Verbal Autopsy as a Tool for Defining Causes of Death in Specific Healthcare Contexts: Study of Applicability through a Traditional Literature Review

Bailo, Paolo; Gibelli, Filippo; Ricci, Giovanna; Sirignano, Ascanio

doi:10.3390/ijerph191811749

Cited by 8 publications

(4 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most studies evaluating the VA questionnaire, however, have not been validated against the gold standard (conventional diagnostic autopsy), which typically involves medical record interpretation 22 . Therefore, we aimed to define the sensitivity and specificity of the VA questionnaire in situations in which conventional autopsies cannot be performed due to biosafety issues or excess deaths at home without medical assistance 23 .…”

Section: Discussionmentioning

confidence: 99%

Accuracy of the Verbal Autopsy questionnaire in the diagnosis of COVID-19 deaths in a Brazilian capital

Campos,

Cutrim,

Cutrim

et al. 2024

Rev. Inst. Med. trop. S. Paulo

View full text Add to dashboard Cite

The Verbal Autopsy (VA) is a questionnaire about the circumstances surrounding a death. It was widely used in Brazil to assist in postmortem diagnoses and investigate excess mortality during the Coronavirus Disease 2019 (COVID-19) pandemic. This study aimed to determine the accuracy of investigating acute respiratory distress syndrome (ARDS) using VA. This is a cross-sectional study with prospective data collected from January 2020 to August 2021 at the Death Verification Service of Sao Luis city, Brazil. VA was performed for suspected COVID-19 deaths, and one day of the week was randomly chosen to collect samples from patients without suspected COVID-19. Two swabs were collected after death and subjected to reverse transcription-polymerase chain reaction (RT-PCR) for SARS-CoV-2 detection. Of the 250 cases included, the VA questionnaire identified COVID-19-related ARDS in 67.2% (52.98% were positive for COVID-19). The sensitivity of the VA questionnaire was 0.53 (0.45–0.61), the specificity was 0.75 (0.64–0.84), the positive predictive value was 0.81 (0.72–0.88), and the negative predictive value was 0.44 (0.36–0.53). The VA had a lower-than-expected accuracy for detecting COVID-19 deaths; however, because it is an easily accessible and cost-effective tool, it can be combined with more accurate methods to improve its performance.

show abstract

Section: Discussionmentioning

confidence: 99%

Accuracy of the Verbal Autopsy questionnaire in the diagnosis of COVID-19 deaths in a Brazilian capital

Campos,

Cutrim,

Cutrim

et al. 2024

Rev. Inst. Med. trop. S. Paulo

View full text Add to dashboard Cite

show abstract

“…Validation studies comparing VA assigned causes of death to hospital records have shown mixed results [ 15 , 16 , 43 ]. Therefore, though deaths assigned using VA are important in settings with weak systems for registering and determing causes of death, they should be interpreted with caution.…”

Section: Discussionmentioning

confidence: 99%

Verbal autopsy analysis of childhood deaths in rural Gambia

et al. 2023

View full text Add to dashboard Cite

Background In low-resource settings, it is challenging to ascertain the burden and causes of under-5 mortality as many deaths occur outside health facilities. We aimed to determine the causes of childhood deaths in rural Gambia using verbal autopsies (VA). Methodology We used WHO VA questionnaires to conduct VAs for deaths under-5 years of age in the Basse and Fuladu West Health and Demographic Surveillance Systems (HDSS) in rural Gambia between September 01, 2019, and December 31, 2021. Using a standardized cause of death list, two physicians assigned causes of death and discordant diagnoses were resolved by consensus. Results VAs were conducted for 89% (647/727) of deaths. Of these deaths, 49.5% (n = 319) occurred at home, 50.1% (n = 324) in females, and 32.3% (n = 209) in neonates. Acute respiratory infection including pneumonia (ARIP) (33.7%, n = 137) and diarrhoeal diseases (23.3%, n = 95) were the commonest primary causes of death in the post-neonatal period. In the neonatal period, unspecified perinatal causes of death (34.0%, n = 71) and deaths due to birth asphyxia (27.3%, n = 57) were the commonest causes of death. Severe malnutrition (28.6%, n = 185) was the commonest underlying cause of death. In the neonatal period, deaths due to birth asphyxia (p-value<0.001) and severe anaemia (p-value = 0.03) were more likely to occur at hospitals while unspecified perinatal deaths (p-value = 0.01) were more likely to occur at home. In the post-neonatal period, deaths due to ARIP (p-value = 0.04) and diarrhoeal disease (p-value = 0.001) were more likely to occur among children aged 1–11 months and 12–23 months respectively. Conclusion According to VA analysis of deaths identified within two HDSS in rural Gambia, half of deaths amongst children under-5 in rural Gambia occur at home. ARIP and diarrhoea, and the underlying cause of severe malnutrition remain the predominant causes of child mortality. Improved health care and health-seeking behaviour may reduce childhood deaths in rural Gambia.

show abstract

“…Using death certificates, some surveillance teams have identified drug overdose deaths from cause of death fields available before the final cause of death coding [13][14][15]. Other teams have used text from verbal autopsies, which describe interviews with witnesses or relatives, as opposed to forensic autopsies, which detail findings from extensive physical examinations and toxicology tests [16,17]. Text from forensic autopsy reports has been used for the automatic classification of causes of death unrelated to drug overdoses [18].…”

Section: Introductionmentioning

confidence: 99%

Using Natural Language Processing to Predict Fatal Drug Overdose From Autopsy Narrative Text: Algorithm Development and Validation Study

Tang¹,

Korona-Bailey²,

Zaras³

et al. 2023

JMIR Public Health Surveill

View full text Add to dashboard Cite

Background Fatal drug overdose surveillance informs prevention but is often delayed because of autopsy report processing and death certificate coding. Autopsy reports contain narrative text describing scene evidence and medical history (similar to preliminary death scene investigation reports) and may serve as early data sources for identifying fatal drug overdoses. To facilitate timely fatal overdose reporting, natural language processing was applied to narrative texts from autopsies. Objective This study aimed to develop a natural language processing–based model that predicts the likelihood that an autopsy report narrative describes an accidental or undetermined fatal drug overdose. Methods Autopsy reports of all manners of death (2019-2021) were obtained from the Tennessee Office of the State Chief Medical Examiner. The text was extracted from autopsy reports (PDFs) using optical character recognition. Three common narrative text sections were identified, concatenated, and preprocessed (bag-of-words) using term frequency–inverse document frequency scoring. Logistic regression, support vector machine (SVM), random forest, and gradient boosted tree classifiers were developed and validated. Models were trained and calibrated using autopsies from 2019 to 2020 and tested using those from 2021. Model discrimination was evaluated using the area under the receiver operating characteristic, precision, recall, F1-score, and F2-score (prioritizes recall over precision). Calibration was performed using logistic regression (Platt scaling) and evaluated using the Spiegelhalter z test. Shapley additive explanations values were generated for models compatible with this method. In a post hoc subgroup analysis of the random forest classifier, model discrimination was evaluated by forensic center, race, age, sex, and education level. Results A total of 17,342 autopsies (n=5934, 34.22% cases) were used for model development and validation. The training set included 10,215 autopsies (n=3342, 32.72% cases), the calibration set included 538 autopsies (n=183, 34.01% cases), and the test set included 6589 autopsies (n=2409, 36.56% cases). The vocabulary set contained 4002 terms. All models showed excellent performance (area under the receiver operating characteristic ≥0.95, precision ≥0.94, recall ≥0.92, F1-score ≥0.94, and F2-score ≥0.92). The SVM and random forest classifiers achieved the highest F2-scores (0.948 and 0.947, respectively). The logistic regression and random forest were calibrated (P=.95 and P=.85, respectively), whereas the SVM and gradient boosted tree classifiers were miscalibrated (P=.03 and P<.001, respectively). “Fentanyl” and “accident” had the highest Shapley additive explanations values. Post hoc subgroup analyses revealed lower F2-scores for autopsies from forensic centers D and E. Lower F2-score were observed for the American Indian, Asian, ≤14 years, and ≥65 years subgroups, but larger sample sizes are needed to validate these findings. Conclusions The random forest classifier may be suitable for identifying potential accidental and undetermined fatal overdose autopsies. Further validation studies should be conducted to ensure early detection of accidental and undetermined fatal drug overdoses across all subgroups.

show abstract

Verbal Autopsy as a Tool for Defining Causes of Death in Specific Healthcare Contexts: Study of Applicability through a Traditional Literature Review

Cited by 8 publications

References 25 publications

Accuracy of the Verbal Autopsy questionnaire in the diagnosis of COVID-19 deaths in a Brazilian capital

Accuracy of the Verbal Autopsy questionnaire in the diagnosis of COVID-19 deaths in a Brazilian capital

Verbal autopsy analysis of childhood deaths in rural Gambia

Using Natural Language Processing to Predict Fatal Drug Overdose From Autopsy Narrative Text: Algorithm Development and Validation Study

Contact Info

Product

Resources

About