DOI: https://doi.org/10.21203/rs.3.rs-1487980/v2
The evaluation of mothers' breastfeeding behavior within six weeks postpartum could help health workers to identify maternal breastfeeding shortcomings comprehensively, clarify nursing problems, and provide targeted interventions. However, no prior study was found and the study aimed to develop and validate the reliability and validity of mothers' breastfeeding behavior scale within six weeks postpartum.
The main two-step approach was used: (1) A qualitative pilot study using the purposive sampling method was adopted to test fitness, simplicity, and clarity of items with 30 mothers; (2) A cross-sectional survey using the convenient sampling method was conducted for item analysis and psychometric validation with 600 mothers.
The final version of the scale consisted of 36 items with seven dimensions, explaining 68.852% of the total variance. The Cronbach's α, split-half, and re-test coefficients were 0.958, 0.843, and 0.753, respectively. The validity of the scale: (1) Content validity: The content validity indexes (CVI) range of items were between 0.882 and 1.000. The scale-level-CVI was 0.990. (2) Structure validity: The fitting index were as follows: \({\chi }^{2}/df\)=2.239, RMR=0.049, RMSEA=0.069, TLI=0.893, CFI=0.903, IFI=0.904, PGFI=0.674, PNFI=0.763. (3) Convergent validity: The composite reliability and Average Variance Extraction (AVE) of 7 dimensions were between 0.876 and 0.920, 0.594 and 0.696. (4) Distinguish validity: The correlation coefficients were less than the square root of the AVE, excepting self-decision behavior, self-coping behavior, and self-control behavior. However, the fitting index of the original three-factor model was better than other new models with significant differences (P<0.001). (5) Calibration validity: The area under the curve was 0.860 or 0.898 when the scale was used to predict exclusive or any breastfeeding at 42 days. The correlation coefficient of maternal breasting feeding evaluation scale, breastfeeding self-efficacy short-form scale, and the scale were 0.569 and 0.674.
The newly developed mothers' breastfeeding behavior scale within six weeks postpartum consists of thirty-six items belonging to seven dimensions with good reliabilities and validities, a reliable and valid instrument to be used in future maternal breastfeeding behavior assessments and interventions.
Breastfeeding has been proven to protect mothers against breast cancer and cardiovascular diseases in later life [1, 2], offer infant better performance on mental and intellectual development [3], strengthen mother-infant bind and save related healthcare costs [4]. The World Health Organization, among other institutions, recommends that all infants should receive exclusive breastfeeding until six months postpartum and sets the goals for achieving six-month exclusive breastfeeding rate of at least 50% in 2025 [5]. However, despite the substantial evidence and many policies indicate that breastfeeding is a healthy behavior for infants, mothers and society, the recent estimates have shown that the Chinese exclusive breastfeeding rate at six months is 29.5%, there is 30.71% countries' six-months exclusive breastfeeding rate in the world at less 20%, which have a certain gap from the target of the WHO [6, 7]. Therefore, how to effectively increase the exclusive breastfeeding rate under six months has become a common problem that many countries and international organizations need to solve together. Many articles had discussed the situation of exclusive breastfeeding rate, and a rule was summarized that the beginning breastfeeding incidence is higher and drops sharply after being discharged from the hospital over time [8, 9]. Recently, a survey was conducted that 67.9% of mothers stopped exclusive breastfeeding within the first six weeks postpartum among mothers who gave up under the first six months, which emphasized the significance of six weeks postpartum and provided a unique perspective to upgrade the exclusive breastfeeding rate [10].
The first six weeks postpartum is also called the puerperium, a variable and particular period for maternal psychic recovery, social and emotional modifications [11]. The birth of a newborn breaks the original balance and pushes forward women to experience role change, complete role adaptation, and achieve maternal role attainment finally [12]. In the process, the mother needs to play the new role of mother and take responsibility for meeting the expectations given by society, such as breastfeeding [13]. The six weeks postpartum is the transition period for breastfeeding, and the transitions theory proposed by Professor Meleis defines the transition as a process from one stable state to another stable state when needs change, with an unstable phase in the midst[14]. To achieve breastfeeding, mothers within six weeks postpartum need to form new stable behavior patterns to replace actual behavior via learning breastfeeding knowledge and skills, coping with breastfeeding challenges and adjusting negative emotion [15, 16]. The new behavior pattern may appear as an inherent behavioral feature to affect maternal cognition, decision-making or action on breastfeeding during the six months postpartum or even the subsequent pregnancy [17]. However, due to the maternal own physical or/and psychological vulnerabilities and insufficient support from health professionals within six weeks postpartum, mothers are prone to emerge behavioral disorders and develop an ineffective breastfeeding behavior pattern resulting in poor breastfeeding condition eventually [18, 19]. Hence, our research team deem that exploration of breastfeeding behavior within six weeks postpartum is one of the breakthroughs to promote breastfeeding practice.
The term of breastfeeding behavior is widely used, but it seems to be a summary concept for breastfeeding mode and specific breastfeeding technologies and not yet a distinct concept in the literature, making it difficult to operationalize [20–23]. What is more, to our knowledge, the mothers' breastfeeding behavior within six weeks postpartum has not yet received attention from other researchers so far. In our previous research, we were inspired by the COM-B system, transitions theory, maternal role attainment theory and related literature to complete the conceptual analysis [24–29]. The mothers' breastfeeding behavior within six weeks postpartum was defined as mothers performed breastfeeding psychological reaction or movement depending on the internal regulation from own capability, motivation and opportunity under the external stimulation from the social environment, social resources, and infant behavior, and its attributes include self-regulation behavior, resource utilization behavior, and at-the-breast feeding behavior. A scientific and practical instrument is the necessary condition for subsequent relevant research. These existing scales focus on measuring why mothers give up breastfeeding early, such as breastfeeding knowledge, skills, attitudes, satisfaction, self-efficacy, and competency[6, 30–34]. Compared with other indicators, the observation and evaluation of maternal breastfeeding behavior are more intuitively and directly, which is more valuable for nurses and midwives to comprehensively identify potential shortcomings, clarify nursing problems, and provide targeted interventions. However, as far as we know, no prior study was found to develop a scale for evaluating breastfeeding behavior, especially for mothers within six weeks postpartum hence warrants careful study. To fill in this gap, this study aimed to develop the mothers' breastfeeding behavior scale within six weeks postpartum (MBBC-6W) and validate its reliability and validity to provide an assessment tool for related evaluation and intervention in future research.
A sequential exploration mixed-method study was carried out in the departments of obstetrics of Fujian Maternity and Child Health Hospital, a tertiary and baby-friendly specialist hospital in Fuzhou, Fujian province, P.R. of China. This scale development and scale validation study was implemented by a qualitative pilot study (Step one) and a formal cross-sectional investigation (Step two). Step one collected the reaction and suggestions of the target population to verify face validity by a face-to-face structured interview. Step two conducted a cross-sectional survey with the convenient sampling method to analyze item performance and finish psychometric validation.
Two samples were used to implement the development and validation study in this study. The first sample (Sample one) was collected to understand whether the designed items was fitness, simplicity, and clarity in the target population of mothers within six weeks postpartum. After generating a list of postpartum women from medical records, the purposive sampling strategy was employed to ensure the sample was diverse in terms of parity, delivery mode, education level, and infant whereabouts. Thirty mothers within six weeks postpartum who met included criteria were interviewed to participate in the pilot study from July 1, 2020, to July 15, 2020 [35]. The second sample (Sample two) was collected to encompass the item analysis to form the final scale version and finish psychometric validation by three key steps: (1) sampling, the potential participants delivered women in the obstetrics department was randomly recruited in one interaction with each mother separately between July 22, 2020, and October 7, 2020. (2) introducing, the author provided mothers with the information of the purpose of the survey, the process of the data collection, and the right to anonymity, confidentiality and withdrawal at any time. (3) screening, potential participants who were interested to participate in this study were interviewed to sign an informed consent and complete a final checklist for determining whether they met the inclusion criteria prior to starting the investigation via further communication. Regarding previous studies, 10–15 participants per item are necessary for construct validity, and 10%-20% of invalid questionnaires should be considered [6]. Therefore, the final calculated sample size interval was between 516 and 774, and the research group made the decision that 600 mothers were recruited informal investigation. The inclusion criteria of sample one and sample two were as follows: (1) mothers within six weeks postpartum have a singleton, full-term and healthy infant (one minute Apgar score > 8), (2) mothers feed their infant with breastfeeding, (3) the age of mother ≥ 20 years old, (4) mothers had attended middle school or above, and could communicate in Chinese generally. Mothers were excluded in the study if: (1) women were experiencing mother-infant separation, (2) women-infant dyads who were prohibited EBF by the professional health provider as medical conditions or other illnesses, (3) infants suffered from a severe disease, or (4) mothers with serious childbirth complications, emotional disorders or cognitive impairment.
For the pilot study, an author distributed the MBBC-6W (version-I) and interviewed participants to express their perspectives concerning the scale item, and maternal reactions and suggestions were recorded. And an initial consensus solution between mothers and investigators was proposed, which could provide a reference to form subsequent item generation of MBBC-6W (version-Ⅱ) in the following group discussion. For the formal investigation, an author invited eligible participants to complete the investigation as the following process: (1) add the WeChat or other messaging method with the consent of participants, (2) invite participants to select a time in the four-time periods of "3–10 days, 11–19 days, 20–29 days, 30–42 days" to finish the questionnaire. (3) sent the personal information form, MBBC-6W (version-Ⅱ), Maternal Breastfeeding Evaluation Scale (MBFES), and Breastfeeding Self-Efficacy Short Form Scale (BSES-SF) online at the appointed time. (4) selected randomly thirty participants to fill out the MBBC-6W (version-Ⅱ) again after the two weeks of completing the questionnaire. (5) sent the feeding mode questionnaire at the six weeks postpartum.
In this study, six research questionnaires were used for data collection. (1) The first questionnaire, the version-I of MBBC-6W, was formed by qualitative content analysis in our previous research, which consists of forty-five items categorized into three primary conceptual dimensions and seven secondary conceptual dimensions. (2) The personal information form designed by the researcher team included maternal socio-demographic questions such as maternal age, marital status, education, monthly income, ethnicity, occupation, maternity leave, parity, and delivery mode. (3) The version-Ⅱ of MBBC-6W, the updated version through discussing group discussions after completing the pilot study, was a 45-item scale according to five-point Likert scale (1 = it is strongly true about me, 2 = it is true about me, 3 = uncertainly; 4 = it is false about me, 5 = it is strongly false about me). (4) The MBFES was designed to measure maternal perception for breastfeeding experience belonging to three dimensions: maternal enjoyment/role attainment, infant satisfaction/ growth, and lifestyle/maternal body image [33]. The Chinese version of MBFES was used as one of the calibration questionnaires in this study, which consists of twenty-nine items with the Cronbach's α coefficient of 0.952 [36]. (5) The BSES-SF, another calibration questionnaire, is a fourteen-item scale with a Cronbach's α coefficient of 0.927 to evaluate maternal breastfeeding confidence and predict feeding mode at 4 and 8 weeks postpartum [34, 37]. (6) The last questionnaire was the feeding mode questionnaire, which consists of one question "How did you feed your baby between the first six months postpartum". Seven responses were set based on the breastfeeding definition from the United Nations Children's Fund [38].
The data were managed and analyzed using SPSS 25.0, Amos 25.0, and Stata 15.1 software. The result of the pilot study and item performance verification was analyzed to the develop final version scale. And then the reliability and validity were verified to test the psychometric validation relying on the data of the cross-sectional investigation. All of the continuous data were presented as mean-standard (\(\stackrel{-}{x}\pm s\)), and the categorical variables were presented as frequencies or percentages. In the stage of scale development, the research group discussed the suggestions and initial consensus formed by the participants and investigator in the pilot study and revised them one by one to form a new version of MBBC-6W (version-II). And then, the coefficient of variation (CV) method, critical ratio (CR) method, Cronbach's α coefficient method, correlation coefficient method, and exploratory factor analysis (EFA) were used for items analysis one by one. The items that met the above five screening methods were retained, and the selection criteria were as follows [6]: (1) The value of CV was calculated as CV=\(\stackrel{-}{x}/s\), and item with the CV<0.15 was deleted. (2) Based on the total score of the scale, if the CR value is less than 3.00 between the high group (top 27%) and low group (bottom 27%) or there was no statistically significant (P>0.05), the item was considered to delete. (3) The Cronbach's α coefficient of the remaining items were calculated when the items were deleted one by one. Moreover, the item that caused a significant change of coefficient was deleted. (4) The Pearson correlation method was used to calculate the correlation coefficient between items and items, corresponding dimensions, and scales. If the coefficient between two items with a high value (> 0.8), and the coefficient of the item and its dimension or scale was less than 0.4 or no statistical difference (P > 0.05), the item was considered for deletion. (5) The Kaiser-Meyer-Olkin (KMO) test (> 0.50) and the Bartlett test of sphericity (P<0.05) were estimated to verify whether the scale was suitable for factor analysis. An item with commonality<0.4, factor loading values<0.4, or factor loading on different common factors ≥ 0.4 was deleted.
In the stage of scale validation, Cronbach's α coefficient, split-half reliability, and re-test reliability were used to verify the scale's reliability, and their fitting criteria were 0.70, 0.80, and 0.70, respectively [6, 39]. The content validity, structure validity, convergence validity, distinguish validity, and calibration validity were used to verify the validity of this scale. (1) Content validity was assessed by the item-level content validity index (I-CVI) and average scale-level CVI (S-CVI/Ave) based on the experts' importance score, and their qualified values were 0.8 and 0.9 [39]. (2) Structure validity was verified by EFA and confirmatory factor analysis (CFA). In the EFA, the principal axis factor analysis extracted the common factors, and the Promax rotation was used to obtain the factor loading matrix. The standard for common factor extraction is as follows [40]: (a) eigen-values > 1.0, (b) scree plot, (c) cumulative variance contribution rate > 50% and (d) the number of items contained in any common factor ≥ 3 [6]. Meanwhile, to make up the methodological defects of scree plot, the parallel analysis (PA) and minimum average partial-correction (MPA) were used for determining the final number of common factors. In the CFA, considering the feature of each fitting index, the following index was adopted to assess the fitting effect: the normed chi-square (χ2/df), root mean square residual (RMR), root mean square error of approximation (RMSEA), tucker-lewis index (TLI), comparative fit index (CFI), incremental fit index (IFI), parsimony goodness of fit index (PGFI) and parsimony adjusted normed fit index (PNFI). If χ2/df ≤ 3, RMR < 0.05, RMSEA < 0.08, TLI > 0.90, CFI > 0.90, IFI > 0.90, PGFI > 0.50 and PNFI > 0.50, and it indicated that the model had a good fit. Among the 526 valid questionnaires, the numbered 1 to 263 were used for EFA, and the numbered 264 to 526 were used for CFA. (3) Convergence validity was assessed by standardized factor loadings, composite reliability (CR) and Average Variance Extraction (AVE), and their threshold level were 0.5, 0.6 and 0.5, respectively [41, 42]. (4) Distinguish validity was tested by comparing the value of AVE and the determination coefficient (r2). If the value of r2 was less than the square root of the AVE, the model was regarded as a good distinguish validity [43]. (5) Calibration validity consists of predictive validity and correlation validity. Receiver operating characteristic (ROC) analysis was used to test whether MBBC-6W could predict exclusive breastfeeding or any breastfeeding at 42 days. The Area Under ROC Curve (AUC) greater than 0.7 was considered predictive validity [44]. In addition, the correlation coefficient of MBFES, BSES-SF, and MBBC-6W were calculated to verify the correlation validity, and the acceptable value was 0.5 [45].
In the pilot study, all thirty invited mothers agreed to participate in this study, ranging from 24 to 36 years old, with a mean of 30 ± 3.06 years old. Participants covered different education levels of middle school (3 cases), high school (5 cases), college (18 cases), master and above (7 cases), delivery mode of vaginal childbirth (17 cases) and cesarean section (13 cases), parity of primipara (16 cases) and multipara (14 cases), whereabouts of rooming-in (17 cases) and neonatal observation unit (13 cases). In the formal investigation, 562 mothers completed the questionnaire in the 600 recruited potential participants with a response rate of 87.67%, and 448 mothers finally reported the feeding mode at six weeks postpartum with a response rate of 79.72%. The detailed flowchart is shown in Fig. 1. The sociodemographic information of included participants is reported in Table 1.
Items |
N (%) |
Items |
N (%) |
---|---|---|---|
Age (years) |
Ethnicity |
||
< 35 |
441 (83.84) |
Han nationality |
518 (98.48) |
≥ 35 |
85 (16.16) |
Other minorities |
8 (1.52) |
Marital status |
Occupation |
||
Married |
514 (97.72) |
Full-time mother |
173 (32.89) |
Unmarried |
10 (1.90) |
Full-time job |
319 (60.65) |
Other |
2 (0.38) |
Part-time job |
34 (6.46) |
Education |
Maternity leave |
||
Junior high school |
40 (7.60) |
< 42 days |
11 (2.09) |
High school |
68 (12.93) |
42 days-6 months |
269 (51.14) |
College |
337 (70.92) |
>6 months |
246 (46.77) |
Master and above |
45 (8.56) |
Parity |
|
Monthly income (yuan) |
Primipara |
295 (56.08) |
|
< 5000 |
48 (9.13) |
Multipara |
231 (43.92) |
5000–9999 |
186 (35.36) |
Delivery mode |
|
10000–14999 |
153 (29.09) |
Vaginal childbirth |
338 (64.26) |
≥ 15000 |
139 (26.43) |
Cesarean section |
188 (35.74) |
In the pilot study, twenty-two mothers thought all items of MBBC-6W (version-Ⅰ) were understandable and simple, and eight mothers proposed their confusions and suggestions for six items, mainly related to the problem of semantic expression. Based on the comments and the initial solution that had reached a consensus between the investigator and mother, the research group revised the controversial items without changing the evaluating purpose. And the expression of two items was modified, and four items were merged into two items. After this phase, version-Ⅱ of MBBC-6W was formed, consisting of forty-three items with the Likert 5-level point scale. The result of the items analysis for the MBBC-6W (version-Ⅱ) is shown in Table 2. Item 14 was excluded because the CV value was less than 0.15, and all of the remaining items met the retention standards of CV value, CR value, and Cronbach's α coefficient test. The results of the correlation coefficient method suggested that the item-item correlations were between 0.074 and 0.775, the item-corresponding dimension correlations were between 0.566 and 0.886, and the item-scale correlations were 0.295 and 0.754, in which the correlation coefficient of item 3 and scale score (r = 0.295) not passed the pre-set standard. Then, a total of six-round EFA were performed, and all of them passed the Bartlett test (P < 0.001) and KMO test ranging from 0.926 to 0.928. Three items (27, 28, 32) were excluded due to the commonality of less than 0.4, and two items (41, 43) were deleted because of the cross-loading in different common factors. After deleting the above five items, the factor loading and commonality of the remaining thirty-six items in the final version of MBBC-6W reached the retention standard, as shown in Table 2.
Item |
CV |
CR |
Cronbach’s α |
Correlation coefficient |
The last EFA result |
||
---|---|---|---|---|---|---|---|
Dimensions |
Scale |
Loading |
Commonality |
||||
1 |
0.160 |
10.540*** |
0.959 |
0.820** |
0.534** |
0.782 |
0.679 |
2 |
0.160 |
11.551*** |
0.958 |
0.813** |
0.587** |
0.868 |
0.594 |
3 |
0.190 |
5.365*** |
0.960 |
0.566** |
0.295** |
NA |
NA |
4 |
0.160 |
12.119*** |
0.958 |
0.813** |
0.634** |
0.707 |
0.709 |
5 |
0.220 |
15.658*** |
0.958 |
0.835** |
0.652** |
0.821 |
0.677 |
6 |
0.250 |
14.266*** |
0.958 |
0.835** |
0.646** |
0.767 |
0.731 |
7 |
0.270 |
19.569*** |
0.958 |
0.805** |
0.691** |
0.494 |
0.604 |
8 |
0.200 |
18.409*** |
0.958 |
0.873** |
0.739** |
0.676 |
0.699 |
9 |
0.190 |
15.998*** |
0.958 |
0.860** |
0.609** |
0.827 |
0.719 |
10 |
0.190 |
14.444*** |
0.958 |
0.819** |
0.635** |
0.688 |
0.615 |
11 |
0.250 |
16.750*** |
0.958 |
0.794** |
0.617** |
0.550 |
0.727 |
12 |
0.200 |
19.865*** |
0.958 |
0.802** |
0.697** |
0.503 |
0.684 |
13 |
0.170 |
19.934*** |
0.958 |
0.830** |
0.745** |
0.583 |
0.425 |
14 |
0.110 |
NA |
NA |
NA |
NA |
NA |
NA |
15 |
0.190 |
16.844*** |
0.958 |
0.809** |
0.655** |
0.651 |
0.539 |
16 |
0.170 |
14.966*** |
0.959 |
0.800** |
0.579** |
0.889 |
0.604 |
17 |
0.170 |
14.324*** |
0.958 |
0.782** |
0.584** |
0.687 |
0.619 |
18 |
0.160 |
13.122*** |
0.959 |
0.801** |
0.569** |
0.600 |
0.521 |
19 |
0.190 |
16.349*** |
0.958 |
0.851** |
0.658** |
0.736 |
0.570 |
20 |
0.200 |
14.576** |
0.958 |
0.824** |
0.611** |
0.616 |
0.482 |
21 |
0.220 |
16.941*** |
0.958 |
0.886** |
0.667** |
0.781 |
0.590 |
22 |
0.190 |
13.402*** |
0.958 |
0.835** |
0.608** |
0.738 |
0.466 |
23 |
0.170 |
14.874** |
0.958 |
0.825** |
0.673** |
0.482 |
0.620 |
24 |
0.170 |
13.285*** |
0.959 |
0.833** |
0.555** |
0.877 |
0.769 |
25 |
0.200 |
18.843*** |
0.958 |
0.861** |
0.700** |
0.735 |
0.568 |
26 |
0.220 |
15.685*** |
0.958 |
0.858** |
0.607** |
0.854 |
0.547 |
27 |
0.180 |
14.494*** |
0.959 |
0.581** |
0.539** |
NA |
NA |
28 |
0.350 |
12.228*** |
0.960 |
0.641** |
0.488** |
NA |
NA |
29 |
0.220 |
15.420*** |
0.958 |
0.806** |
0.602** |
0.868 |
0.721 |
30 |
0.230 |
14.994*** |
0.958 |
0.816** |
0.591** |
0.869 |
0.591 |
31 |
0.200 |
17.032*** |
0.958 |
0.796** |
0.649** |
0.570 |
0.457 |
32 |
0.250 |
12.865*** |
0.959 |
0.640** |
0.471** |
NA |
NA |
33 |
0.210 |
16.587*** |
0.958 |
0.769** |
0.620** |
0.642 |
0.770 |
34 |
0.180 |
17.700*** |
0.958 |
0.735** |
0.614** |
0.613 |
0.619 |
35 |
0.180 |
17.733*** |
0.958 |
0.738** |
0.620** |
0.492 |
0.473 |
36 |
0.340 |
10.812*** |
0.960 |
0.698** |
0.462** |
0.777 |
0.717 |
37 |
0.260 |
19.195*** |
0.958 |
0.831** |
0.646** |
0.935 |
0.668 |
38 |
0.250 |
20.894*** |
0.958 |
0.865** |
0.686** |
0.864 |
0.490 |
39 |
0.220 |
18.546*** |
0.958 |
0.774** |
0.623** |
0.565 |
0.624 |
40 |
0.220 |
21.859*** |
0.958 |
0.810** |
0.703** |
0.548 |
0.619 |
41 |
0.260 |
18.780*** |
0.958 |
0.758** |
0.656** |
NA |
NA |
42 |
0.240 |
21.296*** |
0.958 |
0.792** |
0.742** |
0.505 |
0.629 |
43 |
0.180 |
17.236*** |
0.958 |
0.699** |
0.690** |
NA |
NA |
Note: NA = Not Available, **P < 0.01, ***P < 0.001 |
The Cronbach's α coefficient, split-half coefficient, and re-test coefficient for the final version of MBBC-6W were 0.958, 0.843, and 0.753, respectively. The results of the validity test are as follows: (1) Content validity: According to the important score from seventeen experts, the range of I-CVI values in this scale was between 0.882 and 1.000. For the scale-level, the S-CVI/Ave values for scale was 0.990, and the values for three subscales were between 0.976 and 1.000. (2) Structure validity: The result of EFA presented that the scree plot began to level off after the seventh factor, the PA showed that the eigenvalue of the eighth factor was smaller than the virtual data, and the MPA analysis showed that the square value of average partial correction coefficient was the minimum when seven factors were extracted, as shown in Fig. 2. The above results indicated that seven factors should be extracted in line with the original assumptions. According to the feature of each factor, seven factors were named as self-decision behavior (F1), self-coping behavior (F2), self-control behavior (F3), resource coordination behavior (F4), resource acquisition behavior (F5), breastfeeding operation skills (F6) and breastfeeding self-perception (F7). The standardized path diagram of MBBC-6W is shown in Fig. 3. The seven-factors model accounted for 68.852% of the total variance, and each factor contained at least four items. The result of CFA presented that the seven-factor model has an acceptable fit, with χ2/df = 2.239 (χ2 = 1283.15, df = 573), RMR = 0.049, RMSEA = 0.069, TLI = 0.893, CFI = 0.903, IFI = 0.904, PGFI = 0.674, PNFI = 0.763. (3) Convergent validity: Fig. 3 showed the standardized factor loadings for each item in this scale, ranging from 0.84 to 0.93, whereby all values surpassed the recommended values. The CR values on seven factors from F1 to F7 were 0.879, 0.894, 0.904, 0.920, 0.876, 0.903, 0.917. Correspondingly, the AVE values form F1 to F7 were 0.594, 0.629, 0.654, 0.696, 0.638, 0.609, 0.650. (4) Distinguish validity: The result of the AVE method showed that, excepting F1, F2, and F3, the value of r2 were less than the square root of the AVE, as shown in Table 3. Considering the above three secondary dimensions belong to the same primary dimension, the Chi-square difference test was used to verify further the distinguish validity among them. The three factors were combined to form four new models: model 1 (F1 and F2, F3), model 2 (F1 and F3, F2), model 3 (F2 and F3, F1), and model 4 (F1, F2 and F3), which were compared with the original three-factor model on the fitting index. The result showed that the original model had the best fit and had the significant difference in the value of χ2 (P < 0.001), as shown in Table 4. (5) Calibration validity: The crude AUC was 0.860 when the scale was used to predict exclusive breastfeeding at 42 days postpartum, and the AUC was 0.857 after adjusting parity, delivery mode, and postpartum infant whereabouts. The crude AUC was 0.898 when the scale was used to predict any breastfeeding at 42 days postpartum. Since only two primiparas did not feed their infants with breastfeeding, only two covariates of delivery method and infant whereabouts were adjusted, and its adjusted AUC was 0.903. Furthermore, the correlation coefficient of MBBC-6W, MBFES, and BSES-SF were 0.569 and 0.674.
Factors |
F1 |
F2 |
F3 |
F4 |
F5 |
F6 |
F7 |
---|---|---|---|---|---|---|---|
F1 |
|||||||
F2 |
0.828 |
||||||
F3 |
0.777 |
0.857 |
|||||
F4 |
0.613 |
0.725 |
0.785 |
||||
F5 |
0.650 |
0.732 |
0.781 |
0.685 |
|||
F6 |
0.390 |
0.574 |
0.555 |
0.489 |
0.477 |
||
F7 |
0.363 |
0.522 |
0.531 |
0.450 |
0.470 |
0.727 |
|
\(\sqrt{AVE}\) |
0.771 |
0.793 |
0.809 |
0.834 |
0.799 |
0.780 |
0.806 |
Note: \(\sqrt{AVE}\) = the square root of the average variance extraction |
Model |
\({\chi }^{2}\) |
df |
\({\chi }^{2}/df\) |
RMR |
RMSEA |
TLI |
CFI |
IFI |
PGFI |
PNFI |
---|---|---|---|---|---|---|---|---|---|---|
Original model |
1283.15 |
573 |
2.239 |
0.049 |
0.069 |
0.893 |
0.903 |
0.904 |
0.674 |
0.763 |
Model 1*** |
1417.89 |
579 |
2.449 |
0.520 |
0.074 |
0.875 |
0.886 |
0.886 |
0.661 |
0.755 |
Model 2*** |
1478.16 |
579 |
2.553 |
0.520 |
0.077 |
0.878 |
0.867 |
0.798 |
0.648 |
0.748 |
Model 3*** |
1399.06 |
579 |
2.416 |
0.500 |
0.074 |
0.878 |
0.888 |
0.889 |
0.666 |
0.758 |
Model 4*** |
1560.29 |
584 |
2.672 |
0.053 |
0.080 |
0.856 |
0.867 |
0.868 |
0.647 |
0.745 |
Note: ***P < 0.001 |
The current study aimed to develop a theoretically-driven MBBC-6W and validate its reliability and validity in the Chinese population, which was achieved by following the systematic approach for scale development and comprehensive psychometric validation. In the stage of scale development, the initial 45-item scale was revised and reduced successively to the formal 36-item MBBC-6W through the pre-survey and item analysis. The final scale includes seven dimensions that were self-decision behavior (5 items), self-coping behavior (5 items), self-control behavior (5 items), resource coordination behavior (5 items), resource acquisition behavior (4 items), breastfeeding operation skills (6 items) and breastfeeding self-perception (6 items). All items of the scale are positive with Likert's five-point scale, and the scores are ranging from 36 points to 180 points. Meanwhile, the psychometric validation verified that the MBBC-6W had convincing internal reliability, external reliability, face validity, content validity, structure validity, convergence validity, distinguish validity, and calibration validity, meaning that it is a reliable and valid instrument to access mothers' breastfeeding behavior within six weeks postpartum.
The item was an essential part of the scale, and the items analysis was the critical step in scale development. In this study, five methods were used to analyze the capability of items from the perspectives of sensitivity, differentiation, internal consistency, representativeness, importance, and independence. The CV value reflected the sensitivity of items, and the finding represented that item 14 "For breastfeeding, I pay attention to own lifestyle (such as not drinking the strong tea, strong coffee, and alcoholic beverages, not smoking, not taking drugs, etc.)" had the poor sensitivity (CV < 0.15). The possible reason is related to the traditional Chinese puerperium culture, also known as "Zuo Yue Zi", which deems the lifestyle during the puerperium have a long-term impact on maternal health. Thus, most Chinese mothers and their social support system try their best to keep a healthy lifestyle in the puerperium [46]. The correlation coefficient of the scale scores and item 3 "I decided to breastfeed is not to meet the expectations of my husband, family or others" was lower than the standard, meaning item 3 could not represent the scale. The plausible explanation could be that, with the rise of female consciousness, modern independent women decide to breastfeed their infant because of the benefits for maternal and infant health, rather than to cater to other people [7]. The commonality of item 27 "I can recognize the sign of infant hunger accurately and timely" was less than the standard, presenting that the importance for scale was poor, which could be caused by maternal different understanding for "infant hunger sign". Item 28 "When breastfeeding, I will put the baby's face close to the breast, and align the tip of baby's nose at my nipple instead of mouth" and item 32 "For latch well, I support the breast with a C-shape (Place the thumb on top of the breast, and the other four fingers on the chest wall under the breast)" were deleted due to the lower commonality. The possible reason is the lower completion rate during breastfeeding, which is consistent with the maternal feedback during the daily clinical breastfeeding instruction. Both item 28 and item 32 were designed because the above-mentioned feeding techniques could help infants latch nipples well. The remaining item 33, "During the breastfeeding, my infant can always contain the whole nipple, and most of the areola in the mouth, " could evaluate the latch results more intuitively. Items 41 "I think I have sufficient breastmilk to meet infant demand" and item 43 " Breastfeeding makes me feel like a good mother", were deleted because of the cross-loading, indicating that the independence was not recognized. The deletion of item 43 may be related to it is a comprehensive variable without particularity. The perception of breast milk production is a manifestation of confidence, and the dimension of self-decision behavior also contains the item evaluating maternal confidence on breastfeeding, which may be the reason why item 41 belongs to both dimensions [47]. Fortunately, the measuring purpose of item 40 "I think breastfeeding made baby gain a healthy weight" is similar to item 41, and it could more objectively evaluate whether breast milk is sufficient.
Reliability, reflecting the internal consistency and stability of scale, includes internal and external reliability. The Cronbach's α coefficient and the split-half coefficient were used to verify the internal reliability. The results showed that Cronbach's α coefficient and split-half coefficient of scale were above reference, indicating that the scale has an excellent internal consistency according to the standard classification recommendation [48, 49]. The result showed that the re-test coefficient of the scale was exceeded the suggested value, meaning the scale has the capability for obtaining a stable result under similar external conditions and has acceptable external reliability [50]. The validity, referring to the ability of the scale to reflect the actual characteristics of the measuring target, consists of face validity, content validity, structure validity, convergence validity, distinguish validity, and calibration validity. In the pilot study, mothers with different education levels, delivery mode, parity, infant whereabouts were invited in the investigation, ensuring that comprehensive feedback was received from mothers with different characteristics, which provided a good condition to test scale applicability. Among them, most of the mothers (73.3%) had no doubts about items, and eight mothers pointed out some confusion in expression, which were subsequently resolved by the research group referring to maternal own suggestions thus the final revised scale was understandable with good face validity. Seventeen experts reviewed the content validity, and the findings showed that both item-level and scale-level CVI were acceptable, demonstrating that the MBBC-6W was compatible with the final measuring target. All of the methods often used to extract factors showed that the MBBC-6W yielded seven common factors, which matched the theoretical model. Moreover, the MBBC-6W could explain 68.852% of the total variance, confirming that the scale could capture the main characteristics of the mothers' breastfeeding behavior within six weeks postpartum. On the other hand, the results of CFA showed that, except that TLI is equal to 0.893, all of the remaining fitting indexes of the seven-factor model meet statistical requirements. However, the researcher pointed out that CFI is a valid reference when slightly smaller TLI than the standard value [51]. Therefore, although the value of TLI was less than 0.9, the CFI was equal to 0.903 indicating that the seven-factor MBBC-6W with acceptable structure validity. The factor loadings of items are more significant than the lower limit, indicating the item-belonging dimension with a good convergence validity. Moreover, the value of CR and AVE exceeded the reference, meaning that good convergence validity between different dimensions. Distinguish validity, reflecting the degree of distinction between different dimensions, were analyzed by the AVE method and Chi-square difference test in this study. Although the distinctive degree of F1, F2 and F3 were questioned in the result of the AVE method, the chi-square difference test confirmed that the original three-factor model was significantly better than one-factor model and the two-factor model, indicating that there was a distinction among F1, F2 and F3. The ROC analysis showed that both AUC was more significant than 0.7 when the scale was used to predict exclusive breastfeeding and any breastfeeding at 42 days, and they were not affected by covariances, suggesting that the scale was a valid tool to predict breastfeeding mode at 42 days. The MBBC-6W also exhibited specific correlations with MBFES and BSES-SF, further evidence that MBFES is a valid scale, as breastfeeding behavior is related to breastfeeding satisfaction and self-efficacy [52].
A new scale named MBBC-6W was developed and validated in this study, which was designed to measure breastfeeding behavior among the mothers within six weeks postpartum. Since there is no specific instrument to evaluate breastfeeding behavior, the current study has important theoretical and practical implications. In contrast to the existing scales, the scale is conceptually appealing because it directly concentrates on measuring behavior itself and innovatively focuses on the particular group of mothers within six weeks postpartum, which could lay a foundation for the evaluation and intervention in future research and clinical practice. However, due to the restrictions of time or own conditions, some limitations need to be considered. Firstly, the participants were recruited from the Fujian Maternity and Child Health Hospital, which might not represent the whole Chinese mothers' characteristics of breastfeeding behaviors within six weeks postpartum. Thus, the universality of the scale was limited, which could be enhanced through a multi-center investigation that included mothers of different races, cultures, geography, and medical institutions. Secondly, the convenient sampling method may affect the sampling representativeness, which could be improved through using random sampling in future research. Thirdly, even though a more extensive sample investigation was implemented, the sample size for EFA and CFA was insufficient because the sample was bisected for the credibility of factor analysis. Subsequent research should continue to expand the sample size to get a more stable and reliable model. Fourthly, the psychometric verification of MBBC-6W was based on the classical testing theory (CTT), which has inherent limitations, such as it is challenging to satisfy the assumption of error and accurate score. Further research should use the item response theory or multi-dimensional item response theory to overcome the limitation and provide more information for the psychometric testing of MBBC-6W. Finally, MBBC-6W has not yet been applied in clinical practice. The scale should be further used to investigate the current status and potential risk factors to help policymakers and health workers find problems and formulate corresponding strategies that are conducive to breastfeeding and achieve the goal of optimal breastfeeding practice ultimately.
Behavior is the comprehensive manifestation of psychological reaction or movement, and it is necessary to comprehend the particular breastfeeding behaviors that puerperium mothers gain due to the particularity and importance of the puerperium. The newly developed MBBC-6W is a reliable and valid instrument for accessing the breastfeeding behavior of Chinese mothers within six weeks postpartum, making it possible for us to identify the current state of maternal breastfeeding behavior and the details that should be improved. Further research is needed to explore the specific strategies that can promote best practices in breastfeeding.
Acknowledgements
Thanks to all participants for their valuable contribution to this study.
Authors’ contributions
J.-L.W. designed the study, collect and analyze the data, drafted the first manuscript, and is the first author of the final manuscript as submitted. S.-Q. P. conceptualized the study, reviewed, revised the process of the study design. X.-M. J. provided the survey conditions, obtained funding and revised the paper substantively. Y. L. participated in the data design and data collection. Q.-X. Z. reviewed and revised the paper. All authors agreed to the contribution and approved the final manuscript.
Funding
The study was supported by the Nursing Research Fund of Fujian Maternity and Child Health Hospital [YCXMH20-03] and Innovation and Entrepreneurship Training Program for College Students in Fujian Province [202010393021].
Availability of data and materials
All data generated or analysed during this study are included in this published article.
Ethics approval and consent to participate
Provided details and obtained informed consent to all participants. Those completing the hardcopy version in pilot study provide written consent and those completing an online survey provide an consent after provision of survey information, and prior to the survey items. Approval was granted by the Institute Ethics Committee of Fujian Maternity and Children's Hospital (ethic code = 2019-153).
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Author details
1 School of Nursing, Fujian University of Traditional Chinese Medicine, Fuzhou, Fujian, China
2 Department of nursing, Fujian Maternity and Child Health Hospital College of Clinical Medicine for Obstetrics & Gynecology and Pediatrics, Fujian Medical University, Fuzhou, Fujian, China.
3 Department of obstetrics, Fujian Maternity and Child Health Hospital College of Clinical Medicine for Obstetrics & Gynecology and Pediatrics, Fujian Medical University, Fuzhou, Fujian, China.