Hal Ward

Safety signal identification is founded on the recognition of a deviation from the expected background pattern of clinical events in a population of patients exposed to a product over time. A lack of understanding of the cumulative population of patients exposed, duration of exposure, and subpopulations of exposure by dose level is probably the largest pervasive deficiency in preregistrational safety event processing and interpretation. 

The population exposed is of equal importance to the presence of the event itself. The purpose of this article is to introduce a concept to aid in the determination of the total number of exposed patients (i.e., a rate denominator) in clinical trials. In this way, we will improve our understanding of adverse events (AEs) and their impact both on immediate patient safety for patients in the trial and, ultimately, potential drug efficacy when launched on the market. 

Current regulations and guidance documents only tangentially address exposure understanding as a dynamic and concurrent part of AE evaluation, especially as a prospective part of event evaluation. Current emphasis is on end-of-study or end-of-program summary. The explanation perhaps has less to do with intellectual resistance than a historical difficulty in pragmatically having exposure information available on a timely basis. 

Implicit in the structure of expectedness is a deviation in the frequency of an event. However, based on our experience, the submission of IND safety alerts, triggered by a change in frequency, rarely occur. During clinical trial drug development phases, patients progress from healthy volunteers, to relative health, to uncomplicated patients, to populations of patients reflective of patients with potentially higher risk. It would seem logical that an increase in the rate of AE occurrence would be observed through the development life cycle. But who knows, since the use of a radical concept called the rate seems to not yet be fully accepted in the pharmaceutical and biotechnology industries. 

The concept of a rate is widely used in a variety of industries to determine the significance of an event's occurrence. The use of rates to better understand and interpret the significance of AE occurrence, however, has been minimal. 

Typically, AE data is not fully studied or its ramifications understood until the end of the trial when all the trial data is available. If AE rates could be more accurately tracked and analyzed as a trial progresses, significant trial decisions could be made earlier. These decisions could improve patient safety and safeguard patient health. It could also be used to provide insight into a drug's performance earlier in the trial process. This insight could lead to critical trial changes, such as modifying the patient population or dose rates, among others. 

Why as an industry are we not embracing the interpretation of AE occurrence earlier in the drug development process? Among the most significant challenges in the interpretation of summary AE data is that both the actual number of AEs that have occurred (numerator) and the population of exposed patients (denominator) is unknown or difficult to gain access to during the course of the trial. Yet the key to unleashing this wealth of information has been in our hands all along. IVR technology already captures the necessary data; the only requisite is to recognize the power of the data that the technology holds. Using this technology, it's likely that the integrated use of exposure information will be established as a cornerstone of modern event processing. 

Identifying the actual number of exposed patients and the total duration of exposure is challenging.

  Some of the problems that have been identified include:

under-reporting of AEs by patients and practitioners

The challenges are especially daunting for spontaneous AE reporting.

  In the spontaneous arena, reporting is voluntary for health care providers and patients. Therefore, there is likely to be gross under-reporting, and in this case, even the numerator is unknown. Likewise, the population of exposed patients is unknown because there is no accurate way to determine how many patients receive prescription medication. There are firms that provide prescription data and sales data, but that data merely provides estimates of exposure and is inherently inaccurate. However, in this industry, accuracy is absolutely vital to patient health and is critical for determining denominator data for AE reporting. 

AE reporting is affected by many things besides the safety profile of a particular product. Some external factors that affect AE reporting include, but are not limited to:

adverse publicity associated with a particular product

litigation associated with similar products on the market.

In other words, reporting rate does not equal incidence. 

Before you can understand how an IVR system can aid in understanding AE profiles, we need to take a look at what an IVR system is and what information it captures. 

IVR systems utilize touch-tone telephony and user-friendly voice prompts to capture critical patient and drug supply management data in a centralized database. IVR systems were first used to optimize clinical trial management for pharmaceutical, biotechnology, and medical device trials in 1989 and is a widely used technology in this industry today. Frequent users of IVR systems include sponsors, investigators, site coordinators, monitors, and distribution centers. 

Sponsors use IVR systems for real-time updates. This real-time information can be used to monitor the performance of clinical sites, patient enrollment, drug supply re-ordering needs, and drug supply expiration. The data that IVR systems collect enable sponsors to manage clinical trial criteria. For example, IVR systems typically record inclusion/exclusion criteria, patient demographic data, randomization schemes, and enrollment levels. Real-time reports available to sponsors allow them to query for:  

status by protocol, site, country or region

Sponsors are now also using IVR systems to monitor patient compliance to diary reporting and medication regimens. For example, if patients request resupply late, they may receive reminder calls reinforcing the need for maintaining their medication regimen. Sponsors may receive updates on patient compliance trends. Some IVR systems automatically call patients if they do not report diary information in a timely fashion. 

Sites are also actively using IVR systems. Enrollment levels are monitored, clinical trial personnel receive training/communication updates, subjects can be unblinded for emergency reasons, and sites can be updated with reminders. 

Additionally, patients utilize the IVR system to report information that was traditionally captured in paper patient diaries. In this way, patient symptoms (including frequency and severity), quality of life information, and safety/efficacy information is available in real-time, rather than waiting for diary information to be recorded and provided at the next site visit. Another benefit that can be employed when utilizing IVR patient diaries is to force patients to select from a list of possible answers, reducing ambiguity. By forcing patients to select one of several prompted choices, the information provided may be more expeditiously reviewed and analyzed. 

While IVR systems are now routinely used for patient enrollment, randomization, and drug supply, few sponsors have tapped into the breadth of benefits that the captured data can deliver to their clinical trials. One decisive yet untapped benefit is to utilize the captured data to improve the understanding of AE reporting. As detailed earlier, IVR systems are used to track the dispensing of clinical compounds. The system records exactly how often supply requests are received from investigators. Therefore, an accurate calculation of exposure can be determined with a thoughtful analysis of this data, and those AE challenges that were previously discussed can be overcome. 

The first challenge is data accuracy. The total exposure calculation derived from an IVR system is likely to be much more accurate than the data reported directly from those responsible for coding and collecting AE data. This is because IVR systems, when used for drug supply management, must be engaged for redispensing. In this way, a highly accurate record of exposure is achieved. 

The next challenge is gaining the numerator (incident occurrence) and denominator (population exposure) data. IVR systems capture the critical variables that will give you visibility into the characteristics of the patient population and will provide insight into possible unreported AEs. These variables include: 

Numerator data—Incident reporting: Patient discontinuations.

 IVR systems capture patient discontinuation information. A patient dropping out of the clinical trial or discontinuing the medication indicates a possible AE. By utilizing the discontinuation information captured in the IVR system, an AE can be recorded at the time patient withdrawal is known. 

 Because IVR systems capture requests for resupply, they may be used to track discontinuations or changes in dosing patterns. For instance, the absence of a resupply request would obviously suggest an issue, possibly a discontinuation or noncompliance with a medication regimen. Discontinuations during trials are often a predictor of similar trends and AEs for approved drugs. 

Denominator data—Population exposure: Age and gender

. IVR systems are able to provide records of the ages of exposed patients (and ages of those experiencing AEs). In the same manner, IVR systems are able to provide the gender data for both the exposed population in a clinical trial and also the population experiencing AEs. 

. IVR systems are often used for randomization. In other words, randomization schedules can be built into IVR systems so that patients are randomized according to predetermined disease variables. This stratification can then be analyzed for both the exposed population and the population experiencing adverse events. 

. IVR systems can be programmed to capture the geographic distribution of the exposed patient population. This is especially useful in global trials. For example, if clinical trial enrollment lags in one region, one should expect that the frequency of AEs reported in that region would also lag. By comparison, if enrollment is similar in multiple regions, one would expect AE patterns to be similar in those regions. If AE patterns differ (while enrollment remains similar), there are at least two possible causes:  

There may be some ethnic, racial or other patient difference associated with the AEs being reported.

There may be a pattern of under-reporting or a challenge with monitoring in a region. This provides clinical staff with an opportunity to intervene and improve AE reporting processes.

Thus, IVR systems generate denominator data (or at least denominator approximations). They also enable subset approximations (e.g., age, gender, disease stratification). A researcher who is actively involved in the summation and interpretation of AE data should consider the value that an IVR system offers in providing exposure data, both in patients who experience AEs and in the total patient population. 

Articles

IVR systems can capitalize on exposure data and analyze adverse events much earlier.

Press One for Safety

Sponsors use IVR systems for real-time updates. This real-time information can be used to monitor the performance of clinical sites, patient enrollment, drug supply re-ordering needs, and drug supply expiration. The data that IVR systems collect enable sponsors to manage clinical trial criteria. For example, IVR systems typically record inclusion/exclusion criteria, patient demographic data, randomization schemes, and enrollment levels. Real-time reports available to sponsors allow them to query for: 					

. IVR systems can be programmed to capture the geographic distribution of the exposed patient population. This is especially useful in global trials. For example, if clinical trial enrollment lags in one region, one should expect that the frequency of AEs reported in that region would also lag. By comparison, if enrollment is similar in multiple regions, one would expect AE patterns to be similar in those regions. If AE patterns differ (while enrollment remains similar), there are at least two possible causes: 					