This study adopted a combined approach, involving a questionnaire survey conducted among a relatively large sample and a video observation study conducted among a relatively small sample. The survey was reported according to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines, and the video observation was reported in adherence to the COREQ (Consolidated Criteria for Reporting Qualitative Research) checklist.

Approval to conduct this study was obtained from the Ethics Committee of Peking University (IRB00001052-20047; approval date: August 1, 2020). This study has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All caregivers signed a written informed consent form before any assessment. Adhering to privacy and confidentiality guidelines, all identifiable participant details, including names, hospital numbers, and recognizable features in images, were anonymized or omitted. Participants of the survey received a small toy valued at 5 RMB (US $0.72), while participants of the video observation received 100 RMB (US $14).

Among the 444 participants recruited, 16 did not complete the questionnaire; 2 provided ineligible informed consent; 1 provided a duplicate questionnaire; and 16 had a preterm child, had a child with an abnormal birth weight, or had a child of inappropriate age. Finally, a total of 409 participants were included in the survey. Most caregivers were parents (310/409, 75.8%; including 301 mothers), had a high school education or below (282/409, 68.9%), were unemployed (217/409, 53.1%), and had a normal weight status (207/409, 51.7%). The majority of children were aged 24‐35 months (217/409, 53.1%), were male (214/409, 52.3%), were not the first child in the family (211/409, 51.6%), had a normal weight status (372/409, 93.0%), lived in an extended family (310/409, 75.8%), and had parents with a monthly household income of >5000 RMB (>US $721; 247/409, 60.4%) (Table 1).

A total of 42 caregiver-child pairs were included in the video observation, including 21 pairs each of children aged 6‐23 months and those aged 24‐35 months. In general, the characteristics of the participants in the video observation and survey were similar (Table 1).

The survey measured 7 dimensions of feeding practices. The scores for responsive feeding practices were generally higher than those for nonresponsive feeding practices. The median scores for responsiveness to cues, modeling, and active communication and interaction were 4.00, 3.67, and 3.88, respectively. The median scores for controlling, pressure to eat, the use of food as a reward, and emotional feeding were 3.33, 2.33, 3.00, and 2.67, respectively. The overall responsive feeding score was 3.52 (Table 2).

Video observation additionally assessed the dimension of “creating a good meal environment” that could not be examined by the survey. Except for caregivers’ emotional feeding, all dimensions could be observed by the videos (Table 3). A total of 105 videos were recorded, including 27 videos for breakfast, 37 for lunch, and 41 for dinner. In general, the occurrences of responsive feeding practices (75.6%‐97.6%) were higher than those of nonresponsive feeding practices (0%‐46.3%) at main meals. There was little variation in the occurrence of each practice during breakfast, lunch, and dinner. In addition, among nonresponsive feeding, controlling and pressure to eat were more likely to occur than the use of food as a reward. The practice of the use of food as a reward appeared at breakfast and dinner but not at lunch (Table 3). Notably, the practice of the use of food as a reward is very likely to occur during snacking, which was not recorded in this study. Therefore, the occurrence of the use of food as a reward reported here is likely to be underestimated. In terms of the observed number of practices at each dining occasion, the observed number of responsive feeding practices (2-11) was higher than that of nonresponsive feeding practices (0). The score for creating a good meal environment was high at 4 points. For each feeding practice, the observed number and scores at breakfast, lunch, and dinner were similar (Table 4).

Observations from the videos provided rich and contextual narratives that illustrated responsive/nonresponsive feeding practices. For instance, the practice of “using food as a reward” frequently involved ultraprocessed foods such as sugary drinks and spicy strips. In 1 episode, a mother promised to provide her child with spicy strips for the purpose of encouraging the child’s consumption of the main meal. She said: “OK…spicy strips… If you finish this meal, I will give you spicy strips.” Moreover, the practice of “pressure to eat” was observed in the following three dimensions: (1) continue feeding despite the child’s refusal (eg, turning the head away and crying), (2) use distractions (eg, toys and mobile phones) to coax the child into eating, and (3) insist on eating even after satiety cues (eg, verbally expressing fullness and pushing the bowl away). Video descriptions of “responsiveness to cues” have been presented in our previous publication [6].

The results from univariate analyses that explored factors influencing responsive/nonresponsive feeding practices are shown in Multimedia Appendix 3. After controlling for potential confounders, the results from multivariate analyses are shown in Table 5. For the overall responsive feeding score, caregivers who were nonparents (β=0.13, 95% CI 0.05-0.21) and had a monthly household income of >5000 RMB (>US $721; β=0.07, 95% CI 0.01-0.14) had a higher overall score, while the score was lower among unemployed caregivers (β=−0.07, 95% CI −0.13 to −0.01). For nonresponsive feeding practices, caregivers whose children were not the first child (β=−0.20, 95% CI −0.39 to −0.02) and whose children were aged 24‐35 months (β=−0.20, 95% CI −0.38 to −0.02) had a lower score. The score for pressure to eat was higher among unemployed caregivers (β=0.27, 95% CI 0.11-0.42). The score for the use of food as a reward was lower for caregivers who were nonparents (β=−0.45, 95% CI −0.69 to −0.22). The score for emotional feeding was higher among caregivers whose children were not the first child (β=0.18, 95% CI 0.03-0.34). In terms of responsive feeding practices, the score for responsiveness to cues was higher among caregivers who were nonparents (β=0.43, 95% CI 0.25-0.60), while the score was lower among unemployed caregivers (β=−0.19, 95% CI −0.33 to −0.05). The score for modeling was higher among caregivers whose weight status was underweight (β=0.44, 95% CI 0.07-0.80), and the score was lower among caregivers whose children were aged 24‐35 months (β=0.22, 95% CI 0.07-0.37). Caregivers whose children were not the first child (β=−0.17, 95% CI −0.29 to −0.54) had a lower score for active communication and interaction.

This study used a combination of survey and video observation methods to obtain a comprehensive understanding of the current status of caregivers’ responsive and nonresponsive feeding practices among children aged 6‐35 months and performed multivariate analyses of the survey data to explore factors influencing responsive and nonresponsive feeding practices. Both survey and video observation methods found that the frequencies of responsive feeding practices were higher than those of nonresponsive feeding practices. Video observation revealed that there was little variation in the occurrence/score of each feeding practice in main meals (ie, breakfast, lunch, and dinner). The survey demonstrated that caregivers’ relationships with their children, employment status, and weight status; children’s age and order; and the monthly household income were factors influencing responsive/nonresponsive feeding practices.

Both survey and video observation methods in this study found that caregivers adopted responsive feeding practices more frequently than nonresponsive feeding practices. In our survey, the scores for nonresponsive feeding practices were at a moderate to low level, similar to findings reported in other Chinese studies [16,25]. For example, a study in Hong Kong revealed that the scores for the use of food as a reward and emotional feeding were 2.68 (SD 0.66) and 2.31 (SD 0.70) points, respectively [26]. However, the score for pressure to eat among caregivers of children aged 3‐18 months in Spain was 3.46 (SD 0.82) points [27], which is higher than that in our study. The scores for responsive feeding practices in our study were moderately high and were consistent with the findings of other national and international studies. For instance, a study in the United States revealed that the score for responsiveness to cues among children aged 3‐18 months was 4.49 (SD 0.04) points [28]. The median scores for behaviors related to active communication and interaction among caregivers whose children were aged 18 months in Shanghai were between 4 and 5 points, indicating a relatively high level [18].

The video observation confirmed the results of the survey. In daily meals, caregivers adopted responsive feeding practices more frequently than nonresponsive feeding practices. Caregivers likely prepared a variety of nutritious foods for their children. Responsive feeding practices, such as modeling, are effective strategies for promoting children’s intake of these healthy foods [6]. Furthermore, video observation found that the number of responsive feeding observations was similar among breakfast, lunch, and dinner. Similarly, a video observation study among children aged 12‐36 months in the United States indicated that caregivers’ feeding practices during each main meal had a certain degree of stability. Observation of a single main meal might be an alternative approach to examine responsive feeding in future studies, if resources are limited.

Previous studies on feeding practices were mainly conducted among mothers. Our study adds to the literature by examining responsive and nonresponsive feeding practices among both parents and nonparents. In this study, the scores of overall responsive feeding and the practice of responsiveness to cues among nonparents (99% grandparents) were higher than those among parents. However, the score of the use of food as a reward among nonparents was lower than that among parents. This might be because grandparents were patient and responsible when feeding their grandchildren and had sufficient time to adopt responsive feeding strategies [29,30]. From our video observation, grandparents waited for a while after each feed and did not provide the next feed until the child swallowed the food (data not shown). Moreover, grandparents had previous experience in feeding children. Finally, with the broad use of the internet and smartphones, grandparents were knowledgeable in using appropriate feeding strategies. Our results imply that education on optimal feeding practices should be given to parents. For instance, local maternal and child health centers might deliver responsive feeding guidance (eg, recognizing children’s satiety cues and optimizing mealtime interactions) to parents during children’s routine health check-ups and vaccination visits. On the contrary, some studies failed to report differences in pressure to eat and the use of food as a reward between grandparents and parents, and some studies found that grandparents were more inclined to adopt the pressure to eat practice than parents [29,31]. Children in these studies were older than our study participants [31], which might explain the variation in results between our study and previous studies.

In this study, compared with employed caregivers, unemployed caregivers scored lower in overall responsive feeding and responsiveness to cues but scored higher in pressure to eat. Moreover, the score of overall responsive feeding was lower among caregivers whose monthly household income was ≤5000 RMB (≤US $721). Caregivers with a lower socioeconomic status likely paid less attention to responsive feeding practices. Likewise, a qualitative study in Thailand among factory worker parents of preschool children demonstrated that family income was one of the most influential factors on children’s food choice, as parents prioritized their budget on food over nutritional quality due to economic constraints [32]. Therefore, interventions to enhance responsive feeding in China should attempt to target low-income groups. For example, community health centers might consider offering feeding workshops aligned with intergenerational care norms free of charge for low-income groups. Free personalized consultation could also be provided by community health care workers.

In this study, caregivers of children aged 24‐35 months had a lower score for controlling and a higher score for modeling, in comparison with caregivers of children aged 6‐23 months. It is likely that older children could express their dietary signals well, making it easier for caregivers to respond more appropriately. With an increase in their independence, children are more likely to eat on their own [33]. Caregivers’ use of the modeling strategy could help the development of children’s healthy eating habits. Therefore, specific education on feeding practices should be delivered to caregivers by considering the age of their children. Caregivers of younger children (less than 24 months of age) urgently need responsive feeding education.

Caregivers whose children were not the first child in the family were more likely to use emotional feeding but less likely to use active communication and interaction, and controlling strategies, in comparison with caregivers whose children were the first child. Based on the theory of family systems, the dilution of resources (time, energy, etc) for raising multiple children may affect caregivers’ feeding practices [34]. With an increase in the number of children in the family, caregivers’ time and energy allocated to each child decrease. Therefore, the frequency of caregiver-child interaction and verbal communication during feeding decreases [34]. Caregivers’ practice of controlling children’s eating also decreases. Caregivers are more likely to become impatient and use emotional feeding. However, a cross-sectional study in Thailand reported that the first child was more likely to receive inappropriate complementary feeding [35]. The Thai study focused on children’s dietary intake and diversity, while our study focused on the behaviors of the caregiver-child interaction. Differences in research focus might account for the variations in findings between the studies.

Compared with caregivers having a normal weight status, underweight caregivers in this study adopted more modeling behaviors. The reason may be that underweight caregivers are more concerned about insufficient dietary intake or being underweight in childhood, and thus, they play the role of an eating model for their children and ensure that their children have a good nutritional status.

Notably, there was no difference in responsive/nonresponsive feeding practices when considering the child’s gender. It appears that Chinese traditional patriarchal attitudes did not influence caregivers’ feeding practices among our study participants. Additionally, caregivers’ education, children’s weight status, and the family structure were not associated with caregivers’ responsive/nonresponsive feeding practices in this study.

This study adopted a combined approach involving survey and video observation methods to investigate responsive/nonresponsive feeding practices among caregivers of children aged 6‐35 months. This approach provided an objective and comprehensive assessment of responsive/nonresponsive feeding practices. Findings from the survey and observation assessments were confirmatory and complementary to each other. The questionnaire had good reliability and validity and could comprehensively measure all dimensions of responsive and nonresponsive feeding practices. Video observation could objectively capture the actual feeding environment and caregiver-child interaction and assess the practice of “creating a good meal environment,” which could not be measured by questionnaires. Moreover, this study used multivariate analysis to explore the factors influencing responsive and nonresponsive feeding practices. The findings of this study would be useful in the development of interventions and strategies to promote responsive feeding and in the identification of targeted populations for interventions.

The limitations of this study should be acknowledged. First, this study was conducted in some areas of Hebei Province. The results from this study might not be representative of caregivers living in other areas. Moreover, the caregivers in our survey sample had a higher education level but a lower employment rate than the general population [36]. National data indicate that the population in Hebei Province is at the top level and its economic output (reflected by gross domestic product) is at the middle level in China [36]. The health literacy of residents in Hebei Province [37] was similar to that at the national level (27.78%), according to 2022 surveillance data [38], implying that our findings might represent the Chinese population to some extent. Second, the use of food as a reward is more likely to occur in snacking occasions than during main meals. However, our video observation only recorded feeding episodes during main meals. Thus, the occurrence of the use of food as a reward reported in our video observation might be lower than the actual status. Third, owing to the COVID-19 epidemic, video observation could not be carried out at the planned location (Shijiazhuang city, where the survey was conducted). After careful discussion by the research team, Tangshan city was selected as an alternative location. The 2 cities are located in the same province, with a distance of 420 kilometers, ensuring the homogeneity of the samples to some extent. Fourth, the Hawthorne effect may have existed in the video recording process owing to the presence of the camera or videographer. Nevertheless, a number of efforts were taken to minimize the Hawthorne effect. For instance, we conducted the recording in a familiar environment (ie, home); we explicitly instructed caregivers to behave as usual prior to each recording; and the videographer dressed normally, remained silent, stayed hidden in a corner, and avoided participating in any feeding practices during the recording.

With a combined approach involving survey and video observation methods, this study revealed that the frequency of responsive feeding practices was higher than that of nonresponsive feeding practices among Chinese caregivers whose children were aged 6‐35 months. The combined approach might serve as a methodological reference for research on feeding practices. Moreover, the survey demonstrated that parental, unemployed, and low-income caregivers were less likely to use responsive feeding strategies. Health care providers should consider this finding and target such groups for feeding education and intervention.