The INSPIRE-ASP UTI Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The INSPIRE-ASP UTI Trial (NCT03697096), a randomized na study. The primary outcome is Extended-Spectrum Days Of Antibacterial Therapy (ES-DOT) Per Empiric Day.

2. Clinical Question

This is a cluster-randomized controlled trial with a parallel design and no masking. Because hospitals, rather than individual patients, were randomized, the statistical comparison needs to account for the cluster structure. Treating all patient-level observations as independent could make the apparent precision misleading. The record identifies the primary outcome and its measurement window, but does not report the result or the analytic details needed to interpret a treatment-effect estimate.

3. Trial Design

The registry describes a randomized, parallel, cluster-controlled trial of hospitals. Its summary says that 59 individual HCA hospitals were randomized; the enrollment field lists 55412 and specifically cautions that the enrolled “subjects” represent those hospitals. The enrollment figure should not be interpreted as the number of randomized hospitals. The record does not report how many hospitals were assigned to each arm.

The intervention was the INSPIRE CPOE Smart Prompt. Its description says that a computerized physician order entry alert guided empiric antibiotic choice for UTI in non-ICU patients during the first 3 days of hospitalization. The summary describes the prompt as providing the probability that a non-critically ill adult admitted with UTI is infected with a resistant pathogen. The comparator was routine care, described as continued routine antibiotic stewardship strategies.

The registry identifies the design as randomized, unmasked, and parallel. It lists the study as completed. The recorded start date is 2018-10-01 and the primary completion date is 2024-05-16. The phase is listed as NA. These details describe the registry record; they do not establish when the intervention was implemented at each hospital or how the hospitals were allocated.

The primary outcome concerns extended-spectrum days of antibacterial therapy per empiric day, during the first 3 days of hospitalization for each patient in the 15-month intervention period. Secondary outcomes listed in the record concern vancomycin days of therapy and antipseudomonal antibiotic days of therapy per empiric day. The registry lists 11 outcome measures overall but does not provide analysis results for these outcomes. The record does not report the allocation process, the number of hospitals in each arm, or a 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: 55,412 patients were involved in the intervention period from the 59 participating hospitals.

Overall Study

Analysis population for the primary outcome, as defined in the registry: The total number of participants analyzed reported here is from the intervention period.

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: Number of participant totals reflect total intervention participants. 59 hospitals were randomized to a study group, 29 began the assigned intervention and 30 continued routine care.

6. Primary Endpoint

The registry lists the primary outcome as Extended-Spectrum Days Of Antibacterial Therapy (ES-DOT) Per Empiric Day, measured over First 3 days of hospitalization for each patient during 15-month intervention period.

The primary endpoint is extended-spectrum days of antibacterial therapy per empiric day. The registry defines a day of therapy as a standardized national measure of antibacterial use in hospitals: any amount of a particular extended-spectrum antibiotic given during a day counts as a day of therapy for that drug. Use of different extended-spectrum antibiotics is added across the empiric days for each patient admission. That admission-level amount is then divided by its empiric days to form a per-day measure; admissions shorter than 3 days contribute only their admitted days to the numerator and denominator.

The registry describes the outcome as the number of different extended-spectrum antibacterials received each empiric day, averaged over the empiric days in the first 3 days of hospitalization, and summed across participants. The outcome is intended to characterize the breadth of extended-spectrum antibacterial use during the empiric-treatment window, rather than simply whether any antibiotic was given. A per-day measure accounts for variation in the number of days contributing to the empiric period, as defined in the record.

This kind of endpoint can make patterns of antibiotic use measurable for a stewardship comparison. Its interpretation depends on the exact aggregation and analysis, particularly because hospitals were randomized. The registry record does not report the observed endpoint values, the between-arm comparison, or an uncertainty interval. Therefore, the endpoint definition alone does not establish the direction, size, or precision of any difference.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies a cluster-randomized parallel design. In a cluster-randomized trial, groups such as hospitals are assigned to study arms, and outcomes are then observed among people or admissions within those groups. This differs statistically from randomizing each person independently. Patients treated at the same hospital may have related outcomes because they share local prescribing practices, staffing, or other features. When outcomes within a cluster are correlated, the effective information in the data is less than an analysis assuming every observation is independent would suggest. Appropriate uncertainty estimates need to reflect the hospital-level randomization.

Why does the randomized unit matter?

The registry says 59 individual HCA hospitals were randomized and warns that a simple individual-level re-analysis would be misleading. That caution is important: an analysis must respect the unit of assignment. Depending on the statistical approach, this can involve modeling clusters directly or adjusting uncertainty for clustering. The record does not report which approach the sponsor used, the degree of within-hospital correlation, or the resulting estimate and uncertainty. Those details are not reported in the registry record.

What does the endpoint summarize?

Extended-spectrum days of antibacterial therapy per empiric day is a count-based utilization measure with an exposure period defined by the empiric days. For each admission, the registry’s definition counts a day for a particular extended-spectrum antibiotic when any amount of that drug is given. Different extended-spectrum antibiotics are summed across the empiric days, then divided by the empiric days for the admission. If an admission lasts less than 3 days, only days admitted contribute. The resulting admission-level quantity is described as being averaged over empiric days and summed across participants. The record does not report the final arm-level values or the method used to compare them.

Count outcomes can have distributions that differ from the assumptions of simple normal models. For example, count data may be skewed or show over-dispersion, meaning their variability exceeds what a basic count model would expect. These are general statistical considerations, not claims about the trial’s observed outcome distribution. The record lists the primary parameter as NUMBER and gives no dispersion value. The statistical model, any transformation, and any treatment of over-dispersion are not reported in the registry record.

How should uncertainty be read?

In general, a confidence interval describes a range of effect values compatible with the data and the analysis assumptions at a stated confidence level. Its width reflects precision: a narrower interval indicates greater precision under those assumptions, while a wider interval indicates less. For a cluster-randomized comparison, the interval should reflect clustering. The registry record does not report a confidence interval or state the confidence level used for the primary analysis. A p-value, if reported, would address compatibility with a specified null model; it would not by itself describe the magnitude or practical importance of an effect. No primary p-value is reported in the registry record.

What is known about planning calculations?

The record describes planning illustrations based on the observed effect, using a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. The effect to detect is defined as a stated fraction of the observed effect. For time-to-event illustrations, the record specifies total events using the Schoenfeld approach; other illustrations are expressed as participants per arm. These are planning illustrations, not observed power, and they are not the trial’s own design justification. The registry record does not report the trial’s sample-size justification. It also does not provide an observed-effect estimate from which the planning illustrations could be interpreted.

A sample-size illustration answers a planning question under stated assumptions; it does not establish whether the completed trial had adequate precision. In particular, the number of patients alone does not capture the information loss that can result from cluster randomization. A useful assessment would need the cluster-aware analysis, the number of hospitals in each arm, and the observed effect and its uncertainty. These are not reported in the registry record.

What analysis details are absent?

The registry reports no analyses. It does not describe a statistical model, adjustment variables, handling of repeated admissions, or a method for accounting for hospital clustering. The population analyzed is described as coming from the intervention period, but the number analyzed and the relationship between that population and enrollment are not reported. There is no reported estimate for the primary outcome. Consequently, readers cannot verify the analysis from the registry record or independently assess how modeling choices affected precision.

Statistical methods are not interchangeable labels: the chosen model should match the randomization unit, outcome structure, and intended estimand. The registry record is sufficient to identify why cluster-aware analysis matters, but not to establish how it was carried out. Any numerical interpretation of treatment effect or precision would require results and methods not reported in the registry record.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 11 outcome measures, including the primary extended-spectrum days-of-therapy outcome and secondary outcomes concerning vancomycin and antipseudomonal antibiotic use. It reports no analyses. The approach to multiplicity is not reported in the registry record. Multiplicity matters when several outcomes or comparisons are assessed: the more questions evaluated, the more important it is to distinguish a prespecified primary question from exploratory findings and to consider how false-positive risk is managed. The record does not say whether any adjustment, hierarchy, or other multiplicity strategy was planned.

Missing-data handling is also not reported in the registry record. The primary outcome definition explains how admissions shorter than 3 days contribute to the measure, but that is an endpoint rule, not a description of how missing outcome information was handled. The record does not state how unavailable medication data, incomplete admissions, or other missing measurements were treated, nor does it report the extent of missingness.

The analysis population is not fully described. The primary outcome notes that the total number analyzed is from the intervention period, but the number analyzed is not reported in the registry record. There is no participant or admission flow account, and no description of exclusions from analysis. The record also does not say whether the analysis followed the hospital assignments throughout or how deviations from assigned care were handled.

Because assignment occurred at the hospital level, the missing statistical details are consequential. The record does not report the cluster-aware analysis, the analysis population, or results with uncertainty. These omissions prevent a complete assessment of multiplicity, missing data, and the population represented by the primary comparison.

11. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry’s adverse-event entries list zero serious adverse events and zero other adverse events in each arm. For the routine stewardship arm, the record lists 27505 at risk for serious events and 27505 at risk for other events. For the INSPIRE CPOE Smart Prompt arm, it lists 27907 at risk for serious events and 27907 at risk for other events. These are the registry’s recorded figures; the record does not explain in detail how the risk sets were defined or how adverse events were collected.

Zero reported events should be read as a registry entry, not as proof that no adverse events occurred in all settings or that the approaches are equivalent in safety. The record does not provide event narratives, event-specific terms, or further safety analyses; the serious-event terms list is empty. It also does not report a comparison estimate or uncertainty interval for adverse events. Accordingly, these figures do not establish a difference or lack of difference between arms.

The registry describes hospitals as the randomized units, while adverse-event figures are listed with large at-risk counts. The record does not clarify the relationship between these counts and the hospital-level randomization or provide additional detail about individual-level ascertainment. The entries should therefore be interpreted narrowly as the adverse-event figures recorded in the registry, rather than as a comprehensive safety evaluation.

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

The trial addresses whether an electronic prompt presenting a resistant-pathogen probability can guide empiric antibiotic selection in non-critically ill adults admitted with UTI. The comparator is routine antibiotic stewardship. The registry specifies a primary measure of extended-spectrum antibacterial use during the empiric period, which is the first 3 days of hospitalization as defined in the record. This provides a clear description of the intended comparison and the outcome concept.

The registry record does not report the primary outcome result. It therefore does not show whether the recorded measure was higher or lower with the smart prompt, how large a difference was observed, or how precisely any comparison was estimated. It also does not report results for the listed secondary outcomes. Completion status and a detailed outcome definition do not fill this reporting gap.

The outcome captures antibacterial-use patterns, not every aspect of treatment or patient health. A change in extended-spectrum days of therapy would describe antibiotic use as measured by this endpoint; by itself, it would not establish what happened to resistant infections or other clinical outcomes. Conversely, the absence of a posted estimate is not evidence that the approaches had the same outcome. It means that the comparison cannot be evaluated from the registry results recorded.

The cluster design further shapes what would be needed for interpretation. Hospitals were randomized, so a valid estimate and uncertainty interval should account for hospital-level assignment. The registry warns against simple individual-level re-analysis, but it does not report the sponsor’s analysis result or its statistical methods. The number of hospitals in each arm and the analyzed population are not reported in the registry record.

The adverse-event entries list zero serious and zero other events in both arms, but those entries alone do not establish comparative safety. Taken together, the record supports an account of the intervention, comparator, design, endpoint, and recorded adverse-event figures. It does not support a conclusion about the size, direction, precision, or clinical implications of the primary comparison.

13. Why This Trial Matters Statistically

This trial is statistically notable because it randomizes hospitals while measuring antibiotic-use outcomes among patient admissions. That design can evaluate a hospital-level stewardship strategy, but it requires analysis that respects the clustering of observations within hospitals. A patient-level comparison that assumes independence could misrepresent uncertainty. The registry itself cautions against simple individual-level re-analysis, making the randomization unit central to any interpretation.

The primary outcome also illustrates why endpoint definitions matter. The record defines a standardized days-of-therapy measure, specifies how different extended-spectrum antibiotics contribute, and limits the empiric period to the first 3 days of hospitalization, with a rule for shorter admissions. Readers can understand what the outcome is intended to measure, but cannot infer the treatment contrast without the reported results and analysis.

The record makes a distinction between the enrollment field and the randomized units: enrollment is listed as 55412, while the note says the enrolled “subjects” represent 59 individual HCA hospitals. This is an important reminder that enrollment counts can be ambiguous in cluster trials unless the unit is explicit. The record does not state how the hospitals were distributed across arms.

Finally, the registry lists multiple outcomes but no analyses. It records completion and a defined primary endpoint, yet the primary comparison, uncertainty, and analysis methods are not reported. This illustrates a general reporting principle: design and outcome descriptions are not substitutes for effect estimates, uncertainty intervals, and transparent analysis methods. Those elements are needed to judge the strength and precision of a trial’s statistical evidence.

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 main limitation for interpreting the registry record is the absence of reported efficacy results. No primary outcome estimate, arm-specific result, confidence interval, or p-value is reported. The listed secondary outcomes also have no reported analysis results. As a consequence, the record cannot establish the direction or size of the comparison or its precision.

Methods needed to evaluate the comparison are missing. Although the study is identified as cluster randomized and the record warns against simple individual-level re-analysis, the statistical model and cluster-adjustment method are not reported in the registry record. The record does not provide the number of hospitals assigned to each arm, hospital-level outcome information, or a measure of within-hospital correlation. It also does not report an analysis plan or sample-size justification.

The enrollment description requires care: 55412 appears in the enrollment field, while the record clarifies that the enrolled “subjects” represent 59 individual HCA hospitals randomized. The number of participants analyzed during the intervention period is not reported in the registry record. The record does not give a flow of hospitals, admissions, or participants from enrollment to analysis.

Other interpretive details are also absent. The handling and extent of missing outcome data, multiplicity across the 11 listed outcome measures, and any adjustment variables are not reported in the registry record. The adverse-event entries do not include event terms or a detailed account of ascertainment. The record describes a 15-month intervention period for the primary endpoint but does not provide details about implementation across hospitals. These limitations concern what the registry reports; they do not establish that the underlying trial lacked the information or analyses.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: The INSPIRE-ASP UTI Trial — NCT03697096.
• Registry record last updated 2025-01-30; status last verified 2024-12.
• Results first posted 2025-01-30.
• 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

The INSPIRE-ASP UTI Trial is recorded as a completed, randomized, parallel, unmasked cluster trial sponsored by Harvard Pilgrim Health Care. The registry states that 59 individual HCA hospitals were randomized and lists enrollment as 55412, with a note that “subjects” represents hospitals. The recorded start date is 2018-10-01 and primary completion date is 2024-05-16.

The intervention was an INSPIRE CPOE smart prompt intended to guide empiric antibiotic selection for UTI; the comparator was routine stewardship. The primary outcome was extended-spectrum days of antibacterial therapy per empiric day during the first 3 days of hospitalization in the 15-month intervention period. The registry record does not report primary or secondary outcome analyses, effect estimates, or uncertainty intervals. It lists zero serious and zero other adverse events in both arms.