Disproportionality analysis using empirical Bayes data mining: a tool for the evaluation of drug interactions in the post‐marketing setting

Almenoff, June; DuMouchel, William; Kindman, L. Allen; Yang, Xionghu; Fram, David

doi:10.1002/pds.885

Cited by 102 publications

(71 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, the FDA is exploring these methods in AERS. The FDA and others in the pharmaceutical industry have been attempting to validate a particular Bayesian model and its assumptions (DuMouchel, 1999;Almenoff et al, 2003;Szarfman et al, 2002). Many pharmacovigilance professionals are not familiar with the theoretical concepts, including limitations, biases, and mathematical foundations.…”

Section: Postmarketing Risk Assessmentmentioning

confidence: 99%

Human Milk Biomonitoring Data: Interpretation and Risk Assessment Issues

LaKind

Brent

Dourson³

et al. 2005

Journal of Toxicology and Environmental Health, Part A

View full text Add to dashboard Cite

Biomonitoring data can, under certain conditions, be used to describe potential risks to human health (for example, blood lead levels used to determine children's neurodevelopmental risk). At present, there are very few chemical exposures at low levels for which sufficient data exist to state with confidence the link between levels of environmental chemicals in a person's body and his or her risk of adverse health effects. Human milk biomonitoring presents additional complications. Human milk can be used to obtain information on both the levels of environmental chemicals in the mother and her infant's exposure to an environmental chemical. However, in terms of the health of the mother, there are little to no extant data that can be used to link levels of most environmental chemicals in human milk to a particular health outcome in the mother. This is because, traditionally, risks are estimated based on dose, rather than on levels of environmental chemicals in the body, and the relationship between dose and human tissue levels is complex. On the other hand, for the infant, some information on dose is available because the infant is exposed to environmental chemicals in milk as a "dose" from which risk estimates can be derived. However, the traditional risk assessment approach is not designed to consider the benefits to the infant associated with breastfeeding and is complicated by the relatively short-term exposures to the infant from breastfeeding. A further complexity derives from the addition of in utero exposures, which complicates interpretation of epidemiological research on health outcomes of breastfeeding infants. Thus, the concept of "risk assessment" as it applies to human milk biomonitoring is not straightforward, and methodologies for undertaking this type of assessment have not yet been fully developed. This article describes the deliberations of the panel convened for the Technical Workshop on Human Milk Surveillance and Biomonitoring for Environmental Chemicals in the United States, held at the Hershey Medical Center, Pennsylvania State College of Medicine, on several issues related to risk assessment and human milk biomonitoring. Discussion of these topics and the thoughts and conclusions of the panel are described in this article.

show abstract

Section: Postmarketing Risk Assessmentmentioning

confidence: 99%

Human Milk Biomonitoring Data: Interpretation and Risk Assessment Issues

LaKind

Brent

Dourson³

et al. 2005

Journal of Toxicology and Environmental Health, Part A

View full text Add to dashboard Cite

show abstract

“…There is ongoing research on drug-drug interaction signal detection (van Puijenbroek et al, 2000;Almenoff et al, 2003;Thakrar et al, 2007;Norén et al, 2008;Strandell et al, 2011), but further work is needed to demonstrate its role as a routine tool as Yules' Q (Egberts et al, 2002). Subjective priors based on other data, such as clinical trial results, are also possible and might be a fruitful field of discovery to pursue.…”

Section: Novel Approaches For Quantitative Analysis On Spontaneous Rementioning

confidence: 99%

Quantitative Signal Detection and Analysis in Pharmacovigilance

Bate

Pariente

Hauben

et al. 2014

Mann's Pharmacovigilance

View full text Add to dashboard Cite

“…MGPS is an innovative data-mining algorithm developed by William DuMouchel which introduces the concept of refining the relative reporting ratio calculation by using Bayesian statistics. [19][20][21] The basic assumption behind MGPS is that each observed count (N; drug/vaccine-event combination) is taken from a Poisson distribution with unknown mean (μ), with an interest center on the ratio λ which equals μ divided by E (the expected count estimated by assuming that the count of all reports for the specific drug/vaccine and the count of all reports for the specific event are independent; Table 1). The essential Bayesian contribution supposes that each λ is drawn from a common prior assumed to be a mixture of two g distributions.…”

Section: Methodsmentioning

confidence: 99%

Quantitative signal detection for vaccines: Effects of stratification, background and masking on GlaxoSmithKline’s spontaneous reports database

Zeinoun¹,

Seifert²,

Verstraeten³

2009

Human Vaccines

View full text Add to dashboard Cite

Disproportionality analyses are increasingly gaining acceptance as signal detection tools for drug adverse events. Their application to vaccine adverse events has not been well evaluated. Disproportionality analyses based on the Multi-Item Gamma Poisson Shrinkage principle (MGPS) were applied to spontaneous adverse events reports for five vaccines with different reporting profiles. Sensitivity analyses were used to assess the potential impact of changing key parameters. We used the Company's in-house spontaneous adverse events database. We evaluated the impact of stratification, the comparator dataset and the potential for masking. We conducted a semi-quantitative assessment by comparing the changes in the disproportionality scores and the number of vaccine-event pairs that exceeded an arbitrary threshold as a measure of the impact of any of these choices. The results show that stratification by age and region has a significant impact. The effect of the comparator dataset was dependent on the vaccine being evaluated. The potential for a masking effect was only weakly noticeable. In conclusion, we opt to start with a conservative approach in which we will supplement our primary stratification against the vaccine database with unstratified analyses as well as analyses against the entire database. We recommend that similar studies be performed before introducing disproportionality analyses to a new vaccine adverse events database.

show abstract

Disproportionality analysis using empirical Bayes data mining: a tool for the evaluation of drug interactions in the post‐marketing setting

Cited by 102 publications

References 13 publications

Human Milk Biomonitoring Data: Interpretation and Risk Assessment Issues

Human Milk Biomonitoring Data: Interpretation and Risk Assessment Issues

Quantitative Signal Detection and Analysis in Pharmacovigilance

Quantitative signal detection for vaccines: Effects of stratification, background and masking on GlaxoSmithKline’s spontaneous reports database

Contact Info

Product

Resources

About