David E. Phillips

A global assessment of civil registration and vital statistics systems: monitoring data quality and progress

Increasing demand for better quality data and more investment to strengthen civil registration and vital statistics (CRVS) systems will require increased emphasis on objective, comparable, cost-effective monitoring and assessment methods to measure progress. We apply a composite index (the vital statistics performance index [VSPI]) to assess the performance of CRVS systems in 148 countries or territories during 1980-2012 and classify them into five distinct performance categories, ranging from rudimentary (with scores close to zero) to satisfactory (with scores close to one), with a mean VSPI score since 2005 of 0·61 (SD 0·31). As expected, the best performing systems were mostly in the European region, the Americas, and Australasia, with only two countries from east Asia and Latin America. Most low-scoring countries were in the African or Asian regions. Globally, only modest progress has been made since 2000, with the percentage of deaths registered increasing from 36% to 38%, and the percentage of children aged under 5 years whose birth has been registered increasing from 58% to 65%. However, several individual countries have made substantial improvements to their CRVS systems in the past 30 years by capturing more deaths and improving accuracy of cause-of-death information. Future monitoring of the effects of CRVS strengthening will greatly benefit from application of a metric like the VSPI, which is objective, costless to compute, and able to identify components of the system that make the largest contributions to good or poor performance.

Are well functioning civil registration and vital statistics systems associated with better health outcomes

In this Series paper, we examine whether well functioning civil registration and vital statistics (CRVS) systems are associated with improved population health outcomes. We present a conceptual model connecting CRVS to wellbeing, and describe an ecological association between CRVS and health outcomes. The conceptual model posits that the legal identity that civil registration provides to individuals is key to access entitlements and services. Vital statistics produced by CRVS systems provide essential information for public health policy and prevention. These outcomes benefit individuals and societies, including improved health. We use marginal linear models and lag-lead analysis to measure ecological associations between a composite metric of CRVS performance and three health outcomes. Results are consistent with the conceptual model: improved CRVS performance coincides with improved health outcomes worldwide in a temporally consistent manner. Investment to strengthen CRVS systems is not only an important goal for individuals and societies, but also a development imperative that is good for health.

A composite metric for assessing data on mortality and causes of death: the vital statistics performance index

BackgroundTimely and reliable data on causes of death are fundamental for informed decision-making in the health sector as well as public health research. An in-depth understanding of the quality of data from vital statistics (VS) is therefore indispensable for health policymakers and researchers. We propose a summary index to objectively measure the performance of VS systems in generating reliable mortality data and apply it to the comprehensive cause of death database assembled for the Global Burden of Disease (GBD) 2013 Study.MethodsWe created a Vital Statistics Performance Index, a composite of six dimensions of VS strength, each assessed by a separate empirical indicator. The six dimensions include: quality of cause of death reporting, quality of age and sex reporting, internal consistency, completeness of death reporting, level of cause-specific detail, and data availability/timeliness. A simulation procedure was developed to combine indicators into a single index. This index was computed for all country-years of VS in the GBD 2013 cause of death database, yielding annual estimates of overall VS system performance for 148 countries or territories.ResultsThe six dimensions impacted the accuracy of data to varying extents. VS performance declines more steeply with declining simulated completeness than for any other indicator. The amount of detail in the cause list reported has a concave relationship with overall data accuracy, but is an important driver of observed VS performance. Indicators of cause of death data quality and age/sex reporting have more linear relationships with simulated VS performance, but poor cause of death reporting influences observed VS performance more strongly. VS performance is steadily improving at an average rate of 2.10% per year among the 148 countries that have available data, but only 19.0% of global deaths post-2000 occurred in countries with well-performing VS systems.ConclusionsObjective and comparable information about the performance of VS systems and the utility of the data that they report will help to focus efforts to strengthen VS systems. Countries and the global health community alike need better intelligence about the accuracy of VS that are widely and often uncritically used in population health research and monitoring.

Determinants of effective vaccine coverage in low and middle-income countries: a systematic review and interpretive synthesis

BackgroundMany children in low and middle-income countries remain unvaccinated, and vaccines do not always produce immunity. Extensive research has sought to understand why, but most studies have been limited in breadth and depth. This study documents existing evidence on determinants of vaccination and immunization and presents a conceptual framework of determinants.MethodsWe used systematic review, content analysis, thematic analysis and interpretive synthesis to document and analyze the existing evidence on determinants of childhood vaccination and immunization.ResultsWe documented 1609 articles, including content analysis of 78 articles. Three major thematic models were described in the context of one another. Interpretive synthesis identified similarities and differences between studies, resulting in a conceptual framework with three principal vaccine utilization determinants: 1) Intent to Vaccinate, 2) Community Access and 3) Health Facility Readiness.ConclusionThis study presents the most comprehensive systematic review of vaccine determinants to date. The conceptual framework represents a synthesis of multiple existing frameworks, is applicable in low and middle-income countries, and is quantitatively testable. Future researchers can use these results to develop competing conceptual frameworks, or to analyze data in a theoretically-grounded way. This review enables better research in the future, further understanding of immunization determinants, and greater progress against vaccine preventable diseases around the world.

Childhood vaccines in Uganda and Zambia: Determinants and barriers to vaccine coverage

BACKGROUND Improving childhood vaccine coverage is a priority for global health, but challenging in low and middle-income countries. Although previous research has sought to measure determinants of vaccination, most has limitations. We measure determinants using a clearly-defined hypothetical model, multi-faceted data, and modeling strategy that makes full use of the hypothesis and data. METHODS We use linked, cross-sectional survey data from households, health facilities, patients and health offices in Uganda and Zambia, and Bayesian Structural Equation Modeling to quantify the proportion of variance in childhood vaccination that is explained by key determinants, controlling for known confounding. RESULTS We find evidence that the leading determinant of vaccination is different for different outcomes. For three doses of pentavalent vaccine, intent to vaccinate (on the part of the mother) is the leading driver, but for one dose of the vaccine, community access is a larger factor. For pneumococcal conjugate vaccine, health facility readiness is the leading driver. Considering specifically-modifiable determinants, improvements in cost, facility catchment populations and staffing would be expected to lead to the largest increase in coverage according to the model. CONCLUSIONS This analysis measures vaccination determinants using improved methods over most existing research. It provides evidence that determinants should be approached in the context of relevant outcomes, and evidence of specific determinants that could have the greatest impact in these two countries, if targeted. Future studies should seek to improve our analytic framework, apply it in different settings, and utilize stronger study designs. Programs that focus on a particular determinant should use these results to select an outcome that is appropriate to measure their effectiveness. Vaccination programs in these countries should use our findings to better target interventions and continue progress against vaccine preventable diseases.

The role of implementation science training in global health: from the perspective of graduates of the field’s first dedicated doctoral program

Bridging the ‘know-do gap’ is an enormous challenge for global health practitioners. They must be able to understand local health dynamics within the operational and social contexts that engender them, test and adjust approaches to implementation in collaboration with communities and stakeholders, interpret data to inform policy decisions, and design adaptive and resilient health systems at scale. These skills and methods have been formalized within the nascent field of Implementation Science (IS). As graduates of the worlds first PhD program dedicated explicitly to IS, we have a unique perspective on the value of IS and the training, knowledge, and skills essential to bridging the ‘know-do gap’. In this article, we describe the philosophy and curricula at the core of our program, outline the methods vital to IS in a global health context, and detail the role that we believe IS will increasingly play in global health practice. At this junction of enormous challenges and opportunities, we believe that IS offers the necessary tools for global health professionals to address complex problems in context and raises the bar of success for the global health programs of the future.

Ensemble modelling in verbal autopsy: the Popular Voting method

Abstract Background Verbal autopsy (VA) is a highly valuable tool for assessing causes of death in resource-limited settings without medically certified death certificates. The Population Health Metrics Research Consortium (PHMRC) collected 12 535 VAs in four countries for which the true cause of death was reliably known. This project led to the development of three new computer algorithms to determine cause of death from these VAs, all of which predict underlying cause of death more accurately than the status quo: physician review. Concurrently, ensemble models, or blends of well-performing models, have been shown to have favourable predictive validity and have begun to be implemented in global health metrics settings. Methods We developed a simple ensemble model based on the three top performing PHMRC methods: the Simplified Symptom Pattern (SSP), the Tariff, and the Random Forest (RF). This ensemble method functions at the individual-record level, examining the predicted cause of death from the three component models and selecting cause of death by a simple majority (Popular Voting). Sensitivity analyses revealed that selecting the prediction made by RF in cases where all three models disagreed was preferable, and this ensemble method was adapted accordingly. Findings The Popular Voting method performed better in cause-specific mortality fraction accuracy than did any individual model alone for adults, children, and neonates, and performed better in chance-corrected concordance than did any individual model except SSP in adults. The three component models disagreed in 16% of all cases, and unanimously agreed in 47% of cases. Interpretation As VA continues to be an effective source of data for estimating cause of death, accurate and inexpensive methods for analysing VA interview responses are increasingly important. The recent development of the three highly accurate PHMRC computational models allows for the option of a meta-model such as the ensemble introduced here. This ensemble model for VA achieves superior performance, and could be applied to other VA samples to accurately assess the relative mortality burden from a variety of diseases and injuries. Funding Population Health Metrics Research Consortium.

Worldwide data on causes of death: a systematic assessment of quality and availability of vital registration

Abstract Background Timely, accurate, and unbiased data are essential to evidence-based global health policy and research. Perhaps the most widely applied data pertain to causes of death (CoD), a highly informative indicator of population health. Consequently, an accurate understanding of the availability and quality of vital registration (VR) systems is indispensable to health policy makers and researchers alike. Using the CoD database from the Global Burden of Disease 2010 Study, we measured the quality of available VR data for 187 countries worldwide. Methods Using 2555 site-years of VR data covering 126 countries from 1980 to 2010, we developed a composite index of data quality that incorporated completeness, level of detail, pattern of misclassification of CoD, deaths of unknown age or sex, deaths coded to medically impossible CoD given age or sex, and timeliness of data reporting. This index was applied to the CoD database, providing estimates of overall data quality for 126 countries. Findings Data from VR have improved over time, but vary between and within regions. 39 countries currently have data from 2010 available, and 76 countries have data from 2009. Generally, countries from North America, Europe, Central America and high-income Asia–Pacific were found to have the highest quality data, while countries from south Asia and east Asia were estimated to have lower quality VR data. Further, many countries from sub-Saharan Africa and southeast Asia still have no available VR data. Considerable heterogeneity in VR quality was found in the Caribbean and eastern Europe/central Asia. Countries from the north Africa/Middle East region and Latin America demonstrated substantial improvements in data quality since 1990. Interpretation Metrics about the quality and availability of data from VR systems have important implications for policy and research. With improved information about where death certificates are reliably filled out and reported and where they are not, data collection efforts and health information system strengthening can be focused on areas of high need. High-quality and timely data on population health is critical as research to inform evidence-based policies and funding decisions continues to be of increasing importance in global health. Funding Bill & Melinda Gates Foundation.

Assessing the relationship between vital registration systems, health outcomes, and development: a multivariate analysis

Abstract Background A rapid assessment (RA) tool was developed to rapidly and inexpensively identify key areas of weakness and to yield an overall performance score for vital registration (VR) systems. This standardised tool has now been applied in more than 70 countries, but there has been no formal evaluation of its validity or relationship with measures of socioeconomic development. Similarly, there is very little empirical research on whether better VR systems are associated with better health outcomes. We aimed to compare the RA scores to independent assessments of VR completeness and data quality; explore the relationship between the RA results and selected national socioeconomic indicators; and investigate how the RA scores correlate with key mortality indicators, controlling for levels of national income and other variables. Methods We regressed the overall scores (%) of the RA from 70 countries on a series of development indicators, including national income and education. We measured the correlation between overall and sub-component RA scores against alternative estimates of national VR completeness and quality. We regressed RA scores against child and adult mortality, controlling for national income. Findings The RA scores were consistent with external measures of completeness and quality and were robustly correlated with a number of indicators of socioeconomic development, confirming their predictive value for assessing VR systems. Completeness of civil registration was associated with better health outcomes, independent of level of national income. Interpretation Our research demonstrates the utility of RA as an overall system indicator, as well as its predictive validity in relation to development indicators. VR is not only a by-product of development that will come about organically along with economic, governance, and political maturity, but also contributes directly and indirectly to desirable development outcomes, including better health. The close correlation between assessment scores with widely available development indices enhances the utility of the tool for guiding civil registration and vital statistics development strategies. The correlation between VR systems functionality and key health and development outcomes supports our hypothesis that VR is not merely a source of data but is a driver of development and improved health status in its own right. Funding Australian Overseas Aid.