Following the aims of this study, this research used a descriptive cross-sectional survey design using an anonymous structured questionnaire.

“Fatigue” indicates poor physical condition and sleep deprivation from childbirth and infant care by the postpartum mother.

“Mothers” indicates hospitalized women on 1st to 7th day after birth and who had a normal spontaneous delivery or a C-section.

The authors collected data using a purposive sample of participants from one general hospital in Tokyo Japan. Participants were mothers (N = 301) in their early postpartum period. The inclusion criteria were as follows: 1) after normal delivery or on repeat cesarean section day one to day through seven, 2) fluent in Japanese, and 3) permission to participate by the obstetric primary physician and nurse manager. Exclusion criteria were patients with: 1) emergency cesarean section, and/or 2) severe psychiatric disorders. This large-scale general hospital had a maternity unit and perinatal and life-saving transport support. An average of 750 cases per year in the hospital was assisted delivery. The average age of hospitalized mothers was early 30s. The ratio of cesarean sections was 27.7%. Almost all of the mothers and babies used the rooming-in system. For an adequate multiple regression analysis, it is appropriate that the number of samples is 10 times the number of parameters [14]. Assuming a collection rate of 30%, our calculated target number was 267 participants to ensure 80 responses.

The authors recruited eligible participants, during their hospital stay after childbirth, assisted by the nurse manager of the hospital. After the mothers delivered, the authors and the nurse manager confirmed that they met the inclusion criteria. The authors also informed the mothers verbally and in writing about the study’s purpose as well as about confidentiality, anonymity, and safety of their personal data. The authors provided an explanation of their right to withdraw from the study without penalty, and provided the mothers with a “withdrawal from the study” form. The data collection period was from May to December 2016. The self-report questionnaires were administered during their hospital stay. Mothers responded to the questionnaire in their rooms of the maternity unit and they placed their completed questionnaires in a secure box on the unit, providing convenience and anonymity.

This study was conducted after obtaining approval from the Ethics Committee for Epidemiological Studies at Tokyo Healthcare University Tokyo, Japan (approval no. 27-27). Following the Declaration of Helsinki, a written explanation regarding the study objectives, methods, protection of anonymity, and voluntary basis of participation was provided to each participant. It was also explained that the collected data would only be used for this study.

The quantitative data of demographics, fatigue scale and relaxation scale were collected.

The software programs SPSS version 23.0 (SPSS, Chicago, IL, USA) and AMOS version 23.0 (IBM, Armonk, NY, USA) were used for the data analyses. The significance level was set at 5%. The alpha coefficient and factor loadings were calculated to examine the reliability of the responses on each scale. As the aim of this study was to create a fatigue prediction model for postpartum mothers, the two main variables (fatigue and relaxation) were examined to clarify the reliability and validity of each scale. The Student t-test, one-way analysis of variance (ANOVA), stepwise multiple regression analysis and covariance structure analysis were used in three steps. For the first step, the correlations among the Fatigue, Relaxation, and participants’ characteristics were examined using the Student t-test and one-way ANOVA. For the second step, seven variables were entered as independent variables, to determine which variables affected fatigue as dependent variables, using stepwise multiple regression analysis. For the third step, a fatigue prediction model of factors was examined to determine whether these factors affected fatigue in postpartum mothers using structural equation modeling.

Table 1 shows the demographic characteristics of the participants. The mean age was 33.6 (SD 4.2) years; 151 participants (60.2%) were primipara and 100 participants (39.8%) were multipara. The number of postpartum days was in the range from the 1st to the 7th day. Mothers responding to the survey on the second day after delivery were the most frequent with 26.7%. There were 166 participants (66.1%) with normal vaginal delivery and 85 participants (33.9%) with caesarean section. Of the participants, 9.2% of the mothers had significant previous disease such as potential depression and bronchial asthma; 30.3% of the mothers had pregnancy complication such as threatened premature delivery and pregnancy anemia.

The factor structure was confirmed for each variable of fatigue, and relaxation. The construct validity was confirmed by factor analysis. The results of the factor analysis yielded a factor loading of 0.38 or more for all items, and the cumulative contribution rate was 46.4% or more for all scales. The reliability of the scales was confirmed using the Cronbach’s α coefficient, which indicated from 0.87 and 0.83 respectively. Accordingly, reliability and validity were reconfirmed for fatigue, and relaxation. The results are shown in Table 2.

*Both scale was assessed using exploratory factor analysis.

Fatigue and relaxation as dependent variables and participants’ characteristics as independent variables were analyzed using a two-sample Student t-test and one-way ANOVA. The results are shown in Table 3. Multiple comparisons of the means were performed using the Tukey honest significant difference method for fatigue and relaxation scores. Significant differences in fatigue and relaxation were found depending on the postpartum day. Fatigue was significantly associated with the postpartum days (F = 5.7, p < 0.001). In particular, mothers on postpartum day 7 had significantly higher fatigue scores than mothers on postpartum day 1. (p < 0.01). In addition, the mothers with previous disease had a significantly higher fatigue score than the mothers without previous disease (t [249] = 2.2, p < 0.05). Relaxation was significantly associated with the postpartum days (F = 4.0, p < 0.01). Mothers in postpartum day 7 had significantly lower relaxation scores than mothers during postpartum, day 1 (p < 0.05). The relaxation score of multipara mothers was significantly higher than the primipara mothers’ score (t [249] = 2.5, p < 0.05).

Student’s t-test; one-way ANOVA (Tukey’s muitiple comparisons test). *p < 0.05, **p < 0.01, ***p < 0.001.

To determine which variables affected fatigue, stepwise multiple regression analysis was carried out. Seven variables were entered as independent variables: age, postpartum days, delivery methods, birth experience, previous disease, pregnancy complication, and relaxation. The multiple regression analysis showed an association between fatigue and relaxation, postpartum days, and previous disease (Table 4). The three significant predictors of fatigue were relaxation (β = −0.30, p < 0.001), postpartum days (β = 0.26, p < 0.001), and previous disease (β = 0.14, p < 0.05). Postpartum days and previous disease had a positive impact, whereas relaxation had a negative impact on the fatigue scores. Four variables such as age, delivery methods, birth experience, and pregnancy complication were excluded.

A fatigue prediction model was created using structural equation modeling. The latent variables selected were seven factors: age, postpartum days, delivery methods, birth experience, previous disease, pregnancy complication, and relaxation. The age, delivery methods, birth experience, and pregnancy complication among the latent variables were excluded from the model because they had no significant path coefficient. Thus, the fatigue value for the mothers in the early postpartum period association model used in this study can reliably explain the relationships among the postpartum days, previous disease, and relaxation. The measurement model was then retested resulting in a better model fit index. The final model was created by drawing the partial correlation coefficient for the same error variables. The final values were as follows: GFI = 0.874; AGFI = 0.83; CFI = 0.883, RMSEA = 0.077, AIC =405.6, chi-square score = 317.58, degree of freedom = 127, and chi-square/degree of freedom ratio = 2.501. The final model is shown in Figure 1. The standardized estimates were all significant at the 5% level. The fit indices demonstrated an adequate fit of the model to the data. The fatigue score was significantly affected by relaxation (standardized partial regression coefficient: β = −0.32, p < 0.001), postpartum days (β = 0.31, p < 0.001),

Stepwise multiple regression analysis. *p < 0.05, ***p < 0.001.

and previous disease (β = 0.15, p < 0.05). Therefore, postpartum days and previous disease had a positive effect, whereas relaxation had negative effects on the fatigue score.

Participants’ characteristics are average age of 33.6 years, vaginal deliveries 66.1%, and primipara women 60.2%. The average age of primipara women in Japan was 30.7 years old [1], in other countries was 28.8 years old [19]. However, participants in this study were older than the average Japanese primipara. The caesarean section rate was higher (33.9%) compared to the cesarean section rate in Japan as a whole of 18.5% and the Japanese metropolitan of 25.6% [20]. In addition, 9.2% of participants previous disease and 30.3% had pregnancy complications. A likely explanation was because the hospital where the research was conducted with perinatal and life-saving transport support was located in the metropolitan area, and high-risk pregnant women were referred to the hospital for delivery. Participants in this study were an older and higher-risk group compared to the average Japanese primiparas; however, they were representatives of mothers giving birth at large hospitals in the metropolitan areas.

The factors related to fatigue of mothers in the early postpartum period were more days after birth, no relaxing, and had previous diseases. In particular, mothers with more postpartum days had significantly higher fatigue scores than mothers with fewer postpartum days. In addition to childbirth fatigue, sleep deprivation such as a rooming-in system, frequent breast-feeding, and care for her infants likely contributed to fatigue. Early in the postpartum period, mothers feel tired due to the delivery, and increased fatigue due to breast-feeding and infant-care [21]. There was a strong correlation between fatigue and depressive symptoms among women in the systematic review, although there was a high degree of heterogeneity between studies [22]. Because having previous disease during the postpartum period mothers do not feel fatigue, it is necessary for nurses to promote physical rest such as urging their mothers to take a nap, and to take care of their newborns. At the same time it is also important to encourage mothers’ relaxation to reduce their fatigue. For example, lavender aromatherapy for postpartum mothers resulted in improved physical and mood status [23]. Hand aromatherapy treatments for postpartum mothers increased their relaxation [24]. Hand and foot baths for women after cesarean section stabilized vital signs such as blood pressure and pulse [25]. In particular, it was suggested that necessary for postpartum mothers with previous diseases who are hospitalized to promote rest and relaxation.

A fatigue prediction model was finally selected as the most applicable among several models. For the final model the root mean square error of approximation (RMSEA) was 0.077 meeting the criterion of less than 0.08. Therefore the fit of the model was considered as reasonable or acceptable [26] [27] [28]. The minimum AIC was 405.6, and this final version model was applicated. Although the GFI and comparative fit index (CFI) were 0.9 or less, it met the criterion of GFI being more than AGFI. Significant path coefficients were obtained in all the items between the latent variables and the observed variables (See Figure 1).

The limitations of this research are the following three points. First, the research facility is one of many facilities in the metropolitan area, and data collection entailed convenience sampling, therefore the data are inherently biased. Also, because it was a survey at a facility with many high-risk mothers in the urban area, there was a high caesarean section rate at higher age therefore generalizability is limited. Second, the scales of fatigue and relaxation were used in the survey, were subjective measures of the individual and do not measure objective states. Also, we have not investigated the degree of depression and maternity blues. These states may have influenced the outcomes. Third, 16.4% of mothers could not response and there were missing values. Mothers may not have responded to the questionnaire due to strong fatigue and not being able to relax. Also, we did not confirm the breast feeding method such as complete breast milk, mixed breast milk, complete milk, and that too may have had an effect on fatigue. As a future challenge, for clinical implications, relaxation care for mothers with previous disease is recommended and making necessary alterations in care as postpartum days increase. Future research should include large-scale randomized cross-sectional surveys to, include young low risk mothers from rural areas.