2004年-世界发展银行全球_The_Determinants_of_Child_Health_and_Nutrition___A_Meta-analysis_62页_2mb

Summary of "The Determinants of Child Health and Nutrition: A Meta-Analysis"

Core Content

This paper presents a meta-analysis of the determinants of child health and nutrition, focusing on infant and child mortality and nutritional status. The study aims to identify which factors are most influential in shaping these outcomes and which are most amenable to policy intervention.

Main Viewpoints

1. Theoretical Frameworks

• Mosley-Chen and UNICEF Frameworks: These frameworks distinguish between immediate causes (e.g., lack of food, poor health facilities) and underlying factors (e.g., family income, education, cultural norms). Some variables, like maternal education, can mediate between these causes.
• Classification of Determinants: Determinants are categorized into three groups:
  • Child-specific: such as age, sex, birth order.
  • Household characteristics: including income, expenditure, and access to health services.
  • Community characteristics: such as environmental quality, availability of clean water, and service provision.
• Economic Perspective: The household utility maximization model, adapted from Becker (1981) and Currie (2000), is used to understand how households allocate resources to health and nutrition. This model shows that health and nutrition outcomes are influenced by past health status, current income, food prices, and environmental conditions.

2. Data and Methodology

• Dependent Variables: Infant and child mortality are modeled using dichotomous variables, with probit/logistic or hazards models used. Nutritional status is measured using anthropometric indicators such as height for age (HFA), weight for height (WFH), and height for weight (HFW).
• Z-Scores as Standardized Measures: HFA is recommended as the most appropriate measure for analyzing child nutrition. Z-scores are calculated by standardizing the anthropometric measurements against a reference population (e.g., WHO standards), with thresholds set at -2 and -3 standard deviations for malnutrition.
• Meta-Analysis Approach: The study uses meta-analysis to combine results from multiple regression studies. It calculates the combined t-statistic by summing individual t-statistics and dividing by the square root of the number of studies. However, the method has limitations, such as the assumption that structural models are sufficiently similar across countries and that t-statistics are independent.

3. Estimation Methods and Limitations

• Estimation Techniques: The study includes both hazards models and logit/probit models. Due to the non-comparability of t-statistics from binary models, only linear regression models are used for meta-analysis.
• Exclusion Criteria: Studies not using normalized z-scores, those with non-linear models, and those using weight-for-height measures are excluded due to issues with standardization and sensitivity to short-term fluctuations.
• Population Segmentation: Most studies segment the population by gender or urban/rural location. The maximum age of children considered is generally 60 months, as beyond that, genetic factors may dominate nutritional influences, making HFA less meaningful.

Key Information

Data Sources

• DHS and LSMS Surveys: These are the primary sources of data, especially for nutrition analysis. They provide detailed information on child health, immunization, and anthropometric measurements.
• Data Availability: While many studies use cross-sectional data, some follow children over time (e.g., Fedorov and Sahn, 2003). Mortality data are typically collected from birth histories in household surveys.

Regional Analysis

• Mortality Studies: More studies are available for South and East Asia and Latin America than for Africa. The analysis includes studies from various regions, but some areas (e.g., Europe and Central Asia) have limited data.
• Nutrition Studies: Sub-Saharan Africa has the most studies on child nutrition, while other regions have fewer. Methodological differences across regions (e.g., pooling vs. separate analysis) are noted.

Determinants of Child Health and Nutrition

• Income: A key determinant, with a positive effect on nutrition and a negative effect on mortality. However, income data are not always available, and wealth indices are often used instead.
• Maternal Education: Influences the effectiveness of health interventions and the allocation of household resources.
• Access to Health Services: Plays a crucial role in child health outcomes, especially in the context of immediate health risks.
• Environmental Factors: Such as clean water availability and pollution levels, are important in determining child health and nutrition.
• Household Size and Composition: May influence the allocation of resources and the overall health and nutritional status of children.
• Cultural Norms: Can affect the distribution of resources within the household, particularly with respect to gender biases.

Policy Relevance

• While variables like geographical dummies are statistically significant, they have limited policy relevance as they do not explain the underlying reasons for differences in child health and nutrition.
• The study emphasizes the importance of child-specific and household-level factors in shaping health and nutrition outcomes.

Conclusion

The meta-analysis highlights the complex interplay between child-specific, household, and community-level factors in determining child health and nutrition. It underscores the importance of income, maternal education, access to health services, and environmental conditions. The study also notes the limitations of using aggregated data and the need for careful interpretation of meta-analysis results due to potential correlations and differences in model specifications across studies.