This study was designed as a prospective study, which was performed from January 2015 to August 2018.
Subjects who attended the antenatal clinic at a tertiary care hospital gave written consent. Pregnant women who are usually required to register for obstetrics in the 12th week of pregnancy were included. For obstetric reasons, they are examined on 14 scheduled dates throughout the pregnancy. Subjects were divided into no persistent PGP group and persistent PGP group based on if they had persistent PGP with radiation into one or both legs in the past four weeks. The pain should be severe enough to interfere with daily activities for more than a day.
The exclusion criteria were as follows: a history of other low back pain (LBP) such as specific LBP and other unspecific LBP(N=25), the disease or substance abuse prior to the pregnancy (N=21), severe diabetes, hypertension, and other diseases during the whole pregnancy (N = 9), sustained extreme fatigue (N = 1), major negative events related to the pregnancy such as the accident abortion, severe foetal abnormalities, and other possible reasons for perinatal depression (N = 17) , pregnancy by reproductive treatments (N = 2), and loss follow-up and/or incomplete data (N = 7). In the end, 387 pregnant women participate in the study.
Women experiencing PGP during the pregnancy were arranged for a consultation with a multidisciplinary team, including an orthopaedist, obstetrician, and physiotherapist. During the interview, they could confirm the specific type of LPP, the severity of symptoms, individualized treatments, i.e., medical history. If they fulfilled the inclusion and exclusion criteria and cooperated on further tests, they were scheduled to the multidisciplinary team.
A standardized and reliable assessment  was performed to determine PGP based on a history of pain provocation in different postures or daily living activities, pelvic pain provocations tests, and repeated flexion and extension movements with a wide range in standing and/or lying . The pelvic pain induction test includes sacral thrust , distraction test, compression test, posterior pelvic pain provocation test , and the MAT-test .
The definition of PGP is based on the European Guide . Pain occurs between the posterior iliac crest and the gluteal fold, especially near the sacroiliac joint, in combination or alone with the symphysis with decreased endurance during sitting, standing, and walking, and in positive clinical diagnostic tests to reproduce PGP. In addition, after repeated lumbar movement, there exists no nerve root syndrome, no repetitive pain, and/or symptom changes.
Persistent PGP after pregnancy is defined as a recurrent or persistent pain score ≥3 over a week around 6 months postpartum, as the previous research has shown that persistent PGP generally was significantly improved at this time point  and another study indicated that recurrent or persistent lumbar pain score ≥3 have been disabled to affect the quality of life . Women with this condition were asked to have a check around 6 months postpartum and 2 years postpartum. But the data “2 years” was used for the final analysis. Pain intensity was assessed using a self-reported scale with a range of 0-10 (0 for no pain, 10 for most pain), which was screened for PGP by interview or phone two years after delivery.
Quick Big Five Personality Test (QBFPT)
Personality traits were assessed using the QBFPT developed by Vermulst and Gerris (2005)  in the 12th week of pregnancy when the subjects were included. This five-trait personality measures include agreeableness (interpersonal trust and thoughtfulness), extraversion (sociability and high activity), conscientiousness (determination, diligence, and organization), neuroticism (distress, usableness to control urgency and deal with pressure, and unrealistic ideas), and openness to experience (aesthetic, sympathy diversity and intellectual curiosity). The measure is a 7-point Likert scale from "completely wrong" (1) to "completely right" (7). Thus, scores on each subscale ranged from 6 to 42. The personality types of individuals are not determined according to a certain range of points. Instead, the score of a participant represented a total score from the high and low scores obtained from each category. Cronbach's alpha of each subscale was as follows: 0.86 for conscientiousness, 0.78 for neuroticism, 0.80 for agreeableness, 0.81 for extraversion, and 0.73 for openness .
Data about the pregnancy
Data about the pregnancy were obtained by mail or in the clinic. The authors collected data including age, body mass index (BMI), educational background, annual household income, cesarean delivery, breastfeeding, unexpected sex of the baby, parity, sick leave, no or rare ability to take rest breaks at work, and PGP in the previous pregnancy. In the authors' country, the baby's sex is an important reason for the feelings of the mother and family members. Previous studies have shown that women who live in cultures where greater value is placed on sons are more likely to suffer from depression if they give birth to a girl [31,32]. Therefore, we investigate this important confounder in the present study. These data were completed by the subjects prior to the first evaluation. According to the rule that each variable in the analysis has at least 10 events, the number of variables was required to be limited .
2.4 Sample Size
Using G Power 220.127.116.11, the study power was calculated for the effect size of 0.3, error of the first type 0.05, and the total number of patients with the number of 75. The calculated study power equals 96.23%, which indicates good study power.
2.5 Statistical analysis
Ordinal variables were showed as proportions. Mean and standard deviation, or median and half-quartile ranges, respectively, were used to represent normally and non-normally distributed variables. Continuous variables and dichotomous variables were tested by the Student t-test and chi-square test, respectively. Kruskal-Wallis test was performed to compare nonparametric data at the ordinal level. Pearson correlation coefficient (coefficient, R) was used to test the correlation between the average score in the personality domain and VAS pain scores by controlling the parameter with a correlation value greater than 0.5. A stepwise multivariate logistic regression was used to detect the independent predictors of PGP after the univariate step of all significant variables with P<0.10 as the prerequisite for this stepwise model. Logistic regression analysis was performed to estimate odds ratios (OR) and 95% confidence intervals (CI) to determine the occurrence of PGP for each personality trait. After adjusting the confounding factors, the multivariable logistic model was established through the stepwise elimination of variables of interest in univariate analysis. P-values=0.05 andβ=0.8 were defined as the statistical significance and power analysis, respectively. SPSS version 22 (SPSS; Chicago, IL, USA) was applied in this study.