BACKGROUND: Data mining of spontaneously reported adverse drug reactions (ADRs), using measures of disproportionality, is a valuable first evaluation step for drug safety signal detection. Of all ADRs reported for children and adolescents within VigiBase, vaccine-ADR pairs comprise more than half of the reports. ADRs concerning vaccines differ with respect to type and seriousness from other drugs, and therefore may influence signal detection for non-vaccine drugs if not accounted for appropriately. The potential influence of vaccines on safety signal detection for drugs was recently raised by the CIOMS Working Group VIII, who proposed that it may be appropriate to undertake automatic signal detection using both medicines and vaccines, and some analysis using vaccines only. However, it has not described for which types of ADRs or drugs subgroup analysis is beneficial. OBJECTIVE: The aim of the study was to study the methodological aspects concerning the influence of a high prevalence of vaccine-related ADRs on signal detection within paediatric ADR data. METHODS: We analysed all paediatric Individual Case Safety Reports (ICSRs) received by VigiBase between 2000 and 2006, and calculated the reporting odds ratio (ROR) for all unique drug-ADR pairs with at least three reports. The ROR was additionally calculated in subgroups of vaccine-ADR pairs and non-vaccine-ADR pairs and further in different age groups. A proportional change in the ROR for the different subgroups was calculated and the change in the number of signals of disproportional reporting (SDRs) after subgroup analysis was assessed. RESULTS: Of all paediatric ICSRs (N = 218 840, of which 117 877 were vaccine-related), a total of 26 203 unique drug-ADR pairs were eligible for inclusion (5586 vaccine-related). A total of 1637 vaccine-related SDRs and 13 375 non-vaccine-related SDRs were detected in the crude analysis. Subgroup analysis by restricting to either vaccines or non-vaccines revealed 494 additional SDRs for vaccines (+30.2%) and 821 additional SDRs for non-vaccines (+6.1%). Subgroup analyses were only beneficial for non-vaccines if the ADR of interest was reported uncommonly for non-vaccines and beneficial for vaccines if the ADR was reported uncommonly for vaccines. Subgroup analysis for ADRs that were reported commonly for either vaccines or non-vaccines led to the disappearance of 272 SDRs for vaccines and 2721 SDRs for non-vaccines. We could empirically derive a model that predicts the change in ROR in the subgroups based on the proportion of vaccines within the total dataset. CONCLUSION: The high proportion of vaccine-related reports within paediatric ADR data has a large and mathematically predictable impact on signal detection in paediatric ADR data. Subgroup analysis reveals new SDRs that potentially represent genuine safety signals. The most inclusive and sensitive signal detection method would be the combination of a crude and subgroup-based data mining approach, based on the ratio between the proportion of vaccines within the ADR of interest and within all other ADRs.
BACKGROUND: Data mining of spontaneously reported adverse drug reactions (ADRs), using measures of disproportionality, is a valuable first evaluation step for drug safety signal detection. Of all ADRs reported for children and adolescents within VigiBase, vaccine-ADR pairs comprise more than half of the reports. ADRs concerning vaccines differ with respect to type and seriousness from other drugs, and therefore may influence signal detection for non-vaccine drugs if not accounted for appropriately. The potential influence of vaccines on safety signal detection for drugs was recently raised by the CIOMS Working Group VIII, who proposed that it may be appropriate to undertake automatic signal detection using both medicines and vaccines, and some analysis using vaccines only. However, it has not described for which types of ADRs or drugs subgroup analysis is beneficial. OBJECTIVE: The aim of the study was to study the methodological aspects concerning the influence of a high prevalence of vaccine-related ADRs on signal detection within paediatric ADR data. METHODS: We analysed all paediatric Individual Case Safety Reports (ICSRs) received by VigiBase between 2000 and 2006, and calculated the reporting odds ratio (ROR) for all unique drug-ADR pairs with at least three reports. The ROR was additionally calculated in subgroups of vaccine-ADR pairs and non-vaccine-ADR pairs and further in different age groups. A proportional change in the ROR for the different subgroups was calculated and the change in the number of signals of disproportional reporting (SDRs) after subgroup analysis was assessed. RESULTS: Of all paediatric ICSRs (N = 218 840, of which 117 877 were vaccine-related), a total of 26 203 unique drug-ADR pairs were eligible for inclusion (5586 vaccine-related). A total of 1637 vaccine-related SDRs and 13 375 non-vaccine-related SDRs were detected in the crude analysis. Subgroup analysis by restricting to either vaccines or non-vaccines revealed 494 additional SDRs for vaccines (+30.2%) and 821 additional SDRs for non-vaccines (+6.1%). Subgroup analyses were only beneficial for non-vaccines if the ADR of interest was reported uncommonly for non-vaccines and beneficial for vaccines if the ADR was reported uncommonly for vaccines. Subgroup analysis for ADRs that were reported commonly for either vaccines or non-vaccines led to the disappearance of 272 SDRs for vaccines and 2721 SDRs for non-vaccines. We could empirically derive a model that predicts the change in ROR in the subgroups based on the proportion of vaccines within the total dataset. CONCLUSION: The high proportion of vaccine-related reports within paediatric ADR data has a large and mathematically predictable impact on signal detection in paediatric ADR data. Subgroup analysis reveals new SDRs that potentially represent genuine safety signals. The most inclusive and sensitive signal detection method would be the combination of a crude and subgroup-based data mining approach, based on the ratio between the proportion of vaccines within the ADR of interest and within all other ADRs.
Authors: Suzie Seabroke; Gianmario Candore; Kristina Juhlin; Naashika Quarcoo; Antoni Wisniewski; Ramin Arani; Jeffery Painter; Philip Tregunno; G Niklas Norén; Jim Slattery Journal: Drug Saf Date: 2016-04 Impact factor: 5.606
Authors: Sandra de Bie; Carmen Ferrajolo; Sabine M J M Straus; Katia M C Verhamme; Jan Bonhoeffer; Ian C K Wong; Miriam C J M Sturkenboom Journal: PLoS One Date: 2015-06-19 Impact factor: 3.240
Authors: Osemeke U Osokogu; Federica Fregonese; Carmen Ferrajolo; Katia Verhamme; Sandra de Bie; Geert 't Jong; Mariana Catapano; Daniel Weibel; Florentia Kaguelidou; Wichor M Bramer; Yingfen Hsia; Ian C K Wong; Madlen Gazarian; Jan Bonhoeffer; Miriam Sturkenboom Journal: Drug Saf Date: 2015-02 Impact factor: 5.606
Authors: Antoni F Z Wisniewski; Andrew Bate; Cedric Bousquet; Andreas Brueckner; Gianmario Candore; Kristina Juhlin; Miguel A Macia-Martinez; Katrin Manlik; Naashika Quarcoo; Suzie Seabroke; Jim Slattery; Harry Southworth; Bharat Thakrar; Phil Tregunno; Lionel Van Holle; Michael Kayser; G Niklas Norén Journal: Drug Saf Date: 2016-06 Impact factor: 5.606