Electroconvulsive therapy during pregnancy: a systematic review of case studies

This study aims to explore practice, use, and risk of electroconvulsive therapy (ECT) in pregnancy. A systematic search was undertaken in the databases Medline, Embase, PsycINFO, SveMed and CINAHL (EBSCO). Only primary data-based studies reporting ECT undertaken during pregnancy were included. Two reviewers independently checked study titles and abstracts according to inclusion criteria and extracted detailed use, practice, and adverse effects data from full text retrieved articles. Studies and extracted data were sorted according to before and after year 1970, due to changes in ECT administration over time. A total of 67 case reports were included and studies from all continents represented. Altogether, 169 pregnant women were identified, treated during pregnancy with a mean number of 9.4 ECTs, at mean age of 29 years. Most women received ECT during the 2nd trimester and many were Para I. Main diagnostic indication in years 1970 to 2013 was Depression/Bipolar disorder (including psychotic depression). Missing data on fetus/child was 12 %. ECT parameter report was often sparse. Both bilateral and unilateral electrode placement was used and thiopental was the main anesthetic agent. Adverse events such as fetal heart rate reduction, uterine contractions, and premature labor (born between 29 and 37 gestation weeks) were reported for nearly one third (29 %). The overall child mortality rate was 7.1 %. Lethal outcomes for the fetus and/or baby had diverse associations. ECT during pregnancy is advised considered only as last resort treatment under very stringent diagnostic and clinical indications. Updated international guidelines are urgently needed.


Introduction
For patients with severe psychiatric disorders in the pregnancy period, either medication resistant illness, extremely high suicide risk, psychotic agitation, severe physical decline due to malnutrition or dehydration, electroconvulsive therapy (ECT) still appears as a strong option (Berle et al. 2011;. Previous review publications have advocated ECT to be a relatively safe during pregnancy (Anderson and Reti 2009;Miller 1994;Reyes et al. 2011;Saatcioglu and Tomruk 2011). International ECT guidelines have no clear statements about pregnancy being a contraindication (American Psychiatric 2001;Enns et al. 2010;Royal College of Psychiatrists 2005). Checklists for when ECT is an option during pregnancy have also been provided in textbooks of interface between gynecology and psychiatry (Stewart and Erlick Robinson 2001), without mention of any potential risks to be taken into account.
Prevalence of major depressive episode (MME) during pregnancy is estimated at 12.4 % (Le et al. 2011). Considering that depression is the most common mental disorder (63 %), followed by bipolar disorder (43 %) and schizophrenia (13 %) among deliveries to women with atypical antipsychotic use (Toh et al. 2013), the decision of ECT during pregnancy would not appear uncommon. Although prevalence data on ECT administered during pregnancy is not retrievable, and ECT clearly rarely used during pregnancy in most clinical settings as illustrated by a recent review of contemporary use and practice of ECT worldwide (Leiknes et al. 2012), ECT was noted administered during pregnancy at 10 Polish sites (Gazdag et al. 2009) and also in Spain (Bertolin-Guillen et al. 2006).
Administration of psychotropic drugs during pregnancy requires great caution and benefits must be weighed against potential risks, especially in the first trimester (Stewart and Erlick Robinson 2001). Although evidence for psychotropic medication teratogenicity is generally lacking or limited (Gentile 2010), mood stabilizers such as lithium and valproate are strongly discouraged (Berle and Spigset 2003;Gentile 2010) and carbamazapine controversial (Gentile 2010;Stewart and Erlick Robinson 2001). As for antidepressants, a recent population-based cohort study data from the Danish Fertility Database has found no associated risk with use of SSRIs during pregnancy (Jimenez-Solem et al. 2013). For antipsychotics the risk associated with use during pregnancy is unclear (McCauley-Elsom et al. 2010).
In a systematic review concerning children of women with epilepsy (WWE), no support was found for the common view that epilepsy per se represented a risk for increased congenital malformations (Fried et al. 2004). Conversely, a large population-based register study found a twofold overall risk of malformation in the offspring from WWE compared with those without epilepsy (Artama et al. 2006). Caesarian section in WWE has, also been found to be performed twice as frequently compared with the general population (Olafsson et al. 1998). Total prevalence of major congenital anomalies, is by a large European study (Dolk et al. 2010) reported as 23.9 per 1,000 births for [2003][2004][2005][2006][2007] and 80 % live births. Prevalence of congenital heart disease (the most common birth defect) to be 4-6/1,000 live births by another USA study (Ermis and Morales 2011).
In a previous review of the literature from 1941 to 2007 undertaken by Anderson and Reti (2009), with 57 included studies, ECT was reported administered to 339 women during pregnancy. The same review also reports a partial positive ECT response for pregnant women together with a very low number (N=11) of ECT-related fetal or neonatal abnormalities. Whether these numbers can be reaffirmed and whether there is enough support for APAs the statement that ECT treatment has a "low risk and high efficacy in the management of specific disorders in all three trimesters of pregnancy" (American Psychiatric 2001) is a concern for this present review.
Treatment of mental disorders in pregnancy poses a unique clinical challenge due to potential effects also on the fetus from the intervention. As ECT is utilized worldwide and predominantly in the treatment of women (Leiknes et al. 2012), updated knowledge about safety and risk of ECT treatment during pregnancy for both the mother and fetus/ child is of utmost primary importance.
Against this background, the main objective of this article is to give a systematic case overview of ECT administered during pregnancy, with newer date studies in mind, as well as to report the potential harm (adverse events for mother and fetus/baby).

Data sources and search strategy
A systematic literature search was undertaken in the following databases: Ovid MEDLINE, Embase (Ovid) PsycINFO (Ovid), SveMed, Ovid Nursing Database and CINAHL (EBSCO) ( Table 5 in Appendix 1) in September 2010. The search was updated in January and November 2012 and supplemented with ISI web of Knowledge, Clinical Trials.gov, PROSPERO (CRD), WHO ICTRP, POP-database ( Table 6 in Appendix 1). Search terms intended for Medline were adapted (such) as required for the other databases. Subject headings and free text words used were "electroconvulsive therapy," "electroshock," "electroconvulsive," "ECT," combined with "pregnancy" or "pregnant women" and any of the following "antenatal," "prenatal," "perinatal," "gravid," or "gestation" limited to human studies and dating until today. The search did not exclude the postpartum period to make sure that no articles on the topic were missed. No date limitation was set to find all possible earliest published cases from the 1940s. Relevant references, known to authors of this review from earlier published reviews on this topic or reference lists in retrieved included papers, were also found by hand.

Inclusion and exclusion criteria
Inclusion criteria Studies in the following languages were included: English, Norwegian, Swedish, Danish, Dutch, French, Italian, and Spanish. In addition to authors' European language fluency, the online Google translation tool (http://translate.google.com/) was used when needed.
Exclusion criteria Exclusion criteria include not a data-based study, no or unclear report of ECT undertaken during pregnancy, pseudocyesis, ECT undertaken only in the postpartum period, and not during pregnancy.

Screening of literature
Two reviewers (Kari Ann Leiknes (KAL) and Bjørg Høie (BH)) independently checked the titles, and where available, the abstracts of the studies identified by the electronic database searches. All references appearing to meet inclusion criteria, including those with insufficient details were requested in full text. Reviewers (KAL, BH, and Mary J. Cooke (MJC)), consisting of two pairs independently extracted data from the retrieved full-text articles according to a pre-designed data extraction scheme. All discrepancies were resolved by consensus meeting/discussion, and the final decision was made by the first author (KAL). Ingrid Harboe (IH) undertook the extensive updated literature search. All authors (including Lindy Jarosch-von Schweder (LJS) have contributed to the data presentation and manuscript text.

Data extraction
Briefly, the following aspects were considered: ECT practice and use; publication year and country; diagnoses/indication; mother's age; number of pregnancies (primipara (P1), multipara (P2, 3), etc.); time ECT was administered according to number of gestation weeks (GW), 1st trimester (≤13 GW), 2nd trimester (14-26 GW), 3rd trimester (≥27 GW); total number ECTs administered, ECT administration frequency (two to three times week); ECT parameters (i.e., the manner in which ECT is applied: brief pulse or sine wave current, device type, electrode placement bilateral (BL) or unilateral (UL)); anesthesia type and monitoring (of both mother and fetus); time of birth; and adverse events mother (e.g., genital bleeding, miscarriage, eclampsia, and still birth) and/or baby (e.g., fetal malformations, Apgar score, etc.). As ECT treatment has changed over the years, as for use of anesthesia (termed modified ECT as opposed to unmodified ECT, without anesthesia), device and type of current (mainly from sine wave to brief pulse wave), a clinical cut off for presenting the extracted data was set at 1970.

Study selection
The study selection process, databases searched, and references identified are given in Fig. 1. Altogether, 1,001 references were identified: 681 titles and abstracts screened, 100 full texts screened, 67 included for data extraction, and 33 full texts excluded.

Description of studies
Overview of included case studies (N =67) according to descending publication year, country represented, number of pregnancy cases and fetus and/or baby cases reported are given in Table 1. Overview of full text excluded studies (N =33) and reasons for exclusion are given in Appendix 2. Twelve references were found not relevant to topic (about ECT, but not in pregnancy, e.g., in postpartum or other conditions), 13 had insufficient/ too sparse data, 3 were impossible to find/full text retrieve, and 5 were not relevant, for example, only about anesthesia types or electrical shock accident injury during pregnancy. Detailed extracted data from each included study, such as diagnostic indication, ECT parameters, report of effect and events are presented in Summary of findings tables (N =67), Appendix 3.
A total of 67 case report studies were included, 42 (63 %) from 1970 to 2013 and 25 (37 %) from 1942 to 1970 (Table 1). The literature search included all years, but no studies according to inclusion criteria of this review were found in the 1970s (see Appendix 2 for two excluded 1970s studies (Levine and Frost 1975;Remick and Maurice 1978) lacking ECT data). Studies from all continents were represented as follows: North America (USA and Canada), 32; South America, 1; Europe, 25; Asia (including Middle East), 6; Africa, 2; and Australia, 1. A total of 169 pregnant women were ECT treated from 1942 to 2013. Reports on the fetus or newborn baby/child were found for only 148 cases resulting in 12 % "missing" fetus/baby data (see Table 1).
Altogether, 169 ECT treated pregnant women were identified, exposed to a total number of 1,187 ECTs. Mean and standard deviation (M (SD)) number of ECTs administered per pregnant woman was 9.4 (6.4). Mean age (M (SD) in years) of pregnant women treated with ECT was 28.9 (6.2) and age range 16½-48 years. Overview of ECT-treated pregnant women, number of ECTs, and diagnoses, after and before 1970 is given in Table 2.
Altogether 21 out of 54 (39 %) women were nullipara (Para1) in the later years (from 1970 to 2013) ( Table 2) and for one case in 2011 the pregnancy was by in vitro fertilization (Salzbrenner et al. 2011). The latest ECT administered in pregnancy was at 40 GW (Laird 1955;Schachter 1960) and the earliest at 4 GW (1955). Information about which pregnancy trimester the ECT was undertaken or started was found for 121 women out of 169 (28 % missing). Overview of the ECT reports according to pregnancy trimester for these 121 women is given in Table 3. Most women (53 %) received ECT during the 2nd trimester, although use in the 1st trimester was not uncommon (16 %) and for some, ECT was conducted throughout the entire pregnancy (Pinette et al. 2007).
Generally, the data reported in all studies was very varied concerning the ECT intervention per se, the setting of administration, monitoring, and outcome for both mother and fetus/child.

ECT practice during pregnancy
The setting in which the ECT was administered was usually not recorded. However, ECT undertaken in a surgicalobstetric recovery room or delivery environment was noted by three (Gilot et al. 1999;Wise et al. 1984;Yellowlees and Page 1990).
Monitoring of mother before, during, and after varied. In addition, monitoring of fetus varied greatly from some monitoring to no fetal monitoring by Vanelle et al. (1991). There was some use of cardiotocography (Molina et al. 2010;O'Reardon et al. 2011;Verwiel et al. 1994) but cardiotocography was also noted as not being useful in early pregnancy (before 24 GW) by Lovas et al. (2011). Mother in tilt position during ECT was used in some reports (Brown et al. 2003;Gilot et al. 1999;Livingston et al. 1994;Malhotra et al. 2008;Yang et al. 2011) and by others tilt position was reported not used (Bhatia et al. 1999;Bozkurt et al. 2007;DeBattista et al. 2003).
ECT parameters, such as electrical current type (brief pulse or sine wave), placement of electrodes (UL, BL, bitemporal, and bifrontal) and device manufacture type used was noted in most studies of later date but otherwise very sparsely. (See summary of findings table, Appendix 3). UL placement of electrodes was noted in six studies (Balki et al. 2006;Gahr et al. 2012;Pesiridou et al. 2010;Varan et al. 1985;Wise et al. 1984;Yellowlees and Page 1990). Data on anesthetic agents used combined with muscle relaxant, premedication and 100 % oxygenation was mainly stated in the later date studies (1970 to 2013). Although 13 % of these later date studies (1970 to 2013) were missing anesthesia data, a trend was seen for the following being most used: thiopental (22 %), methohexital (15 %), and propofol (17 %). Anesthesia induced reduced fetal heart rate (FHR) was noted with propofol but not thiamylal in an ECT pregnancy case by Iwasaki et al. (2002). In addition, severe fetal bradycardia by methohexital but not with following propofol anesthesia during ECT administration by De Asis et al. (2013). To  Livingston et al. (1994) USA 1 1(twins) 1 B death Verwiel et al. (1994) N e t h e r l a n d s 1 1 Vanelle et al. (1991) France 5 Evrard (1961) B e l g i u m 1 1 Barten (1961) N e t h e r l a n d s 2 2 Ferrari (1960) I t a l y 8 7 1 B death Sobel (1960) USA 33  avoid pulmonary aspiration, tracheal intubation was preferred by Malhotra et al. (2008) when pregnancy was beyond 1st trimester. Unmodified (without anesthesia) ECT was noted in the earlier studies (from 1942 to 1970), such as in all 8 cases reported by Laird (1955) and in 6 out of 15 cases by Smith (1956). Even use of only muscle relaxant without anesthesia was noted in 7 ECT pregnancy cases by Doan and Huston (1948). Fetus, baby/child-monitoring, and follow-up Fetus or baby/child data was sometimes totally absent even in the later date studies, such as in Gahr et al. (2012) and Ghanizadeh et al. (2009) as well as some earlier ones, for example Russell and Page (1955). Some reported new born baby Apgar score and weight, but most often the information on the newborn infant was meager and the condition of baby/ child noted as normal, "healthy baby," or nothing abnormal.
Information about monitoring of fetus during ECT varied greatly from none at all, to obstetric consultations and ultrasonography between treatment sessions (Espínola-Nadurille et al. 2007;Kasar et al. 2007;Serim et al. 2010) to before and after FHR and Doppler monitoring (O'Reardon et al. 2011).
Although most studies had no follow-up data on the children, some had sparsely noted follow-up at 1 month (Repke and Berger 1984), 3 months (Yellowlees andPage 1990), 18 months (O'Reardon et al. 2011), 2 weeks to 5 months (Sobel 1960), 2½years (Yamamoto et al. 1953), and 6 years (Evrard 1961). A more detailed follow-up study from 1955 by Forssman (1955) of 16 children, whose mothers were given ECT during pregnancy between years 1947 and 1952, was excluded since it contained only data on the children without any ECT during pregnancy data on the mothers.
ECT risk and adverse events No deaths of mother/ECT treated pregnant patient were found in any studies. Overall (all years), child mortality rate was 7.1 % (12/169), and from 1970 to 2013 mortality rate was 9.4 % (5/54) and from 1942 to 1970, 6.1 % (7/115) (see Table 1). Lethal outcomes for the fetus and/or baby were stated to have diverse causes, in one case a long lasting severe grand mal seizure (status epilepticus) induced by ECT (Balki et al. 2006). A combination of insulin coma treatment and ECT was found for 3 early studies in the period 1946 to 1954 by Kent (1947), Gralnick (1946), Wickes (1954)-all with severe very adverse outcome for the fetus/baby. Overview of all reported adverse events for ECT treated pregnant women and fetus and/or baby child are given in Table 4.

Main findings
Altogether 169 ECT treated pregnant women of mean age 29 years, were identified. They were treated with mean number of ECTs 9.4, as treatment for mainly (62 %) severe "psychotic" depression/bipolar disorder. Half (53 %) of pregnant women received ECT during the 2nd trimester. ECT in the 1st trimester was not uncommon (16 %) and for some, ECT was conducted throughout the entire pregnancy. Altogether, 24 % women were nullipara (Para1). Fetus and/or baby report was found missing for 12 %. Child mortality rate was overall (all years) 7.1 %. A total of 67 adverse events were found among 169 women (rate, 0.40). Most common adverse event for mother was premature labor (born between 29 and 37 GW) 19/67 (28 %) and tocolytic treatment often noted. A total of 35 adverse events were found among the reported 148 fetus/baby children (rate 0.24). The most common reported adverse event for fetus/baby child occurring during the ECT intervention was reduced FHR 15/35 (43 %). Whether the reduced FHR event is attributable to the ECT intervention per se or to the anesthetic agent or to both is not possible to say from such descriptive case studies. Due to the complexity of the ECT indication, the intervention per se, previous or concomitant psychotropic medication or other complicating somatic or genetic factors, direct causal inference is not possible to take from case studies. This being said though, having in mind that the risk of fetal malformation in WWE is twofold higher (Artama et al. 2006), and caesarian section performed more often among WWE (Olafsson et al. 1998), the potential risk involved with ECT induced epileptogenic seizures must in each case be considered. Such as illustrated in the recent publication by De Asis et al. (2013), where the ECT induced prolonged seizure duration occurred alongside severe reduced FHR and emergency Caesarian section prepared, but later abandoned when the FHR returned to normal. An earlier study (Balki et al. 2006) also reports severe ECT induced status epilepticus with lethal outcome for the fetus/child.
As for the overall occurrence of serious adverse events, such as stillbirth/neonatal death 8/35 (23 %) and fetal malformation 7/35 (20 %), the rates appear higher than that reported in the general population, i.e. 2.3 % major congenital abnormalities and 80 % live births (2010) and 0.6 % congenital heart disease (Ermis and Morales 2011). Some included studies though claim the miscarriage rate not to be higher than in the general population (Malhotra et al. 2008) and ECT to be less risky than pharmacological treatment (Kasar et al. 2007). However, figures from case studies cannot directly be compared with figures from large observational prevalence studies. This being said, close monitoring of mother and fetus during and after ECT treatment taking into regard the trimester Electroconvulsive therapy during pregnancy: a systematic review situation, is crucial to bear in mind, such as use of cardiotocography, ultrasound between treatments, tilt position for mother including tocolytic treatment to prevent preterm labor. All these monitoring factors varied greatly in the included studies. Direct effect of anesthetic agents on the fetus is still relatively unknown (Iwasaki et al. 2002). FHR variability and reduction under the ECT intervention is often mentioned as something to expect to happen. Propofol's known associated risk of bradycardia calls for alertness from a fetal cardiovascular viewpoint and extra caution is needed where the fetus is immature or has cardiovascular complications. Thiopental (22 %), methohexital (15 %), and propofol (17 %) are the most used anesthetic agents. However, case studies with both anesthesia in favor of propofol (De Asis et al. 2013) and that against it (Iwasaki et al. 2002) are published.
Some factors to bear in mind in the different pregnancy trimesters are mentioned below: 1st trimester Knowledge about when and how to administer ECT in early pregnancy, in order to reduce risk for both mother and fetus, is limited. Cardiotocography monitoring for the fetus, in this early period (before 24 GW) is not so feasible (Lovas et al. 2011). Risk of post ECT vaginal bleeding (indicative of abruptio placenta) and abortion (Vanelle et al. 1991) is mentioned. The complexity of any causal attribution to ECT is illustrated in the case by Yang (Yang et al. 2011) reporting congenital hyaline membrane disease and hypertrophic pyloric stenosis in a premature baby delivered by emergency section, since the mother had been treated with an extensive amount of antipsychotic and antidepressant medication prior to admission due to a 15 year long history of schizophrenia. 2nd trimester Transient FHR reduction (bradycardia) arising during the ECT and subsiding afterwards is commonly reported from this trimester period, likewise post-ECT uterine contractions. The need for both pre-and post-ECT tocolytic treatment in order to avoid preterm labor is considerable (Fukuchi et al. 2003;Malhotra et al. 2008;Polster and Wisner 1999;Prieto Martin et al. 2006;Serim et al. 2010;Yang et al. 2011). 3rd trimester Tilt position is recommended by several, especially in the last trimester in order to reduce risk of gastric reflux. Also inhalation anesthesia is pointed out by Ishikawa et al. (2001) to be beneficial in the last stages of pregnancy in order to reduce uterine contraction and potential uterine relaxation effect of anesthetics.
The overall total number of included studies (N =67) in our review is larger than the 57 by Anderson and Reti (2009). However, overall total number of ECT treated pregnant women (N =169) is much less than the 339 by the same authors (Anderson and Reti 2009). Unlike the Anderson and Reti (2009), numbers of ECT treated pregnant women referred to by others in the general text of the case article, have not been included in this review. Strictly according to the predetermined review criteria, only direct case reports by the study authors are included in the total count number (169) of pregnant ECT treated women by us. For example, only one case is included in this review from the publication by Impastato et al. (1964) as opposed to 159 cases by Anderson and Reti (2009), and we have not included the Forssman (1955) follow-up of 16 infants/children on ECT treated mothers, since this study contains no ECT pregnancy data, i.e. data on the mothers treatment. Likewise the study by Levine and Frost (1975) is excluded by us, since it only contained information about anesthesia type and cardiovascular responses to ECT in a 3rd semester pregnancy and no other information.
Previous studies, such as that by O'Reardon et al. have regarded ECT to be relatively safe during all trimesters of pregnancy. Contrary to this standpoint, our review and overview of recorded adverse events from all case studies call for great clinical caution. Voices of concern, similar to ours, appear also in the included study Pinette et al. (2007) and APA statements regarding ECT as a safe intervention during pregnancy questioned. The previous held opinion by the Miller (1994) review concerning potential complications from ECT during pregnancy to be minimized by improved technique, are also questioned by our results.

Check lists
The study by Salzbrenner et al. (2011) provides a 10-point checklist for pregnant women undergoing ECT. Similarly, a 14-item list for general measures and routine anesthetic measures in order to avoid gastric reflux is provided by O' Reardon et al. (2011). The need for close clinical collaboration between gynecology/obstetrics, anesthesiology and psychiatry together with clear responsibility is evident. Textbook checklists for when ECT is an option during pregnancy (Stewart and Erlick Robinson 2001) need updating of potential risks to be considered.
Our results reveal that all potential risk arising from the complexity of ECT intervention, the grand mal seizure, anesthetic and concomitant or previous psychotropic medications, is of great concern and must be taken into account for both mother and fetus/child, and weighed against the clinical benefits, when deciding to administer ECT during pregnancy.

Ethical issues
Ethical considerations and possible ethical violations for both mother and the unborn non-consenting child are not discussed. Conflicting opinions can easily arise, such as that described by Polster and Wisner (1999) where the obstetrician advised that ECT be discontinued after premature labor treatment in the obstetrics unit, but ECT was continued by the psychiatric unit. All arguments from this review support the need for holistic clinical decision making and caution when ECT is considered as an option during pregnancy.

Strengths and limitations
The strength of this paper is the thorough, systematic review of all published literature without any data limitation. Data extracted from the included studies have strictly been limited to primary case presentations by the authors and not secondary "known to the authors" numbers referred to by the authors in the body text. Likewise all other literature review studies on the subject without any primary case data have also been excluded. The most consistent findings in all included studies was the number of ECTs administered, thereafter the diagnostic indication, pregnancy length, ECT parameters, anesthesia type, condition of both mother and child, the latter was somewhat more dependable in newer date studies. The strength of case study design is the reporting of rare and adverse events, however limitations as for this design must clearly be taken into account.
A limitation is uncertainty in the very oldest published cases, where case presentation is mixed with cases "known to authors" in the manuscript text, to completely document all cases since the introduction of ECT in 1938. The earliest published case reports are also much more likely to be mixed with other treatment forms, such as insulin coma, which is not used and out of date today and these mixed treatment reports therefore not so relevant for today's practice. No prospective or controlled study design of ECT in pregnancy are found, case studies alone in this field provide the knowledge background. Case studies are susceptible to reporting and publication bias, and only descriptive aggregation of study data is possible, no meta-analyses. As cases of ECT during pregnancy where the treatment went well are most likely not published, the included studies in this review might very well be over represented with adverse event reporting.

Clinical implications
ECT during pregnancy should be a last resort treatment. For example in cases of severe depression, catatonia, medication resistant illness, extremely high suicide risk, psychotic agitation, severe physical decline due to malnutrition or dehydration or other life threatening conditions (for example malignant neuroleptic syndrome), where other treatment options are not possible or very inadequate. All potential risks of the ECT treatment, taking into account both mother and fetus, should be weighed against benefits. The ECT should be administered in a hospital emergency setting or delivery room. Information to patients of all possible risks involved should be considered compulsory. ECT during pregnancy should be administered by a highly skilled and competent specialist team consisting of psychiatrist, gynecologist/obstetrician, and anesthesiologist. Monitoring of patient under ECT treatment and also in the recovery room should include midwife and psychiatric nurse. The establishment of a multi-disciplinary specialist team bearing full treatment and follow-up responsibility is fundamental for the safety of the intervention.

Conclusions
Case reports on ECT administered during pregnancy provide vital knowledge. ECT during pregnancy is advised considered only under very stringent diagnostic and clinical indications, weighing all potential risks against benefits. Updated clinical guidelines are urgently needed in this field. Acknowledgments This study has been possible because of research commissioning on the topic "ECT for depression" from the Norwegian Directorate of Health to the Norwegian Knowledge Centre. We thank the Norwegian Knowledge Centre's research librarian Karianne Thune Hammerstrøm (KTH) for designing and undertaking the first primary literature search.

Competing interests None.
Funding statement This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited. Electroconvulsive therapy Topic=(Electroconvulsive Therapy or electroshock* or "electr* shock"*) AND Topic=(pregnan* or gestation* or gravid* or antenatal* or prenatal* or perinatal*) 2 Timespan=1975-2012 Table 7 Excluded studies (N =33) First author (year published) Comments and reason for exclusion: (1) about ECT, but not in pregnancy, e.g., in postpartum or other conditions; (2) commentary, no primary data, too sparse data, review without primary data, letter to editor; (3) parallel other language publication, not possible to find or full text retrieve; and (4) not relevant topic, about anesthesia types or other topic, e.g., electrical shock injury in pregnancy Bader et al. (2010) (2) No study data Passov (2010) (2) Conference abstract about 2 cases of ECT in pregnancy, insufficient data Pinette and Wax (2010) (2) Letter to editor, without study data Anderson and Reti (2009) (2) Literature review, not primary study Nielsen et al. (2007) (2) Literature review, not primary study Richards (2007) (2) Editorial, not primary study Maletzky (2004) (1) About ECT, but not pregnancy Ginsberg (2007) (2) Commentary about another article by Pinette et al. (2007) Howe andSrinivasan (1999) (1) About Cotard's Syndrome, ECT given in postpartum after delivery by cesarean section Berle (1999) (1) Four cases of severe postpartum depression, ECT given in postpartum Cutajar et al. (1998) (1) Case of severe depression in young woman with mild learning disabilities, given ECT in the post-partum period Ratan and Friedman (1997) (1) About Capgras syndrome in puerperium, ECT given in postpartum period Anonymous (1997) (2) Editorial commentary, no primary author, about electrical shock injury Johnson (1996) (1) Case of mania in pregnancy, ECT given in postpartum period Finnerty et al. (1996) (1) Case 33 years, pregnant (para 3) with bipolar disorder. ECT was planned given during pregnancy but due to spontaneous rupture of membranes and Caesarian section at 29 gestation weeks (baby reported ok), ECT was administered in postpartum period. Bernardo et al. (1996) (1) Imaginary pregnancy, not pregnant Bruggeman and de Waart (1994) (2) Letter to editor about another article Eskes and Nijhuis (1994) (2) Commentary to case study by Verwiel et al. (1994) Yoong (1990 (4) Not about ECT, but electrical shock injury and baby died 24 hours after delivery Kramer (1990) (2) Letter to editor about use of ECT in pregnancy Sneddon and Kerry (1984) (1) 55 cases of puerperal psychosis treated with ECT in postpartum Raty-Vohsen (1982) (4) General treatment of postpartum psychoses Levine and Frost (1975) (4) Only about anesthesia type and cardiovascular responses to ECT in 3rd semester pregnancy Remick and Maurice (1978) (2) Letter to editor, without study data Cohn et al. (1977) (1) About postpartum Protheroe (1969) (1) Puerperal psychoses follow-up study and ECT given in postpartum Anderson (1968) (2) Dissertation abstract Marcelino Da Silva and Alexandre (1950) (3) Not able to retrieve/find Impastato and Gabriel (1957) (1) About ECT in postpartum Forssman (1955) (4) Not relevant topic, only information on follow-up of 16 children whose mothers were given ECT in pregnancy between years 1947 to 1952 Forssman (1954) (3) Parallel publication in Swedish to English article of later date by Forssman (1955) Stone and Walker (1949) (4) Article not human (rats) study data Walker (1992) (3) Same clinical case presented as in article by Livingston et al. (1994)            At 31 weeks tocolytic treatment with terbutaline.
At 34 weeks observation in delivery suite needed due to bleeding. Spontaneous labor 37 GW and caesarian section performed FHR reduction after 1st ECT Baby boy, 2,704 g Apgar 3 and 9 Large retro-placental clot confirming abruption placentae diagnoses     Fetal damage among ECT treated is reported as 6 %but type of damage not specified.
Overall sparse data and unclear.
Fetal abnormality 6 % is commented as surprisingly high -and data otherwise lacking.