Intervention to Improve Driving Practices Among High-Risk Teen Drivers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Intervention to Improve Driving Practices Among High-Risk Teen Drivers (NCT04317664), a randomized phase 3 study. The primary outcome is Risky Driving Events.

2. Clinical Question

3. Trial Design

The registry record describes a Phase 3, randomized, unmasked, parallel-group trial. Its status is completed. Enrollment is listed as 240, and the three arm descriptions each list 80 participants. The record gives a start date of 2020-09-28 and a primary completion date of 2024-12-31.

All groups have an in-vehicle device installed. The control group's device has feedback features disabled. Teens in the Feedback Only Group receive access to driving data through a smartphone app, instructions on reviewing the data, and biweekly cumulative reports. In the Feedback and Parent Communication Group, teens and parents can review driving data, and the parent receives communication training. The registry description of that training is incomplete, so its content and delivery are not reported in the registry record.

The primary outcome is risky driving events, expressed as an average rate per 1,000 miles during the six-month/study period. Driving events are detected and recorded by the in-vehicle system. The record names recidivism and parent-teen communication as secondary outcomes and lists seven outcome measures. It does not give further details about the secondary-outcome definitions in the registry record for the primary comparison.

The posted arm summaries give mean rates and standard deviations. Comparisons are presented for Feedback Only versus Control and Feedback and Parent Communication versus Control. The registry record lists zero analyses, so the available comparison calculations should not be read as a detailed account of the sponsor's prespecified analysis. The record does not report the trial's original sample-size justification or statistical analysis plan.

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: Data listed below reflects number of parent-teen dyads.

Overall Study

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: Parents and teens are enrolled in a parent-teen dyad (80 dyads per group, totaling 160 participants per group).

6. Primary Endpoint

The registry lists the primary outcome as Risky Driving Events, measured over Six months/study period.

The primary endpoint is the average number of risky driving events per 1,000 miles driven during the six-month/study period. It is an event rate: the count of detected driving events is related to the distance driven, rather than being expressed only as a count per teen. Using distance as the exposure helps describe event frequency in relation to driving activity.

The in-vehicle system automatically detects and records events. The listed categories include hard braking at or below -0.45 g-force, sudden acceleration above 0.35 g-force, speeding more than 10 miles over the posted speed limit, and speed above 75 mph. The resulting endpoint combines events from these categories into a rate per 1,000 miles.

This measure is relevant to the study question because the intervention is intended to provide feedback about driving behavior, and the endpoint records driving events directly through the device. It does not, by itself, measure traffic violations or recidivism: those are separately named outcomes in the registry record. Nor does an event rate describe the severity or consequences of each event. The record does not report category-specific event rates in the registry record for the primary comparison.

7. Results Posted in the Registry

• Feedback Only Group: 106.20 versus 101.30 (Average number of events per 1000 miles) on control; rate ratio 1.05 (95% CI 0.40–2.74, approximate, Poisson model); p ≈ 0.923.
• Feedback and Parent Communication Group: 77.70 versus 101.30 (Average number of events per 1000 miles) on control; rate ratio 0.77 (95% CI 0.27–2.18, approximate, Poisson model); p ≈ 0.619.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is an event rate per unit of exposure. In general, an event-rate analysis relates the number of recorded events to the amount of time or activity during which events could occur. Here, the specified exposure is distance driven, and the rate unit is events per 1,000 miles. This makes the outcome different from a simple proportion of participants who experienced at least one event: a participant can contribute multiple detected events, and driving distance provides the denominator for the rate.

What does a rate ratio represent?

In general, a rate ratio compares the event rate in one group with the event rate in another. A value of 1 represents equal rates; a value above 1 represents a higher rate in the first group, and a value below 1 represents a lower rate. The posted comparison estimates are 1.048 for Feedback Only versus Control and 0.767 for Feedback and Parent Communication versus Control. These are unadjusted comparison calculations, not adjusted estimates reported with a detailed sponsor analysis in the registry record.

Why is a Poisson model used for event counts?

In general, a Poisson model is one way to analyze counts of events, often with an exposure term when participants contribute different amounts of observation time or activity. It treats the event count as arising from a process whose average frequency is related to exposure. A rate ratio can then express the relative event frequency between groups. The posted comparison calculations are identified as Poisson-based, and their exposure calculation uses an assumed one year because the endpoint timeframe was not parseable for that calculation.

Poisson methods often rely on a relationship between the mean and variance of the count process. In general, over-dispersion occurs when observed variability is greater than the model's Poisson assumption would allow. If it is present and not accommodated, uncertainty can be understated. The registry record gives standard deviations alongside the arm means, but does not report a dispersion assessment, model diagnostics, or an alternative model. The posted summaries alone do not establish whether over-dispersion occurred or how it was handled in any sponsor analysis.

How should the confidence intervals be read?

In general, a confidence interval describes uncertainty around an estimate under the method used to construct it. A narrow interval indicates more precision than a wide interval, all else equal. Here, each posted interval spans the rate-ratio reference value of 1 and includes values both below and above it. The interval for Feedback Only extends to 2.745; the interval for Feedback and Parent Communication extends to 2.181. The lower endpoints are below 1. This width means the estimates are compatible with a range of possible relative rates, rather than pinning down a precise contrast.

The posted unadjusted p-values are 0.923 and 0.619. In general, a p-value describes how compatible the observed data are with a specified null model, under the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not measure the size or practical importance of an effect. The registry record does not give the sponsor's statistical testing plan or state whether these comparisons were prespecified.

What does the record not establish about the analysis?

The registry record lists zero analyses. It does not report a full statistical analysis plan, adjustment variables, methods for missing data, an analysis population, or multiplicity handling. It also does not report whether the posted comparison calculations correspond to the sponsor's planned analysis. These omissions prevent a complete assessment of model specification and analysis conduct. General statistical interpretation can explain what a rate ratio, interval, or p-value means, but cannot fill in unreported design choices.

The posted comparison calculations use the arm event-rate summaries and a Poisson approach. The registry record does not provide participant-level counts, individual miles driven, or the distribution of exposure. Without those details, readers cannot independently assess the event-count model, variation in driving exposure, or the effect of participants contributing different amounts of mileage. The reported means and standard deviations remain useful descriptive summaries, but do not replace those missing details.

10. Confidence Intervals and Precision

The comparison estimates are not highly precise. The Feedback Only rate ratio is 1.048, with an interval that extends below 1 and reaches 2.745. The Feedback and Parent Communication rate ratio is 0.767, and its interval extends below 1 and reaches 2.181. Because each interval includes 1, the posted estimates do not distinguish clearly between equal rates and differences in either direction under the calculation used.

A confidence interval is a range of values compatible with the data and the method used to construct it. It is not a range containing a known true value with certainty. Wide intervals indicate that the estimate leaves substantial uncertainty about the size and direction of the comparison. The registry record does not specify the confidence level for these posted intervals, so a particular confidence percentage is not stated.

The arm summaries also show considerable spread: the standard deviations are 140.4 for Control, 108.6 for Feedback Only, and 89.2 for Feedback and Parent Communication, alongside the respective mean rates. These descriptive summaries reinforce that the average alone does not convey the variability among observed rates. The record does not give enough information to determine the contribution of mileage variation, participant-level clustering, or other sources of uncertainty.

The posted p-values, 0.923 and 0.619, are unadjusted. They do not resolve the imprecision reflected in the intervals, and the registry record does not report whether multiplicity adjustments or other inferential procedures were used by the sponsor.

11. 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 planning figures are participants per arm for 80% power and participants per arm for 90% power. They are not total event counts. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each planning scenario targets the stated fraction of the observed effect. The calculations are planning illustrations based on the observed effect, not observed power.

For Feedback Only versus Control, planning for 100% of the observed effect gives 67,845 participants per arm for 80% power and 90,825 participants per arm for 90% power. Planning for 75% of the observed effect gives 121,312 participants per arm for 80% power and 162,403 participants per arm for 90% power. Planning for 50% of the observed effect gives 274,545 participants per arm for 80% power and 367,537 participants per arm for 90% power.

For Feedback and Parent Communication versus Control, planning for 100% of the observed effect gives 2,538 participants per arm for 80% power and 3,397 participants per arm for 90% power. Planning for 75% of the observed effect gives 4,348 participants per arm for 80% power and 5,820 participants per arm for 90% power. Planning for 50% of the observed effect gives 9,437 participants per arm for 80% power and 12,633 participants per arm for 90% power.

In general, 80% power means a design has an 80% chance of detecting an effect of the specified size under the planning assumptions; 90% power means a higher chance under those assumptions. Smaller target effects require more participants. These calculations are not the trial's original design justification, which is not reported in the registry record. They are illustrations based on the posted observed effects and specified assumptions, and should not be treated as evidence that the completed trial had a particular power.

12. 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, λ₀ and λ₁ are event rates per patient-year, and T is follow-up per participant in years. This assumes Poisson counts and equal follow-up.

Worked calculation for this trial

1. Control rate λ₀ = 0.1013; treatment rate λ₁ = 0.1062 events per patient-year; follow-up T = 1.00 years.
2. n = (1.96 + 0.842)² × (1/0.1013 + 1/0.1062) / (1.00 × [ln(1.048)]²) = 67,844.5, rounded up to 67,845 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
lambda0 <- 0.1013                      # control rate, events per patient-year
lambda1 <- 0.1062                      # treatment rate
tt      <- 1                       # follow-up per participant, years
n_per_arm <- function(power, frac = 1) {
  l1 <- lambda0 * (lambda1 / lambda0)^frac   # smaller effect = fraction of the observed log rate ratio
  ceiling(round((z_a + qnorm(power))^2 * (1 / lambda0 + 1 / l1) / (tt * log(l1 / lambda0)^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 = ",")))

# Simulation 1: Poisson counts. Simulation 2: over-dispersed counts (variance = 1.5 x mean), analysed with the matching variance.
set.seed(4317664)
n    <- n_per_arm(0.80)
reps <- 20000
m0 <- lambda0 * n * tt; m1 <- lambda1 * n * tt
x0 <- rpois(reps, m0); x1 <- rpois(reps, m1)
z  <- log(x1 / x0) / sqrt(1 / x0 + 1 / x1)
cat(sprintf("Simulated power, Poisson counts, n = %s per arm: %.3f\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE)))
phi <- 1.5
y0 <- rnbinom(reps, mu = m0, size = m0 / (phi - 1)); y1 <- rnbinom(reps, mu = m1, size = m1 / (phi - 1))
z2 <- log(y1 / y0) / sqrt(phi * (1 / y0 + 1 / y1))
cat(sprintf("Simulated power, over-dispersed counts (phi = 1.5), same n: %.3f\n", mean(abs(z2) > z_a, na.rm = TRUE)))

Output from running the code

100% of observed effect: n per arm = 67,845 (80% power), 90,825 (90% power)
 75% of observed effect: n per arm = 121,312 (80% power), 162,403 (90% power)
 50% of observed effect: n per arm = 274,545 (80% power), 367,537 (90% power)
Simulated power, Poisson counts, n = 67,845 per arm: 0.803
Simulated power, over-dispersed counts (phi = 1.5), same n: 0.634

The first simulation draws Poisson counts and should land close to the target power. The second allows extra variability (variance 1.5 times the mean, a purely illustrative value), which lowers the power at the same sample size, so rate trials usually need more participants than the Poisson formula suggests.

13. Multiplicity, Missing Data and Analysis Populations

The registry record names several outcomes: risky driving events as the primary outcome, with recidivism and parent-teen communication as secondary outcomes. It lists seven outcome measures. The posted comparisons include two contrasts against the control group for the primary event-rate outcome. When multiple outcomes or comparisons are examined, the chance of at least one apparently unusual result can increase. In general, multiplicity methods are used to address that issue, but the registry record does not report whether any adjustment was planned or applied.

The comparison p-values are explicitly unadjusted calculations. They should therefore be understood as unadjusted rather than as results corrected for examining multiple arms or outcomes. The record does not report a prespecified hierarchy of outcomes, a familywise error approach, or another multiplicity strategy. It also lists zero analyses, so the sponsor's analysis plan and the relationship between the posted comparisons and any planned inferential analysis are not reported in the registry record.

Missing-data methods and analysis populations are also not reported in the registry record. The posted arm summaries identify 80 participants in each group, but do not state whether all randomized participants contributed data, whether an intention-to-treat or another population was used, or how participants with unavailable driving data were handled. The record does not report missing mileage, device non-use, withdrawal, or the amount of outcome data available for analysis.

These omissions matter because event-rate outcomes depend on both the event count and the exposure measure. In general, missing driving data or uneven observation could affect the amount of information available and the interpretation of rates. No direction or size of such an effect can be inferred from the information reported. The registry record also does not report adjustment covariates, sensitivity analyses, or imputation methods.

14. 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 non-serious adverse event terms

The adverse-event summaries list 80 participants at risk in each group. Serious adverse events affected 0 participants in Control, 0 in Feedback Only, and 0 in Feedback and Parent Communication. Other adverse events affected 3 participants in each group. Deaths are listed as 0 in all three groups.

These are counts in the posted registry safety record. A count of zero serious adverse events means none were recorded in these group summaries; it does not establish that an intervention has no possible risk. Likewise, the counts do not describe the nature, timing, duration, or severity of the other adverse events. The record does not list serious adverse-event terms, and further details about the other events are not included in the registry record reported for these summaries.

Safety counts should be interpreted alongside the trial's recorded follow-up and event definitions, which are not fully detailed in the registry record. They do not establish a causal difference between groups.

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

The posted results describe average risky-driving-event rates in the three randomized groups and provide unadjusted comparisons of each feedback group with control. The mean rate is 106.2 per 1,000 miles in Feedback Only, 77.7 in Feedback and Parent Communication, and 101.3 in Control. The comparison estimates are 1.048 and 0.767, respectively, and both confidence intervals include the rate-ratio reference value of 1. The posted p-values are 0.923 and 0.619.

These results do not establish that either feedback approach changed the event rate compared with control. The uncertainty intervals are broad, and the registry record does not report a sponsor analysis plan or a full account of model choices. The posted comparisons should not be interpreted as proof of no difference either: intervals that cross 1 can still be compatible with differences in either direction.

The endpoint reflects detected driving events per distance driven, not traffic violations, crashes, injuries, or all aspects of driving safety. Recidivism and parent-teen communication are listed as secondary outcomes, but the comparison figures concern only risky driving events. No conclusion about those secondary outcomes follows from these primary-outcome comparisons.

The results also do not establish how the interventions affect individual teens, how driving behavior changes beyond the recorded study period, or whether any particular event has a specific consequence. The registry record does not report category-specific event results, a detailed analysis population, missing-data handling, or a full statistical analysis. The appropriate interpretation is therefore limited to the descriptive rates and imprecise unadjusted comparisons posted for this outcome.

16. Why This Trial Matters Statistically

This trial raises a statistical challenge common to behavioral and technology-based studies: the outcome is a recurrent event rate tied to exposure, not simply whether a participant did or did not experience an event. Relating event counts to miles driven offers a way to account for different amounts of driving, but it also makes the exposure definition and the event-count model important to interpretation.

The study also distinguishes feedback alone from feedback paired with parent communication training, while retaining a control condition in which the device is installed but feedback features are disabled. That structure makes the comparisons relevant to questions about the added value of feedback and the combined intervention. The posted calculations, however, compare each active group with control; they do not provide a direct comparison between the two active groups.

The results illustrate why an estimate should not be read without its uncertainty. The mean rates and point estimates differ across groups, but the confidence intervals include the rate-ratio reference value of 1 and extend across a broad range. The posted standard deviations also describe substantial variability in the arm summaries. Together, these details make precision and modeling assumptions central to interpretation, rather than allowing a simple reading of the point estimates alone.

The record also illustrates the value of transparent analysis reporting. Information about the sponsor's analysis plan, missing data, analysis populations, and multiplicity is not reported in the registry record. Those details are needed to evaluate how the posted calculations relate to planned inference and how robust conclusions may be. The statistical interest is therefore not only the estimated rates, but also the limits of what the registry record allows readers to assess.

17. 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.
• Rate comparisons use a Poisson approximation unless the sponsor posted confidence limits; real event counts are often over-dispersed.

The registry record describes the primary endpoint timeframe as six months/study period, but the posted comparison calculations use an assumed one-year exposure because the timeframe was not parseable for those calculations. This is a specific limitation when relating the comparisons to the endpoint description. The record does not report whether the sponsor used the same assumption or a different exposure-time method.

The intervention descriptions are not equally complete. The description for Feedback and Parent Communication ends after stating that the parent receives communication training; the training details are not reported in the registry record. The record also does not provide a more detailed description of participant age, eligibility criteria, or the extent to which participants used the device and app.

The primary endpoint combines different detected driving events, and category-specific results are not reported in the registry record for the comparison. Event rates may also depend on how much each participant drove, but individual mileage distributions and exposure summaries are not reported. The posted arm standard deviations indicate variation in the reported rates, while the registry record does not report model diagnostics or whether a method addressing over-dispersion was considered.

Finally, the record lists zero analyses and does not report the sponsor's original sample-size justification, analysis populations, missing-data procedures, multiplicity methods, or adjustment covariates. The posted comparisons are unadjusted calculations. These gaps limit assessment of the intended analysis and how the reported summaries relate to it; they do not, by themselves, establish bias or a particular direction of effect.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Intervention to Improve Driving Practices Among High-Risk Teen Drivers — NCT04317664.
• Registry record last updated 2026-02-23; status last verified 2026-02.
• Results first posted 2026-02-23.
• 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.

21. Record Summary

NCT04317664, Intervention to Improve Driving Practices Among High-Risk Teen Drivers, is listed as a completed Phase 3 randomized, unmasked, parallel-group trial sponsored by Ginger Yang. The registry record lists enrollment of 240, a start date of 2020-09-28, and a primary completion date of 2024-12-31.

The primary outcome is risky driving events, measured as the average number of events per 1,000 miles over the six-month/study period. The three groups are Control, Feedback Only, and Feedback and Parent Communication. The posted mean rates are 101.3, 106.2, and 77.7, respectively. Unadjusted rate-ratio calculations versus control are 1.048 and 0.767, with intervals spanning 1.

Recidivism and parent-teen communication are listed as secondary outcomes. The record lists seven outcome measures and zero analyses. Details of the sponsor's statistical plan, missing-data handling, analysis populations, multiplicity methods, and original sample-size justification are not reported in the registry record.