PMCC PMCC

Search tips
Search criteria

Advanced
Results 1-25 (1362895)

Clipboard (0)
None

Related Articles

1.  Drivers of Inequality in Millennium Development Goal Progress: A Statistical Analysis 
PLoS Medicine  2010;7(3):e1000241.
David Stuckler and colleagues examine the impact of the HIV and noncommunicable disease epidemics on low-income countries' progress toward the Millennium Development Goals for health.
Background
Many low- and middle-income countries are not on track to reach the public health targets set out in the Millennium Development Goals (MDGs). We evaluated whether differential progress towards health MDGs was associated with economic development, public health funding (both overall and as percentage of available domestic funds), or health system infrastructure. We also examined the impact of joint epidemics of HIV/AIDS and noncommunicable diseases (NCDs), which may limit the ability of households to address child mortality and increase risks of infectious diseases.
Methods and Findings
We calculated each country's distance from its MDG goals for HIV/AIDS, tuberculosis, and infant and child mortality targets for the year 2005 using the United Nations MDG database for 227 countries from 1990 to the present. We studied the association of economic development (gross domestic product [GDP] per capita in purchasing-power-parity), the relative priority placed on health (health spending as a percentage of GDP), real health spending (health system expenditures in purchasing-power-parity), HIV/AIDS burden (prevalence rates among ages 15–49 y), and NCD burden (age-standardised chronic disease mortality rates), with measures of distance from attainment of health MDGs. To avoid spurious correlations that may exist simply because countries with high disease burdens would be expected to have low MDG progress, and to adjust for potential confounding arising from differences in countries' initial disease burdens, we analysed the variations in rates of change in MDG progress versus expected rates for each country. While economic development, health priority, health spending, and health infrastructure did not explain more than one-fifth of the differences in progress to health MDGs among countries, burdens of HIV and NCDs explained more than half of between-country inequalities in child mortality progress (R2-infant mortality  = 0.57, R2-under 5 mortality  = 0.54). HIV/AIDS and NCD burdens were also the strongest correlates of unequal progress towards tuberculosis goals (R2 = 0.57), with NCDs having an effect independent of HIV/AIDS, consistent with micro-level studies of the influence of tobacco and diabetes on tuberculosis risks. Even after correcting for health system variables, initial child mortality, and tuberculosis diseases, we found that lower burdens of HIV/AIDS and NCDs were associated with much greater progress towards attainment of child mortality and tuberculosis MDGs than were gains in GDP. An estimated 1% lower HIV prevalence or 10% lower mortality rate from NCDs would have a similar impact on progress towards the tuberculosis MDG as an 80% or greater rise in GDP, corresponding to at least a decade of economic growth in low-income countries.
Conclusions
Unequal progress in health MDGs in low-income countries appears significantly related to burdens of HIV and NCDs in a population, after correcting for potentially confounding socioeconomic, disease burden, political, and health system variables. The common separation between NCDs, child mortality, and infectious syndromes among development programs may obscure interrelationships of illness affecting those living in poor households—whether economic (e.g., as money spent on tobacco is lost from child health expenditures) or biological (e.g., as diabetes or HIV enhance the risk of tuberculosis).
Please see later in the article for the Editors' Summary
Editors' Summary
Background
In 2000, 189 countries adopted the United Nations (UN) Millennium Declaration, which commits the world to the eradication of extreme poverty by 2015. The Declaration lists eight Millennium Development Goals (MDGs), 21 quantifiable targets, and 60 indicators of progress. So, for example, MDG 4 aims to reduce child mortality (deaths). The target for this goal is to reduce the number of children who die each year before they are five years old (the under-five mortality rate) to two-thirds of its 1990 value by 2015. Indicators of progress toward this goal include the under-five mortality rate and the infant mortality rate. Because poverty and ill health are inextricably linked—ill health limits the ability of individuals and nations to improve their economic status, and poverty contributes to the development of many illnesses—two other MDGs also tackle public health issues. MDG 5 sets a target of reducing maternal mortality by three-quarters of its 1990 level by 2015. MDG 6 aims to halt and begin to reverse the spread of HIV/AIDS, malaria, and other major diseases such as tuberculosis by 2015.
Why Was This Study Done?
Although progress has been made toward achieving the MDGs, few if any of the targets are likely to be met by 2015. Worryingly, low-income countries are falling furthest behind their MDG targets. For example, although child mortality has been declining globally, in many poor countries there has been little or no progress. What is the explanation for this and other inequalities in progress toward the health MDGs? Some countries may simply lack the financial resources needed to combat epidemics or may allocate only a low proportion of their gross domestic product (GDP) to health. Alternatively, money allocated to health may not always reach the people who need it most because of an inadequate health infrastructure. Finally, coexisting epidemics may be hindering progress toward the MDG health targets. Thus, the spread of HIV/AIDS may be hindering attempts to limit the spread of tuberculosis because HIV infection increases the risk of active tuberculosis, and ongoing epidemics of diabetes and other noncommunicable diseases (NCDs) may be affecting the attainment of health MDGs by diverting scarce resources. In this study, the researchers investigate whether any of these possibilities is driving the inequalities in MDG progress.
What Did the Researchers Do and Find?
The researchers calculated how far 227 countries were from their MDG targets for HIV, tuberculosis, and infant and child mortality in 2005 using information collected by the UN. They then used statistical methods to study the relationship between this distance and economic development (GDP per person), health spending as a proportion of GDP (health priority), actual health system expenditures, health infrastructure, HIV burden, and NCD burden in each country. Economic development, health priority, health spending, and health infrastructure explained no more than one-fifth of the inequalities in progress toward health MDGs. By contrast, the HIV and NCD burdens explained more than half of inequalities in child mortality progress and were strongly associated with unequal progress toward tuberculosis goals. Furthermore, the researchers calculated that a 1% reduction in the number of people infected with HIV or a 10% reduction in rate of deaths from NCDs in a population would have a similar impact on progress toward the tuberculosis MDG target as a rise in GDP corresponding to at least a decade of growth in low-income countries.
What Do These Findings Mean?
These findings are limited by the quality of the available data on health indicators in low-income countries and, because the researchers used country-wide data, their findings only reveal possible drivers of inequalities in progress toward MDGs in whole countries and may mask drivers of within-country inequalities. Nevertheless, as one of the first attempts to analyze the determinants of global inequalities in progress toward the health MDGs, these findings have important implications for global health policy. Most importantly, the finding that unequal progress is related to the burdens of HIV and NCDs in populations suggests that programs designed to achieve health MDGs must consider all the diseases and factors that can trap households in vicious cycles of illness and poverty, especially since the achievement of feasible reductions in NCDs in low-income countries could greatly enhance progress towards health MDGs.
Additional Information
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1000241.
The United Nations Millennium Development Goals website provides detailed information about the Millennium Declaration, the MDGs, their targets and their indicators
The Millennium Development Goals Report 2009 and its progress chart provide an up-to-date assessment of progress towards the MDGs
The World Health Organization provides information about poverty and health and health and development
doi:10.1371/journal.pmed.1000241
PMCID: PMC2830449  PMID: 20209000
2.  Progress toward Global Reduction in Under-Five Mortality: A Bootstrap Analysis of Uncertainty in Millennium Development Goal 4 Estimates 
PLoS Medicine  2012;9(12):e1001355.
Leontine Alkema and colleagues use a bootstrap procedure to assess the uncertainty around the estimates of the under-five mortality rate produced by the United Nations Inter-Agency Group for Child Mortality Estimation.
Background
Millennium Development Goal 4 calls for an annual rate of reduction (ARR) of the under-five mortality rate (U5MR) of 4.4% between 1990 and 2015. Progress is measured through the point estimates of the United Nations Inter-agency Group for Child Mortality Estimation (UN IGME). To facilitate evidence-based conclusions about progress toward the goal, we assessed the uncertainty in the estimates arising from sampling errors and biases in data series and the inferior quality of specific data series.
Methods and Findings
We implemented a bootstrap procedure to construct 90% uncertainty intervals (UIs) for the U5MR and ARR to complement the UN IGME estimates. We constructed the bounds for all countries without a generalized HIV epidemic, where a standard estimation approach is carried out (174 countries). In the bootstrap procedure, potential biases in levels and trends of data series of different source types were accounted for. There is considerable uncertainty about the U5MR, particularly for high mortality countries and in recent years. Among 86 countries with a U5MR of at least 40 deaths per 1,000 live births in 1990, the median width of the UI, relative to the U5MR level, was 19% for 1990 and 48% for 2011, with the increase in uncertainty due to more limited data availability. The median absolute width of the 90% UI for the ARR from 1990 to 2011 was 2.2%. Although the ARR point estimate for all high mortality countries was greater than zero, for eight of them uncertainty included the possibility of no improvement between 1990 and 2011. For 13 countries, it is deemed likely that the ARR from 1990 to 2011 exceeded 4.4%.
Conclusions
In light of the upcoming evaluation of Millennium Development Goal 4 in 2015, uncertainty assessments need to be taken into account to avoid unwarranted conclusions about countries' progress based on limited data.
Please see later in the article for the Editors' Summary
Editors' Summary
Background
In September 2000, world leaders adopted the United Nations Millennium Declaration, committing member states (countries) to a new global partnership to reduce extreme poverty and improve global health by setting out a series of time-bound targets with a deadline of 2015—the Millennium Development Goals (MDGs). There are eight MDGs and the fourth, MDG 4, focuses on reducing the number of deaths in children aged under five years by two-thirds from the 1990 level. Monitoring progress towards meeting all of the MDG targets is of vital importance to measure the effectiveness of interventions and to prioritize slow progress areas. MDG 4 has three specific indicators, and every year, the United Nations Inter-agency Group for Child Mortality Estimation (the UN IGME, which includes the key agencies the United Nations Children's Fund, the World Health Organization, the World Bank, and the United Nations Population Division) produces and publishes estimates of child death rates for all countries.
Why Was This Study Done?
Many poorer countries do not have the infrastructure and the functioning vital registration systems in place to record the number of child deaths. Therefore, it is difficult to accurately assess levels and trends in the rate of child deaths because there is limited information (data) or because the data that exists may be inaccurate or of poor quality. In order to deal with this situation, analyzing trends in under-five child death rates (to show progress towards MDG 4) currently focuses on the “best” estimates from countries, a process that relies on “point” estimates. But this practice can lead to inaccurate results and comparisons. It is therefore important to identify a framework for calculating the uncertainty surrounding these estimates. In this study, the researchers use a statistical method to calculate plausible uncertainty intervals for the estimates of death rates in children aged under five years and the yearly reduction in those rates.
What Did the Researchers Do and Find?
The researchers used the publicly available information from the UN IGME 2012 database, which collates data from a variety of sources, and a statistical method called bootstrapping to construct uncertainty levels for 174 countries out of 195 countries for which the UN IGME published estimates in 2012. This new method improves current practice for estimating the extent of data errors, as it takes into account the structure and (potentially poor) quality of the data. The researchers used 90% as the uncertainty level and categorized countries according to the likelihood of meeting the MDG 4 target.
Using these methods, the researchers found that in countries with high child mortality rates (40 or more deaths per 1,000 children in 1990), there was a lot of uncertainty (wide uncertainty intervals) about the levels and trends of death rates in children aged under five years, especially more recently, because of the limited availability of data. Overall, in 2011 the median width of the uncertainty interval for the child death rate was 48% among the 86 countries with high death rates, compared to 19% in 1990. Using their new method, the researchers found that for eight countries, it is not clear whether any progress had been made in reducing child mortality, but for 13 countries, it is deemed likely that progress exceeded the MDG 4 target.
What Do These Findings Mean?
These findings suggest that new uncertainty assessments constructed by a statistical method called bootstrapping can provide more insights into countries' progress in reducing child mortality and meeting the MDG 4 target. As demonstrated in this study, when data are limited, uncertainty intervals should to be taken into account when estimating progress towards MDG 4 in order to give more accurate assessments on a country' progress, thus allowing for more realistic comparisons and conclusions.
Additional Information
Please access these websites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001355.
The UN website has more information about the Millennium Development Goals, including country-specific data
More information is available from UNICEF's ChildInfo website about the UN IGME and child mortality
All UN IGME child mortality estimates and data are available via CME Info
Countdown to 2015 tracks coverage levels for health interventions proven to reduce child mortality and proposes new actions to reach MDG 4
doi:10.1371/journal.pmed.1001355
PMCID: PMC3519895  PMID: 23239945
3.  Inequities in maternal and child health outcomes and interventions in Ghana 
BMC Public Health  2012;12:252.
Background
With the date for achieving the targets of the Millennium Development Goals (MDGs) approaching fast, there is a heightened concern about equity, as inequities hamper progress towards the MDGs. Equity-focused approaches have the potential to accelerate the progress towards achieving the health-related MDGs faster than the current pace in a more cost-effective and sustainable manner. Ghana's rate of progress towards MDGs 4 and 5 related to reducing child and maternal mortality respectively is less than what is required to achieve the targets. The objective of this paper is to examine the equity dimension of child and maternal health outcomes and interventions using Ghana as a case study.
Methods
Data from Ghana Demographic and Health Survey 2008 report is analyzed for inequities in selected maternal and child health outcomes and interventions using population-weighted, regression-based measures: slope index of inequality and relative index of inequality.
Results
No statistically significant inequities are observed in infant and under-five mortality, perinatal mortality, wasting and acute respiratory infection in children. However, stunting, underweight in under-five children, anaemia in children and women, childhood diarrhoea and underweight in women (BMI < 18.5) show inequities that are to the disadvantage of the poorest. The rates significantly decrease among the wealthiest quintile as compared to the poorest. In contrast, overweight (BMI 25-29.9) and obesity (BMI ≥ 30) among women reveals a different trend - there are inequities in favour of the poorest. In other words, in Ghana overweight and obesity increase significantly among women in the wealthiest quintile compared to the poorest. With respect to interventions: treatment of diarrhoea in children, receiving all basic vaccines among children and sleeping under ITN (children and pregnant women) have no wealth-related gradient. Skilled care at birth, deliveries in a health facility (both public and private), caesarean section, use of modern contraceptives and intermittent preventive treatment for malaria during pregnancy all indicate gradients that are in favour of the wealthiest. The poorest use less of these interventions. Not unexpectedly, there is more use of home delivery among women of the poorest quintile.
Conclusion
Significant Inequities are observed in many of the selected child and maternal health outcomes and interventions. Failure to address these inequities vigorously is likely to lead to non-achievement of the MDG targets related to improving child and maternal health (MDGs 4 and 5). The government should therefore give due attention to tackling inequities in health outcomes and use of interventions by implementing equity-enhancing measure both within and outside the health sector in line with the principles of Primary Health Care and the recommendations of the WHO Commission on Social Determinants of Health.
doi:10.1186/1471-2458-12-252
PMCID: PMC3338377  PMID: 22463465
4.  The social context of adolescent women’s use of modern contraceptives in Zimbabwe: a multilevel analysis 
Reproductive Health  2014;11:64.
Background
Efforts aimed at reducing maternal mortality as per the Millennium Development Goal 5 (MDG 5) include reducing early childbearing through increased adolescent contraceptive use. Despite a substantial attempt to study factors influencing adolescent contraceptive use in Sub-Saharan Africa (SSA), few studies have explored the role of community level characteristics on adolescent modern contraceptive use. This study examines the influence of both individual, household and community variables in influencing adolescent contraceptive use in Zimbabwe. This study posits that community characteristics are more critical predictors of adolescent contraceptive use in Zimbabwe than other individual and household characteristics.
Methods
Data from the 2010/11 Zimbabwe Demographic Health Survey (ZDHS), supplemented by additional data from the Measure DHS consultants were used. A total weighted sample of 457 non-pregnant adolescent women aged 15 to 19 years who had their last sex within 12 months preceding the 2010/11 ZDHS was analysed. Univariate, bivariate and multilevel binary logistic regression analysis were performed using generalized linear mixed models (GLMM).
Results
The odds of contraceptive use were higher for adolescent women with one or more children ever born (Odds Ratio (OR), 13.6) and for those ever married (OR, 2.5). Having medium and high access to media also increased the odds of using contraceptives (OR, 1.8; 2.1 respectively). At community level, the odds of modern contraceptive use decreased with an increase in the mean number of children ever borne per woman (OR, 0.071), an increase in the mean number of school years per women (OR, 0.4) and an increase in the proportion of women with at least secondary education (OR, 0.5). It however increased with an increase in the proportion of women experiencing at least one problem accessing health care (OR, 2.0). Individual and community level variables considered successfully explained the variation of adolescent contraceptive use across provinces.
Conclusions
Both individual and community characteristics were important predictors of adolescent contraceptive use in Zimbabwe. Reproductive program interventions aimed at increasing adolescent contraceptive use should take into account both individual and community factors. There is need for further research that examines other community characteristics influences that include political and cultural factors.
doi:10.1186/1742-4755-11-64
PMCID: PMC4134336  PMID: 25108444
Contraceptive use; Adolescent women; Socioeconomic development; Multilevel analysis; Zimbabwe
5.  Child Mortality Estimation: Appropriate Time Periods for Child Mortality Estimates from Full Birth Histories 
PLoS Medicine  2012;9(8):e1001289.
Jon Pedersen and Jing Liu examine the feasibility and potential advantages of using one-year rather than five-year time periods along with calendar year-based estimation when deriving estimates of child mortality.
Background
Child mortality estimates from complete birth histories from Demographic and Health Surveys (DHS) surveys and similar surveys are a chief source of data used to track Millennium Development Goal 4, which aims for a reduction of under-five mortality by two-thirds between 1990 and 2015. Based on the expected sample sizes when the DHS program commenced, the estimates are usually based on 5-y time periods. Recent surveys have had larger sample sizes than early surveys, and here we aimed to explore the benefits of using shorter time periods than 5 y for estimation. We also explore the benefit of changing the estimation procedure from being based on years before the survey, i.e., measured with reference to the date of the interview for each woman, to being based on calendar years.
Methods and Findings
Jackknife variance estimation was used to calculate standard errors for 207 DHS surveys in order to explore to what extent the large samples in recent surveys can be used to produce estimates based on 1-, 2-, 3-, 4-, and 5-y periods. We also recalculated the estimates for the surveys into calendar-year-based estimates. We demonstrate that estimation for 1-y periods is indeed possible for many recent surveys.
Conclusions
The reduction in bias achieved using 1-y periods and calendar-year-based estimation is worthwhile in some cases. In particular, it allows tracking of the effects of particular events such as droughts, epidemics, or conflict on child mortality in a way not possible with previous estimation procedures. Recommendations to use estimation for short time periods when possible and to use calendar-year-based estimation were adopted in the United Nations 2011 estimates of child mortality.
Editors' Summary
Background
In 2000, world leaders set, as Millennium Development Goal 4 (MDG 4), a target of reducing global under-five mortality (the number of children who die before their fifth birthday to a third of its 1990 level (12 million deaths per year) by 2015. (The MDGs are designed to alleviate extreme poverty by 2015.) To track progress towards MDG 4, the under-five mortality rate (also shown as 5q0) needs to be estimated both “precisely” and “accurately.” A “precise” estimate has a small random error (a quality indicated by a statistical measurement called the coefficient of variance), and an “accurate” estimate is one that is close to the true value because it lacks bias (systematic errors). In an ideal world, under-five mortality estimates would be based on official records of births and deaths. However, developing countries, which are where most under-five deaths occur, rarely have such records, and under-five mortality estimation relies on “complete birth histories” provided by women via surveys. These are collected by Demographic and Health Surveys (DHS, a project that helps developing countries collect data on health and population trends) and record all the births that a surveyed woman has had and the age at death of any of her children who have died.
Why Was This Study Done?
Because the DHS originally surveyed samples of 5,000–6,000 women, estimates of under-five mortality are traditionally calculated using data from five-year time periods. Over shorter periods with this sample size, the random errors in under-five mortality estimates become unacceptably large. Nowadays, the average DHS survey sample size is more than 10,000 women, so it should be possible to estimate under-five mortality over shorter time periods. Such estimates should be able to track the effects on under-five mortality of events such as droughts and conflicts better than estimates made over five years. In this study, the researchers determine appropriate time periods for child mortality estimates based on full birth histories, given different sample sizes. Specifically, they ask whether, with the bigger sample sizes that are now available, details about trends in under-five mortality rates are being missed by using the estimation procedures that were developed for smaller samples. They also ask whether calendar-year-based estimates can be calculated; mortality is usually estimated in “years before the survey,” a process that blurs the reference period for the estimate.
What Did the Researchers Do and Find?
The researchers used a statistical method called “jackknife variance estimation” to determine coefficients of variance for child mortality estimates calculated over different time periods using complete birth histories from 207 DHS surveys. Regardless of the estimation period, half of the estimates had a coefficient of variance of less than 10%, a level of random variation that is generally considered acceptable. However, within each time period, some estimates had very high coefficients of variance. These estimates were derived from surveys where there was a small sample size, low fertility (the women surveyed had relatively few babies), or low child mortality. Other analyses show that although the five-year period estimates had lower standard errors than the one-year period estimates, the latter were affected less by bias than the five-year period estimates. Finally, estimates fixed to calendar years rather than to years before the survey were more directly comparable across surveys and brought out variations in child mortality caused by specific events such as conflicts more clearly.
What Do These Findings Mean?
These findings show that although under-five mortality rate estimates based on five-year periods of data have been the norm, the sample sizes currently employed in DHS surveys make it feasible to estimate mortality for shorter periods. The findings also show that using shorter periods of data in estimations of the under-five mortality rate, and using calendar-year-based estimation, reduces bias (makes the estimations more accurate) and allows the effects of events such as droughts, epidemics, or conflict on under-five mortality rates to be tracked in a way that is impossible when using five-year periods of data. Given these findings, the researchers recommend that time periods shorter than five years should be adopted for the estimation of under-five mortality and that estimations should be pegged to calendar years rather than to years before the survey. Both recommendations have already been adopted by the United Nations Inter-agency Group for Child Mortality Estimation (IGME) and were used in their 2011 analysis of under-five mortality.
Additional Information
Please access these websites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001289.
This paper is part of a collection of papers on Child Mortality Estimation Methods published in PLOS Medicine
The United Nations Childrens Fund (UNICEF) works for children's rights, survival, development, and protection around the world; it provides information on Millennium Development Goal 4, and its Childinfo website provides detailed statistics about child survival and health, including a description of the United Nations Inter-agency Group for Child Mortality Estimation; the 2011 IGME report on Levels and Trends in Child Mortality is available
The World Health Organization also has information about Millennium Development Goal 4 and provides estimates of child mortality rates (some information in several languages)
Further information about the Millennium Development Goals is available
Information is also available about Demographic and Health Surveys of infant and child mortality
doi:10.1371/journal.pmed.1001289
PMCID: PMC3429388  PMID: 22952435
6.  Rank Order Entropy: why one metric is not enough 
The use of Quantitative Structure-Activity Relationship models to address problems in drug discovery has a mixed history, generally resulting from the mis-application of QSAR models that were either poorly constructed or used outside of their domains of applicability. This situation has motivated the development of a variety of model performance metrics (r2, PRESS r2, F-tests, etc) designed to increase user confidence in the validity of QSAR predictions. In a typical workflow scenario, QSAR models are created and validated on training sets of molecules using metrics such as Leave-One-Out or many-fold cross-validation methods that attempt to assess their internal consistency. However, few current validation methods are designed to directly address the stability of QSAR predictions in response to changes in the information content of the training set. Since the main purpose of QSAR is to quickly and accurately estimate a property of interest for an untested set of molecules, it makes sense to have a means at hand to correctly set user expectations of model performance. In fact, the numerical value of a molecular prediction is often less important to the end user than knowing the rank order of that set of molecules according to their predicted endpoint values. Consequently, a means for characterizing the stability of predicted rank order is an important component of predictive QSAR. Unfortunately, none of the many validation metrics currently available directly measure the stability of rank order prediction, making the development of an additional metric that can quantify model stability a high priority. To address this need, this work examines the stabilities of QSAR rank order models created from representative data sets, descriptor sets, and modeling methods that were then assessed using Kendall Tau as a rank order metric, upon which the Shannon Entropy was evaluated as a means of quantifying rank-order stability. Random removal of data from the training set, also known as Data Truncation Analysis (DTA), was used as a means for systematically reducing the information content of each training set while examining both rank order performance and rank order stability in the face of training set data loss. The premise for DTA ROE model evaluation is that the response of a model to incremental loss of training information will be indicative of the quality and sufficiency of its training set, learning method, and descriptor types to cover a particular domain of applicability.
This process is termed a “rank order entropy” evaluation, or ROE. By analogy with information theory, an unstable rank order model displays a high level of implicit entropy, while a QSAR rank order model which remains nearly unchanged during training set reductions would show low entropy. In this work, the ROE metric was applied to 71 data sets of different sizes, and was found to reveal more information about the behavior of the models than traditional metrics alone. Stable, or consistently performing models, did not necessarily predict rank order well. Models that performed well in rank order did not necessarily perform well in traditional metrics. In the end, it was shown that ROE metrics suggested that some QSAR models that are typically used should be discarded. ROE evaluation helps to discern which combinations of data set, descriptor set, and modeling methods lead to usable models in prioritization schemes, and provides confidence in the use of a particular model within a specific domain of applicability.
doi:10.1021/ci200170k
PMCID: PMC3428235  PMID: 21875058
7.  Child Mortality Estimation: A Comparison of UN IGME and IHME Estimates of Levels and Trends in Under-Five Mortality Rates and Deaths 
PLoS Medicine  2012;9(8):e1001288.
Leontine Alkema and Danzhen You compare and summarize differences in underlying data and modelling approaches used by two key groups who publish data on global under-5 mortality rates
Background
Millennium Development Goal 4 calls for a reduction in the under-five mortality rate (U5MR) by two-thirds between 1990 and 2015. In 2011, estimates were published by the United Nations Inter-agency Group for Child Mortality Estimation (UN IGME) and the Institute for Health Metrics and Evaluation (IHME). The difference in the U5MR estimates produced by the two research groups was more than 10% and corresponded to more than ten deaths per 1,000 live births for 10% of all countries in 1990 and 20% of all countries in 2010, which can lead to conflicting conclusions with respect to countries' progress. To understand what caused the differences in estimates, we summarised differences in underlying data and modelling approaches used by the two groups, and analysed their effects.
Methods and Findings
UN IGME and IHME estimation approaches differ with respect to the construction of databases and the pre-processing of data, trend fitting procedures, inclusion and exclusion of data series, and additional adjustment procedures. Large differences in U5MR estimates between the UN IGME and the IHME exist in countries with conflicts or civil unrest, countries with high HIV prevalence, and countries where the underlying data used to derive the estimates were different, especially if the exclusion of data series differed between the two research groups. A decomposition of the differences showed that differences in estimates due to using different data (inclusion of data series and pre-processing of data) are on average larger than the differences due to using different trend fitting methods.
Conclusions
Substantial country-specific differences between UN IGME and IHME estimates for U5MR and the number of under-five deaths exist because of various differences in data and modelling assumptions used. Often differences are illustrative of the lack of reliable data and likely to decrease as more data become available. Improved transparency on methods and data used will help to improve understanding about the drivers of the differences.
Please see later in the article for the Editors' Summary.
Editors' Summary
Background
In 2010, more than seven million children died before they reached their fifth birthday, and the global under-five mortality rate (also denoted in the literature as U5MR and 5q0) was 57 deaths per 1,000 live births. Most deaths before the age of five years occur in developing countries (about half occur in just five countries—India, Nigeria, the Democratic Republic of the Congo, Pakistan, and China), and most are caused by preventable or treatable diseases such as pneumonia, diarrhea, and malaria. Faced with this largely avoidable loss of young lives, in 1990, the United Nations (UN) World Summit for Children pledged to improve the survival of children. Later, in 2000, world leaders set a target of reducing under-five mortality to one-third of its 1990 level (12 million) by 2015, as Millennium Development Goal 4 (MDG 4). This goal, together with seven others, is designed to improve the social, economic, and health conditions in the world's poorest countries.
Why Was This Study Done?
Although progress towards MDG 4 is accelerating, MDG 4 is unlikely to be reached. It is important, therefore, to know which countries are making poor progress towards MDG 4 so that extra resources can be concentrated in these areas. To monitor both national and global progress, accurate, up-to-date estimates of U5MR are essential. The first step in estimating U5MR is the collection of data on child deaths, usually through vital registration systems (which record all births and deaths) in developed countries and through surveys that ask women about their living and dead children in developing countries. Country-specific U5MR estimates that are comparable over time and across countries are obtained from these data using a statistical process called trend fitting. Two groups—the UN Inter-agency Group for Child Mortality Estimation (UN IGME) and the Institute for Health Metrics and Evaluation (IHME)—recently published new estimates of the levels and trends in U5MR and under-five deaths across the world. However, their estimates differ somewhat and, for some countries, disagree on the progress being made towards MDG 4. Here, the researchers examine the differences in the underlying data and the trend fitting approaches used by the UN IGME and the IHME to try to understand why their estimates are different.
What Did the Researchers Do and Find?
The researchers first compared the estimates produced by the two groups. From 1990 to 2010, the UN IGME's global estimates of U5MR and under-five deaths were consistently slightly higher than those of the IHME. For example, in 2010, the UN IGME and IMHE estimates of U5MR were 56.7 and 53.9 deaths per 1,000 births, respectively. However, although the global estimates from the two groups were broadly similar, there were important differences between the two sets of estimates at the country level, particularly in countries where there was conflict or civil unrest (for example, Somalia) or high HIV prevalence. The researchers then examined the data used by the two groups to estimate under-five deaths and U5MR, the method used for U5MR trend fitting, and additional adjustment procedures (for example, the UN IGME incorporates feedback from experts and country consultations in its estimates). The UN IGME and IHME estimation approaches included differences in all of these areas, but differences in the data used caused on average larger differences in the estimates than the use of different trend fitting methods did.
What Do These Findings Mean?
These findings show that the substantial country-specific differences between UN IGME and IHME estimates for U5MR and the number of under-five deaths are the result of several differences between the data and trend fitting methods used by the two groups. In particular, the findings indicate that the lack of reliable data in many developing countries, especially those where there is civil unrest or ongoing conflicts, is often responsible for differences in estimates. These differences should, therefore, decrease as more reliable data become available. For now, though, the differences between the UN IGME and IHME national estimates of child mortality may cause confusion about the true extent of progress towards MDG 4 and could foster policy inactivity if the reasons for the discrepancies are not made clear. The researchers call, therefore, for more transparency on the methods and data used in the estimation of U5MR and for a concerted effort by governments, UN agencies, and non-governmental organizations to improve the collection of reliable data on child deaths.
Additional Information
Please access these websites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001288.
This paper is part of a collection of papers on Child Mortality Estimation Methods published in PLOS Medicine
The United Nations Childrens Fund (UNICEF) works for children's rights, survival, development, and protection around the world; it provides information on Millennium Development Goal 4, and its Childinfo website provides detailed statistics about child survival and health, including a description of the UN Inter-agency Group for Child Mortality Estimation and a link to its database; the 2011 UN IGME report on Levels and Trends in Child Mortality is available
The Institute for Health Metrics and Evaluation website includes a summary of their 2011 analysis of U5MR and under-five deaths
The World Health Organization also has information about Millennium Development Goal 4 and provides estimates of child mortality rates (some information in several languages)
Further information about the Millennium Development Goals is available
doi:10.1371/journal.pmed.1001288
PMCID: PMC3429386  PMID: 22952434
8.  Child Mortality Estimation: Estimating Sex Differences in Childhood Mortality since the 1970s 
PLoS Medicine  2012;9(8):e1001287.
Cheryl Sawyer uses new methods to generate estimates of sex differences in child mortality which can be used to pinpoint areas where these differences in mortality merit closer examination.
Introduction
Producing estimates of infant (under age 1 y), child (age 1–4 y), and under-five (under age 5 y) mortality rates disaggregated by sex is complicated by problems with data quality and availability. Interpretation of sex differences requires nuanced analysis: girls have a biological advantage against many causes of death that may be eroded if they are disadvantaged in access to resources. Earlier studies found that girls in some regions were not experiencing the survival advantage expected at given levels of mortality. In this paper I generate new estimates of sex differences for the 1970s to the 2000s.
Methods and Findings
Simple fitting methods were applied to male-to-female ratios of infant and under-five mortality rates from vital registration, surveys, and censuses. The sex ratio estimates were used to disaggregate published series of both-sexes mortality rates that were based on a larger number of sources. In many developing countries, I found that sex ratios of mortality have changed in the same direction as historically occurred in developed countries, but typically had a lower degree of female advantage for a given level of mortality. Regional average sex ratios weighted by numbers of births were found to be highly influenced by China and India, the only countries where both infant mortality and overall under-five mortality were estimated to be higher for girls than for boys in the 2000s. For the less developed regions (comprising Africa, Asia excluding Japan, Latin America/Caribbean, and Oceania excluding Australia and New Zealand), on average, boys' under-five mortality in the 2000s was about 2% higher than girls'. A number of countries were found to still experience higher mortality for girls than boys in the 1–4-y age group, with concentrations in southern Asia, northern Africa/western Asia, and western Africa. In the more developed regions (comprising Europe, northern America, Japan, Australia, and New Zealand), I found that the sex ratio of infant mortality peaked in the 1970s or 1980s and declined thereafter.
Conclusions
The methods developed here pinpoint regions and countries where sex differences in mortality merit closer examination to ensure that both sexes are sharing equally in access to health resources. Further study of the distribution of causes of death in different settings will aid the interpretation of differences in survival for boys and girls.
Please see later in the article for the Editors' Summary.
Editors' Summary
Background
In 2000, world leaders agreed to eradicate extreme poverty by 2015. To help track progress towards this global commitment, eight Millennium Development Goals (MDGs) were set. MDG 4, which aims to reduce child mortality, calls for a reduction in under-five mortality (the number of children who die before their fifth birthday) to a third of its 1990 level of 12 million by 2015. The under-five mortality rate is also denoted in the literature as U5MR and 5q0. Progress towards MDG 4 has been substantial, but with only three years left to reach it, efforts to strengthen child survival programs are intensifying. Reliable estimates of trends in childhood mortality are pivotal to these efforts. So, since 2004, the United Nations Inter-agency Group for Child Mortality Estimation (UN IGME) has used statistical regression models to produce estimates of trends in under-five mortality and infant mortality (death before age one year) from data about childbearing and child survival collected by vital registration systems (records of all births and deaths), household surveys, and censuses.
Why Was This Study Done?
In addition to estimates of overall childhood mortality trends, information about sex-specific childhood mortality trends is desirable to monitor progress towards MDG 4, although the interpretation of trends in the relative mortality of girls and boys is not straightforward. Newborn girls survive better than newborn boys because they are less vulnerable to birth complications and infections and have fewer inherited abnormalities. Thus, the ratio of infant mortality among boys to infant mortality among girls is greater than one, provided both sexes have equal access to food and medical care. Beyond early infancy, girls and boys are similarly vulnerable to infections, so the sex ratio of deaths in the 1–4-year age group is generally lower than that of infant mortality. Notably, as living conditions improve in developing countries, infectious diseases become less important as causes of death. Thus, in the absence of sex-specific differences in the treatment of children, the sex ratio of childhood mortality is expected be greater than one and to increase as overall under-five mortality rates in developing countries decrease. In this study, the researcher evaluated national and regional changes in the sex ratios of childhood mortality since the 1970s to investigate whether girls and boys have equal access to medical care and other resources.
What Did the Researcher Do and Find?
The researcher developed new statistical fitting methods to estimate trends in the sex ratio of mortality for infants and young children for individual countries and world regions. When considering individual countries, the researcher found that for 92 countries in less developed regions, the median sex ratio of under-five mortality increased between the 1970s and the 2000s, in line with the expected changes just described. However, the average sex ratio of under-five mortality for less developed regions, weighted according to the number of births in each country, did not increase between the 1970s and 2000s, at which time the average under-five mortality rate of boys was about 2% higher than that of girls. This discrepancy resulted from India and China—the two most populous developing countries—having sex ratios for both infant and under-five mortality that remained constant or declined over the study period and were below one in the 2000s, a result that indicates excess female mortality. In China, for example, infant mortality was found to be 12% higher for boys than for girls in the 1970s, but 24% lower for boys than for girls in the 2000s. Finally, although in the less developed regions (excluding India and China) girls went from having a slight survival disadvantage at ages 1–4 years in the 1970s, on average, to having a slight advantage in the 2000s, girls remained more likely to die than boys in this age group in several Asian and African countries.
What Do These Findings Mean?
Although the quality of the available data is likely to affect the accuracy of these findings, in most developing countries the ratio of male to female under-five mortality has increased since the 1970s, in parallel with the decrease in overall childhood mortality. Notably, however, in a number of developing countries—including several each in sub-Saharan Africa, northern Africa/western Asia, and southern Asia—girls have higher mortality than boys at ages 1–4 years, and in India and China girls have higher mortality in infancy. Thus, girls are benefitting less than boys from the overall decline in childhood mortality in India, China, and some other developing countries. Further studies are needed to determine the underlying reasons for this observation. Nevertheless, the methods developed here to estimate trends in sex-specific childhood mortality pinpoint countries and regions where greater efforts should be made to ensure that both sexes have equal access to health care and other important resources during early life.
Additional Information
Please access these websites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001287.
This paper is part of a collection of papers on Child Mortality Estimation Methods published in PLOS Medicine
The United Nations Childrens Fund works for children's rights, survival, development, and protection around the world; it provides information on Millennium Development Goal 4, and its Childinfo website provides detailed statistics about child survival and health, including a description of the United Nations Inter-agency Group for Child Mortality Estimation; the 2011 UN IGME report Levels & Trends in Child Mortality is available
The World Health Organization also has information about Millennium Development Goal 4 and provides estimates of child mortality rates (some information in several languages)
Further information about the Millennium Development Goals is available
A 2011 report by the United Nations Department of Economic and Social Affairs entitled Sex Differentials in Childhood Mortality is available
doi:10.1371/journal.pmed.1001287
PMCID: PMC3429399  PMID: 22952433
9.  Back to the future: what would the post-2015 global development goals look like if we replicated methods used to construct the Millennium Development Goals? 
Background
The Millennium Development Goals (MDGs) were ‘top-down’ goals formulated by policy elites drawing from targets within United Nations (UN) summits and conferences in the 1990s. Contemporary processes shaping the new post-2015 development agenda are more collaborative and participatory, markedly different to the pre-MDG era. This study examines what would the outcome be if a methodology similar to that used for the MDGs were applied to the formulation of the post-2015 development goals (Post-2015DGs), identifying those targets arising from UN summits and conferences since the declaration of the MDGs, and aggregating them into goals.
Methods
The UN Department of Economic and Social Affairs (DESA) list of major UN summits and conferences from 2001 to 2012 was utilised to examine targets. The DESA list was chosen due to the agency’s core mission to promote development for all. Targets meeting MDG criteria of clarity, conciseness and measurability were selected and clustered into broad goals based on processes outlined by Hulme and Vandemoortele. The Post-2015DGs that were identified were formatted into language congruent with the MDGs to assist in the comparative analysis, and then further compared to the 12 illustrative goals offered by the UN High-Level Panel of Eminent Persons on the Post-2015 Development (High-Level Panel) Agenda’s May 2013 report.
Results
Ten Post-2015DGs were identified. Six goals expressly overlapped with the current MDGs and four new goals were identified. Health featured prominently in the MDG agenda, and continues to feature strongly in four of the 10 Post-2015DGs. However the Post-2015DGs reposition health within umbrella agendas relating to women, children and the ageing. Six of the 10 Post-2015DGs incorporate the right to health agenda, emphasising both the standing and interconnection of the health agenda in DESA’s summits and conferences under review. Two Post-2015DGs have been extended into six separate goals by the High-Level Panel, and it is these goals that are clearly linked to sustainable development diaspora.
Conclusions
This study exposes the evolving political agendas underplaying the current post-2015 process, as targets from DESA’s 22 major UN summits and conferences from 2001 to 2012 are not wholly mirrored in the HLP’s 12 goals.
doi:10.1186/1744-8603-10-19
PMCID: PMC4008441  PMID: 24708796
Millennium Development Goals; Global health policy; Post-2015 development agenda
10.  Measuring Coverage in MNCH: A Prospective Validation Study in Pakistan and Bangladesh on Measuring Correct Treatment of Childhood Pneumonia 
PLoS Medicine  2013;10(5):e1001422.
Background
Antibiotic treatment for pneumonia as measured by Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) is a key indicator for tracking progress in achieving Millennium Development Goal 4. Concerns about the validity of this indicator led us to perform an evaluation in urban and rural settings in Pakistan and Bangladesh.
Methods and Findings
Caregivers of 950 children under 5 y with pneumonia and 980 with “no pneumonia” were identified in urban and rural settings and allocated for DHS/MICS questions 2 or 4 wk later. Study physicians assigned a diagnosis of pneumonia as reference standard; the predictive ability of DHS/MICS questions and additional measurement tools to identify pneumonia versus non-pneumonia cases was evaluated.
Results at both sites showed suboptimal discriminative power, with no difference between 2- or 4-wk recall. Individual patterns of sensitivity and specificity varied substantially across study sites (sensitivity 66.9% and 45.5%, and specificity 68.8% and 69.5%, for DHS in Pakistan and Bangladesh, respectively). Prescribed antibiotics for pneumonia were correctly recalled by about two-thirds of caregivers using DHS questions, increasing to 72% and 82% in Pakistan and Bangladesh, respectively, using a drug chart and detailed enquiry.
Conclusions
Monitoring antibiotic treatment of pneumonia is essential for national and global programs. Current (DHS/MICS questions) and proposed new (video and pneumonia score) methods of identifying pneumonia based on maternal recall discriminate poorly between pneumonia and children with cough. Furthermore, these methods have a low yield to identify children who have true pneumonia. Reported antibiotic treatment rates among these children are therefore not a valid proxy indicator of pneumonia treatment rates. These results have important implications for program monitoring and suggest that data in its current format from DHS/MICS surveys should not be used for the purpose of monitoring antibiotic treatment rates in children with pneumonia at the present time.
Please see later in the article for the Editors' Summary
Editors' Summary
Background
Pneumonia is a major cause of death in children younger than five years across the globe, with approximately 1.2 million children younger than five years dying from pneumonia every year. Pneumonia can be caused by bacteria, fungi, or viruses. It is possible to effectively treat bacterial pneumonia with appropriate antibiotics; however, only about 30% of children receive the antibiotic treatment they need. The Millennium Development Goals (MDGs) are eight international development goals that were established in 2000. The fourth goal (MDG 4) aims to reduce child mortality, specifically, to reduce the under-five mortality rate by two-thirds, between 1990 and 2015. Given that approximately 18% of all deaths in children under five are caused by pneumonia, providing universal coverage with effective treatments for pneumonia is an important part of MDG 4.
To ensure that MDG 4 targets are met, it is important to measure progress in providing effective treatments. For pneumonia, one of the key indicators for measuring progress is the proportion of children with pneumonia in a population who receive antibiotic treatment, also known as the antibiotic treatment rate. The antibiotic treatment rate is often measured using surveys, such as the Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS), which collect nationally representative data about populations and health in developing countries.
Why Was This Study Done?
Concerns have been raised about whether information collected from DHS and MICS is able to accurately identify cases of pneumonia. In a clinical setting, pneumonia is typically diagnosed based on a combination of physical symptoms, including coughing, rapid breathing, or difficulty breathing, and a chest X-ray. The surveys rely on information collected from interviews of mothers and primary caregivers using structured questions about whether the child has experienced physical symptoms in the past two weeks and whether these were chest-related. The DHS survey labels this condition as “symptoms of acute respiratory infection,” while the MICS survey uses the term “suspected pneumonia.” Thus, these surveys provide a proxy measure for pneumonia that is limited by the reliance on the recall of symptoms by the mother or caregiver. Here the researchers have evaluated the use of these surveys to discriminate physician-diagnosed pneumonia and to provide accurate recall of antibiotic treatment in urban and rural settings in Pakistan and Bangladesh.
What Did the Researchers Do and Find?
The researchers identified caregivers of 950 children under five years with pneumonia and 980 who had a cough or cold but did not have pneumonia from urban and rural settings in Pakistan and Bangladesh. Cases of pneumonia were identified based on a physician diagnosis using World Health Organization guidelines. They randomly assigned caregivers to be interviewed using DHS and MICS questions with either a two- or four-week recall period. They then assessed how well the DHS and MICS questions were able to accurately diagnose pneumonia and accurately recall antibiotic use. In addition, they asked caregivers to complete a pneumonia score questionnaire and showed them a video tool showing children with and without pneumonia, as well as a medication drug chart, to determine if these alternative measures improved the accuracy of pneumonia diagnosis or recall of antibiotic use. They found that both surveys, the pneumonia score, and the video tool had poor ability to discriminate between children with and without physician-diagnosed pneumonia, and there were no differences between using two- or four-week recall. The sensitivity (proportion of pneumonia cases that were correctly identified) ranged from 23% to 72%, and the specificity (the proportion of “no pneumonia” cases that were correctly identified) ranged from 53% to 83%, depending on the setting. They also observed that prescribed antibiotics for pneumonia were correctly recalled by about two-thirds of caregivers using DHS questions, and this increased to about three-quarters of caregivers when using a drug chart and detailed enquiry.
What Do These Findings Mean?
The findings of this study suggest that the current use of questions from DHS and MICS based on mother or caregiver recall are not sufficient for accurately identifying pneumonia and antibiotic use in children. Because these surveys have poor ability to identify children who have true pneumonia, reported antibiotic treatment rates for children with pneumonia based on data from these surveys may not be accurate, and these surveys should not be used to monitor treatment rates. These findings should be interpreted cautiously, given the relatively high rate of loss to follow-up and delayed follow-up in some of the children and because some of the settings in this study may not be similar to other low-income settings.
Additional Information
Please access these websites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001422.
More information is available on the United Nations goal to reduce child mortality (MDG 4)
The World Health Organization provides information on pneumonia, its impact on children, and the global action plan for prevention and control of pneumonia
More information is available on Demographic and Health Surveys and Multiple Indicator Cluster Surveys
KidsHealth, a resource maintained by the Nemours Foundation (a not-for-profit organization for children's health) provides information for parents on pneumonia (in English and Spanish)
MedlinePlus provides links to additional information on pneumonia (in English and Spanish)
doi:10.1371/journal.pmed.1001422
PMCID: PMC3646205  PMID: 23667339
11.  Stability of variable importance scores and rankings using statistical learning tools on single-nucleotide polymorphisms and risk factors involved in gene × gene and gene × environment interactions 
BMC Proceedings  2007;1(Suppl 1):S58.
Risk of complex disorders is thought to be multifactorial, involving interactions between risk factors. However, many genetic studies assess association between disease status and markers one single-nucleotide polymorphism (SNP) at a time, due to the high-dimensional nature of the search space of all possible interactions. Three ensemble methods have been recently proposed for use in high-dimensional data (Monte Carlo logic regression, random forests, and generalized boosted regression). An intuitive way to detect an association between genetic markers and disease status is to use variable importance measures, even though the stability of these measures in the context of a whole-genome association study is unknown. For the simulated data of Problem 3 in the Genetic Analysis Workshop 15 (GAW15), we examined the variability of both rankings and magnitude of variable importance measures using 10 variables simulated to participate in gene × gene and gene × environment interactions. We conducted 500 analyses per method on one randomly selected replicate, tallying the rankings and importance measures for each of the 10 variables of interest. When the simulated effect size was strong, all three methods showed stable rankings and estimates of variable importance. However, under conditions more commonly expected to be encountered in complex diseases, random forests and generalized boosted regression showed more stable estimates of variable importance and variable rankings. Individuals endeavoring to apply statistical learning methods to detect interaction in complex disease studies should perform repeated analyses in order to assure variable importance measures and rankings do not vary greatly, even for statistical learning algorithms that are thought to be stable.
PMCID: PMC2367584  PMID: 18466558
12.  Alternative Strategies to Reduce Maternal Mortality in India: A Cost-Effectiveness Analysis 
PLoS Medicine  2010;7(4):e1000264.
A cost-effectiveness study by Sue Goldie and colleagues finds that better family planning, provision of safe abortion, and improved intrapartum and emergency obstetrical care could reduce maternal mortality in India by 75% in 5 years.
Background
Approximately one-quarter of all pregnancy- and delivery-related maternal deaths worldwide occur in India. Taking into account the costs, feasibility, and operational complexity of alternative interventions, we estimate the clinical and population-level benefits associated with strategies to improve the safety of pregnancy and childbirth in India.
Methods and Findings
Country- and region-specific data were synthesized using a computer-based model that simulates the natural history of pregnancy (both planned and unintended) and pregnancy- and childbirth-associated complications in individual women; and considers delivery location, attendant, and facility level. Model outcomes included clinical events, population measures, costs, and cost-effectiveness ratios. Separate models were adapted to urban and rural India using survey-based data (e.g., unmet need for birth spacing/limiting, facility births, skilled birth attendants). Model validation compared projected maternal indicators with empiric data. Strategies consisted of improving coverage of effective interventions that could be provided individually or packaged as integrated services, could reduce the incidence of a complication or its case fatality rate, and could include improved logistics such as reliable transport to an appropriate referral facility as well as recognition of referral need and quality of care. Increasing family planning was the most effective individual intervention to reduce pregnancy-related mortality. If over the next 5 y the unmet need for spacing and limiting births was met, more than 150,000 maternal deaths would be prevented; more than US$1 billion saved; and at least one of every two abortion-related deaths averted. Still, reductions in maternal mortality reached a threshold (∼23%–35%) without including strategies that ensured reliable access to intrapartum and emergency obstetrical care (EmOC). An integrated and stepwise approach was identified that would ultimately prevent four of five maternal deaths; this approach coupled stepwise improvements in family planning and safe abortion with consecutively implemented strategies that incrementally increased skilled attendants, improved antenatal/postpartum care, shifted births away from home, and improved recognition of referral need, transport, and availability/quality of EmOC. The strategies in this approach ranged from being cost-saving to having incremental cost-effectiveness ratios less than US$500 per year of life saved (YLS), well below India's per capita gross domestic product (GDP), a common benchmark for cost-effectiveness.
Conclusions
Early intensive efforts to improve family planning and control of fertility choices and to provide safe abortion, accompanied by a paced systematic and stepwise effort to scale up capacity for integrated maternal health services over several years, is as cost-effective as childhood immunization or treatment of malaria, tuberculosis, or HIV. In just 5 y, more than 150,000 maternal deaths would be averted through increasing contraception rates to meet women's needs for spacing and limiting births; nearly US$1.5 billion would be saved by coupling safe abortion to aggressive family planning efforts; and with stepwise investments to improve access to pregnancy-related health services and to high-quality facility-based intrapartum care, more than 75% of maternal deaths could be prevented. If accomplished over the next decade, the lives of more than one million women would be saved.
Please see later in the article for the Editors' Summary
Editors' Summary
Background
Every year, more than half a million women—most of them living in developing countries—die from pregnancy- or childbirth-related complications. About a quarter of these “maternal” deaths occur in India. In 2005, a woman's lifetime risk of maternal death in India was 1 in 70; in the UK, it was only one in 8,200. Similarly, the maternal mortality ratio (MMR; number of maternal deaths per 100,000 live births) in India was 450, whereas in the UK it was eight. Faced with the enormous maternal death toll in India and other developing countries, in September 2000, the United Nations pledged, as its fifth Millennium Development Goal (MDG 5), that the global MMR would be reduced to a quarter of its 1990 level by 2015. Currently, it seems unlikely that this target will be met. Between 1990 and 2005, global maternal deaths decreased by only 1% per annum instead of the 5% needed to reach MDG 5; in India, the decrease in maternal deaths between 1990 and 2005 was about 1.8% per annum.
Why Was This Study Done?
Most maternal deaths in developing countries are caused by severe bleeding after childbirth, infections soon after delivery, blood pressure disorders during pregnancy, and obstructed (difficult) labors. Consequently, experts agree that universal access to high-quality routine care during labor (“obstetric” care) and to emergency obstetrical care is needed to reduce maternal deaths. However, there is less agreement about how to adapt these “ideal recommendations” to specific situations. In developing countries with weak health systems and predominantly rural populations, it is unlikely that all women will have access to emergency obstetric care in the near future—so would beginning with improved access to family planning and to safe abortions (unsafe abortion is another major cause of maternal death) be a more achievable, more cost-effective way of reducing maternal deaths? How would family planning and safe abortion be coupled efficiently and cost-effectively with improved access to intrapartum care? In this study, the researchers investigate these questions by estimating the health and economic outcomes of various strategies to reduce maternal mortality in India.
What Did the Researchers Do and Find?
The researchers used a computer-based model that simulates women through pregnancy and childbirth to estimate the effect of different strategies (for example, increased family planning or increased access to obstetric care) on clinical outcomes (pregnancies, live births, or deaths), costs, and cost-effectiveness (the cost of saving one year of life) in India. Increased family planning was the most effective single intervention for the reduction of pregnancy-related mortality. If the current unmet need for family planning in India could be fulfilled over the next 5 years, more than 150,000 maternal deaths would be prevented, more than US$1 billion saved, and at least half of abortion-related deaths averted. However, increased family planning alone would reduce maternal deaths by 35% at most, so the researchers also used their model to test the effect of combinations of strategies on maternal death. They found that an integrated and stepwise approach (increased family planning and safe abortion combined with consecutively increased skilled birth attendants, improved care before and after birth, reduced home births, and improved emergency obstetric care) could eventually prevent nearly 80% of maternal deaths. All the steps in this strategy either saved money or involved an additional cost per year of life saved of less than US$500; given one suggested threshold for cost-effectiveness in India of the per capita GDP (US$1,068) per year of life saved, these strategies would be considered very cost-effective.
What Do These Findings Mean?
The accuracy of these findings depends on the assumptions used to build the model and the quality of the data fed into it. Nevertheless, these findings suggest that early intensive efforts to improve family planning and to provide safe abortion accompanied by a systematic, stepwise effort to improve integrated maternal health services could reduce maternal deaths in India by more than 75% in less than a decade. Furthermore, such a strategy would be cost-effective. Indeed, note the researchers, the cost savings from an initial focus on family planning and safe abortion provision would partly offset the resources needed to assure that every woman had access to high quality routine and emergency obstetric care. Thus, overall, these findings suggest that MDG 5 may be within reach in India, a conclusion that should help to mobilize political support for this worthy goal.
Additional Information
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1000264.
UNICEF (the United Nations Children's Fund) provides information on maternal mortality, including the WHO/UNICEF/UNFPA/The World Bank 2005 country estimates of maternal mortality
The World Health Organization also provides information on maternal health and about MDG 5 (in several languages)
The United Nations Millennium Development Goals Web site provides detailed information about the Millennium Declaration, the MDGs, their targets and their indicators, and about MDG 5.
The Millennium Development Goals Report 2009 and its progress chart provide an up-to-date assessment of progress toward all the MDGs
Computer simulation modeling as applied to health is further discussed at the Center for Health Decision Science at Harvard University
doi:10.1371/journal.pmed.1000264
PMCID: PMC2857650  PMID: 20421922
13.  An AUC-based permutation variable importance measure for random forests 
BMC Bioinformatics  2013;14:119.
Background
The random forest (RF) method is a commonly used tool for classification with high dimensional data as well as for ranking candidate predictors based on the so-called random forest variable importance measures (VIMs). However the classification performance of RF is known to be suboptimal in case of strongly unbalanced data, i.e. data where response class sizes differ considerably. Suggestions were made to obtain better classification performance based either on sampling procedures or on cost sensitivity analyses. However to our knowledge the performance of the VIMs has not yet been examined in the case of unbalanced response classes. In this paper we explore the performance of the permutation VIM for unbalanced data settings and introduce an alternative permutation VIM based on the area under the curve (AUC) that is expected to be more robust towards class imbalance.
Results
We investigated the performance of the standard permutation VIM and of our novel AUC-based permutation VIM for different class imbalance levels using simulated data and real data. The results suggest that the new AUC-based permutation VIM outperforms the standard permutation VIM for unbalanced data settings while both permutation VIMs have equal performance for balanced data settings.
Conclusions
The standard permutation VIM loses its ability to discriminate between associated predictors and predictors not associated with the response for increasing class imbalance. It is outperformed by our new AUC-based permutation VIM for unbalanced data settings, while the performance of both VIMs is very similar in the case of balanced classes. The new AUC-based VIM is implemented in the R package party for the unbiased RF variant based on conditional inference trees. The codes implementing our study are available from the companion website: http://www.ibe.med.uni-muenchen.de/organisation/mitarbeiter/070_drittmittel/janitza/index.html.
doi:10.1186/1471-2105-14-119
PMCID: PMC3626572  PMID: 23560875
Random forest; Conditional inference trees; Variable importance measure; Feature selection; Unbalanced data; Class imbalance; Area under the curve.
14.  Non-Participation during Azithromycin Mass Treatment for Trachoma in The Gambia: Heterogeneity and Risk Factors 
Background
There is concern that untreated individuals in mass drug administration (MDA) programs for neglected tropical diseases can reduce the impact of elimination efforts by maintaining a source of transmission and re-infection.
Methodology/Principal Findings
Treatment receipt was recorded against the community census during three MDAs with azithromycin for trachoma in The Gambia, a hypo-endemic setting. Predictors of non-participation were investigated in 1–9 year olds using random effects logistic regression of cross-sectional data for each MDA. Two types of non-participators were identified: present during MDA but not treated (PNT) and eligible for treatment but absent during MDA (EBA). PNT and EBA children were compared to treated children separately. Multivariable models were developed using baseline data and validated using year one and two data, with a priori adjustment for previous treatment status. Analyses included approximately 10000 children at baseline and 5000 children subsequently. There was strong evidence of spatial heterogeneity, and persistent non-participation within households and individuals. By year two, non-participation increased significantly to 10.4% overall from 6.2% at baseline, with more, smaller geographical clusters of non-participating households. Multivariable models suggested household level predictors of non-participation (increased time to water and household head non-participation for both PNT and EBA; increased household size for PNT status only; non-inclusion in a previous trachoma examination survey and younger age for EBA only). Enhanced coverage efforts did not decrease non-participation. Few infected children were detected at year three and only one infected child was EBA previously. Infected children were in communities close to untreated endemic areas with higher rates of EBA non-participation during MDA.
Conclusions/Significance
In hypo-endemic settings, with good coverage and no association between non-participation and infection, efforts to improve participation during MDA may not be required. Further research could investigate spatial hotspots of infection and non-participation in other low and medium prevalence settings before allocating resources to increase participation.
Author Summary
As the target year for Global Elimination of Trachoma (GET2020) approaches, the scale up of mass drug administration (MDA) with azithromycin will lead to more endemic areas becoming low prevalence settings. In such areas, identification of those at highest risk of Chlamydia trachomatis infection and at highest risk of non-participation during MDA could inform control planning, especially if correlation is present. We investigated non-participation in children aged 1–9 years during three annual MDAs in The Gambia, a low prevalence setting. We found evidence that non-participation is associated with household membership and decision-making, as seen in medium and high prevalence settings in East Africa. In addition, we demonstrate geographical heterogeneity (spatial clustering) of non-participation, persistent non-participation behaviour over time and different non-participator types. Between the first and third MDA, non-participation increased significantly overall from 6.2% to 10.4%, whilst spatial clusters became smaller with non-participation more focused in single households or small groups of households. There was no evidence of association between infection and non-participation. In low prevalence settings with no evidence to suggest non-participation as a risk factor for infection, resources to improve participation may not be required. Spatial hotspot analysis could address this research question in areas with more infection.
doi:10.1371/journal.pntd.0003098
PMCID: PMC4148234  PMID: 25165994
15.  Fecal Contamination of Drinking-Water in Low- and Middle-Income Countries: A Systematic Review and Meta-Analysis 
PLoS Medicine  2014;11(5):e1001644.
Robert Bain and colleagues conduct a systematic review and meta-analysis to assess whether water from “improved” sources is less likely to contain fecal contamination than “unimproved” sources and find that access to an “improved source” provides a measure of sanitary protection but does not ensure water is free of fecal contamination.
Please see later in the article for the Editors' Summary
Background
Access to safe drinking-water is a fundamental requirement for good health and is also a human right. Global access to safe drinking-water is monitored by WHO and UNICEF using as an indicator “use of an improved source,” which does not account for water quality measurements. Our objectives were to determine whether water from “improved” sources is less likely to contain fecal contamination than “unimproved” sources and to assess the extent to which contamination varies by source type and setting.
Methods and Findings
Studies in Chinese, English, French, Portuguese, and Spanish were identified from online databases, including PubMed and Web of Science, and grey literature. Studies in low- and middle-income countries published between 1990 and August 2013 that assessed drinking-water for the presence of Escherichia coli or thermotolerant coliforms (TTC) were included provided they associated results with a particular source type. In total 319 studies were included, reporting on 96,737 water samples. The odds of contamination within a given study were considerably lower for “improved” sources than “unimproved” sources (odds ratio [OR] = 0.15 [0.10–0.21], I2 = 80.3% [72.9–85.6]). However over a quarter of samples from improved sources contained fecal contamination in 38% of 191 studies. Water sources in low-income countries (OR = 2.37 [1.52–3.71]; p<0.001) and rural areas (OR = 2.37 [1.47–3.81] p<0.001) were more likely to be contaminated. Studies rarely reported stored water quality or sanitary risks and few achieved robust random selection. Safety may be overestimated due to infrequent water sampling and deterioration in quality prior to consumption.
Conclusion
Access to an “improved source” provides a measure of sanitary protection but does not ensure water is free of fecal contamination nor is it consistent between source types or settings. International estimates therefore greatly overstate use of safe drinking-water and do not fully reflect disparities in access. An enhanced monitoring strategy would combine indicators of sanitary protection with measures of water quality.
Please see later in the article for the Editors' Summary
Editors' Summary
Background
Access to clean water is fundamental to human health. The importance of water to human health and wellbeing is encapsulated in the Human Right to Water, reaffirmed by the United Nations in 2010, which entitles everyone to “sufficient, safe, acceptable and physically accessible and affordable water for personal and domestic uses.” A step towards such universal access to water is Millennium Development Goal (MDG) target 7c that aims to halve the proportion of the population without sustainable access to safe drinking-water. One of the indicators to help monitor progress towards this target used by the Joint Monitoring Project (JMP—an initiative of the World Health Organization and UNICEF) is “use of an improved source.” Improved sources include piped water into a dwelling, yard, or plot, or a standpipe, borehole, and protected dug well. Unimproved sources are those that do not protect water from outside contamination, such as unprotected wells, unprotected springs, and surface waters.
Why Was This Study Done?
While this simple categorization may reflect established principles of sanitary protection, this indicator has been criticized for not adequately reflecting safety, suggesting that reported access to safe water might be overestimated by billions of people by not accounting for microbial water safety or more fully accounting for sanitary status. So the researchers conducted a systematic review and meta-analysis to investigate whether water from improved sources is less likely to exceed health-based guidelines for microbial water quality than water from unimproved sources and to what extent microbial contamination varies between source types, between countries, and between rural and urban areas.
What Did the Researchers Do and Find?
The researchers comprehensively searched the literature to find appropriate studies that investigated fecal contamination of all types of drinking-water in low and middle-income countries. The researchers included studies that contained extractable data on Escherichia coli or thermotolerant coliform (the WHO recommended indicators of fecal contamination) collected by appropriate techniques. The authors also assessed studies for bias and quality and used a statistical method (random effects meta-regression) to investigate risk factors and settings where fecal contamination of water sources was most common.
Using these methods, the authors included 319 studies reporting on 96,737 water samples. Most studies were from sub-Saharan Africa, southern Asia, or Latin America and the Caribbean. They found that overall, the odds (chance) of contamination within a given study were considerably lower for “improved” sources than “unimproved” sources (odds ratio = 0.15). However, in 38% of 191 studies, over a quarter of samples from improved sources contained fecal contamination. In particular, protected dug wells were rarely free of fecal contamination. The researchers also found that water sources in low-income countries, and rural areas were more likely to be contaminated (both had odds ratios of 2.37).
What Do These Findings Mean?
These findings show that while water from improved sources is less likely to contain fecal contamination than unimproved sources, they are not consistently safe. This study also provides evidence that by equating “improved” with “safe,” the number of people with access to a safe water source has been greatly overstated, and suggests that a large number and proportion of the world's population use unsafe water. As studies rarely reported stored water quality or sanitary risks, the accuracy of these findings may be limited. Nevertheless, the findings from this study suggest that the Global Burden of Disease 2010 may greatly underestimate diarrheal disease burden by assuming zero risk from improved water sources and that new indicators are needed to assess access to safe drinking water. Therefore, greater use should be made of other measures, such as sanitary inspections, to provide a complementary means of assessing safety and to help identify corrective actions to prevent water contamination.
Additional Information
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001644.
PLOS Medicine has a published series on water and sanitation
The WHO/UNICEF Joint Monitoring Programme for Water Supply and Sanitation has extensive information including data and country files
UN Water is the inter-agency coordination mechanism for all freshwater and sanitation related matters
The UN also provides information on the human right to water and sanitation
doi:10.1371/journal.pmed.1001644
PMCID: PMC4011876  PMID: 24800926
16.  Sustained high serum malondialdehyde levels are associated with severity and mortality in septic patients 
Critical Care  2013;17(6):R290.
Introduction
There is a hyperoxidative state in sepsis. The objective of this study was to determine serum malondialdehyde (MDA) levels during the first week of follow up, whether such levels are associated with severity during the first week and whether non-surviving patients showed higher MDA levels than survivors during the first week.
Methods
We performed an observational, prospective, multicenter study in six Spanish Intensive Care Units. Serum levels of MDA were measured in 328 patients (215 survivors and 113 non-survivors) with severe sepsis at days one, four and eight of diagnosis, and in 100 healthy controls. The primary endpoint was 30-day mortality and the secondary endpoint was six -month mortality. The association between continuous variables was carried out using Spearman’s rank correlation coefficient. Cox regression analysis was applied to determine the independent contribution of serum MDA levels on the prediction of 30-day and 6-month mortality. Hazard ratio (HR) and 95% confidence intervals (CI) were calculated as measures of the clinical impact of the predictor variables.
Results
We found higher serum MDA in septic patients at day one (p < 0.001), day four (p < 0.001) and day eight (p < 0.001) of diagnosis than in healthy controls. Serum MDA was lower in surviving than non-surviving septic patients at day one (p < 0.001), day four (p < 0.001) and day eight (p < 0.001). Serum MDA levels were positively correlated with lactic acid and SOFA during the first week. Finally, serum MDA levels were associated with 30-day mortality (HR = 1.05; 95% CI = 1.02-1.09; p = 0.005) and six-month mortality (hazard ratio (HR) = 1.05; 95% CI = 1.02-1.09; p = 0.003) after controlling for lactic acid levels, acute physiology and chronic health evaluation (APACHE)-II, diabetes mellitus, bloodstream infection and chronic renal failure.
Conclusions
To our knowledge, this is the largest series providing data on the oxidative state in septic patients to date. The novel finding is that high serum MDA levels sustained throughout the first week of follow up were associated with severity and mortality in septic patients.
doi:10.1186/cc13155
PMCID: PMC4055989  PMID: 24326199
17.  Bias in random forest variable importance measures: Illustrations, sources and a solution 
BMC Bioinformatics  2007;8:25.
Background
Variable importance measures for random forests have been receiving increased attention as a means of variable selection in many classification tasks in bioinformatics and related scientific fields, for instance to select a subset of genetic markers relevant for the prediction of a certain disease. We show that random forest variable importance measures are a sensible means for variable selection in many applications, but are not reliable in situations where potential predictor variables vary in their scale of measurement or their number of categories. This is particularly important in genomics and computational biology, where predictors often include variables of different types, for example when predictors include both sequence data and continuous variables such as folding energy, or when amino acid sequence data show different numbers of categories.
Results
Simulation studies are presented illustrating that, when random forest variable importance measures are used with data of varying types, the results are misleading because suboptimal predictor variables may be artificially preferred in variable selection. The two mechanisms underlying this deficiency are biased variable selection in the individual classification trees used to build the random forest on one hand, and effects induced by bootstrap sampling with replacement on the other hand.
Conclusion
We propose to employ an alternative implementation of random forests, that provides unbiased variable selection in the individual classification trees. When this method is applied using subsampling without replacement, the resulting variable importance measures can be used reliably for variable selection even in situations where the potential predictor variables vary in their scale of measurement or their number of categories. The usage of both random forest algorithms and their variable importance measures in the R system for statistical computing is illustrated and documented thoroughly in an application re-analyzing data from a study on RNA editing. Therefore the suggested method can be applied straightforwardly by scientists in bioinformatics research.
doi:10.1186/1471-2105-8-25
PMCID: PMC1796903  PMID: 17254353
18.  Geographical Inequalities in Use of Improved Drinking Water Supply and Sanitation across Sub-Saharan Africa: Mapping and Spatial Analysis of Cross-sectional Survey Data 
PLoS Medicine  2014;11(4):e1001626.
Using cross-sectional survey data, Rachel Pullan and colleagues map geographical inequalities in use of improved drinking water supply and sanitation across sub-Saharan Africa.
Please see later in the article for the Editors' Summary
Background
Understanding geographic inequalities in coverage of drinking-water supply and sanitation (WSS) will help track progress towards universal coverage of water and sanitation by identifying marginalized populations, thus helping to control a large number of infectious diseases. This paper uses household survey data to develop comprehensive maps of WSS coverage at high spatial resolution for sub-Saharan Africa (SSA). Analysis is extended to investigate geographic heterogeneity and relative geographic inequality within countries.
Methods and Findings
Cluster-level data on household reported use of improved drinking-water supply, sanitation, and open defecation were abstracted from 138 national surveys undertaken from 1991–2012 in 41 countries. Spatially explicit logistic regression models were developed and fitted within a Bayesian framework, and used to predict coverage at the second administrative level (admin2, e.g., district) across SSA for 2012. Results reveal substantial geographical inequalities in predicted use of water and sanitation that exceed urban-rural disparities. The average range in coverage seen between admin2 within countries was 55% for improved drinking water, 54% for use of improved sanitation, and 59% for dependence upon open defecation. There was also some evidence that countries with higher levels of inequality relative to coverage in use of an improved drinking-water source also experienced higher levels of inequality in use of improved sanitation (rural populations r = 0.47, p = 0.002; urban populations r = 0.39, p = 0.01). Results are limited by the quantity of WSS data available, which varies considerably by country, and by the reliability and utility of available indicators.
Conclusions
This study identifies important geographic inequalities in use of WSS previously hidden within national statistics, confirming the necessity for targeted policies and metrics that reach the most marginalized populations. The presented maps and analysis approach can provide a mechanism for monitoring future reductions in inequality within countries, reflecting priorities of the post-2015 development agenda.
Please see later in the article for the Editors' Summary
Editors' Summary
Background
Access to a safe drinking-water supply (a water source that is protected from contamination) and to adequate sanitation facilities (toilets, improved latrines, and other facilities that prevent people coming into contact with human urine and feces) is essential for good health. Unimproved drinking-water sources and sanitation are responsible for 85% of deaths from diarrhea and 1% of the global burden of disease. They also increase the transmission of parasitic worms and other neglected tropical diseases. In 2000, world leaders set a target of reducing the proportion of the global population without access to safe drinking water and basic sanitation to half of the 1990 level by 2015 as part of Millennium Development Goal (MDG) 7 (“Ensure environmental sustainability”; the MDGs are designed to improve the social, economic, and health conditions in the world's poorest countries). Between 1990 and 2010, more than 2 billion people gained access to improved drinking-water sources and 1.8 billion gained access to improved sanitation. In 2011, 89% of the world's population had access to an improved drinking-water supply, 1% above the MDG target, and 64% had access to improved sanitation (the MDG target is 75%).
Why Was This Study Done?
Despite these encouraging figures, the WHO/UNICEF Joint Monitoring Programme for Water Supply and Sanitation (JMP) estimates that, globally, 768 million people relied on unimproved drinking-water sources, 2.5 billion people did not use an improved sanitation facility, and more than 1 billion people (15% of the global population) were defecating in the open in 2011. The JMP estimates for 2011 also reveal national and sub-national inequalities in drinking-water supply and sanitation coverage but a better understanding of geographic inequalities is needed to track progress towards universal coverage of access to improved water and sanitation and to identify the populations that need the most help to achieve this goal. Here, the researchers use cross-sectional household survey data and modern statistical approaches to produce a comprehensive map of the coverage of improved drinking-water supply and improved sanitation at high spatial resolution for sub-Saharan Africa and to investigate geographic inequalities in coverage. Cross-sectional household surveys collect health and other information from households at a single time-point, including data on use of safe water and improved sanitation.
What Did the Researchers Do and Find?
The researchers extracted data on reported household use of an improved drinking-water supply (for example, a piped water supply), improved sanitation facilities (for example, a flushing toilet), and open defecation from 138 national household surveys undertaken between 1991 and 2012 in 41 countries in sub-Saharan Africa. They developed statistical models to fit these data and used the models to estimate coverage at the district (second administrative) level across sub-Saharan Africa for 2012. For ten countries, the estimated coverage of access to improved drinking water at the district level within individual countries ranged from less than 25% to more than 75%. Within-country ranges of a similar magnitude were estimated for coverage of access to improved sanitation (21 countries) and for open defecation (16 countries). Notably, rural households in the districts with the lowest coverage of access to improved water supply and sanitation within a country were 1.5–8 times less likely to access improved drinking water, 2–18 times less likely to access improved sanitation, and 2–80 times more likely to defecate in the open than rural households in districts with the best coverage. Finally, countries with high levels of inequality in improved drinking-water source coverage also experienced high levels of inequality in improved sanitation coverage.
What Do These Findings Mean?
These findings identify important geographic inequalities in the coverage of access to improved water sources and sanitation that were previously hidden within national statistics. The accuracy of these findings depends on the accuracy of the data on water supplies and sanitation provided by household surveys, on the researchers' definitions for improved water supplies and sanitation, and on their statistical methods. Nevertheless, these findings confirm that, to achieve universal coverage of access to improved drinking-water sources and sanitation, strategies that target the areas with the lowest coverage are essential. Moreover, the maps and the analytical approach presented here provide the means for monitoring future reductions in inequalities in the coverage of access to improved water sources and sanitation and thus reflect a major priority of the post-2015 development agenda.
Additional Information
Please access these websites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001626.
A PLOS Medicine Collection on water and sanitation is available
The World Health Organization (WHO) provides information on water, sanitation, and health (in several languages)
The WHO/UNICEF Joint Monitoring Programme for Water Supply and Sanitation is the official United Nations mechanism tasked with monitoring progress toward MDG7, Target 7B; the JMP 2013 update report is available online (also available in French and Spanish through the JMP website)
The sub-national predictions resulting from this study and the final sub-national maps are available as a resource for researchers and planners
doi:10.1371/journal.pmed.1001626
PMCID: PMC3979660  PMID: 24714528
19.  Random KNN feature selection - a fast and stable alternative to Random Forests 
BMC Bioinformatics  2011;12:450.
Background
Successfully modeling high-dimensional data involving thousands of variables is challenging. This is especially true for gene expression profiling experiments, given the large number of genes involved and the small number of samples available. Random Forests (RF) is a popular and widely used approach to feature selection for such "small n, large p problems." However, Random Forests suffers from instability, especially in the presence of noisy and/or unbalanced inputs.
Results
We present RKNN-FS, an innovative feature selection procedure for "small n, large p problems." RKNN-FS is based on Random KNN (RKNN), a novel generalization of traditional nearest-neighbor modeling. RKNN consists of an ensemble of base k-nearest neighbor models, each constructed from a random subset of the input variables. To rank the importance of the variables, we define a criterion on the RKNN framework, using the notion of support. A two-stage backward model selection method is then developed based on this criterion. Empirical results on microarray data sets with thousands of variables and relatively few samples show that RKNN-FS is an effective feature selection approach for high-dimensional data. RKNN is similar to Random Forests in terms of classification accuracy without feature selection. However, RKNN provides much better classification accuracy than RF when each method incorporates a feature-selection step. Our results show that RKNN is significantly more stable and more robust than Random Forests for feature selection when the input data are noisy and/or unbalanced. Further, RKNN-FS is much faster than the Random Forests feature selection method (RF-FS), especially for large scale problems, involving thousands of variables and multiple classes.
Conclusions
Given the superiority of Random KNN in classification performance when compared with Random Forests, RKNN-FS's simplicity and ease of implementation, and its superiority in speed and stability, we propose RKNN-FS as a faster and more stable alternative to Random Forests in classification problems involving feature selection for high-dimensional datasets.
doi:10.1186/1471-2105-12-450
PMCID: PMC3281073  PMID: 22093447
20.  Equity in reproductive and maternal health services in Bangladesh 
Background
The target date for achieving the Millennium Development Goals (MDGs) is now closer than ever. There is lack of sufficient progress in achieving the MDG targets in many low- and middle-income countries. Furthermore, there has also been concerns about wide spread inequity among those that are on track to achieve the health-related MDGs. Bangladesh has made a notable progress towards achieving the MDG 5 targets. It is, however, important to assess if this is an inclusive and equitable progress, as inequitable progress may not lead to sustainable health outcomes. The objective of this study is to assess the magnitude of inequities in reproductive and maternal health services in Bangladesh and propose relevant recommendations for decision making.
Methods
The 2007 Bangladesh demographic and health survey data is analyzed for inequities in selected maternal and reproductive health interventions using the slope and relative indices of inequality.
Results
The analysis indicates that there are significant wealth-related inequalities favouring the wealthiest of society in many of the indicators considered. Antenatal care (at least 4 visits), antenatal care by trained providers such as doctors and nurses, content of antenatal care, skilled birth attendance, delivery in health facility and delivery by caesarean section all manifest inequities against the least wealthy. There are no wealth-related inequalities in the use of modern contraception. In contrast, less desired interventions such as delivery by untrained providers and home delivery show wealth-related inequalities in favour of the poor.
Conclusions
For an inclusive and sustainable improvement in maternal and reproductive health outcomes and achievement of MDG 5 targets, it essential to address inequities in maternal and reproductive health interventions. Under the government’s stewardship, all stakeholders should accord priority to tackling wealth-related inequalities in maternal and reproductive health services by implementing equity-promoting measures both within and outside the health sector.
doi:10.1186/1475-9276-12-90
PMCID: PMC3842788  PMID: 24228997
21.  Do we have the right models for scaling up health services to achieve the Millennium Development Goals? 
Background
There is widespread agreement on the need for scaling up in the health sector to achieve the Millennium Development Goals (MDGs). But many countries are not on track to reach the MDG targets. The dominant approach used by global health initiatives promotes uniform interventions and targets, assuming that specific technical interventions tested in one country can be replicated across countries to rapidly expand coverage. Yet countries scale up health services and progress against the MDGs at very different rates. Global health initiatives need to take advantage of what has been learned about scaling up.
Methods
A systematic literature review was conducted to identify conceptual models for scaling up health in developing countries, with the articles assessed according to the practical concerns of how to scale up, including the planning, monitoring and implementation approaches.
Results
We identified six conceptual models for scaling up in health based on experience with expanding pilot projects and diffusion of innovations. They place importance on paying attention to enhancing organizational, functional, and political capabilities through experimentation and adaptation of strategies in addition to increasing the coverage and range of health services. These scaling up approaches focus on fostering sustainable institutions and the constructive engagement between end users and the provider and financing organizations.
Conclusions
The current approaches to scaling up health services to reach the MDGs are overly simplistic and not working adequately. Rather than relying on blueprint planning and raising funds, an approach characteristic of current global health efforts, experience with alternative models suggests that more promising pathways involve "learning by doing" in ways that engage key stakeholders, uses data to address constraints, and incorporates results from pilot projects. Such approaches should be applied to current strategies to achieve the MDGs.
doi:10.1186/1472-6963-11-336
PMCID: PMC3260120  PMID: 22168915
22.  Monitoring of Mass Distribution Interventions for Trachoma in Plateau State, Nigeria 
Mass drug administration (MDA) with antibiotics is a key component of the SAFE strategy for trachoma control. Guidelines recommend that where MDA is warranted the whole population be targeted with 80% considered the minimum acceptable coverage. In other countries, MDA is usually conducted by salaried Ministry of Health personnel (MOH). In Plateau State, Nigeria, the existing network of volunteer Community Directed Distributors (CDD) was used for the first trachoma MDA. We conducted a population-based cluster random survey (CRS) of MDA participation to determine the true coverage and compared this to coverage reported from CDD registers. We surveyed 1,791 people from 352 randomly selected households in 24 clusters in three districts in Plateau State in January 2011, following the implementation of MDA. Households were enumerated and all individuals present were asked about MDA participation. Household heads were questioned about household-level characteristics and predictors of participation. Individual responses were compared with the CDD registers. MDA coverage was estimated as 60.3% (95% CI 47.9–73.8%) by the survey compared with 75.8% from administrative program reports. CDD registration books for comparison with responses were available in 19 of the 24 clusters; there was a match for 658/682 (96%) of verifiable responses. CDD registers did not list 481 (41.3%) of the individuals surveyed. Gender and age were not associated with individual participation. Overall MDA coverage was lower than the minimum 80% target. The observed discrepancy between the administrative coverage estimate from program reports and the CRS was largely due to identification of communities missed by the MDA and not reported in the registers. CRS for evaluation of MDA provides a useful additional monitoring tool to CDD registers. These data support modification of distributor training and MDA delivery to increase coverage in subsequent rounds of MDA.
Author Summary
The World Health Organization recommends that mass drug administration for trachoma control reach a minimum of 80% of the target population. Previous evaluations of MDA coverage have demonstrated that administrative reports can bias coverage estimates. A survey of participation in mass drug administration for trachoma control was implemented in three districts in Plateau State, Nigeria in 2011 to validate coverage calculated from treatment registers. A total of 352 households were surveyed from 24 randomly selected communities. Heads of household were interviewed to identify household-level characteristics and predictors of participation. Individual household members were enumerated and those present at the time of interview were asked to report individual participation in the MDA. Responses were verified against the community-drug distributor registration log. Approximately 60% of the sample reported receiving either tetracycline eye ointment or azithromycin for trachoma control. Administrative data on treatment estimated coverage at 76% for the three LGAs. The discrepancy between the coverage estimate from administrative data (calculated by the program) and the survey data suggest that cluster random surveys of MDA provide a useful monitoring tool to validate administrative data on treatment coverage. These data support modification of distributor training and MDA delivery to increase coverage in subsequent rounds of MDA.
doi:10.1371/journal.pntd.0001995
PMCID: PMC3542118  PMID: 23326617
23.  A Multifaceted Intervention to Improve the Quality of Care of Children in District Hospitals in Kenya: A Cost-Effectiveness Analysis 
PLoS Medicine  2012;9(6):e1001238.
A cost-effective analysis conducted by Edwine Barasa and colleagues estimates that a complex intervention aimed at improving quality of pediatric care would be affordable and cost-effective in Kenya.
Background
To improve care for children in district hospitals in Kenya, a multifaceted approach employing guidelines, training, supervision, feedback, and facilitation was developed, for brevity called the Emergency Triage and Treatment Plus (ETAT+) strategy. We assessed the cost effectiveness of the ETAT+ strategy, in Kenyan hospitals. Further, we estimate the costs of scaling up the intervention to Kenya nationally and potential cost effectiveness at scale.
Methods and Findings
Our cost-effectiveness analysis from the provider's perspective used data from a previously reported cluster randomized trial comparing the full ETAT+ strategy (n = 4 hospitals) with a partial intervention (n = 4 hospitals). Effectiveness was measured using 14 process measures that capture improvements in quality of care; their average was used as a summary measure of quality. Economic costs of the development and implementation of the intervention were determined (2009 US$). Incremental cost-effectiveness ratios were defined as the incremental cost per percentage improvement in (average) quality of care. Probabilistic sensitivity analysis was used to assess uncertainty. The cost per child admission was US$50.74 (95% CI 49.26–67.06) in intervention hospitals compared to US$31.1 (95% CI 30.67–47.18) in control hospitals. Each percentage improvement in average quality of care cost an additional US$0.79 (95% CI 0.19–2.31) per admitted child. The estimated annual cost of nationally scaling up the full intervention was US$3.6 million, approximately 0.6% of the annual child health budget in Kenya. A “what-if” analysis assuming conservative reductions in mortality suggests the incremental cost per disability adjusted life year (DALY) averted by scaling up would vary between US$39.8 and US$398.3.
Conclusion
Improving quality of care at scale nationally with the full ETAT+ strategy may be affordable for low income countries such as Kenya. Resultant plausible reductions in hospital mortality suggest the intervention could be cost-effective when compared to incremental cost-effectiveness ratios of other priority child health interventions.
Please see later in the article for the Editors' Summary
Editors' Summary
Background
According to latest global estimates from UNICEF, 7.6 million children currently die every year before they reach five years of age. Half of these deaths occur in children in sub-Saharan Africa and tragically, most of these deaths are due to a few treatable and preventable diseases, such as pneumonia, malaria, and diarrhea, for which effective interventions are already available. In order to meet the target of the 4th Millennium Development Goal—which aims to reduce the under-five child mortality rate by two-thirds from 1990 levels by 2015—delivering these interventions is essential.
In Kenya, the under-five child mortality rate must be reduced by half from its 2008 level in order to meet the Millennium Development Goal (MDG) target and so improving the management of serious child illness might help achieve this goal. A study published last year in PLoS Medicine described such an approach and included the development and implementation of evidence-based clinical practice guidelines linked to health worker training, follow-up supervision, performance feedback, and facilitation in eight district hospitals in Kenya.
Why Was This Study Done?
In the study mentioned above, the researchers compared the implementation of various processes of care in intervention and control hospitals at baseline and 18 months later and found that performance improved more in the intervention hospitals than in the control hospitals. However, while this strategy was effective at improving the quality of health care, it is unclear whether scaling up the approach would be a good use of limited resources. So in this study, the same researchers performed a cost-effectiveness analysis (which they conducted alongside the original trial) of their quality improvement intervention and estimated the costs and effects of scaling up this approach to cover the entire population of Kenya.
What Did the Researchers Do and Find?
In order to perform the cost part of the analysis, the researchers collected the relevant information on costs by using clinical and accounting record reviews and interviews with those involved in developing and implementing the intervention. The researchers evaluated the effectiveness part of the analysis by comparing the implementation of their improved quality of care strategy as delivered in the intervention hospitals with the partial intervention as delivered in the control hospitals by calculating the mean percentage improvement in the 14 process of care indicators at 18 months. Finally, the researchers calculated the costs of scaling up the intervention by applying their results to the whole of Kenya—121 hospital facilities with an estimated annual child admission rate of 2,000 per facility.
The researchers found that the quality of care (as measured by the process of care indicators) was 25% higher in intervention hospitals than in control hospitals, while the cost per child admission was US$50.74 in intervention hospitals compared to US$31.1 in control hospitals. The researchers calculated that each percentage improvement in the average quality of care was achieved at an additional cost of US$0.79 per admitted child. Extrapolating these results to all of Kenya, the estimated annual cost of scaling up the intervention nationally was US$3.6 million, about 0.6% of the annual child health budget in Kenya.
What Do These Findings Mean?
The findings of this cost-effectiveness analysis suggests that a comprehensive quality improvement intervention is effective at improving standards of care but at an additional cost, which may be relatively cost effective compared with basic care if the improvements observed are associated with decreases in child inpatient mortality. The absolute costs for scaling up are comparable to, or even lower than, costs of other, major child health interventions. As the international community is giving an increasing focus to strengthening health systems, these findings provide a strong case for scaling up this intervention, which improves quality of care and service provision for the major causes of child mortality, in rural hospitals throughout Kenya and other district hospitals in sub-Saharan Africa.
Additional Information
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1001238.
The researchers' original article appeared in PLoS Medicine in 2011: Ayieko P, Ntoburi S, Wagai J, Opondo C, Opiyo N, et al. (2011) A Multifaceted Intervention to Implement Guidelines and Improve Admission Paediatric Care in Kenyan District Hospitals: A Cluster Randomised Trial. PLoS Med 8(4): e1001018. doi:10.1371/journal.pmed.1001018
The IDOC Africa provides further information on the ETAT+ strategy
The World Health Organization (WHO) provides information on MDG 4, including strategies to reduce global child mortality) and the WHO pocket-book “Hospital care for children” includes guidelines for the management of common but serious childhood illnesses in resource-limited settings
UNICEF www.unicef.org also publishes information on global child mortality rates and the Countdown to 2015 website tracks coverage levels for health interventions proven to reduce child mortality
doi:10.1371/journal.pmed.1001238
PMCID: PMC3373608  PMID: 22719233
24.  Screening large-scale association study data: exploiting interactions using random forests 
BMC Genetics  2004;5:32.
Background
Genome-wide association studies for complex diseases will produce genotypes on hundreds of thousands of single nucleotide polymorphisms (SNPs). A logical first approach to dealing with massive numbers of SNPs is to use some test to screen the SNPs, retaining only those that meet some criterion for futher study. For example, SNPs can be ranked by p-value, and those with the lowest p-values retained. When SNPs have large interaction effects but small marginal effects in a population, they are unlikely to be retained when univariate tests are used for screening. However, model-based screens that pre-specify interactions are impractical for data sets with thousands of SNPs. Random forest analysis is an alternative method that produces a single measure of importance for each predictor variable that takes into account interactions among variables without requiring model specification. Interactions increase the importance for the individual interacting variables, making them more likely to be given high importance relative to other variables. We test the performance of random forests as a screening procedure to identify small numbers of risk-associated SNPs from among large numbers of unassociated SNPs using complex disease models with up to 32 loci, incorporating both genetic heterogeneity and multi-locus interaction.
Results
Keeping other factors constant, if risk SNPs interact, the random forest importance measure significantly outperforms the Fisher Exact test as a screening tool. As the number of interacting SNPs increases, the improvement in performance of random forest analysis relative to Fisher Exact test for screening also increases. Random forests perform similarly to the univariate Fisher Exact test as a screening tool when SNPs in the analysis do not interact.
Conclusions
In the context of large-scale genetic association studies where unknown interactions exist among true risk-associated SNPs or SNPs and environmental covariates, screening SNPs using random forest analyses can significantly reduce the number of SNPs that need to be retained for further study compared to standard univariate screening methods.
doi:10.1186/1471-2156-5-32
PMCID: PMC545646  PMID: 15588316
25.  A Case-Control Study to Assess the Relationship between Poverty and Visual Impairment from Cataract in Kenya, the Philippines, and Bangladesh 
PLoS Medicine  2008;5(12):e244.
Background
The link between poverty and health is central to the Millennium Development Goals (MDGs). Poverty can be both a cause and consequence of poor health, but there are few epidemiological studies exploring this complex relationship. The aim of this study was to examine the association between visual impairment from cataract and poverty in adults in Kenya, Bangladesh, and the Philippines.
Methods and Findings
A population-based case–control study was conducted in three countries during 2005–2006. Cases were persons aged 50 y or older and visually impaired due to cataract (visual acuity < 6/24 in the better eye). Controls were persons age- and sex-matched to the case participants with normal vision selected from the same cluster. Household expenditure was assessed through the collection of detailed consumption data, and asset ownership and self-rated wealth were also measured. In total, 596 cases and 535 controls were included in these analyses (Kenya 142 cases, 75 controls; Bangladesh 216 cases, 279 controls; Philippines 238 cases, 180 controls). Case participants were more likely to be in the lowest quartile of per capita expenditure (PCE) compared to controls in Kenya (odds ratio = 2.3, 95% confidence interval 0.9–5.5), Bangladesh (1.9, 1.1–3.2), and the Philippines (3.1, 1.7–5.7), and there was significant dose–response relationship across quartiles of PCE. These associations persisted after adjustment for self-rated health and social support indicators. A similar pattern was observed for the relationship between cataract visual impairment with asset ownership and self-rated wealth. There was no consistent pattern of association between PCE and level of visual impairment due to cataract, sex, or age among the three countries.
Conclusions
Our data show that people with visual impairment due to cataract were poorer than those with normal sight in all three low-income countries studied. The MDGs are committed to the eradication of extreme poverty and provision of health care to poor people, and this study highlights the need for increased provision of cataract surgery to poor people, as they are particularly vulnerable to visual impairment from cataract.
Hannah Kuper and colleagues report a population-based case-control study conducted in three countries that found an association between poverty and visual impairment from cataract.
Editors' Summary
Background.
Globally, about 45 million people are blind. As with many other conditions, avoidable blindness (preventable or curable blindness) is a particular problem for people in developing countries—90% of blind people live in poor regions of the world. Although various infections and disorders can cause blindness, cataract is the most common cause. In cataract, which is responsible for half of all cases of blindness in the world, the lens of the eye gradually becomes cloudy. Because the lens focuses light to produce clear, sharp images, as cataract develops, vision becomes increasingly foggy or fuzzy, colors become less intense, and the ability to see shapes against a background declines. Eventually, vision may be lost completely. Cataract can be treated with an inexpensive, simple operation in which the cloudy lens is surgically removed and an artificial lens is inserted into the eye to restore vision. In developed countries, this operation is common and easily accessible but many poor countries lack the resources to provide the operation to everyone who needs it. In addition, blind people often cannot afford to travel to the hospitals where the operation, which also may come with a fee, is done.
Why Was This Study Done?
Because blindness may reduce earning potential, many experts believe that poverty and blindness (and, more generally, poor health) are inextricably linked. People become ill more often in poor countries than in wealthy countries because they have insufficient food, live in substandard housing, and have limited access to health care, education, water, and sanitation. Once they are ill, their ability to earn money may be reduced, which increases their personal poverty and slows the economic development of the whole country. Because of this potential link between health and poverty, improvements in health are at the heart of the United Nations Millennium Development Goals, a set of eight goals established in 2000 with the primary aim of reducing world poverty. However, few studies have actually investigated the complex relationship between poverty and health. Here, the researchers investigate the association between visual impairment from cataract and poverty among adults living in three low-income countries.
What Did the Researchers Do and Find?
The researchers identified nearly 600 people aged 50 y or more with severe cataract-induced visual impairment (“cases”) primarily through a survey of the population in Kenya, Bangladesh, and the Philippines. They matched each case to a normally sighted (“control”) person of similar age and sex living nearby. They then assessed a proxy for the income level, measured as “per capita expenditure” (PCE), of all the study participants (people with cataracts and controls) by collecting information about what their households consumed. The participants' housing conditions and other assets and their self-rated wealth were also measured. In all three countries, cases were more likely to be in the lowest quarter (quartile) of the range of PCEs for that country than controls. In the Philippines, for example, people with cataract-affected vision were three times more likely than normally sighted controls to have a PCE in the lowest quartile than in the highest quartile. The risk of cataract-related visual impairment increased as PCE decreased in all three countries. Similarly, severe cataract-induced visual impairment was more common in those who owned fewer assets and those with lower self-rated wealth. However, there was no consistent association between PCE and the level of cataract-induced visual impairment.
What Do These Findings Mean?
These findings show that there is an association between visual impairment caused by cataract and poverty in Kenya, Bangladesh, and the Philippines. However, because the financial circumstances of the people in this study were assessed after cataracts had impaired their sight, this study does not prove that poverty is a cause of visual impairment. A causal connection between poverty and cataract can only be shown by determining the PCEs of normally sighted people and following them for several years to see who develops cataract. Nevertheless, by confirming an association between poverty and blindness, these findings highlight the need for increased provision of cataract surgery to poor people, particularly since cataract surgery has the potential to improve the quality of life for many people in developing countries at a relatively low cost.
Additional Information.
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.0050244.
This study is further discussed in a PLoS Medicine Perspective by Susan Lewallen
The MedlinePlus encyclopedia contains a page on cataract, and MedlinePlus also provides a list of links to further information about cataract (in English and Spanish)
VISION 2020, a global initiative for the elimination of avoidable blindness launched by the World Health Organization and the International Agency for the Prevention of Blindness, provides information in several languages about many causes of blindness, including cataract. It also has an article available for download on blindness, poverty, and development
Information is available from the World Health Organization on health and the Millennium Development Goals (in English, French, and Spanish)
The International Centre for Eye Health carries out research and education activities to improve eye health and eliminate avoidable blindness with a focus on populations with low incomes
doi:10.1371/journal.pmed.0050244
PMCID: PMC2602716  PMID: 19090614

Results 1-25 (1362895)