i-Matter: Investigating an mHealth Texting Tool for Embedding Patient-reported Data Into Diabetes Management: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for i-Matter: Investigating an mHealth Texting Tool for Embedding Patient-reported Data Into Diabetes Management (NCT03652389), a randomized na study. The primary outcome is HbA1c Level at Baseline.

2. Clinical Question

3. Trial Design

The registry describes i-Matter as a randomized, single-masked, parallel-group study sponsored by NYU Langone Health. It is marked completed. The recorded start date is 2018-12-18, and the primary completion date is 2024-01-31. The study is listed as not applicable for phase.

The study description presents a mixed-methods project with a formative phase and a clinical-efficacy phase. The formative work was intended to adapt the mobile platform to patients and primary care providers, integrate it into practice workflows and the electronic health record, and assess usability. The clinical-efficacy phase was described as a randomized comparison of MJS DIABETES with usual care. This distinction matters: the formative aims concern usability and integration, while the randomized phase is framed around HbA1c.

In the intervention arm, participants were to receive and respond to daily patient-reported outcome messages during the 12-month study. Participants who were insulin-dependent were also to report self-monitored blood glucose. Usual care is described as standard diabetes treatment. The registry does not report additional details about the allocation process, concealment, the identity of the masked party, or whether intervention use varied across participants.

The record contains an enrollment figure of 246, but its brief summary describes the efficacy trial as involving 282 patients. The relationship between those figures is not reported in the registry record. The arm-level baseline outcome entries list 117 for MJS DIABETES and 120 for usual care; they do not resolve the enrollment discrepancy, and the figures should not be treated as a reconciliation of it.

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.

Before assignment: N= 246 participants were enrolled, N=9 of these participants were not randomized to treatment.

Overall Study

Reasons for not completing this period:

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.

6. Primary Endpoint

The registry lists the primary outcome as HbA1c Level at Baseline, measured over Baseline.

HbA1c is a measure of hemoglobin that has glucose attached to it, and the registry records it as a percentage of total hemoglobin. In diabetes research it is used as a clinical measure of glycemic status over time. The registry says HbA1c was extracted from patient medical records.

The primary endpoint listed in the registry is HbA1c at baseline. Its reported group values are model-adjusted means with 95% confidence intervals, and the description says multiple imputation was used. The MJS DIABETES mean is 8.37, with a 95% confidence interval from 8.06 to 8.68. The usual-care mean is 8.29, with a 95% confidence interval from 7.98 to 8.6.

Baseline measurement can describe the participants’ starting status and help characterize the groups. It is not, by itself, a measure of change during treatment or a comparison of outcomes after the intervention. Although the study summary frames the efficacy question around HbA1c reduction at 12 months, the registry’s listed primary endpoint and posted primary values are at baseline. A 12-month HbA1c result, change estimate, or between-group treatment effect is not reported in the registry record. Therefore, the baseline endpoint does not establish an intervention effect.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the study as randomized and parallel-group. Random allocation is intended to make treatment assignment independent of participant characteristics, on average, and a parallel design follows groups assigned to different conditions. These design features support comparisons of outcomes when the outcomes and analyses are reported. The registry record does not give enough detail to assess the allocation sequence, concealment, or the specific masking procedures.

What do the means describe?

A mean is the arithmetic average of a continuous measure. Here, HbA1c is recorded as a percentage of total hemoglobin, while the secondary diabetes self-care measures are scale scores. A mean summarizes the group’s central level, but it does not describe the full spread of individual observations. The self-care outcomes also report standard deviations, which describe variation among observations around the mean. A standard deviation is not the same as a confidence interval: the former describes variability in the data, while the latter describes uncertainty in an estimated quantity.

The primary baseline means are described as model-adjusted. An adjusted mean is estimated using a statistical model rather than being only the raw average. The registry does not report the model specification, the covariates included, or the adjustment rationale. Those details are needed to reproduce or fully evaluate the adjusted estimates.

How should confidence intervals be read?

A confidence interval conveys the precision of an estimate under the assumptions of the analysis. A narrower interval indicates greater precision than a wider interval for the same kind of estimate, although precision does not establish clinical importance. The primary outcome’s 95% confidence intervals describe uncertainty around the adjusted baseline means. They are not intervals for a treatment effect, and they should not be interpreted as containing a specified proportion of individual participants’ measurements.

For continuous outcomes, a mean difference expresses the difference between group means in the scale’s units. In the recalculated secondary comparisons, the reported estimates and p-values are simple, unadjusted recalculations from the posted summary statistics. They are not identified as the sponsor’s reported inferential analyses; no reported analyses are listed for these secondary outcomes. The recalculated results do not account for additional model adjustments or other design features that are not available from the summary data.

What does a p-value tell the reader?

A p-value measures how incompatible the observed data are with a specified null hypothesis under the assumptions of a statistical test. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of an effect. The secondary p-values are unadjusted. Because they concern baseline self-care measures, they should not be mistaken for tests of whether the intervention changed those measures.

Some continuous outcomes can have more variability than a simple model assumes, or their distributions can depart from model assumptions. In general, over-dispersion means the observed variability exceeds what a chosen model expects; ignoring it can make uncertainty estimates too small. Whether over-dispersion was assessed or relevant to these analyses is not reported in the registry record. No distributional diagnostics are reported.

What do the sample-size illustrations mean?

The planning illustrations are based on the observed secondary effect estimates and are not the trial’s original design justification. Their stated assumptions are a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. The illustrations vary the effect to detect as a fraction of the observed effect and give participants per arm for 80% and 90% power. They are prospective planning aids: they do not represent observed power, do not establish that the study was adequately powered, and should not be read as results of the completed trial. The registry does not report the original sample-size calculation or its assumptions.

Planning numbers are also sensitive to the effect assumed. If a future study aims to detect an effect smaller than the observed estimate, more participants may be needed under the same general assumptions. That is a general statistical relationship, not a finding about this trial’s design. The illustrations for the secondary outcomes do not substitute for a sample-size calculation for the planned 12-month HbA1c endpoint.

How does missing-data handling affect interpretation?

The primary outcome description states that multiple imputation was used. In general, multiple imputation replaces missing values with draws from a model, analyzes the resulting completed datasets, and combines the estimates to reflect uncertainty from missingness. Its validity depends on assumptions about why data are missing and on the imputation model. The registry does not report the amount or pattern of missing data, the imputation model, the variables included, or sensitivity analyses. It is therefore not possible to assess those assumptions from the record.

The registry identifies the primary values as model-adjusted and imputed, but it does not report the analysis population, the handling of protocol deviations, or a detailed statistical analysis plan. These omissions limit the ability to reproduce the analysis and to judge whether the reported approach matches the planned efficacy question.

10. Multiplicity, Missing Data and Analysis Populations

The registry record does not report a statistical analysis plan describing a primary hypothesis test, a multiplicity strategy, or an order of testing across outcomes. It lists 43 secondary outcomes and reports no analyses for the secondary outcomes. Without a documented testing strategy, the recalculated unadjusted secondary p-values should not be treated as confirmatory. When researchers examine multiple outcomes, chance alone can produce apparently notable findings; methods such as prespecifying a hierarchy or adjusting inference can address that issue. Whether any such method was used is not reported in the registry record.

Missing-data information is limited. The primary outcome description says that multiple imputation was used, but the record does not report the quantity of missing data, reasons for missingness, imputation model, imputed variables, or sensitivity analyses. In general, multiple imputation can account for uncertainty associated with missing values when its model and assumptions are appropriate. The available description is not sufficient to evaluate those assumptions or to judge how sensitive the estimates are to alternative missing-data approaches.

The analysis population is not reported in the registry record. The record does not specify whether the primary analysis followed intention-to-treat principles, how participants with incomplete follow-up were handled, or whether any per-protocol or other analysis population was used. It also does not report an analysis plan against which the posted results could be compared. These gaps limit assessment of selection, adherence, and robustness.

Finally, the key planned efficacy outcome is not available in the posted results: the brief summary describes HbA1c reduction at 12 months, while the listed primary endpoint and values concern baseline HbA1c. The registry does not report the 12-month result, an analysis of change, or the corresponding between-group estimate. Multiplicity and missing-data questions are important for interpreting such an analysis, but the absence of that outcome result is the more fundamental limitation for the clinical-efficacy question.

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 adverse-event figures report zero serious adverse events and zero deaths in each arm. The registry lists 117 at risk for serious adverse events in MJS DIABETES and 120 in usual care, and the same arm-specific figures at risk for other adverse events. It reports zero other adverse events in both groups as well.

These entries should be read as registry-reported counts, not as proof that no adverse events occurred outside the recorded definitions, observation period, or reporting process. The registry does not provide serious-event terms, and no serious-event terms are listed. It also does not report additional details about how adverse events were collected, how long safety follow-up continued, or whether the counts reflect all randomized participants. The figures alone do not permit a detailed comparison of safety between groups.

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

The posted results can describe baseline status, but they do not answer whether the text-message platform changed HbA1c. The baseline HbA1c means and confidence intervals characterize the groups at the start point recorded as the primary endpoint. They are not follow-up values and are not estimates of a between-group treatment effect. The study summary’s intended efficacy question concerns HbA1c reduction at 12 months; that result is not reported in the registry record.

The same distinction applies to the secondary self-care scores. Their baseline means and standard deviations show group-level starting values. The recalculated unadjusted comparisons describe those baseline summaries only. They do not show changes over time or establish that the platform improved self-care behaviors. A randomized design supports causal comparisons when relevant outcome data and an appropriate analysis are available, but the design label alone cannot supply a missing treatment-effect estimate.

Statistical uncertainty and clinical importance are also different. Confidence intervals communicate precision under a model; p-values describe compatibility with a null hypothesis under test assumptions. Neither alone establishes whether a difference matters to patients or clinicians. The registry does not report a clinically meaningful threshold for HbA1c change, and no such threshold should be inferred from baseline data.

The zero adverse-event counts are limited to what the registry records and do not by themselves establish comparative safety. Further information about definitions, surveillance, and follow-up would be needed for a fuller safety interpretation. Overall, the record supports a description of the trial design, baseline measures, and recorded adverse-event counts. It does not support a conclusion about the clinical efficacy of MJS DIABETES at 12 months.

13. Why This Trial Matters Statistically

This trial raises a useful statistical distinction between a study’s intended question and the outcome values that are actually reported. The summary describes a randomized efficacy comparison centered on HbA1c reduction at 12 months, but the listed primary endpoint is baseline HbA1c. Baseline measurements can describe a study population; they cannot stand in for a follow-up comparison. Keeping the time point and estimand clear prevents descriptive starting values from being mistaken for evidence of treatment effect.

The record also illustrates why the statistical details accompanying an estimate matter. The primary values are described as model-adjusted and based on multiple imputation, yet the model, missing-data assumptions, and analysis population are not reported. These details influence how precisely an estimate can be interpreted and whether it can be reproduced. Confidence intervals are informative, but their meaning depends on the analysis that generated them.

Secondary outcomes create another interpretive challenge. Numerous outcomes can be reported, while unadjusted comparisons may be vulnerable to chance findings. Here, the listed secondary analyses are not reported, and the available recalculations concern baseline scores. Prespecification and transparent multiplicity methods help readers distinguish confirmatory evidence from exploratory patterns.

Finally, the enrollment figure in the record differs from the number described in the study summary. The record does not explain the difference. Statistical reporting is strongest when participant flow, analysis populations, endpoint timing, and planned comparisons can be followed consistently. The gaps do not erase the descriptive value of the posted baseline data, but they constrain what those data can establish.

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 main limitation for interpreting efficacy is the mismatch between the planned clinical question and the posted primary endpoint. The summary describes HbA1c reduction at 12 months, whereas the listed primary outcome is baseline HbA1c. The 12-month result and a between-group estimate of change are not reported in the registry record. This means the posted primary data do not establish an intervention effect.

The registry also contains different enrollment descriptions: it records enrollment of 246, while the brief summary describes a randomized clinical-efficacy trial among 282 patients. The record does not explain the difference. Arm-level outcome entries list 117 for MJS DIABETES and 120 for usual care, but these outcome entries do not clarify the enrollment figures or participant flow.

Important statistical details are absent, including the original sample-size justification, full analysis plan, model specification, analysis population, allocation concealment, identity of the masked party, and any interim analyses. The registry does not report the amount or pattern of missing data, the multiple-imputation model, or sensitivity analyses. These omissions limit reproducibility and assessment of analytic robustness.

The available secondary results are baseline measures, and the recalculated comparisons are unadjusted. The registry lists 43 secondary outcomes but no reported analyses for them. The absence of a reported multiplicity strategy also limits confirmatory interpretation. Safety entries show zero recorded events, but event definitions, collection procedures, and safety follow-up details are not reported. These limitations call for a narrow reading of the record as descriptive rather than as a complete account of efficacy and safety.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: i-Matter: Investigating an mHealth Texting Tool for Embedding Patient-reported Data Into Diabetes Management — NCT03652389.
• Registry record last updated 2025-02-19; status last verified 2025-01.
• Results first posted 2025-02-19.
• 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

i-Matter is recorded as a completed, randomized, single-masked, parallel-group study sponsored by NYU Langone Health. It compares MJS DIABETES, a text-message patient-reported outcome tool, with usual care for type 2 diabetes. The registry lists baseline HbA1c as the primary endpoint and provides model-adjusted means with 95% confidence intervals, using multiple imputation. Its brief summary describes a planned 12-month HbA1c reduction question, but the 12-month efficacy result is not reported in the registry record.

The record lists baseline self-care measures and zero serious adverse events and deaths in both groups. The secondary comparisons are unadjusted recalculations rather than reported analyses. The enrollment figure differs between the registry entry and study summary, with no explanation recorded. The available information supports descriptive interpretation of baseline outcomes, not a conclusion about the planned clinical efficacy question.