This study was approved by the First Affiliated Hospital Ethics Committee of USTC (2019-P-013). All participants were informed regarding the aims, content, and procedures of the study. A self-reported survey questionnaire was disseminated between January 2017 and December 2017 in the Obstetric Department of Anhui Provincial Hospital, that is, the First Affiliated Hospital of the University of Science and Technology of China (USTC), a tertiary care hospital.
Participants, recruitment process and data collection procedures
Based on previous estimates of the prevalence of postpartum depression (30 pregnant women diagnosed with TPL and hospitalized), we assumed prevalence of PPD to be 75%. Using a formula and parameters  (N = 400Q/P, Q = 1-P, P = the prevalence of PPD) for calculating the sample size of our counting data, we needed a minimal sample size of 142.
Inclusion criteria for participants were as follows: (1) pregnant women diagnosed with TPL and hospitalized (according to the diagnostic criteria for the diagnosis of threatened premature birth by the obstetrician with abundant clinical experience; that is, before 37 weeks of pregnancy and appearing as irregular contractions, a small amount of vaginal bleeding, lower abdominal bulge, and other preterm clinical manifestations, and dilatation of the uterine mouth < 2 cm); (2) over 18 years of age; (3) without other severe neurologic diseases or severe psychiatric disorders; (4) having adequate cognitive ability to provide credible information; (5) having at least one caregiver who was one of the pregnant women’s family members and who provided informal care.
In total, 214 pregnant women were eligible. Among these eligible women, we had 9 cases of foetal death or spontaneous abortions, 10 women transferred to the Intensive Care Unit during the study, 31 women unwilling to participate in the study after delivery, and 14 women not providing valid answers (e.g., all the answers were the same, or missing data constituted over 20% of the total items); and these were excluded. As a result, a total of 150 pregnant women were included in the study. The flow chart of our study procedures is displayed in Fig. 1.
Women were approached one week after delivery by two trained senior nurses. Due to the physical discomfort during the pregnancy and labour and for other reasons, women were not very cooperative or interested in participating in the survey before giving birth. After delivery, the puerpera were encouraged to recall their experiences of being hospitalized with TPL; and they reported their social, psychologic and emotional status during the days in the hospital before and after delivery. Women were informed regarding the purpose of the study and asked whether they would be willing to participate and fill out the questionnaire. The potential participants were also advised that the survey was voluntary, that their responses would be kept confidential, and that they could withdraw from the study at any time. The questionnaires were anonymous but stamped with an ID number. The questionnaires took approximately 20–30 minutes to complete, and in order to facilitate their freedom to express their actual feelings, women were approached in the absence of their caregivers. For those cases when the women were unable to complete the questionnaire by themselves, the ward nurses read the questions to them, women responded to the questions, and the nurses then wrote the responses on paper. After each questionnaire was submitted, the nurses assessed whether all questions had been answered. If there were some unanswered questions, women were kindly asked whether they had any confusion or issues regarding the questions, and we encouraged them to complete the unanswered items.
Sociodemographic variables and clinical variables
The pregnant women’s sociodemographic characteristics and clinical information were obtained from electronic medical records. The information included women’s parity (unipara/multipara), age, occupation (unemployed/employed), residence (rural/urban), education (their highest level achieved, whether senior high school or below/junior college, or Bachelor/Master’s degree/Doctorate), number of terminations, having a scarred uterus before current delivery (no/yes), assisted reproductive technologies used (ART) (no/yes), length of hospital stay for TPL before delivery (days), premature delivery (no/yes), and delivery mode of present birth (natural labour/caesarean section). Information regarding insurance and expenses during TPL hospitalization were also obtained. However, since birth insurance can only be used once a year, in our study all pregnant women with TPL chose not to use birth insurance for their TPL hospitalization; and we did not include expenses in our analysis, since we believed that the economic aspect of SPB can more comprehensively and directly reflect the economic burden perceived by women.
Edinburgh Postnatal Depression Scale (EPDS).
The use of the EPDS during early postnatal days has been proven to be a simple and useful screening instrument for the onset of early postnatal depression that typically occurs within 1 week postpartum. In our study, a woman’s early sign of postpartum depression was assessed using a self-administered, locally validated EPDS . This scale consisted of 10 statements on common depressive symptoms, using 0~3 Likert-type responses to reflect the severity of symptoms. In our study, the recommended cut-off score of 9.5 (sensitivity, 82%; specificity, 86%) was used to evaluate the level of antepartum depression. The EPDS’s Cronbach’s α in this study was 0.918.
Self-Perceived Burden Scale (SPBS).
SPBS was used to measure women’s Self-Perceived Burden during pregnancy when hospitalized with TPL Cousineau et al.  first proposed a conceptual framework of SPB and developed a 25-item scale SPBS18.0 Subsequently, an abbreviated version of the SPBS was developed that consisted of 10 items. In this study, SPB was measured with a validated, abbreviated Chinese version of the SPBS  that contains three dimensions (physical, emotional, and economic aspects) and 10 items. It uses a 5-point Likert scale and each score is summed to create a score ranging from 0 (no burden) to 50 (maximal burden), with higher scores indicating more severe SPB. Scoring above 20 points indicates a noteworthy level of SPB . The Cronbach’s alpha of the SPBS in our study was 0.799.
Multidimensional Scale of Perceived Social Support(MSPSS)
We also used MSPSS to measure women’s perceived social support when they were hospitalized with TPL during pregnancy , which is a 12-item self-report scale that assesses perceived social support from three groups, namely family, friends, and significant others. For each group, it contains four specific statements and entails a 7-point Likert scale ranging from 1 (very strongly disagree) to 7 (very strongly agree). Item scores were aggregated, and a summed score ranging from 12 to 84 was obtained, with higher scores indicating greater perceived social support. Cronbach's α for the overall scale was 0.808.
Descriptive statistics were used to quantitatively describe or summarize basic features of the sample in our study (e.g., means, SDs, frequencies). We examined the associations of SPB with 1-week postpartum depression (PPD) using Spearman's correlation . To explore the risk factors for PPD, we used a multiple logistic regression model  with a stepwise approach, including SPB, perceived social support, and all sociodemographic and clinical factors available in the study as potential factors. We conducted all statistical analyses using SPSS 22.0. A P value was set at 0.05 for significance in the analysis.