Rapid Identification and Phenotypic Susceptibility Testing for Gram-Negative Bacteremia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Rapid Identification and Phenotypic Susceptibility Testing for Gram-Negative Bacteremia (NCT03218397), a randomized na study. The primary outcome is Hours to First Antibiotic Modification.

2. Clinical Question

3. Trial Design

RAPIDS-GN is described in the registry record as a multi-center, prospective, randomized, controlled trial with a parallel-group model and single masking. The record does not identify who was masked. The listed phase is NA. Duke University is the sponsor, and the status is completed.

The two assigned strategies were standard blood culture and antimicrobial susceptibility testing (AST), or rapid organism identification and AST using the Accelerate PhenoTest™ BC Kit on the Accelerate Pheno™ System (AXDX). Antimicrobial stewardship was included in both arms. The rapid-testing arm also received standard culture and AST, so the intervention was an added rapid-testing strategy rather than a replacement of standard testing.

The registry record lists enrollment as 500. For the primary endpoint, the posted arm summaries include 226 subjects in the standard-testing group and 222 in the rapid-testing group. The primary outcome description restricts the population to subjects who completed the study; the record does not give further detail about how that population was defined or how participants without an endpoint value were handled.

The recorded study start was 2017-10-09, and primary completion was 2018-11-03. The listed secondary outcomes include mortality within 30 days, hospital length of stay, ICU status through 72 hours, and times to antibiotic escalation or de-escalation. The registry record does not report the original sample-size justification, the masking procedures, or other protocol details beyond those stated above.

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: Subjects who completed the 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: Subjects who completed the study.

6. Primary Endpoint

The registry lists the primary outcome as Hours to First Antibiotic Modification, measured over 72 hours after randomization.

The primary endpoint is mean hours from randomization to the first modification of antibiotic therapy, assessed within 72 hours after randomization. The registry describes the endpoint as applying to subjects who completed the study. A modification is the event being timed; the measure does not, by itself, distinguish the clinical reason for that modification or establish whether a change was beneficial.

Using time to an event can capture how quickly clinicians alter therapy after bacteremia is confirmed. Reporting the mean gives a straightforward summary of the observed timing in each arm, while the standard deviation describes how much individual timings varied around that mean. Because the endpoint is restricted to a 72-hour window and to the stated study-completer population, it should be interpreted within those bounds rather than as a measure of all later treatment decisions.

The registry does not report additional operational details about what qualified as a modification, how repeated modifications were treated, or how participants without a modification during the window were handled. Those details matter to the endpoint’s exact interpretation and are not inferable from the arm means alone.

7. Results Posted in the Registry

• Rapid Organism Identification and AST: mean 19.00 versus 24.70 on control; difference -5.70 (95% CI -10.10 to -1.30), standardized effect d = -0.24; z-test p = 0.011.

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.

Length of Stay in the Hospital

• 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:
  • Rapid Organism Identification and AST: mean 9.80 versus 8.20 on control; difference 1.60 (95% CI -0.21 to 3.41), standardized effect d = 0.17; z-test p = 0.083.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Rapid Organism Identification and AST and the control. The interval gives the range of effects that remain compatible with the data.

Time to First Antibiotic Escalation

• 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:
  • Rapid Organism Identification and AST: mean 45.20 versus 52.70 on control; difference -7.50 (95% CI -13.08 to -1.92), standardized effect d = -0.25; z-test p = 0.008.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Rapid Organism Identification and AST than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Time to First Gram-negative Antibiotic Escalation

• 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:
  • Rapid Organism Identification and AST: mean 47.50 versus 53.90 on control; difference -6.40 (95% CI -11.93 to -0.87), standardized effect d = -0.21; z-test p = 0.023.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Rapid Organism Identification and AST than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary results span outcomes related to mortality, hospital stay, ICU status, and antibiotic escalation or de-escalation. Across the secondary comparisons with numerical results available for re-calculation, the pattern is not uniform: escalation timings tend to be shorter in the rapid-testing arm, while the length-of-stay estimate is higher in that arm. The length-of-stay re-calculated interval includes no difference, whereas the escalation comparisons’ re-calculated intervals do not. These observations are secondary to the primary comparison and should not be treated as a single overall treatment effect.

The escalation pattern is directionally compatible with the primary finding of a shorter mean time to first antibiotic modification in the rapid-testing group. However, the registry record does not establish that the primary endpoint and the separate escalation outcomes measure the same event or use identical analysis rules. Secondary outcomes can clarify which aspects of care may differ, but they do not by themselves explain the primary result.

Unadjusted secondary p-values carry less weight when many outcomes are examined. General statistical context: testing multiple hypotheses increases the chance that at least one result appears unusual by chance, even when no corresponding effect exists. The registry record does not report a multiplicity adjustment or a prespecified hierarchy governing interpretation of the secondary endpoints. Accordingly, the secondary findings are best regarded as exploratory signals rather than definitive confirmations.

The registry lists additional secondary outcomes, but the available numerical summaries do not support a complete interpretation of every listed outcome. In particular, the record does not report enough detail to establish the direction and precision of every secondary comparison. Absence of a numerical result in the record should not be read as evidence of no difference.

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

• Subjects Who Experienced Mortality Within 30 Days of Randomization
• Length of Stay in the Hospital
• ICU Status Through 72 Hours Post-randomization
• Time to First Antibiotic Escalation
• Time to First Gram-negative Antibiotic Escalation
• Time to First Gram-positive Antibiotic Escalation
• Time to First Antibiotic De-escalation
• Time to First Gram-negative Antibiotic De-escalation

9. Statistical Interpretation

10. Statistical Methods Explained

What does a mean difference describe?

A mean is the arithmetic average of the measured values in an arm. For this endpoint, each value is time in hours until the first antibiotic modification within the specified window. A difference in means compares the two arm averages. Its sign depends on the order of subtraction: the sponsor’s reported positive difference is consistent with standard testing minus rapid testing, whereas the simple re-calculation expresses rapid testing minus standard testing and is negative. These are opposite sign conventions for the same direction of contrast, not opposite findings.

The standard deviations describe the spread of individual observations around each arm’s mean. The primary arm standard deviations are 24.6 hours for standard testing and 22.9 hours for rapid testing. General statistical context: a mean difference summarizes the center of each group, but it does not show the distribution of individual times or explain why an individual’s therapy changed.

What does the t-test assess?

The sponsor reports a two-sided t-test for the difference in means. In general, a t-test evaluates how compatible an observed mean difference is with a null hypothesis of no difference, under assumptions appropriate to the test. A two-sided test allows evidence in either direction. The posted sponsor p-value is 0.013. A p-value is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of a difference.

The simple re-calculation uses the posted arm sizes, means, and standard deviations to make an unadjusted comparison. Its p-value is 0.011. This calculation does not incorporate individual-level data or any adjustment beyond the summary-statistic comparison. It should be distinguished from the sponsor’s trial analysis, which is reported separately in the registry record.

How should the confidence interval be read?

A confidence interval describes uncertainty around an estimated difference. In general, an interval that is narrower indicates greater precision, while a wider interval leaves a broader range of values compatible with the data and method. The sponsor’s 95% interval runs from 1.2 to 10 hours under its contrast orientation. The re-calculated 95% interval runs from -10.1 to -1.3 hours under the reverse orientation. The intervals express the same direction after accounting for subtraction order.

A 95% confidence interval is not a statement that there is a 95% probability that this particular interval contains the true effect. In repeated use of the same method under its assumptions, the stated confidence level describes the long-run coverage of the procedure. The interval also does not account for uncertainty from choices not reflected in the calculation, such as unreported missing-data procedures or multiplicity adjustments.

Why does the analysis population matter?

The primary outcome description says “Subjects who completed the study.” That wording identifies the stated population but does not fully define the analysis set. The registry record does not report whether the primary analysis followed intention-to-treat assignment, how completion was determined, how participants with missing endpoint data were treated, or how timing was assigned when no modification occurred within the observation window. These details can affect both the estimate and its uncertainty.

General statistical context: excluding participants with incomplete follow-up can affect a comparison if completion is related to treatment assignment or to the outcome. This is a reason to look for an explicit participant-flow account and a missing-data plan. Such details are not reported in the registry record, so no conclusion about their impact can be drawn from the summaries alone.

Why can over-dispersion matter for count outcomes?

Over-dispersion occurs when the variability in count data is greater than a simple model assumes. General statistical context: if present and ignored, it can make standard errors too small and confidence intervals too narrow. The primary outcome is a continuous time measure analyzed as a mean, not a count endpoint, so this concept does not directly explain the primary result. The registry record does not give enough model detail to assess how count-type secondary outcomes, if any, were analyzed.

11. Confidence Intervals and Precision

The sponsor reports a 95% confidence interval of 1.2 to 10 hours for the primary difference in means, using a contrast consistent with standard testing minus rapid testing. The simple re-calculation reports -10.1 to -1.3 hours for rapid testing minus standard testing. Since the signs reverse with the contrast order, the intervals tell a consistent directional story. Neither interval includes zero under its stated orientation, but the intervals still allow a range of plausible difference sizes.

The reported standard deviations, 24.6 hours and 22.9 hours, indicate substantial variation in individual times relative to the arm means. General statistical context: individual variability and the number of observations both contribute to the precision of an estimated mean difference. A confidence interval represents uncertainty about the group-level contrast, not a range that contains most individual participants’ times.

The sponsor interval and re-calculated interval are close but not identical. The registry record does not flag a discrepancy. The sponsor’s interval is the result of the reported trial analysis; the re-calculated interval is an unadjusted calculation from the posted summaries and may differ because it is based on rounded aggregate information and a simple method. Neither interval addresses uncertainty arising from analysis choices that the registry record does not describe.

Precision should not be confused with clinical importance. The registry record does not state a threshold for a meaningful difference in hours. It also does not report a detailed missing-data analysis or an adjustment for examining multiple outcomes. Those omissions limit how fully the interval can characterize uncertainty in the broader trial question.

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 planning illustrations are based on the observed effect, not on the trial’s original design assumptions. The stated 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 at the stated fraction of the observed effect. For these continuous outcomes, the listed figures are participants per arm, not total events. The record’s general note says time-to-event planning entries, when present, give total events; that is not the format for this mean-based primary endpoint.

The 80% power column describes the number of participants per arm that the stated planning method estimates would be needed to have an 80% chance of detecting the specified effect under its assumptions. The 90% power column gives the corresponding planning estimate for a higher chance of detecting that same effect. These are design-planning quantities, not guarantees about any future study.

For the primary endpoint, planning for 100% of the observed effect gives 274 participants per arm for 80% power and 366 participants per arm for 90% power. Planning for 75% of the observed effect gives 486 participants per arm for 80% power and 650 participants per arm for 90% power. Planning for 50% of the observed effect gives 1,093 participants per arm for 80% power and 1,463 participants per arm for 90% power. Smaller effects require more participants to detect under the same planning assumptions.

These figures are not observed power. They are illustrations calculated from the observed effect and stated assumptions, not evidence that the completed trial had a particular probability of detecting its result. The registry record does not report the trial’s original sample-size justification or the assumptions used when its enrollment target was set.

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 = -5.700; pooled standard deviation σ = 23.773.
2. n = 2 × (1.96 + 0.842)² × 23.773² / -5.700² = 273.1, rounded up to 274 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 <- -5.7                      # observed difference in means
sigma <- 23.7728228787268                     # 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(3218397)
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 = 274 (80% power), 366 (90% power)
 75% of observed effect: n per arm = 486 (80% power), 650 (90% power)
 50% of observed effect: n per arm = 1,093 (80% power), 1,463 (90% power)
power.t.test: 274 
Simulated power at n = 274 per arm: 0.798 (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. Analyses Reported by the Sponsor

The sponsor’s posted primary analysis is a two-sided t-test. It reports a difference in means of 5.6 hours, a 95% confidence interval from 1.2 to 10, and p = 0.013. The sign of that estimate is consistent with subtracting the rapid-testing mean from the standard-testing mean. The posted arm means are 24.7 hours for standard testing and 19.0 hours for rapid testing.

The simple unadjusted re-calculation instead reports rapid testing minus standard testing: -5.7 hours, with a 95% confidence interval from -10.1 to -1.3 and p = 0.011. The reversed sign reflects the reversed subtraction order. Both analyses indicate a lower mean time in the rapid-testing arm. The registry record does not flag a discrepancy between the posted and re-calculated results.

The sponsor result is the trial analysis recorded in the registry; the re-calculation is a transparent comparison based on the available group summaries. Small differences in the estimates, intervals, and p-values are possible when calculations use rounded summary values rather than the underlying participant-level analysis. The registry record does not provide enough detail to reconcile the small numerical differences further. Neither analysis alone addresses unreported analysis-population or missing-data decisions.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary endpoint and multiple secondary outcomes, and it reports results for multiple analyses. The primary endpoint has a stated time window and population description, but the record does not report a detailed prespecified statistical analysis plan. It also does not report whether any interim analyses were planned or conducted. Those matters cannot be inferred from the presence of posted results.

The sponsor’s primary analysis is identified as a two-sided t-test. The registry record does not state whether the primary comparison was adjusted for any covariates, whether any alternative analyses were prespecified, or whether assumptions were checked. It does not describe a multiplicity strategy for the secondary outcomes. General statistical context: when many hypotheses are tested, interpreting each unadjusted p-value as if it were the sole test can overstate the strength of the overall evidence.

The primary population is described as subjects who completed the study. The record does not provide a detailed account of the analysis population, participant flow by arm, reasons for non-completion, or how missing outcome values were handled. It also does not explain how participants without a first modification during the 72-hour window were treated. These omissions limit assessment of possible selection or missing-data effects.

The posted arm sample sizes differ across outcomes, but the registry record does not explain all reasons for those differences. No missing-data method, sensitivity analysis, or imputation procedure is reported in the record. Therefore, it is not possible to determine from the posted summaries whether incomplete observations could have influenced the comparisons or how robust the findings are to alternative assumptions.

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

The primary result describes a difference in average time to first antibiotic modification within the specified 72-hour period among the stated study-completer population. The posted means are lower in the rapid-testing arm, and the sponsor’s confidence interval and p-value indicate evidence of a difference under its reported analysis. The unadjusted re-calculation reaches the same directional interpretation with the contrast expressed in reverse order.

This finding does not establish that rapid testing improved survival, reduced hospital stay, or improved the appropriateness of antibiotic choices. Those are distinct questions. The trial includes secondary outcomes related to some of them, but secondary comparisons have their own uncertainty and the record does not report an adjustment for multiple testing. The primary endpoint is a measure of treatment timing, not a direct measure of patient benefit.

The intervention was a testing strategy used alongside stewardship, and the rapid-testing group also received standard culture and AST. The results therefore concern the comparison of those strategies as implemented in the trial, not rapid testing in isolation or a replacement of standard testing.

The interpretation is bounded by the endpoint window and the described study-completer population. The registry record does not report detailed missing-data methods, complete analysis-population rules, or the protocol’s original sample-size justification. It also does not state a threshold for a clinically meaningful change in time. These limitations leave questions about generalizability and clinical importance that cannot be settled by the mean difference and its confidence interval alone.

17. Why This Trial Matters Statistically

This trial illustrates how a diagnostic strategy can be evaluated through a downstream process measure: the time until antibiotic therapy is first modified. That endpoint links testing to clinical workflow without being equivalent to a patient-centered outcome. Keeping that distinction clear is important when translating a statistically detectable difference into a claim about health outcomes.

The design also highlights the importance of describing the actual strategy contrast. Both arms included antimicrobial stewardship, and the rapid-testing arm also underwent standard culture and AST. The comparison is therefore between care strategies with shared components and an additional rapid-testing component, rather than a simple test-versus-no-test contrast.

Statistically, the sponsor’s reported analysis and a summary-statistic re-calculation point in the same direction despite a reversed subtraction convention and small numerical differences. Reporting the contrast orientation, interval, and analysis population helps readers understand what a positive or negative estimate means. The secondary outcomes further illustrate why a consistent pattern across related endpoints can be informative but does not remove the need to account for multiple testing and uncertainty.

Finally, the planning illustrations show how required sample size depends on the effect size targeted and desired power. They are not a substitute for the trial’s original design rationale, which is not reported in the registry record, and they are not evidence of observed power. Distinguishing planning calculations from trial findings is essential for sound interpretation.

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.

The registry record gives limited detail about eligibility criteria beyond confirmed gram-negative bacillus bacteremia, and it does not identify who was masked despite describing the trial as single-masked. The exact procedures for allocation concealment and masking are not reported. These omissions make it difficult to assess how treatment assignment and knowledge of assignment may have affected conduct or outcome recording.

The primary endpoint is described as mean time to first antibiotic modification among subjects who completed the study. The record does not explain how a modification was defined, how participants with no modification during the window were handled, or how missing times were addressed. It also does not give a detailed participant-flow account or analysis-set definition. These are important details for interpreting a time-based outcome.

The intervention arm received both rapid testing and standard culture and AST, while stewardship was included in both groups. As a result, the comparison estimates the effect of the overall testing strategies as described; it cannot isolate the contribution of each component from the registry summaries alone. The record also does not provide a clinical threshold for judging the importance of a difference in hours.

The record does not report the trial’s original sample-size justification, a detailed statistical analysis plan, interim-analysis information, or a multiplicity procedure. Several secondary outcomes are listed, but the available numerical results do not fully resolve every one. These reporting gaps limit evaluation of robustness and secondary findings. The posted summaries support interpretation of the stated comparisons, not conclusions beyond them.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Rapid Identification and Phenotypic Susceptibility Testing for Gram-Negative Bacteremia — NCT03218397.
• Registry record last updated 2019-11-26; status last verified 2019-11.
• Results first posted 2019-11-26.
• 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

RAPIDS-GN (NCT03218397) is a completed, randomized, parallel-group, single-masked trial sponsored by Duke University. The registry describes a multi-center, prospective comparison of standard blood culture and AST with a rapid organism identification and AST strategy, with antimicrobial stewardship in both arms and standard testing also performed in the rapid-testing arm.

The primary endpoint was mean hours to first antibiotic modification within 72 hours after randomization among subjects who completed the study. The posted means were 24.7 hours with standard testing and 19.0 hours with rapid testing. The sponsor reported a difference of 5.6 hours, a 95% confidence interval from 1.2 to 10, and p = 0.013. The simple unadjusted re-calculation expressed the contrast in the reverse direction.

The registry lists enrollment as 500 and records study start as 2017-10-09 and primary completion as 2018-11-03. Original design assumptions, detailed missing-data procedures, and multiplicity methods are not reported in the registry record.