Adverse Events Recurrence

The Adverse Events Recurrence Report generates analyses for studies with one or more treatment arms and computes treatment differences between all treatments for studies with two or more treatment arms.

Most analyses of adverse events (AEs) are based on the presence or absence of events at the patient level. For example, if a patient experiences one or more headaches, they are considered a “success” in that the event occurred (= 1). Patients who do not experience a headache are considered failures (= 0). The number of patients who experience one or more headaches are tabulated for each treatment arm, and the proportion (or odds ratio or risk ratio) of subjects experiencing a headache are compared between treatment arms. This comparison is conducted separately for every AE that occurs within a study. Here, patients that experience one event versus many events are weighted identically with no distinction made for the number of events a patient may experience (save for a patient who experiences no events).

Recurrence analysis acknowledges that an AE may occur more than once within a patient and accounts for these re-appearances within the analysis. Going further, recurrence analysis accounts for the timing at which events occur beyond comparing the final total of events that patients may experience.

Methodology

When using CDISC standards, the AE or ADAE data sets summarize the events that were experienced by patients in the study, but more information is needed. In order to conduct a recurrence analysis, dates representing the start and end of an appropriate follow-up period need to be defined for all patients, not just those who experience a particular AE. Further, the derivation of the appropriate follow-up dates will depend on the types of AEs of interest for analysis (see below).

Figure 1:

The figure above displays start and end dates appropriate for different types of events. Dates used to define follow up period are described below, and they are considered in the order of presentation.

Note: Visit End Date is equivalent to Visit Start Date if it is not available.

All Events include any AEs that may occur on study once informed consent is signed until the patient ends study participation.

•

Start date: Consent Date or Enrollment Date or Earliest Visit Start Date

•

End date: End of Participation or End of Study or Maximum of (Latest Visit End Date, Latest date for Disposition Events from the DS domain)

Treatment Emergent Events include any AEs that may occur on or after the first dose of study therapy determined by the treatment arm.

•

Start date: First Treatment Date

•

End date: End of Participation or End of Study or Maximum of (Latest Visit End Date, Latest date for Disposition Events from the DS domain)

On Treatment Events include any AEs that may occur on or after the first dose of study therapy and before or at the last dose of study therapy.

•

Start date: First Treatment Date

•

End date: Last Treatment Date

Pre-Treatment Events include any AEs that may occur prior to the first dose of study therapy.

•

Start date: Consent Date or Enrollment Date or Earliest Visit Start Date

•

End date: First Treatment Date

Off Treatment Events include any AEs that may occur after the last dose of study therapy.

•

Start date: Last Treatment Date

•

End date: End of Participation or End of Study or Maximum of (Latest Visit End Date, Latest date for Disposition Events from the DS domain)

For Screen Failures, the following changes apply:

All Events

•

No changes

Pre-Treatment Events

•

Any events are considered to have occurred before treatment since treatment was never received

•

End date: End of Participation or End of Study or Maximum of (Latest Visit End Date, Latest date for Disposition Events from the DS domain)

Excluded from Treatment Emergent, On treatment, and Off Treatment events since no treatment was received.

Note: The Recurrence Report is not valid for crossover studies due to the limitations of the statistical methods in computing treatment differences (which assumes treatment arms are independent).

The Adverse Events Recurrence report utilizes the JMP Recurrence Analysis platform to obtain the mean cumulative function (MCF), which shows the total cost per unit as a function of time. In a clinical trial, the MCF shows the total number of events per patient as a function of time. Refer to the documentation for JMP's Recurrence Analysis platform for information on the statistical methods used to assess survival.

Note: Records marked as not occurring (xxOCCUR = N) are excluded from this analysis.

Usage and Interpretation

The Adverse Events Recurrence report was run using the Nicardipine sample data set to demonstrate the usage and interpretation of the results.

Note: Click on the figures below to enlarge.

Example1: Recurrence of On Treatment Adverse Events

In this example, the report options were set as shown in the following table:

**Option Settings**
Option Type	Option	Setting
Data	Event Type	On treatment events
All other options are set to default.

Results Summary

For studies with two or more treatment arms, differences between the treatment arms are compared at the minimum of the maximum follow-up times for the treatments. Here, the minimum follow-up time is 17.5 days.

Figure 2: Dot-Forest Plot Summary of all AEs

For studies with two or more treatment arms, differences between the treatment arms are compared at the minimum of the maximum follow-up times for the treatments. Figure 2 shows the minimum follow-up time is 17.4 days. The dot plot in the left panel shows the value of the MCF at 17.4 days while the forest plot displays the 95% confidence interval for the difference in NIC. 15 – Placebo MCFs. This difference can be reversed by altering the control arm with the combo box. From the dot plot, the MCF 0.3844 and 0.0756 for NIC. 15 and Placebo, respectively, for phlebitis. In other words, patients can expect to experience 0.3844 or 0.0756 phlebitis events by 17.4 days for NIC. 15 and Placebo, respectively. So, one in every 1/0.3844 = 2.6 patients will experience a phlebitis event by 17.4 days.

Note: For single arm studies, the dot-forest plot will be replaced with a forest plot that presents the 95% confidence intervals for the MCF computed at the maximum follow-up time within the treatment arm.

Figure 3: Data Filter

The data filter, shown in Figure 3, has variables to subset the dot-forest plot by the overall event total, the maximum by treatment event total, or the MCF, though other variables may be added.

Figure 4: Max Difference Plot

While the Dot-Forest Plot is useful for identifying treatment differences among AEs at the end of follow-up (the minimum of the maximum of follow-up times), this ignores any differences that could be present earlier in the study. The Max Difference Plot, shown in Figure 4, provides a summary of all AEs across the entire duration of the study as an alternative way to explore specific events in detail. Simply put, the Max Difference Plot takes the maximum observed value of the mean cumulative function (MCF) across all treatments and subtracts the minimum observed value of the MCF, thereby summarizing the maximum difference at each time point across all treatments. In this case, the minimum value of the y-axis would be 0 (since differences would always be positive). In the special case of two treatments, the Max Difference Plot communicates direction between the two treatments by taking the differences in the MCFs between the first treatment minus the second treatment. Effectively, this definition makes the plot similar to the Differences in Mean Cumulative Function Plot below, except here all AEs are presented. Note that this plot is not produced in the case where only a single treatment arm is available.

Summarizing all AEs across time within a single plot can be particularly crowded. To make the Max Difference Plot more interpretable, the Section Filter utilizes the Maximum of Absolute Max Difference to subset to the most interesting events. By default, this value is 10%, meaning that only AEs that experience at least a 10% difference among the treatments at least once over the course of the study are presented. For the Nicardipine study presented above, only 5 AEs meet this threshold. Phlebitis, Hypotension, and Isosthenuria show increased risk for Nicardipine (since the result is generally positive), while Hypertension and Vasoconstriction show increased risk for Placebo (since the result is generally negative).

Recall that for more than two treatments, the differences presented in the Max Difference Plot are the maximum differences observed across all pairs of treatments at any given time point. This enables the user to find AEs where large differences are occurring; subsetting to these interesting events will allow them to better understand the direction of these differences among the treatments across time.

Plots by Adverse Event

The remaining plots in the report summarize a single event which can be selected from the Display Filter. By default, the event that maximizes the difference in MCF between the first and last treatments (based on sort order) will be displayed.

Event plots are produced to show when AEs occur across the follow-up time at the patient level (left) and treatment level (right).

Figure 5: Event Plots

Patients along the y-axis in the left hand plot in Figure 5 are sorted according to the length of their follow-up time. The treatment-level event plot could be interpreted as the events occurring early and not beyond 13 days. However, this is due to a handful of participants with long follow-up times. As seen in the left hand plot, the phlebitis events occur pretty evenly across the follow-up time, particularly for the NIC.15 patients. Note that the bubble size is indicative of the number of events occurring at a given time. Should follow-up start and end dates not include a time component, larger bubbles would indicate multiple events occurring on a single day for a patient. This is less likely to occur when time is considered.

Note: For single arm studies, event plots are presented vertically to allow for greater visibility of the plots.

Figure 6: MCF Plot and Difference in MCF Plot

The top panel in Figure 6 displays a plot of the MCF for each treatment and the point-wise 95% confidence band. The approximately straight line for the MCF (at least until about 13 days) shows that phlebitis AEs occur at a relatively constant rate over time. The MCF becomes horizontal beyond approximately 13 days indicating no additional phlebitis events occur. The bottom panel plots the differences between the two arms.

The Number at Risk table, displayed at the bottom of the figure when All Patients is selected using the Number at Risk display option, communicates the number of patients who have yet to have an event or be censored. When Cumulative Events is selected using the Number at Risk display option, the total number of patients with events up to and including a particular time are presented in the risk table, which has now been renamed Cumulative Events. Numbers are presented every 30 days and at the last day of follow up.

Note: An MCF curve that is concave down would indicate that the time between phlebitis events is increasing, showing a decrease in phlebitis events. In other words, safety is improving for phlebitis events. An MCF curve that is concave up would indicate that the time between phlebitis events is decreasing, showing an increase in phlebitis events. In other words, safety is worsening for phlebitis events.

Note: For single arm studies, a Difference in Mean Cumulative Function plot is not applicable and is therefore not created.

Figure 7: Forest Plots of MCF and Difference in MCF Confidence Intervals

(click image to enlarge)

The dual forest plot, shown in Figure 7, presents the 95% confidence intervals graphically and in text of the MCF for each treatment as well as the pairwise differences for MCF at the minimum treatment follow-up time. Like the plots shown above, the control group can be selected using the combo box. The points in the left forest plot correspond to the dot plot; here, the confidence intervals are presented to show additional information.

Note: For single arm studies, the dual forest plot becomes an individual forest plot only showing the confidence interval graphically and tabularly.

Data Tables

Finally, should additional analysis insight be of interest from the JMP Recurrence Analysis platform, the Recurrence data table where Record Flag equals Patient can be used.

Click

to view the underlying data table.

(click image to enlarge)

Click

to open the data table as a JMP table.

Select Analyze > Reliability and Survival > Recurrence Analysis to open the Recurrence Analysis platform.

Set the following options:

•

Y, Age, Event Timestamp will use the start date of the adverse event.

•

Label will use Unique Subject Identifier.

•

Cost will use Frequency.

•

Grouping will use Treatment.

•

Timestamp at Start/End will use the dates.

•

By will use event name, typically Dictionary-Derived Term if analyzing preferred terms.

Continue the analysis in JMP as described in the Recurrence Analysis platform documentation.

Options

Data

Term Level

View/Hide Details

Use this option to specify which level of terms to use as categories.

Possible term levels are report-specific and are listed in the following table:

Term Level

Corresponding Variable

Function

Preferred Term (PT)

AEDECOD in AE

•

Uses a dictionary-derived text description of the adverse event, listed in the AEDECOD column in AE, that defines the corresponding value in the AETERM or AEMODIFY column.

•

A preferred term is restricted to a single medical concept.

Verbatim Term (VT)

AETERM in AE

•

Select this option to use the term as listed in the adverse event term variable.

•

Verbatim terms are identified by values within the reported term (AETERM) variable of AE.sas7bdat data set.

Body System

AEBODSYS in AE

•

Uses a dictionary-derived text description of the affected body system, listed in the AEBODSYS column in AE.

High(er) Level Term (HLT)

AEHLT in AE

•

A superordinate descriptor for multiple preferred terms (PTs).

High(er) Level Group Term (HLGT)

AEHLGT in AE

•

A superordinate descriptor for multiple higher level terms (HLTs)

Standard MedDRA Query

•

Grouping of MedDRA terms, at a defined level, that describes a specific medical condition.

Primary System Organ Class (SOC)

AESOC in AE

•

Uses Primary System Organ Class level of adverse events as categories.

Lowest Level Term (LLT)

AELLT in AE

•

Uses Lowest Level Term level of adverse events as categories.

Body System or Organ Class

AEBODSYS in AE

•

Uses Body System or Organ Class level of adverse events as categories.

Dictionary-Derived Term

AEDECOD in AE

•

Uses Dictionary-Derived Term level of adverse events as categories.

Reported Term for the Adverse Event

AETERM in AE

•

Uses Reported Term for the Adverse Event level of adverse events as categories.

Reported Name

either CMTRT (for Concomitant Medications) or SUTRT (for Substance Use)

•

Uses Reported Name level of intervention as categories.

Standardized Name

either CMDECOD (for Concomitant Medications) or SUDECOD (for Substance Use)

•

Uses Standardized Name level of intervention as categories.

Note: Input data determines the availability of options. Not all term levels are available in all of the reports in which term levels are specified.

Note: The dictionary name and version used to generate the descriptions should be included as part of the submission.

Include serious adverse events only

Event Type

View/Hide Details

Use this feature to specify the inclusion of particular events in the analysis. The selection limits the report to only adverse events of that type.

By default, all events are included in the analysis.

Available options are listed in the following table:

Option:

Select when you want to:

All event types

•

Include all events.

Pre treatment events

•

Include events that occur before dosing for treated subjects.

Treatment emergent adverse events

•

Include events that occur on or after the first dose of study drug.

Note: For information about how treatment emergent-adverse events (TEAEs) are defined, please refer to How does JMP Clinical determine whether an Event Is a Treatment Emergent Adverse Event?.

On treatment events

•

Include events that occur on or after the first dose of study drug and at or before the last dose of drug (+ the offset for end of dosing). The date of last dosing is normally taken from EXENDTC. If EXENDTC does not exist, the date is taken from EXSTDTC instead.

Off treatment follow up events

•

Include events that occur after the last dose of drug (+ the offset). The date of last dosing is normally taken from EXENDTC. If EXENDTC does not exist, the date is taken from EXSTDTC instead.

Note: When crossover is detected, only those events that occur within the specified treatment period dates are shown. This is true even when All Event Types has been selected. For example, pre-treatment events are not assigned to a treatment and therefore not shown in the report.

Ignore available treatment emergent flags

Collapse Events

Number at Risk Interval (days)

Color Risk Table by Treatment

Additional Filters - Adverse Events

View/Hide Details

The Adverse Events field is an optional parameter that enables you to filter rows from the AE data set to choose the appropriate records to subset to adverse events of interest. Risk indicators are calculated using this subset of adverse events only.

This filter uses JMP WHERE clauses to Select Rows that meet the specified criteria. To use this filter, you must enter the JSL (JMP Scripting Language) expressions that constitute the WHERE clause in the appropriate text box for specific filtering of AEs. Note: This is an advanced concept that not only requires some basic knowledge of the JSL WHERE clause syntax (similar to, but different from, SQL), but also an understanding of the exact variables in the tables in the study.

To filter subjects with a specific event or finding, one could also use the Subpopulation Builder on any domain of interest. For example, filter to all subjects that exhibit cardiac failure ( :Customized Query 01 Name == "Cardiac failure" ) and run all reports on those.