Introduction

Stunkard et al. [1] first proposed the definition of “Night Eating Syndrome (NES)” as a condition characterized by morning anorexia, evening hyperphagia, and insomnia. More recent studies further expanded the syndrome’s diagnostic criteria by including nocturnal snacking or nighttime awakenings with food intake [2]. Furthermore, NES is now listed in the “Otherwise Specified Feeding and Eating Disorders” section in the fifth version of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [3]. NES is also considered a sleep disorder characterized by a circadian delay of food intake with insomnia and poor sleep quality [4]. Night eating has also been found to be associated with binge eating/loss of control over eating [5], as well as potentially presenting as a risk factor for weight gain [6] and a barrier to weight loss [7], especially in obese populations. Cross-sectional studies showed that the incidence of NES was remarkably higher in obese populations than in normal-weight individuals. For example, in a sample of obese persons seeking medical or surgical weight loss, NES prevalence was as high as 64% [8]. Moreover, symptoms of psychological distress (e.g., depression and stress) are consistently associated with NES [7, 9, 10]. Thus, NES has various negative health implications beyond its potential association with obesity.

Eating practices during adolescence are influenced by a multitude of factors [11, 12], but the family environment may be one of the most fundamental ones [13, 14]. Indeed, children’s eating behaviors are closely related to their interactions around food, eating, and other related factors [15,16,17]. In general, positive parent–child interactions around food or eating are protective for healthy eating and against unhealthy eating. For example, parental monitoring of children’s dietary intake has been associated with children’s healthy eating practices [18]. However, negative child–parent interactions, such as excessive food control and food reward, contribute to the development of unhealthy eating [17]. These child–parent interactions around eating and food, or so-called food-related parenting practices, are commonly guided by general parenting styles [19, 20]. For example, overprotective parents have been found more likely to encourage their children to be involved in meal planning and preparation and give pressure to let their children eat more food; however, authoritarian parents are more likely to attempt to control their children’s food intake to decrease or maintain their children’s weight [21].

Over the past three decades, numerous studies have explored the effects of specific parenting styles on children’s physical and mental development and children’s eating behaviors. For example, research has shown that positive parenting styles such as parental warmth and involvement, emotional support, appropriate granting of autonomy, and clear communication are positively associated with healthy eating in children and adolescents [15, 22]. However, dysfunctional parenting styles such as overprotective, controlling, and rejection have been identified as risk factors for developing disordered eating behaviors [23,24,25,26].

While there have been a number of studies that explored the link between parenting styles and children’s/adolescents’ eating behaviors (e.g., binge eating [27]), to the best of our knowledge, none of these studies examined the associations between parenting styles and night eating. Thus, more studies are needed to understand the relationship between parenting styles and children’s night eating. Moreover, previous studies on the relationships between parenting styles and eating behaviors primarily used variable-centered approaches [28], which focused on whether specific parenting styles (e.g., overprotection) could predict or be related to certain eating behaviors [15, 23, 27, 29, 30]. However, parenting styles are multidimensional, and each parenting style is an integrated combination of behaviors; thus, using variable-centered approaches that separate each parenting style and analyze the independent associations with eating behaviors fails to capture the influence from different combinations of parenting styles on eating behaviors.

Due to the limitations of previous studies, the present study aimed to bridge the research gaps using a person-centered approach (i.e., Latent Profile Analysis; LPA) to further the understandings of the link between parenting styles and night eating. Moreover, there have been well-defined/researched correlates of night eating, such as BMI [6], sleep quality [31], loss of control over eating/binge eating [32], and psychological distress [33]. Thus, to have a more comprehensive understanding of the link between parenting styles and night eating, we also examined how different patterns of parenting styles were related to these correlates.

As one previous study conducted in a sample of Chinese high school students found three parenting profiles: positive parenting, mixed parenting, and negative parenting [34], we hypothesized that three types of parenting styles would emerge. Moreover, given that previous findings generally supported that positive parenting styles are protective while negative parenting styles are detrimental, we hypothesized that adolescents in the positive parenting group would have the lowest levels of night eating, weight status, loss of control over eating/binge eating, and psychological distress, but the best sleep quality; on the contrary, those in the negative parenting group would have the highest levels of night eating, weight status, loss of control over eating/binge eating, and psychological distress, but the poorest sleep quality; and for those in the mixed parenting group, their scores of the night eating and correlates would be in the middle.

Methods

Participants and procedure

From a large middle school in Wuhan, Hubei (Central China), with the approval of the school administration, three classes of 7th grade, four classes of 8th grade, and two classes of 9th grade were randomly selected. The investigators described the project to the students in the selected classes and invited students to participate. If students agreed, consent forms were sent to their parents for further approval. When parental consent forms were obtained, students were asked to complete a set of questionnaires. Finally, a total of 455 adolescents, aged 12–15 years (Mean = 13.17, SD = 0.90), participated in the survey. According to Tein et al. [35], the sample size of 455 had the adequate statistical power to detect the correct number of classes in LPA. Of these participants, 44.6% were boys, 54.5% were girls, and 4 did not report their gender. The percentage expected weight for height (%EWH) was calculated based on participant’s self-reported height and weight (Mean = 85.32, SD = 14.10). Specific demographic information of the participants is shown in Table 1. In addition, it should be noted that a paper with a different topic has been published based on the same dataset [36].

Table 1 Descriptive statistics for the total sample (n = 455)
Full size table

Measures

Perceived parenting styles

The short form of Egna Minnen Beträffande Uppfostran: one’s Memories of Upbringing (s-EMBU; [37]) was used to measure adolescent’s perceived parenting styles in this study. This measurement was developed based on the early 81-items EMBU and scores appear to be consistent with the original measure. Forms for the mother and father each consist of three subscales: rejection, emotional warmth, and overprotection. Generally, emotional warmth represents an appropriate parenting style, while rejection and overprotection represent dysfunctional parenting styles [38]. The Chinese version of the s-EMBU for each parent form has 19 items [37] with a four-point Likert-type scale: 1: No, never; 2: Yes, occasionally; 3: Yes, often; 4:Yes, always. The Chinese version of the s-EMBU has been shown to have adequate psychometric characteristics such as reliability and construct validity [37]. In this study, father’s rejection, emotional warmth, and overprotection had a Cronbach’s α of 0.82, 0.86, 0.64, while mother’s rejection, emotional warmth, and overprotection had a Cronbach’s α of 0.81, 0.86, 0.70.

Night eating

The Night eating Questionnaire (NEQ) [39, 40] was used to assess the severity of the Night Eating Syndrome. The NEQ consists of 13 items evaluating four factors: morning anorexia, evening hyperphagia, mood/sleep, and nocturnal snacking. Participants responded on a 5-point Likert-type scale, and in this study, the total score was calculated by summing the scores of the four subscales. According to Allison et al. [40], a total score of 25 (high sensitivity) or 30 (high specificity) indicates the appearance of the night eating syndrome. Good psychometric properties were revealed for the Chinese version of the NEQ among Chinese samples [39, 41]. The simplified Chinese version of NEQ [39] was used; however, as the NEQ uses an item-skipping response format, the Cronbach’s \(\alpha\) of the NEQ was not calculated.

Loss of control over eating

Loss of control (LOC) over eating was assessed via the Loss of Control Overeating Scale-Brief (LOCES-B; [42, 43]. The LOCES-B comprises seven items rated on a 5-point Likert scale ranging from “never” to “always.” Previous studies suggested the LOCES-B comprises a single factor structure and had sound psychometric properties [42, 43]. In this study, the Chinese version of the LOCES-B [42] was used, and it had a Cronbach’s α of 0.89 in the current sample.

Sleep quality

Participant’s sleep quality was assessed with the self-report section of the Pittsburgh Sleep Quality Index (PSQI; [44]), a brief validated questionnaire consisting of 19 items to indicate sleep quality over the past month. The 19 items are combined to form seven component scores, including subjective sleep quality (item 6), sleep latency (items 2 and 5a), sleep duration (item 4), habitual sleep efficiency (items 1, 3, and 4), sleep disturbance (items 5b-j), sleep medication use (item 7), and daytime dysfunction (items 8 and 9). Each component score can range from 0 to 3. The seven component scores are summed to yield a global score ranging from 0 to 21, with higher scores indicating inferior sleep quality. The psychometric characteristics of the PSQI are well documented in Western and Chinese clinical and non-clinical samples [45,46,47,48].

Psychological distress

Kessler Psychological Distress Scale (K10) was used to assess the participants’ psychological distress. The K10 includes 10 self-report items for assessing psychological distress [49]. Each item is rated on a 5-point Likert scale ranging from “none of the time” to “all of the time.” Summing the 10 items results in a total score and a higher total score indicates a greater level of psychological distress. A number of studies showed good psychometric properties of the K10 for assessing psychological distress [49,50,51]. In this study, the Chinese version of the K10 was used [52]. The Cronbach’s \(\alpha\) of K10 in the current sample was 0.90.

Demographic information

Information regarding the age, sex, mother’s and father’s education background, height, and weight were collected by a self-report questionnaire. Options for parent’s education levels were as follows: primary school or below, junior high school, high school, college, and master’s degree or above.

Statistical analysis

Based on the procedures of conducting latent profile analysis (LPA; [53]), there were two main steps for the data analysis used in the current study. The first main step included exploring the latent profiles of parenting styles, with the six dimensions of s-EMBU being the indicators of LPA. In addition, for ease of interpretation, in line with He et al. [54], the indicators were standardized before being subjected to LPA. Moreover, to estimate model parameters, as recommended in He and Fan [53], we used Mplus version 8.3 [55] to conduct LPA with the robust maximum likelihood estimation (MLR) and a large number of starting values (500 random starting value sets, 200 best solutions). To identify the best fitting model, we tested models by starting a model with one class and adding one more class until the best was identified. The following fit indicators were used for evaluating the fitness of each model: Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Sample Size Adjustment BIC (SABIC), Bootstrapped Likelihood Ratio Test (BLRT), Lo Mendell–Rubin Adjusted Likelihood Ratio Test (LMRT) and Entropy. Among the fit indicators, the smaller the values of AIC, BIC, and SABIC, the better the degree of fitting of the model to the data, while a larger value of the entropy suggests the higher accuracy of the classification is. BLRT and LMRT are used to compare the differences between classes. Specifically, for k-class and k-1 class models, if the BLRT and the LMRT have significant p values, then k-class models are better fitting models than k-1 class models. In the three-step approach [56], after determining the optimal profile model, we used the BCH method to explore further whether there were any profile differences in participants’ night eating and related symptoms, including BMI, loss of control over eating, sleep quality, and psychological distress. In addition, for interpreting group differences, Cohen’s d was used, for which values of 0.2, 0.5, and 0.8 were considered small, medium, and large, respectively [57].

Results

Descriptive statistics

Table 1 describes the participant’s demographic information and the scores of variables used in the current study. In particular, based on the total scores of the NEQ, 19 participants had a night eating syndrome with a cutoff point of 25, while 4 participants had a night eating syndrome with a cutoff point of 30.

Latent profile analysis of perceived parenting styles

Table 2 shows the values of the fit indicators for each LPA model. The results showed that the values of AIC, BIC, and SABIC decreased with the increase of the latent profiles, suggesting that the 5-profile solution had the best fit. However, the p value of LMRT was larger than 0.05, indicating that the 5-profile solution was not significantly superior to the 4-profile solution. Therefore, considering that the 4-profile solution was parsimonious, could be interpreted easily, and had the highest entropy value, we chose the 4-profile solution as the best model.

Table 2 Fit indices and class proportions for 1- to 5-profile models
Full size table

Characteristics of the profiles

Figure 1 and Table 3 graphically and numerically show the characteristics of the four selected parenting style patterns. Specifically, profile 1 (42.20% of the participants) had the lower scores in all indicators, indicating that parents of the participants in this profile were less likely to engage in their children’s lives; thus, profile 1 was labeled as lowly engaged parenting. Profile 2 (31.43% of the participants) had higher scores in the positive parenting style (i.e., emotional warmth) and lower scores in the negative parenting styles (i.e., rejection and overprotection), so it was labeled as positive parenting. The characteristics of profile 3 (19.56% of the participants) were contrary to those of profile 1, with all scores being higher than average, indicating that the parents of the participants in this profile were very likely to engage in their children’s life; therefore, this profile was named as highly engaged parenting. The last profile, profile 4 (6.81% of the participants), was labeled as negative parenting because it was featured by lower scores in the positive parenting style and higher scores in the negative parenting styles.

Fig. 1
figure 1

Description of the selected LPA profiles of the lowly engaged parenting, positive parenting, highly engaged parenting and negative parenting

Full size image
Table 3 Descriptive statistics for the four latent profiles
Full size table

Profiles differences on night eating syndrome symptoms and correlates

With the three-step approach, we further examined whether or not the participants in each profile differed in night eating and its correlates. As shown in Table 4, participants in different profiles had statistically significant differences in all covariates collected in the current study.

Table 4 Profile differences on BMIz, night eating, LOC eating, sleep quality and psychological distress
Full size table

Specifically, for night eating, there were significant differences among the four potential profiles (\({\chi }^{2}\) = 33.11, p < 0.01), with participants in the negative parenting group showing the highest level of the night eating symptoms, followed by those in the low engaged parenting and highly engaged parenting group; while participants in the positive parenting had the lowest level of the night eating symptoms (e.g., Cohen’s d for the comparison of positive parenting vs. negative parenting was 1.12, a large effect size). As for %EWH, significant differences were revealed across the four parenting style patterns (\({\chi }^{2}\) = 8.83, p < 0.05), and participants in the positive parenting profile reported the highest %EWH, but it was not significantly different from the scores from participants in the negative parenting. For the correlate of LOC eating, there were also significant differences found (\({\chi }^{2}\) = 17.23, p < 0.01) with participants in the positive parenting group describing the lowest level of LOC eating (e.g., Cohen’s d for the comparison of positive parenting vs. negative parenting was 0.67, a medium effect size). Moreover, sleep quality also significantly differed across profiles (\({\chi }^{2}\) = 16.99, p < 0.01), and participants in the positive parenting showing the best sleep quality, followed by those in the low and highly engaged parenting groups, while participants in the negative parenting group had the poorest sleep quality (e.g., Cohen’s d for the comparison of positive parenting vs. negative parenting was 0.95, a large effect size). Finally, there were also significant differences among the four profiles regarding psychological distress (\({\chi }^{2}\) = 74.62, p < 0.01), with participants in the negative parenting group showed the highest level of psychological distress, followed by those in the low and highly engaged parenting profiles, while participants in the positive parenting had the lowest level of psychological distress (e.g., Cohen’s d for the comparison of positive parenting vs. negative parenting was 1.92, a large effect size).

Discussion

To the best of our knowledge, the current study was the first study to evaluate associations between parenting style patterns and night eating among adolescents. Specifically, using a person-centered approach (i.e., LPA), we explored how different patterns of perceived parenting styles were related to adolescent’s night eating syndrome symptoms and correlates. The results of LPA showed four distinct profiles of perceived parenting styles, namely low engaged parenting pattern (low appropriate and low dysfunctional), positive parenting pattern (high appropriate and low dysfunctional), highly engaged parenting pattern (high appropriate and high dysfunctional), and negative parenting pattern (low appropriate and high dysfunctional). Subsequent analyses indicated that adolescents in these profiles significantly differed in night eating syndrome symptoms and correlates (i.e., sleep quality, psychological distress, and %EWH). Specifically, while parenting style profiles including more appropriate parenting were less likely to be characterized by night eating syndrome symptoms and correlates, profiles including dysfunctional parenting styles were more likely to exhibit night eating and related symptomatology.

The four emergent parenting style profiles are generally in line with the findings from a previous study by Wu et al. [34] who revealed three patterns of parenting styles for a group of Chinese adolescents: positive parenting, mixed parenting, and negative parenting. The patterns of positive parenting and negative parenting in our study completely corresponded with the two patterns in Wu et al. [34]. However, in our study, we also found two types of mixed parenting (i.e., lowly engaged and highly engaged parenting). The generally consistent finding on the patterns of parenting styles may indicate that these sub-types of perceived parenting styles of Chinese adolescents are relatively stable, supporting the generalizability of findings based on these profiles.

We further investigated whether night eating syndrome symptoms and correlates differed across parenting style profiles. Results showed that adolescents in four parenting style patterns were significantly different in night eating syndrome symptoms and correlates. Specifically, adolescents in the profile of positive parenting (characterized by low rejection, low overprotection, but high emotional warmth) showed the least night eating syndrome symptoms, while adolescents in the profile of negative parenting (characterized by high rejection, high overprotection, but low emotional warmth) had the most night eating syndrome symptoms. Our finding is concordant with previous literature, which shows that dysfunctional parenting practices (e.g., rejection and overprotection) are associated with disordered eating symptoms (e.g., body dissatisfaction, drive for thinness, bulimia, emotional eating; [58,59,60]), while appropriate parenting practices (e.g., emotional warmth) are protective factors for disordered eating behaviors [61]. Thus, our finding suggests that parenting style patterns may be an important contributing factor in the development of night eating syndrome.

Furthermore, we also found that adolescents in the positive parenting group had the lowest LOC eating, sleep quality, and psychological distress scores. In contrast, those in the negative parenting group presented the highest scores in sleep quality and psychological distress. Those in the two mixed pattern profiles were in the middle. These results align with previous literature that parenting styles have comprehensive effects on adolescents’ physical and mental health [22, 62,63,64]. Furthermore, these findings suggest that different parenting style patterns may also have comprehensive effects on adolescent’s night eating. The pattern of positive parenting may decrease the likelihood of night eating and reduce the risks of night eating by lowering the correlates of night eating.

To our knowledge, this study firstly investigated the potential effects of different patterns of perceived parenting styles on adolescent’s night eating and related symptomatology. Based on the findings of the current study as well as the fact that parenting styles influence adolescents in both states (e.g., mood [65]) and traits (e.g., personality [66]), and within the context of transdiagnostic approaches to understanding eating disorders [67], we proposed that parenting style might influence night eating behaviors via promoting the risk factors of both general eating disorders (e.g., psychological distress) as well as the specific risk factors for night eating syndrome (e.g., sleep quality). However, this should be confirmed in future studies. Additionally, based on the current study’s findings, the examination of parenting practices may be an important area to examine when exploring risk factors for the development of night eating behaviors. For example, parental consultation and discussions about how parenting styles influence the relationship with their children may be included as a part of the treatment of night eating for adolescents. Similar to Gerards et al. [68], interventions targeting at night eating [69] may benefit from also including strategies for addressing dysfunctional parenting styles.

Several limitations should be taken into consideration. First, the sample in this study was limited to one high school, so the results may not be generalizable to adolescents in other developmental stages, regions, countries, and cultures. Future research may consider similar studies with large, diverse, population-based samples or samples from other cultures. Second, the participants of the current study were recruited via convenience sampling method, so the scores of the data (e.g., NEQ scores) might not be comparable to the norms. Third, the current study used a cross-sectional design to explore the relationships between the patterns of parenting style and night eating and its correlates; thus, the causal relationships between the patterns of parenting styles and night eating or correlates cannot be guaranteed. Future studies may consider using longitudinal designs to explore the casual relations of these variables. Finally, it should be noted that the parenting styles measured in the current study were “perceived parenting styles” by adolescents. Thus, future studies are encouraged to confirm the findings in the current study by measuring parenting styles directly from parents.

Conclusion

Using a person-centered approach (i.e., LPA), the present study identified four distinct patterns of parenting styles in a sample of Chinese adolescents, with night eating and related symptomatology differing across each profile. Future studies about exploring the risk/protective factors of night eating syndrome among adolescents should consider the potential influence of different patterns of perceived parenting styles.

What is already known on this subject?

Parenting styles are related to adolescents’ eating behaviors (e.g., binge eating). However, previous studies mainly used a variable-centered approach (e.g., linear regression analysis) to research the topic, and little research has been conducted on the relationship between parenting styles and night eating.

What does this study add?

Using a person-centered approach (i.e., Latent Profile Analysis), the present study examined the associations between parenting styles and adolescents’ night eating syndrom symptoms and correlates. Latent profile analysis identified four different patterns of parenting styles (i.e., positive parenting, negative parenting, highly engaged parenting, and lowly engaged parenting), and distinct patterns of parenting styles exhibited significant differences in night eating syndrome symptoms and correlates.