학술논문
Prediction Model with Validation for Polioseronegativity in Malnourished Children from Poliomyelitis Transmission High-Risk Area of the Democratic Republic of the Congo (DRC).
Document Type
Article
Author
Mwamba, Guillaume Ngoie; Kabamba, Michel; Hoff, Nicole A; Mukadi, Patrick K; Musene, Kamy Kaminye; Gerber, Sue K; Halbrook, Megan; Sinai, Cyrus; Fuller, Trevon; Voorman, Arie; Mawaw, Paul Makan; Numbi, Oscar Luboya; Wemakoy, Emile Okitolonda; Mechael, Patricia N; Tamfum, Jean Jacques Muyembe; Mapatano, Mala Ali; Rimoin, Anne W; Dikassa, Paul-Samson Lusamba
Source
Subject
*POLIOMYELITIS vaccines
*POLIO
*RECEIVER operating characteristic curves
*PREDICTION models
*MODEL validation
*
*
*
*
Language
ISSN
1179-7266
Abstract
Background: Malnutrition is identified as a risk factor for insufficient polio seroconversion in the context of a vaccine-derived poliovirus (VDPV) outbreak-prone region. In the Democratic Republic of Congo (DRC), underweight decreased from 31% (in 2001) to 26% (in 2018). Since 2004, VDPV serotype 2 outbreaks (cVDPV2) have been documented and were geographically limited around the Haut-Lomami and Tanganyika Provinces. Methods: To develop and validate a predictive model for poliomyelitis vaccine response in malnourished infants, a cross-sectional household study was carried out in the Haut-Lomami and Tanganyika provinces. Healthy children aged 6 to 59 months (n=968) were enrolled from eight health zones (HZ) out of 27, in March 2018. We performed a bivariate and multivariate logistics analysis. Final models were selected using a stepwise Wald method, and variables were selected based on the criterion p < 0.05. The association between nutritional variables, explaining polio seronegativity for the three serotypes, was assessed using the receiver operating characteristic curve (ROC curve). Results: Factors significantly associated with seronegativity to the three polio serotypes were underweight, non-administration of vitamin A, and the age group of 12 to 59 months. The sensitivity was 10.5%, and its specificity was 96.4% while the positive predictive values (PPV) and negative (PNV) were 62.7% and 65.3%, respectively. We found a convergence of the curves of the initial sample and two split samples. Based on the comparison of the overlapping confidence intervals of the ROC curve, we concluded that our prediction model is valid. Conclusion: This study proposed the first tool which variables are easy to collect by any health worker in charge of vaccination or in charge of nutrition. It will bring on top, the collaboration between the Immunization and the Nutritional programs in DRC integration policy, and its replicability in other low- and middle-income countries with endemic poliovirus. [ABSTRACT FROM AUTHOR]