Identifying the factor structure of the Mother-to-Infant Bonding Scale for post-partum women and examining its consistency during pregnancy.

Abstract

Child abuse by mothers is increasingly recognized as an important issue worldwide. Identifying those at risk of being abusers among pregnant and early post-partum women is very important for the prevention of abuse. The Mother–Infant Bonding Questionnaire,1 established by Kumar and colleagues, can be used to evaluate bonding disorders during the post-partum period. Taylor and colleagues revised Kumar's Mother–Infant Bonding Questionnaire to develop the Mother-to-Infant Bonding Scale (MIBS).2 Marks and colleagues modified the MIBS by changing the wording of a few items and adding a new item. This version of the MIBS was translated into Japanese by Yoshida as the MIBS-J.3 In Japan, the MIBS-J is widely used in the fields of perinatal care and welfare services, as well as in research. To identify psychological and obstetric factors that affect bonding between a mother and a child, it is necessary to begin by accurately clarifying the factor structure of the MIBS. In the present study, we examined the factor structure of the MIBS-J using data on a large sample of women at 4 weeks after birth, exploring whether the factor structure obtained in the analysis of post-partum women was consistent during pregnancy by comparing MIBS-J data collected in early (approximately 12–15 weeks) and late (approximately 30–34 weeks) pregnancy. We conducted the Perinatal Mental Health Research Project jointly with the Department of Obstetrics and Gynecology, Niigata University Medical and Dental Hospital and 33 associated obstetric institutions in Niigata Prefecture. This study was approved by the Ethics Committee of Niigata University and the ethics committees of the participating obstetric institutions. This study included women in the early pregnancy stage (12–15 weeks) as well as those in late pregnancy (30–34 weeks) and post-partum at the start of the study. In total, 2471 women (aged 32.7 ± 5.0 years) consented to participate in this study. In the post-partum group (n = 1755), 46.0%, 73.5%, and 2.8% of women had no history of delivery excepting this time, had a natural conception this time, and had premature delivery this time, respectively. As our MIBS-J data were strongly skewed, we used the value after log transformation in the following exploratory factor analysis and confirmatory factor analysis. First, we randomly divided the post-partum women for whom we had MIBS-J data into two groups. Using the first group (n = 906), we conducted an exploratory factor analysis (spss, Version 25.0; IBM, Armonk, NY, USA) with promax rotation using the factor number determined by the scree test.4 We then created subscales consisting of items for which the loading on each factor was 0.3 or higher. Second, in the second group of post-partum women (n = 849), we conducted a confirmatory factor analysis (AMOS, Version 25.0.0; IBM, Armonk, NY, USA) using the factor structure extracted from the exploratory factor analysis. Third, analyses of multiple-group invariance were performed to determine the extent to which the factor structure that we found in the post-partum group was comparable across the different groups, such as early pregnancy (n = 1120) and late pregnancy (n = 1915). Four levels of measurement invariance are sequentially tested (configural, metric, scalar, and residual invariance), where each level introduces more equality constraints across groups.5 We decided to adopt the comparative fit index (CFI) and the root-mean-square error of approximation (RMSEA) as indices of evaluation for the goodness of fit between the models and the data. In the exploratory factor analysis with the first set of post-partum MIBS-J data, Factor 1 (Items 1, 4, 6, 8, and 10) and Factor 2 (Items 3, 5, and 9) were extracted (Table S1). Next, regarding these two extracted factors, we conducted a confirmatory factor analysis using the data from the remaining group of post-partum subjects. A two-factor structure was confirmed with good model fit (CFI = 0.948, RMSEA = 0.031; Fig. S1). Finally, examining measurement invariance between post-partum and late pregnancy, we obtained the following results: configural invariance, CFI = 0.947, RMSEA = 0.048; metric invariance, CFI = 0.942, RMSEA = 0.047; and scalar invariance, CFI = 0.870, RMSEA = 0.065. In the same way, examining measurement invariance between post-partum and early pregnancy, we obtained the following results: configural invariance, CFI = 0.941, RMSEA = 0.052; metric invariance, CFI = 0.938, RMSEA = 0.050; and scalar invariance, CFI = 0.838, RMSEA = 0.074. Accordingly, up to the step of metric invariance was ΔCFI ≤ 0.01, which could be said to be ‘weak measurement invariance.’6 Previous studies conducting factor analyses using MIBS-J data reported a two-factor structure, with the factors of ‘lack of affection’ (Items 1, 6, 8, and 10) and ‘anger/rejection’ (Items 2, 3, 5, and 7).3, 7 Using the two-factor structure, we also performed the confirmatory factor analyses with our second group of post-partum women (n = 849; CFI = 0.931, RMSEA = 0.036). The present study, which conducted exploratory and confirmatory factor analyses using MIBS-J data at 4 weeks after birth, also showed that the MIBS-J has a two-factor structure (Factor 1: Items 1, 4, 6, 8, and 10; and Factor 2: Items 3, 5, and 9). Considering the content of these items, it is possible to regard our Factor 1 as ‘lack of affection’ and our Factor 2 as ‘anger/rejection.’ Furthermore, examining measurement invariance among the post-partum period and the two stages of pregnancy, we found ‘weak measurement invariance.’ Therefore, when using the MIBS-J in pregnancy, it should be used with the recognition that post-partum bonding is only somewhat predictable. Although not done in this study, it is necessary to modify the wording of the MIBS-J so that it is tailored to the period of pregnancy and to confirm that the modified MIBS-J accurately predicts post-partum bonding. The authors have no conflicts of interest to declare. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.