Verbal autopsy completion rate and factors associated with undetermined cause of death in a rural resource-poor setting of Tanzania

Abstract

BackgroundVerbal autopsy (VA) is a widely used tool to assign probable cause of death in areas with inadequate vital registration systems. Its uses in priority setting and health planning are well documented in sub-Saharan Africa (SSA) and Asia. However, there is a lack of data related to VA processing and completion rates in assigning causes of death in a community. There is also a lack of data on factors associated with undetermined causes of death documented in SSA. There is a need for such information for understanding the gaps in VA processing and better estimating disease burden.ObjectiveThe study's intent was to determine the completion rate of VA and factors associated with assigning undetermined causes of death in rural Tanzania.MethodsA database of deaths reported from the Ifakara Health and Demographic Surveillance System from 2002 to 2007 was used. Completion rates were determined at the following stages of processing: 1) death identified; 2) VA interviews conducted; 3) VA forms submitted to physicians; 4) coding and assigning of cause of death. Logistic regression was used to determine factors associated with deaths coded as "undetermined."ResultsThe completion rate of VA after identification of death and the VA interview ranged from 83% in 2002 and 89% in 2007. Ninety-four percent of deaths submitted to physicians were assigned a specific cause, with 31% of the causes coded as undetermined. Neonates and child deaths that occurred outside health facilities were associated with a high rate of undetermined classification (33%, odds ratio [OR] = 1.33, 95% confidence interval [CI] (1.05, 1.67), p = 0.016). Respondents reporting high education levels were less likely to be associated with deaths that were classified as undetermined (24%, OR = 0.76, 95% CI (0.60, -0.96), p = 0.023). Being a child of the deceased compared to a partner (husband or wife) was more likely to be associated with undetermined cause of death classification (OR = 1.35, 95% CI (1.04, 1.75), p = 0.023).ConclusionEvery year, there is a high completion rate of VA in the initial stages of processing; however, a number of VAs are lost during the processing. Most of the losses occur at the final step, physicians' determination of cause of death. The type of respondent and place of death had a significant effect on final determination of the plausible cause of death. The finding provides some insight into the factors affecting full coverage of verbal autopsy diagnosis and the limitations of causes of death based on VA in SSA. Although physician review is the most commonly used method in ascertaining probable cause of death, we suggest further work needs to be done to address the challenges faced by physicians in interpreting VA forms. There is need for an alternative to or improvement of the methods of physician review.

Investigation of ill-defined causes of death: assessment of a program's performance in a State from the Northeastern region of Brazil

“…In this study, the use of the VA3 form in home investigations represented an important gain for the definition of the underlying cause of death, since 80% were reclassified (using VA alone or together with the MoH form). Studies from other countries in which VA was used for the definition of cause of death had lower propotions, probably because the definition of CoD was the responsibility of two certifying physicians and, in case of disagreement, of a third one 17 .…”

Main causes of death in Dande, Angola: results from Verbal Autopsies of deaths occurring during 2009–2012

“…In case of discordance between the two physicians, the questionnaire was reviewed by a third physician and the majority rule was applied. Otherwise, if the third physician determined a different cause of death, the case was registered as indeterminate [ 24 , 25 ].…”

Trends in Causes of Adult Deaths among the Urban Poor: Evidence from Nairobi Urban Health and Demographic Surveillance System, 2003–2012

“…The InterVA model has been widely tested in several settings, producing cost-effective, quick, and comparable results. [17][18][19][36][37][38][39][40] While challenges remain concerning the efficacy of InterVA, particularly its lower sensitivity relative to other electronic methods, 20 this approach to interpreting VA data remains a useful, easy to use tool, and alternative to the expensive and timeconsuming practice of physician coding in countries without vital registration and medical certification of causes of death. 21 In order to utilize InterVA-4 to interpret our VA data, we first captured the VA questionnaires electronically into an SQL database and then converted the data into a comma separated value (.csv) file.…”