Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Surveillance in easy to access population subgroups as a tool for evaluating malaria control progress: A systematic review

  • Sanie S. S. Sesay ,

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Validation, Visualization, Writing – original draft, Writing – review & editing

    sanie.sesay@sanofi.com

    Affiliations Malawi-Liverpool-Wellcome Trust Clinical Research Programme, Blantyre, Malawi, Liverpool School of Tropical Medicine, Liverpool, United Kingdom

    ORCID http://orcid.org/0000-0001-6233-7854

  • Emanuele Giorgi,

    Roles Writing – original draft, Writing – review & editing

    Affiliations Medical School, Lancaster University, Lancaster, United Kingdom, Institute of Infection and Global Health, University of Liverpool, Liverpool, United Kingdom

  • Peter J. Diggle,

    Roles Methodology, Supervision, Writing – review & editing

    Affiliations Medical School, Lancaster University, Lancaster, United Kingdom, Institute of Infection and Global Health, University of Liverpool, Liverpool, United Kingdom

  • David Schellenberg,

    Roles Methodology, Writing – review & editing

    Affiliation London School of Hygiene and Tropical Medicine, London, United Kingdom

  • David G. Lalloo,

    Roles Methodology, Supervision, Validation, Writing – original draft, Writing – review & editing

    Affiliations Malawi-Liverpool-Wellcome Trust Clinical Research Programme, Blantyre, Malawi, Liverpool School of Tropical Medicine, Liverpool, United Kingdom

  • Dianne J. Terlouw

    Roles Conceptualization, Funding acquisition, Methodology, Supervision, Validation, Writing – original draft, Writing – review & editing

    Affiliations Malawi-Liverpool-Wellcome Trust Clinical Research Programme, Blantyre, Malawi, Liverpool School of Tropical Medicine, Liverpool, United Kingdom

Surveillance in easy to access population subgroups as a tool for evaluating malaria control progress: A systematic review

  • Sanie S. S. Sesay, 
  • Emanuele Giorgi, 
  • Peter J. Diggle, 
  • David Schellenberg, 
  • David G. Lalloo, 
  • Dianne J. Terlouw
PLOS
x

Abstract

Background

The need for surveillance systems generating targeted, data-driven, responsive control efforts to accelerate and sustain malaria transmission reduction has been emphasized by programme managers, policy makers and scientists. Surveillance using easy-to-access population subgroups (EAGs) may result in considerable cost saving compared to household surveys as the identification and selection of individuals to be surveyed is simplified, fewer personnel are needed, and logistics are simpler. We reviewed available literature on the validation of estimates of key indicators of malaria control progress derived from EAGs, and describe the options to deal with the context specific bias that may occur.

Methods

A literature search was conducted of all documents reporting validation of estimates of malaria control indicators from EAG surveys before the 31st of December 2016. Additional records were identified through cross-reference from selected records, other applicable policy documents and grey literature. After removal of duplicates, 13, 180 abstracts were evaluated and 2,653 eligible abstracts were identified mentioning surveillance in EAGs, of which 29 full text articles were selected for detailed review. The nine articles selected for systematic review compared estimates from health facility and school surveys with those of a contemporaneous sample of the same population in the same geographic area.

Results

Review of the available literature on EAGs suitable for surveillance of malaria control progress revealed that little effort has been made to explore the potential approach and settings for use of EAGs; and that there was wide variation in the precision of estimates of control progress between and within studies, particularly for estimates of control intervention coverage. Only one of the studies evaluated the geospatial representativeness of EAG samples, or carried out geospatial analyses to assess or control for lack of geospatial representativeness. Two studies attempted to measure the degree of bias or improve the precision of estimates by controlling for bias in a multivariate analysis; and this was only successful in one study. The observed variability in accuracy of estimates is likely to be caused by selection and/or information bias due to the inherent nature of EAGs. The reviewed studies provided insight into the design and analytical approaches that could be used to limit bias.

Conclusion

The utility EAGs for routine surveillance of progress in malaria control at the district or sub-district programmatic level will be driven by several factors including whether serial point estimates to measure transmission reduction or more precise geospatial distribution to track ‘hot-spots’ is required, the acceptable degree of precision, the target population, and the resources available for surveillance. The opportunities offered by novel geostatistical analyses and hybrid sampling frames to overcome bias justify a renewed exploration of use of EAGs for malaria monitoring and evaluation.

Background

The need for surveillance systems that inform accelerated and sustained control efforts to accelerate and sustain malaria transmission reduction has been emphasized by programme managers, policy makers and scientists. A key element of these surveillance systems will be their cost and whether they can easily be integrated with current malaria control activities. Routine health facility-based passive case reporting, for example through Health Management Information Systems (HMISs), has been and continues to be at the forefront of malaria surveillance [1, 2]. A well-functioning HMIS will provide regular data from all health facilities nationally allowing accurate measurement of malaria control progress across the healthcare system. This has largely not been the case for most HMISs in malaria endemic countries, with problems like incomplete reporting and lack of diagnostic confirmation being comparatively common [3, 4]. Malaria indicator surveys (MISs) provide single cross-sectional national assessments of disease burden [5], but are usually expensive and logistically demanding to undertake. The goal of MISs is to generate nationally representative estimates and are thus not powered to detect local-level variability[6, 7]. The interval between serial MISs also affect their direct relevance for monitoring short- and medium-term trends in malaria control progress. Supplementary approaches are thus needed to provide timely estimates of malaria control progress at the district and sub-district level, complementary to current malaria surveillance systems, particularly as malaria transmission intensity falls and its distribution becomes more localized [8].

Representative subsets of the population or disease at-risk groups routinely assemble at easily accessible locations (e.g. schools or health facilities) making them logistically attractive to sample [9]. Alternatively, representative subgroups or the whole population of interest may be easily accessible during public health intervention activities such as mass drug administration and catch-up vaccination campaigns [10]. The opportunistic nature of surveillance in the so called Easy Access Groups (EAGs) could thus save costs by simplifying the identification and selection of individuals to be surveyed, requiring simplified logistics and fewer study personnel compared to household surveys [9, 11]. Evidence from school surveys indicate that EAGs are suitable for surveillance when they are potentially representative of an at-risk stratum of the population [9]. However, there are concerns about the inherent biased nature of such a sample, as such non-probability samples depend on natural systems of selection which are likely to result in the selection of a non-representative sample of the population of interest [12]. In this systematic review, we studied the available literature on the validation of estimates of key indicators of malaria control progress [13] derived from EAGs, focusing on EAGs that may be suitable for surveillance at the district or (sub)district level.

Methods

Search strategy

We searched EMBASE® (EMBASE, Medline, EMBASE Classic), PubMed® and ScienceDirect® bibliographic databases without language restrictions from inception to 31st December 2016 for articles with the following search terms in their keywords, title or abstract: "malaria" AND "survey”; or “malaria” AND “surveillance”, or "malaria" AND “monitoring” AND “evaluation”; or "malaria" AND "transmission" AND "measurement. We also searched the online WHO document centre [14] for relevant policy documents and for grey literature from the WHO historical documents database on malaria (1947–2000) [15]. We also included pertinent articles that were not picked up by our search from other sources including recommendation from key experts in the field of malaria surveillance.

We compiled the results into a searchable database in EndNote X8.0.1 (Thomson Reuters). We searched this database for abstracts detailing validation of estimates from EAGs predetermined to be most suitable for routine malaria surveillance at the (sub)district-level by a review of historical evidence of previous use for malaria surveillance. We also added EAGs that had not been previously used for malaria surveillance but demonstrated this potential through surveillance of other diseases. Selected EAGs were further validated by examination against general criteria used to evaluate the suitability of a surveillance system [16], adapted to malaria surveillance (Table 1). Based on our review we postulated that the following EAGs were suitable for the routine surveillance of malaria control progress (Table 2):

  1. School children
  2. Health facility attendees, including:
    1. All health facility attendees including accompanying persons
    2. Children coming for sick or routine “well” child visits
    3. Women attending ANC or coming for delivery
  3. Population targeted by public health intervention campaign such as mass drug administration
  4. Population attending rural community markets
thumbnail
Table 1. Criteria evaluating the suitability of EAGs for malaria surveillance.

https://doi.org/10.1371/journal.pone.0183330.t001

thumbnail
Table 2. Advantages and disadvantages of EAGs suitable for malaria surveillance.

https://doi.org/10.1371/journal.pone.0183330.t002

We then searched the EndNote database for articles with the following keywords in their abstract:

  1. “school” AND “survey”, “school AND “surveillance”, “school” AND “monitoring” AND “evaluation”, and “school” AND “transmission” AND “measurement”
  2. “health” AND “facility” OR “centre” AND “survey”, health” AND “facility” OR “centre” AND “surveillance”, “health” AND “facility” OR “centre” AND “monitoring” AND “evaluation”, and “health” AND “facility” OR “centre” AND “transmission” AND measurement
  3. “antenatal clinic” AND “survey”, “antenatal clinic” AND “surveillance”, “antenatal clinic” AND “monitoring” AND “evaluation”, “antenatal clinic” AND “transmission” AND “measurement”, “pregnancy” OR “delivery” AND “survey”, “pregnancy” OR “delivery” AND “surveillance”, “pregnancy” OR “delivery” AND “monitoring” AND “evaluation”, and “pregnancy” OR “delivery” AND “transmission” AND “measurement”
  4. “market” AND “survey”, “market” AND “surveillance”, “market” AND “monitoring” AND “evaluation”, and “market” AND “transmission” AND “measurement”
  5. “public health” AND “intervention” OR “campaign” AND “survey”, “public health” AND “intervention” OR “campaign” AND “surveillance”, “public health” AND “intervention” OR “campaign” AND “monitoring” AND “evaluation”, and “public health” AND “intervention” OR “campaign” AND “transmission” AND “measurement”

Inclusion criteria

A total of 13, 180 records were compiled into a searchable database, at which the key word search resulted in the selection of 2,653 eligible abstracts for further review. These abstracts were reviewed for specific mention of the comparison of malaria indicator estimates from an EAG sample with population sample (Fig 1) and 29 articles were selected for full text review.

thumbnail
Fig 1. PRISMA flow diagram for studies comparing estimates between EAG and population surveys.

https://doi.org/10.1371/journal.pone.0183330.g001

Exclusion criteria

We searched for the full text of the selected 29 publications, and excluded studies in which estimates of malaria control indicators from EAGs were not compared to a contemporaneous random population sample from the same geographic area. Since the distribution of Plasmodium falciparum infection in the population is determined by environmental factors that influence the density of competent anopheline mosquitos, location-specific vector behaviour, and human factors like at-risk status (e.g. age and pregnancy) and behaviour (e.g. ITN use) that increase exposure to infectious mosquito bites [1719]; to increase the accuracy of EAG PfPR estimates, we excluded all studies that did not compare EAG samples to population samples from the same age or other at-risk stratum.

Selection of studies

Twenty of the twenty-nine studies selected for full review satisfied one or more exclusion criteria and were not included in the systematic review (Fig 1). Six of the studies were excluded because the data collected was not sufficiently synchronous between the EAG and population sample [2025]. In fourteen studies the validity of EAG estimates could not be determined either because the population sample was not random [26], the same indicators were not collected from both samples [2731], or both samples were otherwise not comparable [3239]. In the nine selected studies, information was recorded on the type of EAG, comparator population, sampling frame, sampling methodology, sample size and sampling units. Data on the first author, year of survey, survey site, year of publication, malaria transmission intensity (e.g. PfPR), and estimates of control progress were extracted for the systematic review.

Definitions

Anaemia prevalence rate (APR)–Proportion of the population with a haemoglobin measurement of <8 g/dL.

Antibody prevalence rate (AbPR)–Proportion of the population seropositive to defined malaria antigens.

Household bed net ownership–Proportion of households with at least one bed net.

Household ITN ownership–Proportion of households with at least one ITN.

Individual bed net use–Proportion of population that slept under a bed net the previous night.

Individual insecticide treated bed net (ITN) use–Proportion of population that slept under an ITN the previous night.

Indoor residual spraying (IRS) coverage–Proportion of households sprayed with IRS in the past 12 months.

Plasmodium falciparum prevalence rate (PfPR)–Proportion of the population with malaria infection detected by rapid diagnostic test (RDT), microscopy or polymerase chain reaction (PCR).

Sick child visit–Health facility visits during childhood for an illness episode.

Well child visit–Routine health facility visits that occur during childhood that may include immunizations, growth and development assessments, physical examination and other health risk assessments.

Statistical analysis

Data analysis was done using Stata version 13.1® (StataCorp, Texas, USA). Using the presented data from tables in the selected publications, we calculated point estimates of control progress indicators derived from EAGs and compared that to estimates from contemporaneous population samples. In one publication [40], due to absence of the numerator, we derived the numerator from the reported rates and the denominator, and then calculated point estimates and corresponding 95% confidence intervals. Where surveys were repeated either seasonally or after a specific period [41, 42], we presented these estimates separately to account for seasonal or temporal effect respectively. We assessed the degree of accuracy in estimates derived from EAG samples by examining the absolute difference in prevalence difference and corresponding 95% confidence intervals and Pearson’s χ2 p values. Mean prevalence was derived for the overall individual level estimates from the EAG and population samples. The estimates for PfPR were derived individually for each method of detection of parasitaemia e.g. blood film, rapid diagnostic test. Malaria endemicity was classified according to the revised Global Malaria Eradication Program classification [43]. Due to the inherent differences in EAGs and the paucity of studies, we did not derive pooled estimate effects for each malaria control indicator. To evaluate the effect of population coverage of control interventions and transmission intensity on the validity of EAG estimates of control interventions and PfPR respectively, where possible, we correlated the prevalence difference with the population prevalence. We also evaluated the potential for any of the EAG samples to misclassify an area into the wrong malaria endemicity category by comparing the classification of each area by population PfPR to that from EAG estimates.

Results

Description of studies

Nine studies were included in the systematic review (Table 3), all of which were from sites with intense stable or moderate stable malaria transmission. Six studies assessed the accuracy of estimates from health facilities [4042, 4446], two studies assessed the accuracy of estimates from school surveys [47, 48], and one study assessed the accuracy of estimates from antenatal clinics [49]. Three studies compared estimates from children less than 5 years old [40, 42, 44], two studies compared estimates from older children [47, 48], and four studies compared estimates from all presenting individuals at health facilities (including ANC) regardless of age [41, 45, 46, 49].

thumbnail
Table 3. Description of studies comparing estimates between EAG and population surveys.

https://doi.org/10.1371/journal.pone.0183330.t003

Comparison of estimates

Estimates of coverage of control interventions.

Seven studies assessed the accuracy of estimates of coverage of control interventions. (Table 4) [4042, 44, 4749]. The estimates of coverage of different control interventions derived from EAGs were significantly higher than those of the population in three studies [40, 44, 47], except for the estimates of household ITN ownership which was concordant with the population in one of these studies [47]. In three studies, estimates of control intervention coverage were significantly lower in EAGs [41, 48, 49]. In one study, estimates derived from parents/guardians of children aged 6–30 months coming for well child visits in Malawi were concordant in the first year of survey (2005) but significantly higher in the second survey (2008) [42]. In 2005, the estimates of individual bed net use derived from this EAG (PR = 41.0%, 95% CI 38.9%, 47.4%) were slightly lower than that in the same age stratum in the population (PR = 45.4%, 95% CI 39.0%, 51.7%, p = 0.0339), though this difference is not significant due to overlapping confidence intervals. Similarly, the estimate of individual ITN use derived from the EAG in the same survey (PR = 36.7%, 95% CI 31.1%, 42.4%) was not significantly different from that of the population (PR = 41.0%, 95% CI 34.1%, 40.5%, p = 0.0311). The study by Stevenson et al [48] investigated the concordance in school and catchment area-based estimates of control intervention coverage across a range of circumferential distances around each school. Estimates of individual bed net use derived from school children living 601-1000m (PR = 31.3%, 95%CI 29.1%, 33.5%) and >1000m (PR = 32.9%, 95%CI 29.1%, 33.5%) from the school were not significantly different from those from school children within 600m of the school (PR = 33.4%, 95% CI 31.2%, 35.6%), indicating that inaccuracy remained relatively constant with changes in circumferential area within the school’s catchment area. In the same study, estimates of IRS coverage from school children living 601-1000m (PR = 70.7%, 95%CI 68.5%, 72.8%) and >1000m (PR = 72.9%, 95%CI 68.5%, 72.8%) from the school were not significantly different from those from school children within 600m (PR = 68.3%, 95%CI 66.1%, 70.4%) of the school again indicating the inaccuracy was not affected by circumferential area within the school’s catchment area.

thumbnail
Table 4. Comparison of estimates of coverage of control interventions between EAGs and the population.

https://doi.org/10.1371/journal.pone.0183330.t004

Estimates of malaria morbidity.

Six studies assessed the accuracy of estimates of malaria morbidity (Table 5) [41, 42, 4446, 48]. All six studies evaluated estimates of Plasmodium falciparum prevalence rate (PfPR) either by rapid diagnostic test (RDT), microscopy or polymerase chain reaction (PCR). In the studies where PfPR was determined by microscopy, slides were double read [41, 44, 45] or single read by an expert microscopist [42]. As an additional measure, in two studies there was external quality control [41, 42], and in one study PCR was used to complement missing second reads and to disambiguate discordant species read results [45]. In three studies, estimates of PfPR derived from EAGs were significantly higher than those of the population [45, 46, 48]. In one study [44], estimates of PfPR derived from children attending health facilities for sick visits were not only concordant with population estimates but there was also concordance between results derived by microscopy and PCR. The accuracy of estimates PfPR by RDT (Paracheck®, Orchid Biomedical Systems, India) derived from school children with circumferential distance was assessed in one study [48], and the estimate from this EAG remained consistently higher with increasing distance within the school catchment area.

thumbnail
Table 5. Comparison of estimates of coverage of malaria morbidity between EAGs and the population.

https://doi.org/10.1371/journal.pone.0183330.t005

Three studies assessed the ability of EAGs to measure changes in PfPR as result of changes in coverage of interventions [42, 45] or seasonal transmission [41]. When data was collected before and one year after an ITN campaign in Papua New Guinea, the derived reduction in PfPR by RDT in patients with a history of fever attending health facilities (absolute Risk Difference or RD = 23.3%, 95%CI 20.1%, 26.5%) was almost thrice that in the population (RD = 7.6%, 95%CI 6.1%, 9.1%) [45]. After a period of intense distribution of ITNs and a change in first line therapy of malaria from sulphadoxine-pyrimethamine to artemether-lumefantrine in Malawi, the reduction in PfPR by malaria microscopy measured in children 6–30 months attending well child clinics (RD = 17.4%, 95%CI 14.6%, 20.2%) was higher than that in the same age strata in the population (RD = 10.5%, 95% CI 7.4%, 13.7%) [41], probably due to significantly higher EAG estimates in the first survey (Table 5). The study by Oduro et al [41] assessed the effect of seasonality on summary estimates PfPR by malaria microscopy in all patients attending HFs in six ecologically diverse areas in Gambia, a country with intensely seasonal malaria transmission. In patients attending health facility regardless of cause, the reduction in PfPR between the rainy season and the dry season (RD = 22.8%, 95%CI 21.6%, 24.1%) was almost twice that from the HF catchment population (RD = 10.4%, 95%CI 9.3%, 11.6%), probably due to the significantly higher estimates in the rainy/post-rainy season.

Two studies compared estimates of anaemia prevalence rate (APR) between EAGs and the population [41, 42]. In the study by Mathanga et al [42], estimates of APR from children attending well child clinics were not only concordant with values derived from the same age strata in the population but this metric in children attending well child clinics (RD = 2.8%, 95% CI 0.4%, 5.3%) accurately reflected the reduction in the population (RD = 5.3%, 95% CI 2.6%, 8.1%). The other study in Gambia assessed the impact of seasonality on estimates of APR derived from patients of all ages [41], and the difference between the rainy and dry season estimates from this EAG (RD = 2.0%, 95%CI 0.8%, 3.2%) was similar to that in the population (RD = 4.0%, 95%CI 3.0%, 5.0%) though both EAG estimates were consistently higher than population estimates (Table 5).

Two studies compared estimates of antibody prevalence between EAGs and the population [41, 48]. In the study in Gambia where malaria is intensely seasonal with one seasonal peak [41], the difference in Merozoite Surface Protein 119 (MSP119) seroprevalence between the seasons in the EAG (RD = 12.5%, 95%CI 10.4%, 14.6%) was higher than the population (RD = -0.1%, 95%CI -2.0%, 1.8%) due to overestimation of the population value in the rainy season (Table 5). In a moderately stable malaria transmission setting where there are two seasonal peaks of transmission (one major and the other minor), an assessment of AbPR using a number of antigens including Apical Membrane Antigen 1 (AMA1) and MSP1[48] in the month immediately after the major peak revealed that the estimate from school children (AbPR = 51.5%, 95% CI 49.2%, 53.8%) was concordant with that of the same age strata in the population (AbPR = 51.5%, 95% CI 49.9%, 53.1%, p = 1.000), and remained so with increasing distance within the school catchment area.

Assessment of accuracy of EAG estimates

Except for the study by Ndyomugyenyi et al [47] were estimates of household ITN ownership derived from primary school children accurately represented population coverage (RD = 0, 95% CI -0.02, 0.02, p = 0.9759), estimates of control intervention coverage derived from EAGs were subject to widely varying degrees of inaccuracy (RD range: -0.24–0.42), with EAGs estimates more commonly but not exclusively over-estimating population values (Fig 2). In the two studies that assessed the accuracy of multiple indicators of intervention coverage [40, 42], estimates of related indicators usually had a similar level of inaccuracy (Fig 2). In the study by Mathanga et al [42], serial estimates of control intervention exhibited similar degree of bias in estimates of individual bed net and ITN use in the first survey but were markedly different in the subsequent survey (Fig 2). In the study by Skarbinski et al [40], the degree of accuracy in estimates of household bed net ownership, individual bed net and ITN use was the same for both well and sick child visits in Rufiji and ITN use in Lindi, whilst estimates of household bed net ownership and individual bed net use were much higher in Lindi (Table 2, Fig 2) indicting regional-specific bias (Fig 2).

thumbnail
Fig 2. Absolute prevalence difference in estimates of standard malaria indicators.

ANC = Antenatal Clinic OPD = All OPD SC = School children WSC = Well or sick child BS = Blood slide PCR = Polymerase chain reaction RDT = Rapid diagnostic test.

https://doi.org/10.1371/journal.pone.0183330.g002

Estimates of PfPR were on average more consistent than estimates of intervention coverage (Fig 2). In the study by Gahutu et al [44], estimates of PfPR by microscopy and PCR derived from EAGs at different health facility levels were concordant with population values (Fig 2). In the study by Mathanga et al [42], though serial estimates of PfPR from children aged 6–30 months attending well child clinics accurately detected transmission reduction in the same age strata in the population, the estimate of PfPR from this EAG was slightly higher than that in the population in 2005 (RD = 0.06, 95% CI 0.02, 0.10, p = 0.002). Estimates of APR derived from EAGs in two studies [41, 42] were overall a more consistent estimation of population prevalence than PfPR (Fig 2). The close approximation of EAG estimates of APR together with its accurate measurement of a reduction in population prevalence suggests that it is a good surrogate indicator for APR in the population [42]. Estimates of AbPR derived from EAGs were more accurate in the dry season in the Gambia [41], with rainy season estimates being higher than population estimates (RD = 0.12, 95% CI 0.10, 0.02, p<0.001).

Two of the studies attempted to measure the degree of inaccuracy or improve the precision of estimates by controlling for bias [40, 42]. After controlling for potential confounders (age in months, child’s sex, survey type and study area)in a multivariable analysis, in the study by Skarbinski et al [40], the adjusted odds ratio (aOR) between the health facility survey and the EAG survey for individual bed net use (aOR = 2.05, 95% CI 1.36, 3.08) and ITN use (aOR = 2.41, 95% CI 1.69, 3.44) still indicated an overestimation of population coverage. In the study by Mathanga et al [42], after adjusting for confounders in a multivariate analysis, parasitaemia in 2008 vs 2005 in children attending well child clinic (aOR = 0.31, 95% CI 0.22, 0.46) was equivalent to that in the same age strata in the population (aOR = 0.40, 95% CI 0.30, 0.52), and this was similar for anaemia (Hb<8.0d/dl) in this EAG (aOR = 0.85, 95% CI 0.65, 1.65) compared to the population (aOR = 0.74, 95% CI 0.59, 0.94).

For EAG to guide control efforts, it should correctly classify the uptake of control interventions and malaria endemicity. The prevalence difference in bed net use suggested that EAG surveys overestimated population levels up to a certain point (population coverage of approximately 72%), after which they overestimated population values, but this trend was not statistically significant (p = 0.993) (Fig 3A). The prevalence difference in PfPR overestimated population prevalence with increasing transmission (p = 0.979) (Fig 3B), but our assumptions are also limited by the fact that the studies included in this review only covered moderately stable and unstable endemic transmission intensities. Based on the classification of malaria endemicity from the PfPR results, most of the EAG surveys (13/14) were concordant with that of the population (Table 6). During the post-ITN survey in Papua New Guinea [45], population PfPR dropped to unstable endemic levels (PfPR = 2.5%, 95%CI 1.8%; 3.2%) but was wrongly classified to be moderate stable by the EAG (PfPR = 7.5%, 95%CI 5.5%; 9.5%.).

thumbnail
Fig 3. Prevalence difference of bed net use and PfPR with population levels.

https://doi.org/10.1371/journal.pone.0183330.g003

thumbnail
Table 6. Relationship between the results of the classification of malaria endemicity between EAG and population sampling.

https://doi.org/10.1371/journal.pone.0183330.t006

Discussion

Monitoring control progress is important to assess the effectiveness and coverage of malaria control programmes. Easy access group surveys are easier to conduct than population surveys and could provide accurate monitoring of control progress if the EAG sample is representative of our population stratum of interest [30, 33, 5053]. Review of the available literature on EAGs suitable for district or sub-district surveillance of malaria control progress revealed a wide variation in the precision of estimates between and within studies, particularly for estimates of control intervention coverage. The small number of studies in this review shows how little effort has been made to explore the potential approach and settings for use of EAGs, probably due to the inherent assumption of bias in such opportunistic samples. Our study has potential limitations. Our search strategy may not have identified all the relevant papers or there may be other sources of grey literature that may have been missed. We phrased our search terms as simply as possible to allow a wider inclusion of possible papers and in this regard, we may have missed some papers with highly selective titles. The studies selected for the systematic review only included health facility (including ANC) and school surveys, and were from settings with moderate and intense stable malaria transmission, so our results may not be applicable to other EAGs or transmission settings. Our literature search was guided by categories of EAGs with historical evidence of use for malaria surveillance or which we theorized would be suitable for malaria surveillance at the district or sub-district level. This may have excluded publications on other potential EAGs. We limited our review to studies that compared EAG samples to populations samples of the same age or other at-risk stratum. Whilst this may improve the accuracy of EAG estimates of PfPR, especially in moderate to severe transmission settings, this does not mean that EAGs could not be used to estimate control intervention coverage in any population stratum or PfPR at the lower end of the transmission spectrum in other population at-risk strata. Given the pace of developments in analytical technics, this is an area where substantial gains can be made and we discuss this below.

Dealing with bias in EAG surveillance

The main cause of bias in EAG surveillance is due to the selection of an unrepresentative sample of the population of interest. The opportunistic nature of the sampling frame in EAGs is inherently susceptible to selection bias when EAG sampling captures an unrepresentative subset of the population of interest. Particularly, if the reason for inclusion in the EAG sample is associated with the indicator of interest. For example, given the fact that those who are wealthier and more educated are more likely to attend health facilities, and have access to or use ITNs [54, 55], self-reported ITN possession and use from health facility surveys is likely to over-estimate ITN coverage in the population. This could be corrected using the verification rate measured from a small random sample of the catchment population. Also, the representativeness of estimates of PfPR from health facilities is likely to be affected by the difference in transmission between malaria seasons, overall malaria transmission and the prevalence of non-malaria fevers. This could be limited by the use of EAGs excluding individuals coming for sick visits [56, 57] or prioritizing indictors that are less sensitive to short-term changes in transmission like AbPR [58, 59]. Population APR is also less sensitive to short term changes in transmission [60], but whether this makes it an appropriate indicator to measure changes in transmission is debatable. Though malaria is an important correlate of anaemia in children, the aetiology of anaemia is multifactorial and in particular the role of other infections, poor nutrition and the interaction between malaria and nutrition needs to be clarified [61]. Where there is a high probability of inclusion in the EAG sample, the difference in the estimates of an indicator measured from individuals who are and are not included in the EAG sample is likely to be less significant, and the EAG sample is more likely to be representative of the true situation in the population. For example, coverage rates of public health interventions were similar between vaccinated and unvaccinated children if population vaccine coverage was over 60% [62]. Most of the standard methods for analysis of data from convenience samples are based on the questionable assumption that selection bias can be exclusively ascribed to measured risk factors for malaria. Novel geostatistical methods have been recently developed to relax this assumption [63]. By combining data from unbiased gold-standard surveys and opportunistic samples, these methods are able to correct for the selection bias in the convenience samples that is induced by both measured and unmeasured risk factors. Though the aetiology of health facility access and utilization is multifactorial [6469], health facility utilization follows a geographic pattern [7073] and if this can be accurately measured through a small geospatially random sample of the population and accounted for in the model, will allow correction for bias and the production of accurate maps of control progress. Where point estimates are required, combining the EAG sample with a small and presumably far less expensive random sample of the population [74], the so-called hybrid sampling methodology will generate more accurate hybrid prevalence estimates. Pooling data from multiple EAGs in our area of interest is also likely to improve the precision of point estimates [47, 75].

Another cause of bias in EAG samples mainly affecting reported coverage of control interventions is social desirability bias. Survey respondents may answer questions in a manner they consider favourable to the interviewee leading to erroneously high self-reporting of coverage of control interventions [76]. This may be further compounded by the inability to directly validate the presence and use of household-level and individual control measures as in population surveys. Few studies have assessed the effect of social desirability bias on the effect of bed net use [7780], and the wide range in verification rate of bed net use after self-report (60.9–96.2%) suggests variability in the effect of social desirability bias from setting to setting. Social desirability bias can be limited by modifying the standard MIS questionnaire [13] to include questions aimed at detecting and measuring social desirability bias so it can be directly accounted for in analysis [81], improving interviewee confidence by indicating the anonymity of their responses [81], or indirectly by correcting for the verification rate measured from a random sample of the catchment population.

Use of EAG surveys to measure progress in malaria control

Before EAGs can be routinely used to measure malaria control progress, there are a few issues to address. Firstly, how much inaccuracy we are willing to tolerate? If the purpose of the survey is to measure trends in point estimates of control progress, some degree of inaccuracy is tolerable if EAG data displays similar trends to population data; as evidenced by the successful demonstration of transmission reduction from health facility surveys in some endemic countries [30, 33, 5053], and increasing endorsement by WHO as a surveillance tool in different transmission settings and phases of control [1]. One study suggested that estimates of population PfPR from health facility surveys might misclassify malaria endemicity at the lower end of the transmission spectrum [45], but the population in this study (i.e. all health facility attendees) may not be the most suitable to capture the most at-risk population at low transmission settings. When more accurate point estimates are required or accurate data is required over a large geographic area, pooling data from multiple similar EAGs [47, 75] or hybrid sampling methodology [74] may improve precision. If the purpose of the survey is to measure changes in the the geospatial distribution of uptake of control interventions and transmission intensity, to identify areas of low intervention coverage and potential hotspots respectively for targeted control intervention delivery; the smaller sized EAG catchment areas compared to community-based surveys [7073] means the maps derived using EAG sampling will not be consistent with those derived using community-level data and would require geospatial statistical methods to correct for bias [63].

Secondly, are EAGs surveys more cost effective than standard approaches? Because of the ease of EAG sampling, conducting an EAG survey should theoretically be cheaper than a population survey in the same catchment area. Reports from school surveys in Kenya seem to suggest that the financial cost of school surveys is less than half that of a household survey [9, 11]. Though a detailed economic costs analysis of school surveys has not been done in comparison to those from household surveys, and the lower financial costs has not been validated in other EAGs; the decreased expenditure on personnel, transportation and communication in school surveys compared to household surveys suggest that surveillance in EAGs is likely to be more cost-effective [9].

Thirdly, when are EAG surveys most likely to be beneficial i.e. to complement malaria programmatic efforts? At moderate to high malaria transmission intensity, surveillance systems rely on passive surveillance (e.g. HMISs) supplemented by large serial populations surveys (e.g. MISs), with data reported at the national, regional and sometimes district level. Surveillance in EAGs in such settings will be beneficial in providing more detailed sub-district estimates from “problematic” districts with poor control progress compared to national average, estimates from hard-to-reach communities (e.g. opportunistic surveys during MDA) who would otherwise not be covered by population surveys, or when data is required to assess at-risk stratum specific control interventions (e.g. ANC and delivery surveys to assess the impact of Intermittent Preventive Treatment in pregnancy or IPTp). These EAG surveys should be carried out at the same time as population surveys i.e. every 2–3 years, so that the estimates can be interpreted within the context of a wider perspective of population control progress. As transmission intensity falls and we approach the elimination phase, reorientation of programmatic efforts are required to identify hotspots [8] and special high-risk populations [82, 83] both of which serve as reservoirs of infection that should be targeted for malaria elimination. Population surveys become less logistically attractive and less practical given the fact that more regular (e.g. quarterly) local (sub-district) level data is required on control progress. Surveillance in EAGs becomes more attractive as a more sustainable method of surveillance including the high-risk groups (e.g. rural community market surveys at border crossings).

Finally, how do we integrate surveillance in EAGs with current control strategies? EAG surveillance can provide timely data of reasonable accuracy on control progress that reflect local variation at the district or sub-district level, and is complementary to national community-based surveys like MISs [13]. EAG surveys can provide a means of rapid assessment of areas known to have poor coverage or key population risk-strata. The ease of sampling and low costs allows more frequent or even continuous surveys providing timely data and encouraging reactive targeted control. EAG surveillance in health facilities may have a motivational impact on health workers at the district and sub-district level through the provision of continuous locally appropriate data on intervention coverage and malaria transmission, and its flexibility allows it to adapt to new programmatic needs over time. Sufficient person-time is however needed for successful data acquisition in health facilities and to ensure no duplication with recurrent data collection. Implementing and scaling up EAG surveillance will require minimal reorientation and structuring of the health system, including determining which health facility personnel should be dedicated to malaria surveillance, and some preparation and buy-in is required by both national and global health players.

Conclusions

This review describes the previous experiences with the validation of estimates of malaria control progress from different EAGs and highlights the potential of surveillance in EAGs as a complementary approach to current surveillance systems. The utility of an EAG for routine surveillance of progress in malaria control at the district or sub-district programmatic level will be driven by several factors including whether serial point estimates or more precise geospatial distribution is required, the degree of precision accepted, the desired population of interest (e.g. at-risk groups), and the resources available for surveillance (both financial and otherwise).The low cost of EAG surveillance, its flexibility and potential to offer locally applicable timely estimates of control which could improve programmatic responses suggest that further validation and optimization is required.

Supporting information

Acknowledgments

SSSS is supported by the Malaria Capacity Development Consortium which is funded by The Wellcome Trust (Grant number WT084289MA), DJT is supported by the ACT consortium (Grant number ACTia main/VA10). We would also like to acknowledge Marc Vittori of Sanofi Pasteur Library Services Marcy L’Etoile who assisted with the updated bibliographic search.

References

  1. 1. WHO. Disease surveillance for malaria control: an operational manual. Geneva2012 [cited 2017 13th April]. Available from: http://apps.who.int/iris/bitstream/10665/44851/1/9789241503341_eng.pdf?ua=1.
  2. 2. WHO. Disease surveillance for malaria elimination: an operational manual. Geneva: World Health Organization; 2012 [cited 2017 13th April]. Available from: http://apps.who.int/iris/bitstream/10665/44852/1/9789241503334_eng.pdf?ua=1.
  3. 3. Cibulskis RE, Aregawi M, Williams R, Otten M, Dye C. Worldwide incidence of malaria in 2009: estimates, time trends, and a critique of methods. PLoS medicine. 2011;8(12):e1001142. Epub 2011/12/30. pmid:22205883; PubMed Central PMCID: PMCPMC3243721.
  4. 4. Rudan I, Lawn J, Cousens S, Rowe AK, Boschi-Pinto C, Tomaskovic L, et al. Gaps in policy-relevant information on burden of disease in children: a systematic review. Lancet. 2005;365(9476):2031–40. Epub 2005/06/14. pmid:15950717.
  5. 5. RBM. Malaria Indicator Survey: Basic Documentation for Survey Design and Implementation: Roll Back Malaria; 2005 [cited 2017 13th April]. Available from: http://malariasurveys.org/toolkit.cfm.
  6. 6. Greenwood BM. The microepidemiology of malaria and its importance to malaria control. Transactions of the Royal Society of Tropical Medicine and Hygiene. 1989;83 Suppl:25–9. Epub 1989/01/01. pmid:2576161.
  7. 7. Snow RW, Rowan KM, Lindsay SW, Greenwood BM. A trial of bed nets (mosquito nets) as a malaria control strategy in a rural area of The Gambia, West Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene. 1988;82(2):212–5. Epub 1988/01/01. pmid:3055456.
  8. 8. Bousema T, Griffin JT, Sauerwein RW, Smith DL, Churcher TS, Takken W, et al. Hitting hotspots: spatial targeting of malaria for control and elimination. PLoS medicine. 2012;9(1):e1001165. Epub 2012/02/04. pmid:22303287; PubMed Central PMCID: PMCPMC3269430.
  9. 9. Brooker S, Kolaczinski JH, Gitonga CW, Noor AM, Snow RW. The use of schools for malaria surveillance and programme evaluation in Africa. Malaria journal. 2009;8:231. Epub 2009/10/21. pmid:19840372; PubMed Central PMCID: PMCPMC2768743.
  10. 10. Santos LM, Paes-Sousa R, Silva Junior JB, Victora CG. National Immunization Day: a strategy to monitor health and nutrition indicators. Bulletin of the World Health Organization. 2008;86(6):474–9. Epub 2008/06/24. pmid:18568277; PubMed Central PMCID: PMCPMC2647466.
  11. 11. Gitonga CW, Karanja PN, Kihara J, Mwanje M, Juma E, Snow RW, et al. Implementing school malaria surveys in Kenya: towards a national surveillance system. Malaria journal. 2010;9:306. Epub 2010/11/03. pmid:21034492; PubMed Central PMCID: PMCPMC2984573.
  12. 12. Law GR, Pascoe S, W. Foundations of Epidemiology. In: Law GR, Pascoe S, W., editors. Statistical Epidemiology. Boston: CAB International; 2013. p. 1–35.
  13. 13. MEASURE Evaluation, MEASURE DHS, PMI, RBM, UNICEF, World Health Organization Malaria Action Programme. Household Survey Indicators for Malaria Control: WHO; 2013. Available from: http://www.malariasurveys.org/documents/Household%20Survey%20Indicators%20for%20Malaria%20Control.pdf.
  14. 14. WHO. WHO Document centre 2017 [cited 2017 13th April]. Available from: http://www.who.int/malaria/publications/en/.
  15. 15. WHO/MAL. WHO/MAL series of OFFSET documents (1947–2000). Geneva: World Health Organization; 2012. Available from: http://www.who.int/malaria/publications/atoz/whomal_2012.1119.pdf.
  16. 16. Thacker SB, Parrish RG, Trowbridge FL. A method for evaluating systems of epidemiological surveillance. World health statistics quarterly Rapport trimestriel de statistiques sanitaires mondiales. 1988;41(1):11–8. Epub 1988/01/01. pmid:3269210.
  17. 17. Boyd MF PP, Christophers R. Malariology A comprehensive survey of all aspects of this group of diseases from a global standpoint. Philadelphia and London: W.B. Saunders; 1949.
  18. 18. Molineaux LG G. The Garki Project: Research on the epidemiology and control of malaria in the Sudan Savanna of West Africa. Geneva: World Health Organization; 1980.
  19. 19. Desai M, ter Kuile FO, Nosten F, McGready R, Asamoa K, Brabin B, et al. Epidemiology and burden of malaria in pregnancy. The Lancet infectious diseases. 2007;7(2):93–104. Epub 2007/01/26. pmid:17251080.
  20. 20. Mitja O, Paru R, Selve B, Betuela I, Siba P, De Lazzari E, et al. Malaria epidemiology in Lihir Island, Papua New Guinea. Malaria journal. 2013;12:98. pmid:23497296; PubMed Central PMCID: PMCPMC3606355.
  21. 21. Stewart L, Gosling R, Griffin J, Gesase S, Campo J, Hashim R, et al. Rapid assessment of malaria transmission using age-specific sero-conversion rates. PloS one. 2009;4(6):e6083. Epub 2009/06/30. pmid:19562032; PubMed Central PMCID: PMCPMC2698122.
  22. 22. Hagmann R, Charlwood JD, Gil V, Ferreira C, do Rosario V, Smith TA. Malaria and its possible control on the island of Principe. Malaria journal. 2003;2:15. Epub 2003/07/24. pmid:12875660; PubMed Central PMCID: PMC166171.
  23. 23. Howard N, Enayatullah S, Mohammad N, Mayan I, Shamszai Z, Rowland M, et al. Towards a strategy for malaria in pregnancy in Afghanistan: analysis of clinical realities and women's perceptions of malaria and anaemia. Malaria journal. 2015;14:431. Epub 2015/11/06. pmid:26537247; PubMed Central PMCID: PMCPMC4633046.
  24. 24. Ilombe G, Maketa V, Mavoko HM, da Luz RI, Lutumba P, Van geertruyden JP. Performance of HRP2-based rapid test in children attending the health centre compared to asymptomatic children in the community. Malaria journal. 2014;13:308. Epub 2014/08/12. pmid:25108305; PubMed Central PMCID: PMCPMC4248466.
  25. 25. Parekh FK, Hernandez JN, Krogstad DJ, Casapia WM, Branch OH. Prevalence and risk of Plasmodium falciparum and P. vivax malaria among pregnant women living in the hypoendemic communities of the Peruvian Amazon. Am J Trop Med Hyg 2007;77(3):451–7. pmid:17827359
  26. 26. Rulisa S, Kateera F, Bizimana JP, Agaba S, Dukuzumuremyi J, Baas L, et al. Malaria prevalence, spatial clustering and risk factors in a low endemic area of Eastern Rwanda: a cross sectional study. PloS one. 2013;8(7):e69443. Epub 2013/08/13. pmid:23936018; PubMed Central PMCID: PMCPMC3720654.
  27. 27. Hanson K, Marchant T, Nathan R, Mponda H, Jones C, Bruce J, et al. Household ownership and use of insecticide treated nets among target groups after implementation of a national voucher programme in the United Republic of Tanzania: plausibility study using three annual cross sectional household surveys. BMJ (Clinical research ed). 2009;339:b2434. Epub 2009/07/04. pmid:19574316; PubMed Central PMCID: PMCPMC2714691.
  28. 28. Marchant T, Hanson K, Nathan R, Mponda H, Bruce J, Jones C, et al. Timing of delivery of malaria preventive interventions in pregnancy: results from the Tanzania national voucher programme. Journal of epidemiology and community health. 2011;65(1):78–82. Epub 2009/11/10. pmid:19897468.
  29. 29. Parise ME, Lewis LS, Ayisi JG, Nahlen BL, Slutsker L, Muga R, et al. A rapid assessment approach for public health decision-making related to the prevention of malaria during pregnancy. Bulletin of the World Health Organization. 2003;81(5):316–23. Epub 2003/07/12. pmid:12856049; PubMed Central PMCID: PMCPMC2572460.
  30. 30. Nyarango PM, Gebremeskel T, Mebrahtu G, Mufunda J, Abdulmumini U, Ogbamariam A, et al. A steep decline of malaria morbidity and mortality trends in Eritrea between 2000 and 2004: the effect of combination of control methods. Malaria journal. 2006;5:33. Epub 2006/04/26. pmid:16635265; PubMed Central PMCID: PMCPMC1501031.
  31. 31. Owusu-Agyei S, Awini E, Anto F, Mensah-Afful T, Adjuik M, Hodgson A, et al. Assessing malaria control in the Kassena-Nankana district of northern Ghana through repeated surveys using the RBM tools. Malaria journal. 2007;6. pmid:17683584
  32. 32. Sahu SS, Gunasekaran K, Vanamail P, Jambulingam P. Persistent foci of falciparum malaria among tribes over two decades in Koraput district of Odisha State, India. Malaria journal. 2013;12:72. Epub 2013/02/26. pmid:23433186; PubMed Central PMCID: PMCPMC3598688.
  33. 33. Rodrigues A, Schellenberg JA, Kofoed PE, Aaby P, Greenwood B. Changing pattern of malaria in Bissau, Guinea Bissau. Tropical medicine & international health: TM & IH. 2008;13(3):410–7. Epub 2008/02/27. pmid:18298608.
  34. 34. Pacific Malaria Initiative Survey Group on behalf of the Ministries of Health of V, Solomon I. Malaria on isolated Melanesian islands prior to the initiation of malaria elimination activities. Malaria journal. 2010;9:218. Epub 2010/07/28. pmid:20659316; PubMed Central PMCID: PMCPMC2921077.
  35. 35. Bouma MJ, Parvez SK, Nesbit R, Winkler AMF. Malaria control using permethrin applied to tents of nomadic Afghan refugees in northern Pakistan. Bulletin of the World Health Organization. 1996;74(4):413–21. WOS:A1996VF12000011. pmid:8823964
  36. 36. Dicko A, Barry A, Dicko M, Sagara I, Rogier C, Doumbo OK. Seasonality and age specific malaria morbidity in Didieni, District Of Kolokani, Mali. ASTMH 62nd Annual Meeting November 13–17, 2013 Marriott Wardman Park, Washinton DC: ASTMH; 2010. p. 289.
  37. 37. Drakeley C, Stevenson J, Gitonga C, Spencer L, Oduor A, Oriango R, et al. A comparison of school-and community-based cross-sectional surveys for collection of serological measures of malaria transmission. ASTMH 62nd Annual Meeting November 13–17, 2013 Marriott Wardman Park, Washington DC: ASTMH; 2010. p. 231.
  38. 38. Stresman GH, Stevenson J, Owaga C, Marube E, Odongo W, Shaggari S, et al. Operational approaches for detecting foci of malaria infection: How do school and health facility surveys compare against a community based approach. ASTMH 62nd Annual Meeting November 13–17, 2013 Marriott Wardman Park, Washington DC: ASTMH; 2013. p. 436–7.
  39. 39. Onori E. Distribution of Plasmodium ovale in the eastern, western and northern regions of Uganda. Bulletin of the World Health Organization. 1967;37(4):665–8. Epub 1967/01/01. pmid:5301742; PubMed Central PMCID: PMCPMC2554359.
  40. 40. Skarbinski J, Winston CA, Massaga JJ, Kachur SP, Rowe AK. Assessing the validity of health facility-based data on insecticide-treated bednet possession and use: comparison of data collected via health facility and household surveys—Lindi region and Rufiji district, Tanzania, 2005. Tropical medicine & international health: TM & IH. 2008;13(3):396–405. Epub 2008/04/10. pmid:18397401.
  41. 41. Oduro AR, Bojang KA, Conway DJ, Corrah T, Greenwood BM, Schellenberg D. Health centre surveys as a potential tool for monitoring malaria epidemiology by area and over time. PloS one. 2011;6(11):e26305. pmid:22073155; PubMed Central PMCID: PMCPMC3208541.
  42. 42. Mathanga DP, Campbell CH Jr., Vanden Eng J, Wolkon A, Bronzan RN, Malenga GJ, et al. Comparison of anaemia and parasitaemia as indicators of malaria control in household and EPI-health facility surveys in Malawi. Malaria journal. 2010;9:107. Epub 2010/04/23. pmid:20409342; PubMed Central PMCID: PMCPMC2864286.
  43. 43. Hay SI, Smith DL, Snow RW. Measuring malaria endemicity from intense to interrupted transmission. Lancet Infect Dis. 2008;8:369–78. pmid:18387849
  44. 44. Gahutu JB, Steininger C, Shyirambere C, Zeile I, Cwinya-Ay N, Danquah I, et al. Prevalence and risk factors of malaria among children in southern highland Rwanda. Malaria journal. 2011;10:134. Epub 2011/05/20. pmid:21592380; PubMed Central PMCID: PMCPMC3121650.
  45. 45. Hetzel MW, Reimer LJ, Gideon G, Koimbu G, Barnadas C, Makita L, et al. Changes in malaria burden and transmission in sentinel sites after the roll-out of long-lasting insecticidal nets in Papua New Guinea. Parasites & vectors. 2016;9(1):340. Epub 2016/06/16. pmid:27301964; PubMed Central PMCID: PMCPMC4908799.
  46. 46. Karyana M, Burdarm L, Yeung S, Kenangalem E, Wariker N, Maristela R, et al. Malaria morbidity in Papua Indonesia, an area with multidrug resistant Plasmodium vivax and Plasmodium falciparum. Malaria journal. 2008;7:148. Epub 2008/08/05. pmid:18673572; PubMed Central PMCID: PMCPMC2518158.
  47. 47. Ndyomugyenyi R, Kroeger A. Using schoolchildren's reports of bed net use monitored by schoolteachers as a proxy of community coverage in malaria endemic areas of Uganda. Tropical medicine & international health: TM & IH. 2007;12(2):230–7. Epub 2007/02/16. pmid:17300630.
  48. 48. Stevenson JC, Stresman GH, Gitonga CW, Gillig J, Owaga C, Marube E, et al. Reliability of school surveys in estimating geographic variation in malaria transmission in the western Kenyan highlands. PloS one. 2013;8(10):e77641. Epub 2013/10/22. pmid:24143250; PubMed Central PMCID: PMCPMC3797060.
  49. 49. Briand V, Le Hesran JY, Mayxay M, Newton PN, Bertin G, Houze S, et al. Prevalence of malaria in pregnancy in southern Laos: a cross-sectional survey. Malaria journal. 2016;15(1):436. pmid:27566274; PubMed Central PMCID: PMCPMC5002160.
  50. 50. Ceesay SJ, Casals-Pascual C, Erskine J, Anya SE, Duah NO, Fulford AJ, et al. Changes in malaria indices between 1999 and 2007 in The Gambia: a retrospective analysis. Lancet. 2008;372(9649):1545–54. Epub 2008/11/06. pmid:18984187; PubMed Central PMCID: PMCPMC2607025.
  51. 51. Ceesay SJ, Casals-Pascual C, Nwakanma DC, Walther M, Gomez-Escobar N, Fulford AJ, et al. Continued decline of malaria in The Gambia with implications for elimination. PloS one. 2010;5(8):e12242. Epub 2010/09/02. pmid:20805878; PubMed Central PMCID: PMCPMC2923605.
  52. 52. Erhart A, Thang ND, Xa NX, Thieu NQ, Hung LX, Hung NQ, et al. Accuracy of the health information system on malaria surveillance in Vietnam. Transactions of the Royal Society of Tropical Medicine and Hygiene. 2007;101(3):216–25. Epub 2006/09/19. pmid:16979202.
  53. 53. O'Meara WP, Bejon P, Mwangi TW, Okiro EA, Peshu N, Snow RW, et al. Effect of a fall in malaria transmission on morbidity and mortality in Kilifi, Kenya. Lancet. 2008;372(9649):1555–62. Epub 2008/11/06. pmid:18984188; PubMed Central PMCID: PMCPMC2607008.
  54. 54. Zegers de Beyl C, Koenker H, Acosta A, Onyefunafoa EO, Adegbe E, McCartney-Melstad A, et al. Multi-country comparison of delivery strategies for mass campaigns to achieve universal coverage with insecticide-treated nets: what works best? Malaria journal. 2016;15:58. Epub 2016/02/05. pmid:26842732; PubMed Central PMCID: PMCPMC4740992.
  55. 55. Babalola S, Ricotta E, Awantang G, Lewicky N, Koenker H, Toso M. Correlates of Intra-Household ITN Use in Liberia: A Multilevel Analysis of Household Survey Data. PloS one. 2016;11(7):e0158331. Epub 2016/07/13. pmid:27403877; PubMed Central PMCID: PMCPMC4942134.
  56. 56. UNICEF. Expanded Immunization Coverage 2017 [cited 2017 13th April]. Available from: http://www.unicef.org/immunization/index_coverage.html.
  57. 57. UNICEF. Antenatal Care 2017 [cited 2017 13th April]. Available from: https://data.unicef.org/topic/maternal-health/antenatal-care/.
  58. 58. Voller A, Draper CC. Immunodiagnosis and sero-epidemiology of malaria. British medical bulletin. 1982;38(2):173–7. Epub 1982/05/01. pmid:6810986.
  59. 59. Corran P, Coleman P, Riley E, Drakeley C. Serology: a robust indicator of malaria transmission intensity? Trends Parasitol. 2007;23(12):575–82. pmid:17988945.
  60. 60. Korenromp EL, Armstrong-Schellenberg JR, Williams BG, Nahlen BL, Snow RW. Impact of malaria control on childhood anaemia in Africa—a quantitative review. Tropical medicine & international health: TM & IH. 2004;9(10):1050–65. Epub 2004/10/16. pmid:15482397.
  61. 61. McCuskee S, Brickley EB, Wood A, Mossialos E. Malaria and Macronutrient Deficiency as Correlates of Anemia in Young Children: A Systematic Review of Observational Studies. Annals of Global Health. 2014;80(6):458–65. http://doi.org/10.1016/j.aogh.2015.01.003. pmid:25960095
  62. 62. Cibulskis RE, Pujari S, Otten MW. Do estimates of intervention coverage obtained from children at immunization clinics provide a reasonable approximation to population values? The Journal of infectious diseases. 2012;205 Suppl 1:S91–102. Epub 2012/02/15. pmid:22315392.
  63. 63. Giorgi E SS, Terlouw DJ, Diggle PJ. Combining data from multiple spatially referenced prevalence surveys using generalized linear geostatistical models. J R Stat Soc A. 2015;178(2):445–64. Epub February 2015
  64. 64. Zyaambo C, Siziya S, Fylkesnes K. Health status and socio-economic factors associated with health facility utilization in rural and urban areas in Zambia. BMC health services research. 2012;12:389. Epub 2012/11/14. pmid:23145945; PubMed Central PMCID: PMCPMC3536624.
  65. 65. Baker JB, Liu L. The determinants of primary health care utilization: a comparison of three rural clinics in Southern Honduras. GeoJournal. 2006;66(4):295–310.
  66. 66. Wyss K, Whiting D, Kilima P, McLarty DG, Mtasiwa D, Tanner M, et al. Utilisation of government and private health services in Dar es Salaam. East African medical journal. 1996;73(6):357–63. Epub 1996/06/01. pmid:8840594.
  67. 67. Ewing VL, Lalloo DG, Phiri KS, Roca-Feltrer A, Mangham LJ, SanJoaquin MA. Seasonal and geographic differences in treatment-seeking and household cost of febrile illness among children in Malawi. Malaria journal. 2011;10:32. pmid:21303538; PubMed Central PMCID: PMCPMC3049750.
  68. 68. Nyamongo IK. Health care switching behaviour of malaria patients in a Kenyan rural community. Social science & medicine (1982). 2002;54(3):377–86. Epub 2002/02/05. pmid:11824914.
  69. 69. Asenso-Okyere WK, Anum A, Osei-Akoto I, Adukonu A. Cost recovery in Ghana: are there any changes in health care seeking behaviour? Health policy and planning. 1998;13(2):181–8. Epub 1998/05/07. pmid:10180407.
  70. 70. Guagliardo MF. Spatial accessibility of primary care: concepts, methods and challenges. International journal of health geographics. 2004;3(1):3. Epub 2004/02/28. pmid:14987337; PubMed Central PMCID: PMCPMC394340.
  71. 71. Delamater PL, Messina JP, Shortridge AM, Grady SC. Measuring geographic access to health care: raster and network-based methods. International journal of health geographics. 2012;11(1):15. Epub 2012/05/17. pmid:22587023; PubMed Central PMCID: PMCPMC3511293.
  72. 72. Shannon GW, Skinner JL, Bashshur RL. Time and distance: the journey for medical care. International journal of health services: planning, administration, evaluation. 1973;3(2):237–44. Epub 1973/01/01. pmid:4716559.
  73. 73. Alegana VA, Wright JA, Pentrina U, Noor AM, Snow RW, Atkinson PM. Spatial modelling of healthcare utilisation for treatment of fever in Namibia. International journal of health geographics. 2012;11:6. Epub 2012/02/18. pmid:22336441; PubMed Central PMCID: PMCPMC3292929.
  74. 74. Hedt BL, Pagano M. Health indicators: eliminating bias from convenience sampling estimators. Statistics in medicine. 2011;30(5):560–8. Epub 2011/02/04. pmid:21290401; PubMed Central PMCID: PMCPMC3951147.
  75. 75. Rowe AK. Potential of integrated continuous surveys and quality management to support monitoring, evaluation, and the scale-up of health interventions in developing countries. The American journal of tropical medicine and hygiene. 2009;80(6):971–9. Epub 2009/05/30. pmid:19478260.
  76. 76. Skarbinski J, Patel M, Winston CA, Patrick Kachur S, Massaga JJ, Bloland PB, et al. Monitoring insecticide-treated bednet possession and use: Comparison of data collected via health facility and household surveys—Lindi Region and Rufiji District, Tanzania, 2005. The American journal of tropical medicine and hygiene. 2006;75(5):3-. WOS:000242343900011.
  77. 77. Gobena T, Berhane Y, Worku A. Low long-lasting insecticide nets (LLINs) use among household members for protection against mosquito bite in Kersa, Eastern Ethiopia. BMC public health. 2012;12:914. Epub 2012/10/31. pmid:23107071; PubMed Central PMCID: PMCPMC3529106.
  78. 78. Graves PM, Ngondi JM, Hwang J, Getachew A, Gebre T, Mosher AW, et al. Factors associated with mosquito net use by individuals in households owning nets in Ethiopia. Malaria journal. 2011;10:354. Epub 2011/12/15. pmid:22165821; PubMed Central PMCID: PMCPMC3258293.
  79. 79. Astatkie A. Knowledge and practice of malaria prevention methods among residents of arba minch town and arba minch zuria district, southern ethiopia. Ethiopian journal of health sciences. 2010;20(3):185–93. Epub 2010/11/01. pmid:22434978; PubMed Central PMCID: PMCPMC3275842.
  80. 80. Bennett A, Smith SJ, Yambasu S, Jambai A, Alemu W, Kabano A, et al. Household possession and use of insecticide-treated mosquito nets in Sierra Leone 6 months after a national mass-distribution campaign. PloS one. 2012;7(5):e37927. Epub 2012/06/06. pmid:22666414; PubMed Central PMCID: PMCPMC3362537.
  81. 81. Nederhof AJ. Methods of coping with social desirability bias: A review. European Journal of Social Psychology. 1985;15(3):263–80.
  82. 82. Cotter C, Sturrock HJ, Hsiang MS, Liu J, Phillips AA, Hwang J, et al. The changing epidemiology of malaria elimination: new strategies for new challenges. Lancet. 2013;382(9895):900–11. Epub 2013/04/19. pmid:23594387.
  83. 83. Sturrock HJ, Hsiang MS, Cohen JM, Smith DL, Greenhouse B, Bousema T, et al. Targeting asymptomatic malaria infections: active surveillance in control and elimination. PLoS medicine. 2013;10(6):e1001467. Epub 2013/07/16. pmid:23853551; PubMed Central PMCID: PMCPMC3708701.
  84. 84. Gupta S, Snow RW, Donnelly C, Newbold C. Acquired immunity and postnatal clinical protection in childhood cerebral malaria. Proceedings Biological sciences / The Royal Society. 1999;266(1414):33–8. Epub 1999/03/19. pmid:10081156; PubMed Central PMCID: PMCPMC1689649.
  85. 85. Gupta S, Snow RW, Donnelly CA, Marsh K, Newbold C. Immunity to non-cerebral severe malaria is acquired after one or two infections. Nature medicine. 1999;5(3):340–3. Epub 1999/03/23. pmid:10086393.
  86. 86. Filmer D, Pritchett L. The Effect of Household Wealth on Educational Attainment: Evidence from 35 Countries. Population and Development Review. 1999;25(1):85–120.
  87. 87. Wang SJ, Lengeler C, Smith TA, Vounatsou P, Cisse G, Diallo DA, et al. Rapid urban malaria appraisal (RUMA) in sub-Saharan Africa. Malaria journal. 2005;4:40. Epub 2005/09/13. pmid:16153298; PubMed Central PMCID: PMCPMC1249588.
  88. 88. Ashton RA, Kefyalew T, Tesfaye G, Pullan RL, Yadeta D, Reithinger R, et al. School-based surveys of malaria in Oromia Regional State, Ethiopia: a rapid survey method for malaria in low transmission settings. Malaria journal. 2011;10(1):25. Epub 2011/02/04. pmid:21288368; PubMed Central PMCID: PMCPMC3039636.
  89. 89. Afrane YA, Zhou G, Githeko AK, Yan G. Utility of health facility-based malaria data for malaria surveillance. PloS one. 2013;8(2):e54305. Epub 2013/02/19. pmid:23418427; PubMed Central PMCID: PMCPMC3572108.
  90. 90. Agyepong IA, Kangeya-Kayonda J. Providing practical estimates of malaria burden for health planners in resource-poor countries. The American journal of tropical medicine and hygiene. 2004;71(2 Suppl):162–7. Epub 2004/08/28. pmid:15331833.
  91. 91. Rowe AK, Kachur SP, Yoon SS, Lynch M, Slutsker L, Steketee RW. Caution is required when using health facility-based data to evaluate the health impact of malaria control efforts in Africa. Malaria journal. 2009;8:209. Epub 2009/09/05. pmid:19728880; PubMed Central PMCID: PMCPMC2743707.
  92. 92. Carneiro I, Roca-Feltrer A, Griffin JT, Smith L, Tanner M, Schellenberg JA, et al. Age-patterns of malaria vary with severity, transmission intensity and seasonality in sub-Saharan Africa: a systematic review and pooled analysis. PloS one. 2010;5(2):e8988. Epub 2010/02/04. pmid:20126547; PubMed Central PMCID: PMCPMC2813874.
  93. 93. Mathanga DP, Luman ET, Campbell CH, Silwimba C, Malenga G. Integration of insecticide-treated net distribution into routine immunization services in Malawi: a pilot study. Tropical medicine & international health: TM & IH. 2009;14(7):792–801. Epub 2009/06/06. pmid:19497078.
  94. 94. McGregor IA. Epidemiology, malaria and pregnancy. The American journal of tropical medicine and hygiene. 1984;33(4):517–25. Epub 1984/07/01. pmid:6383091.
  95. 95. Brabin BJ. An analysis of malaria in pregnancy in Africa. Bulletin of the World Health Organization. 1983;61(6):1005–16. Epub 1983/01/01. pmid:6370484; PubMed Central PMCID: PMCPMC2536236.
  96. 96. Fried M, Domingo GJ, Gowda CD, Mutabingwa TK, Duffy PE. Plasmodium falciparum: chondroitin sulfate A is the major receptor for adhesion of parasitized erythrocytes in the placenta. Experimental parasitology. 2006;113(1):36–42. Epub 2006/01/25. pmid:16430888.
  97. 97. Duffy PE, Fried M. Antibodies that inhibit Plasmodium falciparum adhesion to chondroitin sulfate A are associated with increased birth weight and the gestational age of newborns. Infection and immunity. 2003;71(11):6620–3. Epub 2003/10/24. pmid:14573685; PubMed Central PMCID: PMCPMC219546.
  98. 98. Boudova S, Divala T, Mawindo P, Cohee L, Kalilani-Phiri L, Thesing P, et al. The prevalence of malaria at first antenatal visit in Blantyre, Malawi declined following a universal bed net campaign. Malaria journal. 2015;14:422. Epub 2015/10/30. pmid:26510414; PubMed Central PMCID: PMCPMC4625940.
  99. 99. Brabin BJ, Warsame M, Uddenfeldt-Wort U, Dellicour S, Hill J, Gies S. Monitoring and evaluation of malaria in pregnancy—developing a rational basis for control. Malaria journal. 2008;7 Suppl 1:S6. Epub 2009/02/20. pmid:19091040; PubMed Central PMCID: PMCPMC2604870.
  100. 100. Ataide R, Mayor A, Rogerson SJ. Malaria, primigravidae, and antibodies: knowledge gained and future perspectives. Trends Parasitol. 2014;30(2):85–94. Epub 2014/01/07. pmid:24388420.
  101. 101. Montagu D, Yamey G, Visconti A, Harding A, Yoong J. Where do poor women in developing countries give birth? A multi-country analysis of demographic and health survey data. PloS one. 2011;6(2):e17155. Epub 2011/03/10. pmid:21386886; PubMed Central PMCID: PMCPMC3046115.
  102. 102. Emerson PM, Ngondi J, Biru E, Graves PM, Ejigsemahu Y, Gebre T, et al. Integrating an NTD with one of "The big three": combined malaria and trachoma survey in Amhara Region of Ethiopia. PLoS Negl Trop Dis. 2008;2(3):e197. pmid:18350115; PubMed Central PMCID: PMCPMC2266804.
  103. 103. Nwokolo C, Ekpechi OL, Nwokolo U. New foci of endemic goitre in Eastern Nigeria. Transactions of the Royal Society of Tropical Medicine and Hygiene. 1966;60(1):97–108.
  104. 104. Price EW. Endemic elephantiasis of the lower legs in Rwanda and Burundi. Tropical and geographical medicine. 1976;28(4):283–90. Epub 1976/12/01. pmid:1014068.