Uganda’s Malaria Problem
In 2015 an estimated 212 million cases of malaria occurred globally and approximately 429,000 people died.11 The majority of these deaths were from children living in Africa. While malaria can be fatal, it is often managed by either the immune system (and may manifest as asymptomatic malaria) or by anti-malarial drugs. In regions that are closer to the equator, transmission of malaria is more intense and is transmitted year round—this makes high-endemicity countries in Africa, such as Uganda, targets of interest for malaria interventions.
While most endemic countries have experienced reductions in their malaria burden and related mortalities, Uganda remains plagued by the disease. Malaria accounts for about 28% of hospital admissions and 34% of outpatient visits, and 9-14% of hospital deaths.6,10 Uganda has a population of roughly 39 million people and 100% of those people are at risk of developing malaria infections, specifically involving the Plasmodium falciparum strain.6,7 According to the President’s Malaria Initiative,6 the Uganda Malaria Reduction Strategic Plan calls for a reduction in 2013-level malaria deaths to near zero by 2020 and a reduction in the parasite prevalence to less than 7%.6 Yet the malaria burden in Uganda has not decreased in recent years and some studies even indicate that it may be increasing.10 One singular approach to intervening in high-burden areas is unlikely to yield much success. Thus future intervention strategies should take a multifaceted approach to tackling malaria in specific world regions and communities. A combination of indoor residual spraying, treated bed nets, mass drug administration, mass screening, surveillance, and strict adherence to these strategies when they are implemented is the most likely approach to success in combatting malaria. These approaches should not only target obvious cases of malaria, but should take into account asymptomatic cases of malaria as well.
Background on Asymptomatic Malaria
Asymptomatic malarial infections are typically defined as the presence of parasitemia of any density in the absence of fever or other acute symptoms in those who have not received recent anti-malarial treatment.1 Asymptomatic malaria infections are sometimes referred to as chronic, as individuals who are asymptomatic still face low-grade hemolysis and even intermittent symptomatic bouts.1 Chronic carriage of the parasites can be detrimental to developing fetuses in infected mothers; placental malaria infection is associated with placental inflammation, fibrosis, and functional insufficiency, leading directly to miscarriage, preterm delivery, low birth weight, and peripartum hemorrhage and, thus, increased maternal and neonatal mortality.1
The ability of an individual’s immune system to suppress high loads of parasitemia and thus malarial symptoms may appear to be advantageous, but still poses a threat regarding local spread of the disease. Transmission of malaria from a chronically infected or asymptomatic individual to a healthy individual is still possible. In fact, P. vivax infection can be transmitted even before symptoms arise.2 Gametocytes, which are the transmissible stage of the malarial life cycle in humans, can remain in peripheral blood for several weeks after infection has been cleared.2 A small study which focused on comparing transmissibility based on gametocyte density in asymptomatic versus symptomatic individuals in Brazil found a small proportion of mosquitoes (1.2%) that fed on asymptomatic individuals developed oocytes in their mid-guts.2 This finding proved that malaria is still a local threat even when it appears to be individually manageable. Therefore, identification and treatment of asymptomatic individuals should be seen as a critical step in malaria control and interventions. Having strong mosquito surveillance programs is one step to controlling for the disease. Ugandan surveillance of malaria, however, is in need of improvement; according to the WHO, there is no mass screening for infection and “uncomplicated” cases of P. falciparum are not routinely admitted to health clinics.7 Mathematical models predict that accounting for asymptomatic individuals in community-wide drug treatment interventions will have a much larger impact on the prevention of disease transmission than interventions which target symptomatic individuals only.8
Treating Asymptomatic Malaria
Mass drug administration (MDA) has become both a treatment and prevention method of choice for rolling out infectious disease interventions. MDA covers all individuals in a population of interest (except when contraindicated). This means that both symptomatic and asymptomatic individuals will receive the antimalarial drug of choice in an intervention. While MDA can be rapidly effective in its reduction of disease transmission, reinfection, morbidity and mortality, if an intervention is unsuccessful in its attempts to eliminate malaria, then drug resistant strains of Plasmodium may become established. In an effort to curb positive selection for drug resistant genes within P. falciparum, the World Health Organization Malaria Policy Advisory Committee does not recommend the general use of MDA except in areas approaching elimination, experiencing epidemics, or complex emergencies.3
Artemisinin-based combination therapies (ACT) are of particular interest in MDA-focused interventions. This is due to the fact that artemisinin as a monotherapy has faced drug-resistant challenges from P. falciparum in regions of the world such as Cambodia, Thailand, Myanmar, Laos, and Vietnam.4 One research study found that isolates within asymptomatic individuals actually face higher rates of mutation compared to isolates within symptomatic individuals.4 These mutations occurred in the genes targeted by drugs such as artemisinin. Since asymptomatic malarial infections can result in local transmission as well as harbor genes for drug-resistance, it is important to emphasize the use of combination therapy in MDA interventions, rather than a mono-therapeutic approach. However, resistance to long-acting partner drugs utilized in ACT is concerning, both in regard to drugs that already suffer from resistance (such as amodiaquine) and those for which resistance may be selected for (such as lumefantrine, dihydroartemisinin).10 Using ACT promises an adequate cure rate and a delay in further drug-resistance development. ACTs have expanded globally in recent years. At the end of 2014, ACTs became the first-line treatment in 81 countries and approximately 392 million ACT treatment courses were delivered to public sectors in endemic countries.5 While MDA interventions appear to reduce the initial risk of malaria in their applied settings, few studies have shown sustained impact beyond six months following the use of MDA, and those that did were conducted on small islands or in highland settings.9,13 Ultimately, more studies are needed in order to fairly assess the effectiveness of MDA campaigns that specifically use ACT, rather than mono-therapies that may face threats from drug resistance.
Assessing the Effectiveness of Past MDA Interventions
In the majority of settings, a combination of IRS, treated bed nets, and ACT have shown success in reducing the burden of malaria in some endemic areas.9,10,12 Successes have included regions near Uganda such as Zanzibar and Coastal Kenya.10 Marked improvements in disease indicators have occurred mainly in areas with low baseline transmission intensity (whereas Uganda has one of the highest rates of malaria transmission in the world).6,7,10 Not only do these areas have lower baseline transmission rates, but they also tend to have better public health infrastructure and infrastructure for monitoring and researching malaria.10 That being said, more interventions with more refined approaches need to be conducted in areas of high transmission, such as Uganda. For instance, one study sought to decrease the burden of malaria in a high transmission setting, Sierra Leone, during the recent Ebola outbreak. Due to fears of being sent to Ebola holding centers or contracting the viral illness after a hospital visit, patient attendance to health facilities fell by about 40% from May to September of 2014.12 Such a phenomenon required a focused effort on mosquito control as the main form for preventing an increased disease burden of malaria during this time period. It was estimated that a lack of health facility attendance led to an increase in untreated malaria cases by about 88%.12 MDA using ACT was conducted in this particular study and once health facility attendance went back to normal levels, data showed that there was a significant reduction in the number of malaria inpatient cases reported throughout a 4-week period.12 However, the effects were relatively short-lived and served as an affirmation of the WHO guidelines for MDA use in high transmission settings—that it is appropriate during public health emergencies, but is otherwise not currently recommended.3,5
Specific to Uganda, past studies which have conducted ACT campaigns (using artesunate-SP) have had “unacceptably high” failure rates.10 Conversely, other ACT campaigns that utilized artemether-lumefantrine were much more efficacious.4,9,10 Despite a 100% cure-rate of malarial infections when this type of ACT has been used, recurrent parasitemia was seen within one month in half of all studied children.10 According to a systematic review, malaria prevalence in high-endemicity regions, such as Palestine and Cambodia, sustained reduced levels of the disease post-MDA for 4 months and 12 months, respectively.9 Interestingly enough, MDA appears to have a greater impact on P. falciparum compared to P. vivax,9 which indicates promise for ACT campaigns in high-endemicity African countries such as Uganda. Even greater support for conducting ACT in Uganda is provided by the fact that mosquito-resistance to new ACT does not appear to be a problem.10
One pilot study in Cambodia tested the use of ACT to control for malaria in a high-endemicity area. In this study, parasite rates were dramatically reduced from 52.3% to 2.6% after three years. The P. falciparum rate in children decreased from 37.0% to 1.4%, reaching 0% in eight of 17 villages, and in a second field study that included one additional mass treatment of artemisinin-piperaquine, the P. falciparum rate in children was reduced from 20.8% to 0% within six months.13 Ultimately, more studies that focus on ACT interventions need to be conducted to assess the long term effectiveness of this type of MDA on malaria burden, as this type of intervention appears to have great potential for high-endemicity regions.
Evaluating the Effectiveness of ACT for Reducing the Transmission of Malaria in Uganda
Objective
The primary goal of this study is to determine how effective the use of artemisinin-based combination therapy in MDA is in reducing the incidence of malaria (specifically P. falciparum) in Uganda. A secondary aspect of this study is to compare the effectiveness of MDA campaigns in combination with the use of treated bed nets to decrease blood parasitemia levels in individuals.
Design
For this study, a double-blind cluster randomized controlled trial design will be implemented. The study population will be on the group level and will be derived from 6 villages in Uganda: Arua, Apac, Tororo, Kyenjojo, Kanungu, and Mubende. These sites were selected based on a particular study that identified Arua, Apac, and Tororo as having higher average daily entymological inoculation rates, while Kyenjojo, Kanungu, and Mubende had more manageable parasitemia prevalence as a result of a much lower average daily entymological inoculation rate.14 Households from these 6 sites will serve as the clusters that will receive the intervention.
The intervention will take place a few weeks prior to the start of the rainy season in March; this should provide enough time for the drugs to have their best effect on individual protection from malaria infection. Drug dose variations must also be taken into account when administering to participating children and infants, in order to control for toxicity and adverse drug events. Cross-sectional surveys will be conducted with the Ugandan Ministry of Health to follow up with these sites. Measures of association of particular interest in this study are the risk ratio and relative risk reduction. In order to test the effectiveness of ACT in controlling malaria burden, two different test arms will be used. The first arm will contain an intervention providing ACT and treated bed nets and the second arm will serve as the control arm, providing a placebo drug and treated bed nets. Having a placebo ACT provides a means for blinding; the participants will not know whether or not they have received a working drug, nor will the health workers delivering the ACT and placebos know which is which.
A randomization algorithm will determine which of the households will receive the intervention containing ACT and treated bed nets versus which of the remaining households receive the placebo and treated bed nets. The counterfactual group in this study is the households not receiving the intervention. If possible, clusters will be based on geographically confined groups of households. There will not be a maximum or minimum population size for clusters, but the minimum distance between clusters will be set at 1 kilometer. Any between-cluster households will be considered “buffer zones”16
The primary outcome of interest is the incidence of malaria in the 6 villages (a population-level outcome). The secondary outcome of interest is blood parasitemia level, which will account for low levels of parasitemia which often lead to asymptomatic malaria (an individual-level outcome). Measuring such low levels of parasitemia density is now possible with diagnostic tools such as the cell microarray chip system. This tool is highly sensitive and provides rapid malaria diagnosis; in one study conducted in Uganda, the chip was able to detect parasitemia ranging from 0.0039% to 2.3438%.15 This type of tool will make detection of asymptomatic infection possible.
Sources of Data
Data collection will be conducted prospectively from participating households located in the 6 respective villages in Uganda. The sample size will be 30 clusters, from which 15 clusters will be randomized to the intervention arm and the other 15 to the control arm. The goal is to have approximately 1,500-3,000 individuals in each study arm. Informed consent will be necessary before this intervention is able to proceed. Population-level information such as disease incidence can be gathered from the Ugandan Ministry of Health, providing a rough idea of incidence per region and thus each village in the study. Individual-level information such as parasitemia levels will need to be collected by doing direct blood sampling and diagnostic tests to confirm parasitemia levels. If the cell microarray chip system is used for diagnostic testing then this process will be fast, efficient, and cheap (the inclusive costs of a cell microarray chip, fluorescent dye, and a push column are less than US $2.00).15 Health facilities near these 6 villages will be used for this diagnostic testing.
Initial data collection prior to intervention implementation will be performed so that a baseline incidence of disease and blood parasitemia levels can be established. Following this preliminary data collection, the various assigned interventions will take place and follow-ups will be conducted after 3 months, after 6 months, and after 12 months. These follow-ups will include resampling blood from a randomly selected group of participants to assess parasitemia levels and cross-sectional surveys to the Ugandan Ministry of Health to evaluate whether a reduction in incidence has occurred. Random selection for follow-up will aim to collect 50 observations per cluster.
Statistical Analysis Plan
Intra-cluster correlation coefficients (ICC) will be generated as well in order to determine whether or not there is within cluster variation. ICCs are generated by squaring the between-cluster variance and dividing it by the sum of the squared between-cluster variance and the within-cluster variance. ICCs are used to inform calculations for determining the effective sample size. Effective sample size will then be used to calculate power during the design phase of this study.
Poisson regression models using data collected from cross-sectional surveys from the Ugandan Ministry of Health will be used to analyze malaria incidence across villages. These regression models will be used to generate risk ratios and 95% confidence intervals. The calculated risk ratio can then be used to calculate the relative risk reduction to elucidate the percentage reduction in the incidence of disease in the population. In this case, the exposure is mosquitoes carrying malaria, specifically P. falciparum. Since this is a two-arm trial, there will be two of each measure of association reported.
Interpretation of Results
The assumption of the null hypothesis of this study is that there is no change in the incidence of malaria within the villages receiving ACT compared to those not receiving ACT. Conversely, should we calculate a p-value below a significance level of 0.05, then the assumption of the alternative hypothesis would be supported and the conclusion would be that there is a change in the incidence of malaria within the villages receiving ACT.
When interpreting a risk ratio, a value below one indicates a protective effect created by the intervention, or in this case, implementing ACT to control malaria transmission and blood parasitemia levels. If the risk ratio is greater than one then a harmful effect occurs due to a lack of an intervention, or not implementing ACT. In either situation, the further away the risk ratio moves from a value of one, the stronger the association becomes. The use of a 95% confidence interval in this study supports the smaller sample size (in this case when n = 30) and allows inference that the true population effect will fall between the calculated upper and lower bounds of the interval. If the confidence interval were to include a value of one within its range, then it can be concluded that there is no difference between the groups of villages in this study.
The ICC values obtained during the design phase will provide information on whether or not being a member of a certain cluster during the study impacts the results. If is equal to zero then cluster membership is not particularly informative regarding the variability in the outcome; if is equal to one then cluster membership is highly informative of this variability; and if equals anything in between, then cluster membership is somewhat informative, depending on how close to either extreme it is. If cluster membership does explain the variability in the outcome (having a large ICC), then outcomes within each cluster will be more correlated than if cluster membership had no effect on the outcome.
Strengths and Limitations
Cluster randomized controlled trials are well suited and commonly used to evaluate public health, health policy and health system interventions. They are ideal for testing interventions when the decision about whether or not to implement the intervention will be taken on behalf of a group rather than individuals. Cluster randomized controlled trials are also useful when the nature of the intervention carries a high risk of “contamination”, that is, when individuals randomized to different comparison groups are in frequent contact with one another and thus may be influenced by the alternative treatment. In the case of this study, participants from one village are less likely to come into contact with participants from a different village, due to geographic distance. The addition of a time-series approach to evaluating the effectiveness of the interventions lends support for the validity of the results, since changes in the effect of the interventions over time will be recorded.
Conversely, cluster randomized controlled trials face challenges that are not so problematic in individual randomized trials. These challenges include difficulty with blinding and analysis, such as the need to account for between-cluster variation. Ensuring that there is an adequate number of clusters for statistical power is an important point that must be accounted for when conducting these types of trials. Designs with more clusters and fewer observations per cluster are usually optimal from a statistical perspective, but are often sub-optimal from a cost and logistical perspective. A limitation specific to this study is sample size per village. While the sample size of this study should lend sufficient enough statistical power, an issue that may arise is that there are numerous individuals in each village who will not be participating in this study but who will still account for the malaria burden and will indirectly contribute to its local transmission (these are the households located in the “buffer zones” mentioned earlier). This means that within-cluster interactions could present a problem, especially if there is sharing of study resources among participants and non-participants. A potential solution would be to provide the intervention to those “buffer zone” households and simply exclude them from the study’s evaluation of ACT effectiveness.
This study is superior to previous studies because it will be focused on a type of MDA that does not face threats of drug-resistance from P. falciparum.10 Furthermore, this study seeks to expand the geographic distribution of a study area (see Figure 4), so that aspects such as elevation, climate differences, and behavioral differences can be further assessed regarding their contribution to malaria incidence. Having such information can help inform future intervention designs and perhaps increase efficiency and efficacy. A final point of difference with this study and previous ones is that this study seeks to assess the effectiveness of ACT in a high-endemicity region, whereas many previous studies have tested MDA combination therapies in low-endemicity regions.
Ensuring Community Engagement
While this intervention is designed to benefit participating villagers in Uganda, it is important to not assume that a hypothesized long term beneficial outcome will guarantee 100% compliance from study participants. One reason for not complying with the intervention might be due to the sharing of resources among individuals, such as neighbors, who are not participating in the study. This could result in participant parasitemia not being adequately treated (due to not taking the full dose of ACT) or in local malaria incidence not being accurately reflected (due to individual parasitemia not being fully treated). Furthermore, some participants may completely refuse to take the drugs offered for the intervention. Full participation without interactions among participating and non-participating villagers is one challenge that will be difficult to overcome. However, some studies have utilized “village malaria volunteers”13 which proved to be an effective method for monitoring and following up with villagers regarding proper ACT delivery and administration. The success of this intervention will depend on well-trained local health workers and village volunteers. It will also be important to enlist the cooperation of village leaders and to educate the general population to encourage active participation for the mass treatment to be successful.16 Arranging meetings with the villages prior to rolling out the intervention could be beneficial. In these meetings, the intervention layout and purposes would be explained to village leaders who would then provide consent and serve as a source of support for this intervention. Obtaining consent from households located in the 6 villages will also be an important step to ensuring participation; during these household stops, the purpose and procedures involved in the study will be explained and demographic information regarding the members of each household will be collected.16 The cost of this general approach should be feasible since the prices of certain ACTs are relatively low ($2 per treatment per person) and payments for local village malaria volunteers, according to a study that utilized them, is affordable.13
Concluding Remarks
Malaria is a complex disease that presents various challenges to its elimination in certain global settings. Understanding the different stages of the malarial lifecycle as well as the intricacies of transmission intensity is a key piece to the intervention puzzle. Future interventions must account for the risk of creating selective pressures for mosquitoes to develop drug-resistance, which clearly hamper intervention efforts. The use of treated bed nets should be a baseline requirement for these types of interventions, as they have been continuously proven to reduce malaria burden when actively used. Rolling out effective MDA interventions remains a challenge due to a lack of health infrastructure in many high-endemicity regions, such as Uganda. Before these interventions can have more realistically achievable goals, malaria surveillance and health infrastructure in certain regions must be heavily focused on for improvement.
Sources
1Chen, I., Clarke, S., Gosling, R., Hamainza, B., Killeen, G., Magill, A., O’Meara, W., Price, R., Riley, E. 2016. “Asymptomatic” malaria: A chronic and debilitating infection that should be treated. PLoS Med, 13(1):1-11.
2Lindblade, K., Steinhardt, L., Samuels, A., Kachur, P., Slutsker, L. 2014. The silent threat: Asymptomatic parasitemia and malaria transmission. Expert Review of Anti-Infective Therapy, 11(6):623-639.
3World Health Organization. 2015. The role of mass drug administration, mass screening and treatment, and focal screening and treatment for malaria. http://www.who.int/malaria/publications/atoz/role-of-mda-for-malaria.pdf?ua=1
4Nyunt, M., Shein, T., Zaw, N., Han, S., Muh, F., Lee, S., Han, JH., Thant, K., Han, ET., Kyaw, M. 2017. Molecular evidence of drug resistance in asymptomatic malaria infections, Myanmar, 2015. Emerging Infectious Diseases, 23(3):517-520.
5World Health Organization. 2016. Overview of malaria treatment. http://www.who.int/malaria/areas/treatment/overview/en/
6U.S. President’s Malaria Initiative. 2016. Uganda. https://www.pmi.gov/docs/default-source/default-document-library/country-profiles/uganda_profile.pdf?sfvrsn=24
7World Health Organization. 2017. http://www.who.int/malaria/publications/country-profiles/profile_uga_en.pdf?ua=1
8Bousema, T., Okell, L., Felger, I., Drakeley, C. 2014. Asymptomatic malaria infections: detectability, transmissibility, and public health relevance. Nature Reviews Microbiology, 1-8.
9Poirot, E., Skarbinski, J., Sinclair, D., Kachur, SP., Slutsker, L., Hwang, J. 2013. Mass drug administration for malaria. The Cochrane Database of Systematic Reviews, 12:1-160.
10Yeka, A., Gasasira, A., Mpimbaza, A., Achan, J., Nankabirwa, J., Nsobya, S., Staedke, S., Donnelly, M., Wabwire-Mangen, F., Talisuna, A., Dorsey, G., Kamya, M., Rosenthal, P. 2012. Malaria in Uganda: Challenges to control on the long road to elimination. I. Epidemiology and current control efforts. Acta Tropica, 121(3), 184–195.
11Centers for Disease Control and Prevention. 2017. Malaria. https://www.cdc.gov/malaria/index.html
12Aregawi, M., Smith, S., Sillah-Kanu, M., Seppeh, J., Kamara, A., Williams, R., Aponte, J., Bosman, A., Alonso, P. 2016. Impact of the mass drug administration for malaria in response to the ebola outbreak in Sierra Leone. Malaria Journal, 15:480.
13Song, J., Socheat, D., Tan, B., Dara, P., Deng, C., Sokunthea, S., Seila, S., Ou, F., Jian, H., Li, G. 2010. Rapid and effective malaria control in Cambodia through mass administration of artemisinin-piperaquine. Malaria Journal, 9:57.
14Okello, P., Bortel, W., Byaruhanga, A., Correwyn, A., Roelants, P., Talisuna, A., D’Alessandro, U., Coosemans, M. 2006. Variation in malaria transmission intensity in seven sites throughout Uganda. The American Journal of Tropical Medicine and Hygiene, 75(2):219-225.
15Yatsushiro, S., Yamamoto, T., Yamamura, S., Abe, K., Obana, E., Nogami, T., Hayashi, T., Sesei, T., Oka, H., Okello, J., Odongo, E., Alai, M., Olia, A., Anywar, D., Sakurai, M, Palacpac, N., Mita, T., Horii, T., Baba, Y., Kataoka, M. (2016). Application of a cell microarray chip system for accurate, highly sensitive, and rapid diagnosis for malaria in Uganda. Scientific Reports, 6, 30136.
16Shekalaghe, S., Drakeley, C., van den Bosch, S., Braak, R., van den Bijllaardt, W., Mwanziva, C., Semvua, S., Masokoto, A., Mosha, F., Teelen, K., Hermsen, R., Okell, L., Gosling, R., Sauerwein, R., Bousema, T. 2011. A cluster-randomized trial of mass drug administration with a gametocytocidal drug combination to interrupt malaria transmission in a low endemic area of Tanzania. Malaria Journal, 10:247.