At present, China’s maternity protection system mainly focuses on providing economic reimbursement to working women from state organizations, enterprises, and other types of economic and social organizations during maternity leave. However, there is a lack of economic and material systems to protect women’s breastfeeding rights. Since the beginning of 2016, following the implementation of the universal two-child policy, 26 provinces and cities have issued new regulations on maternity leave. Among them, Zhejiang, Jiangsu, Guangdong and other southeast coastal provinces, Beijing, Shanghai, and other megalopolises have extended maternity leave to 128 days, while most of the other provinces and cities have extended it to 158 days. On the whole, the vast majority of working women will return to work within six months after childbirth.

Regulations Concerning the Labor Protection of Female Staff and Workers of China stipulates that “Units employing female staff and workers with a baby under one year of age shall grant them two feeding (including bottle feeding) breaks of 30 minutes each during each shift.” However, the 30-minute feeding break includes both the feeding time and transportation time to and from the workplace. With the extension of cities, most people live far away from their workplace and spend quite a long time commuting.

Breastfeeding rooms in public and private entities have increased gradually in recent years. The All-China Federation of Trade Unions claimed that a total of 296,000 grass-root enterprises and labor unions have established women workers’ breastfeeding rooms throughout the country, covering 18,894 million women employees by September 2017. However, the proportion of qualified workplaces or public places remains low and worrisome conditions are not uncommon. Non-lactating people can enter the temporary breastfeeding room, which makes it embarrassing and insufficiently intimate. Some breastfeeding rooms have poor sanitary condition or lack the necessary facilities.

Maternal separation during the daytime reduces the child’s sucking, and in turn this affects the secretion of breast milk, which is a challenge to continued breastfeeding. Therefore, for the sake of career development, many working women choose to prepare breast milk in advance or add formula milk and complementary food supplements to their babies’ diets in order to wholeheartedly return to work.

Social support was introduced into sociological research in the 1970s, and has experienced long-term development. However, a consensus on the definition of social support has not yet been reached. Some researchers propose that social support is the availability of free assistance and services that the society provides for vulnerable populations. Other researchers believe that, from the perspective of the relationship between psychosocial stimulation and individual mental health, social support should be defined as the assistance a person receives from social ties that can help to relieve mental stress and improve social adaptability (He, 2002). Thus, social support is not just one-way care or assistance. In most cases it is a type of social exchange, as well as a social interaction between people. Nowadays, a social support network is generally regarded as a social network whereby individuals can obtain a variety of social supports (such as money, emotion, information, etc.).

In early studies on social support networks, there were no strict distinctions in types of social support. Social ties were generally considered beneficial to resolving difficulties in daily life. Later, researchers began to make distinctions concerning social support from different perspectives. Wellman and Wortley (1989) provided five different dimensions of social support, including emotional aid, minor services, major services, financial aid, and companionship. Cutrona and Russell (1990) indicated that social support included emotional aid, social integration support, reassurance of worth support, material aid, and information aid. Chen (2000) divided social support into objective support (including direct material aid and participation in social networks and groups), and subjective support (emotional support as perceived by individuals).

In this study, we divided the social support network of working women during lactation into material support, emotional support, and informational support, as illustrated in Figure 1.

Figure 1. 
Framework of social support theory.

• (1) Material support refers to direct material assistance, which is objective and visible. For lactating women, the regulations on breastfeeding time, scope of work, and facilities like a female workers’ health room, lounge for pregnant women, and breastfeeding rooms within the “Special Labor Protection Provisions on Female Workers” all pertain to material support.
• (2) Emotional support refers to subjective and perceived emotional aid. It refers to individuals’ emotional experience and satisfaction with the respect and support from society. During lactation, the emotional support for new mothers is mainly obtained from relatives and friends. Baranowski et al. (1983) found that Mexican-American women received most support from their mothers during lactation. With their mothers supporting breastfeeding, new mothers were more likely to breastfeed their babies; their mothers also provided emotional support, tangible aid, and informational support (Lupton & Whelan, 1998). In addition, supportive attitudes and assistance from friends also positively affected the decision to breastfeed (Clifford & McIntyre, 2008), whereas if friends did not support them or exerted negative pressure, new mothers were very likely to stop breastfeeding (Lupton & Whelan, 1998).
• (3) Informational support refers to the functions of the social network in promoting the flow of knowledge and information, and providing advice or guidance on resolving existing problems. With the development of modern society and an accelerated pace of life, a variety of breast milk substitutes has emerged, including the benefits of insisting that breastfeeding has become a new form of knowledge. Previous studies found that disseminating the benefits of breastfeeding could improve the likelihood of new mothers adopting breastfeeding behaviors. New mothers’ knowledge on breastfeeding was one of the independent factors affecting early breastfeeding behavior (Zhang et al., 2012).

This study used Beijing’s two-child women’s maternity survey, which was implemented by the Science and Technology Research Institute of the National Health and Family Planning Commission. The survey began on September 18, 2015, and covered 16 districts in Beijing. Among those who qualified for the universal two-child policy or the “two-child fertility policy for couples where either the husband or the wife is from a single-child family” in August, pregnant women who have already given birth to one child or women with two children but not currently pregnant, were eligible interviewees. To avoid potential errors in determining the number of children, the study only sampled those first marriage couples that were registered as permanent residents in Beijing. The total sample size was 500; pregnant women who had already given birth to one child accounted for 50% of the sample, and women with two children but not currently pregnant accounted for the other 50%. Among the latter group, 150 were between 20 and 34 years, and 100 were between 35 and 49.

The sampling process included two stages. In the first stage, 50 communities and towns were sampled from 331 communities and towns in 16 districts in Beijing, using the method of probability proportionate to size sampling (PPS). The larger the size of the district’s population, the more streets and towns were sampled. In the second stage, a quota sampling strategy was adapted to sample the target population within the selected streets and towns. Among the whole sample at this stage, 25 individuals were also randomly chosen for qualitative research. The quantitative study sample featured a wide spatial distribution coverage to better represent the needs of women of childbearing age in Beijing. Because this was not a random sample, we could not calculate the precision indicators, such as sampling errors, thus could not evaluate the representativeness of the sample with quantitative methods. However, the representativeness of this study sample was still an improvement over other nonrandom samples.

This study analyzed pregnancy, confinement, contraception, and abortion during the birth interval, and basic demographic information on education, residency, and income of the target population’s family members. Given the study’s aims, only those who were qualified according to the universal two-child policy, had already given birth to the second child, and had complete sample information were included. In total, 247 women were included in this study.

Cox regression (or proportional hazards regression) is a method for investigating the effect of several variables upon the time a specified event—in this case, weaning—takes to happen. It assumes all the covariates are multiplicatively related to the hazard, and reports hazard ratios as the effect of the estimated covariates. An important advantage is that Cox regression is suitable to analyzing censored data.

The formula of the Cox proportional hazards regression model is:

• X = (X₁, X₂, …X_p)

Where h(t, X) denotes the risk of weaning at time t, under the condition (X₁, X₂, … X_p); h₀(t) is the baseline risk. (X₁, X₂, … X_p) are independent variables including material support, emotional support, informational support, and control variables.

The formula of the hazard ratio is:

Dependent variable. breastfeeding time of the second child. The average breastfeeding time was 7.37 months (N = 247). Among the sample, 31.98% (N = 79) of the women had already weaned, and 68.02% (N = 168) were still breastfeeding their babies at the time of the survey.

Independent variable. independent variables were selected according to the social support network framework. Material support variables include “whether there is an independent office room or breastfeeding room at workplace” and “time from home to workplace”; emotional support includes “whether [one] received any practices and knowledge about breastfeeding from seniors within the family when giving birth to the second child” and “whether [one] received any practices and knowledge about breastfeeding from colleagues or friends when giving birth to the second child.” In fact, receiving breastfeeding information from seniors, relatives, or friends means more emotional support. Informational support includes “guidance received about breastfeeding when giving birth to the second child.” Guidance about breastfeeding from hospital or community doctors implies a positive attitude toward breastfeeding the second child as the percentage of infants 0–5 months of age who are fed exclusively with breast milk in China is low at about 20.8% by 2013 (UNICEF, 2018).

Control variables. control variables include age, whether the mother is an only child, occupation, years of education, family income, age at giving birth to the second child, confinement mode, childbirth interval, etc. (See Table 1 for the definition of variables).

First, we descriptively explored broad patterns among the mothers who had given birth to their second child. The sample included 247 women who were qualified according to the universal two-child policy and had already given birth to their second child. Among them, 66% were between 25 and 34 years, and 64% of the mothers were themselves only children. The majority had secondary education or higher (89%). Nearly 70% of the mothers were employed, among which nearly half were working at government institutions or state-owned enterprises. Characteristics of the sample are shown in Table 2.

Figure 2 illustrates the times of breastfeeding. Among the sample, 79 mothers had already weaned their second child. Their average breastfeeding time was 8.35 months, far below the two years suggested by the World Health Organization; 54.4% of the women breastfed for less than six months and only 10.1% breastfed for more than one year (WHO, 2017).

Figure 2. 
Time distribution of breastfeeding for the second child.

Figure 3 shows the distribution of occupation and breastfeeding time. The peaks of weaning time for women with and without jobs (or farmers) were all concentrated between three months (34%) and six months (34.6%). However, the second weaning-time peak differed significantly. This was between 9 and 12 months for professional women (28.3%), but was more than 12 months for women without a job or for farmers (19.2%). The descriptive results pointed to notable differences between professional women and those without a job (or farmers) in the likelihood of weaning after six months.

Figure 3. 
Weaning time and employment distribution.

Based on the framework of social support network for professional women during lactation, we performed the Cox regression on the breastfeeding time of women who had given birth to their second child, as shown in Table 3.

Log-log Kaplan-Meier curves were drawn to conduct the hypothesis test of significant variables in the Cox proportional hazards regression. Results showed that curves of “Whether there is an independent office room or breastfeeding room at the workplace,” “Whether the time from home to workplace is less than 15 minutes,” and “Whether [one] received maternal nutrition guidance” were parallel; curves of “Whether the mother is an only child” were not parallel in the first two months, but then became parallel the rest of the time, thus failing to reject the proportional hazard hypothesis. “Whether [one] received any practices and knowledge about breastfeeding from relatives or friends” resulted in rejecting the proportional hazard hypothesis (see Figure 4). As “Whether [one] received any practices and knowledge about breastfeeding from relatives or friends” resulted in rejecting the proportional hazard hypothesis, we adopted the General Stratified Cox (SC) Model to perform further analysis. The formula was as follows:

• g = 1, 2, …, k^*

Where g refers to variables that rejected the proportional hazard hypothesis; in this study, g referred to “Whether [one] received any practices and knowledge about breastfeeding from relatives and friends.”

Figure 4. 1. 
Whether there is independent office room or breastfeeding room at workplace

Figure 4. 2. 
Whether the time from home to workplace is less than 15 minutes

Figure 4. 3. 
Whether [one] received maternal nutrition guidance

Figure 4. 4. 
Whether the mother is an only child

Figure 4. 5. 
Whether [one] received any practices and knowledge about breastfeeding from relatives and friends

Based on the above analysis on the risk factors for discontinuing breastfeeding of professional women given the framework of social support network, we obtained the results described in Table 4 and can conclude the following:

Material support. Material support significantly affected professional women’s continuing to breastfeed. Women with an independent breastfeeding room in the workplace were less likely to discontinue breastfeeding compared with those who did not have access to a breastfeeding room as these rooms provide a private space for women to take and store breast milk, and avoid the trouble of returning home to breastfeed infants.

The time from home to workplace was also a significant factor affecting the discontinuation of breastfeeding. When commuting took more than 15 minutes from home to workplace, the risk of discontinuing breastfeeding increased correspondingly. Therefore, it is very necessary to establish breastfeeding rooms at the workplace.

Emotional support. Emotional support was an important risk factor affecting the discontinuation of breastfeeding. However, contrary to expectations, women who have received suggestions and practice guidance from relatives or friends were more likely to discontinue breastfeeding. The results of the qualitative interviews showed that incorrect suggestions from relatives or friends more than offsets their emotional support, and made new mothers prefer to discontinue breastfeeding. Therefore, in addition to providing new mothers with various information and services on breastfeeding, friends and relatives of the new mothers should also be educated on the benefits of breastfeeding, in order to build a family friendly social environment.

Informational support. The more knowledge on breastfeeding that professional women have, the less likely they are to discontinue breastfeeding. Receiving maternal nutrition guidance was an important factor that affected the discontinuation of breastfeeding. Women who have received professional guidance on maternal nutrition were less likely to discontinue breastfeeding than those who did not. Among those who have already weaned, the average breastfeeding time for professional women who have received maternal nutrition guidance was 8.2 months, three months longer than those who have not.

Individual characteristics. The effect of professional women’s individual characteristics on their breastfeeding decisions also cannot be ignored. Compared with non-only-child professional women, those only-child mothers were more likely to discontinue breastfeeding. It is generally believed that only children are more self-centered and demanding, and the double pressures of breastfeeding and work often make the only-child professional women discontinue breastfeeding prematurely.