Effects of Stratification on Data Mining in the US Vaccine Adverse Event Reporting System (VAERS)

AbstractBackground: Vaccines are administered differentially according to age and sex, and disease patterns also vary among people of different age and sex groups. Estimates of disproportionality should be calculated based on comparisons of groups that have a similar likelihood of receiving similar vaccines and experiencing similar adverse events, to prevent false disproportionality from occurring. Stratified empirical Bayesian (EB) methods have been compared with crude, but not stratified, proportional reporting ratios (PRRs) in their performance on adverse event data. Objectives: (i) to implement stratification of PRR; (ii) to quantify and compare vaccine-event pairs that are highlighted by PRR and EB05 (the lower bound of the 90% CI of the EB geometric mean), for both crude and stratified; and (iii) to evaluate the effects of stratification by age and sex, in identifying adverse events that are accepted to be caused by vaccines. Methods: We applied EB and PRR data mining methods to data from the US Vaccine Adverse Event Reporting System (VAERS). We stratified PRR and EB05 by age and sex. To study the effects of stratification, we compared the crude PRR and stratified PRR. We also assessed the crude EB05 and stratified EB05, and then compared the effects of stratification on EB05 and PRR. Results: Stratification not only changed the number of vaccine-event pairs that were highlighted, but also changed which pairs were highlighted. There were 283 vaccine-event pairs that were highlighted by the crude EB05, but not the stratified; 12 that were highlighted by the stratified EB05, but not the crude; and 162 that were highlighted by both. Similarly, there were 701 vaccine-event pairs that were highlighted by the crude PRR, but not the stratified; 139 that were highlighted by the stratified PRR, but not the crude; and 895 that were highlighted by both. There were 1466 vaccine-event pairs in which the effect of stratification was different for EB05 and PRR. Conclusion: To our knowledge, this is the first published analysis using stratified PRRs. In this analysis of passive surveillance data, stratification revealed and reduced confounding in EB and PRR, and also unmasked some vaccine-event pairs that the crude values did not highlight. Stratification should be applied if confounding is suspected. By decreasing the total number of highlighted vaccine-event pairs, stratification is likely to increase efficiency and therefore might reduce workload.

[1]  M. Braun,et al.  Data mining in the US Vaccine Adverse Event Reporting System (VAERS): early detection of intussusception and other events after rotavirus vaccination. , 2001, Vaccine.

[2]  S S Ellenberg,et al.  The complicated task of monitoring vaccine safety. , 1997, Public health reports.

[3]  Manfred Hauben,et al.  Signal detection in pharmacovigilance: empirical evaluation of data mining tools , 2005, Pharmacoepidemiology and drug safety.

[4]  R. O’Neill,et al.  Use of Screening Algorithms and Computer Systems to Efficiently Signal Higher-Than-Expected Combinations of Drugs and Events in the US FDA’s Spontaneous Reports Database , 2002, Drug safety.

[5]  R. Johnston,et al.  Adverse Events Associated with Childhood Vaccines: Evidence Bearing on Causality , 1994 .

[6]  W. DuMouchel,et al.  Novel Statistical Tools for Monitoring the Safety of Marketed Drugs , 2007, Clinical pharmacology and therapeutics.

[7]  Robert T. Chen,et al.  Understanding vaccine safety information from the Vaccine Adverse Event Reporting System. , 2004, The Pediatric infectious disease journal.

[8]  D. Burwen,et al.  Comparing data mining methods on the VAERS database , 2005, Pharmacoepidemiology and drug safety.

[9]  W. Haenszel,et al.  Statistical aspects of the analysis of data from retrospective studies of disease. , 1959, Journal of the National Cancer Institute.

[10]  Joseph M. Tonning,et al.  Perspectives on the Use of Data Mining in Pharmacovigilance , 2005, Drug safety.

[11]  S. Cochi,et al.  The Vaccine Adverse Event Reporting System (VAERS). , 1994, Vaccine.

[12]  D. Burwen,et al.  Postmarketing safety surveillance for typhoid fever vaccines from the Vaccine Adverse Event Reporting System, July 1990 through June 2002. , 2004, Clinical infectious diseases : an official publication of the Infectious Diseases Society of America.

[13]  E. Woo,et al.  Adverse events after hepatitis A B combination vaccine. , 2006, Vaccine.

[14]  R. Ball.  Methods of ensuring vaccine safety , 2002, Expert review of vaccines.

[15]  William DuMouchel,et al.  Bayesian Data Mining in Large Frequency Tables, with an Application to the FDA Spontaneous Reporting System , 1999 .

[16]  S. Evans,et al.  Use of proportional reporting ratios (PRRs) for signal generation from spontaneous adverse drug reaction reports , 2001, Pharmacoepidemiology and drug safety.

[17]  W. DuMouchel,et al.  Comparative Performance of Two Quantitative Safety Signalling Methods , 2006, Drug safety.

[18]  Photophobia following smallpox vaccination. , 2005, Vaccine.