Skip to main content

Table 1 Key features of the Million Death Study

From: Performance criteria for verbal autopsy-based systems to estimate national causes of death: development and application to the Indian Million Death Study

Feature Purpose
Random sample of deaths surveyed Ensures results are representative of India (based on rural and urban strata for major states, and at the state level for smaller states)
Continuous enumeration of deaths and births Ensures follow-up of the same houses to enable prospective analyses of risk factors (such as education, smoking and alcohol), and familiarity by households to the SRS field staff
3% to 5% random household resample of deaths by independent team Quality check on the reliability of data, and is a disincentive for faulty field work
Structured survey questions, half-page local language narrative, and guiding cardinal symptom lists Guides surveyors to fully capture chronology of key symptoms by age group, so as to aid physician diagnosis
Extraction of VA field data into web-based reports for coding Concise reports increase speed and efficiency of coding, custom extraction of data retains confidentiality
Independent, anonymous and random physician double coding (stratified only by language) Increases cross-state comparability (in particular for about half the records which are recorded in Hindi or English), and decreases local biases in coding
Web-based centralized medical coding application, with logical checks, clinical guidelines, and differential diagnoses Coding application with a user interface which includes searchable ICD-10 codes, standardised clinical guidelines and differential diagnoses, age/sex restrictions (for example, no cervical cancer in males, or senility before old age), and highlighting of keywords; increases the speed, repeatability, and quality of coding versus a paper-based system
Reconciliation and adjudication stages for coding disagreements Double coding with reconciliation and adjudication helps train new coders on correct use of coding, is a check on coding quality, and a disincentive for faulty coding
Financial incentives for quality of coding Payment is made per record that has cleared the reconciliation stage rather than per code assigned, thus decreasing incentives for random or faulty coding
Online recruitment and e-training for physicians ( Physicians train remotely as their schedule allows and are evaluated before entering the system; increases efficiency and quality of coding