Human Reproduction Update Advance Access published November 17,2016 Human Reproduction Update, pp. I-1 I, 2016 doi: 10.1093/humupd/dmw043 human reproduction update Interpregnancy interval following miscarriage and adverse pregnancy outcomes: systematic review and meta-analysis Chrishny Kangatharan1, Saffi Lab ram2, and Sohinee Bhattacharya 2,* Department of Medical Education, University of Malta St Paul's Street, Valletta VLTI2I 6, Malta Institute of Applied Health Sciences, University of Aberdeen, Foresterhill Health Campus, Aberdeen AB25 2ZL, UK correspondence address. Obstetric Epidemiology, Dugald Baird Centre for Research on Women's Health, Aberdeen Maternity Hospital, Aberdeen AB25 2ZL, UK. Tel: +44-1224-438441; E-mail: sohinee.bhattacharya@abdn.ac.uk Submitted on May 23, 20/6; resubmitted on September 14, 20/6; editorial decision on October 27, 20/6; accepted on October 31, 2016 TABLE OF CONTENTS • Introduction • Methods Ethical approval Review protocol Literature search Review methods Quality assessment and risk of bias Statistical analysis • Results Further miscarriage Preterm birth Live birth Stillbirth Low birthweight Preeclampsia • Discussion • Conclusion BACKGROUND: A short interpregnancy interval (IPI) following a delivery is believed to be associated with adverse outcomes in the next pregnancy. The optimum IPI following miscarriage is controversial. Based on a single large-scale study in Latin and South America, the World Health Organization recommends delaying pregnancy for 6 months after a miscarriage to achieve optimal outcomes in the next pregnancy. OBJECTIVE AND RATIONALE: Our aim was to determine if a short IPI (<6 months) following miscarriage is associated with adverse outcomes in the next pregnancy. SEARCH METHODS: Studies were retrieved from MEDLINE, Embase and Pubmed, with no time and language restrictions. The search strategy used a combination of Medical Subject Headings terms for miscarriage, IPI and adverse outcomes. Bibliographies of the retrieved articles were also searched by hand. All studies including women with at least one miscarriage, comparing subsequent adverse pregnancy outcomes for I Pis of less than and more than 6 months were included. Two independent reviewers screened titles and abstracts for © The Author 20 16. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please email: journals.permissions@oup.com 2 Kangatharan et al inclusion. Characteristics of the studies were extracted and quality assessed using Critical Appraisal Skills Programme criteria. A systematic review and meta-analysis were conducted to compare short (<6 months) versus long (>6 months) I PI following miscarriage in terms of risk of further miscarriage, preterm birth, stillbirth, preeclampsia and low birthweight babies in the subsequent pregnancy. Review Manager 5.3 was used for conducting meta-analyses. OUTCOMES: Sixteen studies including I 043 840 women were included in the systematic review and data from 10 of these were included in one or more meta-analyses (977 972 women). With an IPI of less than 6 months, the overall risk of further miscarriage (Risk ratio (RR) 0.82 95% CI 0.78, 0.86) and preterm delivery (RR 0.79 95% CI 0.75, 0.83) were significantly reduced. The pooled risks of stillbirth (RR 0.88 95% CI 0.76, 1.02); low birthweight (RR 1.05 95% CI 0.48, 2.29) and preeclampsia (RR 0.95 95% CI 0.88, 1.02) were not affected by IPI. Similar findings were obtained in subgroup analyses when IPI of <6 months was compared with IPI of 6-12 months and > 12 months. WIDER IMPLICATIONS: This is the first systematic review and meta-analysis providing clear evidence that an IPI of less than 6 months following miscarriage is not associated with adverse outcomes in the next pregnancy. This information may be used to revise current guidance. Key words: interpregnancy interval / miscarriage / recurrent miscarriage / pregnancy outcomes / preterm birth / live birth / stillbirth / low birthweight / preeclampsia Introduction Miscarriage is a relatively common occurrence, affecting 10-15% of all pregnancies in the UK (Bhattacharya et al, 2008). It is defined as any pregnancy loss that occurs in the first 24 weeks (Bhattacharya et al, 2008), although the gestational week cut off varies according to availability of neonatal care. Loss of a pregnancy through miscarriage is associated not only with psychological distress but may also affect the outcomes of the subsequent pregnancy resulting in further miscarriage, preeclampsia and preterm delivery (Bhattacharya et al, 2008). Birth spacing after an initial miscarriage may help mitigate some of these risks. The time between the end of a pregnancy and the start of another one is defined as the interpregnancy interval (IPI) (Bentolila et al, 2013). The optimum IPI after a live birth has been reported to be 18-23 months, for better maternal and perinatal outcomes in the next pregnancy (Conde-Agudelo et al, 2006). In their meta-analysis of observational studies, Conde Agudelo et al (2006) found J shaped associations between IPI following a live birth and adverse outcomes in the subsequent pregnancy. Intervals shorter than 20 months and longer than 60 months conferred the highest risk of preterm birth, low birthweight, and small for gestational age; while intervals shorter than 6 months and longer than 50 months were associated with the highest risk of perinatal deaths. The optimum IPI after a miscarriage is, however, controversial. Some clinicians advise couples not to delay conceiving the next pregnancy, as an increasing IPI after a miscarriage does not appear to improve birth outcomes (Basso et al, 1998; Goldstein et al, 2002; Love et al, 2010). Others suggest delaying pregnancy for at least 18 months based on the optimum IPI after a live birth (Conde-Agudelo et al, 2006). The World Health Organization (WHO) guidelines recommend waiting for at least 6 months before trying to conceive again after a miscarriage (WHO, 2005). These guidelines were based on a single multicentre study in Latin and South America, which found that an IPI of less than 6 months following miscarriage was associated with adverse outcomes in the next pregnancy (Conde-Agudelo et al, 2004). This study however, was unable to distinguish between miscarriage and induced abortion and this may have affected their findings. As increased maternal age is independently associated with increased risk of miscarriage (Aref-Adib et al, 2008), delaying conception after a miscarriage may further increase this risk. We therefore performed a systematic review with meta-analyses looking at the relationship between a short IPI (less than 6 months) compared to 6 months or more following a miscarriage and adverse outcomes in the next pregnancy. Methods Ethical approval As this study was a systematic review and meta-analysis of aggregated published data, formal ethical approval was not required. Review protocol At first a specific protocol was designed where the review question was formulated using the Population, Exposure, Comparison and Outcome (PECO) format. The population (P) of interest was women with at least one pregnancy following a miscarriage, exposure (E) was IPI of less than 6 months compared (C) to IPI of 6 months or more. The pre-specified outcomes (O) of interest were further miscarriage, preterm birth, stillbirth, pre-eclampsia and low birthweight in the pregnancy following miscarriage. All types of study design were assessed for eligibility. The criteria used to identify, include and exclude studies and the methods for analysing data were all derived from this format and agreed a priori in the review protocol. The review was conducted and reported according to the guidelines of the Meta-analysis of Observational studies in Epidemiology group (MOOSE checklist). The protocol was registered with PROSPERO (registration number CRD42016038424). Literature search A search strategy was initially developed in Ovid Medline then modified and run in other databases - PubMed (U.S. National Library of Medicine), Embase (Elsevier) and Scopus. The search strategy used a combination of Medical Subject Headings (MeSH) terms for miscarriage, IPI and adverse outcomes. The terms for miscarriage were: miscarriages, abortion, Optimum interpregnancy interval after miscarriage 3 spontaneous abortion, early pregnancy loss. Other terms for I PI were inter-conception interval, time to birth, birth spacing and birth interval. Terms for adverse outcomes were pregnancy outcomes, adverse outcomes. A further search was conducted using specific terms for IPI: long IPI, short IPI, more than 6 months IPI, less than 6 months IPI. A specific search was also conducted for the names of each adverse outcome, these terms were: further miscarriage, pregnancy loss, stillbirth, preterm birth, low birth-weight, pre-eclampsia. These search terms were combined using Boolean operators 'AND' or 'OR' as appropriate. No time or language restrictions were applied to the search strategy. Two reviewers (C.K. and S.L.) independently ran the searches. Review methods The titles and abstracts of the articles identified by this search were independently screened by two reviewers (C.K. and S.L.) for inclusion in the review and the full texts of those that appeared relevant were retrieved. Bibliographies of the retrieved articles were also searched by hand. Where there was inadequate information in the published article, authors were contacted to request additional data. All the retrieved full text articles were then assessed for inclusion in the review using the predefined exclusion and inclusion criteria. The criteria determining whether an article was going to be included were: (i) If the populations studied were women with at least one miscarriage. The studies with women with no miscarriage but just live births or induced abortions were excluded. (ii) If the studies used IPI as exposure. Studies were excluded if they did not include IPI or the women did not have any further pregnancies. (iii) If they had studied I Pis for less and more than 6 months. Studies were excluded if they did not have comparison groups or did not report findings for I Pis of less than 6 months. Nevertheless, authors were contacted to see if they could provide appropriate data if the range of IPI was inconsistent with this inclusion criterion. (iv) If the studies had the outcomes that were relevant to this review. Outcomes were broadly categorised into primary and secondary outcomes based on frequency and consistency of association reported in the literature, biological plausibility and clinical importance. Primary outcomes were defined as further miscarriage (less than 24 weeks of gestation) and preterm delivery (delivery before 37 weeks of gestation). Secondary outcomes were live birth, stillbirth, pre-eclampsia, and low birthweight. Studies were included if they had adverse outcomes in the next pregnancy and excluded if they only reported adverse outcomes in the same pregnancy. Studies were also excluded if they were case reports, reviews or editorials. Quality assessment and risk of bias Once the potentially eligible articles were retrieved, they were assessed for methodological quality using the Critical Appraisal Skills Programme (CASP) checklist for cohort studies (Critical Appraisal Skills Programme (CASP), 2016). The following were extracted from each included article: titles, authors' names, the type of study, characteristics of the population studied, the setting of the study (the geographical location), the outcomes studied, the measured exposure IPI. Statistical analysis Meta-analysis was performed where appropriate using the software Review Manager 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2014. Copenhagen, Denmark). Data were entered for each outcome if there were at least two studies addressing that outcome. The raw numbers for each outcome in each group of IPI (>6 months or <6 months) as reported in the primary studies were entered in the software to calculate the crude risk ratio (RR) and the 95% confidence interval (CI) using >6 months as the reference category. These were then weighted and pooled to produce forest plots and pooled RRs with 95% CI. Statistical heterogeneity was assessed using the P statistic. Where I2 was more than 50% signifying moderate to large statistical heterogeneity, a random effects model was used. If a study varied significantly in terms of methodology or findings from all other included studies, we performed a sensitivity analysis excluding those studies from the meta-analysis. In subgroup analyses, we split the comparator group of >6 months into 6-12 months and > 12 months for the primary outcomes of further miscarriage and preterm birth. Results Figure I shows the process for the search and identification of studies. The bibliographic searches identified 151 publications and I 8 others were found from a hand search of the references. Of these, 38 publications were considered relevant and the full text reviewed for inclusion. Of these, 13 cohort studies (Wyss et ai, 1994; Basso et ai, 998; Goldstein et ai, 2002; Buchmayer et ai, 2004; Conde-Agudelo et ai, 2004; DaVanzo et ai, 2007, 2012; Cox et ai, 2010; Love et ai, 2010; Morgan-Ortiz etai, 2010; Bentolila etai, 2013; El Behery et ai, 2013; Sapra et ai, 2014) and three RCTs (Kaandorp et ai, 2014; Makhlouf et ai, 2014; Wong et ai, 2015) met the inclusion criteria. However, six of these articles had insufficient data for inclusion in meta-analysis; the authors of these papers were contacted but were unable to provide additional data. Therefore, 10 (Wyss et ai, 1994; Buchmayer et ai, 2004; Conde-Agudelo et ai, 2004; DaVanzo et ai, 2007, 2012; Love et ai, 2010; Morgan-Ortiz et ai, 2010; Bentolila et ai, 2013; Makhlouf et ai, 2014; Wong et ai, 2015) studies were included in the meta-analyses. Table I shows the characteristics of the included studies (13 cohort and 3 RCTs) along with their quality assessment scores. The authors also carried out a secondary cohort analysis of the women in the three RCTs to look at the effect of a short IPI after a previous loss (Kaandorp et ai, 2014; Makhlouf et ai, 2014; Wong et ai, 2015). Out of the 16 studies, four were set in the USA (Goldstein et ai, 2002; Makhlouf et ai, 2014; Sapra et ai, 2014; Wong et ai, 2015), two in Bangladesh (DaVanzo et ai, 2007, 2012), two in the Netherlands (Cox et ai, 2010; Kaandorp et ai, 2014) and one each in Scotland (Love et ai, 2010), Denmark (Basso et at, 1998), Sweden (Buchmayer et ai, 2004), Egypt (El Behery et ai, 2013) Israel (Bentolila et ai, 2013), Switzerland (Wyss et ai, 1994), Uruguay (Conde-Agudelo et ai, 2004) and Spain (Morgan-Ortiz et ai, 2010). Most studies looked at IPI in months, while two studies looked at IPI in terms of menstrual cycles in days (Goldstein et ai, 2002; Sapra et ai, 2014). All the studies used a population of women with one miscarriage or recurrent miscarriages. Eight studies provided data on preterm birth (Wyss et ai, 1994; Buchmayer et ai, 2004; Conde-Agudelo et ai, 2004; Love et ai, 2010; Morgan-Ortiz et ai, 2010; Bentolila et ai, 2013; Makhlouf et ai, 2014; Wong et ai, 2015), seven on further miscarriage (Wyss et ai, 1994; DaVanzo et ai, 2007, 2012; Love et ai, 2010; 4 Kangatharan et ai c o % LI ■o CJ 8 Records identified through Additional records identified catasase searching through other sources (n - 1S1) (n ■ 18J from reference lists r 1 Records after duplicates removed :- ■ inj [ Records screened (rt • 88] Records excluded (n • 51) o o 3 Full-text articles assessed for eligibility (n - 38) Studies included in Qualitative synthesis [n - 16) I Studies included in quantitative synthesis (meta-analysis) (n - 10) Full-text articles excluded, with reasons (n - 22) Reasons: reviews (4). insufficient data (2), no mterpregnancy interval (8), OS miscarriages (4). editorials (2), other outcomes (2) Figure I Flow diagram of study identification and selection process for systematic review of the association between interpgnancy interval following miscarriage and subsequent pregnancy outcomes. Morgan-Ortiz et ai, 2010; Bentolila et ai, 2013; Wong et ai, 2015), four on live births (DaVanzo et ai, 2007, 2012; Love et ai, 2010; Wong et ai, 2015), four on stillbirths (DaVanzo et ai, 2007, 2012; Love etai, 2010; Wong et ai, 2015), five on preeclampsia (Conde-Agudelo et ai, 2004; Love et ai, 2010; Bentolila et ai, 2013; Makhlouf et ai, 2014; Wong et ai, 201 5) and four on low birthweight (Conde-Agudelo et ai, 2004; Love et ai, 2010; Bentolila et ai, 2013; Makhlouf et ai, 2014). The study by Conde-Agudelo et ai (2004) did not distinguish between spontaneous and induced abortions and a sensitivity analysis was performed including and excluding this study. The average quality assessment score using CASP criteria was 9.4 out of I I, therefore all the included studies were of good quality with low risk of bias. Publication bias was investigated using a funnel plot for the outcome further miscarriage but showed no appreciable evidence of this bias (PI Supplementary Fig. SI). Further miscarriage Seven of the 10 studies provided data on further miscarriage after a previous miscarriage. The risk of having a further miscarriage with IPI of less than 6 months was significantly reduced when compared to IPI of more than 6 months, with a pooled RR (95% CI) of 0.82 (0.78, Table I Characteristics and quality of 16 studies included in a systematic review on interpregnancy interval following miscarriage and adverse pregnancy outcomes. Reference Design Setting Population Exposure (IPI) Outcome Confounders QA Score Wong et al. (2015) Kaandorp et al. (2014) Makhlouf et al. (2014) Sapra et al. (2014) Bentolila et al. (2013) DaVanzo et al. (2012) El Behery et al. (2013) Love et al. (2010) RCT/analysed Four clinical trial sites as cohort in USA RCT/cohort RCT/cohort Cohort Cohort Cohort Cohort Cohort Morgan-Ortiz Cohort et al. (2010) Cox et al. Cohort (2010) DaVanzo et al. Cohort (2007) Conde-Agudelo Cohort et al. (2004) Buchmayer et al. Cohort (2004) ALIFE trial Netherlands (2004-2009) Eunice Kennedy Shriver National Institute RCT (2003-2008) Michigan and Texas, USA (2005-2009) RPL clinic in the Soroka University Medical Center, Israel Matlab DHSS Bangladesh (1977-2008) Zagazig and Suez, Canal University Hospitals (2009 to 2012) Scotland (1981-2000) Mexico 38 fertility centres in the Netherlands Matlab, Bangladesh (1982-2002) Latin and South America (1985-2002) Sweden (1987-2000) Women with > I previous miscarriage Women with unexplained recurrent miscarriage Women with previous miscarriage Women with miscarriage Women with 2 or more consecutive miscarriage Women with miscarriage Women with first pregnancy miscarriage 3 monthly intervals 0 to > 12 6, 12 and 24 months <6, 6-12, > 12 months No. of menstrual cycles <6 and >6 months 3 and 6 month intervals <6 months and > 12 months Women with first pregnancy miscarriage 6 monthly intervals from <6 to >24 Women with early 6 months pregnancy loss in last pregnancy Women with > I previous 6-18 months miscarriage All pregnancies including miscarriage Women delivering singleton with previous history of abortion (spontaneous or induced). Women with previous pregnancy loss <6, 6-14, 15-26, 27-50, 51-74 and >74 months IPI (in months): <2, 3-5, 6-1 I, 12-17, 18-23, 24-59, >60 0-3, 3-6, 6-12 and > 12 intervals Live birth; pregnancy loss Weeks to conception; time to live birth Preterm birth, pre-eclampsia, foetal/ neonatal death, birthweight Pregnancy Adverse outcomes in the next pregnancy Miscarriage, termination; stillbirth; early, late and post neonatal mortality Miscarriage, ectopic, termination, stillbirth, live birth, pre-eclampsia, placenta praevia, abruption, PPH, low birthweight, preterm delivery Miscarriage, ectopic, live birth, stillbirth; preeclampsia, placenta praevia, placental abruption, induction of labour, caesarean, preterm, low birthweight Further miscarriage, preterm birth and perinatal outcomes: agpar <7 Spontaneous ongoing pregnancy Live birth, stillbirth, miscarriage Multiple adverse pregnancy outcomes Preterm delivery Age, BMI, race, gestational age I I of previous loss Age, BMI, no. of miscarriages, 7 intervention, previous live birth, factor V Leiden mutation Age, BMI, race, smoking, education, I I marital status Age, BMI, smoking, caffeine and 8 alcohol intake Age, ethnicity I I Age, education, geographic area, gravidity, calendar year Age, BMI, smoking, voluntary/ involuntary IPI, gynaecological history Age, social class, smoking, calendar year None Age, duration of subfertility, sperm motility, post-coital test Age, parity, education, household space, religion, planned pregnancy, calendar year Age, parity, education, marital status, smoking BMI, gestational weight gain, geographic area, hospital type, calendar year Age, relationship with father, smoking, mother's birth country, calendar year 10 10 Continued 9I0Z '0Z -raqureAON uo oo]J3)e^\ jo ÄiisjsAiuf) )e /§jo'S]Eiunorpjojxo'pdnraraj//:d))ij raojj pspEojuMOQ 6 Kangatharan et al. < E c Q D N o QJ C o o OJ QJ QJ Golds (2002; Basso (1998; Wyss (1994; < a 0.86) (Fig. 2A). Compared to an IPI of 6-12 months, IPI of <6 months reduced the risk of further miscarriage (pooled RR 0.82, 95% CI 0.77, 0.88). Similarly this risk was further reduced (pooled RR 0.78, 95% CI 0.74, 0.83) when compared with IPI > 12 months. Preterm birth Out of the 10 studies included in meta-analysis, eight reported on preterm deliveries. We performed a meta-analysis including and excluding the study by Conde-Agudelo et al. (2004). The metaanalysis including the study by Conde-Agudelo et al. (2004) resulted in a pooled RR of 0.93(95% CI 0.58, 1.48) (Fig. 2B). The incidence of preterm deliveries was significantly lower (P < 0.01) when women with IPI of less than 6 months were compared to those with an IPI of more than 6 months: pooled RR (95% CI) of 0.79 (0.75, 0.83) (Fig. 2B) when the study by Conde-Agudelo et al. (2004) was excluded. There was no significant increase in the risk of preterm birth when compared with IPI of 6 to 12 months (pooled RR 1. 10, 95% CI 0.64, 1.89) or with IPI of > I 2 months (pooled RR 1.06, 95% CI 0.57, 1.97). The study by Conde-Agudelo et al. (2004) was included in the latter two meta-analyses. Live birth Four studies presented data on live births after a miscarriage. Live births were observed to be significantly higher when women had an IPI of less than 6 months after a miscarriage (P < 0.01), 40% higher compared to an IPI of 6 months or more, RR of (95% CI) 1.06 (1.01, I.I I) (Fig. 2C). Stillbirth The reported risk of stillbirths in women after a miscarriage was not significantly different in the two IPI groups (P = 0.09) RR (95% CI) of 0.88 (0.76, 1.02). The risk varied from 1.56 to 0.71 across the four studies included in the meta-analysis (Fig. 2D). Low birthweight Four studies presented data on low birthweight, three of the studies defined low birthweight as less than 2500 g (Conde-Agudelo et al., 2004; Love et al., 2010; Bentolila et al., 2013) and I as less than the fifth percentile for gestational age adjusted by sex and race (Makhlouf et al., 2014). The overall risk of having low birthweight babies after a miscarriage was not significantly different in women with an IPI of less than 6 months (P = 0.07), compared to women with an IPI of 6 months or more including the study by Conde-Agudelo et al. (2004) RR (95% CI) of 1.05(0.48, 2.29) (Fig. 2E). When this study was excluded, the risk of low birthweight was significantly lower with IPI of <6 months (pooled RR 0.74 95% CI 0.68, 0.81) (Fig. 2E lower panel). Pre-eclampsia The rate of preeclampsia did not appear to differ in women with IPI of less than 6 months after a miscarriage compared to IPI >6 months, including the study by Conde-Agudelo et al. (2004) pooled RR (95% CI) of 0.95 (0.88, 1.02) (Fig. 2F) and excluding the study 1.00 (0.90, Optimum interpregnancy interval after miscarriage 7 Less than 6 months 6 months or more Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year Wyss 1994 91 443 19 101 1.1% 1.09 [0.70, 1.70] 1994 DaVanzo 2007 112 1366 95 1171 3.6% 1.01 [0.78, 1.31] 2007 Love 2010 1308 12744 2311 18193 67.8% 0.81 [0.76, 0.86] 2010 Morgan-Ortiz 2010 42 286 76 572 1.8% 1.11 [0.78, 1.57] 2010 DaVanzo 2012 422 4596 681 5839 21.4% 0.79 [0.70, 0.88] 2012 Bentolila 2013 21 113 63 212 1.6% 0.63 [0.40, 0.97] 2013 Wong 2015 96 445 57 232 2.7% 0.88 [0.66, 1.17] 2015 Total (95% CI) 19993 26320 100.0% 0.82 [0.78, 0.86] Total events 2092 3302 Heterogeneity: Chi2 = 9.22, df = 6(P = 0.16); l2 = 35% Test for overall effect Z = 7.51 (P < 0.00001) Risk Ratio M-H, Fixed, 95% CI 0.1 0.2 0.5 1 2 6 months or more <6 months 10 Less than 6 months 6 to 12 months Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% C Year Love 2010 1308 12744 1004 7791 74.7% 0.80 [0.74, 0.86] 2010 DaVanzo 2012 422 4596 302 2920 22.1% 0.89 [0.77, 1.02] 2012 Wong 2015 96 445 36 155 3.2% 0.93 [0.66, 1.30] 2015 Total (95% CI) 17785 10866 100.0% 0.82 [0.77, 0.88] Total events 1826 1342 Heterogeneity: Chi2 = 2.30, df = 2(P = 0.32); I2 = 13% Test for overall effect Z = 5.83 (P < 0.00001) Risk Ratio M-H, Fixed, 95% CI 0.1 0.2 0.5 1 2 6 to 12 months <6 months 10 O o Study or Subgroup Less than 6 months Events Total More than 12 months Events Total Weight Risk Ratio M-H, Fixed, 95% CI Year Love 2010 1308 12744 1307 10402 69.5% 0.82 [0.76, 0.88] 2010 DaVanzo 2012 422 4596 379 2919 22.4% 0.71 [0.62, 0.81] 2012 El Behery 2013 99 2422 117 2197 5.9% 0.77 [0.59,1.00] 2013 Wong 2015 96 445 27 77 2.2% 0.62 [0.43, 0.87] 2015 Total (95% CI) 20207 15595 100.0% 0.78 [0.74, 0.83] Total events 1925 1830 Heterogeneity: Chi2 = 5.49, df=3(P = 0.14); l2 = 45% Test for overall effect Z = 7.84 (P < 0.00001) Risk Ratio M-H, Fixed, 95% CI 0.1 0.2 0.5 1 2 >12 months <6 months 10 Figure 2 Forest plots presenting the association of interpregnancy interval (IPI) following miscarriage with subsequent pregnancy outcomes. (A) Forest plot presenting the association of I Pis following miscarriage with further miscarriage. (B) Forest plot presenting the association of I Pis following miscarriage with subsequent preterm birth. (C) Forest plot presenting the association of I Pis following miscarriage with subsequent live birth. (D) Forest plot presenting the association of I Pis following miscarriage with subsequent stillbirth. (E) Forest plot presenting the association of I Pis following miscarriage with subsequent delivery of low birthweight babies. (F) Forest plot presenting the association of I Pis following miscarriage with subsequent pre-eclampsia. 1. 12) (Fig. 2F lower panel). Five of the ten studies provided data on preeclampsia. Discussion Birth spacing is an important element of reproductive counselling. Couples experiencing a miscarriage need to know the optimal time to conceive another pregnancy in order to have the best possible outcomes. In this systematic review, we evaluated 6 different outcomes and found that an IPI of less than 6 months following a miscarriage was associated with lower risks of having a further miscarriage and preterm delivery, and increased odds of having live births. There were no differences in the risks of stillbirth, preeclampsia and low birthweight babies between an IPI of less than 6 months and of 6 months or more. Based on the published evidence from 10 studies we can therefore conclude that delaying a pregnancy for more than 6 months after a miscarriage is unnecessary as a short IPI (less than 6 months) results in no worse pregnancy outcomes but may also be associated with better outcomes in terms of a lower risk of further miscarriage and preterm birth and increased chance of live birth in the next pregnancy. This systematic review was carried out in compliance with the criteria in the MOOSE checklist. At first a focussed review question was framed using the PECO format, from which a robust search strategy and inclusion and exclusion criteria were developed. The studies were carefully assessed for quality independently by two reviewers and data extracted for meta-analyses. The meta-analysis in this review included 10 studies. The study by Conde-Agudelo et al. (2004) provided outcome data on further miscarriage, preterm delivery, low birthweight and preeclampsia. While this was a large retrospective study on which the WHO guidelines for delaying pregnancy for at least 8 Kangatharan et at. B Less than 6 months 6 months or more Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year Wyss 1994 36 443 12 101 11.3% 0.68 [0.37,1.27] 1994 Buchmayer 2004 1023 15451 1448 17128 13.9% 0.78 [0.73, 0.85] 2004 Conde Agudelo 2004 2466 11615 25842 246493 14.0% 2.03 [1.95, 2.10] 2004 Morgan-Ortiz 2010 16 286 47 572 11.7% 0.68 [0.39,1.18] 2010 Love 2010 964 10856 1571 13977 13.9% 0.79 [0.73, 0.85] 2010 Bentolila2013 14 92 21 149 11.2% 1.08 [0.58, 2.02] 2013 Makhlouf2014 31 395 60 645 12.6% 0.84 [0.56,1.28] 2014 Wong 2015 27 445 15 232 11.3% 0.94 [0.51,1.73] 2015 Total (95% CI) 39583 279297 100.0% 0.93 [0.58,1.48] Total events 4577 29016 Risk Ratio M-H, Random, 95% CI Heterogeneity: Tau2 = 0.41; Chi2 = 869.52, df = 7 (P < 0.00001); I2 Test for overall effect: Z = 0.32 (P = 0.75) ■■ 99% 0.1 0.2 0.5 1 2 6 months or more <6 months 10 Study or Subgroup Less than 6 months Events Total 6 months or more Events Total Risk Ratio Weight M-H, Random, 95% CI Year Wyss 1994 36 443 12 101 0.7% 0.68 [0.37,1.27] 1994 Buchmayer 2004 1023 15451 1448 17128 47.1% 0.78 [0.73, 0.85] 2004 2CC1 Morgan-Ortiz 2010 16 286 47 572 0.9% 0.68 [0.39,1.18] 2010 Love 2010 964 10856 1571 13977 48.2% 0.79 [0.73, 0.85] 2010 Bentolila2013 14 92 21 149 0.7% 1.08 [0.58, 2.02] 2013 Makhlouf2014 31 395 60 645 1.6% 0.84 [0.56,1.28] 2014 Wong 2015 27 445 15 232 0.7% 0.94 [0.51,1.73] 2015 Total (95% CI) 27968 32804 100.0% 0.79 [0.75, 0.83] Total events 2111 3174 Heterogeneity: Tau2 = 0.00; Chi2 = 1 .90,df = 6(P = 0.93); I2 = 0% Test for overall effect: Z = 8.81 (P < 0.00001) Risk Ratio M-H, Random, 95% CI 0.1 0.2 0.5 1 2 6 months or more <6 months 10 O o Less than 6 months 6 to 12 months Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year Conde Agudelo 2004 2466 11615 3752 39491 21.9% 2.23(2.13,2.34] 2004 Buchmayer 2004 1023 15451 541 6792 21.8% 0.83(0.75,0.92] 2004 Love 2010 964 10856 500 6158 21.8% 1.09(0.99,1.21] 2010 Makhlouf2014 31 395 19 216 17.9% 0.89(0.52,1.54] 2014 Wong 2015 27 445 12 155 16.6% 0.78(0.41,1.51] 2015 Total (95% CI) 38762 52812 100.0% 1.10 [0.64,1.89] Total events 4511 4824 Risk Ratio M-H, Random, 95% CI Heterogeneity: Tau2 = 0.35; Chi2 = 414.87, df = 4 (P < 0.00001); I2 = 99% Test for overall effect: Z = 0.34 (P = 0.73) 0.1 0.2 0.5 1 2 6 to 12 months <6 months 10 Less than 6 months More than 12 months Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year Conde Agudelo 2004 2466 11615 18113 207002 18.2% 2.43 [2.34, 2.52] 2004 Buchmayer 2004 1023 15451 907 10336 18.1% 0.75 [0.69, 0.82] 2004 Love 2010 964 10856 743 7823 18.1% 0.93 [0.85,1.02] 2010 ElBehery 2013 139 1944 162 1614 17.8% 0.71 [0.57, 0.88] 2013 Makhlouf2014 31 395 41 429 16.6% 0.82 [0.53, 1.28] 2014 Wong 2015 27 445 3 77 11.1% 1.56 [0.48, 5.01] 2015 Total (95% CI) 40706 227281 100.0% 1.06 [0.57,1.97] Total events 4650 19969 Heterogeneity: Tau2 = 0.56; Chi2 = 971.00, df=5(P < 0.00001); I2 = 99% Risk Ratio M-H, Random, 95% CI Test for overall effect: Z = 0.17 (P = 0.87) 0.1 0.2 0.5 1 2 >12 months <6 months 10 Figure 1 Continued 6 months (WHO, 2005) is based, it did not differentiate between induced and spontaneous abortions and used data from many countries where induced abortion is illegal (Conde-Agudelo et ai, 2004). Therefore, the conclusions from this study should be interpreted in context. The meta-analyses were repeated with and without this study in sensitivity analyses. The exclusion of this study had large effects on the pooled outcome estimates. In several cases, such as preterm birth, a shorter IPI was associated with more favourable outcomes. Optimum interpregnancy interval after miscarriage 9 Less than 6 months More than 6 months Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year DaVanzo 2007 1198 1366 1018 1171 27.0% 1.01 [0.98,1.04] 2007 Love 2010 10856 12744 13977 18193 30.0% 1.11 [1.10,1.12] 2010 DaVanzo 2012 3949 4596 4756 5839 29.3% 1.05 [1.04,1.07] 2012 Wong 2015 349 445 169 232 13.6% 1.08 [0.98,1.18] 2015 Total (95% CI) 19151 25435 100.0% 1.06 [1.01,1.11] Total events 16352 19920 Heterogeneity: Tau2 = 0.00; Chi2 = 49.60,df = 3(P < 0.00001); I2 = 94% Risk Ratio M-H, Random, 95% CI Test for overall effect: Z = 2.53 (P = 0.01) 0.1 0.2 0.5 1 2 6 months or more <6 months 10 Less than 6 months More than 6 months Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year DaVanzo 2007 41 1366 49 1171 14.5% 0.72 [0.48, 1.08] 2007 Love 2010 70 12744 129 18193 29.3% 0.77 [0.58, 1.04] 2010 DaVanzo 2012 176 4596 230 5839 55.8% 0.97 [0.80, 1.18] 2012 Wong 2015 3 445 1 232 0.4% 1.56 [0.16,14.95] 2015 Total (95% CI) 19151 25435 100.0% 0.88 [0.76,1.02] Total events 290 409 Heterogeneity: Chi2 = 2.99,df = 3(P = 0.39); I2 = 0% Test for overall effect Z= 1.69 (P = 0.09) Risk Ratio M-H, Fixed, 95% CI 0.1 0.2 0.5 1 2 6 months or more <6 months 10 O o Less than 6 months 6 months or more Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year Conde Agudelo 2004 2264 11615 22514 246493 27.0% 2.13 [2.05, 2.22] 2004 Love 2010 776 10856 1358 13977 26.9% 0.74 [0.68, 0.80] 2010 Bentolila 2013 10 92 26 149 22.4% 0.62 [0.32, 1.23] 2013 Makhlouf2014 20 395 28 645 23.7% 1.17 [0.67, 2.04] 2014 Total (95% CI) 22958 261264 100.0% 1.05 [0.48, 2.29] Total events Risk Ratio M-H, Random, 95% CI 3070 Heterogeneity: Tau2 = 0.58; Chi2 = 539.34, df Test for overall effect: Z = 0.13 (P = 0.89) 23926 = 3 (P< 0.00001); I2 = 99% 0.2 0.5 1 2 6 months or more <6 months Less than 6 months 6 months or more Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year Risk Ratio M-H, Fixed, 95% CI Love 2010 776 10856 1358 13977 96.7% 0.74 [0.68, 0.80] 2010 Bentolila 2013 10 92 26 149 1.6% 0.62 [0.32, 1.23] 2013 Makhlouf2014 20 395 28 645 1.7% 1.17 [0.67, 2.04] 2014 Total (95% CI) 11343 14771 100.0% 0.74 [0.68, 0.81] Total events 806 1412 Heterogeneity: Chi2 = 2.80,df = 2(P = 0.25); I2 = 29% Test for overall effect Z = 7.07 (P< 0.00001) 0.1 0.2 0.5 1 2 6 months or more <6 months 10 Figure 2 Continued Meta-analyses and systematic reviews can be limited by a number of factors. Original data collection varied across the different studies as some used the mother's recall of the previous pregnancies while others used information from databases. Thus quality of the original data is a limiting factor. In addition, studies varied in their definition of certain outcomes such as miscarriage. While some studies made distinctions between women with spontaneous and induced abortions, others could not - possibly due to legal constraints and religious and cultural stigmas associated with induced abortions. Another potential bias is publication bias, and although the literature search was rigorous we were unable to search unpublished studies, which may affect our results. We investigated this possibility using a funnel plot which did not demonstrate any appreciable publication bias for the outcome of further miscarriage, but may have been present for some of the secondary outcomes with fewer publications. Furthermore confidence in the results could be limited due to the small number of studies used in the meta-analyses. A number of factors are associated with pregnancy outcomes, including age, ethnicity, social class, smoking, alcohol, BMI and previous obstetric history. However other than maternal age, the studies also varied in addressing potential confoun-ders. Failure to address all the potential confounders in the primary studies included in this review could be due to the fact that they were not recorded in the databases, or either not measured or poorly measured. Thus this can be recognised as a potential limitation 10 Kangatharan et al. Less than 6 months 6 months or more Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year Conde Agudelo 2004 421 11615 9839 246493 58.9% 0.91 [0.83, 1.00] 2004 Love 2010 485 10856 634 13977 36.9% 0.98 [0.88, 1.11] 2010 Bentolila 2013 6 92 9 149 0.5% 1.08 [0.40, 2.93] 2013 Makhlouf2014 31 395 42 645 2.1% 1.21 [0.77, 1.88] 2014 Wong 2015 39 445 19 232 1.7% 1.07 [0.63, 1.81] 2015 Total (95% CI) 23403 261496 100.0% 0.95 [0.88,1.02] Total events 982 10543 Risk Ratio M-H, Fixed, 95% CI Heterogeneity: Chi2 = 2.58, df = 4 (P = 0.63); I2 = Test for overall effect: Z = 1.51 (P = 0.13) 0% 0.1 0.2 0.5 1 2 6 months or more <6 months 10 Less than 6 months 6 months or more Risk Ratio Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year Risk Ratio M-H, Fixed, 95% CI ^smlm'igadeM Love 2010 Bentolila 2013 Makhlouf2014 Wong 2015 485 6 31 39 10856 92 395 445 11788 Total (95% CI) Total events 561 Heterogeneity: Chi2 = 0.82, df = 3 (P = 0.84); K Test for overall effect: Z = 0.01 (P = 0.99) Figure 1 Continued = 0% 634 704 10100 0.01U 13977 89.7% 9 149 1.1% 42 645 5.2% 19 232 4.0% c.~ IC.™,'.cc; zzza 0.98 [0.88, 1.11] 2010 1.08 [0.40, 2.93] 2013 1.21 [0.77, 1.88] 2014 1.07 [0.63, 1.81] 2015 15003 100.0% 1.00 [0.90,1.12] < ► 0.1 0.2 0.5 I I I 2 5 10 6 months or more <6 months O o in this study as it can lead to over or under estimated results. Despite this, a consistent effect was reported by all the studies conducted in a variety of countries and settings, which leads us to believe that these associations are likely to exist. The results of this systematic review are consistent with other studies (Basso et al., 1998; Goldstein et al., 2002; El Behery et al., 2013) that could not be included in this meta-analysis as they did not have appropriate data. The study by El Behery et al. (2013) shows that women conceiving within 6 months of a miscarriage had good reproductive outcomes and a reduced incidence of complications, and they noted that live births were highest when conceiving within 6 months (79.31%) compared to conceiving after 12 months (71.6%). However, they did not focus on an IPI of more than 6 months, but looked only at less than 6 months IPI and more than 12 months IPI. Hence this study could not be included in the main meta-analysis but only in the subgroup analysis comparing IPI of less than 6 months with that of more than 12 months (El Behery et al., 2013). Studies by Basso et al. (1998) and Goldstein et al. (2002) show that there are no adverse outcomes associated with short I PI s but also that adverse outcomes increase as IPI increases (Basso et al., 1998). However they did not use the same IPI groups as this systematic review therefore could not contribute towards the meta-analyses. In their systematic review of mechanisms underpinning short and long IPI with adverse pregnancy outcomes, Conde Agudelo et al. (2012) found evidence to support hypotheses of maternal nutritional depletion, folate depletion, cervical insufficiency, vertical transmission of infections and abnormal remodelling of endometrial blood vessels as possible explanations for the association of adverse outcomes with short IPI. Women's natural decline in reproductive capacity with age was the only hypothesis proposed to explain the association between long I Pis and adverse outcomes (Conde Agudelo et al., 2012). In cases where the IPI starts with a miscarriage, the woman's body may behave differently to that after a live birth. For example, the nutritional depletion or folate depletion hypothesis suggests that from the fifth month of pregnancy until a prolonged time after delivery, the stores of maternal nutrients, such as folate, remain low leading to folate insufficiency in women with a short IPI after a live or stillbirth. However after a miscarriage, there is a very small burden on the folate reserve and thus miscarriage is not very likely to lead to folate deficiency in the postpartum period. This could explain the reduced risk of adverse outcomes in a short IPI after a miscarriage (Smits and Essed, 2001). In support of this hypothesis, there is evidence to suggest that late miscarriages (after 12 weeks of gestation) are associated with worse outcomes in the subsequent pregnancy (Edlow et al., 2007). In addition, most women who attempt another pregnancy soon after a miscarriage are likely to be motivated to take better care of their health and consequently result in better pregnancy outcomes (DaVanzo et al., 2007). Another plausible reason may be that those who conceive soon after a miscarriage are naturally more fertile and consequently have better pregnancy outcomes. This is the first systematic evidence synthesis to assess the effect of short versus long IPI and based on the available evidence we can conclude that a short IPI (less than 6 months) following miscarriage is not associated with adverse outcomes in the subsequent pregnancy. Couples wishing to conceive after a miscarriage can be counselled that delaying pregnancy does not necessarily improve outcomes. Further research needs to look at an IPI of less than 3 months to determine an optimum cut off, if there is one. Individual patient data meta-analysis can offer opportunities to study small subgroups and/or stratify by other risk factors to determine a personalised optimum IPI after miscarriage. Optimum interpregnancy interval after miscarriage Conclusion The results of this systematic review and meta-analyses show that an IPI of less than 6 months is associated with no increase in the risks of adverse outcomes in the pregnancy following miscarriage compared to delaying pregnancy for at least 6 months. In fact, there is some evidence to suggest that chances of having a live birth in the subsequent pregnancy are increased with an IPI of less than 6 months. There is now ample evidence to suggest that delaying a pregnancy following a miscarriage is not beneficial and unless there are specific reasons for delay couples should be advised to try for another pregnancy as soon as they feel ready. Supplementary data Supplementary data are available at http://humupd.oxfordjournals.org/. Authors' roles C.K. conducted the initial literature searches, reviewed the included papers, conducted the meta-analyses and wrote the first draft. S.L repeated the searches, quality assessed the included studies and commented on the draft. SB designed the review question, developed the protocol, supervised C.K. and S.L. Funding This research project did not receive any funding. Conflict of interest The authors declare that they have no conflict of interest. References Aref-Adib M, Freeman-Wang T, Ataullah I. The older obstetric patient. Obstet Gynaecol Reprod Med 2008; 18:43^8. Basso O, Olsen J, Christensen K. Risk of preterm delivery, low birth weight and growth retardation following spontaneous abortion: a regis try-based study in Denmark. IntJ Epidemiol 1998;27:642-646. Bentolila Y, Ratzon R, Shoham-Vardi I, Serjienko R, Mazor M, Bashiri A. Effect of interpregnancy interval on outcomes of pregnancy after recurrent pregnancy loss. J Matern Fetal Neonatal Med 2013;26:1459-1464. Bhattacharya S, Townend J, Shetty A, Campbell D, Bhattacharya S. Does miscarriage in an initial pregnancy lead to adverse obstetric and perinatal outcomes in the next continuing pregnancy? BJOG 2008; 115:1623-1629. Buchmayer SM, Sparen P, Cnattingius S. Previous pregnancy loss: risks related to severity of preterm delivery. Obstet Gynecol 2004; 191:1225-1231. Critical Appraisal Skills Programme (CASP). Critical Appraisal Skills Programme (CASP) [Internet]. 2016. http://www.casp-uk.net/#!casp-tools-checklists/c 18f8. Conde-Agudelo A, Belizan JM, Breman R, Brockman SC, Rosas-Bermudez A. Effect of the interpregnancy interval after an abortion on maternal and perinatal health in Latin America. Int J Gynaecol Obstet 2004;89:S34-S40. Conde-Agudelo A, Rosas-Bermüdez A, Kafury-Goeta A. Birth spacing and risk of adverse perinatal outcomes: a meta-analysis. JAMA 2006;295:1809-1823. Conde-Agudelo A, Rosas-Bermudez A, Castaho F, Norton MH. Effects of birth spacing on maternal, perinatal, infant, and child health: a systematic review of causal mechanisms. Stud Farn Plann 2012;43:93-1 14. Cox T, van der Steeg JW, Steures P, Hompes PG, van der Veen F, Eijkemans MJ, van Leeuwen JH, Renckens C, Bossuyt PM, Mol BW. Time to pregnancy after a previous miscarriage in subfertile couples. Fertil Steril 2010;94:485-488. DaVanzo J, Hale L, Razzaque A, Rahman M. Effects of interpregnancy interval and outcome of the preceding pregnancy on pregnancy outcomes in Matlab, Bangladesh. BJOG 2007; I 14:1079-1087. DaVanzo J, Hale L, Rahman M. How long after a miscarriage should women wait before becoming pregnant again? Multivariate analysis of cohort data from Matlab, Bangladesh. BMJ Open 20I2;2. Edlow AG, Srinivas SK, Elovitz MA. Second-trimester loss and subsequent pregnancy outcomes: What is the real risk? Am J Obstet Gynecol 2007; 197. El Behery MM, Siam S, Seksaka MA, Ibrahim ZM. Reproductive performance in the next pregnancy for nulliparous women with history of first trimester spontaneous abortion. Arch Gynecol Obstet 2013;288:939-944. Goldstein RRP, Croughan MS, Robertson PA. Neonatal outcomes in immediate versus delayed conceptions after spontaneous abortion: a retrospective case series. Obstet Gynecol 2002;186:1230-1236. Kaandorp SP, van Mens TE, Middeldorp S, Hutten BA, Hof MH, van der Post JA, van der Veen F, Goddijn M. Time to conception and time to live birth in women with unexplained recurrent miscarriage. Hum Reprod 20I4;29:1 146-1 152. Love ER, Bhattacharya S, Smith NC, Bhattacharya S. Effect of interpregnancy interval on outcomes of pregnancy after miscarriage: retrospective analysis of hospital episode statistics in Scotland. BMJ 20I0;34I:3967. Makhlouf MA, Clifton RG, Roberts JM, Myatt L, Hauth JC, Leveno KJ, Varner MW, Thorp JM Jr, Mercer BM, Peaceman AM et al. Adverse pregnancy outcomes among women with prior spontaneous or induced abortions. Am J Perinatol 2014;31:765-772. Morgan-Ortiz F, Muhoz-Acosta J, Valdez-Quevedo R, Quevedo-Castro E, Bäez-Barraza J. Effect of post-abortion interpregnancy interval on obstetric and perinatal outcomes. Ginecol Obstet Mex 2010;78:46-52. Report of a WHO Technical Consultation on Birth Spacing. WHO [Internet] 2005. http://www.who.int/maternal_child_adolescent/documents/bi rth_spacing.pdf. Sapra KJ, McLain AC, MaisogJM, Sundaram R, Buck Louis GM. Successive time to pregnancy among women experiencing pregnancy loss. Hum Reprod 20I4;29: 2553-2559. Smits LJ, Essed GG. Short interpregnancy intervals and unfavourable pregnancy outcome: role of folate depletion. Lancet 2001 ;358:2074-2077. Wong LF, Schliep KC, Silver RM, Mumford SL, Perkins NJ, Ye A, Galai N, Wactawski-Wende J, Lynch AM, Townsend JM et al. The effect of a very short interpregnancy interval and pregnancy outcomes following a previous pregnancy loss. ObstetGynecol 20l5;2l2:375.el-375.el I. Wyss P, Biedermann K, Huch A. Relevance of the miscarriage-new pregnancy interval. J Perinat Med 1994;22:235-241.