Valproate (All indications)

Study	Exclusion reasons	Rmk	Reference
Christensen (Valproate) (Controls exposed to Lamotrigine, sick) (Mixed indications), 2019		Less than 90% of exposed mothers have epilepsy. Overlapping: Christensen 2019 totally included in a larger study published by Dreier 2023 (longer study period and 5 countries) => use of Dreier 2023 data.	Christensen JAMA Netw Open 2019; 2:e186606 10.1001/jamanetworkopen.2018.6606
Christensen (Valproate) (Controls unexposed NOS) (Mixed indications), 2019		Less than 90% of exposed mothers have epilepsy. Overlapping: Christensen 2019 totally included in a larger study published by Dreier 2023 (longer study period and 5 countries) => use of Dreier 2023 data.	Christensen JAMA Netw Open 2019; 2:e186606 10.1001/jamanetworkopen.2018.6606
Coste (Valproate) (Controls unexposed, sick), 2020		EXCLUDED: this control group doesn't exist in the article (only in the report). Children with a diagnosis of brain malformation (ICD-10 codes Q00 to Q04 and Q05.0 to Q05.4) during their stay in the maternity unit are excluded.	Coste Sci Rep 2020; 10:17362 10.1038/s41598-020-74409-x
Daugaard (Valproate) (Controls exposed to Lamotrigine, sick) (Mixed indications), 2020		Less than 90% of exposed mothers have epilepsy. Overlapping between Bjork 2022 and Daugaard 2020 for Intellectual disabilities (2 outcomes), with more pregnancies in Bjork 2022 because longer study period and 5 countries => use of Bjork 2022 data.	Daugaard JAMA Netw Open 2020; 3:e2025570 10.1001/jamanetworkopen.2020.25570
Daugaard (Valproate) (Controls unexposed NOS) (Mixed indications), 2020		Less than 90% of exposed mothers have epilepsy. Overlapping between Bjork 2022 and Daugaard 2020 for Intellectual disabilities (2 outcomes), with more pregnancies in Bjork 2022 because longer study period and 5 countries => use of Bjork 2022 data.	Daugaard JAMA Netw Open 2020; 3:e2025570 10.1001/jamanetworkopen.2020.25570
He (Valproate) (Controls exposed to Lamotrigine, sick), 2017	repeat population groups, duplicate reports (most recent study included)	Overlapping: He 2017 totally included in Li 2023, with a longer study period and more pregnancies => Use of Li 2023. Patients who developed epilepsy during pregnancy or after delivery were not included.	He Epilepsy Behav 2017; 74:10-14 10.1016/j.yebeh.2017.05.034
He (Valproate) (Controls unexposed, sick), 2017	repeat population groups, duplicate reports (most recent study included)	Overlapping: He 2017 totally included in Li 2023, with a longer study period and more pregnancies => Use of Li 2023. Patients who developed epilepsy during pregnancy or after delivery were not included.	He Epilepsy Behav 2017; 74:10-14 10.1016/j.yebeh.2017.05.034
Gopinath (Valproate), 2015	repeat population groups, duplicate reports (most recent study included)	Overlapping: Thomas 2022 included data on language and cognitive delay, also evaluated in Sreedharan et al., 2018; Thomas 2007 and Gopinath 2015, but with more exposed pregnancies, more relevant control group and older children => Use of Thomas 2022.	Gopinath Epilepsy Res 2015; 117:58-62 10.1016/j.eplepsyres.2015.09.003
Thomas (Valproate), 2007	repeat population groups, duplicate reports (most recent study included)	Overlapping: Thomas 2022 included data on language and cognitive delay, also evaluated in Sreedharan et al., 2018; Thomas 2007 and Gopinath 2015, but with more exposed pregnancies, more relevant control group and older children => Use of Thomas 2022.	Thomas Epilepsia 2007; 48:2234-40 10.1111/j.1528-1167.2007.01376.x