Artemisinin‐based combination therapy for treating uncomplicated malaria
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To compare the effects of ACTs with other available ACT and non‐ACT antimalarial combinations for treating uncomplicated P. falciparum malaria.
A secondary objective is to explore the effects of the combinations on P. vivax infection.
Background
Malaria is a disease of global public health importance. Its social and economic burden is a major obstacle to human development in many of the worlds’ poorest countries. In heavily affected countries, malaria alone accounts for as much as 40% of public health expenditure, 30% to 50% of hospital admissions, and up to 60% of outpatient visits (WHO 2007). It has an annual incidence of over 500 million episodes and is the cause of more than a million deaths, most of them in infants, young children, and pregnant women (WHO 2007).
Malaria is transmitted from person to person by the bite of mosquitoes infected with the protozoan parasite Plasmodium. Four Plasmodium species are capable of causing malaria in humans: P. falciparum, P. vivax, P. malariae, and P. ovale. Of these P. falciparum is responsible for over 90% of cases in Africa and almost all of the malaria deaths worldwide (RBM 2005). P. vivax is also common and often presents as a co‐infection with P. falciparum in a single illness (Mayxay 2004). Uncomplicated malaria is the mild form of the disease and presents as a febrile illness with headache, tiredness, muscle pains, abdominal pains, rigors (severe shivering), and nausea and vomiting. If left untreated P. falciparum malaria can rapidly develop into severe malaria with anaemia (low haemoglobin in the blood), hypoglycaemia (low blood sugar), renal failure (kidney failure), pulmonary oedema (fluid in the lungs), convulsions (fitting), coma, and eventually death (WHO 2006). A clinical diagnosis of malaria can be confirmed by detection of the malaria parasite in the patient's blood. This has traditionally been done by light microscopy but increasingly rapid diagnostic tests are being used.
Resistance of P. falciparum to the traditional antimalarial drugs (such as chloroquine, amodiaquine, mefloquine, and sulfadoxine‐pyrimethamine) is a growing problem and is thought to have contributed to increased malaria mortality in recent years (WHO 2006). Chloroquine resistance has now been documented in all regions except Central America and the Caribbean. There is high‐level resistance to sulfadoxine‐pyrimethamine throughout South‐East Asia and increasingly in Africa. Mefloquine resistance is common in the border areas of Cambodia, Myanmar, and Thailand, but it is uncommon elsewhere. Resistance of P. vivax to sulfadoxine‐pyrimethamine is also increasing, and chloroquine resistance has been reported in some parts of Asia and Oceania (WHO 2006).
Artemisinin‐based antimalarials
Artemisinin and its derivatives (such as artesunate, artemether, and dihydroartemisinin) are relatively new antimalarial drugs, with a unique structure and mode of action. The first published report of clinical trials appeared in the Chinese Medical Journal in 1979 (QACRG 1979). Until recently there had been no reported resistance to the artemisinin derivatives; however, the possibility of emerging resistance on the Thai‐Cambodian border is currently being investigated (WHO 2008).
Artemisinin derivatives have been shown to produce faster relief of clinical symptoms and faster clearance of parasites from the blood than other antimalarial drugs (McIntosh 1999; Adjuik 2004; WHO 2006). When used as monotherapy, the short half‐life of the artemisinin derivatives (and rapid elimination from the blood) means that patients must take the drug for at least seven days (Meshnick 1996; Adjuik 2004). Failure to complete the course, due to the rapid improvement in clinical symptoms, can lead to a high treatment failure rate even in the absence of drug resistance. Artemisinin derivatives are therefore usually given with another longer‐acting drug, with a different mode of action, in a combination known as artemisinin‐based combination therapy or ACT. These combinations can then be taken for shorter durations than artemisinin alone, and they have a lower treatment failure rate (White 1999; WHO 2006).
The artemisinin derivatives may also reduce the development of gametocytes (the sexual form of the malaria parasite that is capable of infecting mosquitoes) and consequently the carriage of gametocytes in the peripheral blood (Price 1996; Targett 2001). This reduction in infectivity has the potential to reduce the post‐treatment transmission of malaria (particularly in areas of low or seasonal transmission) with significant public health benefits (WHO 2006).
Artemisinin and its derivatives are generally reported as being safe and well tolerated, and the safety profile of ACTs may be largely determined by the partner drug (WHO 2006; Nosten 2007). Studies of artemisinin derivatives in animals have reported significant neurotoxicity (brain damage), but this has not been seen in human studies (Price 1999). Animal studies have also shown adverse effects on the early development of the fetus, but the artemisinin derivatives have not been fully evaluated during early pregnancy in humans (Nosten 2007). Other reported adverse events include gastrointestinal disturbance (stomach upset), dizziness, tinnitus (ringing in the ears), neutropenia (low levels of white blood cells), elevated liver enzymes (a marker for liver damage), and electrocardiographic (ECG) abnormalities (changes in cardiac conduction). Most studies, however, have found no evidence of ECG changes, and only non‐significant changes in liver enzymes (WHO 2006; Nosten 2007). The incidence of type 1 hypersensitivity (allergic) reactions is reported to be approximately 1 in 3000 patients (Nosten 2007).
Assessing antimalarial efficacy
The World Health Organization (WHO) recommends that first‐line antimalarials should have a treatment failure rate of less than 10%, and failure rates higher than this should trigger a change in treatment policy (WHO 2006). Treatment failure can be classified as early (with failure to clear the parasite from the blood or progression from uncomplicated to severe malaria) or late (with reappearance of the parasite after initial clearance) (Bloland 2003).
The late reappearance of P. falciparum parasites in the blood can be due to failure of the drug to completely clear the original parasite (a recrudescence) or due to a new infection, which is especially common in areas of high transmission. A molecular genotyping technique called polymerase chain reaction (PCR) can be used in clinical trials to distinguish between recrudescence and new infection, giving a clearer picture of the efficacy of the drug and its post‐treatment prophylactic effect (White 2002; Cattamanchi 2003).
The WHO now recommends a minimum follow‐up period of 28 days for antimalarial efficacy trials, but longer periods of follow may be required for antimalarials with long elimination half‐lives (White 2002; Bloland 2003). This is because treatment failure due to true recrudescence of malaria parasites may be delayed until the drug concentration falls below the minimum concentration required to inhibit parasite multiplication, which may be beyond 28 days. The WHO recommends 42 days' follow up for trials involving lumefantrine and 63 days for trials of mefloquine (Bloland 2003).
P. vivax malaria
P. vivax differs from P. falciparum in generally producing a milder illness and in having a liver stage known as a hypnozoite. These hypnozoites can lie dormant in the liver following an acute infection and cause spontaneous relapses at later dates.
As P. vivax often co‐exists with P. falciparum in a single illness, it is important to assess the effect of ACTs on the P. vivax parasite (Mayxay 2004; WHO 2006). ACTs have been shown to clear P. vivax from the peripheral blood, but they do not have a substantial effect on the liver stage of the parasite (Pukrittayakamee 2000). Although ACTs cannot provide a radical cure from P. vivax, their ability to delay the eventual relapse of P. vivax and provide a prolonged malaria free period may produce significant public health benefits.
It is important to note that when P. vivax parasitaemia occurs following initial treatment, PCR is unable to distinguish a recrudescence of the original infection (due to failure to clear the parasite from the peripheral blood) from a spontaneous relapse (due to failure to clear the liver stage) (WHO 2006).
Choice of combination treatment
The WHO now recommends that P. falciparum malaria is always treated using a combination of two drugs that act at different biochemical sites within the parasite (WHO 2006). If a parasite mutation producing resistance arises spontaneously during treatment, the parasite should then be killed by the partner drug, thereby reducing or delaying the development of resistance to the artemisinin derivatives and increasing the useful lifetime of the individual drugs (White 1996; White 1999; WHO 2006). This policy emerged at the time when ACTs were primarily being considered, but other possibilities such as amodiaquine combined with sulfadoxine‐pyrimethamine (non‐ACTs) are also available.
The decision of which ACT to adopt into national malaria control programmes has been based on a combination of research and expert opinion. Systematic reviews can contribute to this decision by providing evidence on the:
-
relative effects on cure between combinations;
-
absolute cure levels achieved by a drug in a particular region;
-
safety and risk of adverse effects of the combination;
-
impact on gametocytes;
-
impact on haemoglobin levels; and
-
relative effects on P. vivax.
Other information that is also important to decision‐making include:
-
the appropriateness of the partner drug within a locality, based on informed judgements related to regional and national overviews of drug resistance and the degree of intensity of malaria transmission;
-
public health judgements based on known pharmacology, epidemiology, and biology of the parasite to avoid the development of parasite drug resistance;
-
the simplicity of the treatment regimen (co‐formulated products are generally preferred as they reduce the availability and use of monotherapy, which may in turn reduce the development of resistance);
-
the cost (since the ACT is likely to represent a large percentage of the annual health expenditure in highly endemic countries); and
-
other concerns such as fetal toxicity and teratogenicity.
To contribute to informed decision‐making, we will examine the comparative effects of ACTs for which co‐formulated products are currently available or shortly to be made available. We will also include trials that have used co‐packaged or loose preparations of these same ACTs to provide information on relative effects of different treatment options. While recent Cochrane Reviews have synthesized the evidence around individual ACT comparisons (Bukirwa 2005; Omari 2005; Bukirwa 2006; Omari 2006), this review broadens the inclusion criteria and will pool the data in a single Cochrane Review. A comprehensive list of the available drugs and the treatment comparisons that we will assess is shown in Table 1.
| Question | Analysis | Comparisons |
| 1. How does artesunate plus amodiaquine perform?
| 1.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 1.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 1.3 | vs artemether‐lumefantrine (6 doses) | |
| 1.4 | vs dihydroartemisinin‐piperaquine | |
| 1.5 | vs artesunate plus mefloquine | |
| 2. How does artemether‐lumefantrine (6 doses) perform?
| 2.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 2.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 2.3 | vs artesunate plus amodiaquine | |
| 2.4 | vs dihydroartemisinin‐piperaquine | |
| 2.5 | vs artesunate plus mefloquine | |
| 3. How does dihydroartemisinin‐piperaquine perform?
| 3.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 3.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 3.3 | vs artesunate plus amodiaquine | |
| 3.4 | vs artemether‐lumefantrine (6 doses) | |
| 3.5 | vs artesunate plus mefloquine | |
| 4. How does artesunate plus mefloquine perform?
| 4.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 4.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 4.3 | vs artesunate plus amodiaquine | |
| 4.4 | vs artemether‐lumefantrine (6 doses) | |
| 4.5 | vs dihydroartemisinin‐piperaquine |
aTo contribute to informed decision‐making, the review is limited to artemisinin combination therapies (ACTs) for which co‐formulated products are currently available or shortly to be made available (trials using co‐packaged or loose preparations of these same ACTs are included).
Objectives
To compare the effects of ACTs with other available ACT and non‐ACT antimalarial combinations for treating uncomplicated P. falciparum malaria.
A secondary objective is to explore the effects of the combinations on P. vivax infection.
Methods
Criteria for considering studies for this review
Types of studies
Randomized controlled trials. Quasi‐randomized studies will be excluded.
Types of participants
Adults and children (including pregnant women and infants) with symptomatic, microscopically confirmed, uncomplicated P. falciparum malaria.
Trials that include participants with P. vivax co‐infection and mono‐infection are also eligible.
Types of interventions
Intervention
Three‐day course of an ACT (fixed dosed, co‐blistered, or individually packaged (loose)).
Control
Three‐day course of an alternative ACT or other antimalarial treatment.
As described inTable 1, this review version is limited to the following drugs: artesunate plus amodiaquine; artemether‐lumefantrine (six doses); dihydroartemisinin‐piperaquine; artesunate plus mefloquine; artesunate plus sulfadoxine‐pyrimethamine; and amodiaquine plus sulfadoxine‐pyrimethamine.
Types of outcome measures
Primary
Total failure at days 28, 42, and 63; PCR‐adjusted and PCR‐unadjusted.
Secondary
-
Gametocyte carriage at day 7 or 14 (preference for day 14 in data analysis).
-
Gametocyte development (negative at baseline, and positive at follow up).
-
Gametocyte clearance (positive at baseline, and negative at follow up).
-
Change in haemoglobin from baseline (minimum 28 day follow up).
-
P. vivax parasitaemia at day 28, 42, or 63 (all participants).
-
P. vivax parasitaemia at day 28, 42, or 63 (only participants with P. vivax at baseline).
Adverse events
-
Deaths occurring during follow up.
-
Serious adverse events (life threatening, causing admission to hospital, or discontinuation of treatment).
-
Haematological and biochemical adverse effects (eg neutropenia, liver toxicity).
-
Early vomiting.
-
Other adverse events.
Search methods for identification of studies
We will attempt to identify all relevant trials regardless of language or publication status (published, unpublished, or in press).
We will search the following databases using the search terms detailed in Table 2: Cochrane Infectious Disease Group Specialized Register; Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library; MEDLINE; EMBASE; and LILACS. We will also search the metaRegister of Controlled Trials (mRCT) using 'malaria' and 'arte* OR dihydroarte*' as search terms.
| Search set | CIDG SRa | CENTRAL | MEDLINEb | EMBASEb | LILACSb |
| 1 | malaria | malaria | malaria | malaria | malaria |
| 2 | arte* | arte* | arte* | arte* | arte* |
| 3 | dihydroarte* | dihydroarte* | dihydroarte* | dihydroarte* | dihydroarte* |
| 4 | amodiaq* | amodiaq* | amodiaq* | amodiaq$ | amodiaq$ |
| 5 | lumefantrine | lumefantrine | lumefantrine | lumefantrine | lumefantrine |
| 6 | Coartem* | Coartem* | Coartem* | Coartem$ | Coartem$ |
| 7 | mefloquine | mefloquine | mefloquine | mefloquine | mefloquine |
| 8 | 2 or 3 | 2 or 3 | 2 or 3 | 2 or 3 | 2 or 3 |
| 9 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 |
| 10 | 1 and 8 and 9 | 1 and 8 and 9 | 1 and 8 and 9 | 1 and 8 and 9 | 1 and 8 and 9 |
| 11 | — | — | Limit 10 to humans | Limit 10 to human | — |
aCochrane Infectious Diseases Group Specialized Register.
bSearch terms used in combination with the search strategy for retrieving trials developed by The Cochrane Collaboration (Levebvre 2008); upper case: MeSH or EMTREE heading; lower case: free text term.
We will contact individual researchers working in the field, organizations including the World Health Organization, and pharmaceutical companies (Atlantic, Guilin, Holleykin, HolleyPharm, Mepha, Novartis, Parke‐Davis, Pfizer, Sanofi‐Aventis, Roche) for information on unpublished trials.
We will also check the reference lists of all trials identified by the methods described above.
Data collection and analysis
Selection of studies
David Sinclair (DS) and Babalwa Zani (BZ) will review the results of the literature search and obtain full‐text copies of all potentially relevant trials. DS will scrutinize each trial report for evidence of multiple publications from the same data set. DS and BZ or Hasifa Bukirwa (HB) will then independently assess the eligibility of each trial for inclusion in this review using an eligibility form based on the inclusion criteria. We will resolve any disagreements through discussion or, where necessary, by consulting Paul Garner (PG). If further clarification is necessary we will attempt to contact the trial authors for further information. We will list the trials that are deemed ineligible and the reasons for their exclusion.
Data extraction and management
DS and BZ/HB will independently extract data using a pre‐tested data extraction form. We will extract data on trial characteristics including methods, participants, interventions, and outcomes as well as data on dose and drug ratio of the combinations. We will resolve disagreements through discussion, review of the trial report, or, where necessary, by consulting PG.
We will extract the number randomized and the number analysed in each treatment group for each outcome. We will calculate and report the loss to follow up in each group.
For dichotomous outcomes, we will record the number of participants experiencing the event and the number of participants in each treatment group. For continuous outcomes, we will extract the arithmetic means and standard deviations for each treatment group together with the numbers of participants in each group. If the data have been reported using geometric means, we will record this information and extract standard deviations on the log scale. If medians have been extracted we will also extract ranges. Since time‐to‐event outcomes are reported using survival analysis, estimates of the hazards ratio and standard errors of the log hazards ratios will be extracted. If these estimates are not available we aim to obtain survival probabilities or the numbers at risk from Kaplan‐Meier curves and life tables (Parmar 1998).
Primary outcome
The primary analysis draws on the WHO's protocol for assessing and monitoring antimalarial drug efficacy (Bloland 2003). This protocol has been used to guide most efficacy trials since its publication in 2003, even though it was designed to assess the level of antimalarial resistance in the study area rather than for comparative trials. As a consequence a high number of randomized participants are excluded from the final efficacy outcome as losses to follow up or voluntary or involuntary withdrawals. For this reason we will conduct a sensitivity analysis where we will aim to restore the integrity of the randomization process (as is usual in trial analysis) and test the robustness of the results to this methodology. (For a summary of the methodology and sensitivity analysis see Table 3.)
| Analysis | Participants | PCRb‐adjusted | PCR‐unadjusted | ||
| Numerator | Denominator | Numerator | Denominator | ||
| Primary analysis | Exclusions after enrolment | Excludedc | Excluded | Excluded | Excluded |
| Missing or indeterminate PCR | Excluded | Excluded | Included as failures | Included | |
| New infections | Excluded | Excluded | Included as failures | Included | |
| Sensitivity analysis 1d | As 'Primary analysis' except: missing or indeterminate PCR | Included as failures | Included | — | — |
| Sensitivity analysis 2e | As 'Sensitivity analysis 1' except: new infections | Included as successes | Included | — | — |
| Sensitivity analysis 3f | As 'Sensitivity analysis 2' except: exclusions after enrolment | Included as failures | Included | Included as failures | Included |
| Sensitivity analysis 4g | As 'Sensitivity analysis 2' except: exclusions after enrolment | Included as successes | Included | Included as successes | Included |
aNote: participants who were found to not satisfy the inclusion criteria after randomization are removed from all calculations.
bPCR: polymerase chain reaction.
c"Excluded" means removed from the calculation.
dTo re‐classify all indeterminate or missing PCR results as treatment failures in the PCR‐adjusted analysis.
eTo re‐classify all PCR‐confirmed new infections as treatment successes in the PCR‐adjusted analysis. (This analysis may overestimate efficacy as PCR is not wholly reliable and some recrudescences may be falsely classified as new infections. Also some participants may have gone on to develop a recrudescence after the new infection.)
fTo re‐classify all exclusions after enrolment (losses to follow up, withdrawn consent, other antimalarial use, or failure to complete treatment) as treatment failures. For PCR‐unadjusted total failure this represents a true worse‐case scenario.
gTo re‐classify all exclusions after enrolment (losses to follow up, withdrawn consent, other antimalarial use, or failure to complete treatment) as treatment successes.
We will initially present the results grouped by continent (before pooling them) to aid the interpretation of the forest plots by specifying where the drugs have been studied.
The primary outcome (total failure) relates solely to failure due to P. falciparum. For both PCR‐adjusted and PCR‐unadjusted total failure, participants who experience P. vivax during follow up will be retained in the calculation if they were treated with chloroquine and continued in the follow up. As long as they did not go on to develop P. falciparum parasitaemia they will be classified as treatment successes. We will exclude from the calculation those participants who experienced P. vivax and were removed from the trial's follow up at the time of P. vivax parasitaemia.
PCR‐adjusted total failure
PCR‐adjusted total failure (P. falciparum) will be calculated as the sum of early treatment failures and late treatment failures due to PCR‐confirmed recrudescence. Early treatment failure is defined as the development of danger signs or severe malaria on days one, two, or three in the presence of parasitaemia. Late treatment failure will be defined as recurrence of P. falciparum parasitaemia on or after day four (this includes the WHO's definitions of late clinical failure and late parasitological failure (Bloland 2003)). Late treatment failures that occur between days four and 14 can be assumed to be recrudescences of the original parasite without the need for PCR genotyping (unless genotyped in the trial). It is not always possible to guarantee that individual trials used the standard WHO definitions. We will accept the trial authors' data unless we have specific reason to reclassify an individual participant or reject the data. Where this is done we will state clearly the reasons for doing so.
We will exclude from the calculations participants who were found not to fulfil the inclusion criteria after randomization and those who were lost to follow up, withdrew consent, took other antimalarials, or failed to complete treatment. Participants with indeterminate PCR results, missing PCR results, or PCR‐confirmed new infections will be treated as involuntary withdrawals and also excluded from the calculation.
PCR‐unadjusted total failure
PCR‐unadjusted total failure (P. falciparum) will be calculated as the sum of early treatment failures and late treatment failures (without PCR adjustment). We will exclude from the calculation those participants who were found not to fulfil the inclusion criteria after randomization and those who were lost to follow up, withdrew consent, took other antimalarials, or failed to complete treatment.
Secondary outcomes and adverse events
In a secondary analysis we will examine the effects of ACTs on P. vivax. We anticipate four types of study: trials of P. falciparum where P. vivax co‐infections were excluded but P. vivax occurred during follow up; trials of P. falciparum where P. vivax co‐infections were included at baseline; trials that included both P. falciparum and P. vivax mono‐infections as well as co‐infections at baseline; and trials that only recruited patients with P. vivax. We will report the occurrence rate of P. vivax parasitaemia during follow up at days 28, 42, and 63. Where possible, we will stratify this analysis into participants with P. vivax infection at baseline and those negative for P. vivax at baseline. There are no methods available to distinguish a recrudescence of P. vivax from a relapse so these rates may reflect either failure of the ACT or the ability of the ACT to delay the eventual relapse.
Other secondary outcomes will be presented using forest plots, tables, or narrative summaries as appropriate.
Adverse events will be presented in a table with a narrative summary as appropriate. We will specifically seek to report deaths, serious adverse events, biochemical, and haematological effects and early vomiting of the drug. Other adverse events will be grouped into their pathophysiological system. Where we are able to perform a meta‐analysis of adverse events, the denominator will only include participants from trials where we are sure that the trial investigators were adequately monitoring for this adverse event. For example, for biochemical adverse events we will only include trials that measured biochemical indicators routinely, and for adverse symptoms we will only include trials that stated that participants were actively and routinely questioned about possible adverse symptoms.
Assessment of risk of bias in included studies
DS and BZ/HB will independently assess the risk of bias for each trial using The Cochrane Collaboration's tool for assessing the risk of bias (Higgins 2008). Differences of opinion will be discussed with PG. We will follow the guidance to make judgements on the risk of bias in six domains: sequence generation; allocation concealment; blinding (of participants, personnel, and outcome assessors); incomplete outcome data; selective outcome reporting; and other sources of bias. We will categorize these judgements as 'yes' (low risk of bias), 'no' (high risk of bias), or 'unclear'. Where our judgment is unclear we will attempt to contact the trial authors for clarification.
This information will be used to guide the interpretation of the data that is presented. In some instances, the risk of bias assessment may lead to removal of the trial data from the meta‐analysis or complete exclusion of the trial from this review. This judgement will be done in discussion with PG and the reasoning clearly stated.
Measures of treatment effect
We will analyse data using Review Manager 5. Dichotomous data will be presented and combined using risk ratios. For continuous data summarized by arithmetic means and standard deviations, data will be combined using mean differences. Where continuous data have been summarized using geometric means or when hazard ratios are available, they will be combined on the log scale using the generic inverse variance method and reported on the natural scale. Risk ratios and mean differences will be accompanied by 95% confidence intervals. Medians and ranges will only be reported in a table.
Dealing with missing data
If data from the trial reports are insufficient, unclear, or missing, we will attempt to contact the trial authors for additional information. If we judge the missing data to render the result uninterpretable, we will exclude the data from the meta‐analysis and state the reason.
Assessment of heterogeneity
We will assess for heterogeneity amongst trials by inspecting the forest plots, applying the chi‐squared test with a 10% level of statistical significance, and also using the I2 test with a value of 50% used to denote moderate levels of heterogeneity.
Assessment of reporting biases
We will investigate the possibility of publication bias using a funnel plot.
Data synthesis
Treatments will be compared directly using meta‐analyses of randomized pair‐wise comparisons in the main analysis. A secondary analysis will compare several treatments simultaneously using multiple treatment comparison (MTC) methodology. MTC methodology involves using both direct evidence from randomized pair‐wise comparisons and indirect (non‐randomized) evidence from across trials to produce pooled treatment effect estimates (Lu 2004). The assumption of exchangeability must hold for MTC methods to be appropriate. Exchangeability will be considered by comparing clinical and methodological characteristics of the trials (Bucher 1997).
We will stratify the analyses by the control treatment. For outcomes that are measured at different time points, we will stratify the analysis by the time point.
Subgroup analysis and investigation of heterogeneity
We will conduct the following subgroup analyses to explore potential sources of heterogeneity: participant age; intensity of malaria transmission (low to moderate versus high malaria transmission); and drug dose (comparing regimens where there are significant variations in drug dose). A pertinent example of a drug dose subgroup analysis is the comparison of mefloquine as a once daily dose of 8 mg/kg for three days with the traditional 15 mg/kg on day one and 10 mg/kg on day two.
Sensitivity analysis
We will conduct a series of sensitivity analyses to investigate the robustness of the methodology used in the primary analysis. We will aim to restore the integrity of the randomization process by adding excluded groups back into the analysis in a stepwise fashion (see Table 3 for details).
| Question | Analysis | Comparisons |
| 1. How does artesunate plus amodiaquine perform?
| 1.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 1.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 1.3 | vs artemether‐lumefantrine (6 doses) | |
| 1.4 | vs dihydroartemisinin‐piperaquine | |
| 1.5 | vs artesunate plus mefloquine | |
| 2. How does artemether‐lumefantrine (6 doses) perform?
| 2.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 2.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 2.3 | vs artesunate plus amodiaquine | |
| 2.4 | vs dihydroartemisinin‐piperaquine | |
| 2.5 | vs artesunate plus mefloquine | |
| 3. How does dihydroartemisinin‐piperaquine perform?
| 3.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 3.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 3.3 | vs artesunate plus amodiaquine | |
| 3.4 | vs artemether‐lumefantrine (6 doses) | |
| 3.5 | vs artesunate plus mefloquine | |
| 4. How does artesunate plus mefloquine perform?
| 4.1 | vs amodiaquine plus sulfadoxine‐pyrimethamine |
| 4.2 | vs artesunate plus sulfadoxine‐pyrimethamine | |
| 4.3 | vs artesunate plus amodiaquine | |
| 4.4 | vs artemether‐lumefantrine (6 doses) | |
| 4.5 | vs dihydroartemisinin‐piperaquine | |
| aTo contribute to informed decision‐making, the review is limited to artemisinin combination therapies (ACTs) for which co‐formulated products are currently available or shortly to be made available (trials using co‐packaged or loose preparations of these same ACTs are included). | ||
| Search set | CIDG SRa | CENTRAL | MEDLINEb | EMBASEb | LILACSb |
| 1 | malaria | malaria | malaria | malaria | malaria |
| 2 | arte* | arte* | arte* | arte* | arte* |
| 3 | dihydroarte* | dihydroarte* | dihydroarte* | dihydroarte* | dihydroarte* |
| 4 | amodiaq* | amodiaq* | amodiaq* | amodiaq$ | amodiaq$ |
| 5 | lumefantrine | lumefantrine | lumefantrine | lumefantrine | lumefantrine |
| 6 | Coartem* | Coartem* | Coartem* | Coartem$ | Coartem$ |
| 7 | mefloquine | mefloquine | mefloquine | mefloquine | mefloquine |
| 8 | 2 or 3 | 2 or 3 | 2 or 3 | 2 or 3 | 2 or 3 |
| 9 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 | 4 or 5 or 6 or 7 |
| 10 | 1 and 8 and 9 | 1 and 8 and 9 | 1 and 8 and 9 | 1 and 8 and 9 | 1 and 8 and 9 |
| 11 | — | — | Limit 10 to humans | Limit 10 to human | — |
| aCochrane Infectious Diseases Group Specialized Register. | |||||
| Analysis | Participants | PCRb‐adjusted | PCR‐unadjusted | ||
| Numerator | Denominator | Numerator | Denominator | ||
| Primary analysis | Exclusions after enrolment | Excludedc | Excluded | Excluded | Excluded |
| Missing or indeterminate PCR | Excluded | Excluded | Included as failures | Included | |
| New infections | Excluded | Excluded | Included as failures | Included | |
| Sensitivity analysis 1d | As 'Primary analysis' except: missing or indeterminate PCR | Included as failures | Included | — | — |
| Sensitivity analysis 2e | As 'Sensitivity analysis 1' except: new infections | Included as successes | Included | — | — |
| Sensitivity analysis 3f | As 'Sensitivity analysis 2' except: exclusions after enrolment | Included as failures | Included | Included as failures | Included |
| Sensitivity analysis 4g | As 'Sensitivity analysis 2' except: exclusions after enrolment | Included as successes | Included | Included as successes | Included |
| aNote: participants who were found to not satisfy the inclusion criteria after randomization are removed from all calculations. | |||||
