Situation: A public health nurse at the Rural Health Unit (RHU) is preparing a study of handwashing and diarrhea among children under 5 years in a municipality of 18 barangays. She has a complete list of households in each barangay. She reads a report in which the provincial health office itself chose the 10 barangays with the most diarrhea cases to receive a new school handwashing program, while 10 other barangays continued their usual activities. Diarrhea incidence among children in the two groups of barangays was compared over the next year. How is this design classified?

The investigators decided which barangays received the program, so the study is experimental (an intervention study), not observational. Because whole communities were assigned but not by random allocation, it is a nonrandomized (quasi-experimental) community trial. Choosing the worst-affected barangays means the groups may differ at the start, so its evidence is weaker than that of a randomized trial.

Study design classification

The scenario describes an intervention study in which the provincial health office assigned whole barangays to receive a school handwashing program or to continue usual activities, and then compared diarrhea incidence between the two groups over the following year. The key feature is that the assignment was made by the investigators, not by chance, and the unit of assignment was the community rather than the individual child.

Because the investigators decided which barangays received the program, this is an experimental design, not an observational one. The term “experimental” in epidemiology means that the researcher controls the exposure. In contrast, a cohort study would simply observe handwashing practices that already exist without assigning them.

The correct classification is a nonrandomized community trial, also called a quasi-experimental community trial. The word “community” reflects that entire barangays were allocated as clusters, while “nonrandomized” reflects that the provincial health office selected the intervention barangays based on having the most diarrhea cases rather than using a random process.

Watch out! A cluster randomized trial would require random allocation of barangays to intervention or control. Here, the selection was deliberate and based on disease burden, so randomization is absent.

Key point! Selecting the 10 barangays with the most diarrhea cases introduces a major threat to validity. These barangays may differ from the comparison barangays in sanitation, water access, health-seeking behavior, or socioeconomic status at baseline. Because the groups may not be comparable at the start, the evidence from this design is weaker than that from a randomized trial.

Design feature	Nonrandomized community trial	Cluster randomized trial	Prospective cohort study	Ecological correlation study
Investigator assigns exposure	Yes	Yes	No	No
Unit of assignment	Community	Community	Individual	Group or population
Allocation method	Deliberate or convenience	Random	Not applicable	Not applicable
Main weakness	Baseline differences between groups	Fewer weaknesses; stronger evidence	Confounding by indication	Ecological fallacy

The term quasi-experimental is used in the literature to describe exactly this kind of design. In a multi-country study of community-oriented handwashing interventions, the authors explicitly described their approach as a quasi-experimental multi-country study with matched comparisons, where intervention sites were selected and then compared with matched control sites ^[2]. Similarly, a sanitation and hygiene intervention in Ethiopia was conducted as a community-based quasi-experimental study among households, with intervention and comparison groups chosen by the research team rather than by random assignment ^[4]. A non-randomised cluster trial of a WaSH intervention in rural India also used the label quasi-experimental, non-randomised cluster study to signal that clusters were assigned without randomization ^[3].

These examples reinforce the classification logic for the licensure exam. When the question states that the health office chose the barangays and the outcome is compared over time, the design is experimental because an intervention was delivered, but the absence of random allocation makes it nonrandomized. The fact that the intervention was delivered at the school or barangay level, with outcomes measured in children, confirms that the community is the unit of assignment.

The reason this design produces weaker evidence than a randomized trial is that the selection of high-burden barangays may create systematic differences between groups at baseline. Any observed reduction in diarrhea could be due to the intervention, but it could also be due to regression to the mean, secular trends, or pre-existing differences in risk factors. A randomized design would balance both measured and unmeasured confounders across the two groups, which is why randomization is preferred when feasible ^[1].