This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Acute Respiratory Illness Surveillance (AcRIS) With Mobile Application in a Low-Interventional Decentralized Study. (NCT04748445), a registered phase not stated study. The primary outcome is Change From Baseline in Self-Reported Symptom Scores From Well-to-Sick State Through Day 56.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT04748445 |
| Phase | Phase not stated |
| Condition | Healthy |
| Allocation | Not Stated |
| Design model | Not Stated |
| Masking | Not Stated |
| Enrollment | 9151 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Pfizer |
| Results posted | Yes |
2. Clinical Question
Population
The primary analysis population comprised participants who tested negative for SARS-CoV-2, influenza and/or RSV at Swab 1, had no acute symptoms, and later tested positive for one of these viruses at Swab 2 while symptomatic. The registry lists the condition as healthy and enrollment as 9151; that enrollment figure is not the number analyzed for the primary endpoint.
Intervention
AcRIS gathered self-reported symptoms in an electronic diary and voice recordings through a mobile application, with remote self-swabbing and RT-PCR testing for SARS-CoV-2, influenza and RSV. The registry describes the study as observational and lists the diagnostic test as an intervention.
Comparator
No separate comparison group is described. The registry lists all participants together, and the record does not describe a randomized treatment allocation. The primary analysis instead follows symptom scores across the transition from a well state to a sick state.
Primary question
Among participants meeting the virologic and symptom criteria, how did self-reported total symptom scores change from baseline across the well-to-sick period through Day 56? The score combined ratings for respiratory and other symptoms, with higher values indicating more severe total symptoms. The numerical primary estimate is not reported in the registry record.
3. Trial Design
AcRIS is an observational, low-interventional decentralized study of mobile-app monitoring for acute respiratory illness. Its stated purpose was to characterize the relationship between symptoms and voice features among participants with acute illness confirmed by RT-PCR for SARS-CoV-2, influenza virus or RSV. The study also explored remote recruitment, participant engagement and sample collection as approaches that could inform future clinical studies. These objectives describe the study’s intended use of collected data; they are not evidence that a voice algorithm’s detection performance was established.
The registry lists Pfizer as sponsor, enrollment of 9151, a start date of 2021-04-12, primary completion on 2022-04-22 and status as completed. It lists a single all-participant group rather than randomized treatment arms. Allocation is not stated, and the registry record does not describe a like-for-like control comparison. The RT-PCR test is listed as a diagnostic-test intervention, not as a treatment comparison.
Participants used an electronic diary to record symptoms at least once daily and posted voice recordings. The study also involved self-swabbing, with test results used to identify participants who moved from a negative result at Swab 1 to a positive result at Swab 2 while becoming symptomatic. The primary endpoint followed symptom scores from baseline through Day 56. Secondary outcomes concerned diary and voice-recording compliance and quality, self-swab completion and validity, positive swab results, and timing of swabbing relative to symptoms.
The registry lists 17 outcome measures and 214 analyses. Those counts indicate a broad set of recorded outcomes and analyses, but the registry record does not describe how the analyses were prioritized or whether adjustments were made for multiple testing. Numerical results for the primary endpoint and the listed secondary outcomes are not reported in the registry record.
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
| Stage | All Participants |
|---|---|
| STARTED | 9151 |
| COMPLETED | 8520 |
| NOT COMPLETED | 631 |
Reasons for not completing this period:
| Reason | All Participants |
|---|---|
| No longer met eligibility criteria | 12 |
| Other | 36 |
| Withdrawal by Subject | 553 |
| Lost to Follow-up | 30 |
Analysis population for the primary outcome, as defined in the registry: Participants who initially were SARS-CoV-2, Influenza and/or RSV negative (at Swab 1) and had no acute symptoms, and became SARS-CoV-2, Influenza virus or RSV positive (at Swab 2) and symptomatic during the study were included. Here, 'Overall Number of Participants Analyzed' signifies participants evaluable for this outcome measure and 'Number Analyzed' signifies participants evaluable for the specified rows.
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: Full analysis set included all participants who signed the informed consent document and were enrolled in the study.
| Characteristic | Unit | All Participants |
|---|---|---|
| Participants analysed | 9151 | |
| Age, Continuous | Years | 41.03 (SD 10.63) |
| Sex: Female, Male: Female | Participants | 7235 |
| Sex: Female, Male: Male | Participants | 1916 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 793 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 7915 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 443 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 116 |
| Race (NIH/OMB): Asian | Participants | 89 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 18 |
| Race (NIH/OMB): Black or African American | Participants | 1127 |
| Race (NIH/OMB): White | Participants | 7112 |
| Race (NIH/OMB): More than one race | Participants | 452 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 237 |
6. Primary Endpoint
The registry lists the primary outcome as Change From Baseline in Self-Reported Symptom Scores From Well-to-Sick State Through Day 56, measured over Baseline up to Day 56.
| Element | Registry entry |
|---|---|
| Result type | Least Squares Mean |
| Measure of spread | Standard Error |
| Unit | Scores on a scale |
| Endpoint classification | Continuous Endpoint |
The primary endpoint measures change from baseline in self-reported symptom scores as participants moved from a well state to a sick state, through Day 56. Participants recorded symptoms at least once daily in an electronic diary. The listed symptoms included fever, cough, difficult breathing, fatigue, nasal symptoms, sore throat, loss of taste or smell, chills, muscle pain, gastrointestinal symptoms, headache, nausea, rigors and wheezing. Ratings used a scale from 0:none to 4:severe for fever and from 0:none to 7:severe for other symptoms. A total symptom score was the sum of symptom ratings in a recording session; daily values were averaged across available sessions. The stated range was 0 to 109, with higher scores indicating more severe total symptoms.
Baseline was defined using the average of values up to 7 days before the first occurrence of new or increased symptoms. When there were no values in that period, the registry describes using the average of endpoint values for the closest 3 days prior to the 7-day period. The analysis used a linear mixed-effects model to estimate change over the well-to-sick period, with day-related slope terms and random effects. The reported parameter was the population-level estimate for the fixed effect of slope.
A combined symptom score can describe the overall burden of several symptoms in a single measure, while repeated diary entries can track its course over time. That makes the endpoint relevant to the study’s aim of examining symptom changes alongside voice features. It does not, by itself, establish diagnostic accuracy or show that a mobile application can identify infection. The registry record does not report the numerical primary endpoint estimate or its uncertainty.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry describes an observational analysis of symptom-score changes among participants who were initially virus-negative and without acute symptoms, then became virus-positive and symptomatic. This is a within-course question about symptom patterns, not a randomized comparison of treatments. The record does not describe a separate control group, so the primary endpoint should not be read as an estimate of a treatment effect relative to a comparator.
The endpoint is a total symptom score formed from ratings in an electronic diary. Its meaning depends on the listed symptom set, rating scales and diary entries: higher values represent more severe total symptoms on the defined score. The endpoint description specifies how baseline was set and how the score was summarized over available sessions. Those definitions help interpret what an estimated change would represent, but they do not reveal the size or direction of the observed change.
The registry describes the model as a linear mixed-effects model, with the population-level fixed effect of slope reported for change from baseline during the well-to-sick period. In general, a slope-based estimate summarizes a pattern of change over time rather than a single isolated diary reading. The registry record does not report the estimate, standard error, confidence interval, test statistic or p-value for the primary endpoint. Therefore, the magnitude, precision and statistical compatibility of the primary result cannot be assessed from the registry record.
The record lists 17 outcome measures and 214 analyses. A large analysis set can involve multiple opportunities to identify patterns, so the prespecification and handling of multiplicity matter when interpreting statistical evidence. The multiplicity strategy and the detailed statistical analysis plan are not reported in the registry record. The record also does not state whether particular analyses were designated as confirmatory or exploratory.
Enrollment is listed as 9151, but this is not the primary analysis denominator. The endpoint population was defined by initial negative testing and absence of acute symptoms followed by a positive test and symptoms; the number meeting those criteria is not reported in the registry record. Consequently, enrollment alone cannot establish how much information contributed to the primary analysis, and it should not be substituted for the analyzed population size.
The study’s stated purpose included characterizing relationships between voice features and symptoms and generating data to support algorithm development. The primary endpoint is a symptom measure, not a direct evaluation of an algorithm’s sensitivity, specificity or other diagnostic performance. Those performance results are not reported in the registry record. The study’s observational design and the lack of a separate comparison group also mean that symptom patterns should not be interpreted as evidence that a particular monitoring approach caused a change.
Overall, the registry identifies the intended population, score construction and model framework, but does not provide the numerical primary result needed to judge its estimated size or precision. Any interpretation should distinguish the study’s goals from results actually reported in the registry record.
9. Statistical Methods Explained
The registry describes repeated symptom measurements analyzed with a linear mixed-effects model. Such models are generally used when observations are collected repeatedly from the same participant: readings from one person may be more alike than readings from another person, and a model can represent that within-person structure while estimating an overall pattern. The registry says that slope terms were included as fixed effects and random effects, and that the fixed-effect slope was reported as the population-level estimate. The full model specification, including details such as covariance structure and any additional covariates, is not reported in the registry record.
What does the slope represent?
In general, a slope describes how an outcome changes with time or another ordered measure. For this endpoint, the model addressed symptom-score change from baseline during the well-to-sick period, with day-related slope terms. A population-level fixed-effect slope summarizes the modeled average pattern for the analysis population. Random effects allow individual patterns to vary around that population-level pattern. The registry description does not give the numerical slope estimate, so its direction and magnitude cannot be stated.
The endpoint is described as a least-squares mean, with standard error as the dispersion measure. In general, a least-squares mean is a model-based average, which may reflect the fitted model rather than a simple arithmetic mean of all recorded values. A standard error describes the estimated precision of a model-based quantity; it is not a measure of how widely individual participants’ scores vary. The registry record does not include the primary estimate or its standard error, and therefore does not permit an assessment of the estimate’s precision.
Why use a mixed-effects model?
General statistical rationale: repeated observations within a participant are correlated, so treating each diary entry as if it came from a different, unrelated person can misrepresent uncertainty. Mixed-effects models can account for the clustered nature of repeated measurements and can represent participant-level differences in trajectories. They are a common framework for longitudinal outcomes. The particular assumptions and diagnostics used in this analysis are not reported in the registry record, so the suitability of those assumptions cannot be evaluated from the registry description alone.
In general, a confidence interval expresses a range of values compatible with an estimate under the model and its assumptions. Its width conveys precision: a narrower interval indicates more precise estimation than a wider interval. No confidence interval for the primary endpoint is reported in the registry record. A standard error is reported as the endpoint’s dispersion type, but the numerical standard error is not reported in the registry record.
How is the baseline defined?
The registry defines baseline as the average of symptom-score values recorded up to 7 days before the first occurrence of new or increased symptoms. If no data were available in that 7-day interval, the baseline used the average of endpoint values for the closest 3 days prior to the interval. This rule attempts to anchor change to a pre-symptom period while allowing a baseline to be derived when diary values were absent in the preferred window. The number of participants using the fallback definition is not reported in the registry record.
Symptom scores were based on diary sessions, and the daily total was averaged across available sessions. This means the measured daily value depends on the symptom ratings recorded and the sessions available for that day. The registry lists percentage of compliant days and percentage of quality voice recordings among secondary outcomes, but numerical results for these outcomes are not reported in the registry record. The record does not describe a specific imputation method for missing diary values or explain how incomplete recording patterns were handled beyond the stated use of available sessions.
What can planning illustrations say?
In general, sample-size calculations can illustrate the number of participants per arm that would be needed to detect a chosen effect under stated assumptions. The planning approach described for this study uses a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation, with the target effect expressed as a fraction of the observed effect. Such calculations are planning illustrations, not observed power and not evidence that the study was designed or powered that way. This study is not a randomized two-arm comparison, and the registry record does not report the observed effect needed to assess an illustration. The registry record also does not state that these planning assumptions formed part of the study’s original design.
General considerations about over-dispersion apply when count data vary more than a simple count model expects; accounting for that extra variability can affect uncertainty estimates. The primary endpoint is a continuous symptom score, not a count outcome, and the registry description does not report an over-dispersion model. That concept should therefore not be mistaken for a method stated for this endpoint.
The registry lists many analyses alongside many outcome measures. In general, when multiple statistical questions are examined, the chance of at least one apparently unusual result can increase if each is judged without accounting for the wider set. Adjustments, hierarchical testing or a clear distinction between confirmatory and exploratory analyses can address that issue in different ways. The statistical plan and any multiplicity procedure are not reported in the registry record. Without numerical results and those analysis details, the primary endpoint’s statistical evidence cannot be independently assessed.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists 17 outcome measures and 214 analyses, including the primary symptom-score endpoint and secondary outcomes relating to diary compliance, voice-recording quality, positive self-swab results, swab administration, timing and validity. When many outcomes and analyses are considered, multiplicity is relevant: in general, repeated testing can make apparently notable findings more likely by chance unless the analysis plan accounts for the number and structure of comparisons. The registry record does not report a multiplicity adjustment, a testing hierarchy, or which analyses were intended to be confirmatory.
The registry describes symptom values as averages across available diary sessions and defines a fallback baseline when values are absent in the preferred pre-symptom interval. It does not report the extent of missing diary data, the number of participants using the fallback baseline, or a broader missing-data strategy. The record does not state whether any imputation was used, what assumptions were made about missingness, or whether sensitivity analyses assessed those assumptions. Mixed-effects models can use repeated observations when some measurements are absent, but their interpretation depends on assumptions; the registry description does not document those assumptions or how they were evaluated.
The primary analysis population was restricted to participants who were initially negative and asymptomatic, then became positive and symptomatic. The number of participants meeting those criteria and the number included in the primary analysis are not reported in the registry record. Enrollment of 9151 should not be treated as the endpoint denominator. The registry also does not report the analysis population for each secondary outcome or explain how participants with incomplete self-swab, diary or voice data were handled.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| All Participants | 0 | 9151 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| All Participants | 0 of 9151 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| All Participants | 28 of 9151 |
Most frequent non-serious adverse event terms
| Event term | Organ system | All Participants |
|---|---|---|
| Headache | Nervous system disorders | 11 of 9151 |
| Pyrexia | General disorders | 7 of 9151 |
| Diarrhoea | Gastrointestinal disorders | 6 of 9151 |
| Pain | General disorders | 5 of 9151 |
| Nasal congestion | Respiratory, thoracic and mediastinal disorders | 5 of 9151 |
| Rhinorrhoea | Respiratory, thoracic and mediastinal disorders | 5 of 9151 |
| Nausea | Gastrointestinal disorders | 4 of 9151 |
| Chills | General disorders | 4 of 9151 |
| Fatigue | General disorders | 4 of 9151 |
| Cough | Respiratory, thoracic and mediastinal disorders | 4 of 9151 |
The registry lists safety information for the all-participant group. Serious adverse events were reported for 0 participants among 9151 at risk, other adverse events for 28 participants among 9151 at risk, and deaths for 0. No serious adverse-event terms are listed. These are descriptive registry counts; they do not establish that an event was caused by study participation or by a particular component of the mobile monitoring and testing process.
The record does not give further detail about the 28 other adverse events, such as their types, timing, severity or relationship to study procedures. It also does not report a comparative safety analysis. Because the study is observational and has no separate comparison group described, these figures cannot be used to estimate a relative risk against another group. The registry counts should be read as reported safety information for the listed group, not as a causal assessment or a complete account of every possible event.
12. What the Result Does, and Does Not, Mean
The study addresses whether self-reported symptom patterns and voice features could be characterized in participants with acute respiratory viral illness using remote, app-based data collection. Its primary endpoint was designed to describe symptom-score change from a pre-symptom baseline through the well-to-sick period. That focus can inform understanding of how recorded symptom burden changes over time in the defined group, if the numerical estimates and their precision are available.
The registry record does not report the numerical primary endpoint result. It therefore does not establish how much scores changed, whether the modeled trend was precise, or how the results compare with a meaningful threshold. It also does not report numerical findings for the listed secondary outcomes. The record describes planned aims and measurement methods, but those descriptions are not substitutes for reported results.
The endpoint does not directly test whether voice recordings can diagnose a viral infection. Although the study’s purpose included using collected data as a basis for voice and symptom algorithms, the primary endpoint is a symptom-score measure. No algorithm-performance estimates are reported in the registry record. Without such results, diagnostic performance cannot be inferred from the study’s stated goals or from its enrollment figure.
The observational design is also important. The registry lists all participants together and does not describe randomized allocation or a separate control group. The study can characterize patterns among the eligible participants, but it is not a controlled comparison of monitoring strategies or treatments. It cannot establish that the app, diary, voice recording or self-swabbing caused a particular symptom pattern.
The registry lists enrollment of 9151, while the primary endpoint applied to a more specifically defined population. The number meeting those endpoint criteria is not reported in the registry record. That distinction matters because enrollment does not indicate how many participants contributed to a given analysis or how complete their repeated measurements were. The record also does not report enough detail to assess missing-data assumptions or multiplicity handling.
In short, the record describes an approach to decentralized respiratory-illness data collection and defines a longitudinal symptom endpoint. It does not provide the estimates needed to judge the observed symptom trajectory, its precision, or the performance of any resulting algorithm. The study’s stated potential uses should be distinguished from findings documented in the registry record.
13. Why This Trial Matters Statistically
AcRIS is statistically relevant because it combines repeated symptom measurement, voice capture and remote testing in a decentralized observational study. These methods create a longitudinal data structure: diary scores are collected over time, while laboratory results and symptom onset help define the subgroup of interest. Understanding the structure matters because repeated measurements from the same participant are not generally independent, and the registry describes a mixed-effects model intended to estimate population-level change while accounting for participant-level variation.
The score also illustrates a common measurement choice: combining ratings across multiple symptoms to represent overall symptom burden. Such a composite can provide a single trajectory to analyze, but its interpretation depends on which symptoms are included, how each is rated, and how session-level and daily scores are combined. AcRIS specifies these elements, including different rating ranges for fever and other symptoms and an overall score with a stated range of 0 to 109. These details help readers understand what the endpoint represents rather than treating it as a generic measure of illness.
The study’s decentralized methods raise important questions about data completeness and quality. Diary compliance, quality voice recordings, self-swab completion and validity appear among the secondary outcomes. These are not merely operational details: the availability and quality of repeated measurements influence what can be learned from longitudinal analyses and from future algorithm development. Their numerical results, however, are not reported in the registry record.
Finally, a record listing 17 outcome measures and 214 analyses underscores the importance of transparent analysis plans and multiplicity handling. A clear account of the primary question, missing-data approach and analysis populations helps readers distinguish robust, prespecified evidence from exploratory patterns. Those details and numerical outcome results are not reported in the registry record, limiting statistical interpretation despite the detailed endpoint definition.
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 principal limitation for interpreting this record is that numerical results for the primary endpoint are not reported in the registry record. The same is true for the listed secondary outcomes. Without estimates and uncertainty measures, readers cannot assess the observed symptom-score trajectory, the precision of that estimate, or the findings for diary compliance, recording quality and self-swab outcomes.
The analyzed population is narrower than the listed enrollment: the primary endpoint concerns participants who initially tested negative and had no acute symptoms, then tested positive and became symptomatic. The number meeting those criteria and the number included in the analysis are not reported in the registry record. Enrollment of 9151 therefore does not establish the endpoint sample size or the amount of repeated data available.
The registry describes an observational design and does not describe a separate comparison group or randomized allocation. As a result, the primary analysis is not a controlled comparison, and findings about symptom trajectories cannot establish that the mobile application or another study procedure caused those trajectories. The stated algorithm-development aim is also distinct from demonstrated diagnostic performance; no performance estimates are reported.
Several analysis details that would support evaluation are not reported in the registry record: the multiplicity strategy, the broader missing-data approach, sensitivity analyses, model diagnostics, and full model specification. The record defines a baseline fallback and use of available diary sessions, but does not say how often those rules were needed or how incomplete data affected results. Safety reporting is similarly limited: the count of other adverse events is listed without event descriptions or a comparative assessment.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Acute Respiratory Illness Surveillance (AcRIS) With Mobile Application in a Low-Interventional Decentralized Study. — NCT04748445.
- Registry record last updated 2024-02-23; status last verified 2024-02.
- Results first posted 2024-02-23.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
AcRIS (NCT04748445) is a completed observational study sponsored by Pfizer, with enrollment listed as 9151. The registry gives a start date of 2021-04-12 and primary completion date of 2022-04-22. The study used a mobile application for symptom diaries and voice recordings and included remote self-swabbing with RT-PCR testing for SARS-CoV-2, influenza and RSV.
The primary endpoint was change from baseline in self-reported symptom scores through Day 56 among participants who were initially virus-negative and asymptomatic, then became positive and symptomatic. The registry describes a linear mixed-effects model and a population-level slope estimate. It lists 17 outcome measures and 214 analyses. Numerical primary and secondary outcome results, the primary analysis count, a multiplicity strategy and a broader missing-data plan are not reported in the registry record.
Safety entries list 0 serious adverse events, 28 other adverse events and 0 deaths among 9151 at risk. The registry does not provide further descriptions of the other adverse events or a comparative safety analysis.