This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Behavioral Insights to Encourage Judicious Prescribing of Opioids (NCT02790476), a randomized na study. The primary outcome is Average Change Over Time in Dispensed Narcotics.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT02790476 |
| Phase | NA |
| Condition | Substance-Related Disorders |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None (open label) |
| Enrollment | 851 |
| Arms in the primary outcome | n/a |
| Lead sponsor | University of Southern California |
| Results posted | Yes |
2. Clinical Question
The trial asks whether sending a factual, non-judgmental letter to a prescriber is associated with a different change over time in the amount of dispensed narcotics than not sending a letter. Its stated primary measure is monthly morphine milligram equivalents (MME), comparing the average change over time for the letter group with the average change for the control group.
Population
Clinicians and allied health professionals in California with scheduled drug prescribing privileges who prescribed a schedule II, III, or IV drug to a person who died following an accidental overdose involving a schedule II, III, or IV drug in San Diego County between 1 July 2015 and 30 June 2016.
Intervention
Letters sent to prescribers in San Diego County. The described letter stated that a patient of the provider who was treated with prescription narcotics died of an opioid poisoning and encouraged judicious prescribing, including use of CURES before prescribing.
Comparator
Prescribers assigned to the control group, who did not receive letters.
Primary question
Was the average change over time in monthly dispensed MME different for prescribers receiving a letter than for control prescribers? The registry record does not report the primary outcome result.
The population is defined by prescribing privileges and a link between a prescriber’s prescription and a patient who died after an accidental overdose in San Diego County. It is not a general sample of all prescribers or all opioid prescriptions. The intervention is a communication directed to prescribers, rather than a medication or clinical procedure delivered to patients. The stated comparison concerns prescribing behavior recorded over time. The investigators’ stated hypothesis was that making opioid risk more cognitively available would encourage more careful prescribing. The record also names CURES use and new-start prescriptions as secondary outcomes, but does not report their results.
3. Trial Design
The registry describes a randomized, parallel trial with no masking. Enrollment was 851, and the study status is completed. The sponsor was the University of Southern California. The listed phase is NA. The study began on 2017-01-27, and primary completion was 2017-05-27.
The intervention was a letter sent to prescribers in San Diego County; the control group consisted of prescribers who did not receive letters. The study summary says investigators would identify opioid poisonings over the prior 12 months in collaboration with the San Diego Medical Examiner’s Office and California’s CURES system. Letters were to be sent when at least one prescription from a provider was filled by a patient who died of opioid poisoning in San Diego County. The letters were described as factual and non-judgmental, and encouraged judicious prescribing, including checking CURES before prescribing.
The primary outcome description concerns the average change over time in monthly dispensed narcotic dose, measured in MME. The stated observation timeframe is 12 months. The summary describes an evaluation of prescribing practices over 24 months, with 12 months before and 12 months after the letter. The primary hypothesis compares the average change over time in the letter group with that in the control group. These design details describe the intended comparison, not an outcome finding.
Random allocation can support a comparison between groups because allocation is intended to reduce systematic baseline differences. In an unmasked design, participants and study personnel are not blinded to group assignment. The registry record does not report how allocation was generated or concealed, whether any participants crossed between groups, or how the analysis accounted for the timing of the letter. The record also does not report a detailed statistical analysis plan. Those omissions limit assessment of how closely the analysis followed the intended design.
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
| Stage | Letter Intervention | Control |
|---|---|---|
| STARTED | 404 | 447 |
| COMPLETED | 386 | 425 |
| NOT COMPLETED | 18 | 22 |
Analysis population for the primary outcome, as defined in the registry: Clinicians and allied health professionals with scheduled drug prescribing privileges in California who prescribed a schedule II, III, or IV drug to a person who died as a result of a schedule II, III, or IV accidental overdose between the period of 1 July 2015 and 30 June 2016 in San Diego County.
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: We removed prescribers with missing data or lack of prescriptions during the study period, resulting in an analysis population of 811 prescribers.
| Characteristic | Unit | Letter Intervention | Control | Total |
|---|---|---|---|---|
| Participants analysed | 386 | 425 | 811 | |
| Age, Categorical: <=18 years | Participants | — | — | 0 |
| Age, Categorical: Between 18 and 65 years | Participants | — | — | 0 |
| Age, Categorical: >=65 years | Participants | — | — | 0 |
| Sex: Female, Male: Female | Participants | — | — | 0 |
| Sex: Female, Male: Male | Participants | — | — | 0 |
| Race and Ethnicity Not Collected | Participants | — | — | 0 |
| Region of Enrollment: United States | Participants | 386 | 425 | 811 |
6. Primary Endpoint
The registry lists the primary outcome as Average Change Over Time in Dispensed Narcotics, measured over 12 months.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | 95% Confidence Interval |
| Unit | Morphine milligram equivalents |
| Endpoint classification | Continuous Endpoint |
The primary endpoint is the average change over time in dispensed narcotics, represented as monthly morphine milligram equivalents. MME expresses opioid quantities on a common scale, allowing amounts of different narcotic medications to be described using a shared measure. The registry identifies the endpoint as a mean and gives a 95% confidence interval as its dispersion measure.
The endpoint is intended to capture prescribing behavior through dispensed prescriptions. The trial summary says prescribing practices would be evaluated using CURES data. The stated comparison is not simply the amount dispensed in one group at one time: it is the average change over time for prescribers receiving letters compared with the average change over time for controls. The description specifies 12 months for the primary outcome timeframe and describes a 12-month pre-letter period and a 12-month post-letter period in the study summary.
Using dispensed MME provides a quantitative measure relevant to the trial’s behavioral question, but it represents recorded dispensing rather than every aspect of prescribing judgment or patient opioid exposure. The registry record does not report the observed mean changes, the group contrast, or the confidence interval. It also does not report the precise calculation used to summarize the change over time. As a result, the endpoint’s definition can be described, but its observed magnitude and precision cannot be assessed from the registry record.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record identifies a randomized comparison and a primary question about average changes in monthly dispensed MME. It does not report the numerical primary outcome results. In particular, no group-specific change, between-group estimate, confidence interval, or test result for the primary outcome is reported in the registry record. The record therefore does not establish the direction or size of the observed difference, or how precisely it was estimated.
The investigators’ hypothesis was that letters would make opioid risk more cognitively available and lead physicians to prescribe more carefully. That is a proposed mechanism and expected direction, not a result. The fact that the study is marked completed does not itself show whether the observed prescribing patterns were consistent with the hypothesis. The named secondary outcomes are frequency of CURES use and number of new-start prescriptions; their results are not reported in the registry record either.
The endpoint is a mean, so an estimate would describe an average pattern across the analyzed observations. A mean can conceal variation among prescribers: some may change substantially while others change little or in a different direction. A confidence interval, when available, would communicate uncertainty around an estimate under the analysis assumptions. Without the estimate and interval, it is not possible to judge either the size of the observed contrast or its precision. A significance statement alone, even if one were reported, would not substitute for the estimated magnitude and uncertainty.
The registry also reports serious adverse event and death counts by group. For the Letter Intervention group, 386 are listed as at risk and the recorded counts are zero for serious adverse events, other adverse events, and deaths. For Control, 425 are listed as at risk and those recorded counts are also zero. These are registry counts, not evidence that all possible harms were assessed or that no event occurred outside the reported categories or observation period. The record does not provide event definitions or further details on how safety events were collected.
Enrollment is reported as 851, while the adverse-event entries list 386 and 425 at risk in the two groups. The registry record does not explain the relationship between the enrollment figure and these safety counts. They should therefore be read as separate recorded quantities rather than treated as interchangeable denominators. No inference about outcome completeness, exclusions, or analysis population can be made from the counts alone.
Overall, the available record supports a description of the study question, design, and selected safety counts, but not an estimate of the intervention’s effect on prescribing. The direction, magnitude, and precision of the primary comparison remain not reported in the registry record.
9. Statistical Methods Explained
The statistical descriptions in the registry identify a continuous primary outcome summarized as a mean, with a 95% confidence interval. They do not specify the complete analysis method. The concepts below explain how such quantities are commonly interpreted in general; they should not be taken as a description of an unreported model or procedure used in this trial.
What does a mean change describe?
For a continuous outcome such as monthly dispensed MME, a mean summarizes the arithmetic average of observations in a defined group. A change-over-time outcome compares measurements across time rather than describing a single measurement alone. The registry’s wording identifies an average change in the letter group and an average change in the control group, then states that these changes are to be compared. The record does not specify whether the comparison was calculated from individual prescriber changes, group-level summaries, a repeated-measures model, or another approach.
The exact unit of analysis matters. Prescribing measurements may be recorded repeatedly for a prescriber, and repeated observations from the same prescriber are generally related rather than independent. General statistical methods for repeated measurements account for that dependence in some way. The registry record does not report the analytic unit, the method used to account for repeated measurements, or whether any adjustment variables were included. It would be inappropriate to infer a particular model from the endpoint description alone.
How should a 95% confidence interval be read?
In general, a confidence interval gives a range of values compatible with the observed data under a specified statistical method and its assumptions. It conveys uncertainty about an estimate, not a range containing a fixed proportion of individual participants’ outcomes. A narrow interval generally indicates greater precision than a wide interval, while the interval’s location describes which effect sizes are compatible with the analysis. Interpretation requires the estimate, interval limits, outcome scale, and method. Although the registry lists a 95% confidence interval as the dispersion measure for the primary outcome, it does not report the observed interval or the method used to construct it.
Precision is distinct from clinical or policy importance. An estimate can be precise but small, or imprecise while including a range of potentially important effects. Statistical uncertainty should therefore be considered alongside the scale and practical meaning of the endpoint. In this record, the numerical estimate and interval are not reported, so neither precision nor practical magnitude can be evaluated.
Why can prescribing data be variable?
In general, prescribing measures can vary across people and over time. A small number of unusually high values can influence an arithmetic mean. When variability is greater than assumed by a simple model, standard errors and confidence intervals based on that model may not describe uncertainty well. This is called over-dispersion in settings where the observed variation exceeds the model’s expected variation. Other approaches may be used when data are skewed, clustered, or repeatedly measured. These are general considerations only: the registry does not describe the distribution of MME data or report whether diagnostics, transformations, robust methods, or other variance procedures were used.
The measure is based on dispensed prescriptions according to the study summary. Such a measure can describe recorded dispensing but does not, by itself, establish whether a prescription was taken, how it was used, or the clinical circumstances behind it. This is a distinction between the measured endpoint and broader concepts such as patient exposure or prescribing quality. The endpoint can still be relevant to the behavioral question, but interpretation should remain tied to what was measured.
What does randomization contribute?
In general, random assignment is used to make treatment allocation independent of participant characteristics on average. It can strengthen causal interpretation of a between-group contrast when assignment is implemented as intended and analysis preserves the comparison. Randomization does not guarantee identical groups in every respect, nor does it address all sources of bias after allocation. For example, incomplete outcome data, changes in measurement, or departures from assigned groups can affect interpretation. The registry identifies randomized allocation but does not report allocation procedures, baseline characteristics, adherence to assignment, exclusions, or analytic handling of missing observations.
The design is also listed as unmasked. In general, awareness of assignment can affect behavior or data collection, depending on the outcome and context. Here, the intervention itself is a letter, so prescribers necessarily know whether they received it; the record does not describe masking of outcome assessors or analysts. No further masking procedure is reported. These points do not demonstrate bias, but they are relevant considerations when appraising an unmasked behavioral intervention.
The registry gives assumptions for planning illustrations: a two-sided significance level of 0.05, equal allocation, and a normal approximation, with the effect to detect expressed as a fraction of the observed effect. These are described as planning illustrations based on an observed effect, not as the trial’s own design justification. The record does not report the underlying observed effect or numerical planning results. Such illustrations are not observed power and cannot replace the missing outcome estimate or interval. The registry does not report the original sample-size justification.
10. Multiplicity, Missing Data and Analysis Populations
The registry identifies five outcome measures and nine analyses. It names frequency of CURES use and number of new-start prescriptions as secondary outcomes, in addition to describing the primary outcome of average change over time in dispensed MME. The full outcome definitions, the mapping of analyses to outcomes, and numerical results are not reported in the registry record.
When a study examines several outcomes or performs several analyses, the chance of obtaining at least one apparently notable result can increase. In general, multiplicity procedures can control or describe that issue, but the appropriate approach depends on which outcomes and comparisons were planned and how they relate to the study question. The registry record does not report whether a multiplicity adjustment or a prespecified hierarchy of outcomes was used. It is therefore not possible to determine how multiple analyses were handled or how any individual test should be interpreted in that context.
Information about missing data is also not reported in the registry record. The record does not state how many participants contributed to each outcome, whether measurements were missing, how missing values were handled, or whether sensitivity analyses were conducted. Enrollment is 851, while the safety entries list 386 at risk in the Letter Intervention group and 425 in Control; no explanation of their relationship is reported. Those figures do not establish the number analyzed for the primary outcome.
The analysis populations are not described. The registry record does not say whether analyses followed randomized assignment, included only prescribers with complete measurements, or used another population definition. Nor does it report whether prescribers with no observed dispensing during a measurement period were included and how such observations were treated. These details matter because exclusions and missingness can affect the comparison and its generalizability. Without them, the extent to which the reported design is reflected in the analysis cannot be assessed.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Letter Intervention | 0 | 386 | 0.0% |
| Control | 0 | 425 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Letter Intervention | 0 of 386 |
| Control | 0 of 425 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Letter Intervention | 0 of 386 |
| Control | 0 of 425 |
The registry lists safety counts separately for the Letter Intervention and Control groups. In the Letter Intervention group, 386 are listed as at risk; the recorded counts are zero for serious adverse events, other adverse events, and deaths. In Control, 425 are listed as at risk, with zero recorded for each of those categories as well.
These entries describe what is reported in the registry, not a comprehensive account of all possible harms. The registry record does not report the definitions used for serious or other adverse events, the period over which events were collected, or the process for identifying them. It also does not provide event narratives or information about severity beyond the category labels. A zero count in a registry entry does not by itself show that every event was actively sought, that all participants had complete follow-up, or that no unrecorded event occurred.
The two group counts listed as at risk are safety-record quantities. Enrollment is reported separately as 851, and the record does not explain the relation between enrollment and the safety counts. These figures should not be treated as proof of the number included in the primary outcome analysis. The safety entries do not inform the missing primary outcome estimate or establish the overall safety profile of the letter strategy.
12. What the Result Does, and Does Not, Mean
The study addresses a specific behavioral question: whether a letter to prescribers is associated with a different change in recorded opioid dispensing than no letter. Its randomized, parallel design is relevant to that comparison, and the registry describes a pre-letter and post-letter measurement period. The intervention’s stated aim was to encourage more judicious prescribing, including use of CURES.
What the record does not provide is the primary outcome estimate. The average change in monthly MME for either group, the comparison between groups, and its 95% confidence interval are not reported in the registry record. The named secondary outcomes are also not accompanied by results. Consequently, the posted information cannot establish whether prescribing changed, whether the groups differed, or how large or precise any difference was. The hypothesis in the study summary should not be mistaken for an observed finding.
Even with a numerical difference, interpretation would depend on how the outcome was calculated, which prescribers contributed data, how repeated observations and missing values were handled, and whether the analysis followed the randomized assignment. Those details are not reported in the registry record. The endpoint also concerns dispensed MME, which is a measure of recorded dispensing rather than a complete measure of prescribing intent, use, or patient experience.
The safety entries record zero serious adverse events, other adverse events, and deaths in both groups, with the listed at-risk counts. Those figures are limited to the categories and information recorded; they cannot answer questions that the registry does not document, such as the event collection process or completeness of follow-up.
The most defensible interpretation is therefore narrow. The registry describes a completed randomized evaluation of letters, a defined primary prescribing measure, and selected safety counts. It does not report the outcome data needed to judge the intervention’s observed effect or the uncertainty around it. No conclusion about a benefit, lack of benefit, or harm can be drawn from the absent efficacy results.
13. Why This Trial Matters Statistically
This trial is statistically interesting because it applies random allocation to a behavioral intervention aimed at prescribers. Rather than assigning a drug or procedure, the intervention consists of a letter that communicates a patient’s death from opioid poisoning and encourages judicious prescribing and CURES use. The key statistical question is whether change over time differs between prescribers receiving letters and those assigned to control.
That question is more informative than examining change in the letter group alone. Prescribing may change over time for reasons unrelated to a letter, so a comparison group can help distinguish a general time trend from a difference associated with assignment. Randomization can strengthen that comparison, provided allocation and analysis are appropriately implemented. The registry record does not report enough analytical detail to assess those implementation steps fully.
The endpoint also raises an important measurement issue. A monthly MME measure derived from dispensing data can quantify a pattern in recorded opioid dispensing, but it is not identical to a complete judgment of prescribing quality. Likewise, CURES use and new-start prescriptions capture particular behaviors, not every possible response to a letter. Being explicit about what each outcome measures helps keep statistical conclusions aligned with the actual data.
Finally, a statistical design is only as informative as its reported estimates. The record describes the primary question and lists a confidence-interval measure, but does not report the primary effect estimate or interval. Without them, readers cannot assess magnitude, direction, or precision. This illustrates why transparent outcome reporting is essential: a completed trial and a plausible randomized comparison do not, by themselves, reveal what the data showed.
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.
A central limitation for interpretation is that primary outcome results are not reported in the registry record. The record does not give group-specific average changes in monthly MME, a between-group estimate, or its 95% confidence interval. Results for the named secondary outcomes are also not reported. This prevents assessment of the observed effect, its precision, and consistency across outcomes.
The population is narrowly defined: clinicians and allied health professionals with scheduled drug prescribing privileges who prescribed a schedule II, III, or IV drug to a person who died following an accidental overdose involving a schedule II, III, or IV drug in San Diego County during the stated period. Findings from such a population, if reported, would not automatically apply to prescribers without that history or to other settings. The intervention is specifically a letter sent to prescribers in San Diego County, and the control group did not receive letters.
The intervention is unmasked, and the registry does not report whether outcome assessors or analysts were masked. It also does not describe allocation concealment, the statistical model, handling of repeated measurements, missing data, analysis populations, or adjustment for multiple analyses. These details are necessary for a fuller appraisal of potential bias and uncertainty, but their absence from the registry record does not establish that any particular flaw occurred.
The primary endpoint is recorded dispensing expressed as MME. It does not by itself describe how much medication was consumed, the clinical context for prescribing, or all dimensions of judicious prescribing. The registry record also does not explain how enrollment relates to the at-risk counts in the adverse-event entries. Together, these reporting gaps restrict both the quantitative interpretation and the scope of conclusions that can be drawn.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Behavioral Insights to Encourage Judicious Prescribing of Opioids — NCT02790476.
- Registry record last updated 2024-04-29; status last verified 2023-11.
- Results first posted 2024-04-29.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
The registry describes NCT02790476, titled Behavioral Insights to Encourage Judicious Prescribing of Opioids, as a completed randomized, parallel, unmasked study sponsored by the University of Southern California. Enrollment was 851. The study began on 2017-01-27, with primary completion on 2017-05-27.
Prescribers in San Diego County were assigned to receive a factual letter or to a control group that received no letter. The primary outcome was the average change over time in monthly dispensed MME, with a stated 12-month timeframe. The study summary describes 12 months before and 12 months after the letter. Named secondary outcomes were frequency of CURES use and number of new-start prescriptions.
Numerical primary and secondary outcome results are not reported in the registry record. Safety entries list zero serious adverse events, other adverse events, and deaths in both groups, with 386 at risk in the Letter Intervention group and 425 in Control. The record does not explain the relationship between those counts and total enrollment.