The changing epidemiology of malaria requires adaptation of control measures to address shifts in geographical, behavioural and demographic risk characteristics. A deeper understanding of possible determinants of change is critically important. Local knowledge of the burden and features of the disease will be necessary to adapt interventions and maintain cost-effectiveness and equity. Factors that should be monitored include: changes in vector habits and insecticide resistance, shifts in risks of different demographic groups, drug resistance, climate, socio-economic changes, and coverage and impact of interventions.
Beyond Garki Project (named in recognition of the contributions of the Garki Project to understanding of malaria in Africa during 1969-76) is led by Malaria Consortium and implemented in collaboration with the Ethiopian Public Health Institute and Ministries of Health in Ethiopia and Uganda, alongside Regional/District Health Offices in the study sites.
The present study involved a longitudinal collection of data in selected study sites in Ethiopia and Uganda to monitor changes in malaria epidemiology in the context of interventions that have been implemented. Here, we present data on trends on several variables, including malaria epidemiology, vector behaviour and insecticide resistance, demographic and socio-economic factors, treatment-seeking behaviour, and coverage and impacts of interventions in four study sites from four surveys carried out between 2012 and 2014.
Cross-sectional surveys were repeatedly conducted in selected study sites in Uganda and Ethiopia. Two study sites were selected per country, representing different epidemiological settings in rural environments: Aduku (Apac District, Norther Region) and Butemba (Kyankwanzi District, Central Region) in Uganda; and Hembecho (Boloso Sore District) and Guba (Halaba District) both in Southern Nations, Nationalities and Peoples, in Ethiopia.
The main components include: household surveys, malariometric and serological surveys, entomological surveys, health facility-based morbidity studies, and climate studies. Simple random samples of households were selected in each site. The household surveys included interviews with household heads and women aged 15-49 years of age using handheld devices.
Blood samples were collected from each member of the sampled households, except infants under six months. Febrile subjects were tested with malaria rapid diagnostic tests (mRDTs) and treated. Dried blood spots were used for serological analysis of antibody responses to assess malaria transmission intensity over an extended period.
Anopheles mosquitoes were sampled to determine species composition, densities, behaviour and insecticide resistance using CDC light traps, exit traps, room searches, pyrethrum spray catch and human landing catch (HLC) methods. Laboratory analysis included molecular identification of species and detection of insecticide resistance markers, and sporozoite rate determination. Insecticide susceptibility tests were carried out in the Uganda sites.
The highest malaria prevalence rates were recorded in Butemba followed by Aduku, during all survey rounds. Both Ethiopia sites had low prevalence rates. Overall, the average prevalence rates were 7.0%, 21.1%, 1.0% and 0.5% in Aduku, Butemba, Hembecho and Guba sites, respectively. Nearly all infections in the Uganda sites were due to Plasmodium falciparum (with few infections caused by P. ovale and P. malariae). In Ethiopia, P. vivax was the most common, with the remainder of infections caused by P. falciparum.
Malaria prevalence declined steadily in the Ethiopia sites. In Uganda, prevalence increased significantly in both sites in rounds 2 and 3 (2013) compared with round 1. In round 4, prevalence significantly decreased in Aduku, but remained relatively high in Butemba although it declined slightly. Prevalence was relatively at its highest level in round 3 in both Uganda sites.
Malaria prevalence varied considerably among age groups in the Ugandan sites. In Ethiopia, prevalence was low and no differences were observed among the age groups. In Uganda, the variation among age groups was most pronounced in Butemba, with infection confined mainly to children under the age of 15, indicating high transmission intensity; the peak malaria prevalence was observed in children 5-9 years old.
The benefit of using insecticide treated nets (ITNs) was investigated by comparing malaria infection rates in individuals who slept under an ITN the previous night and those who did not. In both Uganda sites, use of ITNs was associated with significantly lower infection. In the Ethiopia sites, the difference was not significant, probably due to the very low overall prevalence.
Malaria prevalence also varied between socio-economic levels in both Uganda sites, whereas all socio-economic status (SES) groups had similarly low infection rates in both Ethiopia sites. In the Uganda sites, generally there was a trend of reduced prevalence rates with increasing SES levels.
In Aduku, Anopheles funestus s.s. dominated during survey rounds 1 and 2; A. arabiensis was more abundant than A. gambiae s.s. In Butemba, A. gambiae s.l. was the dominant species complex collected during all four rounds, which primarily consisted of A. gambiae s.s. Anopheles arabiensis. A. funestus s.l. and other anophelines were collected in small numbers.
In Ethiopia, all anophelines collected were A. gambiae s.l. except mosquitoes collected in round 4 in Hembecho which were all other anophelines. All A. gambiae s.l. were assumed to be A. arabiensis.
Assuming that by 22:00h all residents would be under an ITN, the proportion of contact before this time could be used as a proxy for the potential risk of malaria exposure. In round 1 the largest proportion of human-vector contact in Aduku for A. funestus s.l. (65%) took place before 22:00h, but in subsequent rounds this pattern was not observed although the sample sizes were too low to be conclusive. The majority of contact with A. gambiae s.l. took place after 22:00h.
A. gambiae s.l. (of which A. gambiae s.s. was dominant) was the main contributor to the human biting rate (HBR) in Butemba. Almost all human-vector contact took place indoors in all rounds and after 22:00h.
Infection rates varied across rounds in the main vector species. Although A. coustani s.l. was commonly observed in Aduku, none of the analysed samples were infected. In Aduku, the daily entomological inoculation rate (EIR) was highest in rounds 1 and 3 (0.19 and 0.14, respectively) while in rounds 2 and 4 it varied between 0.00 and 0.01. In Butemba, the EIR varied between 0.02 and 0.10.
In Aduku, there was no evidence of bendiocarb resistance in A. gambiae s.l. (mainly A. arabiensis) during rounds 1 and 2, but in rounds 3 and 4 some survivors were seen and test results were classified as suspected resistant. There was no resistance detected against the organophosphate pirimiphos-methyl. In rounds 1 and 4, full susceptibility was observed and tests primarily consisted of A. arabiensis. In rounds 2 and 3, both suspected resistance and confirmed resistance were observed. Permethrin was tested twice: in one test (which consisted mainly of A. arabiensis) all mosquitoes were susceptible while in another test done exclusively with A. gambiae s.s. resistance was detected.
In Butemba, no resistance of A. gambiae s.s. was detected against bendiocarb or pirimiphos-methyl. When mosquitoes were pre-exposed to the synergist piperonyl butoxide (PBO), mortality after exposure with deltamethrin or permethrin increased suggesting at least some contribution of a metabolic resistance mechanism to the observed resistance. No A. arabiensis survived following PBO pre-exposure suggesting that this mechanism is stronger in this species compared to A. gambiae s.s., but more studies are needed to confirm this observation.
In Butemba, kdr-L1014S (kdr-east) frequencies were uniformly high (>89.6%) in A. gambiae s.s. Almost all A. arabiensis analysed were homozygous susceptible at the kdr-L1014S locus.
In Aduku, A. gambiae s.s. collected by various trapping techniques showed a high frequency of kdr-L1014S (80.8 and 100%), whereas those from the resistance tests showed varying levels (41.7-87.5%).
Kdr-L1014S genotype frequencies were determined for A. gambiae s.s. survivors and non-survivors following exposure to pyrethroids or DDT. In Aduku, A. gambiae s.s. that survived the exposure had a significantly higher kdr-L1014S genotype frequency than non-survivors (X2=24.2, p<0.001). In Butemba, a similar observation was made (X2=15.1, p=0.001), although for both survivors and non-survivors kdr-L1014S genotype frequencies were high.
The kdr-L1014F mutation was observed in low frequencies in A. arabiensis in Aduku. Frequencies were slightly higher in A. gambiae s.s. from Butemba where genotype frequency was 15.0 and 8.2% in rounds 3 and 4, respectively.
The acetylcholinesterase (ace-1 G119S) target mutation which confers resistance against carbamate and organophosphate insecticides was investigated for a subset of A. gambiae s.s. and A. arabiensis from the collections and tests. All samples tested were homozygous susceptible for the ace-1 G119S mutation for each site.
ITN coverage varied between countries, sites and rounds. In the Uganda sites, the proportion of households with enough nets to cover all of their residents (universal coverage, assuming one net for every two people) was below 50% during rounds 1-3. An increase in coverage was observed following the 2013-14 universal coverage campaign and 68.1% and 51.7% of households owned enough nets to cover all of their residents in round 4 in Apac and Butemba, respectively.
In Ethiopia, coverage was low in Guba with less than 6% of households having enough nets to cover all of their residents during all rounds. In Hembecho almost all households (98.7%) owned at least one ITN during round 1, but universal coverage was only achieved in 24.9% of households. Net coverage reduced in subsequent rounds due to net attrition: 9.2% of households owned enough nets to cover all of their residents in round 4.
ITN ownership increased with socioeconomic status in the two Uganda sites and in Guba, while in Hembecho no statistically significant differences were observed in the percent of households that had enough nets to cover all of their residents by wealth quintile.
In the Uganda sites, the majority of nets were obtained from public sources. In Ethiopia, almost all nets were received from public sources. Overall, 99% of untreated nets and 14% of LLINs were purchased during the study period in the Uganda sites. The proportion of nets purchased increased with increasing wealth.
In the Uganda sites, ITN use was highest in round 4: 71.5% and 64.3% slept under an ITN the night before the survey in Aduku and Butemba respectively. In Hembecho, use was high in round 1 (69.3%) but declined over time and in round 4 only 16.3% of people had slept under an ITN. In Guba, ITN use was low in all rounds and ranged between 11.1% in round 2 and 23.1% in round 3. Net use among those with access indicates whether available nets were being used. ITN use rates among those with access were significantly more than general use rates in all study sites, indicating that access was an important determinant of use.
In the Ethiopia sites, nets being too old or dirty were often mentioned as reasons for not using nets. In Aduku, nets were most often not used because they were not hung or the usual user did not sleep in the house, followed by nets being too old. In Butemba, main reasons for not using a net were ‘net too old’ followed by ‘no space to hang net’.
Indoor residual spraying (IRS) was implemented in all sites except Butemba. Aduku was sprayed twice a year with bendiocarb from 2010 onwards. Guba was sprayed with deltamethrin only in 2010 and 2012. Hembecho was sprayed annually during 2010-2013 and twice in 2012. Deltamethrin was used until February 2012 and propoxur was used in August 2012. Spray coverage was estimated from the household questionnaire by asking respondents if their house had been sprayed in the preceding 12 months. In Aduku, spray coverage in rounds 1 and 2 was above the desired 80% coverage level. In rounds 3 and 4 coverage was 79.4 and 76.4%, respectively. In both Ethiopian sites, coverage was greater than 90% during years when IRS was implemented.
In the Uganda sites, women of child bearing age were asked about their use of intermittent preventive treatment with sulfadoxine-pyrimethamine (SP) during their last pregnancy (IPTp) (IPTp is not implemented in Ethiopia). In Aduku and Butemba, 38.6% and 40.3% of women who gave birth within two years preceding the survey took three or more doses of IPTp in round 4.
Treatment was sought for the majority of febrile children in Aduku (77-98%), Butemba (77-95%) and Guba (70-84%). Treatment seeking behaviour in Hembecho was lower and varied between 42% and 61%. If treatment was sought, antimalarials were given to more than three quarters of the children in Aduku and Butemba. There was a general decline in the percentage of children who received antimalarials in Guba across rounds while in Hembecho use rates were lower in rounds 1 and 4.
In Aduku and Butemba, the proportion of febrile children treated with an antimalarial in the first 24 hours of onset of fever varied between 36% and 61% across all rounds and no significant change was detected over time. In Guba, a significant increase was seen in early treatment of fevers in rounds 2-4 compared to round 1.
Among children for whom treatment was sought, the majority of cases (57-100%) in Ethiopia received a diagnostic test (either mRDT or microscopy) and no significant differences were seen between rounds in each site. In Aduku, 56-80% of cases seen were tested for malaria while in Butemba this varied between 30.0 and 65% and no significant differences were seen between rounds in both sites. The majority of these diagnostic tests were performed in public facilities for all sites and rounds except in Aduku where more tests were performed in private facilities compared to the other sites. Generally, the majority of diagnostic tests were mRDTs in the Ugandan sites and microscopy in the Ethiopian sites.back