Validation of verbal autopsy to determine the cause of 137 neonatal deaths in Karachi, Pakistan
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Verbal autopsy (VA) aims to estimate a community's mortality experience in the absence of contact with formal registration or health care systems. Application of VA to neonatal deaths is problematic as the agonal phase of a neonatal death tends to be indistinct. This is the first attempt to validate the technique exclusively on newborns who died. Seriously ill neonates (n = 137) were enrolled from the Civil Hospital, Karachi, Pakistan, between 31 October 1993 and 31 July 1994. All died as newborns, and caregivers were interviewed at home 3-230 days later. Surveillance physicians completed case questionnaires in the hospital, and investigator physicians assigned the main and associated causes of death using clinical criteria. Field questionnaires including a verbatim open-ended history, and syndrome modules were completed by a field worker, and investigator physicians again assigned the main and associated causes of death based on three diagnostic methods: verbatim alone, modules alone and verbatim and modules combined. We assessed the validity of VA by comparing field against hospital diagnoses by diagnostic (verbatim vs. modules vs. both) and analytic method (main vs. any diagnosis). VA identified at least one diagnosis accurately in 71% of the newborns. VA underdiagnosed low birthweight and prematurity in the field. Verbatim and modules diagnostic method comparing any field against main hospital diagnoses revealed high sensitivities for too early/too small syndrome (90%) and neonatal tetanus (84%). VA correctly identified some important causes of neonatal death in the field. Assigning multiple diagnoses using both open- and closed-ended questions increases the likelihood of correct ascertainment.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it