Study characteristics

Methods

Dates of study: 2014‐2017

Location of study: Zambia

Malaria endemicity (prevalence): Clusters stratified by ≤ 10% and > 10% parasite prevalence prior to randomization. This study includes the low transmission strata [Low].

Transmission season: January to May

Malaria species: Plasmodium falciparum

Vector species: Anopheles funestus (hereafter An funestus) and Anophelesgambiae (hereafter An gambiae)

Antimalarial drug resistance context: Widespread resistance to chloroquine and sulfadoxine‐pyrimethamine, but no evidence of resistance to artemisinin.

Study design: Cluster‐randomized trial (3 arms: MDA, fMDA, and a no fMDA or MDA control; only MDA and control arms included in this review)

Statistical power: 80% power (alpha = 0.05) to detect a 50% reduction in parasite prevalence accounting for clustering, and 80% power (alpha = 0.05) to detect a 50% reduction in parasite incidence accounting for clustering

For cluster‐RCTs

Unit of randomization: Health facility catchment area (HFCA)

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using the study‐provided ICC to calculate effective sample sizes

Adjustment method: Random intercept for HFCA

ICC (estimated at the HFCA level in the low transmission strata): 0.06517 (in 2014); 0.01753 (in 2015); 0.01515 (in 2016); 0.04184 (in 2017)

Number of clusters randomized: 20

Number of clusters analysed: 20

Number of people: 82,866 (low), 184,928 (total)

Average cluster size: 9246

Features: Stratified by low vs high (below and above 10%) transmission strata and HFCA population size prior to randomization of clusters

Participants

Age groups included: All ages above 3 months. Children < 3 months and pregnant women in first trimester excluded from MDA with dihyrdroartemisinin piperaquine, but offered appropriate dose of antimalarial treatment in accordance with national policy if found to be RDT positive (note: testing was performed in fMDA and MDA arms, but all eligible persons were treated in the MDA arm, irrespective of RDT result).

Population targeted

Intervention: 37,694

Comparison: 45,172

Interventions

Intervention:

Drug/dose:

• Ages ≥ 14 years (> 45 kg): Dihydroartemisinin plus piperaquine (120 mg/960 mg as 3 tablets) given for three days

• Ages 8‐13 years (25 to 40 kg): Dihydroartemisinin plus piperaquine (80 mg/640 mg as 2 tablets) given for three days

• Ages 1‐7 years (10 to 23 kg): Dihydroartemisinin plus piperaquine (40 mg/320 mg as 1 tablet) given for three days

• Ages 3 months to 1 year (8 kg): Dihydroartemisinin plus piperaquine (20 mg/160 mg as ½ tablet) given for three days

Number of rounds (timing/dates): 4 (December 2014 at the end of the dry season, February‐March 2015 at the start of the rainy season, September‐October 2015 during dry season, and February‐March 2016 at start of the rainy season)

Interval: Variable

Duration implemented: 15 months

Coverage (%): 79% in round 1, 63% in round 2, 76% in round 3, and 66% in round 4

Co‐interventions: Baseline IRS household coverage in the preceding 12 months was 6.9%; Baseline household LLIN coverage of at least 1 net was 70.3%; Enhanced standard of care was scaled up in the study area, which consisted of RDT or microscopic confirmation of all suspected cases presenting to health facility and treating positives with artemether‐lumefantrine and reactive case detection in areas with manageable case counts.

Comparison:

Type: No MDA and no placebo

Co‐interventions: Baseline IRS household coverage in the preceding 12 months was 16.9%; Baseline household LLIN coverage of at least 1 net was 75.3%; Enhanced standard of care was scaled up in the study area, which consisted of RDT or microscopic confirmation of all suspected cases presenting to health facility and treating positives with artemether‐lumefantrine and reactive case detection in areas with manageable case counts.

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys of P falciparum prevalence by RDT in children ≥ 3 months and < 6 years

Time points: Pre‐MDA (April‐May 2014), During MDA (April‐May 2015), and Post‐MDA 1‐3 months (April‐May 2016)

Sample size (range): 372‐545 (intervention); 361‐453 (comparison)

Parasitaemia incidence

Measurement: Prospective cohort of persons ≥ 3 months of age to capture P falciparum incidence by RDT

Time points: Followed through 2 months post‐MDA (January 2015 ‐ May 2016)

Sample size: 410 (intervention); 326 (comparison)

Confirmed malaria illness incidence

Measurement: Passive surveillance as measured by outpatient department confirmed malaria cases (by microscopy or RDT) reported through routine health management information systems data

Time points: Pre‐MDA (January‐May 2013 combined with January‐May 2014) and Post‐MDA 1‐3 months (January 2015 ‐ May 2016)

Adverse effects (AEs) (reported in both MDA and fMDA arms)

A total of 687 AEs (0.24% of participants and 0.43% of treatments) were reported and one was a serious AE. The most common AEs reported were stomach pains, dry cough, and vomiting.

Notes

ClinicalTrials.gov: NCT02329301

Outcomes stratified by low and high transmission strata as specified a priori by study design (defined as above and below 10% parasite prevalence)

Two out of ten low transmission strata clusters received two additional rounds of programmatic MDA (for a total of six rounds) at 10 and 12 months following the fourth round of trial MDA, however no outcomes were analysed following the first round of programmatic MDA.

Abbreviations:

ICC = intracluster correlation coefficient, fMDA = focal mass drug administration, IRS = indoor residual spraying, LLIN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, RDT = rapid diagnostic test

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Following stratification by transmission setting and health facility catchment area population size, clusters were randomly allocated to study arms using random allocation rule.

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

Low risk

The proportion of households that received IRS in the previous 12 months was higher at baseline in control (17%) compared to MDA (7%) clusters, but other baseline characteristics were balanced. Analysis adjusted for baseline differences.

Contamination protection

Low risk

Households within a 3 km buffer around HFCA borders were excluded from the sampling frame for parasite prevalence and incidence.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation, but access to antimalarials through enhanced standard of care was comparable across arms.

Blinding of participants and personnel (performance bias)
Parasitaemia incidence

Low risk

Participants were not blinded to allocation, but parasitaemia incidence was assessed through active follow‐up in a prospective cohort cleared of parasitaemia at baseline.

Blinding of participants and personnel (performance bias)
Confirmed malaria illness incidence

Unclear risk

Participants were not blinded to allocation, which may have impacted care‐seeking.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Unclear risk

Outcome assessment by microscopy not described.

Blinding of outcome assessment (detection bias)
Parasitaemia incidence

Low risk

Blood samples for malaria testing collected from all individuals, irrespective of study arm at follow‐up visits.

Blinding of outcome assessment (detection bias)
Confirmed malaria illness incidence

Unclear risk

Blinding of health facility staff to allocation arm was not described.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Simple random sample of children at baseline and follow‐up.

Incomplete outcome data (attrition bias)
Parasitaemia incidence

Unclear risk

"No statistically significant differences in primary outcomes or covariates between individuals" included in analysis and those lost to follow‐up, "although fewer individuals were lost to follow up in high transmission areas than in low transmission areas". Based on this information provided, it is unclear whether attrition bias was a concern once lost to follow‐up is stratified by baseline endemicity.

Incomplete outcome data (attrition bias)
Confirmed malaria illness incidence

Low risk

Based on routine data from health management information system with established reporting from January 2011 onwards.

Selective reporting (reporting bias)

Low risk

All pre‐stated outcomes of interest were reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Mixed models were used to adjust for clustering by investigators; however, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 2014 to 2017

Location of study: Zambia

Malaria endemicity (prevalence): Clusters stratified by ≤ 10% and > 10% parasite prevalence prior to randomization. This study includes the high transmission strata [High].

Transmission season: January to May

Malaria species: Plasmodium falciparum

Vector species: An funestus and Angambiae

Antimalarial drug resistance context: Widespread resistance to chloroquine and sulfadoxine‐pyrimethamine, but no evidence of resistance to artemisinin.

Study design: Cluster‐randomized trial (3 arms: MDA, fMDA, and a no fMDA or MDA control; only MDA and control arms included in this review)

Statistical power: 80% power (alpha = 0.05) to detect a 50% reduction in parasite prevalence accounting for clustering, and 80% power (alpha = 0.05) to detect a 50% reduction in parasite incidence accounting for clustering

For cluster RCTs

Unit of randomization: Health facility catchment area (HFCA)

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using the study‐provided ICC to calculate effective sample sizes

Adjustment method: Random intercept for HFCA

ICC (estimated at the HFCA level for the high transmission strata): 0.16142 (in 2014); 0.11301 (in 2015); 0.08066 (in 2016); 0.13479 (in 2017)

Number of clusters randomized: 20

Number of clusters analysed: 20

Number of people: 102,062 (high), 184,928 (total)

Average cluster size: 9246

Features: Stratified by low vs high (below and above 10%) transmission strata and HFCA population size prior to randomization of clusters

Participants

Age groups included: All ages above 3 months. Children < 3 months and pregnant women in first trimester excluded from MDA with dihydroartemisinin piperaquine, but offered appropriate dose of antimalarial treatment in accordance with national policy if found to be RDT positive (note: testing was performed in fMDA and MDA arms, but all eligible persons were treated in the MDA arm, irrespective of RDT result).

Population targeted

Intervention: 45,442

Comparison: 56,620

Interventions

Intervention:

Drug/dose:

• Ages ≥ 14 years (> 45 kg): Dihydroartemisinin plus piperaquine (120 mg/960 mg as 3 tablets) given for three days

• Ages 8‐13 years (25 to 40 kg): Dihydroartemisinin plus piperaquine (80 mg/640 mg as 2 tablets) given for three days

• Ages 1‐7 years (10 to 23 kg): Dihydroartemisinin plus piperaquine (40 mg/320 mg as 1 tablet) given for three days

• Ages 3 months to 1 year (8 kg): Dihydroartemisinin plus piperaquine (20 mg/160 mg as ½ tablet) given for three days

Number of rounds (timing/dates): 4 (December 2014 at the end of the dry season, February to March 2015 at the start of the rainy season, September to October 2015 during dry season, and February to March 2016 at start of the rainy season)

Interval: Variable

Duration implemented: 15 months

Coverage (%): 79% in round 1, 63% in round 2, 76% in round 3, and 66% in round 4

Co‐interventions: Baseline IRS household coverage in the preceding 12 months was 6.9%; Baseline household LLIN coverage of at least 1 net was 70.3%; Enhanced standard of care was scaled up in the study area, which consisted of RDT or microscopic confirmation of all suspected cases presenting to health facility and treating positives with artemether‐lumefantrine and reactive case detection in areas with manageable case counts.

Comparison:

Type: No MDA and no placebo

Co‐interventions: Baseline IRS household coverage in the preceding 12 months was 16.9%; Baseline household LLIN coverage of at least 1 net was 75.3%; Enhanced standard of care was scaled up in the study area, which consisted of RDT or microscopic confirmation of all suspected cases presenting to health facility and treating positives with artemether‐lumefantrine and reactive case detection in areas with manageable case counts.

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys of P falciparum prevalence by RDT in children ≥ 3 months and < 6 years

Time points: Pre‐MDA (April to May 2014), During MDA (April to May 2015), and Post‐MDA 1 to 3 months (April to May 2016)

Sample size (range): 348 to 490 (intervention); 332 to 505 (comparison)

Parasitaemia incidence

Measurement: Prospective cohort of persons ≥ 3 months of age to capture P falciparum incidence by RDT

Time points: Followed through 2 months post‐MDA (January 2015 to May 2016)

Sample size: 371 (intervention); 368 (comparison)

Confirmed malaria illness incidence

Measurement: Passive surveillance as measured by outpatient department confirmed malaria cases (by microscopy or RDT) reported through routine health management information systems data

Time points: Pre‐MDA (January to May 2013 combined with January to May 2014) and Post‐MDA 1 to 3 months (January to May 2015)

Adverse effects (AEs) (reported in both MDA and fMDA arms)

A total of 687 AEs (0.24% of participants and 0.43% of treatments) were reported and one was a serious AE. The most common AEs reported were stomach pains, dry cough, and vomiting.

Notes

ClinicalTrials.gov: NCT02329301

Outcomes stratified by low and high transmission strata as specified a priori by study design (defined as above and below 10% parasite prevalence)

Abbreviations:

ICC = intracluster correlation coefficient, fMDA = focal mass drug administration, IRS = indoor residual spraying, LLIN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, RDT = rapid diagnostic test

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Following stratification by transmission setting and health facility catchment area population size, clusters were randomly allocated to study arms using random allocation rule.

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

Low risk

The proportion of households that received IRS in the previous 12 months was higher at baseline in control (17%) compared to MDA (7%) clusters, but other baseline characteristics were balanced. Analysis adjusted for baseline differences.

Contamination protection

Low risk

Households within a 3 km buffer around HFCA borders were excluded from the sampling frame for parasite prevalence and incidence.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation, but access to antimalarials through enhanced standard of care was comparable across arms.

Blinding of participants and personnel (performance bias)
Parasitaemia incidence

Low risk

Participants were not blinded to allocation, but parasitaemia incidence was assessed through active follow‐up in a prospective cohort cleared of parasitaemia at baseline.

Blinding of participants and personnel (performance bias)
Confirmed malaria illness incidence

Unclear risk

Participants were not blinded to allocation, which may have impacted care‐seeking.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Unclear risk

Outcome assessment by microscopy not described.

Blinding of outcome assessment (detection bias)
Parasitaemia incidence

Low risk

Blood samples for malaria testing collected from all individuals, irrespective of study arm at follow‐up visits.

Blinding of outcome assessment (detection bias)
Confirmed malaria illness incidence

Unclear risk

Blinding of health facility staff to allocation arm was not described.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Simple random sample of children at baseline and follow‐up.

Incomplete outcome data (attrition bias)
Parasitaemia incidence

Unclear risk

"No statistically significant differences in primary outcomes or covariates between individuals" included in analysis and those lost to follow‐up, "although fewer individuals were lost to follow up in high transmission areas than in low transmission areas". Based on this information provided, it is unclear whether attrition bias was a concern once lost to follow‐up is stratified by baseline endemicity.

Incomplete outcome data (attrition bias)
Confirmed malaria illness incidence

Low risk

Based on routine data from health management information system with established reporting from January 2011 onwards.

Selective reporting (reporting bias)

Low risk

All pre‐stated outcomes of interest were reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Mixed models were used to adjust for clustering by investigators. However, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 1960 to 1961

Location of study: Burkina Faso

Malaria endemicity (prevalence): 56.1% prevalence in children 0 to 10 years at baseline in control [High]

Transmission season: June to December

Malaria species: P falciparum, P ovale, P malariae

Vector species: An gambiae, An funestus, Anopheles nili (hereafter An nili)

Antimalarial drug resistance context: Not described

Study design: Controlled before‐and‐after study (6 arm study: CQ+PQ or AQ+PQ every 4 weeks, CQ+PQ or AQ+PQ every 2 weeks, CQ+PQ or AQ+PQ every 4 weeks with IRS, CQ+PQ or AQ+PQ every 2 weeks with IRS, IRS only, non‐IRS control; the 3 arms with IRS are excluded in this review since the population size and number of villages for the IRS only control were not reported)

Statistical power: Not described

Participants

Age groups included: All ages

Population targeted

Intervention (CQ+PQ or AQ+PQ every 4 weeks): 1890 in 5 villages

Intervention (CQ+PQ or AQ+PQ every 2 weeks): 2560 in 3 villages

Comparison (non‐IRS control): 6 villages, population size not described

Interventions

Drug/dose (for all intervention arms receiving MDA):

• Ages ≥10 years: Chloroquine‐primaquine (600 mg/15 mg) or amodiaquine‐primaquine (600 mg/15 mg) as a single dose

• Ages 5 to 9 years: Chloroquine‐primaquine (400 mg/10 mg) or amodiaquine‐primaquine (400 mg/10 mg) as a single dose

• Ages 0 to 4 years: Chloroquine‐primaquine (200 mg/5 mg) or amodiaquine‐primaquine (200 mg/5 mg) as a single dose

Intervention (CQ+PQ or AQ‐PQ every 4 weeks, "low frequency"):

Number of rounds (timing/dates): 8 (June, July, August, September, October, November, December 1960)

Interval: Every 28 days

Duration implemented: 7 months (June to December 1960)

Coverage (%): 75% to 91% per round

Co‐interventions: None

Intervention (CQ+PQ or AQ+PQ every 2 weeks, "high frequency"):

Number of rounds (timing/dates): 15 (June to December 1960)

Interval: Every 14 days

Duration implemented: 7 months (June to December 1960)

Coverage (%): 84% to 97% per round

Co‐interventions: None

Comparison (non‐IRS control):

Type: No MDA and no placebo

Co‐interventions: None

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys in all children ages 2 to 9 years every 4 months (microscopy)

Time points: Pre‐MDA (June 1960), During‐MDA (October 1960), and at 3 months post‐MDA (March 1961)

Sample size (range): 274 to 348 (intervention: CQ+PQ or AQ+PQ every 4 weeks); 390 to 467 (intervention: CQ+PQ or AQ+PQ every 2 weeks); 217 to 691 (comparison: non‐IRS control)

Gametocytaemia prevalence

Measurement: Cross‐sectional surveys in all children ages 2 to 9 years every 4 months (microscopy)

Time points: Pre‐MDA (June 1960), During MDA (October 1960), and at 3 months post‐MDA (March 1961)

Sample size (range): 274 to 348 (intervention: CQ+PQ or AQ+PQ every 4 weeks); 390 to 467 (intervention: CQ+PQ or AQ+PQ every 2 weeks); 217 to 691 (comparison: non‐IRS control)

Notes

Samples for outcome assessment in June 1960 were collected prior to MDA distribution; therefore the survey in June 1960 was considered as "pre‐MDA".

Abbreviations:

AQ = amodiaquine, CQ = chloroquine, IRS = indoor residual spraying, MDA = mass drug administration, PQ = primaquine

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

High risk

Assignment to intervention or control was not randomized although drug assignment within intervention was randomized in arms with more than one drug.

Allocation concealment (selection bias)

High risk

Non‐randomized controlled study (controlled before‐and‐after study)

Baseline imbalance (selection bias)

Unclear risk

Baseline characteristics were not described.

Contamination protection

Unclear risk

No information is provided to assess the risk of contamination.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

High risk

Microscopists were not blinded to study arm.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Outcomes were assessed in all children ages 2 to 9 years in the study area.

Selective reporting (reporting bias)

Low risk

All pre‐stated outcomes were reported.

Other bias

Low risk

No other bias detected.

Study characteristics

Methods

Dates of study: 2013 to 2015

Location of study: Kayin (Karen) state, Myanmar

Malaria endemicity (prevalence): Plasmodium falciparum prevalence 11.0% in MDA villages and 5.4% in control villages at baseline by ultrasensitive polymerase chain reaction (uPCR); P vivax prevalence 18.9% in MDA villages and 17.5% in control villages at baseline by uPCR [Low ‐ estimated P falciparum slide prevalence 1.2%]

Transmission season: June to October

Malaria species: P falciparum and P vivax

Vector species: Anopheles minimus s.l., An maculatus s.l., and An dirus s.l.

Antimalarial drug resistance context: “Area where artemisinin resistance is firmly established”

Statistical power: For the multi‐country trial (Landier 2017 MMRa; Tripura 2018 KHM; Pongvongsa 2018 LAO; von Seidlein 2019 VNM), 80% power (alpha = 0.05) to detect a 95% reduction in parasite prevalence from a baseline prevalence of 10%

For cluster RCTs

Unit of randomization: Village

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using ICC values estimated from the study data to calculate effective sample sizes

Adjustment method: Generalized estimating equations

ICC: 0.03512 (P falciparum outcomes at baseline before MDA), 0 (P falciparum outcomes at post‐MDA 1 to 3, 4 to 6, and 7 to 12 months); 0 (P vivax outcomes at baseline before MDA, post‐MDA 4 to 6 months and 7 to 12 months), 0.000798 (P vivax outcomes at post‐MDA 1 to 3 months)

Number of clusters randomized: 4

Number of clusters analysed: 4

Number of people: 3238

Average cluster size: 810

Features: Two village pairs (4 villages) were established by geographical proximity, population size, and parasite prevalence. Within each pair, one village was randomly selected to receive early MDA, while the other village received deferred MDA

Participants

Age groups included: All ages ≥ 6 months. All pregnant women in their first trimester were excluded from MDA and pregnant women in any trimester were excluded from primaquine.

Population targeted

Intervention: 1434

Comparison: 1804

Interventions

Intervention:

Drug/dose: Dihydroartemisinin (7 mg/kg) plus piperaquine (55 mg/kg) administered once a day for 3 days with a single dose of primaquine (0.25 mg/kg)

Number of rounds (timing/dates): 3 (May, June, July 2013 or June, July, August 2013)

Interval: Every 1 month

Duration implemented: 3 months

Coverage (%): 66% in round 1, 56% in round 2, and 65% in round 3

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Comparison:

Type: Deferred MDA administered in control villages in January, February, and March 2014

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Outcomes

Parasitaemia prevalence (P falciparum andP vivax )

Measurement: Cross‐sectional surveys in all ages every 3 months by uPCR; all individuals present at the time of the survey in the study villages were sampled.

Time points: Pre‐MDA (May 2013) and at 1 (Aug 2013), 4 (Nov 2013), and 7 (Jan 2014) months post‐MDA

Sample size (range): 419 to 689 (intervention) and 750 to 848 (control)

Confirmed malaria illness incidence (P falciparum andP vivax )

Measurement: Passive case detection at malaria posts for measured or self‐reported fever (≥ 37.5C) and confirmed P falciparum or P vivax infection by RDT or microscopy.

Time points: 7 (May 2013 to January 2014) months post‐MDA

Adverse effects (AEs)

AEs monitored by active surveillance through structured interviews on the second, third, and seventh day of MDA treatment course and by passive surveillance via reporting to medical assistants in mobile clinics located in study villages during MDA rounds. From interviews, no serious AEs were reported. The most common AEs were dizziness (n=192) and pruritus (n=17). There were three reports of black urine, from glucose 6‐phosphate dehydrogenase (G6PD) deficient (report prior to PQ), G6PD‐normal (48 hours after PQ), and G6PD‐heterozygous individuals. Among participants reporting passively to mobile clinics, there were 23 serious AEs and 15 deaths that were not drug‐related, 9 moderate AEs of which 4 were possibly related to the study drug, and 191 mild AEs of which 1 was highly likely related and 7 were possibly related to the study drug.

Notes

One of four sites from a multi‐country trial in Southeast Asia (ClinicalTrials.gov: NCT01872702)

Abbreviations:

ICC = intracluster correlation coefficient, LLITN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, uPCR = ultrasensitive polymerase chain reaction

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"The randomisation was based on computer‐generated random numbers provided by the trial statistician"

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

High risk

Small number of clusters (4 villages) randomized.

Contamination protection

High risk

Intervention and control clusters were pair‐matched by geographical proximity, which likely led to a high risk of contamination due to population movement.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation.

Blinding of participants and personnel (performance bias)
Confirmed malaria illness incidence

Unclear risk

Participants were not blinded to allocation, which may have impacted care‐seeking.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Laboratory staff performing PCR were unaware of the study arm allocation of samples.

Blinding of outcome assessment (detection bias)
Confirmed malaria illness incidence

Unclear risk

Unclear if the health facility staff performing malaria testing were aware of which study arm participants were assigned to.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Parasitaemia surveys were performed in all individuals aged six months or older residing in the study villages.

Incomplete outcome data (attrition bias)
Confirmed malaria illness incidence

Low risk

Data collected at malaria health posts with dedicated study staff.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Although no adjustment for clustering was performed by investigators, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes.

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 2014 to 2017

Location of study: Myanmar

Malaria endemicity (prevalence): P falciparum < 20 cases per 1000 population per year and P falciparum/P vivax 2.7% at baseline by rapid diagnostic test (RDT) [Low]

Transmission season: June to August

Malaria species: P falciparum, P vivax, P ovale, P malariae

Vector species: Not described

Antimalarial drug resistance context: Artemisinin resistance reported; Kelch 13 in 57% (54/94) of samples at baseline.

Study design: Cluster‐randomized trial

Statistical power: Not statistically powered for outcomes

For cluster RCTs

Unit of randomization: Village

Adjusted analyses for clustering: No; however, in this review, we performed cluster adjustment of the raw data using the study‐provided ICC to calculate effective sample sizes

Adjustment method: Not applicable

ICC: 0.056

Number of clusters randomized: 16

Number of clusters analysed: 16

Number of people: 8721

Average cluster size: 554

Features: Intervention and control clusters were pair‐matched based on P falciparum prevalence (+/‐ 8%), geographical proximity and distance to main road

Participants

Age groups included: All ages ≥ 1 year and pregnant women in first trimester excluded. No primaquine administered to pregnant women in other trimesters.

Population targeted

Intervention: 5481

Comparison: 3240

Interventions

Intervention:

Drug/dose: Dihydroartemisinin (7 mg/kg) plus piperaquine (55 mg/kg) administered once a day for three days with a single dose of primaquine (0.25 mg/kg)

Number of rounds (timing/dates): 3 (March, April, May 2015, during dry season)

Interval: Every 1 month

Duration implemented: 3 months

Coverage (%): 90% completed at least one round; Round 1: 86%, Round 2: 86%, Round 3: 88%

Co‐interventions: LLITNs distributed at the start of study; routine malaria control by village health workers.

Comparison:

Type: No MDA and no placebo

Co‐interventions: LLITNs distributed at the start of study; routine malaria control by village health workers.

Outcomes

Parasitaemia prevalence (P falciparum and P vivax)

Measurement: Cross‐sectional surveys of P falciparum and P vivax prevalence by ultrasensitive PCR in adults 18 to 55 years; up to 2 participants sampled from randomly selected households at baseline and the same participants were sampled at follow‐up surveys.

Time points: Pre‐MDA (January 2015) and at < 1 (3‐8 June 2015), 3 (24‐29 August 2015), 8 (15‐26 January 2016), 13 (7‐14 June 2016), 19 (2‐19 December 2016), 25 (13‐18 June 2017), and 31 (1‐13 December 2017) months post‐MDA

Sample size (range): 620 to 1106 (intervention); 412 to 543 (comparison)

Adverse effects (AEs)

A total of 151 (1.4% of all doses, 3.6% of treated individuals) adverse effects reported: 12 (7.9%) not related to drug, 40 (26.5%) unlikely related to drug, 81 (53.6%) possibly related to drug, and 18 (11.9%) probably related to drug. 6 serious AEs: 1 possibly related, 5 unrelated

Notes

Prior to randomization, villages selected based on baseline PCR prevalence survey conducted in 58 villages in study area. Selected villages: (1) had a population between 75 to 1200 people (excluded 15 villages), and (2) were a hotspot village defined as > 30% parasite prevalence (all types) or > 10% P falciparum prevalence by PCR (excluded 27 non‐hotspot villages)

Abbreviations:

ICC = intracluster correlation coefficient, LLITN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, PCR = polymerase chain reaction

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

High risk

Randomization of clusters was performed by flipping a coin.

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

Low risk

Baseline malaria risk and co‐intervention coverage were similar across intervention arms.

Contamination protection

High risk

Intervention and control clusters were pair‐matched by geographical proximity, which likely led to a high risk of contamination due to population movement.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Laboratory staff performing PCR were unaware of the study arm allocation of samples.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

High risk

Outcome data was assessed only in participants aged 18 to 55 years. Up to 2 participants from randomly selected households at baseline were surveyed at each time point and alternative participants (similarly matched by gender, age, and occupation) were surveyed if selected participants were unavailable during follow‐up. Substantial variation in denominators suggesting different participants sampled (non‐randomly) during each survey.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Although no adjustment for clustering was performed by investigators, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 1970 to 1976 (included data from 1970 to 1973 in this review)

Location of study: Nigeria

Malaria endemicity (prevalence): 46% in all ages [High]

Transmission season: April to October

Malaria species: Plasmodium falciparum, P malariae, P ovale

Vector species: Anopheles gambiae, An funestus

Antimalarial drug resistance context: Not described

Study design: Controlled before‐and‐after study (4 arms: no intervention, IRS only, low frequency MDA+IRS, and high frequency MDA+IRS; IRS only, low frequency+IRS, and high frequency+IRS arms included in this review)

Statistical power: Consideration of statistical power was mentioned, but the parameters were not described.

Participants

Age groups included: All ages, but infants not included in MDA until their first malaria episode.

Population targeted

Intervention (low frequency MDA+IRS): 14,129

Intervention (high frequency MDA+IRS): 1810

Comparison (IRS only): 32,828

Interventions

Drug/dose (for all intervention arms receiving MDA):

• Ages ≥10 years: Sulfalene‐pyrimethamine (500 mg/25 mg as 1 tablet) as a single dose

• Ages 5 to 9 years: Sulfalene‐pyrimethamine (250 mg/12.5 mg as ½ tablet) as a single dose

• Ages 1 to 4 years: Sulfalene‐pyrimethamine (230 mg/12.0 mg as 30 drops syrup) as a single dose

• Ages 6 to 11 months: Sulfalene‐pyrimethamine (150 mg/7.5 mg as 20 drops syrup) as a single dose

• Ages < 6 months: Sulfalene‐pyrimethamine (90 mg/4.5 mg as 12 drops syrup) as a single dose

Intervention (Low frequency MDA+IRS group):

Number of rounds (timing/dates): 9 (April 1972 ‐ October 1973)

Interval: Every 10 weeks

Duration implemented: 18 months

Coverage (%): 73% to 92%

Co‐interventions: IRS using propoxur 3 to 4 rounds per year

Intervention (High frequency MDA+IRS group):

Number of rounds (timing/dates): 23 (April 1972 ‐ October 1973)

Interval: Every two weeks during the wet season (May‐October 1972 and May‐October 1973) and every 10 weeks during the dry season (December 1972, March 1973, and October‐November 1973)

Duration implemented: 18 months

Coverage (%): 72% to 91%

Co‐interventions: IRS using propoxur 3 to 4 rounds per year

Comparison (IRS only)

Type: No MDA and no placebo

Co‐interventions: IRS using propoxur 3 to 4 rounds per year

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys in selected village clusters (all ages) (microscopy)

Time points: Pre‐MDA (8 surveys), During MDA (8 surveys)

Sample size (range): 1257 to 2099 (intervention: low frequency MDA+IRS); 1486 to 1679 (intervention: high frequency MDA+IRS); 1104 to 1171 (comparison: IRS only)

Gametocytaemia prevalence

Measurement: Cross‐ sectional surveys in selected villages clusters (all ages) (microscopy)

Time points: Pre‐MDA (8 surveys), During MDA (8 surveys)

Sample size (range): 1257 to 2099 (intervention: low frequency MDA+IRS); 1486 to 1679 (intervention: high frequency MDA+IRS); 1104 to 1171 (comparison: IRS only)

Notes

Abbreviations:

IRS = indoor residual spraying, MDA = mass drug administration

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

High risk

Non‐randomized controlled study

Allocation concealment (selection bias)

High risk

Non‐randomized controlled study

Baseline imbalance (selection bias)

Unclear risk

Similar malaria characteristics between groups at baseline, but unclear if other demographic factors were balanced.

Contamination protection

Low risk

Evaluation villages in both arms were surrounded by similarly treated buffer zones to mitigate possible contamination due to migration.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation, but there was limited access to antimalarials outside of MDA.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Microscopists were blinded to study arm allocation through the use of a numeric identification code. Slides were read independently by two microscopists.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Parasitaemia surveys were performed in all ages in selected study villages.

Selective reporting (reporting bias)

Low risk

All pre‐stated outcomes were reported.

Other bias

Low risk

No other bias detected.

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 2016 to 2017

Location of study: Zanzibar (three districts in Unguja, Central, South, and West districts)

Malaria endemicity (prevalence): 2.5% (range between clusters 0.7‐4.5%) in control group at baseline by quantitative polymerase chain reaction (qPCR) [Very low ‐ estimated Plasmodium falciparum slide prevalence 0.2%]

Transmission season: April to August

Malaria species: P falciparum (predominant), P malaria, P ovale, and P vivax (rare)

Vector species: Anopheles gambiae s.l., An arabiensis, An merus and An funestus

Antimalarial drug resistance context: No evidence of resistance to first line treatment artesunate‐amodiaquine, with 100% efficacy in clinical trial conducted in 2017.

Study design: Cluster‐randomized trial (allocation of shehias, or administrative wards, within the trial arms conducted using computerized block randomization based on shehia population size)

Statistical power: Assuming a coefficient of variation of 0.35 and baseline malaria incidence of 12 per 1000, there was 80% power (alpha = 0.05) to detect a 50% reduction in confirmed malaria illness incidence in the intervention group. Study was not powered for parasitaemia prevalence.

For cluster‐RCTs

Unit of randomization: Shehia (hotspot shehia, defined as a shehia with an with an annual parasite index of greater than 8 confirmed malaria cases per 1000 population)

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using an estimated ICC to calculate effective sample sizes

Adjustment method: Generalized estimating equations

ICC: Not determined

Number of clusters randomized: 16

Number of clusters analysed: 16

Number of people: 23,251

Average cluster size: 1453

Participants

Age groups included: All ages > 6 months. Pregnant women in their first trimester and anyone on concurrent antimalarial treatment at the time of MDA were excluded from MDA with dihydroartemisinin piperaquine. Pregnant women (all trimesters) or women breast feeding infants < 6 months were excluded from receiving low dose primaquine.

Population targeted

Intervention: 10,944

Comparison: 12,307

Interventions

Intervention:

Drug/dose:

• Ages ≥ 14 years (> 40 kg): Dihydroartemisinin plus piperaquine (120 mg/960 mg as 3 tablets) given for three days with a single dose of primaquine (15 mg as 2 tablets)

• Ages 8 to 13 years (21 to 40 kg): Dihydroartemisinin plus piperaquine (80 mg/640 mg as 2 tablets) given for three days with a single dose of primaquine (7.5 mg as 1 tablet)

• Ages 2 to 7 years (10 to 20 kg): Dihydroartemisinin plus piperaquine (40 mg/320 mg as 1 tablet) given for three days with a single dose of primaquine (4 mg as 4 cc solution)

• Ages 6 months to 1 year (5 to 9.9 kg): Dihydroartemisinin plus piperaquine (20 mg/160 mg as ½ tablet) given for three days with a single dose of primaquine (2 mg as 2 cc solution)

Number of rounds (timing/dates): 2 (30 April to 7 May 2016 at the start of high transmission season, and 28 May to 4 June 4 2016 during the peak of high transmission season)

Interval: Every 4 weeks

Duration implemented: 6 weeks

Coverage (%): 91% in round 1 and 88% in round 2 (dihydroartemisinin piperaquine (DHAp)); 86% in round 1 and 80% in round 2 (low dose primaquine).

Co‐interventions: IRS (single round in March 2016 with pirimiphos methyl; 85% of households sprayed at baseline) and ITNs (universal distribution campaign in 2015‐2016; self‐reported ITN use among all ages 75% at baseline).

Comparison:

Type: No MDA and no placebo

Co‐interventions: IRS (single round in March 2016 with pirimiphos methyl; 85% of households sprayed at baseline)

and ITNs (universal distribution campaign in 2015‐2016; self‐reported ITN use among all ages 71% at baseline).

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys of P falciparum (and Plasmodium) prevalence by two‐step pooled 18s‐quantitative PCR (qPCR) with first step screening by cytochrome b (Cytb) qPCR and 18s‐qPCR in participants of all ages from households randomly selected (50% of households randomly selected at each time point).

Time points: During MDA (30 April to 7 May 2016) and 3 months post‐MDA (30 August to 9 September 2016)

Sample size (range): 4402 to 4896 (intervention); 3875 to 4905 (comparison)

Confirmed malaria illness incidence

Measurement: Passive case detection at health facilities via malaria case notification system which captures confirmed malaria infections in real time.

Time points: Pre‐MDA (May ‐ November 2015) and at 2 (May‐August 2016), 5 (May‐November 2016), 10 (May 2016 ‐ April 2017), and 14 (May 2016 ‐ August 2017) months post‐MDA

Adverse effects (AEs)

Among participants receiving MDA, 11.6% and 3.2% reported at least one AE after the first and second round, respectively. An additional 85 and 29 AE reports were passively identified at health facilities after rounds 1 and 2, respectively. The most commonly reported AEs were: nausea and vomiting (33.1%), stomach pain and diarrhoea (18.9%), and dizziness, headache, and fatigue (23.5%). Across all AEs, 44.1% were considered mild, 52.0% as moderate, and 0.5% as severe. There were no MDA‐associated deaths or other serious AEs.

Notes

ClinicalTrials.gov: NCT02721186

The timing of the first parasitaemia survey coincided with the first round of MDA and continued for 10 days following the end of the first MDA distribution; therefore, the first parasitaemia survey was considered as "during MDA".

Abbreviations:

ICC = intracluster correlation coefficient, IRS = indoor residual spraying, ITN = insecticide‐treated bed net, MDA = mass drug administration, PCR = polymerase chain reaction

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Random allocation of shehias to trial arms was conducted using computerized block randomization based on shehia population size.

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

High risk

Baseline malaria prevalence was higher in control (2.5%) compared to intervention (0.8%) shehias. Baseline vector control interventions and demographic characteristics were balanced across arms.

Contamination protection

Unclear risk

Distance between hotspot shehias was not described and there was no mention of buffer zones.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Allocation concealment was not described.

Blinding of participants and personnel (performance bias)
Confirmed malaria illness incidence

Unclear risk

Participants were not blinded to allocation, which may have impacted care‐seeking.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Laboratory staff performing PCR were unaware of the study arm allocation of samples.

Blinding of outcome assessment (detection bias)
Confirmed malaria illness incidence

Unclear risk

Unclear if the health facility staff performing malaria testing were aware of which study arm participants were assigned to.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Outcomes were assessed in participants of all ages residing in randomly sampled households. 50% of households randomly selected at each survey

Incomplete outcome data (attrition bias)
Confirmed malaria illness incidence

Low risk

Malaria cases reported in real time at health facilities through the malaria case notification system (MCN).

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes were reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Clustering accounted for using generalized estimating equations; however, in this review, we performed cluster adjustment of the raw data using an estimated ICC to calculate effective sample sizes.

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 2016 to 2017

Location of study: Savannakhet Province, Laos

Malaria endemicity (prevalence): Plasmodium falciparum prevalence 4.8% in MDA villages and 17.5% in control villages at baseline by ultrasensitive polymerase chain reaction (uPCR); P vivax prevalence 2.3% in MDA villages and 14.7% in control villages at baseline by uPCR [Low ‐ estimated P falciparum slide prevalence 5.3%]

Transmission season: May to October

Malaria species: P falciparum and P vivax

Vector species: Not described

Antimalarial drug resistance context: Not described

Statistical power: For the multi‐country trial (Landier 2017 MMRa; Tripura 2018 KHM; Pongvongsa 2018 LAO; von Seidlein 2019 VNM), 80% power (alpha = 0.05) to detect a 95% reduction in parasite prevalence from a baseline prevalence of 10%

For cluster‐RCTs

Unit of randomization: Village

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using ICC values estimated from the study data to calculate effective sample sizes

Adjustment method: Generalized estimating equations

ICC: 0.1416 (P falciparum outcomes at baseline before MDA), 0.0944 (P falciparum outcomes at post‐MDA 1‐3 months), 0.05406 (P falciparum outcomes at post‐MDA 4‐6 months), 0.04319 (P falciparum outcomes at post‐MDA 7‐12 months); 0.1438 (P vivax outcomes at baseline before MDA), 0.0913 (P vivax outcomes at post‐MDA 1‐3 months), 0.02963 (P vivax outcomes at post‐MDA 4‐6 months), 0.01452 (P vivax outcomes at post‐MDA 7‐12 months)

Number of clusters randomized: 4

Number of clusters analysed: 4

Number of people: 1889

Average cluster size: 472

Features: Two village pairs (4 villages) were established by geographical proximity, population size and parasite prevalence. Within each pair, one village was randomly selected to receive early MDA, while the other village received deferred MDA.

Participants

Age groups included: All ages ≥ 6 months. All pregnant women were excluded from MDA.

Population targeted

Intervention: 1006

Comparison: 883

Interventions

Intervention:

Drug/dose: Dihydroartemisinin (7 mg/kg) plus piperaquine (55 mg/kg) administered once a day for 3 days with a single dose of primaquine (0.25 mg/kg).

Number of rounds (timing/dates): 3 (April, June, and July 2016)

Interval: Every 1 month

Duration implemented: 3 months

Coverage (%): 81% in round 1, 80% in round 2, and 82% in round 3

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Comparison:

Type: Deferred MDA administered in control villages in April, June, and July 2017

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Outcomes

Parasitaemia prevalence (P falciparum and P vivax)

Measurement: Cross‐sectional surveys in all ages every 3 months by uPCR; all individuals present at the time of the survey in the study villages were sampled.

Time points: Pre‐MDA (April 2016, just prior to first MDA round) and at 1 (August 2016), 4 (late October 2016), 7 (January 2017), and 10* (April 2017) months post‐MDA

Sample size (range): 745 to 859 (intervention) and 618 to 802 (control)

Adverse effects (AEs)

AEs were assessed by home visits from village volunteers and clinicians following a report of an AE. Following MDA rounds, 282 individuals reported 295 AEs: 291 (99%) were mild, 3 (1%) were moderate, and 1 (< 1%) was severe (case of pneumonia requiring hospitalization). The most common AEs were common cold (17%), gastritis (8%), diarrhoea (8%), vomiting (7%), dizziness (6%), pruritus (6%), watery stool (4%), nausea (3%), and headache (3%).

Notes

One of four sites from a multi‐country trial in Southeast Asia (ClinicalTrials.gov: NCT01872702)

Abbreviations:

ICC = intracluster correlation coefficient, LLITN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, uPCR = ultrasensitive polymerase chain reaction

* Data from this survey was analysed as post‐MDA 7 to 12 month time period

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"The randomisation was based on computer‐generated random numbers provided by the trial statistician"

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

High risk

Baseline prevalence of both P falciparum and P vivax substantially higher in control compared to intervention villages. Small number of clusters (4 villages) randomized.

Contamination protection

High risk

Intervention and control clusters were pair‐matched by geographical proximity, which likely led to a high risk of contamination due to population movement.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Laboratory staff performing PCR were unaware of the study arm allocation of samples.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Parasitaemia surveys were performed in all individuals aged six months or older residing in the study villages.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Although no adjustment for clustering was performed by investigators, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 1953 to 1954

Location of study: Kenya

Malaria endemicity (prevalence): 37.6% (control) or 23% (intervention) at baseline in all ages by microscopy [Moderate]

Transmission season: May to July

Malaria species: P falciparum, P malariae

Vector species: Anopheles gambiae, An funestus

Antimalarial drug resistance context: Not described

Study design: Controlled before‐and‐after study

Statistical power: Not described

Participants

Age groups included: All ages

Population targeted

Intervention (mean): 101,000

Comparison (mean): population not specified, but control area spans an entire district (Tiriki)

Interventions

Intervention:

Drug/dose:

• Ages > 12 years: Pyrimethamine (50 mg as 2 tablets) as a single dose

• Ages 1 to 12 years: Pyrimethamine (25 mg as 1 tablet) as a single dose

• Ages < 1 years: Pyrimethamine (12.5 mg as ½ tablet) as a single dose

Number of rounds (timing/dates): 2 (May 1953 and May 1954, just prior to the start of the rainy season)

Interval: Every 1 year

Duration implemented: 13 months

Coverage (%): 95% in round 1 and 93% in round 2

Co‐interventions: None

Comparison:

Type: No MDA and no placebo

Co‐interventions: None

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys conducted in a sub‐sample of study population (see notes); 14 surveys in total (microscopy)

Time points: Pre‐MDA, During MDA, and at 1, 2, 3, 4, 6, and 7 months post‐MDA

Sample size (range): 300 to 2100 (intervention); 300 to 2100 (comparison)

Notes

Roberts 1964 KEN (1956 article) states: “Three hundred blood films were taken at each of the two places from people in the following age groups: 0‐10 years (100 films), 11‐20 years (100 films), 21 years and over (100 films)”. Therefore, we assumed that the total participants examined at each survey was 300 in intervention and 300 in comparison groups in order to calculate number of events (malaria cases) for parasitaemia prevalence. We also assumed that samples at each survey time point were independent and aggregated data within follow‐up time point categories.

Abbreviations:

MDA = mass drug administration

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

High risk

Non‐randomized controlled study

Allocation concealment (selection bias)

High risk

Non‐randomized controlled study

Baseline imbalance (selection bias)

High risk

Baseline parasitaemia in the control group was much higher than the intervention group. No other baseline characteristics are described.

Contamination protection

High risk

The control area was located 10 miles from the centre of the intervention area with several major roads connecting both areas. No buffer zone.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation and unclear if this affected outcomes since the sampling methodology at each parasitaemia survey was not described.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Unclear risk

Examination of slides not described. Unclear if multiple reads were taken or if microscopists were blinded.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Data from three hundred slides at each survey sampled in 3 age groups are presented as described.

Selective reporting (reporting bias)

Low risk

All pre‐stated outcomes were reported.

Other bias

Low risk

No other bias detected.

Study characteristics

Methods

Dates of study: 2008

Location of study: Tanzania

Malaria endemicity (prevalence): 0% in all ages at baseline by microscopy [Very low]

Transmission season: March to May, October to November

Malaria species: Plasmodium falciparum

Vector species: Not described

Antimalarial drug resistance context: Not described

Study design: Cluster‐randomized trial

Statistical power: 80% power (alpha = 0.05) to detect a 10‐fold lower malaria incidence in the intervention arm vs control (assuming 0.5 episodes per child) accounting for repeated measures and clustering.

For cluster‐RCTs

Unit of randomization: Geographical clusters of households

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using an estimated ICC to calculate effective sample sizes

Adjustment method: Generalized estimating equations

ICC: Not described

Number of clusters randomized: 16

Number of clusters analysed: 16

Number of people: 3457

Average cluster size: 216

Participants

Age groups included: Ages > 1 year, but individuals who had received a full dose of artemisinin‐based combination therapy in the two weeks before the intervention were excluded. Pregnant women and individuals who were anaemic did not receive primaquine.

Population targeted

Intervention: 1110

Comparison: 2347

Interventions

Intervention:

Drug/dose:

• Ages ≥ 1 year: Sulfadoxine‐pyrimethamine (25 mg + 1.25 mg/kg as a single dose on the first day) plus artesunate (4 mg/kg/day for three days) with primaquine (0.75 mg/kg as a single dose on the third day)

• Anaemic individuals: No primaquine. Sulfadoxine‐pyrimethamine plus artesunate as described above.

• Pregnant women: No primaquine. Sulfadoxine‐pyrimethamine as described above plus amodiaquine (10 mg/kg once daily for three days) instead of artesunate.

Number of rounds (timing/dates): 1 (February‐March 2008)

Interval: Not applicable

Duration implemented: 16 days

Coverage (%): 94.6% received at least one dose and 93% received a complete dose of an efficacious anti‐malarial drug prior to the transmission season or immediately upon arrival to the area

Co‐interventions: Reported ITN use 36.1% (2007) and a single treatment campaign for trachoma with azithromycin was undertaken by a non‐governmental organization.

Comparison:

Type: Placebo administered to all persons in eight clusters once daily over three days.

Co‐interventions: Reported ITN use 36.1% (2007) and a single treatment campaign for trachoma with azithromycin was undertaken by a non‐governmental organization.

If Placebo:

Number of rounds (timing/dates): 1 (February‐March 2008)
Interval: Not applicable
Duration implemented: 16 days
Coverage (%): Not described

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys of P falciparum prevalence by microscopy and QT‐NASBA in 50 randomly‐selected individuals per cluster

Time points: Pre‐MDA (January‐February 2008) and at < 1 (April 2008), 2 (May 2008), 3 (June 2008), and 4 (July 2008) months post‐MDA

Sample size (range): 261 to 399 (intervention); 212 to 395 (comparison)

Confirmed malaria illness incidence

Measurement: Active case surveillance of 150 randomly‐selected children (ages 1 to 10 years) from each arm, visited every 2 weeks to monitor symptoms and test by RDT if febrile. Passive surveillance in entire population.

Time points: Followed every 2 weeks for 6 months (February‐July 2008)

Gametocytaemia prevalence

Measurement: Cross‐sectional surveys of gametocytaemia prevalence by microscopy and QT‐NASBA in 50 randomly selected individuals per cluster

Time points: Pre‐MDA (January‐February 2008) and at < 1 (April 2008), 2 (May 2008), 3 (June 2008), and 4 (July 2008) months post‐MDA

Sample size (range): 261‐399 (intervention); 212‐395 (comparison)

Adverse effects

One individual was diagnosed with possibly‐drug related severe skin reaction in the week following MDA.

A second individual presented with non‐drug related skin hyperpigmentation on the face. Both individuals were treated with steroids and monitored until symptoms disappeared. In those given primaquine, moderate anaemia (Hb level of < 8 g/dL) was observed in 40% (6/15 individuals) of the G6PD A‐, 11.1% (3/27 individuals) of the G6PD A, and 4.5% (18/399 individuals) of the G6PD B individuals; one case of severe anaemia (haemoglobin level of < 5 g/dL) was observed.

Notes

ClinicalTrials.gov: NCT00509015

Due to the following reasons, we excluded this study from quantitative synthesis in this review: outcome evaluation by 18S QT‐NASBA ended prematurely during the follow‐up period; we were unable to classify events in microscopy outcomes by post‐MDA time point (reported in aggregate in the publication); and the baseline before MDA number of events for multiple outcomes was zero.

Abbreviations:

ICC = intracluster correlation coefficient, ITN = insecticide‐treated bed net, MDA = mass drug administration, QT‐NASBA = real‐time quantitative nucleic acid sequence based amplification, RDT = rapid diagnostic test

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"Clusters were randomized to the intervention or control arm using computer generated randomization tables using excel."

Allocation concealment (selection bias)

Low risk

Placebo‐controlled trial, therefore allocation was concealed.

Baseline imbalance (selection bias)

Low risk

Baseline demographic and malaria characteristics were similar across arms.

Contamination protection

Low risk

"Households that were located between clusters (i.e. within 1 km distance from the boundary of intervention and/or control clusters) were considered as buffer zones. Members of these households received the intervention in order to minimize contamination."

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Low risk

Placebo‐controlled trial and, at each cross‐sectional survey, individuals were randomly selected from computer‐generated random tables. However, placebo tablets appear different from intervention drug.

Blinding of participants and personnel (performance bias)
Confirmed malaria illness incidence

Low risk

Placebo‐controlled trial.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Placebo‐controlled trial; slides were read independently by two microscopists.

Blinding of outcome assessment (detection bias)
Confirmed malaria illness incidence

Low risk

Placebo‐controlled trial.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Random sample of individuals surveyed at baseline and follow‐up.

Incomplete outcome data (attrition bias)
Confirmed malaria illness incidence

Low risk

Active (visit by trained fieldworker every 2 weeks) and passive case detection.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcome measures were reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Clustering accounted for using generalized estimating equations; however, in this review, we performed cluster adjustment of the raw data using an estimated ICC to calculate effective sample sizes.

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 2014‐2016

Location of study: Battambang province, Cambodia

Malaria endemicity (prevalence): Plasmodium falciparum prevalence 0.9% in MDA villages and 2.4% in control villages at baseline by ultrasensitive polymerase chain reaction (uPCR); P vivax prevalence 10.7% in MDA villages and 8.8% in control villages at baseline by uPCR [Very low ‐ estimated P falciparum slide prevalence 0.5%]

Transmission season: May to October

Malaria species: P falciparum and P vivax

Vector species: Not described

Antimalarial drug resistance context: Reduced susceptibility to artemisinins and ACT partner drug resistance

Study design: Pair‐matched cluster‐randomized trial

Statistical power: For the multi‐country trial (Landier 2017 MMRa; Tripura 2018 KHM; Pongvongsa 2018 LAO; von Seidlein 2019 VNM), 80% power (alpha = 0.05) to detect a 95% reduction in parasite prevalence from a baseline prevalence of 10%

For cluster‐RCTs

Unit of randomization: Village

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using ICC values estimated from the study data to calculate effective sample sizes

Adjustment method: Generalized estimating equations

ICC: 0.004175 (P falciparum outcomes at baseline before MDA), 0 (P falciparum outcomes at post‐MDA 1‐3, 4‐6, and 7‐12 months); 0.02167 (P vivax outcomes at baseline before MDA), 0.01637 (P vivax outcomes at post‐MDA 1‐3 months), 0.02229 (P vivax outcomes at post‐MDA 4‐6 months), 0.002531 (P vivax outcomes at post‐MDA 7‐12 months)

Number of clusters randomized: 4

Number of clusters analysed: 4

Number of people: 2,770

Average cluster size: 693

Features: Two village pairs (4 villages) were established by geographical proximity, population size, and parasite prevalence. Within each pair, one village was randomly selected to receive early MDA, while the other village received deferred MDA

Participants

Age groups included: All ages ≥ 6 months. All pregnant women were excluded from MDA.

Population targeted

Intervention: 858

Comparison: 1912

Interventions

Intervention:

Drug/dose: Dihydroartemisinin (7 mg/kg) plus piperaquine tetraphosphate (55 mg/kg) administered once a day for 3 days. Visitors or returning residents were offered a single course of piperaquine tetraphosphate.

Number of rounds (timing/dates): 3 (July, August, September 2015)

Interval: Every 1 month

Duration implemented: 3 months

Coverage (%): 74% in round 1, 60% in round 2, and 71% in round 3

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Comparison:

Type: Deferred MDA administered in control villages in July, August, and September 2016

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Outcomes

Parasitaemia prevalence (P falciparum and P vivax)

Measurement: Cross‐sectional surveys in all ages every 3 months by uPCR; all individuals present at the time of the survey in the study villages were sampled.

Time points: Pre‐MDA (July 2015) and at 1 (Oct 2015), 4 (Jan 2016), 7 (April 2016), and 10* (July 2016) months post‐MDA

Sample size (range): 470 to 543 (intervention) and 583 to 1090 (control)

Confirmed malaria illness incidence (P falciparum and P vivax)

Measurement: Passive case detection at malaria posts for measured or self‐reported fever (≥ 37.5 °C) and confirmed Pfalciparum or P vivax infection by RDT or microscopy.

Time points: Pre‐MDA (July 2014 ‐ June 2015) and at 9 (July 2015 to June 2016) months' post‐MDA

Adverse effects (AEs)

AEs were assessed through active surveillance on days 1, 2, 3, and 7 following MDA rounds. AEs were reported by 46% (n=909) participants and a majority (96%) were mild; 4% (n=40) required medical attention and there were 3 non‐study‐related deaths. No serious AEs were reported. The most common AEs reported were dizziness (22%), headache (18%), fever (10%), and nausea (8%).

Notes

One of four sites from a multi‐country trial in Southeast Asia (ClinicalTrials.gov: NCT01872702)

Abbreviations:

ACT = artemisinin‐based combination therapy, ICC = intracluster correlation coefficient, LLITN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, uPCR = ultrasensitive polymerase chain reaction

* Data from this survey was analysed as post‐MDA 7‐12 month time point

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"The randomisation was based on computer‐generated random numbers provided by the trial statistician"

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

High risk

Baseline prevalence of P falciparum substantially higher in control compared to intervention villages baseline. Small number of clusters (4 villages) randomized.

Contamination protection

High risk

Intervention and control clusters were pair‐matched by geographical proximity, which likely led to a high risk of contamination due to population movement.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation.

Blinding of participants and personnel (performance bias)
Confirmed malaria illness incidence

Unclear risk

Participants were not blinded to allocation, which may have impacted care‐seeking.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Laboratory staff performing PCR were unaware of the study arm allocation of samples.

Blinding of outcome assessment (detection bias)
Confirmed malaria illness incidence

Unclear risk

Unclear if the health facility staff performing malaria testing were aware of which study arm participants were assigned to.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Parasitaemia surveys were performed in all individuals aged six months or older residing in the study villages.

Incomplete outcome data (attrition bias)
Confirmed malaria illness incidence

Low risk

Data collected at malaria health posts with dedicated study staff.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Although no adjustment for clustering was performed by investigators, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 1999

Location of study: The Gambia

Malaria endemicity: 42.9% in children ≤ 5 years in control at baseline [High]

Transmission season: June to December

Malaria species: Plasmodium falciparum

Vector species: Not described

Antimalarial drug resistance context: Not described

Study design: Cluster‐randomized trial

Statistical power: 90% power (alpha = 0.05) to detect a 40% reduction in malaria incidence among children < 11 years assuming coefficient of variation between pair‐matched villages of 0.25, 20% loss to follow‐up, and mean incidence of malaria of 1 attack per child per week during the 20 week transmission season.

For cluster‐RCTs

Unit of randomization: Villages

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using an estimated ICC to calculate effective sample sizes

Adjustment method: Poisson regression model adjusting for population size

ICC: Not described

Number of clusters randomized: 18

Number of clusters analysed: 18

Number of people: 3655

Average cluster size: 203

Feature: Matched villages by population size, spleen rate in children < 5 years, and distance from the river

Participants

Age groups included: All persons ≥ 6 months old; pregnant women excluded.

Population targeted

Intervention: 12,331

Control: 1686

Interventions

Intervention:

Drug/dose:

• Adults: Sulfadoxine‐pyrimethamine (1500 mg/75 mg as 3 tablets) plus artesunate (200 mg as 4 tablets) given in a single day

• Children (< 10 kg): Sulfadoxine‐pyrimethamine (250 mg/37.5 mg as ½ tablet) plus artesunate (4 mg/kg) given in a single day. Additional quarter tablet of sulfadoxine‐pyrimethamine given for every 5 kg increment in weight

Number of rounds (timing/dates): 1 (June 1999)

Interval: Not applicable

Duration implemented: 1 month

Coverage (%): 89% in total population, 90.8% in evaluated group

Co‐interventions: None

Comparison:

Type: Placebo

Co‐interventions: None

If placebo:

Number of rounds (timing/dates): 1 (June 1999)
Interval: Not applicable
Duration implemented: 1 month
Coverage (%): 89% in total population, 90.8% in evaluated group

Outcomes

Parasitaemia prevalence

Measurement: Cross‐sectional surveys of P falciparum prevalence by microscopy in children and weekly surveillance in all ages

Time points: Pre‐MDA (children ≤ 5 years, May 1999) and at 6 months post‐MDA (children < 11 years, November 1999)

Sample size (range): 808 to 985 (intervention); 605 or 606 (comparison)

Parasitaemia incidence

Measurement: Weekly surveillance in children aged < 11 years for cases with temperature ≥ 37.5 °C and P falciparum parasitaemia > 5000 parasites per µL by microscopy

Time points: at 5 months post‐MDA (20 July 1999 to 2 December 1999)

Sample size: 769 (intervention); 607 (comparison)

Gametocytaemia prevalence

Measurement: Cross‐sectional surveys of gametocytaemia prevalence by microscopy in children

Time points: Pre‐MDA (children ≤ 5 years) and at 6 months post‐MDA (children < 11 years)

Sample size (range): 808 to 985 (intervention); 605 or 606 (comparison)

Malaria‐specific mortality

Measurement: Verbal autopsy confirmed by 3 physicians in children < 11 years

Adverse effects (AEs)

Monitored through passive and active surveillance. Active surveillance consisted of asking 90 randomly selected individuals across all 42 study area villages about AEs one month following MDA.

AEs reported (passive surveillance system): 1 episode of pruritus in intervention group.

AEs reported (active surveillance system): 25 of 75 individuals in intervention group remembered one or more complaints within 2 days of taking the drug including dizziness (13), fever (6), diarrhoea (5), vomiting (5) and itching (4). In the comparison group, 2 of 15 individuals (13%) who had received placebo remembered complaints

Notes

Abbreviations:

ICC = intracluster correlation coefficient, MDA = mass drug administration

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

Method of randomization not described, but previous author correspondence revealed that randomization was computer generated.

Allocation concealment (selection bias)

Low risk

One study nurse administered all study drugs to the 18 villages and left the study area following administration. Study personnel and participants were blinded to the intervention status. Placebo‐controlled trial, therefore allocation was concealed.

Baseline imbalance (selection bias)

Low risk

Intervention and control villages were similar across baseline characteristics reported.

Contamination protection

Low risk

All individuals in neighbouring non‐randomized villages in the study area received MDA in order to minimize the risk of contamination.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Low risk

Cluster‐randomized, double‐blind, placebo‐controlled trial in which neither study personnel nor participants were aware of the intervention status of villages. Placebo tablets identical to intervention drug.

Blinding of participants and personnel (performance bias)
Parasitaemia incidence

Low risk

Cluster‐randomized, double‐blind, placebo‐controlled trial in which neither study personnel nor participants were aware of the intervention status of villages. Placebo tablets identical to intervention drug.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Outcome assessment by microscopy was blinded to intervention status.

Blinding of outcome assessment (detection bias)
Parasitaemia incidence

Low risk

Outcome assessment by microscopy was blinded to intervention status. Cases diagnosed at health centre by RDT, but neither study personnel nor participants were aware of the intervention status of villages.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

All children < 11 years in the surveillance villages were surveyed.

Incomplete outcome data (attrition bias)
Parasitaemia incidence

High risk

78‐92% of children were examined weekly by field workers and 87% of planned visits took place. However, malaria cases were defined as a temperature ≥ 37.5 °C and parasitaemia > 5000/µL, which may have precluded asymptomatic or lower density infections and resulted in an underestimate of parasitaemia incidence.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes were reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

One cluster pair (Daro Rahman and Misira villages) was not analysed due to political and logistical problems, but sensitivity analysis indicated similar results. "Analyses which omitted these 2 villages yielded similar results as the analyses including the 2 villages."

Incorrect analysis (cluster RCT)

Low risk

Analysis adjusted for clustering using Poisson regression model adjusting for population size; however, in this review, we performed cluster adjustment of the raw data using an estimated ICC to calculate effective sample sizes.

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.

Study characteristics

Methods

Dates of study: 2013‐2015

Location of study: Binh Phuoc and Ninh Thuan provinces, Vietnam

Malaria endemicity (prevalence): Plasmodium falciparum prevalence 3.9% in MDA villages and 4.1% in control villages at baseline by ultrasensitive polymerase chain reaction (uPCR); P vivax prevalence 6.3% in MDA villages and 7.3% in control villages at baseline by uPCR [Very low ‐ estimated P falciparum slide prevalence 0.9%]

Transmission season: May to November

Malaria species: P falciparum and P vivax

Vector species: Not described

Antimalarial drug resistance context: At the start of the study, no evidence of resistance to piperaquine and cure rates following dihydroartemisinin piperaquine (DHAp) were satisfactory. In 2016, multidrug resistance was first detected and treatment failures with DHAp have increased since then.

Statistical power: For the multi‐country trial (Landier 2017 MMRa; Tripura 2018 KHM; Pongvongsa 2018 LAO; von Seidlein 2019 VNM), 80% power (alpha = 0.05) to detect a 95% reduction in parasite prevalence from a baseline prevalence of 10%

For cluster‐RCTs

Unit of randomization: Village

Adjusted analyses for clustering: Yes; however, in this review, we performed cluster adjustment of the raw data using ICC values estimated from the study data to calculate effective sample sizes

Adjustment method: Generalized estimating equations

ICC: 0.002464 (P falciparum outcomes at baseline before MDA), 0 (P falciparum outcomes at post‐MDA 1‐3 months), 0.003616 (P falciparum outcomes at post‐MDA 4‐6 months), 0.006523 (P falciparum outcomes at post‐MDA 7‐12 months); 0.001502 (P vivax outcomes at baseline before MDA), 0.002539 (P vivax outcomes at post‐MDA 1‐3 months), 0 (P vivax outcomes at post‐MDA 4‐6 and 7‐12 months)

Number of clusters randomized: 4

Number of clusters analysed: 4

Number of people: 2846

Average cluster size: 712

Features: Two village pairs (4 villages) were established by geographical proximity, population size and parasite prevalence. Within each pair, one village was randomly selected to receive early MDA, while the other village received deferred MDA

Participants

Age groups included: All ages ≥ 6 months. All pregnant women in their first trimester were excluded from MDA and pregnant women in any trimester were excluded from primaquine.

Population targeted

Intervention: 1439

Comparison: 1407

Interventions

Intervention:

Drug/dose: Dihydroartemisinin (7 mg/kg) plus piperaquine (55 mg/kg) administered once a day for 3 days with a single dose of primaquine (0.25 mg/kg).

Number of rounds (timing/dates): 3 (November 2013, January and February 2014)

Interval: Every 1 month

Duration implemented: 3 months

Coverage (%): 83% in round 1, 98% in round 2, and 99% in round 3

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Comparison:

Type: Deferred MDA administered in control villages in December 2014, January 2015, and February 2015

Co‐interventions: LLITNs, uninterrupted access to diagnosis and treatment in study villages

Outcomes

Parasitaemia prevalence (P falciparum and P vivax)

Measurement: Cross‐sectional surveys in all ages every 3 months by uPCR; all individuals present at the time of the survey in the study villages were sampled.

Time points: Pre‐MDA (November 2013, just prior to first MDA round) and at 1 (March 2014), 4 (June 2014), 7 (September 2014), and 10* (December 2014) months post‐MDA

Sample size (range): 745 to 859 (intervention) and 618 to 802 (control)

Adverse effects (AEs)

22 AEs were reported which included vomiting, nausea, diarrhoea, labyrinth disorder, leg fracture and urticaria. Seven serious AEs were reported within the first year of the study including death from suicide, sudden death, drowning, decline due to aging, and gastric cancer.

Notes

One of four sites from a multi‐country trial in Southeast Asia (ClinicalTrials.gov: NCT01872702)

Abbreviations:

ICC = intracluster correlation coefficient, LLITN = long‐lasting insecticide‐treated bed net, MDA = mass drug administration, uPCR = ultrasensitive polymerase chain reaction

* Data from this survey was analysed as post‐MDA 7‐12 month time point

Risk of bias

Bias

Authors' judgement

Support for judgement

Random sequence generation (selection bias)

Low risk

"The randomisation was based on computer‐generated random numbers provided by the trial statistician"

Allocation concealment (selection bias)

Low risk

Allocation was conducted by an institution.

Baseline imbalance (selection bias)

Low risk

Small number of clusters (4 villages) randomized, but baseline malaria prevalence is balanced across arms.

Contamination protection

High risk

Intervention and control clusters were pair‐matched by geographical proximity, which likely led to a high risk of contamination due to population movement.

Blinding of participants and personnel (performance bias)
Parasitaemia prevalence

Unclear risk

Participants were not blinded to allocation.

Blinding of outcome assessment (detection bias)
Parasitaemia prevalence

Low risk

Laboratory staff performing PCR were unaware of the study arm allocation of samples.

Incomplete outcome data (attrition bias)
Parasitaemia prevalence

Low risk

Parasitaemia surveys were performed in all individuals aged six months or older residing in the study villages.

Selective reporting (reporting bias)

Low risk

All pre‐specified outcomes reported.

Other bias

Low risk

No other bias detected.

Recruitment bias (cluster RCT)

Low risk

No recruitment following randomization.

Loss of clusters (cluster RCT)

Low risk

No clusters were lost.

Incorrect analysis (cluster RCT)

Low risk

Although no adjustment for clustering was performed by investigators, in this review, we performed cluster adjustment of the raw data using a study‐provided ICC to calculate effective sample sizes

Comparability with individually randomized trials (cluster RCT)

Low risk

MDA, by definition, is applied at the population level, so this criteria is irrelevant for the intervention evaluated in this review.