FINGER Robot Therapy Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for FINGER Robot Therapy Study (NCT02048826), a randomized na study. The primary outcome is Motor and Strength Outcome Measure Using Box and Block Test Measured in Units of Blocks Placed.

2. Clinical Question

3. Trial Design

The FINGER Robot Therapy Study (NCT02048826) is recorded as a randomized, double-blind, parallel-group study. Its listed phase is NA, and its status is completed. The sponsor is the University of California, Irvine. The recorded start was 2013-11, and primary completion was 2015-05. Enrollment is recorded as 31. The record does not provide details about how the random sequence was generated, how allocation was concealed, or who was masked. The meaning and practical implementation of double masking are therefore not described in enough detail to assess.

The two study arms used FINGER I and FINGER II, both described as exercise with the FINGER robot for three weeks. Each arm description specifies a minimum of three days per week and one hour per day, with a different device setting. The record does not provide further details on the settings, adherence, or how exercise sessions were supervised. It also does not describe whether other care or rehabilitation was permitted or balanced between the arms.

For the primary outcome, the posted group summaries each report 15 participants. The registry enrollment figure is 31; the record does not explain how enrollment relates to the number included in these outcome summaries. It does not report participant flow, withdrawals, exclusions, or the analysis population. This means the results can be described using the reported group summaries, but the completeness of follow-up and the handling of participants without outcome data cannot be established from the registry record.

The study focused on motor learning and hand movement after stroke. Its primary endpoint was change in Box and Block score from baseline to one-month post therapy, with Action Research Arm Test and Fugl-Meyer scores listed as secondary outcomes over the same timeframe. No original sample-size justification or prespecified statistical analysis plan is 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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The number of blocks participants able to pick up and place in a box in 60 seconds at one month follow up minus the number of blocks participants able to pick up and place in a box in 60 seconds at baseline.

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 Motor and Strength Outcome Measure Using Box and Block Test Measured in Units of Blocks Placed, measured over From baseline to 1-month post therapy.

The primary endpoint is change in the Box and Block score from baseline to one-month post therapy. The test records how many blocks a participant can pick up and place in a box within 60 seconds. The registry states that higher scores indicate better arm and hand function. The outcome is expressed as a mean change in units of blocks.

Using change from baseline focuses the comparison on how scores moved over the study period, rather than only comparing scores at follow-up. A mean summarizes the average change in each group, while the standard deviation describes how much individual participants' changes varied around that group mean. These summaries are useful for describing the groups, but do not show the full distribution of individual outcomes or whether all participants had similar changes.

The endpoint was measured through one-month post therapy. It therefore describes an outcome at that recorded follow-up point and does not establish whether any observed change persisted beyond it. The registry does not report other follow-up timepoints or additional details about test administration, scoring procedures, or who performed the assessment. The endpoint is a functional measure; it is not, by itself, a complete account of recovery, daily activity, or other outcomes that may matter to participants.

7. Results Posted in the Registry

• FINGER II: mean 2.40 versus 2.30 on control; difference 0.10 (95% CI -2.89 to 3.09), standardized effect d = 0.02; z-test p = 0.948.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Motor and Strength Measure Using Action Research Arm Test Measured as a Score

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • FINGER II: mean 3.50 versus 2.50 on control; difference 1.00 (95% CI -2.11 to 4.11), standardized effect d = 0.23; z-test p = 0.528.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between FINGER II and the control. The interval gives the range of effects that remain compatible with the data.

Motor, Strength and Sensory Using Fugl-Meyer Test Measured as a Score

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • FINGER II: mean 1.80 versus 3.70 on control; difference -1.90 (95% CI -3.83 to 0.03), standardized effect d = -0.71; z-test p = 0.053.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between FINGER II and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes give additional views of motor function and strength over the same baseline-to-one-month-post-therapy timeframe. The Action Research Arm Test and Fugl-Meyer measure domains related to the primary motor and hand-function question, but they are distinct scales and should not be treated as interchangeable with the Box and Block test. Across the outcomes, the posted directions and sizes of the group differences are not uniform. That mixed pattern does not form a consistent secondary signal that clearly reinforces the primary comparison.

The secondary findings are best considered alongside, rather than in place of, the primary result. They can help describe whether differences appear across related measures, but they do not resolve the uncertainty in the primary estimate. Each outcome has its own scale, measurement properties, and variability. Without details on scoring, clinically meaningful thresholds, or an analysis plan, the numerical differences cannot be translated into a common measure of functional importance.

The registry lists two secondary outcomes, and the reported analyses list is empty. This means the record does not provide an account of planned or adjusted secondary analyses. The posted summary comparisons are unadjusted. When several outcomes are evaluated, the chance of observing an apparently striking result among them can increase. Adjusting for multiple comparisons or designating outcomes in advance can affect how strong the evidence should be considered, but the registry does not report whether such procedures were planned or used.

For these reasons, secondary results carry less weight than a clearly prespecified primary analysis when the analysis plan and multiplicity approach are unavailable. They may be useful as descriptive and hypothesis-generating evidence, but should not be used in isolation to claim that one setting is superior or to override the uncertainty in the primary endpoint. No secondary result should be treated as confirmation simply because it concerns a related aspect of motor function.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Motor and Strength Measure Using Action Research Arm Test Measured as a Score
• Motor, Strength and Sensory Using Fugl-Meyer Test Measured as a Score

9. Statistical Interpretation

10. Statistical Methods Explained

The posted outcomes are continuous measures summarized using means and standard deviations. A continuous measure can take values across a range rather than being limited to categories such as event/no event. For a continuous outcome, the mean describes the average score or change within a group. The standard deviation describes the spread of participant values around that average. Neither statistic alone reveals the shape of the data or whether a few unusually high or low values influenced the mean.

What does the mean difference represent?

The primary comparison is a difference between the FINGER II and FINGER I mean changes. The direction is defined by subtracting the FINGER I mean from the FINGER II mean. A positive estimate therefore means the reported mean change was higher in FINGER II; a negative estimate means it was lower. The estimate of 0.1 blocks is the observed difference in the posted group means. It is a point estimate, not a guarantee that the population difference is exactly that value.

The secondary comparisons use the same general form: a difference between group means. They summarize differences in changes in Action Research Arm Test and Fugl-Meyer scores. The registry does not report a more detailed statistical model, covariate adjustment, or a test of whether the results vary according to participant characteristics. The simple group comparisons should not be interpreted as adjusted estimates.

What does a confidence interval convey?

A confidence interval represents uncertainty around an estimated effect under the method used to construct it. Narrower intervals indicate greater precision; wider intervals indicate less precision. A confidence interval that includes zero is compatible with no average difference under the interval's statistical framework, as well as with effects on either side of zero. It does not prove there is no effect. The confidence level and exact interval-construction method are not reported in the registry record, so they cannot be independently characterized from the record alone.

The primary estimate is 0.1 blocks, with an interval from -2.887 to 3.087. Because this range crosses zero, the reported data do not distinguish clearly between a small advantage for either setting and little or no average difference. Interpretation should consider both the point estimate and the interval rather than relying on a p-value alone.

How should the p-values be read?

A p-value describes how unusual data at least as incompatible with a specified null hypothesis would be under the assumptions of the statistical test. It does not state the probability that the null hypothesis is true, and it does not measure the size or practical importance of a difference. The primary unadjusted p-value is 0.948. The registry does not report the test used or the planned threshold for judging results, so the p-value should be understood as a posted unadjusted statistic rather than as a complete account of the trial's inferential plan.

The word “unadjusted” matters because the record includes a primary outcome and secondary outcomes. When several outcomes are examined, some apparently notable results may arise by chance even when there is no underlying effect. Methods for addressing multiple comparisons, if any, are not reported. The p-values should therefore not be read as though each outcome were necessarily the only comparison made.

Why do standard deviations matter?

The standard deviations show that participant changes varied within both arms. Variation affects how precisely a group mean is estimated: greater spread generally makes it harder to estimate a small difference precisely when the group sizes are limited. This is a general statistical point, not evidence of a particular reason for variation in this trial. The registry does not provide participant-level data, distributions, or information that would explain why individual scores differed.

For continuous outcomes, some analyses rely on assumptions about the distribution of observations or the sampling distribution of the mean. Other approaches can be used when those assumptions are not suitable. The registry does not report which method was used, diagnostic checks, or whether any alternative analysis was conducted. It is therefore not possible to assess those assumptions from the posted summaries alone.

What can be inferred from the design?

Random assignment can support a causal comparison between arms when it is implemented appropriately and when follow-up and analysis preserve the randomized comparison. The registry labels the study randomized and double-blind, but does not describe the allocation process, masking procedures, adherence, or analysis population. Those missing details prevent a full assessment of how well the design protections were carried through.

The reported analysis is a comparison of two settings of the FINGER robot. It does not estimate the effect of robot training compared with no training, and it cannot establish the effect of the device relative to other rehabilitation approaches. The endpoint timeframe also restricts interpretation to baseline through one-month post therapy. Longer-term effects are not reported.

11. Confidence Intervals and Precision

Precision is reflected most directly in the interval around the estimated mean difference. For the primary outcome, the estimated difference is 0.1 blocks, and the reported interval ranges from -2.887 to 3.087. The interval includes zero and extends to values in both directions. It therefore indicates uncertainty about both the size and direction of the underlying difference. The point estimate alone would give an incomplete account of that uncertainty.

The interval is wide relative to the estimate. This means the posted results do not pin down the difference closely. A small difference near zero is compatible with the data, but so are larger differences that favor either setting within the interval. Whether any particular difference would matter in practice cannot be determined from these statistics alone; the registry does not report a threshold for a clinically important change.

Confidence intervals are useful because they show a range of estimates consistent with the data and the statistical method, rather than reducing uncertainty to a binary conclusion. However, the confidence level and interval method are not reported in the registry record. The interval should therefore be quoted as posted, without assigning an unreported confidence level or making claims about how it was constructed.

The primary group standard deviations, 3.7 and 4.6 blocks, show variation in individual changes. The reported outcome group sizes are 15 for each arm. Limited information about participant flow and missing measurements makes it difficult to judge whether the group summaries capture everyone enrolled. The registry does not report individual-level results or additional analyses that could clarify the stability of the estimate.

In practical terms, the primary result is not a precise estimate of a near-zero difference. It is an imprecise estimate centered near zero, with uncertainty spanning differences in both directions. The absence of a clearly separated interval does not demonstrate equivalence; doing so would require an appropriate equivalence framework and a prespecified margin, neither of which is reported.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed effect, not a description of the trial's original design calculation. For these continuous outcomes, the figures are participants per arm, not total event counts. The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set as the stated fraction of the observed effect. The registry record does not report the trial's original sample-size assumptions or justification.

The 80% power column describes the number of participants per arm that the planning calculation estimates would be needed to have an 80% chance of detecting the specified effect under those assumptions, if that effect were the true one. The 90% power column uses a higher target chance of detection and consequently gives a larger planning figure. These are conventional planning concepts, not guarantees about what a future study would find.

The rows that use 100%, 75%, and 50% of the observed effect illustrate how the required group size changes when the target effect is smaller. For the primary endpoint, the posted planning figures rise from 27,354 participants per arm for 80% power at 100% of the observed effect, to 48,629 at 75%, and 109,414 at 50%. The corresponding 90% figures are 36,619, 65,100, and 146,474. Smaller effects are harder to distinguish from random variation, so more participants are needed to detect them at the same power target.

The accompanying curve should be read as a visual representation of this relationship between the assumed effect and the planning sample size, not as evidence about participants enrolled or outcomes observed. These planning estimates are not observed power. They are illustrative calculations using the stated assumptions, and should not be treated as the study's actual design target or as a measure of the probability that its findings are correct.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 0.100; pooled standard deviation σ = 4.174.
2. n = 2 × (1.96 + 0.842)² × 4.174² / 0.100² = 27,353.3, rounded up to 27,354 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.1                      # observed difference in means
sigma <- 4.17432629294836                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(2048826)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 27,354 (80% power), 36,619 (90% power)
 75% of observed effect: n per arm = 48,629 (80% power), 65,100 (90% power)
 50% of observed effect: n per arm = 109,414 (80% power), 146,474 (90% power)
power.t.test: 27354.3 
Simulated power at n = 27,354 per arm: 0.803 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Multiplicity, Missing Data and Analysis Populations

The registry identifies a primary endpoint and two secondary outcomes, but does not report a statistical analysis plan or multiplicity procedure. It is therefore not reported in the registry record whether the study intended to control the chance of false-positive findings across outcomes, whether any hierarchy among secondary outcomes was planned, or how multiple comparisons would affect interpretation. The posted secondary comparisons are unadjusted, so their p-values do not account for the broader set of outcomes.

The record lists zero reported analyses and no reported secondary analyses. This describes what is documented in the registry; it does not establish that no analyses were performed. The analysis methods, any covariate adjustments, model diagnostics, and handling of repeated or incomplete measurements are not reported in the registry record. The method used to calculate the posted intervals and p-values is also not specified.

Missing data are an important uncertainty. Participant flow, reasons for missing measurements, and the number with complete follow-up are not reported. The registry does not state whether missing outcomes were imputed, excluded, or handled by another method. It also does not specify whether the primary results use an intention-to-treat population, a per-protocol population, or another analysis population. Consequently, the possible influence of missing data or exclusions cannot be assessed from the summary record.

The record gives an enrollment figure of 31 and reports outcome summaries for 15 participants in each arm, but it does not explain the relationship between these figures. No inference about withdrawals, exclusions, or the treatment of any remaining participant should be made from that discrepancy in counts alone. Further details on participant disposition and analysis populations would be needed to assess whether the reported comparison retained the protections of randomization.

These reporting gaps do not invalidate the posted numbers, but they limit how fully their statistical reliability can be evaluated. They also mean that conclusions about prespecification, completeness of follow-up, and control of false-positive results are not supported by the registry record.

15. 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 record lists no serious adverse events in either arm and no deaths. It also lists no other adverse events in either arm. The record gives 16 participants at risk for serious and other events in FINGER I and 15 in FINGER II; these are the denominators attached to the adverse-event entries and should be read as reported.

These figures are counts recorded in the registry, not a complete description of every possible safety outcome. The record does not provide event definitions, collection procedures, observation duration, severity categories beyond serious and other, or a narrative of events. It does not explain how the adverse-event denominators relate to the outcome group sizes. Zero recorded events in these categories should not be interpreted as proof that no adverse effects occurred or that the settings have identical safety profiles. The figures describe what the registry reports for the study arms, within the information and follow-up recorded.

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

The primary result describes average change in a functional test after a short program using one of two FINGER robot settings. The reported mean changes were close, and the estimated difference between settings was 0.1 blocks. The interval around that estimate includes zero and allows for differences in either direction. Taken together, the results do not show a clear separation between the two settings on the recorded primary endpoint, while leaving substantial uncertainty about the size and direction of any underlying difference.

This is not evidence that the settings are equivalent. Equivalence requires a defined margin for what would count as an acceptably small difference, together with an analysis designed to test that question. Such a margin and analysis are not reported. Nor does the comparison establish the effect of the FINGER robot relative to no robot training, usual care, or another rehabilitation approach, because both study arms used the device.

The outcome was measured from baseline to one-month post therapy. The results therefore do not establish what happened at longer follow-up. The registry also does not report enough information about participant flow, missing outcomes, baseline characteristics, adherence, or the analysis population to determine how those factors may affect interpretation. The posted summaries are averages and do not describe individual responses.

The record concerns people with cerebrovascular accident, but it does not give enough detail about participant characteristics to determine how broadly the findings apply. Results from this study should not be generalized automatically to different populations, treatment programs, settings, or device configurations. The posted numbers are descriptive statistical evidence from the recorded comparison, not medical advice and not a complete assessment of benefit or harm.

17. Why This Trial Matters Statistically

This trial illustrates an important distinction between a small point estimate and strong evidence that two approaches are alike. The observed primary mean difference is close to zero, but the interval is broad and includes values favoring either setting. Statistical interpretation must account for both the estimate and its uncertainty. A result centered near zero can still be too imprecise to rule out differences that may matter.

The study also shows why a comparison needs a clearly reported analysis plan. A primary outcome and secondary measures can tell related but non-identical stories. Without a documented approach to multiple outcomes, missing data, and analysis populations, readers cannot determine which comparisons were confirmatory or how the results were intended to be combined. Transparent reporting of these decisions is part of judging the evidence, not a technical detail separate from it.

Finally, the planning illustrations emphasize the relationship between effect size and study size. Planning to detect a smaller effect requires more participants under the stated assumptions. These calculations are not a retrospective score of the trial; they show how design choices and detectable effects are linked. The registry does not report the original design assumptions, so the illustrations should not be confused with the trial's initial rationale.

In this record, the most defensible statistical reading is restrained: the posted group means are similar, but the primary estimate is not precise enough to establish equivalence or exclude meaningful differences. That distinction is central to interpreting randomized trials with limited reporting and uncertainty around the estimated effect.

18. 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.

Specific limitations in interpreting this study arise from gaps in the registry description. Although the study is labeled randomized and double-blind, the record does not explain sequence generation, allocation concealment, who was masked, or how masking was maintained. It does not describe the two robot settings in enough technical detail to understand the exact contrast. The exercise schedules are specified, but adherence, supervision, and concurrent rehabilitation are not reported.

The registry does not provide baseline characteristics, participant flow, reasons for missing data, or the analysis population. The relation between the enrollment figure and the reported outcome group sizes is not explained. Without these details, it is difficult to judge balance between arms, completeness of follow-up, or whether the analysis preserved the randomized assignment.

The endpoint information is limited to mean changes and standard deviations. Individual data, distributions, scoring details, and a threshold for clinically important change are not reported. The record also does not report the confidence level or calculation method for the intervals, the statistical test used, model checks, an analysis plan, or a multiplicity approach. These gaps restrict independent assessment of the statistical methods and the role of the secondary outcomes.

Follow-up is reported through one-month post therapy. Longer-term outcomes are not reported, and the registry does not provide a comparison against no device use or an alternative treatment. Accordingly, the results address a narrow comparison of device settings at a specified follow-up point rather than the broader effectiveness of robot therapy.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: FINGER Robot Therapy Study — NCT02048826.
• Registry record last updated 2021-03-25; status last verified 2021-03.
• Results first posted 2021-03-25.
• 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.

22. Record Summary

NCT02048826, the FINGER Robot Therapy Study, is recorded as a completed, randomized, double-blind, parallel-group study sponsored by the University of California, Irvine. The listed population has cerebrovascular accident. The arms compared FINGER I and FINGER II, using first and second settings of the FINGER robot during a three-week exercise program.

The primary endpoint was change in Box and Block score from baseline to one-month post therapy. The reported mean difference between FINGER II and FINGER I was 0.1 blocks, with an interval from -2.887 to 3.087 and an unadjusted p-value of 0.948. Two secondary motor-function outcomes were also recorded. The registry does not report the original sample-size justification, statistical analysis plan, multiplicity method, missing-data handling, or analysis population.