Verbal Autopsies for Out-of-Hospital Infant Deaths in Zambia.

OBJECTIVES: In Zambia, a significant number of infants die in the community. It is hypothesized that delays in care contribute to many of these so-called "brought in dead" infants. METHODS: We analyzed free-text narratives from verbal autopsies, in which families narrate the final series of events leading to each infant's death. Using the 3-delays model framework and working iteratively to achieve consensus, we coded each narrative using NVivo software to identify, characterize, and quantify the contribution of delays and other factors to the fatal outcome. RESULTS: Verbal autopsies were collected from 230 families of brought in dead infants younger than 6 months of age. As many as 82.8% of infants had 1 or more delays in care. The most-common delay was in the family's decision to seek care (54.8%), even as severe symptoms were frequently described. Similarly, 27.8% of infants died en route to a health care facility. Delays in receiving adequate care, including infants dying while waiting in line at a clinic or during referral from a clinic to a hospital, occurred in 24.7% of infants. A third of infants had been previously evaluated by a clinician in the days before their death. CONCLUSIONS: Delays in care were the rule rather than the exception in this population of Zambian infants. Accessing care requires families to navigate significant logistic barriers, and balance complex forces in deciding to seek care. Strategies to avoid such delays could save many infants lives.

Can traditional birth attendants be trained to accurately identify septic infants, initiate antibiotics, and refer in a rural African setting?

BACKGROUND: Neonatal sepsis is a major cause of neonatal mortality. In populations with limited access to health care, early identification of bacterial infections and initiation of antibiotics by community health workers (CHWs) could be lifesaving. It is unknown whether this strategy would be feasible using traditional birth attendants (TBAs), a cadre of CHWs who typically have limited training and educational backgrounds. METHODS: We analyzed data from the intervention arm of a cluster-randomized trial involving TBAs in Lufwanyama District, Zambia, from June 2006 to November 2008. TBAs followed neonates for signs of potential infection through 28 days of life. If any of 16 criteria were met, TBAs administered oral amoxicillin and facilitated referral to a rural health center. RESULTS: Our analysis included 1,889 neonates with final vital status by day 28. TBAs conducted a median of 2 (interquartile range 2-6) home visits (51.4% in week 1 and 48.2% in weeks 2-4) and referred 208 neonates (11%) for suspected sepsis. Of referred neonates, 176/208 (84.6%) completed their referral. Among neonates given amoxicillin, 171/183 (93.4%) were referred; among referred neonates, 171/208 (82.2%) received amoxicillin. Referral and/or initiation of antibiotics were strongly associated with neonatal death (for referral, relative risk [RR] = 7.93, 95% confidence interval [CI] = 4.4-14.3; for amoxicillin administration, RR = 4.7, 95% CI = 2.4-8.7). Neonates clinically judged to be "extremely sick" by the referring TBA were at greatest risk of death (RR = 8.61, 95% CI = 4.0-18.5). CONCLUSION: The strategy of administering a first dose of antibiotics and referring based solely on the clinical evaluation of a TBA is feasible and could be effective in reducing neonatal mortality in remote rural settings.