sign in sign up
<<
Home , In vitro fertilisation

-RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

Contents lists available at ScienceDirect

Journal of Psychosomatic Research

journal homepage: www.elsevier.com/locate/jpsychores

Review article Association between treatment and perinatal depressive 7 symptoms: A meta-analysis of observational studies

Shengbao Chen, Tingting Wang, Senmao Zhang, Lijuan Zhao, Lizhang Chen⁎

Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha, Hunan Province, China

ARTICLE INFO ABSTRACT

Keywords: Objective: Recent reports have shown a considerable number of couples received infertility treatment, raising Infertility treatment new concerns about the association between infertility treatment and perinatal depressive symptoms. However, Meta-analysis the conclusions of existing studies were inconsistent. Therefore, we conducted a meta-analysis to determine Perinatal whether infertility treatment increase the risk of developing perinatal depressive symptoms. Depressive symptoms Methods: A systematic literature search was performed in several databases up to July 2018 for relevant articles. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using a random-effects model. Subgroup and sensitivity analyses were performed to explore possible sources of heterogeneity. Results: Twenty-two studies with a total of 69,201 individuals were included in this study. The pooled OR of the association between infertility treatment and perinatal depressive symptoms was 1.01(95% CI: 0.83, 1.23), with substantial heterogeneity (I2 = 63%, P < .001). However, in subgroup analyses, a significantly positive asso- ciation between infertility treatment and depressive symptoms was observed only in some Asian countries (six studies), and the pooled OR was 1.73 (95% CI:1.07, 2.81). An inverse association was found in 6–12 months after delivery (OR = 0.56, 95% CI:0.33, 0.96). Sensitivity analyses validated evidence of the robustness of the find- ings. Conclusion: The results show that women who receive infertility treatment do not appear to be at increased risk of significant perinatal depressive symptoms compared with those after spontaneous conception.

1. Introduction considered to be multifactorial involving psychological and social fac- tors beyond biological aspects [12–15]. Risk factors such as previous Perinatal (PND), defined as depressive symptoms oc- depression [16,17], emotional instability [18], lack of social support curring either during or postpartum, is one of the most [19], cesarean delivery [20] and low socioeconomic status [10], asso- common medical complications in women [1,2]. The pre- ciated with perinatal depressive symptoms were reported in previous valence of antenatal depressive symptoms ranges from 14% to 23% [3], studies. With more and more people using infertility treatment to while over 22% of women develop postpartum depressive symptoms achieve pregnancy, the effect of infertility treatment on the risk of within the first year after delivery [4]. In some high-income countries, perinatal depressive symptoms has also attracted close attention of re- antenatal depressive symptoms are estimated to be 17–18%, which searchers. decrease to 13–15% after delivery [5]. Relatively, the mean prevalence Infertility is defined as the failure to achieve a clinical pregnancy of postpartum depressive symptoms and antenatal depressive symptoms after 12 months or more of regular unprotected sexual intercourse in low and middle-income countries is estimated to be 19.8% and [21,22]. It is a problem for infertile women wishing to have a baby and 15.6% [6]. Perinatal depressive symptoms have adverse effects on the it is also costly for society [23]. It affects 8%–12% of the childbearing couple's relationship, the -infant relationship and the child's population [24], and 56% of individuals seek infertility treatment development [7]. Untreated depressive symptoms can not only bring [25,26]. The World Health Organization recognizes infertility as a multiple adverse outcomes to the mother, but also adversely affect the global public health issue [7,22]. Fortunately, with the rapid develop- mental and emotional health of the child through school age [8–11].As ment of infertility treatment including fertility enhancing drugs and present, the etiology of perinatal depressive symptoms is widely assisted reproductive technologies (ART), it becomes easier for

⁎ Corresponding author at: Department of Epidemiology and Health Statistics, Xiangya School of Public Health, Central South University, NO. 238 Shang Ma Yuan Ling Xiang Xiangya Road, Kaifuya District, Changsha, Hunan Province, China. E-mail address: [email protected] (L. Chen). https://doi.org/10.1016/j.jpsychores.2019.03.016 Received 28 November 2018; Received in revised form 15 March 2019; Accepted 16 March 2019 ‹(OVHYLHU,QF$OOULJKWVUHVHUYHG S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ² countless infertile couples to achieve pregnancy successfully. The beginning of pregnancy to 1 year after delivery in order to detect more number of women who use infertility treatment especially ART is in- cases. creasing. More than 10% of the childbearing population resort to ART [27]. While the increasing availability of infertility treatment can help 2.3. Study selection reduce the burden of infertility worldwide, the experience of infertility and its treatment may bring infertile women stress at the same time At the stage of titles and abstracts screening, we purposely broa- [28]. Infertility treatment may lower the couple's marital satisfaction. It dened the inclusion criteria to obtain any relevant study. First, studies also has an adverse effect on their self-esteem and quality of life, which were considered for inclusion if they were published in Chinese or lasts for a long time during pregnancy and postpartum [7,29]. How- English and reported on infertility treatment with perinatal depressive ever, studies have shown that these factors are associated with an in- symptoms. Then the full texts of all selected studies were reviewed. creased risk of developing perinatal depressive symptoms [30–32]. This Studies were included if they 1) were observational studies targeted on suggests that there may be an association between infertility treatment women who received infertility treatment and had a delivery success- and perinatal depressive symptoms. fully, 2) involved the diagnosis or evaluation of PND, 3) reported odds However, it is worth noting that the findings of studies on the as- ratio (OR) with the 95% confidence intervals (CI) or provided sufficient sociation between infertility treatment and perinatal depressive symp- information to calculate effect sizes. Conversely, studies were excluded toms are inconsistent [33–36]. A previous meta-analysis of observa- if they 1) were review articles, conference abstracts, case reports, ex- tional studies found that women conceived through infertility treatment perimental studies, 2) did not provided any information to calculate are not associated with postpartum depressive symptom [7]. In this effect sizes, 3) had incomplete or unclear data, or 4) were duplicate meta-analysis, only eight studies with small sample size were included publications. If more than one study involved the same population, the which may have resulted in limited statistical power to detect a po- most recent and comprehensive study was selected. tential increase in the risk of postpartum depressive symptoms. More- over, all the included studies were conducted in high-income countries, 2.4. Data extraction and quality assessment which may weaken the generalizability of the results in this study. Therefore, we conducted a meta-analysis of observational studies to Data from the selected relevant studies were extracted by using a further investigate the association between infertility treatment and data extraction form, which was designed by the authors during the perinatal depressive symptoms. protocol stage of this review. The following data were extracted from each study by two investigators (SBC and TTW) independently: first 2. Materials and methods author and year of publication, study location, study design, study period, type of infertility treatment, sample size, assessment method of 2.1. Search strategy perinatal depressive symptoms, assessment time, OR with 95% CI for the risk of perinatal depressive symptoms, whether adjusted for con- The meta-analysis was conducted on the basis of the items outlined founding factors, quality assessment and covariates adjusted for. in the Preferred Reporting Items for Systematic Reviews and Meta- The quality of the studies was assessed and scored by two in- analyses (PRISMA) statement [37]. PubMed, Web of Science, China vestigators independently (SBC and TTW) using the Newcastle-Ottawa National Knowledge Infrastructure (CNKI), Chinese Scientific Journals quality assessment scale (NOS) for case-control studies or cohort studies Full text Database (CQVIP) and wan fang databases were searched for and an 11-item checklist which was recommended by Agency for relevant available articles published up to July 2018. The following Healthcare Research and Quality (AHRQ) for cross-sectional studies. In terms were used in combination: 1) assisted , this meta-analysis, we define articles quality as follows: high quality assisted reproductive technologies, assisted conception, assisted re- ≥7; low quality < 7. production, in vitro fertilization, test tube baby, intracytoplasmic injection, artificial , intrauterine insemination, cervical 2.5. Statistical analyses canal insemination, implantation, ovarian stimulation, ovula- tion induction, infertility treatment and fertility treatment; and 2) To perform the meta-analysis, data were entered and analyzed using prenatal, antenatal, antepartum, perinatal, periparum, pregnancy, ma- Stata Version 12 (Stata Corp, College Station, Texas, USA) and Review ternal, postpartum, puerperal, puerperium and postnatal; and 3) de- Manager Version 5.3 (The Nordic Centre, Cochrane pression, mood disorder, affective disorder, baby blues, depressive Collaboration, Copenhagen, Denmark). Individual and pooled OR and disorder, dysthymia, depressive symptoms, mental health and mental associated 95% CI were calculated using a fixed effects model in the illness. Furthermore, we reviewed citations in the retrieved articles and case of absence of between-study heterogeneity and a random effects relevant reviews to search for additional relevant studies. model in the case of significant between-subject heterogeneity [39]. Hazard ratio (HR), incidence rate ratio and (RR) were di- 2.2. Exposure and outcomes rectly considered as OR. Statistical heterogeneity among studies was evaluated using the Cochran Q test with P < .1 considered as sig- The exposure of interest was any types of infertility treatment which nificant heterogeneity and I2 statistics that was a quantitative indicator included ART and fertility enhancing drugs. ART included artificial used to estimate the proportion of variance between studies (I2 ≤ 25% insemination (AI), fertilization and (IVF-ET), represents low heterogeneity, 26–50% represents moderate hetero- intracytoplasmic sperm injection (ICSI) and other forms of treatment. geneity, 51–75% represents substantial heterogeneity, and 76–100% Fertility enhancing drugs included , clomiphene citrate, represents high heterogeneity). Subgroup analyses for perinatal de- human chorionic , gonadotropin-releasing hormone ana- pressive symptoms were performed by assessment time, study location, logs, and other unspecified fertility drugs. study design, sample size, quality of studies and adjustment status of The outcome of interest was probable depression or depressive factors. Sensitivity analyses were performed to examine symptoms in perinatal period which was indicated by validated as- the influence of different exclusion criteria on the overall results. We sessment tools, such as Edinburgh postnatal depression scale (EPDS), conducted sensitivity analyses in which some studies were removed and Postpartum depression screening scale (PDSS) and Beck depression the rest were analyzed to evaluate whether the results were affected inventory (BDI). However, mental illnesses occurring as late as statistically significantly. was visually estimated by 12 months after are often defined as postpartum [3,38,53]. In generating a funnel plot, and Egger's linear regression test was per- our study, the perinatal period was defined as the time range from the formed to explore the asymmetry of the funnel plot [40]. All reported P

 S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

Fig. 1. Flow diagram of literature search. values were two-sided and P < .05 was considered statistically sig- published between 2000 and 2018, were shown in Table 1. Thirteen nificant unless otherwise specified. studies (59.1%) were cohort in design, five (22.7%) were case-control, and four (18.2%) were cross-sectional. Eight studies (36.4%) were 3. Results conducted in Europe, six (27.3%) in Asia, six (27.3%) in Oceania, and two (9.1%) in North America. Eight studies (36.4%) evaluated the as- 3.1. Literature search sociation of infertility treatment with the risk of perinatal depressive symptoms, as the rests evaluated the association of ART with the risk of The literature search yielded a total of 2684 potentially relevant perinatal depressive symptoms. The time points for depressive symp- fi citations, three additional articles were found through other sources. toms assessment were divided into six parts: rst and second trimester, Among these studies, 250 duplicate studies were removed. Of the re- third trimester, 0–3 months after delivery, 3–6 months after delivery, mains, 2399 studies were excluded by reviewing titles and abstracts. 6–12 months after delivery and unknown evaluation time points. The Then, 38 studies were identified and retrieved for full-text screening, 16 unknown evaluation time point means that studies only reported the additional papers were excluded. The reasons were as following: lit- approximate evaluation time of depressive symptoms. And eighteen erature review articles (3 studies), case report (1 study), not focus on studies (81.8%) only evaluated depressive symptoms during pregnancy fi depressive symptoms (2 studies), mental health of women who were ( rst and second trimester, third trimester) or postpartum (0–3 months ready for IVF treatment (2 studies) and unable to extract outcome after delivery, 3–6 months after delivery and 6–12 months after de- measures (8 studies). Finally, 22 studies [23,33–35,41–58] were in- livery), while others (4/22 studies, 18.2%) assessed depressive symp- fi cluded in the meta-analysis. The flowchart of the literature search is toms during both periods. According to our broad de nition, we con- shown in Fig. 1. sider all of these periods to be an assessment of perinatal depressive symptoms. Most of included studies (15/22 studies, 68%) took the EPDS as the assessment tool, while the rests used different tools such as 3.2. Study characteristics Hospital anxiety and depression scale (HADS), Postpartum depression screening scale (PDSS) and Beck depression inventory (BDI). Eight The characteristics of the 22 studies, which included 69,201 parti- studies (36.4%) adjusted some potential confounding factors (such as cipants (ranged between 126 and 40,337 individuals) and were

 S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

Table 1 Characteristics of studies on the associations between infertility treatment and perinatal depressive symptoms.

Study Study design Sample size Assessment time Time classification Assessment method Quality score

GarciaBlanco et al. (2018) Cohort 243: 60 IVF; 183 controls Pregnancy: the third trimester The third trimester BDI/SF 8 Spain Mori et al. (2018) Japan Cohort 2709: 162 IT; 189 ART; 2352 Postpartum: EPDS ≥ 99 controls the day before hospital 0–3 months discharge 1 month 0–3 months 2 months 0–3 months 4 months 4–6 months 6 months 4–6 months Joelsson et al. (2017) Sweden Cross- 3583: 468 sub-fertile women; NA / EPDS ≥ 12 7 sectional 143 ART; 2972 controls Gambadauro et al. (2017) Cohort 3283: 167 IVF; 3116 controls Pregnancy: EPDS ≥ 12 8 Sweden 17 weeks First and second trimester 32 weeks Third trimester Postpartum: 6 weeks 0–3 months 6 months 4–6 months Vikstrom et al. (2017) Sweden Case-control 12,085: 3532 IVF; 8553 Postpartum: 12 months 6–12 months ICD–10 7 controls Gao et al. (2017) Xi'an, China Case-control 756: 170 Tube baby; 133 OI; NA / EPDS ≥ 12 5 456 controls Tendais et al. (2016) Portugal Cohort 267: 36 IT; 231 controls Pregnancy: EPDS ≥ 12 7 8–15 weeks First and second trimester 20–24 weeks First and second trimester 28–34 weeks Third trimester Postpartum: After childbirth 0–3 months 3 months 0–3 months Lynch et al. (2014) US Cross- 40,337: 503 IT; 39,834 Postpartum: 2–4 months 0–3 months PRAMS questionnaire 7 sectional controls Listijono et al. (2014) Australia Case-control 189: 94 ART; 95 controls Postpartum: 4 weeks 0–3 months EPDS 6 Raguz et al. (2014) Canada Cohort 228: 76 IT; 152 controls Pregnancy: 18–24 weeks First and second EPDS ≥ 13 8 Postpartum: 4 months Trimester 4–6 months Fisher et al. (2013) Australia Cohort 549: 272 ART; 277 controls Pregnancy: 31.6 ± 2.5 weeks Third trimester EPDS ≥ 12 7 Fisher et al. (2013) Australia Cohort 592: 297 ART; 295 controls Pregnancy: 31.6 ± 2.5 weeks Third trimester EPDS ≥ 14 7 Warmelink et al. (2012) Cross- 428: 32 IT; 396 controls Postpartum: 2–6 months 4–6 months HADS ≥ 87 Netherlands sectional McMahon et al. (2011) Cohort 541: 275 ART; 266 controls Postpartum: 4 months 4–6 months MINI and EPDS 7 Australia Akyuz et al. (2010) Turkey Cohort 156: 51 IT; 105 controls Postpartum: 4–6 weeks 0–3 months PDSS ≥ 65 8 Monti et al. (2009) Italy Case-control 64: 25 ART; 39 controls Pregnancy: 30–32 weeks Third trimester EPDS ≥ 12 6 Postpartum: 1 week 0–3 months 3 months 0–3 months Wang et al. (2008) Case-control 274: 4 IVF-ET; 270 controls Postpartum: 6 weeks 0–3 months EPDS ≥ 13 6 Shenyang, China Repokar et al. (2007) Finland Cohort 746: 367 IVF or ICSI; 379 Pregnancy: 18–20 weeks First and second BDI/SF 7 controls trimester Xie et al. (2006) Changsha, Cross- 320 Postpartum: 6 weeks 0–3 months EPDS ≥ 13 6 China sectional Fisher et al. (2005) Australia Cohort 745: 45 IVF; 12 OI and AI; Postpartum: Days 1 and 5 0–3 months EPDS > 12 7 678 controls Yokoyama et al. (2003) Japan Cohort 990: 359 IT; 631 controls NA / DSM-IV 7 Gibson et al. (2000) Australia Cohort 126: 65 IVF; 61 controls Postpartum: 12 months 6–12 months EPDS ≥ 13 7

AI, artificial insemination; ART, assisted reproductive technologies; BDI/SF, Beck depression inventory-sort form; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders; EPDS, Edinburgh postnatal depression scale; HADS, Hospital anxiety and depression scale; ICD–10, International Classification of Diseases,10th edition; ICSI, intracytoplasmic sperm injection; IT, infertility treatment; IVF, in vitro fertilization; IVF-ET, in vitro fertilization and embryo transfer; MINI, Mini international neuropsychiatric interview; OI, induction; PDSS, Postpartum depression screening scale; PRAMS, pregnancy risk assessment monitoring system; NA, not available. age, education, social support, history of depression, type of delivery, 3.3. Meta-analysis results sex of infant and so on), and the remaining studies did not adjusted any factors when estimating the effect. Sixteen studies (72.7%) had a The random-effects model results with the ORs of perinatal de- quality assessment score of greater than or equal to 7 and six studies pressive symptoms in relation to ever experience of infertility treatment (27.3%) had a quality assessment score of < 7, indicating that the are shown in Fig. 2. Among the 22 studies, only 6 studies showed sta- methodological quality was generally good. The detailed characteristics tistically significant association between infertility treatment and of the included studies are shown in Table 1. perinatal depressive symptoms. No significant difference was reported in the incidence of significant perinatal depressive symptoms between

 S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

[TSD]: I2 = 66.8%, P = .08), study location (TSD: I2 = 55.9%, P = .08) and assessment time (TSD: I2 = 55.4%, P = .05) were identified as the first three of the most relevant heterogeneity moderators. However, there were not statistically significantly differences for risks of devel- oping perinatal depressive symptoms in these three subgroup analyses. In subgroup analysis stratified by assessment time, nine studies presented data on prenatal depressive symptoms, while seventeen stu- dies on postpartum depressive symptoms. The pooled ORs and corre- sponding 95% CIs of the association between infertility treatment and perinatal depressive symptoms classified by time points for depressive symptoms assessment (first and second trimester, third trimester, 0–3 months after delivery, 3–6 months after delivery, 6–12 months after delivery and unknown evaluation time points) were 0.95 (95% CI:0.71, 1.26), 0.73 (95% CI:0.36, 1.45), 1.21 (95% CI:0.95, 1.56), 0.89 (95% CI:0.68, 1.16), 0.56 (95% CI:0.33, 0.96) and 1.40 (95% CI:0.93, 2.10), respectively. An inverse association between infertility treatment and perinatal depressive symptoms was only found in 6–12 months after delivery. In subgroup analysis stratified by study design, the pooled ORs for cohort, case-control and cross-sectional studies were 0.94 (95% CI:0.76, 1.16), 1.73 (95% CI:0.69, 4.34), 1.17 (95% CI:0.65, 2.10), respectively. A significantly positive association between infertility Fig. 2. Forest plot for the effect of infertility treatment on perinatal depressive treatment and perinatal depressive symptoms was observed only in symptoms. The size of gray box is positively proportional to the weight assigned some Asian countries (One study is from Turkey, two studies are from to each study, and horizontal lines represent the 95% CIs. (OR, ; CI, Japan, and three studies are from China), and the pooled OR was 1.73 fi con dence interval). (95% CI:1.07, 2.81). The pooled ORs were 0.94 (95% CI:0.79, 1.13) among sixteen studies with high-quality assessment score and 2.08 pregnant women who used infertility treatment and those with a (95% CI:0.87, 4.96) among six studies with low-quality assessment spontaneous pregnancy (OR = 1.01, 95% CI: 0.83, 1.23), with evidence score. In the subgroup analysis stratified by sample size, we observed no of significant between-study heterogeneity (I2 = 63%, P < .001). evidence of positive association for studies with sample size < 500 (OR = 1.24, 95% CI:0.68, 2.26), sample size 500–1000 (OR = 1.06, 95% CI:0.78, 1.45) and sample size > 1000(OR = 0.91, 95% CI:0.80, 3.4. Subgroup analyses 1.05). In subgroup analysis stratified by adjustment status of con- founding factors, the pooled ORs were 0.87 (95% CI:0.64, 1.20) among Table 2 shows the results of the subgroup analyses of the stability of studies that adjusted confounding factors and 1.11 (95% CI:0.85, 1.44) the major findings and explored the sources of potential heterogeneity. among studies that did not adjust. After subgroup analyses, quality score (test for subgroup differences

Table 2 Subgroup analysis for the association between infertility treatments and perinatal depressive symptoms.

Subgroup variables No. of studies Pooled OR(95%CI) I2 (%) P for heterogeneity Test for subgroup differences

I2 (%) P

Total 22 1.01(0.83–1.23) 63.0 < .001 Assessment time First and second trimester 4 0.95(0.71–1.26) 0 .69 55.4 .05 Third trimester 6 0.73(0.36–1.45) 73.4 .002 0–3 months postpartum 11 1.21(0.95–1.56) 46.4 .045 4–6 months postpartum 4 0.89(0.68–1.16) 0 .49 6–12 months postpartum 2 0.56(0.33–0.96) 0 .78 Not available 3 1.40(0.93–2.10) 51.5 .13 Location Europe 8 0.84(0.58–1.20) 71.0 < .001 55.9 .08 Asia 6 1.73(1.07–2.81) 77.0 < .001 Oceania 6 0.86(0.65–1.14) 2.0 .40 North America 2 0.95(0.65–1.41) 0 .97 Study design Cohort 13 0.99(0.89–1.10) 60.0 .003 0 .75 Case-control 5 1.14(0.78–1.66) 77.0 .001 Cross-sectional 4 1.04(0.77–1.40) 66.0 .03 Sample size < 500 10 1.24(0.68–2.26) 73.0 < .001 0 .45 500–1000 7 1.06(0.78–1.45) 58.0 .03 > 1000 5 0.91(0.80–1.05) 6.0 .37 Quality score < 7 6 2.08(0.87–4.96) 79.0 < .001 66.8 .08 ≥7 16 0.94(0.79–1.13) 51.0 .01 Adjustment status of confounding factors Adjusted 8 0.87(0.64–1.20) 68.0 .002 21.0 .26 Unadjusted 14 1.11(0.85–1.44) 58.0 .003

CI, confidence interval; OR, odds ratio.

 S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

4. Discussion

The main goal of our meta-analysis was to examine the association between infertility treatment and the risk of developing perinatal de- pressive symptoms, which showed that women who receive infertility treatment do not appear to be at increased risk of significant perinatal depressive symptoms compared with those after spontaneous concep- tion (OR = 1.01, 95% CI:0.83, 1.23). Only one study in this meta- analysis reported higher EPDS scores in the ART group compared with the control group at 3 months postpartum [35]. To the best of our knowledge, this study may be the most comprehensive meta-analysis assessing the association between infertility treatment and perinatal depressive symptoms, which can not only encourage infertile couples to receive infertility treatment but also eliminate misunderstandings of the general population about infertility treatment. Our findings were basically consistent with those in a good recent and meta-analysis of medically assisted conception Fig. 3. Sensitivity analyses by excluding any study for infertility treatment and perinatal depressive symptoms. and postpartum depressive symptoms published by Gressier et al. [7], which found no significant risk for postpartum depressive symptoms conceived by infertility treatment (OR = 0.93, 95% CI:0.67, 1.31) and 3.5. Sensitivity analyses and publication bias ART (OR = 1.04, 95% CI:0.71, 1.52). Both meta-analyses confined their literature search to observational studies which included data on de- fl Sensitivity analyses were performed to examine the in uence of pressive symptoms during the perinatal period. However, there were ff di erent exclusion criteria on the overall risk estimate for perinatal some limitations in the previous one. Firstly, there were only eight depressive symptoms associated with infertility treatment. In the thir- small sample-size studies in the meta-analysis, which may limit the fi teen studies with ART as an exposure, we found no signi cant change in capacity to identify small or moderate effects on risk for perinatal de- our result (OR = 0.89; 95% CI: 0.63, 1.25),with substantial hetero- 2 pressive symptoms. Secondly, it is hard to generalize the results of geneity (I = 68.8%, P < .001). Exclusion of 6 studies that were low previous meta-analysis in other countries and low socio-economic methodologic quality yielded similar results (OR = 0.94; 95% CI: 0.79, 2 samples as the included studies were only conducted in high-income 1.13), with substantial heterogeneity (I = 50.9%, P = .01). Further countries. exclusion of 14 studies in which the confounding factors were not ad- According to the subgroup analyses, there was a significantly posi- justed when assessing the risk of developing perinatal depressive tive association between the use of infertility treatment and perinatal symptoms associated with infertility treatment also yielded similar re- depressive symptoms in some Asian countries like Turkey, Japan and sults (OR = 0.87; 95%CI: 0.64–1.20), yet heterogeneity was still pre- China, where infertility treatment might be a private problem. Infertile 2 fi sent (I = 68.4%, P = .002).The results showed no signi cant change of women in these countries are not willing to talk about this topic com- the overall risk estimates for perinatal depressive symptoms when any pared with western countries [35]. When considering the time of as- one study was excluded, with a range from 0.97 (95% CI:0.81, 1.17) to sessment of depressive symptoms, women seeking infertility treatment 1.05 (95% CI:0.89, 1.27) (Fig. 3). presented a lower risk of developing depressive symptoms than natu- For publication bias, Egger's linear regression test along with the rally conceived women. Women treated with infertility treatment have result of the funnel plot was applied to assess the asymmetry of the been usually mentally prepared in advance to cope with depressive fi funnel plot and evaluate the signi cance of publication bias. Fig. 4 symptoms [30]. Moreover, because infertility treatment itself was so shows no visual asymmetry in the visual examination of funnel plots stressful, those women who were successful in having a baby with the fi and Egger regression asymmetry tests found no evidence of a signi cant help of infertility treatment tend to be more satisfied during perinatal publication bias for the analysis that investigated the relationship of period for a long time than naturally conceived women [33,59]. infertility treatment with perinatal depressive symptoms (P = 0.686). Assessing heterogeneity is a critical issue in meta-analysis, and it is common to detect heterogeneity between studies [60]. Substantial heterogeneity was demonstrated for the links between infertility treatment and perinatal depressive symptoms. Though the results of subgroup analyses did not find the sources of between-study hetero- geneity, quality score (TSD: I2 = 66.8%, P = .08), study location (TSD: I2 = 55.9%, P = .08) and assessment time (TSD: I2 = 55.4%, P = .05) were the three most likely factors contributing to the between-study heterogeneity. However, both two subgroups showed no statistically significant differences for risks of developing perinatal depressive symptoms. In subgroup analysis stratified by study location, the results indicated that significant heterogeneity was still found in the Asian and European populations (I2 = 76.8%, P = .001; I2 = 71.5%, P = .001), which may be related to the specific genetic and environmental back- grounds. Other factors such as different assessment time and different assessment tools of perinatal depressive symptoms, different thresholds of EPDS and the different variables adjusted for across studies may also lead to the heterogeneity. In additional, EPDS was assessed in different ways, such as telephone interviews [57], face-to-face interviews [35] and self-reported questionnaire [34]. Even we conduct subgroup ana- lyses to find possible sources, heterogeneity still persists. Therefore the Fig. 4. Funnel plot of infertility treatment and perinatal depressive symptoms. results of this meta-analysis should be interpreted cautiously.

 S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

There are several notable strengths in this meta-analysis. Firstly, a obstetric factors may potentially produce biases when observed the large number of cases and participants from 12 countries were included association between infertility treatment and perinatal depressive by a more comprehensive search strategy. Because of the rapid devel- symptoms. In our study, we conducted sensitivity analyses to consider opment of infertility treatment, our results are of great interests to both the effects of those potential confounding factors. Results showed that medical science and the infertile couples. Secondly, the results of the exclusion of studies without any adjustment or with low methodolo- sensitivity analyses based on various exclusion criteria supported the gical quality yielded similar results for all estimates. It is worth noting robustness of the observed association between infertility treatment and that some confounding factors that cannot be adjusted within included perinatal depressive symptoms. Thirdly, all included studies received studies, such as cultural-specific, genetic and environmental factors, an appropriate score based on the criteria from the NOS, and the result may also affect our results. Due to the limitation of obtaining adequate of Egger's test did not find the potential risk of publication bias. In data, we could not present the role of those potential confounding addition, we were also concerned about the antenatal depressive factors in the association between infertility treatment and perinatal symptoms of women who had successfully delivered through infertility depressive symptoms through subgroup analysis processes. treatment, which may summary a more comprehensive result. In addition, few of the included studies have a diagnostic assessment However, the study also has some limitations. Firstly, only openly tool for perinatal depression and the screening tools used for assessing published articles were searched for and included, while other un- perinatal depressive symptoms varies in included studies, so our review published articles or gray literature that meet our inclusion criteria cannot address the true relationship between infertility treatment and might have been neglected. Moreover, although an attempt was made perinatal depression, but the potential risk associated with perinatal to minimize the possible bias in the process of document retrieving with depressive symptoms. Instruments such as Diagnostic and Statistical specific searches in major English-Chinese databases (including master Manual of Mental Disorders by the American Psychiatric Association and doctoral theses), there may still be some unidentified papers. And have high specificity and sensitivity for diagnosing depression, while because our research only included studies published in Chinese or other instruments, such as EPDS or BDI, have low specificity and should English, additional research in other populations was also needed to be regarded as screening tools. It seems that screening tools tend to generalize the results. Based on different cultural backgrounds, it is overestimate the prevalence relative to diagnostic assessment tools, difficult for infertile women from different countries to have a unified which may lead to an overestimated risk for developing perinatal de- attitude towards infertility and its treatment. For example, Chinese pressive symptoms associated with infertility treatment in our meta- women think infertility is a private problem. They are unwilling to analysis. Despite the limitations in validated screening tools of de- disclose their experience of receiving infertility treatment and feel pressive symptoms, these screening tools are still essential for assessing shameful about it [35,61]. Due to the limitations of the original data, depression because they are more convenient and more cost-effective we did not fully consider the impact of potential cultural differences on for use in busy medical clinics and epidemiological surveys than formal the results. diagnostic tools. As an alternative, validated screening tools may be the Secondly, another important limitation was that we were unable to best choice to assess depressive symptoms. explain the effects of underlying infertility on depressive symptoms. On In conclusion, we found that there seems to be no increased risk of the one hand, our meta-analyses cannot distinguish whether the ob- significant perinatal depressive symptoms in women after infertility served risk of depressive symptoms in infertile women after infertility treatment compared with the control group. Even so, psychological treatment was associated with underlying infertility, or the treatment counseling and stress management for infertile women are still essential itself. On the other hand, in all included studies, we did not find a study during and following treatment, especially for women with risk factors that regarded women who had a diagnosis of infertility as an unbiased of perinatal depressive symptoms. It is worth mentioning that we esti- control. Moreover, our study was not looking at all women who went mated the effect of infertility treatment on depressive symptoms at through infertility treatment, as it only looked at women who received different periods, but unfortunately we did not explore the association infertility treatment and had a delivery successfully. Because of the between specific infertility treatment and perinatal depressive symp- limitation in included studies, the observed risk between infertility toms. The association between different types of infertility treatment treatment and perinatal depressive symptoms was not evidence of and perinatal depressive symptoms may present different results. causation. Therefore, further studies are needed to clarify the relation Therefore, more high-quality articles and reviews are needed to further among infertility, infertility treatment and perinatal depressive symp- improve this research. toms. Thirdly, nearly 41% included original studies are case-control stu- Competing interest statement dies or cross-sectional studies, which is particularly vulnerable to po- tential recall bias and information bias. Thus, real relationships be- The authors have no competing interests to report. tween infertility treatment and perinatal depressive symptoms may be hidden due to the bias in these studies. And the subgroup analyses by Acknowledgments specific types of infertility treatment were dropped because of the limited number of included studies on specific types of infertility Thanks for all the authors of this study. treatment. Thus, we performed a sensitivity analysis on data taking into account ART only. More relevant studies should be included in future Funding systematic reviews to provide further support for our results. Fourthly, perinatal depressive symptoms are complex with kinds of This research did not receive any specific grant from funding causative factors, including environmental, psychological and social agencies in the public, commercial, or not-for-profit sectors. factors. Although we extracted the OR and 95%CI adjusted for most covariates, some potential confounding factors were not adjusted for in References some studies [34,35,42–47,49,53,55–58]. Indeed, it is difficult to find an unbiased comparison group that is similar in all characteristics to the [1] Committee on Obstetric Practice, The American College of Obstetricians and infertility treatment group. Given that only eight studies (36.4%) Gynecologists Committee opinion no. 630. Screening for perinatal depression, Obstet. Gynecol. 125 (5) (2015) 1268–1271. [23,33,41,48,50–52,54] adjusted some potential confounding factors [2] American Psychiatric Association, Diagnostic and Statistical Manual of Mental (such as age, education, social support, history of depression, type of Disorders (DSM5), 5th ed., American Psychiatric Publishing, Arlington, VA, 2013. delivery and sex of infant), other unadjusted confounding factors for [3] B.N. Gaynes, N. Gavin, S. Meltzer-Brody, K.N. Lohr, T. Swinson, G. Gartlehner, et al., Perinatal depression: prevalence, screening accuracy, and screening perinatal depressive symptoms such as socio-demographic factors and

 S. Chen, et al. -RXUQDORI3V\FKRVRPDWLF5HVHDUFK  ²

outcomes, Evid. Rep. Technol. Assess. (Summ) 119 (2005) 1–8. Quality of life, postnatal depression and baby gender, J. Clin. Nurs. 17 (3) (2008) [4] S. Elisei, E. Lucarini, N. Murgia, L. Ferranti, L. Attademo, Perinatal depression: a 312–322. study of prevalence and of risk and protective factors, Psychiatr. Danub. 25 (Suppl. [33] C.D. Lynch, M.R. Prasad, Association between infertility treatment and symptoms of 2) (2013) S258–S262. postpartum depression, Fertil. Steril. 102 (5) (2014) 1416–1421. [5] A. Gajaria, A.V. Ravindran, Interventions for perinatal depression in low and [34] D.R. Listijono, S. Mooney, M. Chapman, A comparative analysis of postpartum middle-income countries: a systematic review, Asian J. Psychiatr. 37 (2018) maternal mental health in women following spontaneous or ART conception, J. 112–120. Psychosom. Obst. Gynaecol. 35 (2) (2014) 51–54. [6] J. Fisher, D.M.M. Cabral, V. Patel, A. Rahman, T. Tran, S. Holton, et al., Prevalence [35] F. Monti, F. Agostini, P. Fagandini, G.B. La Sala, I. Blickstein, Depressive symptoms and determinants of common perinatal mental disorders in women in low- and during late pregnancy and early parenthood following assisted reproductive tech- lower-middle-income countries: a systematic review, Bull. World Health Organ. 90 nology, Fertil. Steril. 91 (3) (2009) 851–857. (2) (2012) 139G–149G. [36] S.H. Lee, L.C. Liu, P.C. Kuo, M.S. Lee, Postpartum depression and correlated factors [7] F. Gressier, A. Letranchant, O. Cazas, A.L. Sutter-Dallay, B. Falissard, P. Hardy, Post- in women who received in vitro fertilization treatment, J. Womens partum depressive symptoms and medically assisted conception: a systematic re- Health 56 (4) (2011) 347–352. view and meta-analysis, Hum. Reprod. 30 (11) (2015) 2575–2586. [37] D. Moher, A. Liberati, J. Tetzlaff, D.G. Altman, Preferred reporting items for sys- [8] G. Sui, B. Pan, G. Liu, G. Liu, L. Wang, The long-term effects of maternal postnatal tematic reviews and meta-analyses: the PRISMA statement, Int. J. Surg. 8 (5) (2010) depression on a Child's intelligence quotient: a meta-analysis of prospective cohort 336–341. studies based on 974 cases, J. Clin. Psychiatry 77 (11) (2016) e1474–e1482. [38] M.W. O'Hara, K.L. Wisner, Perinatal mental illness: definition, description and ae- [9] L.E. Sockol, C.N. Epperson, J.P. Barber, The relationship between maternal attitudes tiology, Best Pract. Res. Clin. Obstet. Gynaecol. 28 (1) (2014) 3–12. and symptoms of depression and anxiety among pregnant and postpartum first-time [39] J.P. Higgins, S.G. Thompson, J.J. Deeks, D.G. Altman, Measuring inconsistency in , Arch. Womens Ment. Health 17 (3) (2014) 199–212. meta-analyses, BMJ 327 (7414) (2003) 557–560. [10] M.W. O'Hara, J.E. McCabe, Postpartum depression: current status and future di- [40] M. Egger, S.G. Davey, M. Schneider, C. Minder, Bias in meta-analysis detected by a rections, Annu. Rev. Clin. Psychol. 9 (2013) 379–407. simple, graphical test, BMJ 315 (7109) (1997) 629–634. [11] A.S. Garner, J.P. Shonkoff, Early childhood adversity, toxic stress, and the role of [41] A. Garcia-Blanco, V. Diago, D. Hervas, F. Ghosn, M. Vento, C. Chafer-Pericas, the pediatrician: translating developmental science into lifelong health, Pediatrics Anxiety and depressive symptoms, and stress biomarkers in pregnant women after 129 (1) (2012) e224–e231. in vitro fertilization: a prospective cohort study, Hum. Reprod. 33 (7) (2018) [12] J. Martini, J. Petzoldt, F. Einsle, K. Beesdo-Baum, M. Hofler, H.U. Wittchen, Risk 1237–1246. factors and course patterns of anxiety and depressive disorders during pregnancy [42] B. Gao, X. Cui, R. Liu, Q. Xie, Investigation and analysis of postpartum depression and after delivery: a prospective-longitudinal study, J. Affect. Disord. 175 (2015) and its influencing factors in 756 pregnant women, J. Prev. Med. Chin. People's Lib. 385–395. Army (08) (2017) 1001–1003. [13] A. Di Florio, S. Meltzer-Brody, Is postpartum depression a distinct disorder? Curr. [43] J. Fisher, K. Hammarberg, K. Wynter, J. McBain, F. Gibson, J. Boivin, et al., Assisted Psychiatr. Rep. 17 (10) (2015) 76. conception, maternal age and breastfeeding: an Australian cohort study, Acta [14] I.S. Yim, S.L. Tanner, C.M. Guardino, J. Hahn-Holbrook, S.C. Dunkel, Biological and Paediatr. 102 (10) (2013) 970–976. psychosocial predictors of postpartum depression: systematic review and call for [44] J. Fisher, K. Wynter, K. Hammarberg, J. McBain, F. Gibson, J. Boivin, et al., Age, integration, Annu. Rev. Clin. Psychol. 11 (2015) 99–137. mode of conception, health service use and pregnancy health: a prospective cohort [15] B. Weobong, A.A. Ten, S. Soremekun, A.A. Manu, S. Owusu-Agyei, M. Prince, et al., study of Australian women, BMC Pregnancy Childbirth 13 (2013). Association of antenatal depression with adverse consequences for the mother and [45] A. Akyuz, M. Seven, A. Devran, M. Demiralp, Infertility history is it a for newborn in rural Ghana: findings from the DON population-based cohort study, postpartum depression in Turkish women? J. Perinat. Neonatal Nurs 24 (2) (2010) PLoS One 9 (12) (2014) e116333. 137–145. [16] M.B. Sexton, H.A. Flynn, C. Lancaster, S.M. Marcus, S.C. McDonough, B.L. Volling, [46] R. Wang, Correlation between Postpartum Depression and Psychosocial Factors, et al., Predictors of recovery from prenatal depressive symptoms from pregnancy Chinese Medical Sciences University, 2008. through postpartum, J. Women's Health (Larchmt) 21 (1) (2012) 43–49. [47] L. Repokari, R.L. Punamaki, L. Unkila-Kallio, S. Vilska, P. Poikkeus, J. Sinkkonen, [17] S. Banti, M. Mauri, A. Oppo, C. Borri, C. Rambelli, D. Ramacciotti, et al., From the et al., Infertility treatment and marital relationships: a 1-year prospective study third month of pregnancy to 1 year postpartum. Prevalence, incidence, recurrence, among successfully treated ART couples and their controls, Hum. Reprod. 22 (5) and new onset of depression. Results from the perinatal depression-research & (2007) 1481–1491. screening unit study, Compr. Psychiatry 52 (4) (2011) 343–351. [48] R. Xie, G. He, J. Lei, Study on demographic and obstetrical factors of postpartum [18] D. Marin-Morales, F. Monge, C. Penacoba-Puente, Personality, depressive symptoms depression, J. Chin. Physician (4) (2006) 460–462. during pregnancy and their influence on postnatal depression in Spanish pregnant [49] Y. Yokoyama, Comparison of child-rearing problems between mothers with mul- Spanish women, Ann. Psicol.-Spain 30 (3) (2014) 908–915. tiple children who conceived after infertility treatment and mothers with multiple [19] M.E. Silverman, A. Reichenberg, D.A. Savitz, S. Cnattingius, P. Lichtenstein, children who conceived spontaneously, Res. 6 (2) (2003) 89–96. C.M. Hultman, et al., The risk factors for postpartum depression: a population-based [50] F.L. Gibson, J.A. Ungerer, C.C. Tennant, D.M. Saunders, Parental adjustment and study, Depress. Anxiety 34 (2) (2017) 178–187. attitudes to after in vitro fertilization, Fertil. Steril. 73 (3) (2000) [20] H. Xu, Y. Ding, Y. Ma, X. Xin, D. Zhang, Cesarean section and risk of postpartum 565–574. depression: a meta-analysis, J. Psychosom. Res. 97 (2017) 118–126. [51] J.C. Warmelink, C.A. Stramrood, K.M. Paarlberg, H.H. Haisma, A.J. Vingerhoets, [21] K. Chu, Q. Zhang, H. Han, C. Xu, W. Pang, Y. Ma, et al., A systematic review and W.C. Schultz, et al., Posttraumatic stress disorder, anxiety and depression following meta-analysis of nonpharmacological adjuvant interventions for patients under- conceived through fertility treatments: the effects of medically assisted going assisted reproductive technology treatment, Int. J. Gynaecol. Obstet. 139 (3) conception on postpartum well-being, J. Reprod. Med. 57 (3–4) (2012) 115–122. (2017) 268–277. [52] E. Mori, H. Iwata, K. Maehara, A. Sakajo, K. Tamakoshi, Relationship between the [22] J. Boivin, L. Bunting, J.A. Collins, K.G. Nygren, International estimates of infertility mode of conception and depressive symptoms during the first 6months post-partum prevalence and treatment-seeking: potential need and demand for infertility med- in Japan, Reprod. Med. Biol. 17 (3) (2018) 275–282. ical care, Hum. Reprod. 22 (6) (2007) 1506–1512. [53] J. Vikstrom, G. Sydsjo, M. Hammar, M. Bladh, A. Josefsson, Risk of postnatal de- [23] L.S. Joelsson, T. Tyden, K. Wanggren, M.K. Georgakis, J. Stern, A. Berglund, et al., pression or suicide after in vitro treatment: a nationwide case-control Anxiety and depression symptoms among sub-fertile women, women pregnant after study, BJOG 124 (3) (2017) 435–442. infertility treatment, and naturally pregnant women, Eur. Psychiat. 45 (2017) [54] P. Gambadauro, S. Iliadis, E. Brann, A. Skalkidou, Conception by means of in vitro 212–219. fertilization is not associated with maternal depressive symptoms during pregnancy [24] M.C. Inhorn, P. Patrizio, Infertility around the globe: new thinking on gender, re- or postpartum, Fertil. Steril. 108 (2) (2017) 325–332. productive technologies and global movements in the 21st century, Hum. Reprod. [55] I. Tendais, B. Figueiredo, Parents' anxiety and depression symptoms after successful Update 21 (4) (2015) 411–426. infertility treatment and spontaneous conception: does singleton/twin pregnancy [25] N. Kumar, A.K. Singh, Trends of male factor infertility, an important cause of in- matter? Hum. Reprod. 31 (10) (2016) 2303–2312. fertility: a review of literature, J. Hum. Reprod. Sci. 8 (4) (2015) 191–196. [56] N. Raguz, S.W. McDonald, A. Metcalfe, C. O'Quinn, S.C. Tough, Mental health [26] L. Bunting, J. Boivin, Decision-making about seeking medical advice in an internet outcomes of mothers who conceived using fertility treatment, Reprod. Health 11 sample of women trying to get pregnant, Hum. Reprod. 22 (6) (2007) 1662–1668. (2014). [27] Y. Frederiksen, I. Farver-Vestergaard, N.G. Skovgard, H.J. Ingerslev, R. Zachariae, [57] C.A. McMahon, J. Boivin, F.L. Gibson, J.R.W. Fisher, K. Hammarberg, K. Wynter, Efficacy of psychosocial interventions for psychological and pregnancy outcomes in et al., Older first-time mothers and early postpartum depression: a prospective infertile women and men: a systematic review and meta-analysis, BMJ Open 5 (1) cohort study of women conceiving spontaneously or with assisted reproductive (2015) e6592. technologies, Fertil. Steril. 96 (5) (2011) 1218–1224. [28] A. Eugster, A.J. Vingerhoets, Psychological aspects of in vitro fertilization: a review, [58] J. Fisher, K. Hammarberg, H. Baker, Assisted conception is a risk factor for postnatal Soc. Sci. Med. 48 (5) (1999) 575–589. mood disturbance and early parenting difficulties, Fertil. Steril. 84 (2) (2005) [29] P. Tao, R. Coates, B. Maycock, Investigating marital relationship in infertility: a 426–430. systematic review of quantitative studies, J. Reprod. Infertil. 13 (2) (2012) 71–80. [59] C. Olivius, B. Friden, G. Borg, C. Bergh, Psychological aspects of discontinuation of [30] L.E. Ross, K. McQueen, S. Vigod, C.L. Dennis, Risk for postpartum depression as- in vitro fertilization treatment, Fertil. Steril. 81 (2) (2004) 276. sociated with assisted reproductive technologies and multiple births: a systematic [60] L. Lin, H. Chu, J.S. Hodges, Alternative measures of between-study heterogeneity in review, Hum. Reprod. Update 17 (1) (2011) 96–106. meta-analysis: reducing the impact of outlying studies, Biometrics 73 (1) (2017) [31] M.N. Norhayati, N.H. Hazlina, A.R. Asrenee, W.M. Emilin, Magnitude and risk 156–166. factors for postpartum symptoms: a literature review, J. Affect. Disord. 175 (2015) [61] Y.L. Hsu, B.J. Kuo, Evaluations of emotional reactions and coping behaviors as well 34–52. as correlated factors for infertile couples receiving assisted reproductive technolo- [32] C. de Tychey, S. Briancon, J. Lighezzolo, E. Spitz, B. Kabuth, V. de Luigi, et al., gies, J. Nurs. Res. 10 (4) (2002) 291–302.