Research method
Cluster Randomized Trial
5 studies
A cluster-randomised trial assigns groups — clinics, villages, community sites, physician practices — rather than individual patients, then measures outcomes in people nested inside those groups. Randomisation still aims to balance clusters at baseline, but people in the same cluster share practices, norms and contamination risk, so the effective sample size is closer to the number of clusters than to the headcount. Many cluster trials in this set are also open-label: sites know which programme they received.
Investigators cluster-randomise when the intervention is a service, a reminder system, a sanitation campaign or a community-norm programme that cannot be given to one patient without spilling to the next. It answers 'did assigning this site or village change outcomes?' Its main limitation is that a within-cluster behaviour shift (more toilets, fewer missed TB doses) need not move child health or microbiologic cure, and an open-label 9 mm Hg SBP drop is not a blinded hard-event result.
Studies
5
Findings
5
5 supporting · 0 challenging · 0 qualifying citations
Open tensions
2
Currently
What we know
- A 12-week open-label three-arm pilot cluster-randomised 13 US primary-care sites to 4-week digital-medicine offering, 12-week offering, or usual care for uncontrolled hypertension with type 2 diabetes. Digital-medicine users had about 9 mm Hg greater SBP reduction by week 4, more patients at BP goal by week 12, and larger LDL-C declines; HbA1c trended lower. The design cannot speak to long-term cardiovascular events or to blinded placebo effects.
- Across 36 clusters, 4,173 pulmonary TB patients were assigned to control, SMS, medication monitor, or both. Medication monitors reduced poor adherence by 40–50% versus standard care; SMS alone did not improve adherence but cut loss to follow-up by 58%. Better dose-taking in this report is not the same as proven microbiologic cure, and performance outside China's NTP is untested.
- India's Total Sanitation Campaign was cluster-randomised across 80 villages in rural Madhya Pradesh. Intervention villages gained about 19 percentage points in improved sanitation facilities and a smaller (~10%) reduction in open defecation, but child-health endpoints did not improve after about 21 months. Short follow-up and some control-group contamination mean the trial shows this scaled package did not move health markers here, not that sanitation can never help.
- SASA! trained community activists in Kampala intervention clusters to challenge norms linking gender power, IPV and HIV risk. Past-year physical IPV and men's concurrency were about 50% lower in intervention communities, and attitudes supporting women's right to refuse sex improved. Precision was limited for some IPV estimates (wide CIs), and election-period suspension interrupted delivery.
- In Mashhad primary care, 35 physicians and 240 hypertensive patients were studied after intervention physicians received short experiential communication-skills training. Training was linked to better patient health literacy, medication adherence, self-efficacy and blood-pressure outcomes versus usual care to 6 months. Effects beyond 6 months, and which communication components drove BP change, were not isolated.
Largest unresolved question
Process success and health success are not the same cluster-trial result. TSC moved toilets (~19 percentage points) without moving child health; TB medication monitors moved adherence 40–50% without, in this report, proving cure; SASA! moved IPV and concurrency by about 50% as the intended social outcome. A student who treats 'the cluster intervention worked' as one sentence will misread at least one of these papers.
Common misconceptions
A cluster RCT with thousands of patients is as precise as individually randomising the same headcount.
People in a clinic or village are not independent. The TB trial analysed 4,173 patients across 36 clusters; TSC used 80 villages; digital medicines used 13 sites. The independent units are the clusters that were assigned.
If SMS 'did not work' for TB adherence, it was a useless arm.
SMS alone did not improve missed-dose adherence but cut loss to follow-up by 58%. Medication monitors were the adherence winners (40–50% less poor adherence). Endpoint choice changes which arm looks successful.
More toilets, or a 9 mm Hg SBP drop, means child survival or heart attacks must have improved.
TSC's ~19 point facility gain and ~10% open-defecation reduction did not move child-health endpoints in that follow-up. The digital-medicine pilot's SBP/LDL shifts are 12-week open-label findings, not hard cardiovascular events.
Related
Claim ledger
What the evidence shows
Drawn from 5 studies in this library. Mix labels say which citation roles are present; they are not a strength score. Supports means evidence for a finding; Challenges means evidence against a stated position; Qualifies marks scope.
A 12-week open-label three-arm pilot cluster-randomised 13 US primary-care sites to 4-week digital-medicine offering, 12-week offering, or usual care for uncontrolled hypertension with type 2 diabetes. Digital-medicine users had about 9 mm Hg greater SBP reduction by week 4, more patients at BP goal by week 12, and larger LDL-C declines; HbA1c trended lower. The design cannot speak to long-term cardiovascular events or to blinded placebo effects.
Across 36 clusters, 4,173 pulmonary TB patients were assigned to control, SMS, medication monitor, or both. Medication monitors reduced poor adherence by 40–50% versus standard care; SMS alone did not improve adherence but cut loss to follow-up by 58%. Better dose-taking in this report is not the same as proven microbiologic cure, and performance outside China's NTP is untested.
India's Total Sanitation Campaign was cluster-randomised across 80 villages in rural Madhya Pradesh. Intervention villages gained about 19 percentage points in improved sanitation facilities and a smaller (~10%) reduction in open defecation, but child-health endpoints did not improve after about 21 months. Short follow-up and some control-group contamination mean the trial shows this scaled package did not move health markers here, not that sanitation can never help.
SASA! trained community activists in Kampala intervention clusters to challenge norms linking gender power, IPV and HIV risk. Past-year physical IPV and men's concurrency were about 50% lower in intervention communities, and attitudes supporting women's right to refuse sex improved. Precision was limited for some IPV estimates (wide CIs), and election-period suspension interrupted delivery.
In Mashhad primary care, 35 physicians and 240 hypertensive patients were studied after intervention physicians received short experiential communication-skills training. Training was linked to better patient health literacy, medication adherence, self-efficacy and blood-pressure outcomes versus usual care to 6 months. Effects beyond 6 months, and which communication components drove BP change, were not isolated.
Debates
Tensions and limits
Some items are genuine disagreements on the same question. Others mark different assays, populations, or outcomes.
Process success and health success are not the same cluster-trial result. TSC moved toilets (~19 percentage points) without moving child health; TB medication monitors moved adherence 40–50% without, in this report, proving cure; SASA! moved IPV and concurrency by about 50% as the intended social outcome. A student who treats 'the cluster intervention worked' as one sentence will misread at least one of these papers.
Process success and health success are not the same cluster-trial result. TSC moved toilets (~19 percentage points) without moving child health; TB medication monitors moved adherence 40–50% without, in this report, proving cure; SASA! moved IPV and concurrency by about 50% as the intended social outcome. A student who treats 'the cluster intervention worked' as one sentence will misread at least one of these papers.
- Did India’s toilet campaign improve child health?
- TB pillbox reminders beat SMS for doses
- SASA! cut partner violence in Kampala
Study Role Design N Population Outcome Did India’s toilet campaign improve child health? Supports RCTCluster RCT of 80 villages: Total Sanitation Campaign support vs control N=3390 · Baseline 3,390 children <5 from 1,954 households in 80 villages; ~21-month follow-up Children under five in rural Madhya Pradesh villages Sanitation facility use, open defecation, and child health endpoints TB pillbox reminders beat SMS for doses Supports RCTCluster RCT across 36 sites: control, SMS, medication monitor, or combined N=4173 · Pulmonary TB patients in China’s NTP Pulmonary tuberculosis patients in China’s National Tuberculosis Control Program Poor medication adherence and loss to follow-up SASA! cut partner violence in Kampala Supports RCTCluster RCT of SASA! community activism vs control in Kampala N=2532 · Follow-up interviews: 600+768 intervention and 530+634 control (women+men) Adults in Kampala intervention and control communities in the SASA! CRT Past-year physical IPV and men’s sexual concurrency (community-level contrasts)
PaperFren reads this as a limit on how far one study travels — different assays, populations, or outcomes — not a forced fight between papers.
Open-label clinical clusters and community structural trials ask different causal questions under the same design name. Digital medicines at 13 sites cut SBP by about 9 mm Hg without placebo control; Mashhad trained 35 physicians and followed 240 patients for literacy and BP; SASA! randomised Kampala communities to a norm-change programme. Blinding a village activist campaign is not the same problem as blinding a digital pill, and neither is an individually randomised ITT drug trial.
Open-label clinical clusters and community structural trials ask different causal questions under the same design name. Digital medicines at 13 sites cut SBP by about 9 mm Hg without placebo control; Mashhad trained 35 physicians and followed 240 patients for literacy and BP; SASA! randomised Kampala communities to a norm-change programme. Blinding a village activist campaign is not the same problem as blinding a digital pill, and neither is an individually randomised ITT drug trial.
- Digital pills cut blood pressure faster
- Doctor communication training helps hypertension care
- SASA! cut partner violence in Kampala
Study Role Design N Population Outcome Digital pills cut blood pressure faster Supports RCTOpen-label cluster-randomized three-arm pilot: 4-week DMO, 12-week DMO, usual care N=109 · Modified ITT (40 / 40 / 29) across 13 US primary-care sites Adults with uncontrolled hypertension and type 2 diabetes in US primary care SBP reduction, BP goal attainment, and LDL-C/HbA1c change to 12 weeks Doctor communication training helps hypertension care Supports RCTTwo-arm RCT; physicians received short experiential communication-skills training N=240 · 240 hypertensive patients nested under 35 physicians Hypertensive patients and their primary-care physicians in Mashhad, Iran Patient health literacy, adherence, self-efficacy, and blood pressure to 6 months SASA! cut partner violence in Kampala Supports RCTCluster RCT of SASA! community activism vs control in Kampala N=2532 · Follow-up interviews: 600+768 intervention and 530+634 control (women+men) Adults in Kampala intervention and control communities in the SASA! CRT Past-year physical IPV and men’s sexual concurrency (community-level contrasts)
PaperFren reads this as a limit on how far one study travels — different assays, populations, or outcomes — not a forced fight between papers.
Timeline
How understanding moved
Study years are when the paper was published. Evidence edits are dated changes to this page's claims. Explanations are when PaperFren added a Discovery — not a claim that the science happened that day.
2026
Concept page published
Cluster Randomized Trial
Change log
What changed
Dated edits to this page's evidence: studies added or removed from a claim, claims added or withdrawn, and new explanations tagged here. Coverage edits are distinct from reviewed research changes and corrections. Dates record PaperFren updates; study publication dates appear in the timeline.
- Coverage updated · Concept page published
Papers
5 studies in this library bear on Cluster Randomized Trial, ordered by citations.
- SASA! cut partner violence in Kampala
A Kampala cluster RCT found community mobilization (SASA!) associated with roughly 50% lower past-year physical IPV and male partner concurrency.
- Did India’s toilet campaign improve child health?
A village-randomized sanitation program raised latrine access and cut open defecation modestly but did not improve child diarrhea, parasites, anemia, or growth.
- TB pillbox reminders beat SMS for doses
In a China cluster RCT, medication-monitor reminders cut poor TB adherence 40–50%, while text messages mainly reduced loss to follow-up.
- Digital pills cut blood pressure faster
Patients with uncontrolled hypertension and type 2 diabetes using digital medicines had larger 4-week SBP drops than usual care.
- Doctor communication training helps hypertension care
Training physicians in communication skills improved hypertensive patients’ health literacy, blood pressure control, adherence, and self-efficacy in an Iranian RCT.
Compare studies
Select 2–10 studies. Design and N are labels, not a ranking.
Nothing selected yet.
Questions
What is still open
Process success and health success are not the same cluster-trial result. TSC moved toilets (~19 percentage points) without moving child health; TB medication monitors moved adherence 40–50% without, in this report, proving cure; SASA! moved IPV and concurrency by about 50% as the intended social outcome. A student who treats 'the cluster intervention worked' as one sentence will misread at least one of these papers.
Open-label clinical clusters and community structural trials ask different causal questions under the same design name. Digital medicines at 13 sites cut SBP by about 9 mm Hg without placebo control; Mashhad trained 35 physicians and followed 240 patients for literacy and BP; SASA! randomised Kampala communities to a norm-change programme. Blinding a village activist campaign is not the same problem as blinding a digital pill, and neither is an individually randomised ITT drug trial.
Ask PaperFren about Cluster Randomized Trial
Study this conceptflashcards and short-answer questions
Why can TSC be a 'positive' sanitation trial and a 'null' child-health trial at the same time?
Intervention villages among the 80 gained about 19 percentage points in improved facilities and a smaller (~10%) drop in open defecation, which are process behaviours. Child-health endpoints did not improve at ~21 months. Cluster assignment tested this TSC package's health effect, not the abstract value of sanitation.
In the TB reminder trial, which arm should a programme manager pick if the goal is fewer missed doses, and which if the goal is retaining people in care?
Medication monitors reduced poor adherence by 40–50% versus standard care. SMS alone did not improve adherence but reduced loss to follow-up by 58%. Combined reminders were a third assigned option; the analysed population was 4,173 patients in 36 clusters.
SASA! reported about 50% lower past-year physical IPV in intervention communities. What cluster-trial caveats belong next to that sentence?
Communities, not individuals, were assigned; some IPV estimates had wide CIs; delivery paused during elections. It is evidence about a structural activist programme in Kampala clusters, not a precision estimate for every setting, and not an individually randomised counselling trial.
A colleague treats the digital-medicine 9 mm Hg SBP result as if it were a double-blind drug RCT. What features of the design block that reading?
Thirteen primary-care sites were cluster-assigned, the pilot was open-label, follow-up was 12 weeks, and endpoints were SBP, goal attainment, LDL-C and trending HbA1c — not hard cardiovascular events. Site-level usual care is the comparator, not a matched placebo pill.