Safety and Efficacy Study of Aztreonam for Inhalation Solution (AZLI) in Cystic Fibrosis Patients With P. Aeruginosa: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Safety and Efficacy Study of Aztreonam for Inhalation Solution (AZLI) in Cystic Fibrosis Patients With P. Aeruginosa (NCT00104520), a randomized phase 3 study. The primary outcome is Time to Need for Inhaled or Intravenous (IV) Antipseudomonal Antibiotics.

2. Clinical Question

This is a time-to-event question rather than a comparison of the average number of antibiotic courses or a simple count of participants needing antibiotics. A participant's event time depends on when the documented need first occurred during the stated follow-up. Participants without a recorded event during follow-up can still contribute information up to the point their observation ends. The registry record does not give a hazard ratio or a detailed account of censoring, so the available results do not quantify a relative event rate over follow-up. They report group medians with confidence intervals and a log-rank test.

The endpoint is tied to a clinical decision: a course of antipseudomonal antibiotics was considered needed when a physician documented the need in relation to specified symptoms or other reasons. It therefore reflects both participants' clinical status and the assessment and documentation of the treating physician. The endpoint is not simply a laboratory measure of bacterial burden, lung function, or symptoms. The record does not report additional criteria governing when a physician made or documented the decision.

3. Trial Design

The registry describes a Phase 3, randomized, parallel-group, placebo-controlled trial with quadruple masking. Its official title identifies it as a multicenter, double-blind study of aztreonam lysinate for inhalation in people with cystic fibrosis and pulmonary Pseudomonas aeruginosa who required frequent antibiotics. The sponsor was Gilead Sciences. The registry record gives enrollment as 211 and lists the trial status as completed.

The intervention groups in the posted results are AZLI pooled across the two-times-a-day and three-times-a-day schedules, and placebo pooled across the same schedules. The intervention description identifies AZLI as 75 mg. The registry does not provide arm descriptions beyond the pooled labels. Consequently, the posted estimates address the pooled treatment groups and cannot show whether outcomes differed between the two dosing schedules.

The primary outcome was time to need for inhaled or intravenous antipseudomonal antibiotics. Its stated timeframe was Day 0 to Day 84, the end of the study. The primary population is described as intent to treat, with the additional qualification that participants were randomized and received at least part of one dose of study drug. The posted primary groups include 76 participants in the placebo group and 135 in the AZLI group.

The registry gives a start date of 2005-02 and a primary completion date of 2006-09. The stated secondary outcomes cover a respiratory-symptom questionnaire score, percent change in forced expiratory volume in one second, hospitalization days, and change in sputum Pseudomonas aeruginosa burden. The registry record does not report the exact roles included in quadruple masking, the procedures used to maintain masking, or details of random-sequence generation and allocation concealment. It also does not report the protocol's sample-size justification or a full statistical analysis plan. Those omissions limit what can be inferred about the planned conduct beyond the design labels and outcome descriptions recorded in the registry.

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: Subjects were randomized in 1:1:2:2 ratio to placebo twice a day (BID) or three times a day (TID) and AZLI BID or TID, respectively. A total of 211 subjects were included in the ITT analyses: 38 placebo BID, 38 placebo TID, 69 AZLI BID, and 66 AZLI TID. All analyses were conducted on pooled placebo (n = 76) versus pooled AZLI (n = 135) groups.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Analysis based on intents to treat (ITT) population (all participants randomized to treatment who received at least part of one dose of study drug).

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 Time to Need for Inhaled or Intravenous (IV) Antipseudomonal Antibiotics, measured over Day 0 to Day 84 (end of study).

The primary endpoint measures the time from Day 0 until a course of inhaled or intravenous antipseudomonal antibiotics was judged necessary and that need was documented by a physician. The registry describes the decision as based on the presence of at least one of the listed indicators predictive of pulmonary exacerbation: decreased exercise tolerance, increased cough, increased sputum or chest congestion, decreased appetite, or other. Because the endpoint is time to an event, it captures when the need arose during follow-up rather than only whether an event occurred at any point.

This endpoint is used to represent a clinically consequential change in a participant's course, assessed in the context of symptoms and physician judgment. It is distinct from a scheduled measurement of lung function, a questionnaire score, or sputum bacterial burden. The definition also means that the outcome depends on assessment and documentation, not solely on a centrally measured biological quantity. The registry record does not describe additional adjudication or standardization procedures for these assessments.

The listed primary summary is the median time to the event, with a 95% confidence interval, and the between-group test is a log-rank test. A median gives the time by which half of the observed event-time distribution has experienced the event, under the analysis method. It does not describe every participant's experience or the full shape of the event-time distributions. The record gives Day 0 to Day 84 as the observation window; it does not report a hazard ratio or a more detailed account of censoring.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Time-to-event summaries. A time-to-event analysis considers both whether an event was observed and how long it took to occur. For this trial, the event was a documented need for inhaled or intravenous antipseudomonal antibiotics. The median is a summary of the event-time distribution: it marks the time at which half of the distribution has reached the event, as estimated by the analysis. It is not the mean time, and it does not indicate that every participant experienced the event at that time.

Time-to-event methods can use information from participants who have not had the event by the end of observation or whose observation ends earlier. Such records are commonly treated as censored, meaning the analysis uses their observed follow-up without assuming that the event occurred after observation stopped. This is a general explanation of the method, not a statement about the exact handling in this trial. The registry record does not report censoring details, the number of events, or the follow-up distribution.

Confidence intervals for medians. A confidence interval gives a range of values compatible with the data under the statistical method and its assumptions. It conveys uncertainty around an estimate; it is not a range containing a fixed proportion of individual participant outcomes. The placebo median is reported with a 95% confidence interval of 57 to 97 days. The AZLI result gives a median of 92 days and a lower confidence bound of 89 days, while an upper bound is not reported in the registry record. Without that upper bound, the interval cannot be described completely. The record does not explain why the upper limit is absent.

Log-rank comparison. The log-rank test is a general method for comparing event-time patterns between groups over follow-up. It considers the timing of events across the observation period rather than comparing only the two reported medians. A p-value from this test assesses compatibility of the observed group patterns with the test's null hypothesis, under the test assumptions. The reported p-value is 0.007. This test does not itself provide an estimate of the magnitude of a relative event rate. The registry record reports no hazard ratio, so that measure should not be inferred from the median values or the p-value.

As a general matter, a hazard ratio summarizes a relative event rate at a given time, with the usual interpretation depending on model assumptions. Such a ratio is not interchangeable with a difference between medians. No hazard ratio is reported for this primary outcome, and there is no basis in the registry record for reconstructing one. The value of the log-rank test is therefore to assess evidence of a difference in event-time patterns, while the medians describe one point in each estimated distribution.

Continuous secondary outcomes. The record also includes continuous outcomes analyzed with least-squares means and standard errors, including the questionnaire score and percent change in forced expiratory volume in one second. A least-squares mean is a model-based group estimate; it need not be the arithmetic average of the raw observations. A standard error describes uncertainty in an estimated mean, rather than the spread of individual participant values. The sponsor reports ANCOVA analyses for these outcomes, with a mean difference, confidence interval, and p-value. The registry does not provide the covariate details or model specification, so the precise adjustment cannot be assessed from the record.

Simple unadjusted recalculations from posted summary statistics are not substitutes for the sponsor's model-based analyses. They can be useful as checks on the direction and approximate scale of a comparison, but they cannot reproduce an adjusted analysis without the underlying data and model details. In these results the record does not flag a discrepancy; the sponsor's reported analysis remains the direct report of the planned or conducted analysis, while recalculations should be understood as separate approximations.

Over-dispersion and count-like outcomes. In general, over-dispersion occurs when variability in count data exceeds what a simple model assumes. If it is present and not handled appropriately, estimated uncertainty can be too small. Hospitalization days are reported as means and standard deviations, but the registry does not report a sponsor analysis for the between-group comparison or a model for those data. No specific claim about over-dispersion in this trial can be made from the posted summary alone. General concerns about distributional assumptions are reasons to seek the analysis details, not evidence that a particular assumption failed.

Planning illustrations. The separate sample-size calculations are planning illustrations based on observed effects, not the trial's original design justification and not observed power. They assume a two-sided significance level of 0.05, equal allocation, and a normal approximation; the effect to detect is a stated fraction of the observed effect. The record explains that time-to-event illustrations use total events, while other illustrations use participants per arm. Such calculations answer a hypothetical planning question under their assumptions. They do not show the probability that this completed trial detected an effect, and they do not establish that the original trial was planned using those assumptions. The registry record does not report its own sample-size justification.

10. Analyses Reported by the Sponsor

For the primary outcome, the registry reports the sponsor's log-rank test with p = 0.007 and the group medians and confidence limits. The record does not report a sponsor hazard ratio for this comparison. The displayed primary result should therefore be interpreted through its time-to-event summaries and log-rank test, rather than through an unreported relative-risk estimate. There is no separate recalculated primary effect estimate in the posted results to compare with the sponsor's test.

For the respiratory-symptom questionnaire outcome, the sponsor reports an ANCOVA mean difference of 5.01, with a 95% confidence interval from 0.81 to 9.21 and p = 0.020. A simple unadjusted recalculation from the posted group summaries gives an estimate of 5.0, with an interval from 0.828 to 9.172 and an unadjusted p-value of 0.019. For percent change in forced expiratory volume in one second, the sponsor reports a mean difference of 6.280, a 95% confidence interval from 2.500 to 10.060, and p = 0.001. The simple unadjusted recalculation gives 6.28, an interval from 2.521 to 10.04, and an unadjusted p-value of 0.001.

The recalculated estimates are close to the sponsor-reported estimates, but they are not the same analysis: the sponsor identifies ANCOVA, while the recalculations use posted summary statistics without the sponsor's model details. The record does not flag a discrepancy. Where a sponsor analysis is reported, it is the primary account of that analysis; the recalculations are supplementary checks and should not be presented as replacing it. No sponsor comparison is reported for hospitalization days.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint and several secondary outcomes, but the registry record does not report how the analyses were organized to account for multiple comparisons. It does not state whether secondary outcomes were tested in a prespecified order, whether a family-wise error procedure or another adjustment was used, or whether the secondary tests were intended to be confirmatory. This matters because examining multiple outcomes creates more opportunities for apparently noteworthy results by chance. The unadjusted recalculations should therefore be read as exploratory comparisons, not as multiplicity-controlled evidence.

The record identifies the primary analysis population as randomized participants who received at least part of one dose of study drug. It does not report how missing measurements were handled for the primary or secondary outcomes, whether any participants were excluded after randomization, or whether sensitivity analyses addressed missing data. It also does not detail withdrawals, reasons for missing observations, or the assumptions used for any imputation. These are important because missingness can affect both the estimates and the uncertainty, especially when participants' likelihood of having data is related to their health or treatment experience.

The registry does not report whether interim analyses were planned or performed, whether stopping boundaries were used, or whether the analysis plan changed. The protocol's sample-size justification and the original statistical analysis plan are also not reported in the registry record. These omissions do not establish that such procedures were absent; they mean the registry entry does not allow readers to assess them.

The posted primary result includes a log-rank test, but the record does not give the number of primary events, censoring details, or a hazard ratio. For the secondary continuous outcomes, sponsor-reported ANCOVA results are available for the questionnaire and lung-function measures, but covariate choices and model diagnostics are not reported. The hospitalization-days summary has no posted sponsor comparison. These gaps limit independent assessment of analysis assumptions and the robustness of the findings. They should be kept distinct from the numerical results that are reported: the medians, confidence limits, group summaries, and reported tests remain interpretable as registry results, but their full analytical context is incomplete.

12. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: cough (5.2%), productive cough (4.4%), pulmonary function test decreased (3.0%), dyspnoea (2.6%), respiratory tract congestion (2.2%).

The registry reports serious adverse events affecting 3 participants in the placebo group and 13 in the AZLI group. The corresponding groups at risk are listed as 76 and 135. These are counts of participants reported as affected, not measures of event severity beyond the serious classification, and they do not by themselves describe timing, duration, recurrence, or the relationship of an event to study treatment. The registry record does not report deaths for either group.

The record also lists serious adverse event terms with percentages: cough, 5.2%; productive cough, 4.4%; pulmonary function test decreased, 3.0%; dyspnoea, 2.6%; respiratory tract congestion, 2.2%; and wheezing, 1.5%. The posted terms do not identify a group alongside each percentage, so these figures should not be attributed to one treatment group. They are not a complete account of all serious events or all adverse events.

Other adverse events are reported for 65 participants in the placebo group and 121 in the AZLI group, with 76 and 135 at risk, respectively. The registry record does not provide additional detail about the event definitions, collection period, severity grades, relatedness assessments, or handling of multiple events in one participant. Because group sizes differ and event counts alone do not account for follow-up or exposure, the counts should not be compared as if they were directly standardized risks. These safety summaries describe what the registry reports; they do not establish that a treatment caused or prevented an event.

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

The primary result is a comparison of time to a physician-documented need for inhaled or intravenous antipseudomonal antibiotics in the trial's pooled AZLI and pooled placebo groups. The estimated median was later in the AZLI group than in the placebo group, and the registry reports a log-rank p-value of 0.007. This is evidence of a difference in observed event-time patterns under the test's assumptions. It is not a direct estimate of how much an individual participant's risk changed, and no hazard ratio is reported.

The result does not mean every participant in the AZLI group had a later event than every participant in the placebo group. Group medians summarize distributions and do not describe individual outcomes. Nor does the p-value state the probability that the result arose by chance or the probability that a treatment effect exists. It measures how incompatible the observed data are with the test's null hypothesis under specified assumptions.

The outcome depends on a documented physician assessment in relation to symptoms or other reasons. It does not directly measure all pulmonary exacerbations, bacterial burden, lung function, or quality of life. The registry's secondary outcomes provide additional measures, but differences in timeframe and outcome definition prevent them from being treated as interchangeable confirmation of the primary endpoint. The record does not report how endpoint decisions were standardized or adjudicated.

The posted findings also do not settle whether the same pattern would occur in other populations, settings, or dosing schedules. The primary results pool AZLI schedules of two times a day and three times a day, so they do not supply schedule-specific estimates. The recorded analysis population excludes randomized participants who received no study drug, as defined in the registry. Missing-data procedures, multiplicity handling, and several elements of the analysis plan are not reported. These uncertainties do not erase the reported comparison, but they limit how broadly or conclusively it can be interpreted.

In short, the registry reports a later median time to documented antibiotic need for pooled AZLI than for pooled placebo, with a log-rank test result of p = 0.007. The data describe this trial's comparison and its stated endpoint. They do not, on their own, establish a universal effect, a mechanism, or a patient-specific outcome.

14. Why This Trial Matters Statistically

This trial illustrates why a clinical endpoint's definition matters as much as its label. “Time to need for antibiotics” is not simply a count of prescriptions or a direct physiological measurement. In this record, the event required physician documentation and was linked to symptoms predictive of pulmonary exacerbation or other reasons. Statistical interpretation therefore needs to account for the clinical judgment embedded in the endpoint, while avoiding assumptions about procedures that the registry does not describe.

It also illustrates the value and limits of time-to-event methods. A median time gives an accessible summary, while the log-rank test compares event-time patterns over follow-up. Neither one alone provides every aspect of the treatment contrast. Without a reported hazard ratio, event count, or censoring detail, readers cannot assess a relative event rate or fully reconstruct the analysis. The reported median and test remain informative, but they answer narrower questions than a complete time-to-event analysis might.

The secondary outcomes show why a broad outcome set requires careful interpretation. Symptoms, lung function, hospitalization, and sputum measurements describe different domains. A pattern across measures can add context, but unadjusted comparisons are vulnerable to multiplicity and do not necessarily constitute independent confirmation. The registry record's lack of a reported multiplicity strategy makes that distinction particularly important.

Finally, the trial demonstrates why recalculations from summary statistics should be separated from sponsor analyses. Simple recalculations can provide a check on scale and direction, but they do not recreate model-based analyses such as ANCOVA. The sponsor's reported estimates are the direct record of those analyses, while unadjusted recalculations are supplementary. Clear separation of these sources helps readers distinguish what was reported from what can be approximated from group summaries, and what remains unavailable in the registry.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several limitations specific to this record affect interpretation. First, the primary results pool the two-times-a-day and three-times-a-day schedules in both groups. No separate regimen-specific primary estimate is reported, so the pooled result cannot indicate whether the schedules differed. The arm descriptions contain no further detail about the pooled groups.

Second, the primary outcome uses documented physician assessment. The listed symptoms and the possibility of “other” reasons give the endpoint clinical relevance, but the registry record does not describe assessment rules, adjudication, or consistency of documentation. The outcome may therefore reflect both clinical status and the process by which a need for antibiotics was assessed and recorded.

Third, the primary confidence interval is incomplete in the registry: the AZLI median is reported as 92 days with a lower bound of 89 days, but no upper bound is reported. The record does not explain the omission. It also gives no hazard ratio, event count, or censoring details, limiting assessment of the full time-to-event result.

Fourth, the registry does not report missing-data methods, the number or reasons for withdrawals, interim analyses, multiplicity procedures, the original sample-size justification, or a full statistical analysis plan. The sponsor's ANCOVA covariate details are not reported, and no sponsor comparison is listed for hospitalization days. These details are needed to evaluate the robustness and intended interpretation of the analyses, particularly the secondary outcomes.

Finally, the safety summary does not provide event timing, duration, treatment relatedness, or group attribution for the listed serious-event percentages. Deaths are not reported in the registry record. The registry entry is therefore not a complete basis for comparing detailed safety profiles. These are limitations of the information recorded, not proof that the omitted analyses or procedures were not performed.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Safety and Efficacy Study of Aztreonam for Inhalation Solution (AZLI) in Cystic Fibrosis Patients With P. Aeruginosa — NCT00104520.
• Registry record last updated 2011-03-11; status last verified 2010-09.
• Results first posted 2011-03-11.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry record identifies NCT00104520 as a completed Phase 3 randomized, parallel-group, quadruple-masked, placebo-controlled trial sponsored by Gilead Sciences. It enrolled 211 participants with cystic fibrosis and pulmonary Pseudomonas aeruginosa infection. The study compared AZLI 75 mg, pooled across two-times-a-day and three-times-a-day schedules, with pooled placebo.

The primary outcome was time to a physician-documented need for inhaled or intravenous antipseudomonal antibiotics from Day 0 to Day 84. The reported medians were 71 days for placebo and 92 days for AZLI; the log-rank p-value was 0.007. The AZLI median's upper confidence bound is not reported. Secondary outcomes covered respiratory symptoms, lung function, hospitalization days, and sputum bacterial burden. The registry provides sponsor ANCOVA comparisons for the first two, but does not report a sponsor comparison for hospitalization days.

The record does not report several details needed to fully assess the analyses, including missing-data handling, multiplicity procedures, the sample-size justification, and the full statistical analysis plan. The posted results are consequently best read as registry-reported comparisons with those limitations in view.