Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/279957585
Article in Transactions of the Royal Society of Tropical Medicine and Hygiene · July 2015
DOI: 10.1093/trstmh/trv049
CITATIONS READS
4 378
11 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Maha Adel Elbadry on 10 July 2015.
ORIGINAL ARTICLE
Spatial patterns and determinants of malaria infection during
pregnancy in Zambia
Mulakwa Kamuliwoa, Karen E. Kirkb, Emmanuel Chandac, Maha A. Elbadryd,e, Jailos Lubindaf,
Thomas A. Weppelmannd,e, Victor M. Mukonkag, Wenyi Zhangh, Gabriel Mushingei, Mercy Mwanza-Ingwea
and Ubydul Haqued,j,*
a
Ministry of Health, National Malaria Control Centre, Lusaka, Zambia; bIndependent Consultant, Baltimore, MD, USA; cVector Control
Specialist/Consultant, 11 Granite Street, Plot 33421/917 Kamwa South, Lusaka, Zambia; dEmerging Pathogens Institute, University of
Florida, Gainesville, FL, USA; eDepartment of Environmental and Global Health, College of Public Health and Health Professions, University
of Florida, Gainesville, FL, USA; fMacha Research Trust, Macha, Zambia; gDepartment of Public Health, Copperbelt University, School of
Background: Malaria in pregnancy (MiP) is a major concern in Zambia. Here we aim to determine the burden and
risk factors of MiP.
Methods: Monthly reported district-level malaria cases among pregnant women (count data) from January 2009
to December 2014 were obtained from the Zambian District Health Information System. Negative binomial regres-
sion model was used to investigate the associations between vector control tools, coverage of health care facilities,
transportation networks and population density. Data on MiP treatment were obtained from the 2012 Zambian
Malaria Indicator Survey. Yearly clusters of MiP were investigated using spatial statistics in ArcGIS v 10.1.
Results: The results indicated that MiP decreased in Zambia between 2010 and 2013. MiP was observed throughout
the year, but showed a strong seasonal pattern. Persistent hotspots of MiP were reported in the southeast and
northeast regions of Zambia, with districts that had better access to rail road and presence of water bodies asso-
ciated with decreased prevalence of MiP. Better indoor residual spraying and long-lasting insecticide-treated nets
coverage was demonstrated to be protective against MiP.
Conclusions: Mapping the distribution of MiP to track the future requirements for scaling up essential disease-
prevention efforts in stable hotspots can help the Zambian National Malaria Control Center to further develop
strategies to reduce malaria prevalence in this vulnerable sub-population.
# The Author 2015. Published by Oxford University Press on behalf of Royal Society of Tropical Medicine and Hygiene. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com.
1
M. Kamuliwo et al.
Malaria continues to be a large source of maternal mortality in Sources of data based on collection done between 2009
malaria endemic areas of Zambia.10,11 In 2007, the maternal and 2014
mortality rate for Zambia was estimated at 591 deaths per
Monthly district-level, aggregated malaria cases (count data)
100 000 live births.12 This number decreased to 315 per
from January 2009 to December 2014 among pregnant women
100 000 live births in 2013.13 Though overall maternal mortality
were obtained from the Zambian District Health Information
rates had declined by 4.1% (from 475 to 315 per 100 000 live
System (DHIS). DHIS data were collected at all heath facilities
births) between 2003 and 2013, malaria continued to be a
including public, private, faith-based hospitals, antenatal clinics,
major cause of mortality among pregnant women in the coun-
and rural health centers, in paper form and sent to the District
try.9,10 However, data from the 2007 Demographic Health
Health Office (DHO). The date of visit and results of malaria diag-
Survey (DHS) and the 2008 MIS found that approximately 94%
nosis among MiP were entered in an electronic database at DHO
of pregnant women received ANC during some point in their preg-
and transmitted to the provincial office for further processing and
nancy.6 WHO recommends effective case management to reduce
aggregation by the province. The consolidated provincial data
MiP and maternal mortality rates in high transmission areas
were then transmitted to the NMCC.
by improving access to intermittent preventive treatment in preg-
District level aggregated data on IRS coverage were collected
nancy (IPTp) using sulfadoxine-pyrimethamine (SP).14,23 Zambia’s
from the daily spray forms that were totaled for each year by the
IPTp treatment case management protocol recommends three
Study area
Malaria prevention, diagnosis and treatment
Zambia is located in Southern Africa, bordering eight other
of pregnant women
malaria-endemic countries that include Angola, Botswana,
Democratic Republic of Congo, Malawi, Mozambique, Namibia, Information regarding IPTp was obtained from the 2012 Zambian
Tanzania and Zimbabwe.8 In 2014, the population of Zambia MIS.24 In Zambia, during the first trimester of pregnancy, oral
was 13.2 million living within 752 614 km 2 with a population quinine was recommended for malaria treatment among preg-
density of 20 persons per square kilometer.15 Population at the nant women. In the absence of quinine, artemether-lumefantrine
district level for 2010 was collected from the Central Statistical (AL) was the prescribed treatment.3 During the second and third
Office in Lusaka, Zambia and projected for 2009 and 2011– trimester, use of artemisinin-based combination therapy (ACT),
2014 based on the 2000 and 2010 national censuses.16,17 While specifically AL, was the preferred treatment.3 SP was also the
the country is considered stable politically and economically, it is medicine of choice for IPTp (3 tablets) recommended monthly
also a low-income country with the average GNI per capita of after quickening (i.e., 16 weeks following the last menstrual
3810 USD (2013) and 60% of its population living in poverty cycle). The same adult treatment dose of SP (3 tablets) was
(2010).18 The average life expectancy in 2012 was 57 years.18 recommended monthly (at least 4 weeks apart) during the
Zambia is stratified into three malaria epidemiological zones: second and third trimesters at every scheduled ANC visit. The
stable transmission in the southeastern region of the country total number of doses recommended for the entire duration of
(parasite prevalence of .30%); unstable transmission in the pregnancy was three or more doses.3
northeastern and northwestern regions (parasite prevalence of
20%); and low-transmission in the southwestern region (parasite
prevalence ,5%).8 Zambia has a tropical climate with three sea- Spatial data
sons: rainy season (November to May during which time malaria Spatial data, including district boundary, geo-coded locations of
transmission peaks); cool, dry season (late May to August); and Zambia’s 1426 health facilities, lakes, rivers, roads and railroad, were
hot, dry season (September to November). Malaria prevalence is collected (updated until 2012) from the Ministry of Environment and
highest during the rainy season, however, malaria transmission Natural Resources. Distances were calculated from the centroid of
occurs throughout the year.4 Plasmodium falciparum is the dom- each district to the nearest road/rail station and categorized as
inant malaria parasite species (about 98%) in the country.19 follows: ≤2 km¼good access; .2 km¼limited access.
2
Transactions of the Royal Society of Tropical Medicine and Hygiene
Statistical analysis of MiP rates within districts, between nearest adjacent districts and
among all districts combined. The test statistic, local Moran’s index
Negative binomial regression analysis was completed to investi-
(I), a z-score, p-value, and outlier type were then calculated. Outlier
gate the association at district level between the variables for vec-
analysis tool classifies spatial outliers. To ensure this, the tool com-
tor control tools (e.g., LLINs and IRS) and variables for coverage of
putes a local Moran’s I value, a z-score, a p-value, and a code (high/
health clinics, access to road networks and passenger railways,
low) representing the cluster form for each feature (based on
and population densities (Table 1). These variables were used to
weighted features). The z-scores and p-values characterize the
determine risk factors of MiP. Statistical analyzes were performed
statistical significance of the calculated index values. Here outlier
with STATA version 13 (StataCorp, College Station, TX, USA).
types indicated whether the values observed between two neigh-
boring foci are more similar (a spatial clustering of either high or low
Spatial analysis values) or dissimilar (a spatial outlier) than would be expected at
random. An alpha level of 0.05 was used to determine if the spatial
Yearly maps showing geographic foci of MiP where transmission statistic was statistically significant.
intensity exceeds the average level were created for all MiP cases
between 2009 and 2014 using cluster and outlier analysis
(ArcGIS 10.1 software).25 This data was used to make comparisons Results
Figure 1. Malaria in pregnancy (MiP) in Zambia, 2009–2014. From the left axis the bar shows total malaria cases. From the right axis the line shows MiP.
This figure is available in black and white in print and in color at Transactions online.
3
M. Kamuliwo et al.
Figure 3. Hotspots of malaria in pregnancy in Zambia. This figure is available in black and white in print and in color at Transactions online.
railways, located closer to large water bodies (e.g., lakes and riv- The results of IPTp usage by pregnant women during their last
ers), (Figure 4) and/or high IRS and LLIN coverage, had signifi- birth (in the previous five years), based on the 2007 DHS survey,
cantly lower MiP compared with other districts (Table 2). showed that most pregnant women (85%) reported taking an
4
Transactions of the Royal Society of Tropical Medicine and Hygiene
antimalarial drug for prevention during their last pregnancy and Discussion
71% took two or more doses of IPTp. Regional variations were
observed. Pregnant women living in the Copperbelt (85%) and Approximately 1.5% of the total number of malaria cases
Eastern (85%) Provinces, where malaria prevalence is higher than reported in Zambia between 2009 and 2014 were MiP. There
the national average, had higher rates of receiving two doses of was a noted increase in confirmed MiP cases during this study per-
IPTp during pregnancy as compared to pregnant women in the iod, with seasonality observed. Districts having access to railways,
Southern (57%), Muchinga (54%) and Western (61%) Provinces waterbodies and greater coverage of IRS and LLINs were asso-
and where malaria prevalence rates are lower than the national ciated with lower MiP.
average. Over half of all pregnant women in the country (52%) Using Zambian DHIS data, the results of this study indicate
reported taking three or more doses of IPTp during their last that malaria prevalence among pregnant women continues
pregnancy. to be an issue. However, based on these findings, the prevalence
5
M. Kamuliwo et al.
of MiP (1.47%) in Zambia were much lower than the country’s the northeastern part of the country in Milenge district and in
overall malaria prevalence of 36% (4 738 375 yearly malaria the southeastern region of the country including the Chadiza,
cases on an average from 2009–2014 of Zambian with 13.2 Chipata, Katete, Mambwe and Petauke districts which falls within
million people).10 This study also confirmed that MiP is lower the high transmission region of the country.8 Regularly updated
in Zambia than in other African countries such as Malawi,27 MiP and risk maps33–45 using geographic information systems
Sudan,28,29 Kenya, Mali30 and Gabon.31,32 will help the NMCC in better control efforts.
Despite Zambia being a highly malaria endemic country, per- This study found that transportation networks, particularly
sistent and stable hotspots of MiP have been reported only in access to passenger railways and developed roads, and proximity
to large bodies of water within the district were associated with
reduced MiP. One reason for this association might be that
Table 2. Risk factors of malaria in pregnancy (MiP) in Zambia, these districts have better and faster access to healthcare
2009 to 2014
services.
Findings from this study also support a previous study in those
districts with better LLIN coverage contributed to lower MiP preva-
Risk factor PRR (95% CI) lence.8 A cross-sectional study on the use of LLINs in stable
malaria hotspots among pregnant women would be useful to
Table 3. The use of intermittent preventive treatment (IPTp) in pregnant women presented by region for Zambia between 2009 and 2014a
6
Transactions of the Royal Society of Tropical Medicine and Hygiene
This study has several limitations that accompany ecological 4 NMCC. Guidelines for the Diagnosis and Treatment of Malaria in
level analyses. Not all MiP cases were confirmed by either micros- Zambia. 4th ed. Lusaka: National Malaria Control Center; 2014. http://
copy or RDT. Both RDT and microscopy have limited sensitivity and www.nmcc.org.zm/files/GuidelinesonDiagnosisandTreatmento
specificity (by their inability to detect sub-patent malaria parasite fMalariainZambia_4thEd_2-24-14.pdf [accessed 31 March 2015].
thresholds) which impacts the number of MiP cases identified in 5 Mukonka VM, Chanda E, Haque U et al. High burden of malaria
this study. There is a potential for reporting bias due to the vari- following scale-up of control interventions in Nchelenge District,
ation in level of accuracy in reporting within different district Luapula Province, Zambia. Malar J 2014;13:153.
levels. In addition, there may be an underestimation of true preva- 6 President’s Malaria Initiative. Successes and Challenges for Malaria in
lence of MiP attributed to a potential misreporting by health cen- Pregnancy Programming: A Three-Country Analysis. Washington, DC:
ters. In addition, the different anti-malarial medications used to President’s Malaria Initiative; 2012. http://www.jhpiego.org/files/
treat MiP during this study period were not accessible and there- MIP%20Brief_Three%20Country_Final.pdf [accessed 31 March 2015].
fore their impact on MiP could not be analyzed. 7 President’s Malaria Initiative. Zambia. Washington, DC: President’s
Malaria Initiative; 2014. http://www.pmi.gov/docs/default-source/
default-document-library/country-profiles/zambia_profile.pdf?sfvrsn=14
Conclusions [accessed 31 March 2015].
Though MiP remains a burden in Zambia, the reduced prevalence 8 Eisele TP, Larsen DA, Anglewicz PA et al. Malaria prevention in
Ethical approval: Not required. 17 Central Statistical Office. Census of population and housing. Lusaka,
Zambia: Ministry of Finance and National Planning; 2011.
18 World Bank. 2015. World Development Indicators. Washington, DC:
References World Bank; 2015. http://databank.worldbank.org/data/views/
1 Luxemburger C, McGready R, Kham A et al. Effects of malaria during variableselection/selectvariables.aspx?source=World-Development-
pregnancy on infant mortality in an area of low malaria transmission. Indicators [accessed 22 March 2015].
Am J Epidemiol 2001;154:459–65. 19 WHO. 2014. Malaria. World Malaria Report 2014. Geneva: World Health
2 Schantz-Dunn J, Nour NM. Malaria and pregnancy: a global health Organization; 2014. Available at http://www.who.int/malaria/
perspective. Rev Obstet Gynecol 2009;2:186–92. publications/world_malaria_report_2014/en [accessed 31 March 2015].
3 WHO. Malaria. World malaria report 2013. Geneva: World Health 20 Kilian A, Byamukama W, Pigeon O et al. Evidence for a useful life of
Organization; 2013. http://www.who.int/malaria/publications/world_ more than three years for a polyester-based long-lasting insecticidal
malaria_report_2013/en/index.html [accessed 31 March 2015]. mosquito net in Western Uganda. Malar J 2011;10:299.
7
M. Kamuliwo et al.
21 Djalle D, Gody JC, Moyen JM et al. Performance of Paracheck-Pf, SD 36 Haque U, Sunahara T, Hashizume M et al. Malaria prevalence, risk
Bioline malaria Ag-Pf and SD Bioline malaria Ag-Pf/pan for diagnosis factors and spatial distribution in a hilly forest area of Bangladesh.
of falciparum malaria in the Central African Republic. BMC Infect Dis PLoS One 2011;6:e18908.
2014;14:109. 37 Haque U, Scott LM, Hashizume M et al. Modelling malaria treatment
22 Malaria Indicator Definitions. http://gamapserver.who.int/GlobalAtlas/ practices in Bangladesh using spatial statistics. Malar J 2012;11:63.
docs/Malaria/malariaindicatordefinitions.pdf [accessed 31 March 2015]. 38 Haque U, Hashizume M, Sunahara T et al. Progress and challenges to
23 Republic of Zambia Ministry of Health. National Malaria Communication control malaria in a remote area of Chittagong Hill Tracts, Bangladesh.
Strategy (2011–2015). Washington, DC: USAID. Malar J 2010;9:156.
24 NMCC. Zambia National Malaria Indicator Survey 2012. Lusaka: 39 Reid HL, Haque U, Roy S et al. Characterizing the spatial and temporal
National Malaria Control Center; 2012. http://www.nmcc.org.zm/ variation of malaria incidence in Bangladesh, 2007. Malar J
files/FullReportZambiaMIS2012_July2013_withsigs2.pdf [accessed 2012;11:170.
31 March 2015]. 40 Haque U, Overgaard HJ, Clements AC et al. Malaria burden and control
25 Esri. ArcGIS Resource Center. How Cluster and Outlier Analysis (Anselin in Bangladesh and prospects for elimination:an epidemiological and
Local Moran’s I) works. http://help.arcgis.com/en/arcgisdesktop/10.0/ economic assessment. Lancet Glob Health 2014;2:e98–105.
help/index.html#//005p00000012000000 [accessed 31 March 2015]. 41 Simon C, Moakofhi K, Mosweunyane T et al. Malaria control in
26 Bousema T, Griffin JT, Sauerwein RW et al. Hitting hotspots: spatial Botswana, 2008–2012: the path towards elimination. Malar J