Reliability and Equivalence of Alternate Forms of the Digital Clock Drawing Test: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Reliability and Equivalence of Alternate Forms of the Digital Clock Drawing Test (NCT04496752), a randomized na study. The primary outcome is Equivalence of DCTclock-pen and DCTclock-tablet in Assessing Cognitive Impairment.

2. Clinical Question

This is a measurement-comparison question, not a comparison of treatments intended to change health outcomes. The two forms are alternative ways of administering the digital clock drawing test, and the reference for the agreement assessment is the MMSE. The randomized crossover structure means that participants were assigned to an order in which they received both forms; it does not make the comparison a parallel trial of separate participant groups. The registry record does not report whether the order was accounted for in the primary analysis, or whether any interval or other procedures were used between assessments.

The analysis population was narrower than the group that completed both sessions: it required a valid score from each form. Results therefore describe agreement in the participants whose paired clocks were analyzable, rather than necessarily describing every enrolled participant. Further details about the trial population and the reasons some completed assessments were not analyzable are not reported in the registry record.

3. Trial Design

The study is listed as a completed, randomized crossover trial of DCTclock-pen and DCTclock-tablet. Enrollment was 200. The registry lists the start date as 2020-12-18 and the primary completion date as 2021-11-16. Masking is listed as none, and the phase is listed as NA. Linus Health, Inc. is identified as the sponsor.

Participants were assigned to one of two sequences: DCTclock-pen followed by DCTclock-tablet, or DCTclock-tablet followed by DCTclock-pen. The central design feature is that each participant could be assessed using both forms, while randomization determined which form came first. This permits a within-participant comparison of the forms among people with usable results from both assessments. The registry does not report the randomization method, allocation concealment, the interval between assessments, or any procedures for addressing possible order or carryover effects.

The primary outcome was the equivalence of the two forms in assessing cognitive impairment, measured by comparison of agreement with MMSE using Cohen’s kappa. The stated timeframe was 5 weeks. The registry lists three outcome measures and one reported analysis. A secondary outcome concerns the number of participants with device-related and non-device-related adverse events.

The registry does not report the detailed protocol, inclusion and exclusion criteria, recruitment setting, or participant characteristics. It also does not describe the planned sample-size justification or the statistical plan in enough detail to determine how the crossover structure was handled analytically. Those omissions limit what can be inferred about the intended design and the scope of the posted findings.

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: Screening for inclusion criteria was conducted at the first study session after enrollment and prior to randomization. Of 200 enrolled participants, 195 met inclusion criteria and were randomized to groups.

First Assessment

Second Assessment

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analysis includes only assessments from participants who: (1) completed both assessment sessions; and (2) produced clocks from both assessments which could be analyzed to give a valid score. Of participants who completed both sessions, 170 of 187 produced two analyzable clocks and were included in analysis (84 of 92 in the pen-first group, and 86 of 95 in the tablet-first group).

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 Equivalence of DCTclock-pen and DCTclock-tablet in Assessing Cognitive Impairment, measured over 5 weeks.

The primary endpoint is Cohen’s kappa for each digital clock drawing form in its agreement with the MMSE, together with the reported difference between the forms’ kappa estimates. The registry frames the endpoint as equivalence in assessing cognitive impairment and gives a timeframe of 5 weeks. Kappa is a chance-adjusted measure of agreement for categorical classifications: it describes the observed agreement while accounting for the agreement that could occur by chance under the measure’s assumptions. It is not a direct measure of correlation, and it does not by itself establish that either assessment is accurate against a definitive diagnosis.

The registry reports a kappa of 0.16 for DCTclock-pen, with a 90% confidence interval from 0.05 to 0.27, and a kappa of 0.23 for DCTclock-tablet, with a 90% confidence interval from 0.12 to 0.33. The reported difference in Cohen’s kappa is 0.07, with a 90% confidence interval from -0.05 to 0.19. The interval expresses uncertainty in the estimated difference; it does not specify how large a difference would be acceptable for equivalence.

Equivalence is a distinct statistical claim from finding no evidence of a difference. In general, an equivalence assessment compares an uncertainty interval with a prespecified range of differences considered acceptably small. The registry record does not report that range, so the reported estimates alone do not establish whether the study’s criterion for equivalence was met. The clinical rationale for choosing the MMSE as the agreement reference, and the definitions used to classify cognitive impairment, are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does Cohen’s kappa measure?

In general, Cohen’s kappa quantifies agreement between classifications while adjusting for the agreement expected by chance under the measure’s assumptions. A kappa value is not itself the proportion of classifications that match. Its meaning depends on how categories are defined and how common each category is in the population being assessed. The registry identifies the comparison as agreement with MMSE but does not report the category definitions, underlying classification counts, or the kappa calculation details.

The posted kappa estimates are 0.16 for DCTclock-pen and 0.23 for DCTclock-tablet. Their 90% confidence intervals are 0.05 to 0.27 and 0.12 to 0.33, respectively. A confidence interval represents uncertainty in an estimate under the assumptions of the analysis. It does not show the range of individual participant results, and it does not guarantee that the unknown value falls inside the interval. These intervals also do not by themselves determine whether the forms are equivalent.

How should the difference be read?

The reported difference in Cohen’s kappa is 0.07, with a 90% confidence interval from -0.05 to 0.19. The interval describes uncertainty around the contrast between forms. Because it includes zero, the posted estimate does not rule out no difference in the direction of the comparison. The interval also extends to values on either side of zero, so the point estimate should not be treated as a precise description of the true difference.

Equivalence requires more than failing to establish a difference. In general, investigators define an equivalence margin before examining results. That margin sets the range of differences regarded as small enough for the intended purpose. An equivalence conclusion typically depends on whether the confidence interval is contained within the prespecified acceptable range. The registry identifies an equivalence objective but does not report the margin, the rationale for it, or the exact decision procedure. As a result, the equivalence objective cannot be fully evaluated from the posted comparison alone.

Why does the crossover structure matter?

In a crossover design, participants receive the conditions in an assigned order. Here, the sequences were pen first then tablet, or tablet first then pen. Since the same participants could contribute results for each form, the two measurements are related rather than independent. A suitable comparison generally needs to account for that pairing. In general, order, learning, or carryover can also matter when assessments are repeated. Whether any such effects were modeled, tested, or otherwise addressed is not reported in the registry record.

The primary analysis population required completed sessions and analyzable clocks from both assessments. The registry reports that 170 of 187 participants who completed both sessions met the analyzability condition. This paired analysis population is relevant to the comparison because a participant without a valid result from either form cannot contribute a complete pair in the same way. The record does not provide a full account of missing assessments, invalid clocks, or any sensitivity analyses for missing or unusable results.

What can be said about statistical planning?

The registry lists one reported analysis and does not provide a p-value for it. It also does not report the sample-size justification, the planned statistical model, or a complete analysis plan. General guidance on interpretation should not be mistaken for a description of procedures that the investigators actually used. For example, accounting for paired measurements is a general consideration for a crossover comparison; the specific approach used in this trial is not reported in the registry record.

Any sample-size planning illustration based on an observed effect is a hypothetical planning exercise, not observed power. In general, such an illustration asks what sample size might be required under stated assumptions if an effect of a chosen size were the target. The registry’s planning assumptions specify a two-sided significance level of 0.05, equal allocation, and a normal approximation, but the displayed planning result does not establish the study’s actual design rationale or prove that the trial had a particular probability of detecting an effect. The registry does not report that this was the sponsor’s original sample-size calculation.

What limits interpretation of the estimate?

Kappa depends on the classifications and their prevalence, so its magnitude should be interpreted in the context of the scoring and participant mix. The registry does not report enough detail to examine those features. The confidence intervals describe statistical uncertainty conditional on the analysis and its assumptions, but they do not account automatically for every possible source of bias, such as selection into the analyzable population or unaddressed sequence effects.

Over-dispersion is a general concern for count outcomes when variability exceeds what a simple model assumes. It is not the primary endpoint type in this trial, which reports kappa coefficients, and the registry does not identify an over-dispersion model for the primary analysis. No count-endpoint recalculation or sample-size result should be used to infer the method for the kappa comparison. The central interpretive constraints remain the missing equivalence margin, limited methodological detail, and selection of participants with valid paired assessments.

10. Analyses Reported by the Sponsor

The registry reports the comparison as an equivalence analysis: the difference in Cohen’s kappa is 0.07, with a 90% confidence interval from -0.05 to 0.19. No p-value is listed. This is the posted analysis result; the registry record does not provide a separate sponsor narrative explaining the result or the criterion used to declare equivalence.

The record does not include a separately identified recalculated estimate for this endpoint. Its discrepancy indicator is false, so no conflict between a sponsor result and a simple unadjusted re-calculation is flagged. The reported analysis therefore remains the result to describe, while its limitations are assessed from the information in the registry record. The equivalence margin, analysis method, and decision rule are not reported, so the confidence interval cannot be compared with a stated acceptance range.

A recalculation from summary statistics, when available, would be unadjusted and would not replace an analysis that accounts for the randomized crossover design and its planned assumptions. In this case, no alternative estimate or p-value is reported for comparison. The registry result should not be recast as proof of equivalence solely because it is labelled an equivalence analysis.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists three outcome measures, one reported analysis, and one posted secondary outcome. The primary question concerns equivalence between the two digital forms in agreement with MMSE. The secondary outcome concerns the number of participants with device-related and non-device-related adverse events. The record does not describe how many outcomes were prespecified, how they were prioritized in the protocol, or whether any multiplicity adjustment was planned or used.

Multiplicity arises when several hypotheses or outcomes are examined, because the chance of at least one apparently notable finding can increase across repeated tests. In general, a statistical plan may address this through a prespecified hierarchy, adjustment, or another strategy. The registry record does not report such a strategy. Since no p-value is posted for the primary comparison, there is no reported significance test to interpret as adjusted or unadjusted.

The primary analysis population was restricted to participants who completed both assessment sessions and had analyzable clocks from each form. The registry reports 170 of 187 participants who completed both sessions in this population. It does not provide details about the participants who completed sessions but lacked analyzable paired clocks, or about the reasons for their exclusion. Although enrollment was 200, the handling of enrolled participants who did not complete both sessions is not described in the primary outcome text.

The record does not report an intention-to-treat analysis, a missing-data method, imputation, sensitivity analyses, or analyses using a broader population. It also does not report whether the outcome classifications were assessed without knowledge of sequence or other study information. These details matter because excluding participants with unusable assessments can affect how well the analyzed group represents the enrolled group. No assumptions about the reasons for missing or invalid results should be made from the posted summary alone.

12. 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.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: cholelithiasis (1.0%), death (1.0%), dyspnea (1.0%).

The registry reports adverse events by sequence and period, including the pen visit, tablet visit, and between-visits period. It records no serious adverse events for the pen-first pen visit, tablet-first tablet visit, pen-first tablet visit, or tablet-first pen visit groups. In the between-visits period, the pen-first group has 3 serious adverse events among 97 at risk, while the tablet-first group has 0 serious adverse events among 98 at risk. The record also reports 1 death in the pen-first between-visits group and no deaths in the other listed groups.

The registry lists cholelithiasis, death, and dyspnea among the serious-event terms, each at 1.0%. These terms and percentages are the posted registry entries; they should not be treated as evidence that the device caused the events. The event summaries do not establish causality, and the record does not provide clinical narratives or a detailed attribution assessment for these events.

Because participants could receive both forms and the adverse-event entries are separated by visit and between-visits periods, the groups are not straightforward parallel treatment groups. The registry record does not report the exposure time, detailed event timing, or a statistical comparison of safety between forms. These entries are descriptive counts and should be read with the trial’s crossover structure and the limited attribution information in mind.

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

The results describe how closely each digital clock drawing form agreed with the MMSE in the participants included in the primary analysis. The reported kappa estimates are 0.16 for DCTclock-pen and 0.23 for DCTclock-tablet. The reported difference is 0.07, with a 90% confidence interval from -0.05 to 0.19. These estimates address agreement with the specified reference assessment; they do not establish that either form is accurate against an independent diagnostic standard.

The trial’s stated objective was equivalence, but the registry record does not report the margin that defines an acceptable difference. Without that margin, the result cannot be classified as meeting or failing a prespecified equivalence criterion from the posted information. An interval that includes zero is not by itself evidence that two methods are equivalent: it may also reflect uncertainty that permits differences in either direction.

The findings apply to the analyzable paired assessments, not automatically to every participant enrolled or to every setting in which the forms might be used. The registry does not report full participant characteristics, the scoring thresholds, or details of the analysis model. Those gaps limit assessment of whether the agreement estimates would carry over to populations with different characteristics or to different uses of the test.

The trial does not compare a treatment with a control for a clinical outcome, and its results do not show that a device improves cognition or patient outcomes. Nor do the posted kappa values establish that either form can replace the MMSE or a clinical evaluation. They are statistical estimates of agreement in a specific comparison, with uncertainty that should be considered alongside the absent equivalence margin and limited methodological detail.

14. Why This Trial Matters Statistically

This trial raises a useful statistical distinction: showing that two measurement forms differ little is not automatically the same as demonstrating equivalence. An equivalence claim needs a clinically or operationally justified margin, set out before interpreting the results, and a suitable analysis showing that uncertainty is contained within that margin. The registry identifies an equivalence objective and gives a confidence interval for the difference, but does not report the margin. The posted estimate can therefore be described, while the formal equivalence claim remains difficult to evaluate.

The crossover design also makes the analysis structure important. Each participant could complete both forms in randomized order, creating paired measurements. A comparison that respects this pairing can use information efficiently, while potential order effects require consideration in repeated assessments. The registry does not describe the model or any assessment of order effects, illustrating why outcome summaries alone do not provide a complete account of the statistical analysis.

Another important issue is the definition of the analysis population. The primary analysis was limited to participants with completed sessions and analyzable clocks from both forms. That restriction makes the reported comparison interpretable for paired valid assessments, but it also makes the handling and causes of non-analyzable results relevant. Further details are not reported in the registry record.

Finally, a kappa estimate is a measure of chance-adjusted agreement, not an overall verdict on measurement quality. Its interpretation depends on the categories and population being assessed. The registry’s results are most informative when read narrowly: as estimates of agreement with MMSE for the analyzable participants, accompanied by uncertainty and incomplete information about the equivalence criterion.

15. 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.

Several trial-specific limitations constrain interpretation. The registry identifies the study as an equivalence trial but does not report the equivalence margin or the rationale for choosing one. Without that criterion, the confidence interval for the difference cannot establish whether the result falls within an acceptable range. The registry also does not report the statistical method used for the comparison, the handling of the crossover structure, or whether sequence and order effects were evaluated.

The primary analysis was restricted to participants with analyzable results from both forms. The registry reports that 170 of 187 participants who completed both sessions produced two analyzable clocks. It does not explain why the remaining completed assessments were not analyzable, give detailed information about participants who did not complete both sessions, or report sensitivity analyses addressing those exclusions. The participant characteristics and eligibility criteria are also not reported in the registry record.

The registry does not provide the scoring definitions or classification thresholds used to assess agreement with MMSE, nor enough detail to assess how the reference assessment was applied. It reports kappa estimates and confidence intervals but no p-value for the comparison. The rationale for the stated 5 weeks timeframe and the timing of assessments within that period are not reported.

Safety information is listed by sequence and period, but the record does not provide event narratives, exposure-time information, or a detailed basis for attribution to the device. The crossover design and period-specific entries also limit simple between-group interpretation. These limitations are specific to the information available in the registry record and do not establish that the unreported procedures were or were not performed.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Reliability and Equivalence of Alternate Forms of the Digital Clock Drawing Test — NCT04496752.
• Registry record last updated 2025-03-07; status last verified 2025-03.
• Results first posted 2025-03-07.
• 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.

19. Record Summary

NCT04496752 is a completed randomized crossover study of DCTclock-pen and DCTclock-tablet. Enrollment was 200, and participants were assigned to pen-first or tablet-first sequences. The primary endpoint compared agreement with MMSE using Cohen’s kappa over a stated timeframe of 5 weeks.

The registry reports a kappa of 0.16 for the pen form and 0.23 for the tablet form. The reported difference is 0.07, with a 90% confidence interval from -0.05 to 0.19. The primary analysis included 170 participants with analyzable clocks from both assessments among 187 who completed both sessions. The registry does not report the equivalence margin or enough methodological detail to determine whether a prespecified equivalence criterion was met.

The posted safety entries include adverse events by visit and between-visits period. Their attribution and clinical context are not detailed in the registry record.