Abstract
Physician certification is the most widely used method for interpreting verbal autopsy (VA), yet physicians correctly determine cause of death less than half of the time, according to new research by IHME and the University of Queensland as part of the Population Health Metrics Research Consortium (PHMRC).
Research objective
Physician-certified verbal autopsy (PCVA) is the most widely used approach for analysis of VA. PCVA involves physicians reviewing reported signs and symptoms collected in VA and assigning a cause of death based on that VA. This study is part of ongoing work by IHME to determine the most accurate and efficient methods of predicting causes of death using VA.
This study was designed to explore the performance of PCVA, comparing the method to over 12,000 validated deaths from the PHMRC gold standard VA validation study. The PHMRC has undertaken the five-year study to develop a range of new analytical methods for VA and test these methods using data collected at six sites in four countries (Mexico, Tanzania, India, and the Philippines).
Research findings
Physicians varied in their ability to assign an accurate cause of death. The accuracy of diagnosis improved greatly when physicians were given information on household recall of health care experience. Household recall of health care experience includes any information the caretaker has about the patient's medical treatment, including whether health workers provided documentation for the cause of hospitalization or cause of death.
Without health care experience, physicians identified the correct cause less than 30% of the time for adults and neonates, and 36% of the time for children. Providing the physicians with health care experience information improved performance for adults to 45% and for children to 48%. For neonates, performance only improved to 33%.
PCVA does well for a number of injuries, including violence, road traffic accidents, drowning, fires, falls, and bite of a venomous animal. Accuracy of cause of death assignment was lower when physicians from other countries reviewed the verbal autopsy, indicating that a physician's familiarity with the prevalence of diseases in an area may improve the ability to determine cause of death.
Physicians estimated cause-specific mortality fractions, the fraction of all deaths in a population due to a specific cause, with considerable error for adults, children, and neonates. In all three age groups, cause-specific mortality fraction accuracy improved when household recall of health care experience was available.
Analytical approach
This study used a multisite sample of 12,542 VAs collected as part of the PHMRC gold standard VA validation study. Certification was performed by 24 physicians in four sites, and the assignment of VA was random and blinded. Each VA was certified by one physician. Half of the VAs were reviewed by a different physician with household recall of health care experience included. The physicians’ completed death certificates were processed for automated coding of the underlying cause of death according to the International Classification of Diseases (10th version, ICD-10). PCVA was compared to gold standard cause of death assignment based on strictly defined clinical diagnostic criteria that were collected as part of the PHMRC gold standard VA study.
Policy implications
The challenge for physicians to assign an accurate cause of death on the basis of the recall of signs, symptoms, and health care experience raises questions about the accuracy of medical certification of deaths that occur outside of a health facility.
Given that VA remains the global standard for assessing causes of death and prioritizing health interventions in areas lacking complete vital registration systems, it is essential to develop analytical methods that are low-cost, quick to implement, and consistently accurate. PCVA meets none of these criteria, and yet it is still the most widely implemented method for analysis of VA today. PCVA requires considerable time and cost, and it requires physicians to stop servicing immediate health needs in a population to review VAs.
This research highlights the importance and urgency of developing better methods to more reliably analyze past and future VA to obtain the highest quality mortality data from populations without reliable death certification.