Already a DIA Member? Sign in. Not a member? Join.

Sign in

Forgot User ID? or Forgot Password?

Not a Member?

Create Account and Join

Menu Back to Poster-Presentations-Details

M-02: Impact of Patient Support Programs on the Performance of Adverse Drug Event (ADE) Signal Detection





Poster Presenter

      Inyoung Lee

      • PhD Candidate
      • University of Illinois at Chicago
        United States

Objectives

The study objective was to compare the performance of ADE signal detection using data from patient support programs (PSP) vs. non-PSP data within a multinational pharmaceutical company where PSPs are manufacturer-sponsored programs designed to support patients’ disease or medication management.

Method

We used the manufacturer’s internal safety database from 2015 to 2017 to detect signals. We compared sensitivity and specificity of signal detection between PSP and non-PSP data using a list of confirmed and unconfirmed signals as positive and negative references, respectively.

Results

Within the pharmaceutical company’s internal safety database across the three years of data, approximately 84% of the ADE reports were collected from PSPs. The number of ADE reports from PSPs was 5.25 times the reports from non-PSP sources (e.g., clinical trials, spontaneous reports, case reports, registries, and published literature). Yet, the number of unique drug-adverse event (AE) pairs identified from PSPs was 1.5 times that from non-PSP data. This suggests the redundancy of ADE reports contained in the PSP data. The Multi-item Gamma Poisson algorithm was used to compute Empirical Bayesian Geometric Mean (EBGM) where the fifth percentile of EBGM greater than 2 was used as the threshold for identifying a potential drug-AE association (i.e., signal of disproportionate reporting (SDR)). Among the unique drug-AE pairs identified from PSP data, approximately 5% were detected as SDR compared with 33% SDR identified from non-PSP data. Our preliminary result demonstrate a lower sensitivity of ADE signal detection using PSP data compared to non-PSP data (sensitivity PSP: 19% vs. non-PSP: 33%; p<0.0001). Conversely, using PSP data led to higher specificity of signal detection compared to non-PSP data (specificity PSP: 87% vs. non-PSP: 77%; p<0.0001).

Conclusion

Using data from PSPs resulted in lower sensitivity and higher specificity for ADE signal detection compared to using data from non-PSP sources. This is the first known study comparing performance of signal detection using data from PSPs vs. non-PSPs. Limitations of this analysis were that we did not use reference signals of drug-drug interaction or signals among a certain patient population as they are difficult to identify through quantitative signal detection. Signal detection is an important activity for patient safety and public health that identifies potential associations between drug products and AEs leading to further investigation. The landscape of the main databases used for signal detection is changing with rising number of ADE reports received from manufacturers’ PSPs. As more and more ADE reports are received through PSPs across manufacturers, we need to be aware of the implications of having high volume of reports from PSPs in the safety database. Many of the reports from PSPs may be of minimal incremental value and could inundate the database with redundant information, possibly hindering signal detection activities by masking true signals. Regulatory authorities and manufacturers may refer to the findings of this study to help guide decision making on pharmacovigilance regulations and actions around ADE cases received from PSPs.

Be informed and stay engaged.

Don't miss an opportunity - join our mailing list to stay up to date on DIA insights and events.