This study used data from the China Migrants Dynamic Survey (CMDS) in 2017, which was provided by the Migrant Population Service Center. CMDS is an annual national sample survey of the internal migrants organized by the National Health Commission (NHC), which aims to understand the changing landscape of internal migration, the utilization of public health services, and the management of family planning services (16). The survey is conducted in 32 provincial units, which cover all 31 provinces and the Xinjiang Production and Construction Corps (XPCCs) of China. In order to understand the epidemic status of key diseases, in addition to the original survey, 8 cities were selected for a special survey in 2017. This study is based on this survey. Sampling sites included Qingdao, Suzhou, Guangzhou, Zhengzhou, Changsha, Jiulongpo District, Urumqi, and Xishuangbanna. From the perspective of location, Qingdao, Suzhou, and Guangzhou are located in the east, which is more economically developed; Zhengzhou and Changsha belong to the central region; Jiulongpo District, Xishuangbanna, and Urumqi belong to the western region. The data were standardized to adjust for bias caused by differences between regions.

The participants were selected by using a stratified multi-stage sampling method with a probability-proportional-to-size (PPS) approach. First, 31 provinces (autonomous regions and municipalities) and XPCCs were taken as the first-level sample units, eight representative provinces were selected. Second, one city in each province was selected, as follows: Qingdao, Suzhou, Guangzhou, Zhengzhou, Changsha, Jiulongpo District, Urumqi, and Xishuangbanna. Then, in each selected city according to the administrative division, township (town, street) attributes were sorted, as the third layer. Next, selected townships (towns and streets) by the PPS method. In the selected township (town, street), the village (neighborhood) committee was selected by the same method. All eligible subjects in the selected village (neighborhood) committees were invited to participate in the study. In each village or neighborhood, floating populations' households were selected by simple random sampling according to a random number table. The floating population that lived in the destination for more than 1 month, aged 15 and over, and were not registered in the district (county or city) were included in the study. Similarly, the registered families were selected according to the same sampling method as the floating populations' households. The registered population aged 15 and above at each sampling point was included in the study. A household needs to investigate only one mobile population or registered residence population. Finally, a total of 13,998 floating population and 14,000 registered residence population were surveyed. Information collected included participants' basic information, family members, health and public services, social integration, and epidemic influencing factors of key diseases, etc.

According to the research needs of this study, the dependent variable was children's vaccination. This variable was measured by the following question: “Has your child been vaccinated on time since birth?.” Possible answers were as follows: yes, no, and not applicable. We excluded all “not applicable” responses, resulting in a total of 12,199 participants included.

Data were processed and analyzed using STATA version 14.0. Descriptive statistics and a chi-square independence test were used to describe the information and compare the vaccination status of children under different sociodemographic characteristics. To further examine potential factors associated with risk or protection for children vaccination, a binary logistic regression analysis was used, and odds ratios (ORs) and 95% CIs were calculated. All tests were 2-tailed, and statistical significance was set at the 5% level.

The characteristics of the participants are present in Table 1. Of the 12,199 participants, 49.82% (6,078) participants were registered population, 50.18% (6,121) participants were floating population. Among the registered population, 51.60% participants were men, and the mean age was 33.74 years (SD, 7.66). The majority of participants (87.46%) were of Han nationality. In total, 52.58% of the individuals had received an education of junior college or above and 95.97% were married. For the floating population, 50.60% were men, the mean age was 34.33 years (SD, 8.29), 88.83% were of Han nationality. Compared with the registered population, the education level of the floating population is relatively low, and only 19.02% had a college degree or above. Chi-square testing showed that there were significant differences in registered and floating populations in the regional classification, age, ethnicity, education level, marital status, length of time to the nearest medical institution, health records, chronic disease, and cognition of basic public health service (p < 0.05).

In this study, the acceptance rates of health education for registered and floating populations were 87.07% (5,292/6,078) and 79.58% (4,871/6,121), respectively, and the difference was statistically significant (X² = 123.034, p < 0.001). Meanwhile, the acceptance rate of health education for all types of the floating populations was significantly lower than that of the registered population (p < 0.001; (see Table 2).

The data of this study showed that 94.39% (5,737/6,078) of children in registered residence were completely vaccinated, which was significantly higher than that of the floating children (91.68%, X² = 34.430, p < 0.001). Among the registered population, differences in regional classification (p < 0.001), ethnicity (p < 0.001), an education level (p < 0.001), marital status (p < 0.001), length of time to the nearest medical institution (p < 0.001), and cognition of basic public health service (p = 0.026) aspects between completely and incompletely vaccinated children were statistically significant. For the floating population, the analysis showed that there were significant differences in vaccination rates among floating children of different regional classifications, ethnicity, educational levels, marital status, length of time to the nearest medical institution, health records, cognition of basic public health service, and health education (p < 0.05; see Table 3).

To explore the influencing factors of complete vaccination of children in the registered population and floating population, logistic regression analysis was carried out. Sociodemographic variables and basic public health services (such as, health education and health records) were defined as independent variable X, and vaccination of children was defined as dependent variable Y, as shown in Table 4. The findings indicated that region, education level, and marital status were the main influencing factors of children's vaccination, regardless of registered residence, or floating population. Those who live in the East (OR₁ = 1.546, OR₂ = 1.834) have a high level of education (OR₁ = 2.341, OR₂ = 470), are married (OR₁ = 2.682, OR₂ = 2.094), are more likely to have their children fully vaccinated. Besides, ethnicity (OR = 2.219) and length of time to the nearest medical institution (OR = 1.353) were unique risk factors for complete vaccination of children in registered residence. For the floating population, health records (OR = 1.745) were a unique influencing factor for children's vaccination.

In China, planned immunization for children began in 1978. At present, the goal of reaching 85% of children's immunization rate has been achieved. Yet, with the rapid increase of the floating population, the immunization planning and management of floating children have become the focus of current work (20). Our research showed that 94.39% of children in registered residence were completely vaccinated and only 5.61% were incompletely vaccinated, but the proportion of incomplete vaccination among the floating children had reached 8.32%. These results were similar to others from diverse population groups. A study in southern Ethiopia found that compared with children born to non-migrant mothers, children born to rural-rural migrant mothers had significantly less chance of receiving full immunization coverage (21). Kagoné et al. conducted a qualitative study in Burkina Faso and also reported that migration was an important reason for incomplete vaccination (22). It can be seen that although China has made some achievements in immunization, the vaccination status of floating children still needs to be further improved. Addressing this issue will be of high significance to the goal of achieving a 100% vaccination rate among children in China.

Univariate analysis showed that regardless of the registered or floating population, regional classification, ethnicity, education level, marital status, length of time to the nearest medical institution, and cognition of basic public health service were related to complete vaccination of children. Parents in the eastern region, Han nationality, college degree or above, married, closer to medical service institutions, and familiar with basic public health services have a higher complete vaccination rate for their children. In addition, the difference in the utilization of basic public health services also has a significant impact on the vaccination of children of the floating population. The complete vaccination rate of children of the floating population who establish health records and understand basic public health services was higher. However, this result seems somewhat less significant in the registered population. According to logistic regression analysis, there were also differences in influencing factors of complete vaccination of children between the two groups.