The online version of this article (https://doi.org/10.1186/s12936-018-2460-9) contains supplementary material, which is available to authorized users.
Bhargavi Rao and Thomas S. Churcher are shared last authorship
The number of clinical cases of malaria is often recorded in resource constrained or conflict settings as a proxy for disease burden. Interpreting case count data in areas of humanitarian need is challenging due to uncertainties in population size caused by security concerns, resource constraints and population movement. Malaria prevalence in women visiting ante-natal care (ANC) clinics has the potential to be an easier and more accurate metric for malaria surveillance that is unbiased by population size if malaria testing is routinely conducted irrespective of symptoms.
A suite of distributed lag non-linear models was fitted to clinical incidence time-series data in children under 5 years and ANC prevalence data from health centres run by Médecins Sans Frontières in the Democratic Republic of Congo, which implement routine intermittent screening and treatment alongside intermittent preventative treatment in pregnancy. These statistical models enable the temporal relationship between the two metrics to be disentangled.
There was a strong relationship between the ANC prevalence and clinical incidence suggesting that both can be used to describe current malaria endemicity. There was no evidence that ANC prevalence could predict future clinical incidence, though a change in clinical incidence was shown to influence ANC prevalence up to 3 months into the future.
The results indicate that ANC prevalence may be a suitable metric for retrospective evaluations of the impact of malaria interventions and is a useful method for evaluating long-term malaria trends in resource constrained settings.
Additional file 1: Figure S1. A table of the values of four information criteria for different lag orders, used to determine the lag order of the VAR model. Figures S2–S4. Copies of Fig. 4 whereby the NENL model is fitted to data using a lag order of 1, 2, or 4 months. Figure S5. A copy of Fig. 4 using the NELL model described in the analysis rather than the NENL model.
Additional file 2. This dataset contains monthly time series data for all 5 MSF locations, including ANC visits, ANC prevalence, and clinical incidence in children under 5 years old.
WHO. World malaria report 2017. Geneva: World Health Organization; 2017.
USAID D program-. https://dhsprogram.com/. Accessed 16 Oct 2017.
Gregson S, Dharmayat K, Pereboom M, Takaruza A, Mugurungi O, Schur N, et al. Do HIV prevalence trends in antenatal clinic surveillance represent trends in the general population in the antiretroviral therapy era? The case of Manicaland, East Zimbabwe. AIDS. 2015;29:1845–53. CrossRefPubMedPubMedCentral
Granger CWJ. Investigating causal relations by econometric models and cross-spectral methods. Econometrica. 1969;37:424–38. CrossRef
Lütkepohl H. New introduction to multiple time series analysis. Berlin: Springer; 2005. CrossRef
Pfaff B. VAR, SVAR and SVEC models: implementation within R package vars. J Stat Softw. 2008;27:1–32. CrossRef
Ventzislav I, Lutz K. A practitioner’s guide to lag order selection for VAR impulse response analysis. Stud Nonlinear Dyn Econom. 2005;9:1–36. https://ideas.repec.org/a/bpj/sndecm/v9y2005i1n2.html. Accessed 17 Jul 2018.
- Using ante-natal clinic prevalence data to monitor temporal changes in malaria incidence in a humanitarian setting in the Democratic Republic of Congo
Thomas S. Churcher
- BioMed Central
Neu im Fachgebiet Innere Medizin
Meistgelesene Bücher aus der Inneren Medizin
Mail Icon II