiroli data from 1995–96 to 1997–98. Reported incidence in Projahnmo I increased from 11% to 14% from 2004 to 2005 in the home care arm, and also increased in Gadchiroli up to 2003, probably reflecting improvements in reporting. Blencowe et al. (2011) found an odds ratio of 0.51 for infection incidence combining four studies for the impact of hand washing on cord infection. Infection survival 72.4% (Baqui et al. 2008Improvement in survival in Projahnmo I is assumed to be from treatment of infections. Mortality reduced in the comparison arm by 74.1% the amount in the intervention arm. IPR-NE incidence 1% Assumption 1% derives from 5% birth attendance times 20% of IPR-NE that can be avoided in community settings with appropriate intervention (1% = 5% × 20%). Projahnmo I used wiping, drying, and mouth-to-mouth resuscitation, but not bag and mask resuscitation. IPR-NE survival Not affected Assumption In Projahnmo I, there was little attendance at birth, to either reduce intrapartum injury via obstetric care or to provide immediate neonatal resuscitation. Preterm incidence Not affected (Arifeen et al. 2000Levels of preterm birth are similar between this study and a study from Dhaka slums (16.8%). Since progesterone administration and smoking cessation were not part of Projahnmo I, reduced incidence is unlikely. Preterm survival Not affected Projahnmo I No significant difference found in Projahnmo I cause-specific mortality. ## Results Incidence of the three major neonatal conditions in Projahnmo I were 18.0% (male) and 16.8% (female) for preterm birth (Projahnmo I), 15.9% (male) and 13.8% (female) for suspected clinical infection (Projahnmo I) and 1.34% for IPR-NE (Lawn et al. 2009). The sex ratio for preterm birth in Projahnmo I data (1.1) was lower than the 1.22 to 1.5 that is expected from published data (Cooperstock and Campbell 1996; Ingemarsson 2003). Among preterm births, 83.67% (male) and 83.45% (female) were 33–36 weeks (Projahnmo I). Among suspected community-acquired infections, 3.42% were meningitis according to data pooled from four high mortality countries (WHO Young Infants Study Group 1999). Percentage survival was 95.18% (male) and 94.10% (female) for late preterm birth (Projahnmo I), 77.78% (male) and 86.49% (female) for very preterm birth (Projahnmo I), 43% or 56% for different types of meningitis (Edmond et al. 2010) and 69% for infants with IPR-NE (Lawn et al. 2009). Sequelae resulted for 5% of late preterm infants (estimate), 68% for very preterm infants (Khan et al. 2006) and 9.5–34.7% for infants with meningitis. Forty per cent of infants with IPR-NE developed mild sequelae, and 30% developed moderate to severe sequelae (Lee et al. 2011a). Before intervention, the full model predicted 22/1000 neonates having any preterm sequelae, 9/1000 having IPR-NE sequelae and <1/1000 having meningitis sequelae. The estimate for infection is likely low as only meningitis sequelae were considered due to lack of data on pneumonia, sepsis and viral infection from Southeast Asia. Furthermore, these numbers for ‘sequelae’ do not reflect morbidity severity or duration. The result for IPR-NE sequelae is similar to that found in Nepal (8/1000) (Lee et al. 2011a) and GBD project modelling (6/1000). These figures are not standard incidence calculations, but are products of ‘rolling back’ or taking the weighted average of parameters from each branch of the tree. When morbidity duration is accounted for, the proportional burden of disability is preterm (73%), IPR-NE (24%) and meningitis (3%) (Table 3). Proportions of more specific outcomes are given in Table 4. Table 3 Results per 1000 neonates. YLD (r,K) notation is used [r = discount rate (%), K = age weighting factor (1 present, 0 absent)] Dutch weights Preterm treated/Untreated Simple Percentages by condition Expected value 95% CI Expected value 95% CI Expected value 95% CI Preterm IPR-NE Meningitis YLD (3,0) National life expectancy 195 157 241 199 150 265 120 77 180 73% 24% 3% Regional life expectancy 197 158 243 200 150 266 121 77 181 73% 24% 3% Global standard life expectancy 234 189 288 266 210 336 127 93 174 74% 24% 3% YLD (3,1) National life expectancy 219 176 271 213 159 285 109 74 157 73% 25% 3% Regional life expectancy 221 178 273 214 160 286 131 83 198 73% 24% 3% Global standard life expectancy 254 206 314 289 229 366 162 112 232 74% 24% 3% YLD (6,0) National life expectancy 120 97 148 127 98 164 62 44 87 73% 24% 3% YLD (0,0) National life expectancy 379 303 473 386 277 535 201 131 300 72% 25% 3% Dutch weights Preterm treated/Untreated Simple Percentages by condition Expected value 95% CI Expected value 95% CI Expected value 95% CI Preterm IPR-NE Meningitis YLD (3,0) National life expectancy 195 157 241 199 150 265 120 77 180 73% 24% 3% Regional life expectancy 197 158 243 200 150 266 121 77 181 73% 24% 3% Global standard life expectancy 234 189 288 266 210 336 127 93 174 74% 24% 3% YLD (3,1) National life expectancy 219 176 271 213 159 285 109 74 157 73% 25% 3% Regional life expectancy 221 178 273 214 160 286 131 83 198 73% 24% 3% Global standard life expectancy 254 206 314 289 229 366 162 112 232 74% 24% 3% YLD (6,0) National life expectancy 120 97 148 127 98 164 62 44 87 73% 24% 3% YLD (0,0) National life expectancy 379 303 473 386 277 535 201 131 300 72% 25% 3% Notes: YLD = years lost to disability; IPR-NE = intrapartum-related neonatal encephalopathy. Table 4 Proportion of neonates developing each sequelae Sequelae Total 2.71 Seizure disorder 0.20 Hearing loss 0.15 Motor deficit 0.40 Visual disturbance 0.17 Mild intellectual deficit 0.75 Moderate intellectual deficit 0.19 Severe intellectual deficit 0.17 Profound intellectual deficit 0.06 Speech delay 0.41 Behaviour problems 0.18 Cerebral palsy without intellectual deficit 0.02 Sequelae Total 2.71 Seizure disorder 0.20 Hearing loss 0.15 Motor deficit 0.40 Visual disturbance 0.17 Mild intellectual deficit 0.75 Moderate intellectual deficit 0.19 Severe intellectual deficit 0.17 Profound intellectual deficit 0.06 Speech delay 0.41 Behaviour problems 0.18 Cerebral palsy without intellectual deficit 0.02 YLD expected for 1000 neonates are shown according to different DALY formula assumptions and sequelae branches (Table 3). YLD changed as expected when different life expectancies were used, age-weighting was added and the discount rate was adjusted. Using 1998 GBD weights and the assumption that 85% of preterm patients remain untreated was similar to the reference case. In the simple analysis, while the weighted average of GBD weights and full life expectancies were greater than the weighted average of Dutch weights, we considered shorter life spans for very preterm in the simple analysis (GBD weights) according to a published single estimate (Profit et al. 2010). The main analysis found more disability than the analysis using single disability weights for preterm, meningitis and IPR-NE because the weighted average for the preterm disability weights is higher. If years of healthy life are valued at per-capita Gross National Income (GNI) (CMH 2001),$731 in Bangladesh (deflated from 2012 to 2010) (World Bank 2012), the resulting value of health losses is $142 940 per 1000 neonates. One-way sensitivity analyses highlight the importance of parameters with wide ranges of uncertainty (IPR-NE incidence, percent very preterm sequelae), parameters proximal in the tree (preterm incidence, percent late preterm) and parameters with strong preceding nodes (percent late preterm sequelae) in influencing the YLD estimates (Table 5). Infections parameters were not as influential given the low probability of survival and of sequelae among survivors. Overall, YLD increase with incidence, survival and sequelae. Results are confirmed by tornado diagrams that highlight the importance of IPR-NE and preterm incidence, percent of infections that are meningitis, being late preterm, or having very or late preterm sequelae (Figure 2). With all parameters set at their worst case scenario, YLD (3,0) per 1000 increase to 374 [719 for YLD (0,0)]. Figure 2 Sensitivity analyses, tornado diagram (YLD 3,0) ordinary least squares coefficients *Significant at P < 0.05. Note: IPR-NE = intrapartum-related neonatal encephalopathy. Figure 2 Sensitivity analyses, tornado diagram (YLD 3,0) ordinary least squares coefficients *Significant at P < 0.05. Note: IPR-NE = intrapartum-related neonatal encephalopathy. Table 5 One-way sensitivity analysis Parameter Parameterization YLD % Ref Case Parameter source Reference case 195 100% PT incidence Low 16.6%, 15.38% 184 94.33% Projahnmo I High 19.47%, 18.22% 206 105.48% Projahnmo I Percent LPT Low 80.42%, 80.00% 213 109.19% Projahnmo I High 86.92%, 86.89% 177 90.72% Projahnmo I Percent LPT survive Low 93.12%, 91.71% 195 99.51% Projahnmo I High 97.24%, 96.49% 196 100.37% Projahnmo I Percent VPT survive Low 68.72%, 78.70% 185 94.65% Projahnmo I High 86.83%, 94.28% 207 105.87% Projahnmo I Percent LPT sequelae Low 4.23% 189 96.81% (Moster et al. 2008 High 10% 238 121.92% (Chyi et al. 2008; Huddy et al. 2001 Percent VPT sequelae Low 58% 181 92.63% (Khan et al. 2006 High 78% 210 107.17% (Khan et al. 2006 Infections incidence Low 14.52%, 12.52% 193 98.80% Projahnmo I High 17.26%, 15.14% 197 100.56% Projahnmo I Percent meningitis Low 2.70% 195 99.79% (WHO Young Infants Study Group 1999 High 4.14% 197 100.79% (WHO Young Infants Study Group 1999 Percent Haemophilus influenzae type b Low 20.00% 196 100.29% (Edmond et al. 2010 High 72.00% 195 99.56% (Edmond et al. 2010 Percent Streptococcus pneumoniae Low 14.10% 195 99.96% (Edmond et al. 2010 High 40.20% 197 100.71% (Edmond et al. 2010 Percent Neisseria meningitidis Low 5.10% 195 99.74% (Edmond et al. 2010 High 32.10% 196 100.03% (Edmond et al. 2010 Percent Other infection Low 8.20% 193 99.74% (Edmond et al. 2010 High 28.10% 196 100.07% (Edmond et al. 2010 Survive Haemophilus influenzae type b Low 38.00% 193 98.80% (Watt et al. 2009 High 83.00% 196 100.04% (Watt et al. 2009 Survive Streptococcus pneumoniae Low 29% 194 99.32% (O’Brien et al. 2009 High 56% 196 100.04% (O’Brien et al. 2009 Survive Neisseria meningitidis Low 38.00% 194 99.03% Assumption based on Watt et al. (2009) High 83.00% 196 100.07% Assumption based on Watt et al. (2009) Survive Other infections Low 38.00% 193 98.88% Assumption High 83.00% 195 100.09% Assumption Sequelae Haemophilus influenzae type b Low 7.10% 194 99.33% (Edmond et al. 2010 High 15.20% 196 100.14% (Edmond et al. 2010 Sequelae Streptococcus pneumoniae Low 16.20% 193 98.82% (Edmond et al. 2010 High 35.40% 196 100.22% (Edmond et al. 2010 Sequelae Neisseria meningitidis Low 4.10% 195 99.53% (Edmond et al. 2010 High 11.00% 196 100.06% (Edmond et al. 2010 Sequelae Other infections Low 7.10% 194 99.38% (Edmond et al. 2010 High 21.10% 196 100.26% (Edmond et al. 2010 IPR-NE incidence Low 0.40% 162 83.12% GBD estimate High 2.70% 244 124.77% GBD estimate IPR-NE survive Low 67% 195 99.87% (Lawn et al. 2009 High 80% 197 100.56% (Lawn et al. 2009 IPR-NE sequelae mild Low 23% 191 97.54% GBD estimate High 56% 200 102.10% GBD estimate IPR-NE sequelae moderate/severe Low 13% 193 98.74% GBD estimate High 62% 200 102.23% GBD estimate Parameter Parameterization YLD % Ref Case Parameter source Reference case 195 100% PT incidence Low 16.6%, 15.38% 184 94.33% Projahnmo I High 19.47%, 18.22% 206 105.48% Projahnmo I Percent LPT Low 80.42%, 80.00% 213 109.19% Projahnmo I High 86.92%, 86.89% 177 90.72% Projahnmo I Percent LPT survive Low 93.12%, 91.71% 195 99.51% Projahnmo I High 97.24%, 96.49% 196 100.37% Projahnmo I Percent VPT survive Low 68.72%, 78.70% 185 94.65% Projahnmo I High 86.83%, 94.28% 207 105.87% Projahnmo I Percent LPT sequelae Low 4.23% 189 96.81% (Moster et al. 2008 High 10% 238 121.92% (Chyi et al. 2008; Huddy et al. 2001 Percent VPT sequelae Low 58% 181 92.63% (Khan et al. 2006 High 78% 210 107.17% (Khan et al. 2006 Infections incidence Low 14.52%, 12.52% 193 98.80% Projahnmo I High 17.26%, 15.14% 197 100.56% Projahnmo I Percent meningitis Low 2.70% 195 99.79% (WHO Young Infants Study Group 1999 High 4.14% 197 100.79% (WHO Young Infants Study Group 1999 Percent Haemophilus influenzae type b Low 20.00% 196 100.29% (Edmond et al. 2010 High 72.00% 195 99.56% (Edmond et al. 2010 Percent Streptococcus pneumoniae Low 14.10% 195 99.96% (Edmond et al. 2010 High 40.20% 197 100.71% (Edmond et al. 2010 Percent Neisseria meningitidis Low 5.10% 195 99.74% (Edmond et al. 2010 High 32.10% 196 100.03% (Edmond et al. 2010 Percent Other infection Low 8.20% 193 99.74% (Edmond et al. 2010 High 28.10% 196 100.07% (Edmond et al. 2010 Survive Haemophilus influenzae type b Low 38.00% 193 98.80% (Watt et al. 2009 High 83.00% 196 100.04% (Watt et al. 2009 Survive Streptococcus pneumoniae Low 29% 194 99.32% (O’Brien et al. 2009 High 56% 196 100.04% (O’Brien et al. 2009 Survive Neisseria meningitidis Low 38.00% 194 99.03% Assumption based on Watt et al. (2009) High 83.00% 196 100.07% Assumption based on Watt et al. (2009) Survive Other infections Low 38.00% 193 98.88% Assumption High 83.00% 195 100.09% Assumption Sequelae Haemophilus influenzae type b Low 7.10% 194 99.33% (Edmond et al. 2010 High 15.20% 196 100.14% (Edmond et al. 2010 Sequelae Streptococcus pneumoniae Low 16.20% 193 98.82% (Edmond et al. 2010 High 35.40% 196 100.22% (Edmond et al. 2010 Sequelae Neisseria meningitidis Low 4.10% 195 99.53% (Edmond et al. 2010 High 11.00% 196 100.06% (Edmond et al. 2010 Sequelae Other infections Low 7.10% 194 99.38% (Edmond et al. 2010 High 21.10% 196 100.26% (Edmond et al. 2010 IPR-NE incidence Low 0.40% 162 83.12% GBD estimate High 2.70% 244 124.77% GBD estimate IPR-NE survive Low 67% 195 99.87% (Lawn et al. 2009 High 80% 197 100.56% (Lawn et al. 2009 IPR-NE sequelae mild Low 23% 191 97.54% GBD estimate High 56% 200 102.10% GBD estimate IPR-NE sequelae moderate/severe Low 13% 193 98.74% GBD estimate High 62% 200 102.23% GBD estimate Notes: IPR-NE = intrapartum-related neonatal encephalopathy; PT = preterm birth; LPT = late preterm birth; VPT = very preterm birth. With assumed incremental reductions, our model predicts that 2.0 (95% CI 1.3–2.8) DALYs would be averted per 1000 neonates (Table 6a). This disability is valued at$1427 (95% CI $933–$2037) per 1000 neonates. Applied to the original cost-effectiveness analysis, adding YLD leads to 0.6% additional DALYs averted and reduces the incremental cost-effectiveness ratio (ICER) from $105.23 to$104.62 ($65.15–$266.60) per DALY averted (Table 6b).

Table 6a

Incremental results: Burden of disease

YLD averted

Valuation according to GNI

Expected value 95% CI Expected value 95% CI
YLD (3,0) National life expectancy $1427$933 $2037 Regional life expectancy$1658 $1176$2262
Global standard life expectancy $1942$1371 $2679 YLD (3,1) National life expectancy$1451 $889$2129
Regional life expectancy $1867$1325 $2549 Global standard life expectancy$2117 $1494$2920
YLD (6,0) National life expectancy $962$668 $1341 YLD (0,0) National life expectancy$2110 $1100$3335
YLD averted

Valuation according to GNI

Expected value 95% CI Expected value 95% CI
YLD (3,0) National life expectancy $1427$933 $2037 Regional life expectancy$1658 $1176$2262
Global standard life expectancy $1942$1371 $2679 YLD (3,1) National life expectancy$1451 $889$2129
Regional life expectancy $1867$1325 $2549 Global standard life expectancy$2117 $1494$2920
YLD (6,0) National life expectancy $962$668 $1341 YLD (0,0) National life expectancy$2110 $1100$3335

Notes: YLD = years lost to disability; GNI = Gross National Income.

Table 6b

Incremental results: Cost-effectiveness

Expected value 95% CI
Mortality based analysis
Incremental costs $35 358$32 870 $37 565 YLLs averted 336 130 539 Cost-effectiveness$105.23 $69.73$253.48
Population size 1000
Per person estimates
Incremental cost $35.36$32.87 $37.57 YLLs averted 0.336 0.130 0.539 YLDs averted 0.002 0.001 0.003 DALYs averted 0.338 0.131 0.542 Proportion YLDs 0.5% 1.0% 0.5% Cost-effectiveness$104.62 $65.15$266.60
Expected value 95% CI
Mortality based analysis
Incremental costs $35 358$32 870 $37 565 YLLs averted 336 130 539 Cost-effectiveness$105.23 $69.73$253.48
Population size 1000
Per person estimates
Incremental cost $35.36$32.87 $37.57 YLLs averted 0.336 0.130 0.539 YLDs averted 0.002 0.001 0.003 DALYs averted 0.338 0.131 0.542 Proportion YLDs 0.5% 1.0% 0.5% Cost-effectiveness$104.62 $65.15$266.60

Notes: YLD = years lost to disability; YLL = years of life lost; DALY = disability-adjusted life year.

## Discussion

Calculated from YLD results, our analysis found a burden of morbidity equivalent to approximately 7 deaths per 1000 neonates after intervention [one death equals 28 discounted (3%) DALYs]. Preterm was the biggest contributor to the YLD count, despite most being late preterm with low probability of sequelae, as preterm has the highest incidence and survival. Few neonates develop IPR-NE and 31% die, but it contributes the second largest amount of YLD since 70% of survivors develop neurodevelopmental sequelae. Data are incomplete on outcomes from pneumonia, sepsis and viral infections, and meningitis accounts for the smallest amount of sequelae in our model. Reflecting correct DALY formulation, discounting had a strong effect on results, and age weighting had little effect. Adjusting life expectancy had the strongest effect when the West Life Table was tested with age weighting.

In the United States, 50% of neurological sequelae from neonatal conditions is due to preterm birth (Goldenberg and Culhane 2007), which can be compared with the 73% predicted by our model. YLD per 1000 neonates were 142 for preterm, 48 for IPR-NE and 5 for meningitis; which together are higher than the YLD estimate for overall disability in Bangladesh from the WHO World Report on Disability (10.1/100 people) (WHO 2011). ABD calculations found 7 YLD for LBW, 3 for IPR-NE and 1 for infection per 1000 neonates. Bangladesh has a higher incidence of preterm births (17.4% in Projahnmo I, compared with 6.5% for LBW in Australia) (WHO 2004), but has a slightly lower probability that these infants will survive to develop sequelae. More YLD were incurred due to infection in Bangladesh than Australia because incidence and sequelae were more prevalent in Bangladesh (sequelae were seen in 2.71% of the total population compared with 0.9% in Australia). Overall, both improved survival and neurodevelopmental outcomes may be expected in all but the most premature neonates as quality and access to care improve through time (Vohr et al. 2005; Wilson-Costello et al. 2007; Roberts et al. 2010).

Externally validated to other developing country settings, preterm incidence in the Projahnmo I dataset (17.4%) was higher than the incidence reported for south central Asia, at 11.4% (10.0–12.7) (Beck et al. 2010), although this is consistent with other studies from Bangladesh (Al Mamun et al. 2006; Arifeen et al. 2000). IPR-NE incidence estimates applied to our model are less than ‘asphyxia’ incidence estimates from Africa (3–26.5/1000) (Tafari 1985; Airede 1991a; Airede 1991b) and CHERG modelling (18.6/1000) (Lee et al. 2011c). The incidence of community-acquired meningitis among neonates presenting to health facilities in four high mortality countries in the Young Infants Study Group was 3.4% (WHO Young Infants Study Group 1999) is higher than the laboratory-confirmed community-based estimate of infection from Bangladesh (0.29%) assessed by CHWs (Darmstadt et al. 2009b).

The model structure is relevant to comparable settings in Southeast Asia and sub-Saharan Africa (SSA); however, many published studies in Africa are hospital-based and less useful for community-based parameters (Hodgson et al. 2001; Klingenberg et al. 2003). Some exceptions exist (Hinderaker et al. 2003), and community-based programmes are established in Pakistan and India, which have similar demographic and epidemiologic trends to Bangladesh. Further programmes are developing in Ethiopia, Malawi and Tanzania where national neonatal mortality rates are comparable. Data from these and other clinical trials may inform parameters, although long-term outcomes are difficult to measure (Villar and Shah 2004).

Optimally, impairment from ‘asphyxia’ may be measured epidemiologically according to IPR-NE definitions, although this is not possible for most community settings. New tools are available, which have been validated for use in identifying disabilities at the population level in children under 2 years (Khan et al. 2010; Khan et al. 2012). These tools have the potential to advance our ability to measure the burden of disabilities and impairments in the community.

A non-systematic review of literature was performed to establish ranges of uncertainty according to published sources relevant to the context, which was limited by a scarcity of Bangladeshi data. In generalizing to other settings, incidence estimates may be confounded by socio-economic, health system, cultural, ethnic, environmental or genetic factors. Ways in which these factors can have an effect include access and care-seeking at health services, unsafe traditional practices, unawareness of beneficial health practices, suspicion of care providers perceived as outsiders, limited geographic mobility, and epidemiological patterns such as the increasing incidence of preterm birth (Lawn et al. 2010).

There are several limitations of this analysis. Data are difficult to collect in LMICs as many births do not have a qualified attendant present. In Projahnmo I, 5% of births were attended, particularly relevant for evaluating IPR-NE (although since then the proportion of neonates that receive a visit from a village health worker within 24 hours after birth has improved to 87%) (Shah et al. 2010). Where a complete lack of evidence exists from Bangladesh, data were used from a multi-country trial, pooled estimates or GBD parameterizations. Longitudinal studies varied in their follow-up times which may indicate administrative censoring, and outcomes occur with low frequency in the population, adding to uncertainty in estimates.

Variation between impairment assessments was substantial and data quality was inconsistent, thus comparing articles defining limits for sensitivity analysis was necessarily rough—different outcome variables were measured, using different tools, defined at different thresholds, tested at different ages and including different sample cohorts. Different diagnostic algorithms were used in Projahnmo I and the Young Infants Study Group; the latter may have been less specific since it included infants that the mother or health worker thought were ill generally (Young Infants Clinical Signs Study Group 2008). Data on all relevant infections is incomplete, and survival and sequelae parameterization was based solely on meningitis outcomes (Edmond et al. 2010). Studies in the infections arm reported outcomes for children under 5 (O’Brien et al. 2009; Watt et al. 2009; Edmond et al. 2010), not neonates, which we assume would be more severe. Follow-up times were variable (most sequelae will present within 2 years) (Marlow 2004). Different sample cohorts came from hospital and community-based settings.

While this model uses discrete definitions, it is not always possible to delineate clearly the boundaries between different neonatal conditions, particularly in the context of disaggregating ‘birth asphyxia’ from infection (Shibuya and Murray 1998a). Each of the three disease categories share clinical symptoms and are risk factors for each of the others (except IPR-NE as a risk factor for preterm). All three conditions lead to similar long-term outcomes including functional limitations, cognitive and behavioural deficits, seizure and severe hearing loss. It is difficult to distinguish an intrapartum etiology for IPR-NE in community settings (Nelson and Leviton 1991), although verbal autopsy instruments exist for distinguishing between risk factors. Several IPR-NE symptoms are difficult to assess (Darmstadt et al. 2009a), and agreement on definition is undergoing evolution (Halloran et al. 2009; Lawn et al. 2009). Methods are developing to cross-check community-based data based on computer algorithms and expert review to improve identification of ‘birth asphyxia’ in the community setting (Lee et al. 2008), and the proportion of institutional deliveries may be expected to increase as programmes develop. Diagnostic algorithms for infections are non-specific in community settings where health workers of minimal training cannot distinguish sepsis, meningitis and pneumonia from each other or various conditions with overlapping clinical signs, and which frequently coexist. Even for qualified providers, separate algorithms for preterm, infection and ‘birth asphyxia’ have been poorly validated (Weber et al. 2003; Lee et al. 2008). Improvements in algorithm diagnosis of infection have since been made for sepsis (Baqui et al. 2009c) and neurodevelopmental status (Khan et al. 2010).

Preterm is usually defined based on time since last menstrual period, or a clinical assessment (Beck et al. 2010), although clinical assessments (Dubowitz and Ballard scores) can be too complex for community settings. Last menstrual period has been shown to be relatively accurate in various settings, only slightly overestimating age (depending on quality of measurement) (Rosenberg et al. 2009), with wider uncertainty among populations with lower education levels. Substituting preterm for LBW may be more problematic for incidence and sequelae than mortality because almost all LBW neonatal deaths are preterm (Jamison et al. 2006).

Some relevant conditions were not included such as intra-uterine growth restriction (IUGR) and congenital anomalies. IUGR incidence in Bangladesh was found to be 65% in Dhaka slums (Arifeen et al. 2000), although mortality evidence from Bangladesh is limited to a doctoral research project (SE Arifeen, Johns Hopkins University) and further research is needed (Walker et al. 2007). The CHERG is generating further data, and a crude odds ratio suggests that IUGR neonates may be at risk of developmental deficits, although this relationship is not significant when adjusted for confounders (Wu et al. 2011). While it has a 60.3% prevalence in Nepal (Wu et al. 2011), IUGR is less prevalent in African settings (Manji 2009). The impact of malaria on preterm (Menendez et al. 2000; Ngassa 2000), neonatal outcomes among HIV infected women (Fawzi et al. 2000; Mehta et al. 2008) and bacterial resistance to antibiotics may be additional considerations. For congenital anomalies, only a few preventive or curative interventions exist in the developing country setting (such as folic acid during pregnancy, micronutrient supplementation, and nutritional and substance abuse counselling). The burden of congenital anomalies is being estimated in forthcoming work by WHO/CHERG, emphasizing neural tube defects, heart conditions and chromosomal abnormalities.

Current incidence data are defined according to single conditions. Co-morbidities would reduce incidence in single-condition branches, and occur frequently in Bangladesh (Khan et al. 2006). Survival in co-morbidity branches is likely to be lower than single condition branches (Bang et al. 2005d), the probability of sequelae would be higher, and disability weights would be higher. Co-morbidity data from Gadchiroli, India exist (Bang et al. 2005e; Bang et al. 2005a) and some relevant Dutch disability weights are available (Stouthard et al. 1997), although work is needed to develop relevant disability weights in a Southeast Asian setting. Since co-morbidity is not considered, YLD results are likely to be biased; neonates cannot die twice and resulting disability is unlikely to be additive. Evidence is needed for other outcomes; for example preterm birth, neonatal infection and IPR-NE can all lead to epilepsy. Disability weights should be developed for speech delays and mild behavioural and learning conditions (Khan et al. 2006).

Proportional differences between model arms are based on the best available evidence. Incidence of suspected clinical infection was only measured in one trial arm of Projahnmo I, IPR-NE evidence is not available, and the validity and precision of other evidence is limited due to low birth attendance. Interventions to address neonatal health conditions are reviewed in detail (Darmstadt et al. 2005; Lawn et al. 2006; Lawn et al. 2009; Wall et al. 2009; Barros et al. 2010), and investigating their expected effectiveness is advocated where access to perinatal services is limited. However, care is not always provided in the home, neonatal conditions do not necessarily stimulate care seeking, and improvements to the burden of disease are likely to vary across locations.

Uncertainty around disability weights is being evaluated in the forthcoming version of the GBD study (GBD 2008), which will allow for a more complete stochastic analysis. Disability weights for preterm are forthcoming in the current GBD project (WHO 2010). The Dutch Public Health Status and Forecast study also provides disability weights for very preterm [0.52 (0.446–0.574)], and ranges of uncertainty around estimates for the conditions it included (Stouthard et al. 1997), although it is not clear what disability weight to use for late preterm. A 3% discount rate reflects preferences of programme funders, and is WHO standard (Tan-Torres Edejer et al. 2003); however, people in poverty have more uncertainty about the future, and may value it less, making a 6–7% discount rate more representative of a local perspective (Robberstad 2005). Recognizing the use of high discount rates in other sectors (Harrison 2010), the implications of using them are that long-term health outcomes in developing countries will be valued less, which goes against the objectives of relevant programmes and analysts.

Ongoing research initiatives may address gaps in data, particularly proportional differences between model arms and long-term outcomes. Extrapolation from developed country trends can inform future forecasting as Bangladesh transitions from a high mortality setting and the relative importance of different conditions changes (Tafari 1985; Dunn 1986). The neonatal mortality rate in Bangladesh has declined by 17% in the last decade from 36 to 30 (UNICEF 2009), down from 41–42 in the late 1990s (NIPORT et al. 2005).

Disability weights may differ over life course, such as for Down’s syndrome (Stouthard et al. 1997), and taking that into account would require a Markov model rather than a decision tree. Evidence is mixed whether at-risk and disabled neonates can overcome disability, achieve academically at least equivalent to their peers, and successfully integrate into society (Marlow 2004; Maulik and Darmstadt 2009). Preterm and normal term people report similar values for quality of life (Saigal et al. 2006; Gäddlin et al. 2009; Baumgardt et al. 2012), and parents’ valuations of preterm conditions are usually higher than those of clinicians (Saigal et al. 1999).

While disability is not explicitly mentioned in the Millennium Development Goals (Groce and Trani 2009), results are relevant through the United Nations Convention on the Rights of Persons with Disabilities (CRPD) (UN 2007). Also, the World Report on Disability was recently released (WHO 2011), and disability is a current focus of the CHERG (CHERG 2008). Research suggests a link between infectious disease and intelligence at the national level (Eppig et al. 2010), and long-term sequelae from neonatal conditions is a priority at the Gates Foundation, which is currently conducting a survey of newborns in Dar es Salaam (Manji 2009). The burden of disease results from this study may be of interest to several groups, and care is needed to ensure that the correct skill mix of health workers and professionals is available to maintain or reduce levels of disability in Sylhet, Bangladesh (Ahmed et al. 2011).

## Conclusions

Based on Australian and Global Burden of Disease precedent, our model quantifies the burden of disease resulting from neonatal conditions. It suggests that the burden of neonatal disability in Sylhet, Bangladesh is significant, particularly for preterm birth, with little effect on the interpretation of the cost-effectiveness of the community-based intervention. Quantifying burden of disease can be useful to help identify emerging trends and future needs, determine priorities for expenditure, provide information to educate the public, identify which interventions will have the greatest effect, help set research agendas (Hyder and Morrow 2006), break policy inertia that results from lack of information, and to help programme managers monitor and improve the effectiveness of programmes. Studies examining IPR-NE incidence and longitudinal studies examining the sequelae of all three disease categories in LMICs are needed to improve precision of results and to consider new settings. Based on current evidence, health decision-makers should implement CHW programmes in similar settings and expect a small but significant decrease in morbidity.

## Funding

This research was funded by a grant from the Bill and Melinda Gates Foundation to the US Fund for UNICEF to support the Child Health Epidemiology Reference Group (CHERG) grant number 50140.

None declared.

## Acknowledgements

Theo Vos provided valuable feedback at the design stage of the analysis. Emma Williams, Heather Rosen and Luke Mullany provided insight on data and concepts in neonatal health. Richard Morrow contributed ideas for analysis and model development. Louis Niessen and John Bridges provided valuable feedback on the draft and methods before final revision. Monte Carlo Simulations were made possible by a Visual Basic for Applications macro written by Chris Bombardo. Tom Smith provided instruction for fitting distributions to data more precisely and efficiently.

## References

T
Lim
SS
Mehta
S
, et al.  .
Cost effectiveness analysis of strategies for maternal and neonatal health in developing countries
British Medical Journal
,
2005
, vol.
331
pg.
1107

Ahmed
SM
Hossain
MA
Rajachowdhury
AM
Bhuiya
AU
The health workforce crisis in Bangladesh: shortage, inappropriate skill-mix and inequitable distribution
Human Resources for Health
,
2011
, vol.
9

Airede
AI
Birth asphyxia and hypoxic-ischaemic encephalopathy: incidence and severity
Annals of Tropical Paediatrics
,
1991a
, vol.
11
(pg.
331
-
5
)
Airede
KI
Neonatal seizures and a 2-year neurological outcome
Journal of Tropical Pediatrics
,
1991b
, vol.
37
(pg.
313
-
17
)
Al Mamun
AA
SS
Khatun
M
Maternal health during pregnancy and perinatal mortality in Bangladesh: evidence from a large-scale community-based clinical trial
Paediatric and Perinatal Epidemiology
,
2006
, vol.
20
(pg.
482
-
90
)
Anon
The global burden of preterm birth
The Lancet
,
2009
, vol.
371
(pg.
75
-
84
)
Arifeen
SE
Black
RE
Caulfield
LE
, et al.  .
Infant growth patterns in the slums of Dhaka in relation to birth weight, intrauterine growth retardation, and prematurity
American Journal of Clinical Nutrition
,
2000
, vol.
72
(pg.
1010
-
17
)
Azra Haider
B
Bhutta
ZA
Birth asphyxia in developing countries: current status and public health implications
Current Problems in Pediatric and Adolescent Health Care
,
2006
, vol.
36
(pg.
178
-
88
)
Bang
AT
Bang
RA
Stoll
BJ
, et al.  .
Is home-based diagnosis and treatment of neonatal sepsis feasible and effective? Seven years of intervention in the Gadchiroli field trial (1996 to 2003)
Journal of Perinatology
,
2005a
, vol.
25
(pg.
S62
-
S71
)
Bang
AT
Bang
RA
Baitule
SB
Reddy
HM
Deshmukh
MD
Management of birth asphyxia in home deliveries in rural Gadchiroli: the effect of two types of birth attendants and of resuscitating with mouth-to-mouth, tube-mask or bag–mask
Journal of Perinatology
,
2005b
, vol.
25
(pg.
S82
-
S91
)
Bang
AT
Bang
RA
Reddy
HM
Deshmukh
MD
Baitule
SB
Reduced incidence of neonatal morbidities: effect of home-based neonatal care in rural Gadchiroli, India
Journal of Perinatology
,
2005c
, vol.
25
(pg.
S51
-
S61
)
Bang
AT
Reddy
HM
Bang
RA
Deshmukh
M
Why do neonates die in rural Gadchiroli India? (Part II): estimating population attributable risks and contribution of multiple morbidities for identifying a strategy to prevent deaths
Journal of Perinatology
,
2005d
, vol.
25
(pg.
S35
-
S43
)
Bang
AT
Baitule
SB
Reddy
HM
Deshmukh
MD
Bang
RA
Low birth weight and preterm neonates: can they be managed at home by mother and a trained village health worker?
Journal of Perinatology
,
2005e
, vol.
25
(pg.
S72
-
S81
)
Baqui
A
El Arifeen
S
G
, et al.  .
Effect of community-based newborn-care intervention package implemented through two service-delivery strategies in Sylhet district, Bangladesh: a cluster-randomised controlled trial
The Lancet
,
2008
, vol.
371
(pg.
1936
-
44
)
Baqui
AH
Arifeen
SE
Williams
EK
, et al.  .
Effectiveness of home-based management of newborn infections by community health workers in rural Bangladesh
Pediatric Infectious Disease Journal
,
2009a
, vol.
28
(pg.
304
-
10
)
Baqui
AH
Ahmed
S
El Arifeen
S
, et al.  .
Effect of timing of first postnatal care home visit on neonatal mortality in Bangladesh: an observational cohort study
British Medical Journal
,
2009b
, vol.
339
pg.
b2826

Baqui
AH
Arifeen
SE
Rosen
HE
, et al.  .
Community based validation of assessment of newborn illnesses by trained community health workers in Sylhet district of Bangladesh
Tropical Medicine & International Health
,
2009c
, vol.
14
(pg.
1448
-
56
)
Barendregt
JJ
Bonneux
L
Van Der Maas
PJ
DALYs: the age-weights on balance
Bulletin of the World Health Organization
,
1996
, vol.
74
(pg.
439
-
43
)
Barros
FC
Bhutta
ZA
Batra
M
, et al.  .
Global report on preterm birth and stillbirth (3 of 7): evidence for effectiveness of interventions
BMC Pregnancy and Childbirth
,
2010
, vol.
10
pg.
S3

Baumgardt
M
Bucher
HU
Mieth
RA
Fauchère
JC
Health related quality of life of former very preterm infants in adulthood
Acta Paediatrica
,
2012
, vol.
101
(pg.
e59
-
63
)
Beck
S
Wojdyla
D
Say
L
, et al.  .
The worldwide incidence of preterm birth: a WHO systematic review of maternal mortality and morbidity
Bulletin of the World Health Organization
,
2010
, vol.
88
(pg.
31
-
38
)
Black
RE
Cousens
S
Johnson
H
, et al.  .
Global, regional, and national causes of child mortality in 2008: a systematic analysis
The Lancet
,
2010
, vol.
375
(pg.
1969
-
87
)
Blencowe
H
Cousens
S
Mullany
LC
, et al.  .
Clean birth and postnatal care practices to reduce neonatal deaths from sepsis and tetanus: a systematic review and Delphi estimation of mortality effect
BMC Public Health
,
2011
, vol.
11
pg.
S11

Borghi
J
Thapa
B
Osrin
D
, et al.  .
Economic assessment of a women’s group intervention to improve birth outcomes in rural Nepal
The Lancet
,
2005
, vol.
366
(pg.
1882
-
4
)
Boyle
CA
Boulet
S
Schieve
LA
, et al.  .
Trends in the prevalence of developmental disabilities in US children, 1997–2008
Pediatrics
,
2011
, vol.
127
(pg.
1034
-
42
)
CHERG
Child Health and Epidemiology Reference Group website
2008
Geneva
World Health Organization

Chyi
LJ
Lee
HC
Hintz
SR
Gould
JB
Sutcliffe
TL
School outcomes of late preterm infants: special needs and challenges for infants born at 32 to 36 weeks gestation
Journal of Pediatrics
,
2008
, vol.
153
(pg.
25
-
31
)
CMH
Macroeconomics and Health: Investing in Health for Economic Development
,
2001
Geneva
World Health Organization
Coale
AJ
Demeney
P
Regional Model Life Tables and Stable Populations
,
1983
New York
Cooperstock
M
Campbell
J
Excess males in preterm birth: interactions with gestational age, race and multiple birth
Obstetrics & Gynecology
,
1996
, vol.
88
(pg.
189
-
93
)
D’Amore
A
Broster
S
Le Fort
W
Curley
A
Two-year outcomes from very low birthweight infants in a geographically defined population across 10 years, 1993–2002: comparing 1993–1997 with 1998–2002
Archives of Disease in Childhood. Fetal and Neonatal Edition
,
2011
, vol.
96
(pg.
F178
-
F185
)
GL
Baqui
AH
Choi
Y
, et al.  .
Validation of clinical algorithm to identify neonates with severe illness during routine household visits in rural Bangladesh
Archives of Disease in Childhood
,
2011
, vol.
96
(pg.
1140
-
6
)
GL
Bhutta
ZA
Cousens
S
, et al.  .
Evidence-based, cost-effective interventions: how many newborn babies can we save?
The Lancet
,
2005
, vol.
365
(pg.
977
-
88
)
GL
Baqui
AH
Choi
Y
, et al.  .
Validation of community health workers’ assessment of neonatal illness in rural Bangladesh
Bulletin of the World Health Organization
,
2009a
, vol.
87
(pg.
12
-
19
)
GL
Lee
ACC
Cousens
S
, et al.  .
60 million non-facility births: who can deliver in community settings to reduce intrapartum-related deaths?
International Journal of Gynecology & Obstetrics
,
2009b
, vol.
107
(pg.
S89
-
S112
)
GL
Saha
SK
Choi
Y
, et al.  .
Population-based incidence and etiology of community-acquired neonatal bacteremia in Mirzapur, Bangladesh: an observational study
Journal of Infectious Diseases
,
2009c
, vol.
200
(pg.
906
-
15
)
Dunn
HG
Sequelae of Low Birth Weight: The Vancouver Study
,
1986
Oxford
Blackwells Scientific Publications
Durkin
MS
Schneider
H
Pathania
VS
, et al.  .
Jamison
DT
Breman
J
Measham
A
, et al.  .
Learning and developmental disabilities
Disease Control Priorities in Developing Countries
,
2006
2nd
Washington, DC
The World Bank and Oxford University Press
Edmond
K
Clark
A
Korczak
VS
, et al.  .
Global and regional risk of disabling sequelae from bacterial meningitis: a systematic review and meta-analysis
The Lancet
,
2010
, vol.
10
(pg.
317
-
28
)
Eppig
C
Fincher
CL
Thornhill
R
Parasite prevalence and the worldwide distribution of cognitive ability
Proceedings. Biological Sciences
,
2010
, vol.
277
(pg.
3801
-
8
)
Fawzi
WW
Msamanga
G
Hunter
D
, et al.  .
Randomized trial of vitamin supplements in relation to vertical transmission of HIV-1 in Tanzania
Journal of Acquired Immune Deficiency Syndromes
,
2000
, vol.
23
(pg.
246
-
45
)
Furyk
JS
Swann
O
Molyneux
E
Systematic review: neonatal meningitis in the developing world
Tropical Medicine & International Health
,
2011
, vol.
16
(pg.
672
-
9
)
Gäddlin
PO
Finnström
O
Sydsjö
G
Leijon
I
Most very low birth weight subjects do well as adults
Acta Paediatrica
,
2009
, vol.
98
pg.
1513

GBD
The global burden of diseases, injuries, and risk factors study: operations manual
2008
Cambridge, MA
Harvard Initiative for Global Health, Institute for Health Metrics and Evaluation at the University of Washington, Johns Hopkins University, University of Queensland, World Health Organization
Ghana Health Assessment Project Team
A quantitative method of assessing the health impact of different diseases in less developed countries
International Journal of Epidemiology
,
1981
, vol.
10
(pg.
73
-
80
)
Goldenberg
R
Culhane
JF
Low birth weight in the United States
American Journal of Clinical Nutrition
,
2007
, vol.
85
(pg.
584S
-
90S
)
Gordon
AL
English
M
Dzombo
JT
Karisa
M
Newton
C
Neurological and developmental outcome of neonatal jaundice and sepsis in rural Kenya
Tropical Medicine & International Health
,
2005
, vol.
10
(pg.
1114
-
20
)
Gottlieb
CA
Maenner
MJ
Cappa
C
Durkin
MS
Child disability screening, nutrition, and early learning in 18 countries with low and middle incomes: data from the third round of UNICEF’s Multiple Indicator Cluster Survey (2005–06)
The Lancet
,
2009
, vol.
374
(pg.
1831
-
9
)
Groce
NE
Trani
J-F
Millennium Development Goals and people with disabilities
The Lancet
,
2009
, vol.
374
(pg.
1800
-
1
)
Halloran
DR
McClure
E
Chakraborty
H
, et al.  .
Birth asphyxia survivors in a developing country
Journal of Perinatology
,
2009
, vol.
29
(pg.
243
-
9
)
Harrison
M
Valuing the future: the social discount rate in cost-benefit analysis
2010
Canberra
Australian Government Productivity Commission
Hinderaker
SG
Olsen
BE
Bergsjø
PB
, et al.  .
Avoidable stillbirths and neonatal deaths in rural Tanzania
British Journal of Obstetrics & Gynaecology
,
2003
, vol.
110
(pg.
616
-
23
)
Hodgson
A
Smith
T
Gagneux
S
, et al.  .
Survival and sequelae of meningococcal meningitis in Ghana
International Journal of Epidemiology
,
2001
, vol.
30
(pg.
1440
-
6
)
Hyder
AA
Morrow
RH
Merson
MH
Black
RE
Mills
AJ
Measures of health and disease in populations
International Public Health
,
2006
2nd
Sudbury, MA
Jones and Bartlett Publishers
Huddy
CL
Johnson
A
Hope
PL
Educational and behavioral problems in babies of 32-35 weeks gestation
Archives of Disease in Childhood Fetal and Neonatal Edition
,
2001
, vol.
85
(pg.
F23
-
F28
)
IMF
World Economic Outlook Database
,
2012

International Monetary Fund website. Online at: http://www.imf.org/external/pubs/ft/weo/2012/01/weodata/index.aspx, accessed 17 June 2012
I
Gender aspects of preterm birth
British Journal of Obstetrics & Gynaaecology
,
2003
, vol.
110
(pg.
34
-
8
)
IOM
Improving Birth Outcomes: Meeting the Challenge in the Developing World
,
2003
Washington, DC
Jamison
DT
Shahid-Salles
SA
Jamison
J
Lawn
JE
Zupan
J
Lopez
Mathers
CD
Ezzati
M
Jamison
DT
Murray
CJL
Incorporating deaths near the time of birth into estimates of the Global Burden of Disease
Global Burden of Disease and Risk Factors
,
2006
New York
Oxford University Press
Jang
DH
Sung
IY
Jeon
JY
, et al.  .
Neurodevelopmental outcomes in very low-birth-weight infants in Korea: 1998–2007 vs 1989–1997
Journal of Child Neurology
,
2011
, vol.
26
(pg.
1405
-
10
)
Khan
N
Durkin
M
Zinkin
P
McConachie
H
Framework: Prevalence
Disabled Children in Developing Countries: Clinics in Developmental Medicine. No. 136
,
1995
London
MacKeith Press
Khan
NZ
Muslima
H
Begum
D
, et al.  .
Validation of rapid neurodevelopmental assessment instrument for under-two-year-old children in Bangladesh
Pediatrics
,
2010
, vol.
125
(pg.
e755
-
62
)
Khan
NZ
Muslima
H
Parveen
M
, et al.  .
Neurodevelopmental outcomes of preterm infants in Bangladesh
Pediatrics
,
2006
, vol.
118
(pg.
280
-
9
)
Khan
NZ
Muslima
H
Shilpi
AB
, et al.  .
Validation of a home-based neurodevelopmental screening tool for under 2-year-old children in Bangladesh
Child: Care, Health, and Development
,
2012

Jun 8. doi: 10.1111/j.1365-2214.2012.01393.x
Klingenberg
C
Olomi
R
Oneko
M
Sam
N
Langeland
N
Neonatal morbidity and mortality in a Tanzanian tertiary care referral hospital
Annals of Tropical Paediatrics
,
2003
, vol.
23
(pg.
293
-
9
)
Kuti
O
Owa
JA
Gestational age-specific neonatal mortality among preterm singleton births in a Nigerian tertiary institution
International Journal of Gynecology & Obstetrics
,
2003
, vol.
80
(pg.
319
-
20
)
Lawn
J
K
Stein
C
No cry at birth: global estimates of intrapartum stillbirths and intrapartum-related neonatal deaths
Bulletin of the World Health Organization
,
2005
, vol.
83
(pg.
409
-
17
)
Lawn
JE
Gravett
MG
Nunes
TM
Rubens
CE
Stanton
C
GAPPS Review Group
Global report on preterm birth and stillbirth (1 of 7): definitions, description of the burden and opportunities to improve data
BMC Pregnancy and Childbirth
,
2010
, vol.
10

Suppl. 1
pg.
S1

Lawn
JE
Lee
ACC
Kinney
M
, et al.  .
Two million intrapartum-related stillbirths and neonatal deaths: where, why, and what can be done?
International Journal of Gynecology & Obstetrics
,
2009
, vol.
107
(pg.
S5
-
S19
)
Lawn
JE
Manandhar
A
Haws
RA
GL
Reducing one million child deaths from birth asphyxia: a survey of health systems gaps and priorities
Health Research Policy & Systems
,
2007
, vol.
5
pg.
4

Lawn
JE
Zupan
J
Begkoyian
G
Jamison
DT
Breman
J
Measham
A
, et al.  .
Newborn survival
Disease Control Priorities in Developing Countries
,
2006
2nd
Washington, DC
The World Bank and Oxford University Press
Lee
ACC
Mullany
LC
Tielsch
JM
, et al.  .
Incidence and risk factors of neonatal respiratory depression and encephalopathy in rural Sarlahi, Nepal
Pediatrics
,
2011a
, vol.
128
pg.
e915

Lee
ACC
Cousens
S
Wall
SN
, et al.  .
Neonatal resuscitation and immediate newborn assessment and stimulation for the prevention of neonatal deaths: a systematic review, meta-analysis and Delphi estimation of mortality effect
BMC Public Health
,
2011b
, vol.
11
pg.
S12

Lee
AC
Cousens
S
GL
, et al.  .
Care during labor and birth for the prevention of intrapartum-related neonatal deaths: a systematic review and Delphi estimation of mortality effect
BMC Public Health
,
2011c
, vol.
11
pg.
S10

Lee
ACC
Mullany
LC
Tielsch
JM
, et al.  .
Verbal autopsy methods to ascertain birth asphyxia deaths in a community-based setting in southern Nepal
Pediatrics
,
2008
, vol.
121
pg.
e1372

Lefevre
AE
Shillcutt
SD
Saha
SK
, et al.  .
Cost-effectiveness of skin barrier enhancing emollients to prevent infection and death among preterm neonates in Bangladesh
Bulletin of the World Health Organization
,
2010
, vol.
88
(pg.
104
-
12
)
Lopez
Mathers
CD
Ezzati
M
Jamison
DT
Murray
CJL
Lopez
Mathers
CD
Ezzati
M
Jamison
DT
Murray
CJL
Measuring the global burden of disease and risk factors, 1990–2001
Global Burden of Disease and Risk Factors
,
2006
Geneva
World Health Organization
Manji
K
Situation analysis of newborn health in Tanzania: current situation, existing plans, and strategic next steps for newborn health
2009
Dar es Salaam
Ministry of Health and Social Welfare, Save the Children
Marlow
N
Neurocognitive outcome after very preterm birth
Archives of Disease in Childhood. Fetal and Neonatal Edition
,
2004
, vol.
89
(pg.
F224
-
8
)
Martin
JA
Hamilton
BE
Sutton
PD
, et al.  .
Births: Final data for 2002
National Vital Statistics Reports: Centers for Disease Control
,
2003
, vol.
52
(pg.
1
-
114
)
Mathers
CD
Bernard
C
Iburg
KM
, et al.  .
Global burden of disease in 2002: data sources, methods, and results
2004
Geneva
World Health Organization

Global Programme on Evidence for Health Policy Discussion Paper No 54
Mathers
CD
Lopez
Murray
CJL
Lopez
Mathers
CD
Ezzati
M
Jamison
DT
Murray
CJL
The burden of disease and mortality by condition: data, methods, and results for 2001
Global Burden of Disease and Risk Factors
,
2006a
New York
World Bank
Mathers
CD
Salomon
JA
Ezzati
M
, et al.  .
Lopez
Mathers
CD
Ezzati
M
Jamison
DT
Murray
CJL
Sensitivity and uncertainty analyses for burden of disease and risk factor estimates
Global Burden of Disease and Risk Factors
,
2006b
New York
World Bank
Mathers
CD
Vos
ET
Stevenson
CE
Begg
SJ
The burden of disease and injury in Australia
Bulletin of the World Health Organization
,
2001
, vol.
79
(pg.
1076
-
84
)
Maulik
PK
GL
Childhood disability in low- and middle-income countries: overview of screening, prevention, services, legislation, and epidemiology
Pediatrics
,
2007
, vol.
120

Suppl. 1
(pg.
S1
-
55
)
Maulik
PK
GL
Community-based interventions to optimize early childhood development in low resource settings
Journal of Perinatology
,
2009
, vol.
29
(pg.
531
-
42
)
Mehta
S
Manji
KP
Young
AM
, et al.  .
Nutritional indicators of adverse pregnancy outcomes and mother-to-child transmission of HIV among HIV-infected women
American Journal of Clinical Nutrition
,
2008
, vol.
87
(pg.
1639
-
49
)
Menendez
C
Ordi
J
Ismail
MR
, et al.  .
The impact of placental malaria on gestational age and birth weight
Journal of Infectious Diseases
,
2000
, vol.
181
(pg.
1740
-
5
)
Moster
D
Lie
RT
T
Long-term medical and social consequences of preterm birth
New England Journal of Medicine
,
2008
, vol.
359
pg.
262

Muhe
L
Tilahun
M
Lulseged
S
, et al.  .
Etiology of pneumonia, sepsis and meningitis in infants younger than three months of age in Ethiopia
Pediatric Infectious Disease Journal
,
1999
, vol.
18
(pg.
S56
-
S61
)
Murray
CJL
Lopez
The global burden of disease: a comprehensive assessment of mortality and disability from diseases, injuries, and risk factors in 1990 and projected to 2020
,
1996
Cambridge, MA
Harvard University Press
Musgrove
P
Fox-Rushby
JA
Jamison
D
Breman
J
Measham
A
, et al.  .
Cost-effectiveness analysis for priority setting
Disease Control Priorities in Developing Countries
,
2006
2nd
Washington, DC
The World Bank and Oxford University Press
Nelson
KB
Leviton
A
How much of neonatal encephalopathy is due to birth asphyxia?
American Journal of Diseases of Children
,
1991
, vol.
145
(pg.
1325
-
31
)
Ngassa
C
Malaria parasitaemia and the risk of preterm labour: a re-evaluation of the evidence
African Journal of Reproductive Health
,
2000
, vol.
4
(pg.
53
-
61
)
NIPORT, Mitra & Associates, ORC Macro
Bangladesh Demographic and Health Survey 2004
2005
Dhaka, Bangladesh and Calverton, Maryland (USA)
National Istitute of Population Research and Training (NIPORT), Mitra and Associates, and ORC Macro
O’Brien
KL
Wolfson
LJ
Watt
JP
, et al.  .
Burden of disease caused by Streptococcus pneumoniae in children younger than 5 years: global estimates
The Lancet
,
2009
, vol.
374
(pg.
893
-
902
)
Olusanya
BO
Somefun
AO
Sensorineural hearing loss in infants with neonatal jaundice in Lagos: a community-based study
Annals of Tropical Paediatrics
,
2009
, vol.
29
(pg.
119
-
28
)
Platt
MJ
Cans
C
Johnson
A
, et al.  .
Trends in cerebral palsy among infants of very low birthweight (<1500 g) or born prematurely (<32 weeks) in 16 European centres: a database study
The Lancet
,
2007
, vol.
369
(pg.
43
-
50
)
Profit
J
Lee
D
Zupancic
JA
, et al.  .
Clinical benefits, costs, and cost-effectiveness of neonatal intensive care in Mexico
PLoS Medicine
,
2010
, vol.
7
pg.
e1000379

B
Estimation of private and social time preferences for health in northern Tanzania
Social Science & Medicine
,
2005
, vol.
61
(pg.
1597
-
607
)
Roberts
G
Anderson
PJ
De Luca
C
Doyle
LW
Victorian Infant Collaborative Study Group
Changes in neurodevelopmental outcome at age eight in geographic cohorts of children born at 22–27 weeks’ gestational age during the 1990s
Archives of Disease in Childhood. Fetal & Neonatal Edition
,
2010
, vol.
95
(pg.
F90
-
4
)
Rosenberg
RE
Ahmed
AS
Ahmed
S
, et al.  .
Determining gestational age in a low-resource setting: validity of last menstrual period
Journal of Health, Population & Nutrition
,
2009
, vol.
27
(pg.
332
-
38
)
Saha
SK
Khan
NZ
Ahmed
A
, et al.  .
Neurodevelopmental sequelae in pneumococcal meningitis cases in Bangladesh: a comprehensive follow-up study
Clinical Infectious Diseases
,
2009
, vol.
48
(pg.
S90
-
96
)
Saigal
S
Stoskopf
B
Pinelli
J
, et al.  .
Self-perceived health-related quality of life of former extremely low birth weight infants at young adulthood
Pediatrics
,
2006
, vol.
118
(pg.
1140
-
8
)
Saigal
S
Stoskopf
BL
Feeny
D
, et al.  .
Differences in preferences for neonatal outcomes among health care professionals, parents, and adolescents
Journal of the American Medical Association
,
1999
, vol.
281
(pg.
1991
-
7
)
Saving Newborn Lives
The State of the World’s Newborns: A Report from Saving Newborn Lives
,
2001
Washington, DC

Save the Children
Sendi
PP
Clemen
RT
Sensitivity analysis on a chance node with more than two branches
Medical Decision Making
,
1999
, vol.
19
(pg.
499
-
502
)
Shah
R
Munos
MK
Winch
PJ
, et al.  .
Community-based health workers achieve high coverage in neonatal intervention trials: a case study from Sylhet, Bangladesh
Journal of Health, Population & Nutrition
,
2010
, vol.
28
(pg.
610
-
18
)
K
Murray
CJL
Murray
CJL
Lopez
Birth asphyxia. Chapter 11
Health Dimensions of Sex and Reproduction
,
1998a
Geneva
World Health Organization and Harvard University Press
K
Murray
CJL
Murray
CJL
Lopez
Low birth weight. Chapter 10
Health Dimensions of Sex and Reproduction
,
1998b
Geneva
World Health Organization and Harvard University Press
Stouthard
MEA
Essink-Bot
ML
Bonsel
GJ
Disability weights for diseases: a modified protocol and results for a Western European region
European Journal of Public Health
,
2000
, vol.
10
(pg.
24
-
30
)
Stouthard
MEA
Essink-Bot
ML
Bonsel
GJ
, et al.  .
Disability Weights for Diseases in the Netherlands
,
1997
Rotterdam
Erasmus University
Tafari
N
Sterkey
HG
Tafari
N
Tunnell
R
Epidemiology of birth asphyxia
Breathing and Warmth at Birth: Judging the Appropriateness of Technology
,
1985
Stockholm
Swedish Agency for Research Cooperation with Developing Countries
Tan-Torres Edejer
T
Baltussen
R
T
, et al.  .
WHO Guide to Cost-Effectiveness Analysis
,
2003
Geneva
World Health Organization
Thaver
D
Zaidi
AKM
Burden of neonatal infections in developing countries
Pediatric Infectious Disease Journal
,
2009
, vol.
28
(pg.
S3
-
S9
)
UN
Convention on the rights of persons with disabilities (resolution 61/106)
2007
New York
United Nations
UNICEF
2009
Geneva
United Nations Childrens Fund

Van Haastert
IC
Groenendaal
F
Uiterwaal
C
, et al.  .
Decreasing incidence and severity of cerebral palsy in prematurely born children
Journal of Pediatrics
,
2011
, vol.
159
(pg.
86
-
112
)
Villar
J
Shah
A
WHO global data system on maternal and perinatal health study protocol. A multi-stage stratified sample survey of mode of delivery and maternal and perinatal outcomes
2004
Geneva
World Health Organization
Vohr
BR
Wright
LL
Poole
WK
McDonald
SA
Neurodevelopmental outcomes of extremely low birth weight infants <32 weeks’ gestation between 1993 and 1998
Pediatrics
,
2005
, vol.
116
pg.
635

Walker
SP
Wachs
TD
Gardner
JM
, et al.  .
Child development: risk factors for adverse outcomes in developing countries
The Lancet
,
2007
, vol.
369
(pg.
145
-
57
)
Wall
SN
Lee
ACC
Niermeyer
S
, et al.  .
Neonatal resuscitation in low-resource settings: what, who, and how to overcome challenges to scale up?
International Journal of Gynecology & Obstetrics
,
2009
, vol.
107
(pg.
S47
-
S64
)
Watt
JP
Wolfson
LJ
O’Brien
KL
, et al.  .
Burden of disease caused by Haemophilus influenzae type b in children younger than 5 years: global estimates
The Lancet
,
2009
, vol.
374
(pg.
903
-
11
)
Weber
MW
Carlin
JB
Gatchalian
S
Al
E
Predictors of neonatal sepsis in developing countries
Pediatric Infectious Disease Journal
,
2003
, vol.
22
(pg.
711
-
17
)
WHO
Low birth weight: country, regional, and global estimates
2004
Geneva
World Health Organization
WHO
World Health Report 2005: Making Every Mother and Child Count
,
2005
Geneva
World Health Organization
WHO
Health statistics and health information systems: National tools
2009
Geneva
World Health Organization

WHO
Global Burden of Disease (GBD) 2010 study website
2010
Geneva
World Health Organization

WHO
World Report on Disability
,
2011
Geneva
World Health Organization and the World Bank
WHO Young Infants Study Group
Bacterial etiology of serious infections in young infants in developing countries: results of a multicenter study
Pediatric Infectious Disease Journal
,
1999
, vol.
18
(pg.
S17
-
22
)

WHO. 2012. World Health Organization Choosing Interventions that are Cost-Effective website. Geneva: World Health Organization. Online at: http://www.who.int/choice/en/, accessed 20 Aug 2012
Wilson-Costello
D
Friedman
H
Minich
N
, et al.  .
Improved neurodevelopmental outcomes for extremely low birth weight infants in 2000–2002
Pediatrics
,
2007
, vol.
119
(pg.
37
-
45
)
Winter
S
Autry
A
Boyle
C
Yeargin-Allsopp
M
Trends in the prevalence of cerebral palsy in a population-based study
Pediatrics
,
2002
, vol.
110
(pg.
1220
-
25
)
World Bank
World Development Indicators Data Query website
2012
Washington, DC
World Bank

Wu
LA
Katz
J
Mullany
LC
, et al.  .
The association of preterm birth and small birthweight for gestational age on childhood disability screening using the Ten Questions Plus tool in rural Sarlahi district, southern Nepal
Child: Care, Health, and Development
,
2011
, vol.
38
(pg.
332
-
40
)
Young Infants Clinical Signs Study Group
Clinical signs that predict severe illness in children under age 2 months: a multicentre study
The Lancet
,
2008
, vol.
371
(pg.
135
-
42
)