Diagnostic accuracy of a simplified minimally invasive tissue sampling protocol for stillbirths in low-resource...

Stillbirth rates remain unacceptably high in low- and middle-income countries (LMICs). Understanding the causes of death (CoD) is mandatory to develop effective strategies to reduce this high mortality. Minimally invasive tissue sampling (MITS) is a promising alternative to conventional autopsy (CA) but its validation in stillbirths remains limited. Existing evidence indicates that most samples of conventional MITS (c-MITS) lack diagnostic relevance in stillbirths. This study aimed to validate c-MITS against CA in stillbirths and design and assess a cost-efficient, simplified MITS (s-MITS) protocol.

The study comprised two subsets of stillbirths occurring at Maputo Central Hospital, Mozambique. The CaDMIA-Plus cohort (Cause of Death investigation using Minimally Invasive Autopsy, n=90; 2017-2018), in which both c-MITS and CA were performed, was used to validate c-MITS against the gold standard and to determine the diagnostic value of each sample and design a s-MITS. The MIBio cohort (Mortality Identification Biomarkers, n=98; 2021-2022), which included only s-MITS, was used to evaluate the performance and cost of the s-MITS in comparison to c-MITS in an independent cohort.

Almost perfect overall agreement (Kappa=0.82) was observed between the c-MITS and CA-attributed CoD. Lung and placenta samples were identified as the most informative in c-MITS. When using only lung and placenta results to model an s-MITS, substantial agreement (Kappa=0.79) was found between the s-MITS-derived CoD and those attributed by CA. Similar CoD distributions were observed when applying the s-MITS to the MIBio cohort, while costs were reduced by 55.7%. The leading CoDs were primarily related to maternal conditions and pregnancy complications (70.0-72.4%) and infectious diseases (25.6-27.6%).

c-MITS is a simpler and cost-effective alternative to CA for determining CoD in stillbirths. s-MITS has a diagnostic accuracy similar to that of c-MITS, while significantly reducing costs, making it adequate for implementation in routine clinical practice in LMICs.