Impact of COVID-19 on Marshallese Communities in the U.S.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Impact of COVID-19 on Marshallese Communities in the U.S. (NCT04474496), a registered phase not stated study. The primary outcome is Personal Preventative Behaviors to Decrease Risk of COVID-19 Exposure.

2. Clinical Question

This is a descriptive question about self-reported behavior in a defined population. It does not ask whether one intervention changes behavior relative to another, or whether a particular behavior prevents infection. The planned survey also addressed risk exposure, knowledge of preventive recommendations, barriers and facilitators to prevention, barriers and facilitators to testing when appropriate, and self-care behaviors. Its secondary outcomes covered general health status, access to healthcare, dietary habits, physical activity, weight, HbA1c, and barriers to type 2 diabetes management. These topics place the primary question within a wider account of how the pandemic affected Marshallese communities. The record does not report estimates for the primary question or the secondary outcomes.

3. Trial Design

The University of Arkansas-sponsored study is listed as completed. Its recorded start date is 2020-07-27, and its primary completion date is 2020-11-30. Enrollment is listed as 120. The record describes Marshallese adults living in the United States completing an online or telephone survey. Eligibility required self-reported Marshallese identity, age 18 years or older, and residence in the continental U.S. or Hawaii.

The registry lists one group, “Marshallese Adults in the U.S.” It does not describe a randomized allocation or a like-for-like control comparison. The allocation is listed as not stated. The design information therefore supports describing a survey in one population, not a comparison of assigned treatments. The intervention label in the registry should be understood in that context: the described activity is assessing COVID-19 impact through a survey.

The primary outcome is measured at baseline and asks participants which preventive measures they have taken, allowing multiple selections. The registry lists 18 outcome measures overall and eight named secondary outcome topics. The outcomes concern reported experiences and behaviors, including healthcare access, dietary and physical activity behaviors, and diabetes-related matters. The record does not provide the full measurement instruments, response coding details, or timing for each secondary outcome.

The registry lists zero analyses. It does not report a statistical analysis plan, a sample-size justification, interim analyses, or details about how survey responses were summarized. Although the record contains assumptions for planning illustrations based on an observed effect, those assumptions do not establish that the study was designed as a two-arm trial. Such illustrations are not observed power and do not supply a missing comparison group or an unreported analysis. The completed status and enrollment figure describe the registry record, but do not by themselves establish how many participants answered each outcome item.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Analysis population for the primary outcome, as defined in the registry: N=117 due to missing data on this item for three participants.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: 120 eligible individuals completed at least 10 survey items and were included in the final analytical sample

6. Primary Endpoint

The registry lists the primary outcome as Personal Preventative Behaviors to Decrease Risk of COVID-19 Exposure, measured over Baseline.

The primary endpoint is titled “Personal Preventative Behaviors to Decrease Risk of COVID-19 Exposure.” It concerns participants’ reports of behaviors they regularly engage in to lessen their risk of contracting COVID-19. The survey question asks which measures participants have taken to prevent infection and gives social distancing and wearing a mask as examples. Participants could select all that apply, so the question is not limited to a single mutually exclusive response.

The endpoint is recorded in participants and is classified as a count of participants. In a descriptive survey, counts of participants selecting each behavior can show how common each reported response was among those who answered the item. This is useful for describing reported practices in the surveyed population and for identifying which behaviors participants say they use. The registry record does not report the response counts or any corresponding proportions, so the extent of each behavior cannot be established from the registry information.

The item is specified at baseline. Its population is listed as N=117 because three participants were missing data on this item. That missingness matters when interpreting any count: the item denominator is not necessarily the full enrollment. The endpoint records self-report, not independently verified behavior, and does not measure whether a behavior prevented infection. It also does not compare behavior between assigned groups. Those distinctions define what the measure can describe and what claims it cannot support.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

For a survey outcome, the first statistical question is what was measured and in whom. The primary measure asks participants which preventive behaviors they had taken. It is recorded in participants, at baseline, and permits selecting all applicable responses. The registry describes the endpoint as a count of participants. Such a count is a descriptive summary: it records how many respondents selected a particular response. It is not, by itself, a treatment effect or a measure of clinical benefit. The registry record does not report the observed counts.

Counts, proportions, and denominators. A count gives the number of respondents in a category. A proportion relates that count to the number of people with usable responses for the item. These answer related but distinct questions: the count gives an absolute number, while the proportion expresses a frequency within a defined denominator. The primary item’s listed population is N=117 because three participants had missing item data, whereas enrollment is 120. A valid interpretation must use the denominator appropriate to the item, not assume that every enrolled participant answered every question. No proportion can be stated without a response count.

Because the primary question allows multiple selections, response categories can overlap. One participant may contribute to more than one behavior count. In general, that means category counts should not be interpreted as a partition of respondents, and their sum need not equal the number with item data. A multi-select question describes endorsement of each listed response; it does not identify one exclusive behavior for every person.

Uncertainty and confidence intervals. General statistical explanation: an estimate from a sample is subject to sampling variability. A confidence interval expresses the range of values compatible with a statistical procedure and its assumptions; a narrower interval generally indicates greater precision. Its interpretation depends on the sampling design, measurement, and analytic method. The registry information does not report a confidence interval for the primary endpoint, nor enough outcome data to reconstruct one. Therefore, neither precision nor statistical significance can be assessed from the reported endpoint description alone.

Comparisons and causal interpretation. General statistical explanation: a comparative effect requires a defined contrast between groups or conditions. Random allocation, when used and documented, can help make groups comparable on average, but this registry record does not describe randomized allocation. It lists one group and says allocation was not stated. There is therefore no reported treatment contrast for the primary outcome. The survey can describe reported behaviors and experiences, but its endpoint does not by itself estimate what would have happened under an alternative intervention or establish that a behavior changed because of the study.

Over-dispersion and count models. General statistical explanation: count outcomes are sometimes analyzed with probability models. A simple count model may assume that variation follows a particular relationship to the average count. Over-dispersion occurs when the observed variability is greater than that model allows; using a model that fails to accommodate it can make uncertainty estimates too small. The registry record does not report a count model, a dispersion estimate, or an analysis for the primary endpoint. The endpoint’s count classification alone is not evidence that a particular model was used. In a descriptive presentation, raw counts and clearly defined denominators may be more directly interpretable than a model-based contrast, but the actual analysis is not reported.

Planning illustrations. The registry’s sample-size assumptions describe a planning illustration based on an observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. General statistical explanation: a planning calculation estimates how many participants might be needed under specified assumptions to detect a chosen effect with a chosen power. It is prospective design guidance, not an outcome finding. If it is based on an observed effect, that effect is used as an input to the illustration; it does not make the resulting sample size an observed result or observed power. The registry does not report a sample-size justification for this survey. Since the described study has one listed group and no stated randomized comparison, the two-arm assumptions should not be mistaken for the study’s documented design.

Multiplicity. When a study measures many outcomes or tests many hypotheses, the chance of at least one apparently unusual result can increase if no adjustment or hierarchy is specified. This is a general concern, not evidence that a particular adjustment was necessary or performed in this study. The registry lists 18 outcome measures and zero analyses, and does not report a multiplicity strategy. Without reported analyses, it is not possible to assess how outcomes were selected for emphasis, whether tests were conducted, or how uncertainty was handled across outcomes.

These methods concepts help distinguish a survey description from an inferential comparison. The registry supplies design and measurement context, but not the numerical outcome results or analytic details needed to assess an estimated effect. General statistical principles can clarify what such results would mean; they cannot fill in values that are not reported.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 18 outcome measures and zero analyses. It names secondary outcome topics, including general health status, access to healthcare services, dietary habits, physical activity, weight, HbA1c, and barriers to type 2 diabetes management. The registry record does not report which outcomes were analyzed, how results were selected for presentation, or whether any adjustment or hierarchy was used to address multiple outcomes. General statistical concern: when many outcomes or hypotheses are examined, interpreting isolated findings without accounting for multiplicity can overstate how persuasive they are. This is a caution about interpretation, not a claim that a particular analysis or adjustment occurred.

Missing data are explicitly noted for the primary item: the listed item population is N=117 because three participants were missing data on that question. The record does not describe why those responses are missing, whether the missingness was related to participant characteristics or responses, or how missing data were handled in any analysis. It also does not report item-specific denominators for the secondary outcomes. The enrolled total of 120 should not be assumed to be the analysis population for every measure.

The record does not describe an analysis population definition, imputation procedure, sensitivity analysis, or method for handling incomplete survey responses. Those details are not reported in the registry record. Because outcome results and analyses are not reported, the impact of missing responses and multiplicity on any estimated finding cannot be evaluated.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry lists one group, “Marshallese Adults in the U.S.” For that group, it records 120 participants at risk for serious adverse events and 0 affected; it also records 120 at risk for other adverse events and 0 affected. The deaths figure is 0. No top serious adverse event terms are listed.

These are the adverse-event figures recorded in the registry, not a comparison between treatment and control groups. The record does not describe a randomized comparator, and these counts do not provide a relative risk or a comparative safety estimate. The at-risk counts should also be read in the context of enrollment, rather than assumed to establish complete follow-up or complete ascertainment for every participant.

A recorded count of 0 affected means that the registry lists no affected participants in those adverse-event categories. It does not, by itself, establish how adverse events were collected, what observation period applied, or whether all events were captured. Those details are not reported in the registry record. The figures therefore describe the recorded counts, but do not support broader claims about safety beyond that record.

12. What the Result Does, and Does Not, Mean

The registry record establishes that a completed survey study enrolled 120 people and asked Marshallese adults in the United States about COVID-19 experiences and behaviors. Its primary question concerns self-reported preventive measures, and the item has a listed population of N=117 because three participants were missing data for that question. The record also identifies a broad set of secondary topics, from healthcare access and health behaviors to diabetes management.

What the record does not establish is the distribution of answers. It does not report how many participants selected any preventive behavior, provide an estimate of frequency, or give uncertainty around an estimate. It therefore cannot show which behavior was most commonly reported or how precisely any frequency was measured. The absence of results in the registry record should not be interpreted as a finding of no behaviors or no effects; it is an absence of reported numerical results.

The survey is not described as a randomized comparison. It does not establish that a particular behavior reduced infection, that the survey changed behavior, or that COVID-19 caused a particular health outcome among participants. Self-reported behavior is also not the same as independently verified behavior. The measure captures what respondents said they had done, within the wording and response options of the survey.

The adverse-event figures record zero affected participants and zero deaths in the listed group, but do not provide a comparative safety analysis or details on event collection. The record’s completed status and its planning assumptions likewise do not substitute for outcome estimates. Any sample-size calculation based on an observed effect is an illustration under its stated assumptions, not observed power or evidence of a clinical effect. The most defensible reading is therefore about scope and measurement: the study asked relevant descriptive questions in a specific population, while the numerical findings and analytic details needed to judge the results are not reported in the registry record.

13. Why This Trial Matters Statistically

This study is statistically notable for the distinction between documenting a question and reporting its answer. The record defines a specific population and a concrete primary survey item, but does not give response counts or estimates. That distinction matters because readers can understand the intended measurement without being able to assess the observed distribution or its precision.

The item also illustrates why denominators deserve attention. Enrollment is 120, while the primary item’s population is listed as N=117 because three participants had missing data on that item. A descriptive count must be interpreted against the relevant set of responses. Treating the enrolled total as the denominator for every outcome would ignore the item-level missingness documented by the registry.

Allowing multiple selections changes how category counts should be read. The responses are not mutually exclusive options, so each behavior is best understood as a separate reported measure rather than a part of a single total that assigns every respondent to one category. This is an important design feature for communicating survey results clearly.

The study also calls attention to the boundary between descriptive and causal questions. A survey can characterize what participants report, but without a documented comparison it cannot estimate a randomized treatment effect. The registry’s two-arm planning assumptions do not alter the one-group description or establish a trial comparison. More broadly, the listing of 18 outcome measures alongside zero analyses underscores the difference between measuring many topics and reporting inferential results for them. Statistical interpretation depends on the actual estimates, denominators, uncertainty, and methods; these are not reported in the registry record.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record leaves important details unavailable for interpreting the survey. It does not report primary response counts, estimates, confidence intervals, or p-values. It also does not provide numerical results for the named secondary outcome topics. Consequently, the magnitude, distribution, and precision of the measured outcomes cannot be assessed from the registry information.

The primary endpoint is self-reported and permits multiple responses. Self-report describes participant answers, not independently verified behavior. The record does not provide the full questionnaire, the complete response options, or details about how responses were coded. Although it gives examples of preventive measures and states that participants could select all that apply, further measurement properties are not reported.

Three participants are noted as missing data on the primary item, leaving a listed item population of N=117. The record does not explain the reason for missing responses or how missingness was handled. It also does not report whether secondary measures had missing data or how their analysis populations were defined. The broader survey’s online or telephone format is described, but the registry does not report how mode of response was handled analytically.

Only one group is listed, with no randomized comparator described. The record does not support a treatment-effect interpretation or a causal conclusion from the primary endpoint. It lists zero analyses and does not report an analysis plan, sample-size justification, interim analyses, or a multiplicity strategy. The adverse-event counts are reported, but event-collection methods and observation details are not. These omissions limit assessment of the reliability, precision, and scope of conclusions that might otherwise be drawn from the survey.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Impact of COVID-19 on Marshallese Communities in the U.S. — NCT04474496.
• Registry record last updated 2021-04-05; status last verified 2021-04.
• Results first posted 2021-04-05.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT04474496, “Impact of COVID-19 on Marshallese Communities in the U.S.,” is listed as completed and sponsored by the University of Arkansas. The recorded start date is 2020-07-27, primary completion date is 2020-11-30, and enrollment is 120. Marshallese adults aged 18 years or older living in the continental U.S. or Hawaii were invited to complete an online or telephone survey.

The primary outcome concerns reported personal preventive behaviors at baseline. The registry lists N=117 for that item because three participants had missing data. It lists 18 outcome measures and zero analyses. Numerical primary and secondary outcome results are not reported in the registry record. The adverse-event listing records 0 affected participants for serious and other adverse events and 0 deaths in the group of 120 at risk.