Imperial College London

Jeff Eaton

Faculty of MedicineSchool of Public Health

Senior Lecturer in HIV Epidemiology
 
 
 
//

Contact

 

jeffrey.eaton

 
 
//

Location

 

Norfolk PlaceSt Mary's Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Eaton:2017:10.1097/QAD.0000000000001419,
author = {Eaton, JW and Bao, L},
doi = {10.1097/QAD.0000000000001419},
journal = {AIDS},
pages = {S61--S68},
title = {Accounting for nonsampling error in estimates of HIV epidemic trends from antenatal clinic sentinel surveillance},
url = {http://dx.doi.org/10.1097/QAD.0000000000001419},
volume = {31},
year = {2017}
}

RIS format (EndNote, RefMan)

TY  - JOUR
AB - Objectives: The aim of the study was to propose and demonstrate an approach to allowadditional nonsampling uncertainty about HIV prevalence measured at antenatal clinicsentinel surveillance (ANC-SS) in model-based inferences about trends in HIV incidenceand prevalence.Design: Mathematical model fitted to surveillance data with Bayesian inference.Methods: We introduce a variance inflation parameter s2in fl that accounts for theuncertainty of nonsampling errors in ANC-SS prevalence. It is additive to the samplingerror variance. Three approaches are tested for estimating s2in fl using ANC-SS andhousehold survey data from 40 subnational regions in nine countries in sub-Saharan, asdefined in UNAIDS 2016 estimates. Methods were compared using in-sample fit andout-of-sample prediction of ANC-SS data, fit to household survey prevalence data, andthe computational implications.Results: Introducing the additional variance parameter s2in fl increased the error variancearound ANC-SS prevalence observations by a median of 2.7 times (interquartilerange 1.9–3.8). Using only sampling error in ANC-SS prevalence (s2in fl ¼ 0), coverageof 95% prediction intervals was 69% in out-of-sample prediction tests. This increased to90% after introducing the additional variance parameter s2in fl. The revised probabilisticmodel improved model fit to household survey prevalence and increased epidemicuncertainty intervals most during the early epidemic period before 2005. Estimatings2in fl did not increase the computational cost of model fitting.Conclusions: We recommend estimating nonsampling error in ANC-SS as anadditional parameter in Bayesian inference using the Estimation and Projection Packagemodel. This approach may prove useful for incorporating other data sources such asroutine prevalence from Prevention of mother-to-child transmission testing into futureepidemic estimates.
AU - Eaton,JW
AU - Bao,L
DO - 10.1097/QAD.0000000000001419
EP - 68
PY - 2017///
SN - 0269-9370
SP - 61
TI - Accounting for nonsampling error in estimates of HIV epidemic trends from antenatal clinic sentinel surveillance
T2 - AIDS
UR - http://dx.doi.org/10.1097/QAD.0000000000001419
UR - http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=LinksAMR&KeyUT=WOS:000398210100008&DestLinkType=FullRecord&DestApp=ALL_WOS&UsrCustomerID=1ba7043ffcc86c417c072aa74d649202
UR - http://hdl.handle.net/10044/1/47920
VL - 31
ER -