Effects of Inhaled Cannabis on Driving Performance: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effects of Inhaled Cannabis on Driving Performance (NCT01620177), a randomized na study. The primary outcome is Driving Performance.

2. Clinical Question

3. Trial Design

The registry lists this as a randomized, triple-masked crossover study. Enrollment was 98, and the record describes the study status as completed. The listed study period began in 2012-07, with primary completion in 2014-03. The phase is recorded as NA. The sponsor is Gary R Gaffney.

The interventions were oral alcohol and placebo, and inhaled cannabis containing THC and placebo. The listed conditions combine 0% THC, 2.5-3.5% THC, or 6.0-7.5% THC with either 0.065 g/dL BAC or 0 g/dL BAC. These are condition labels in the registry; they do not establish the precise timing or sequence in which each participant received them.

A crossover design generally means participants receive more than one study condition, so comparisons may use information from the same people under different conditions. That is a general feature of the design, not a description of unreported details in this trial. The registry does not report the treatment sequence, period duration, washout interval, or whether the analysis accounted for period or carryover effects. It also does not report the basis for choosing the enrollment target or a sample-size justification.

The primary outcome was analyzed with a regression model combining data across subjects and conditions, because THC and BrAC levels varied. The registry notes that one subject who was an extreme outlier across driving-performance measures was excluded. Further details about the analysis population, participant flow, and the handling of incomplete observations are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: One subject who was an extreme outlier across driving performance measures was excluded. Due to the variability in THC and BrAC levels across subjects and conditions, a regression model was used which combined all of the data.

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: There was no baseline group, but rather all subjects who completed experienced the baseline dosing condition of placebo cannabis and placebo alcohol.

6. Primary Endpoint

The registry lists the primary outcome as Driving Performance, measured over Through entire drive, 0.5-1.3 hr post cannabis administration..

The primary endpoint was driving performance, measured as the standard deviation of lane position (SDLP) in centimeters. The timeframe was through the entire drive, 0.5-1.3 hr after cannabis administration. SDLP describes variation in a vehicle’s lateral position within its lane: a larger value represents greater lane-position variability. It is a continuous measure, rather than a count of crashes or a direct measure of every aspect of driving.

The registry reports regression coefficients relating SDLP to THC and BrAC concentrations. It defines a THC unit as 1 ng/ml and a BAC unit as 0.01% BAC; the coefficient description expresses the alcohol term in relation to BrAC. Thus, the coefficients describe modeled changes in SDLP per concentration unit, not mean SDLP values for each condition. The reported coefficients are 0.26 for THC, 0.42 for Alcohol - BrAC, and 0.0 for the THC x BrAC interaction. The posted model results do not include confidence intervals or p-values.

SDLP is used in driving research as a measurable indicator of lane-keeping performance. In general, a standardized continuous outcome can capture changes in driving behavior that may be difficult to summarize with a simple event count. It does not, on its own, describe the full safety consequences of a driving condition or establish what happened in real-world driving.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary result uses regression, a statistical method for describing how an outcome varies with one or more predictors. Here, the outcome is SDLP in centimeters, and the predictors are THC concentration, BrAC concentration, and their interaction. The registry describes the equation as SDLP = Intercept + Cthc x THC + Cbrac x BrAC, and also reports a THC x BrAC coefficient. The stated equation is a general description of the model structure; the posted interaction coefficient is 0.0.

What does a regression coefficient mean?

General explanation: a regression coefficient represents the model’s estimated change in the outcome associated with a one-unit increase in a predictor, while the other modeled terms are accounted for. Its interpretation depends on the units used. In this study, a unit for THC is 1 ng/ml and a unit for BAC is 0.01% BAC; the coefficient description refers to BrAC for the alcohol term. The outcome is SDLP in centimeters. The THC coefficient is 0.26 and the BrAC coefficient is 0.42. These are slopes, not the observed SDLP value for a participant or the mean for an exposure group.

A regression intercept is the model’s baseline value when predictors take their reference values. The registry gives the form of the equation but does not report the intercept in the posted primary results. No interpretation of an unreported intercept or prediction at a specific concentration can therefore be made.

Why include an interaction term?

General explanation: an interaction term allows the association between one predictor and the outcome to vary depending on another predictor. For example, the THC-by-BrAC term represents whether the modeled relationship between THC and SDLP changes with BrAC concentration. The posted interaction coefficient is 0.0. This is the fitted estimate as recorded; it should not be treated as proof that the relationship is exactly absent. Precision and uncertainty cannot be assessed from that coefficient alone.

How should the model selection be understood?

The registry says metrics of driving performance were modeled using the SAS GLM Select function to identify changes in driver performance. It describes the resulting values as regression-equation coefficients. The registry does not report the candidate terms considered, the selection criteria, whether selection was prespecified, or validation procedures. Those details can affect how stable selected model estimates are, so the record does not permit a fuller assessment of the model-building process.

General explanation: when predictors are selected using the same data used to estimate their effects, estimates can be sensitive to the data and selection procedure. Independent validation or other safeguards may help assess reproducibility, but whether any such steps were used in this study is not reported in the registry record. No assumptions should be made about them.

What does combining observations across conditions imply?

The registry says the regression model combined all of the data because THC and BrAC levels varied across subjects and conditions. This approach treats concentration levels as predictors rather than relying only on condition labels. General explanation: modeling measured exposure levels can represent variation within nominal conditions, but it depends on the quality and range of those measurements and on the model’s assumptions. The registry lists secondary outcomes for THC concentration in plasma and THC concentration levels in whole blood, but does not give their results in the primary outcome record.

Because the trial is crossover, measurements may be associated within participants. General explanation: repeated observations from the same participant are not necessarily statistically independent, and an analysis may need to account for that dependence. The posted description does not state whether or how the regression accounted for repeated measurements, participant-level clustering, trial periods, or treatment sequence. This is an uncertainty about the reported methods, not evidence that a particular method was or was not used.

How should precision be judged?

A confidence interval is a general statistical way to express uncertainty around an estimate. A narrower interval indicates greater precision under the method used, while a wider interval indicates less precision. The registry record does not report confidence intervals for these coefficients. It also does not report p-values or a standard error for the primary coefficients. As a result, the numerical estimates can be described, but their uncertainty cannot be quantified from the posted results.

The reported sample size for each of the three coefficients is 18. Those are the outcome-level sample sizes attached to the coefficients; they should not be confused with overall enrollment or treated as counts for the six listed exposure combinations. The registry does not explain the difference between these figures. The planning illustrations, where presented, are planning exercises based on an observed effect and assumptions, not observed power and not a substitute for the trial’s own design justification. The registry’s sample-size justification is not reported in the registry record.

Why does over-dispersion matter?

General explanation: over-dispersion occurs when observed variability is greater than a statistical model expects. If it is present and not accounted for, standard errors and confidence intervals can be misleading. The primary endpoint is continuous and is reported through regression coefficients; the registry does not describe an over-dispersion assessment or identify over-dispersion as an issue. Count-outcome methods should not be assumed to apply to this continuous endpoint.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists three outcome measures and two secondary outcomes: THC concentration in plasma and THC concentration levels in whole blood. The primary outcome is driving performance. The record does not report a plan for controlling multiplicity across outcomes, model terms, or possible comparisons. Any adjustment for multiple testing is not reported in the registry record.

The primary result is a regression model with concentration terms and an interaction, rather than a collection of posted pairwise comparisons. The registry states that values are missing or not estimable for a two-arm comparison for this endpoint. It does not give a set of inferential results for comparisons among the listed exposure combinations. The analysis count is recorded as 0; the record does not explain further how that field relates to the regression coefficients.

For the primary analysis, one subject described as an extreme outlier across driving-performance measures was excluded. The registry does not report other exclusion rules, whether the exclusion was prespecified, or sensitivity analyses including that subject. It does not describe missing-data methods, the amount or pattern of missing primary-outcome data, or whether all randomized participants contributed to the reported model.

The outcome-level sample size attached to each coefficient is 18, while enrollment is listed as 98. The record does not provide a participant flow diagram or explain these figures’ relationship. The analysis population and its relationship to randomization are therefore not fully described. An interim-analysis plan, a full statistical analysis plan, and the sample-size justification are not reported in the registry record.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

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 registry reports serious adverse events as 0 in each of the six listed exposure groups. It lists the numbers at risk for those groups as 29, 27, 24, 26, 31, and 27, respectively. Death counts are not reported in the registry record.

Other, non-serious adverse events are reported as 4 for 0% THC With 0.065 g/dL BAC, 4 for 2.5-3.5% THC With 0.065 g/dL BAC, 6 for 6.0-7.5% THC and 0.065 g/dL BAC, 0 for 2.5-3.5% THC With 0 g/dL BAC, 3 for 6.0-7.5% THC With 0 g/dL BAC, and 0 for 0% THC With 0 g/dL BAC. The numbers at risk are listed separately for each group in the registry.

These counts describe the registry’s adverse-event entries, not a complete assessment of risk or a comparison adjusted for the crossover design. The record does not provide event descriptions, severity details for the other events, or a statistical comparison between conditions. Zero reported serious events should be read as a count in the posted record, not as evidence that serious events are impossible or that a condition is risk-free.

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

The results describe a model relating measured THC and BrAC concentrations to SDLP during the drive. The reported coefficients are 0.26 for THC, 0.42 for Alcohol - BrAC, and 0.0 for THC x BrAC. They are concentration-based regression estimates, expressed in relation to the units described in the registry, rather than direct treatment-group differences.

The model’s positive THC and BrAC coefficient estimates indicate positive modeled associations with SDLP. The posted interaction estimate is 0.0. The record does not include confidence intervals or p-values, so the precision and inferential strength of these estimates cannot be judged from the posted results. A coefficient’s numerical size alone does not establish that an association is statistically reliable or practically important.

The study used randomized crossover conditions, but the primary result combines data across subjects and conditions because exposure levels varied. It does not report a direct estimate for each of the listed exposure combinations. The registry also notes that values are missing or not estimable for a two-arm comparison. The coefficients therefore should not be presented as a head-to-head effect of one condition versus another.

SDLP measures lane-position variability, not every component of driving performance, nor real-world crash outcomes. The result does not establish what would happen in circumstances beyond the study’s driving environment, nor does it establish the effect of a particular exposure on every individual. One extreme outlier was excluded from the primary analysis, and the record does not report sensitivity analyses that would show how that decision affects the estimates.

The adverse-event entries are counts by listed condition. They do not supply a complete comparative safety analysis. Overall, the registry supports a description of the fitted concentration-related coefficients and recorded event counts, but not a definitive claim about the size, precision, or generalizability of a causal effect.

13. Why This Trial Matters Statistically

This trial is statistically notable because the primary result is expressed as a regression of driving performance on measured concentrations, rather than as a simple contrast between two groups. In a crossover study with several THC and alcohol conditions, this distinction affects what the estimates answer: the coefficients describe modeled concentration relationships across observations, while a condition comparison would answer a different question.

The registry says exposure levels varied across subjects and conditions, motivating a model that combined the data. That choice makes the units and predictor definitions essential to interpretation. A coefficient per concentration unit is not interchangeable with an average difference between conditions. Readers need the model scale and concentration units to understand what each estimate represents.

The record also illustrates why estimates require uncertainty measures. The coefficients give point estimates, but confidence intervals and p-values are not reported. Without them, the posted values alone cannot show how precisely the relationships were estimated. Likewise, the interaction coefficient is informative as a model term, but a value reported as 0.0 does not establish that interaction is absent.

Finally, the reporting leaves important questions open about the crossover analysis, including how repeated observations and trial periods were handled. The registry’s note that one extreme outlier was excluded is also relevant, while the impact of that exclusion is not quantified. These are not merely technical details: they determine how confidently the estimates can be interpreted and how well they may apply beyond the analyzed data.

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 primary result is reported as coefficients without confidence intervals, p-values, standard errors, or other uncertainty measures. This prevents a direct appraisal of precision. The record gives no intercept and no condition-specific estimates, and it does not report inferential comparisons among the six listed exposure combinations.

The regression combined data across subjects and conditions because THC and BrAC varied. The registry does not describe how within-participant dependence in the crossover design was handled, or whether period, sequence, or carryover effects were considered. The treatment order, period structure, and washout procedures are not reported in the registry record.

One subject described as an extreme outlier across driving-performance measures was excluded. The record does not state whether this rule was set in advance, give sensitivity results with the subject included, or describe other exclusions. It also does not report the amount or handling of missing data, participant flow, or the full analysis population.

The regression model used SAS GLM Select, but details of candidate variables, selection criteria, and validation are not reported. The record also does not state whether multiplicity was addressed across outcomes or model terms. The number of outcome measures is listed as 3, with two secondary outcomes; results for those secondary measures are not included in the primary result.

Finally, SDLP is one driving-performance measure during a specified drive timeframe. It does not capture all driving outcomes, and the registry does not provide enough methodological detail to assess how directly these findings generalize beyond the study conditions.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Effects of Inhaled Cannabis on Driving Performance — NCT01620177.
• Registry record last updated 2018-05-02; status last verified 2018-04.
• Results first posted 2017-06-01.
• 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

NCT01620177, titled Effects of Inhaled Cannabis on Driving Performance, is recorded as a completed randomized, triple-masked crossover study with enrollment of 98. The registry lists oral alcohol and placebo and inhaled cannabis containing THC and placebo as interventions. The primary outcome is SDLP, measured through the entire drive, 0.5-1.3 hr after cannabis administration.

The primary result is a regression model combining data across subjects and conditions. It reports coefficients of 0.26 for THC, 0.42 for Alcohol - BrAC, and 0.0 for THC x BrAC. One subject identified as an extreme outlier across driving-performance measures was excluded. Confidence intervals and p-values are not reported in the registry record.

The registry lists two secondary outcomes concerning THC concentrations in plasma and whole blood. Serious adverse events are recorded as 0 in each listed group; death counts are not reported in the registry record. Direct two-arm comparison values for the primary endpoint are missing or not estimable.