Study design
This study involving 2,558 study participants was conducted with data from the prospective PWC study between June 2017 and June 2019 in the Gansu province of China. The female patients who had miscarriage for the first time, planned to undergo cause examination and continued to have child-bearing desire were enrolled in the prospective PWC study. We intended to explore the weight management of pregnant women and other factors that affect pregnancy outcomes. A face-to-face psychological interview was carried out at that time and the birth outcomes were obtained from the hospital medical record system after 2 years as following. This study was approved by the Ethics Committee of Gansu Provincial Maternity and Child-Care Hospital (REC 2017,GSFY 16).
Study participants
A pregnancy loss or miscarriage is defined as the spontaneous demise of a pregnancy before 24 weeks of gestation[2]. Women who experienced the losses of two or more pregnancies were defined as RPL cases ( their first miscarriage was also counted during the study)[3]. The non-RPL controls were randomly screened out and matched by maternal age. Potential cases and controls were excluded if the woman chose to induce abortion, faced infertility problems, previous ectopic pregnancies, molar pregnancies, stillbirth, neonatal death, suffered from pregnancy complications (i.e. pre-eclampsia, gestational diabetes…), chronic diseases or had a history of psychiatric disorders or addiction and was unavailable for analysis. On matching 1:1 case controls, we re-checked and eliminated those who did not meet the admission criteria and lost follow-up. Thus, a total of 1,132 cases (RPL group) and 1,426 controls (non-RPL group) were included in this study.
Data collection
An in-person structured interview was undertaken with the participants after their first miscarriage during the cohort study by a specially trained nurse at the hospital. Information collected during the interview included socioeconomic characteristics (e.g. maternal age, ethnicity, education, occupation, family monthly income) and lifestyle habits before miscarriage (e.g. active or passive smoking status, alcohol consumption, sleep quality and level of physical exercise). Information on the maternal menstrual and reproductive history (e.g. menstrual cycle, previous liveborn; gestational age, time limit of past pregnancy loss, whether or not have embryonic chromosome abnormalities about this miscarriage) were obtained from the participants’ medical records. Follow-up data of the subsequent pregnancy outcomes (e.g. RPL, no pregnancy, or ≥24 gestational weeks) were obtained via outpatient department visits and telephone interviews until 30 June 2019. The follow-up rate was 88.2%.
Measurements
The self-rating anxiety scale (SAS)[12] and the self-rating depression scale (SDS)[13] were used respectively (Chinese versions) to ascertain the women’s true situations regarding depression and anxiety during the first few days after their first miscarriage. The SDS and SAS both contain 20 items, using a point score from the baseline of one. The point scores indicate the following: 1=‘none or a little of the time’; 2=‘some of the time’; 3=‘a good part of the time’; and 4=‘most or all of the time’. The original total scores of the SDS and SAS of all women ranged from 20 to 80. The SDS and SAS indices were obtained by multiplying the total score on each questionnaire by 1.25 and converting to a 100-point scale. According to the primary screening diagnostic criteria of Chinese anxiety and depression norms: SAS≥50 and SDS≥53 were defined, respectively, as diagnosed anxiety and depression. The alpha Cronbach for SAS and SDS is 0.82 and 0.78.[14].
Statistical analysis
Indicators such as age and the family monthly income are artificially classified by the researchers as: ≤25, 26–29, 30–34, ≥35 and <2000, 2000–3999, 4000–5999, ≥6000[15]; The quantitative variables of the SAS score and SDS score were converted to qualitative variables by the scale from the book named “Manual of Mental Health rating scale (China)”. Anxiety: mild (score 50–59), moderate (score 60–69) and severe (score ≥70); depression: mild (score 53–62), moderate (score 63–72) and severe (score ≥72)[14].
A chi-square test was used to evaluate the statistically significant differences between the RPL and control groups. The multiple logistic regression analysis was performed to analyse the relationship between anxiety and depression symptoms from miscarriage and the occurrence of subsequent RPL. The data used for the regression equation were derived from the statistically significant variables and clinically relevant reported data. Maternal age, ethnicity, family monthly income, education, time of miscarriage at baseline, whether they had gave birth to a child, time limit of past pregnancy loss and foetal abnormalities were analysed as the potential confounding factors.
Furthermore, the effects of anxiety interaction with depression of different levels (no/mild/moderate/severe) on RPL was analysed by using "multiple interaction" from the logistic regression model and “addition interaction statistical analysis model” which were reported by Andersson T[16]. The odds ratios (ORA*B) and 95% confidence intervals (CIA*B) was used to analyse the interaction effect. Moreover, the relative excess risk due to interaction (RERI, RERI = ORAB - ORA - ORB + 1) value with 95% CI, the attributable proportion due to interaction (AP) and the synergy index (S) was used to reflect the additive interaction effect, which was more likely to evaluate the biological interaction between risk factors and the disease. Both RERI and AP are positive, and the confidence interval does not contain 0; S > 1, the confidence interval does not contain 1, indicating that they interact on the additive scale and are synergistic. (if RERP and AP are less than 0 and S is less than 1, then it is antagonistic). The results suggest a synergistic effect of anxiety and depression on the occurrence of RPL when the RERI>0 and the lower limit of 95% CI>0[16].
The SPSS software (SPSS Inc., Chicago, IL, USA, version 19.0) and the Excel software from Andersson T. were used to perform the statistical analyses. A P value of <0.05 was considered statistically significant.