2004-11-02-世界卫生组织-WHO_Technical_Consultation_on_Verbal_Autopsy_Tools_45页_384kb

Summary of WHO Technical Consultation on Verbal Autopsy Tools

I. Introduction

Verbal autopsy (VA) is a method used to determine the cause of death (COD) through interviews with family members or caregivers. It is particularly valuable in low- and middle-income countries where vital registration systems are inadequate. The WHO organized a technical meeting in Talloires, France, in November 2004, to discuss the development of a standardized VA tool. The meeting aimed to reach a consensus on core requirements for such a tool, based on the current use of VA tools and validation studies.

II. Methods

• Collection of VA tools: 36 Demographic Surveillance Sites (DSS) and researchers were contacted, and tools from WHO, CDC, WB, INDEPTH, and AMMP were also reviewed.
• Literature search: A comprehensive search was conducted using Medline, Popline, and the LSHTM verbal autopsy archives. Additionally, internet resources and NGO websites were reviewed.
• Exclusion criteria: Tools that are no longer in use were excluded from the analysis.
• Focus: The review included 18 adult VA questionnaires and 15 child-specific questionnaires.

III. Verbal Autopsy Tools - Overview

A. Questionnaires: Adults

• Common symptoms: Most questionnaires include common signs and symptoms such as fever, weight loss, and cough. These are frequently asked in over 90% of the tools.
• Differences: Some questionnaires are more structured, while others are less formal. The WHO standard questionnaire for children is discussed separately.
• Structure: Some tools combine questions for all age groups on one form, while others have separate sections for neonates and post-neonates.
• Open vs closed-ended sections: There is a trend to include both open and closed-ended sections in a standard tool. Open-ended sections allow for more detailed narratives, while closed-ended ones help in data processing and reduce bias.
• Sequence and layout: A logical sequence of questions is important for accuracy. Filter questions help reduce respondent fatigue and improve data quality.
• Pretesting and language: Pretesting is essential to ensure clarity and reduce ambiguity. Attention to local language and biomedical concepts is necessary to improve validity and cultural relevance.

B. Questionnaires: Children

• Standard tool: WHO developed a standard VA questionnaire for children in 1999, which is being updated.
• Structure: The tool is divided into neonate and post-neonate sections. Some questionnaires focus on birth-related events, while others emphasize disease symptoms.
• Age criteria: There is inconsistency in age ranges covered, with some tools for children up to 12 years and others for 29 days to five years.
• Consensus: A standard VA tool should include questionnaires for at least three age groups to ensure comprehensive coverage.

IV. Mortality Classification Systems

• Frameworks: COD lists are based on either organ systems (e.g., ICD-10) or pathophysiological mechanisms.
• Content and structure: Some systems are more comprehensive, while others are limited in scope. A few do not follow a clear framework.
• Validation need: Differences in content highlight the importance of validation studies to ensure accurate COD assignment.
• User-friendliness: The layout and clarity of classification systems affect inter- and intra-coder reliability. Clear and user-friendly systems reduce the risk of misclassification.

V. Deriving Causes of Death

A. Physician Review

• Dominant method: Physician review is the most commonly used method to derive COD. It involves doctors using clinical judgment to assign causes based on VA data.
• Limitations: It is criticized for being subjective and unreliable. However, reliability is often high due to the reviewers’ familiarity with local epidemiological patterns.

B. Data-Derived Algorithms

• Alternative methods: These include logistic regression, decision trees, and artificial neural networks.
• Use cases: The choice of method depends on the purpose of the analysis, such as comparing mortality patterns or formulating interventions.
• Challenges: These methods are not yet widely applicable due to complexity and the need for specialized software and training.

VI. Validation Studies

• Definition: Validation studies are essential to assess the accuracy of VA tools.
• Issues: These include variability in data quality, potential for bias, and the need for field testing of language and biomedical concepts.
• Children and adults: Validation studies for both age groups were reviewed, highlighting the importance of context-specific data.

VII. The Expanding Role of Verbal Autopsy

• Applications: VA is increasingly used in public health planning, refugee camps, and demographic surveillance.
• Validation studies: Tables in the report summarize validation studies in children, neonates, and adults, emphasizing the need for further research and standardization.

VIII. Conclusion

• Need for standardization: A standardized VA tool is essential to improve data quality and comparability.
• Balancing structure and flexibility: The inclusion of both open and closed-ended sections in a core tool is recommended.
• Validation and pretesting: These are critical steps to ensure reliability and accuracy.
• Incorporating equity and health: Including socioeconomic and health access information in VA questionnaires can provide better insights into health disparities.

IX. References and Annexes

• The report includes references to key studies and annexes detailing the strengths and limitations of VA coding methods, questionnaires, and mortality classification systems.
• These annexes provide detailed tables comparing the content and structure of various VA tools and classification systems.

Key Points

• Core requirements: A standardized VA tool is necessary to improve the reliability and comparability of COD data.
• Questionnaire design: Open and closed-ended sections should be included to balance detail and efficiency.
• Validation studies: These are crucial for improving the accuracy of VA tools and are needed in many regions.
• Equity considerations: Incorporating socioeconomic factors and health service access can enhance the understanding of mortality patterns and health inequalities.