We conducted this review by following the reporting guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [38]. These guidelines include a four-phase flow diagram to guide the inclusion and exclusion of research papers [38]. In addition, the guidelines provide a 27-item checklist outlining standards per review section (eg, title, abstract, introduction, methods, results, discussions, funding) to ensure that reviews are systematically conducted and reported [38].

We searched six electronic databases: Science Direct, PubMed, PsycINFO, PscyARTICLES, Medline, and Cumulative Index to Nursing and Allied Health Literature (CINAHL) between April and June 21, 2016. The search was limited to full-text, peer-reviewed articles published in English between January 2000 and June 2016. Key search terms included: (1) child* OR pediatric*; (2) maternal depress* OR maternal depress* symptoms OR maternal “depressive symptoms”; and (3) “feeding practices” OR “feeding behavior” OR feeding srateg* OR feeding style (see Figure 1). Two authors (ACL, TM) independently examined the titles and abstracts of all citations, and the citations were excluded when both authors determined that the study did not meet the inclusion criteria. The same two authors independently reviewed the retrieved articles and identified studies to be included in this systematic review. They also searched the reference lists [39-46] and other review studies focusing on infants and/or children older than 8 years of age [7,10-16,25,31,47-49].

This systematic review was limited to peer-reviewed studies that included mothers 18+ years old of normally developing children (ie, not born preterm, not diagnosed with physical or mental complications) between 2 and 8 years of ages (ie, early childhood). We identified studies that (1) examined the association between maternal depression and/or maternal depressive symptoms (independent variable) and parental feeding styles, parental feeding practices, and/or family mealtime practices (outcome variables), and (2) measured maternal depression or depressive symptoms with a validated questionnaire or scale at any period prior to or following childbirth. Studies that focused on special groups (eg, teen mothers, children born pre-term or low birth weight, or special needs such as cerebral palsy) or populations with health concerns (eg, mothers diagnosed with HIV) were excluded. Studies that used qualitative methods exclusively were also excluded to simplify comparison of findings across studies. Additionally, studies that focused exclusively on breastfeeding and/or complementary feeding practices were excluded as previous review papers have examined the association between maternal depressive symptoms and infant feeding practices [15,25,31,47,49].

Using the search strategy illustrated in Figure 1 (PRISMA flow diagram), we identified eight observational studies meeting eligibility requirements [39-46]. Two authors (ACL, TM) independently read and completed an article extraction form for all articles. The data extraction form gathered the following information: (1) authors, (2) study setting, (3) sample size, (4) participant characteristics, (5) study design, (6) study aim(s), (7) measure (s) of maternal depressive symptoms, (8) measure(s) of parental feeding styles and practices and family mealtime practices, and (9) study results. The two authors who completed the data extraction forms compared their results and discussed discrepancies, which were resolved with feedback from a third author.

This review extracted data on associations between maternal depressive symptoms (exposure) and parental feeding styles, parental feeding practices, and/or family mealtime practices (outcomes) and summarized the findings. Due to the range of study designs, assessment of exposure, and outcomes, conducting a meta-analysis of the data was not appropriate, and results of this review are presented as a narrative summary.

The first author (ACL) assessed the quality of reviewed studies using a modified version of the Strengthening in the Reporting of Observational Studies in Epidemiology (STROBE) statement [50], which were confirmed by two authors (MLG, JAW). STROBE is an international, collaborative initiative of epidemiologists, methodologists, statisticians, researchers and journal editors who have a common goal of strengthening reporting of observational studies in epidemiology. The combined STROBE checklist for cohort, case-control, and cross-sectional studies includes 22 items [50]. For this review, we included 11 of these items to (1) identify potential sources of bias, and (2) identify possible methodological areas that were insufficiently addressed (see Multimedia Appendix 1). Each question was designed to be answered either “yes” or “no,” with a score of 1 assigned to “yes” response, and a score of 0 to “no” response (range of scores 0-11). Total scores were then used to assign a rating of the study as strong (score> 8), moderate (score between 8-6), or weak (score between 5-0).

We identified 297 unique citations and two authors (ACL, TM) independently examined the titles and abstracts of the identified citations. We excluded 272 citations that did not meet the eligibility criteria, and 35 full-text articles were selected for detailed review and assessed by two authors independently. Eight studies that met the eligibility criteria were included in this systematic review (see Figure 1). Studies included in this review fell into three categories, and this review is organized by the study’s purpose. The three purpose categories were to examine the associations between: (1) maternal depressive symptoms and nonresponsive parental feeding styles, (2) maternal depressive symptoms and nonresponsive parental feeding practices, and (3) maternal depressive symptoms and nonresponsive family mealtime practices.

Eligible studies examined maternal depressive symptoms [51-56] and parental feeding styles [57] and feeding practices [58-64] using a wide variety of tools. Of the 8 included studies, 3 focused on feeding styles [39,41,46], 6 on nonresponsive feeding practices [40,41,43-46] and 3 on nonresponsive family mealtime practices [41,42,44]. Summary characteristics of included studies are presented in Multimedia Appendix 2, whereas synthesized information on methodology and main findings are presented in Multimedia Appendix 3.

The quality of the reviewed studies varied (See Multimedia Appendix 1). Out of the 8 included studies, 6 were assigned a rating of > 8 (strong), and 2 were rated between 8-6 (moderate). Most of the reviewed studies presented methodological limitations: (1) nearly all (6/8) used cross-sectional study designs that prohibit determining causal inferences between maternal depressive symptoms and child-feeding practices and styles [39,41,42,44-46], (2) all studies (8/8) relied on self-reporting of maternal depressive symptoms [39-46], and (3) the majority (6/8) relied on self-reporting of parental feeding styles, parental feeding practices and family mealtime practices [39-42,44,46]. Furthermore, it was not always possible to determine whether analyses controlled for potential confounding factors and analyses were not stratified by child gender.

Moreover, maternal depressive symptoms [51-56] and parental feeding practices [58-64] were assessed by an array of instruments, which made it difficult to compare results across studies. For example, across the 8 reviewed studies, 6 different instruments were used to measure maternal depressive symptoms [51-56], whereas 7 instruments were used to assess parental feeding practices [58-64], 2 instruments were used to assess family mealtime practices [59,61], and 1 to assess parental feeding styles [57].

Out of the 8 included studies [39-46], 3 studies [39,41,46] examined the association between maternal depressive symptoms and nonresponsive feeding styles. Two [39,41] of these studies used the validated Center for Epidemiological Study Depression (CES-D) instrument [51], whereas 1 used the Depression Anxiety Stress Scale 21-item (DASS-21) [53] to assess maternal depressive symptoms. All three studies were cross-sectional [39,41,46] and used the validated Caregiver’s Feeding Styles Questionnaire (CFSQ) [57] to assess parental feeding styles. The CFSQ is a 19-item valid, reliable measure developed for use with caregiver’s of preschool-age children [57].

Of the 3 studies examining associations between maternal depressive symptoms and nonresponsive parental feeding styles in adjusted analysis, only 2 studies found positive associations [39,41], with mothers with elevated depressive symptoms being more likely to exhibit uninvolved [39,41] and permissive styles [41] than those without these symptoms.

A cross-sectional study [41] conducted in the United States with a sample of low-income mothers of whom approximately 30% were Hispanics, found that mothers reporting elevated depressive symptoms reported using more demandingness (eg, encouragement or discouragement of child’s eating behaviors) and permissive feeding styles (eg, fewer authority narratives about feeding) than mothers without elevated depressive symptoms after adjusting for potential confounders. One cross-sectional study [39] conducted in the United States with a sample of low-income mothers, of whom approximately 55% were Hispanic and 45% were African-American, found that after adjusting for potential confounders, mothers employing an uninvolved feeding style (a permissive feeding style) reported less positive affect and more parenting stress than mothers who used authoritative, authoritarian, or indulgent/permissive feeding styles. In addition, mothers with elevated depressive symptoms were more likely to present low authority in child feeding [39].

One cross-sectional study [46] conducted in Australia found that mothers who reported experiencing higher levels of depressive symptoms also reported using higher levels of the authoritarian feeding style. In adjusted analysis, however, none of the maternal psychosocial well-being variables independently contributed to the prediction of authoritarian parental feeding style.

Out of the 8 reviewed studies, 6 [40,41,43-46] examined the association between maternal depressive symptoms and nonresponsive feeding practices. Four studies [41,44-46] employed cross-sectional designs and two used longitudinal designs [40,43]. The 6 studies used five different instruments [51,53-56] to assess maternal depressive symptoms, with the Edinburgh Postnatal Depression Scale (EPDS) [53] being used in 2 [40,43] and the Depression Anxiety Stress Scales 21-item (DASS-21) [54] also being used in 2 studies [43,46]. The validated Child Feeding Questionnaire (CFQ) [58] was the instrument most used to assess parental feeding practices (5/6). In addition, five other validated instruments were used across the 6 studies [59-64]. All 6 reviewed studies [40,41,43-46] provided information on the reliability and validity of the instruments used to assess both maternal depressive symptoms and parental feeding practices (see Multimedia Appendix 3).

In summary, 3 studies reported positive associations between maternal depressive symptoms and use of instrumental feeding (eg, using food as a reward) (3/3) [40,44,45], and pressure to eat (3/6) [40,41,45]. One study reported positive associations between maternal depressive symptoms and restriction of child’s food intake (1/6) [40], and emotional (eg, using food to manage child’s mood) feeding (1/3) [40]. In addition, 2 studies (2/6), one with a cross-sectional design [44] and one using a longitudinal design [43], reported negative associations between maternal depressive symptoms and restriction of child food intake. In contrast, a cross-sectional study (1/6) [45] found that elevated depressive symptoms were not associated with restriction of child food intake. Moreover, 1 study (1/4) reported that elevated depressive symptoms were negatively associated with monitoring of child food intake [43].

Out of the 8 reviewed studies [39-46] 3 [41,42,44], all of which employed cross-sectional designs and were conducted in the United States with low-income mothers, examined the association between maternal depressive symptoms and a number of nonresponsive family mealtime practices. Two [41,42] of these studies used the Center for Epidemiological Study Depression (CES-D) instrument [51], whereas one [44] used the Patient Health Questionnaire-9 (PHQ-9) [55] to assess maternal depressive symptoms. Two different instruments [59,61] were used to assess family mealtime practices by 2 [42,44] of the 3 studies, and 1 study [41] used both semistructured narrative interview and videotaped observations of mother-child feeding situations.

All 3 studies determined that children in households with mothers with elevated depressive symptoms were more likely to be exposed to less optimal mealtime practices and routines than children in households with mothers with low or without any depressive symptoms [41,42,44]. Moreover, all 3 studies found that mothers reporting elevated depressive symptoms were more likely to report nonresponsive feeding practices which were associated with both uninvolved (eg, mother not being present during meals, child skipping breakfast, child eating while watching television) and permissive (eg, lower levels of maternal control over child eating routines, greater child choice over snacking) feeding styles.

One study [41] found that in households of mothers with elevated depressive symptoms, children were less likely to eat at the kitchen or dining table, the television was more likely to be audible during meals, and children were less likely to eat with their mothers. Similarly, 1 study [42] found that maternal depression was significantly associated with lower maternal presence when the child ate, lower levels of maternal control over child eating routines, greater child choice over snacking, and fewer optimal mealtime practices than in homes of mothers without higher depression scores [42]. Likewise, 1 study [44] found in adjusted analyses that mothers reporting mild depressive symptoms were more likely to have children who consumed sweetened drinks daily, who did not eat breakfast regularly, and who ate out in restaurants 3 or more times per week than mothers without depressive symptoms.

The aim of this systematic review was to identify and review existing research examining associations between maternal depressive symptoms and nonresponsive parental feeding styles and parental feeding practices in mothers of young children. The 8 reviewed studies provide mixed support for associations between maternal depressive symptoms and nonresponsive feeding styles, feeding practices, and family mealtime practices. Uninvolved and permissive feeding styles, and feeding practices use of instrumental feeding (eg, use of food as reward) and pressuring children to eat were the most consistently associated with depressive symptoms among studies included in this review. In addition, maternal depressive symptoms were associated with uninvolved and permissive family mealtime practices.

Across the reviewed studies, elevated maternal depressive symptoms were most often associated with uninvolved and permissive parental feeding styles [39,41]. Two of the three studies examining associations between maternal depressive symptoms and nonresponsive feeding styles (uninvolved and permissive) found a positive association [39,41]. These findings concur with results of studies [11,12] and review papers [10,15,25,47,49,65] conducted among mothers with infants. Furthermore, available evidence from the extant literature on maternal mental health and parenting suggests that maternal mental health issues may impair mothers’ responsiveness to, and interactions with, their children. The reduced interaction may manifest in nonresponsive, more controlling, and less-sensitive parenting [10,29,37,47,49,60]. In addition, research suggest that elevated depressive symptoms may contribute to decreased maternal–child interactions [30,66], with mothers being less responsive to their children and choosing strategies for coping that require less cognitive effort [11-13].

Evidence for associations between maternal depressive symptoms and nonresponsive maternal feeding practices was mixed across reviewed studies [40,41,43-46]. Instrumental feeding (eg, use of food as reward; 3/3) [40,44,45] and pressure to eat (3/6) [40,41,45] were the most consistently nonresponsive feeding practices associated with elevated depressive symptoms across the studies included in this review. Evidence from studies [11,12,16] and systematic reviews [25,47] with mothers of infants and toddlers suggests that mothers experiencing elevated depressive symptoms are more likely to use restrictive and controlling feeding practices than mothers without elevated depressive symptoms. Moreover, evidence suggest that mothers experiencing elevated depressive symptoms are less likely to be responsive to their children’s cues of hunger and satiety and less likely to respect their child’s ability to self-regulate food intake [10-13,66]. Previous studies indicate that nonresponsive feeding practices interfere with a child’s natural ability to self-regulate food intake based on hunger and satiety cues [14,19,30,67-69]. Furthermore, research suggests that both parental pressure to eat and feeding restrictions are associated with unrestrained eating and disinhibited eating in later life, excessive weight gain, and increased risk of child obesity in children [5,20,21,67-71].

Studies included in this review provide consistent evidence for the association between maternal depressive symptoms and nonresponsive family mealtime practices [41,42,44]. Children in households with mothers having elevated depressive symptoms were more likely to be exposed to less optimal mealtime practices and routines than children in households with mothers having low or no depressive symptoms [41,42,44]. Elevated depressive symptoms such as low energy and diminished pleasure in activities may contribute to decreased maternal involvement with the child [5,21,67-69,30], resulting in mothers being less responsive to their children and choosing strategies for coping that require less cognitive effort [5]. Suboptimal family mealtime practices have been reported to be associated with children’s unhealthy eating habits [72-77], which in turn have been linked to risk of overweight and obesity [72-77].

Our evaluation of the methodologies of studies included in this systematic review suggests some limitations, and therefore caution in the interpretation of study findings. The majority (6/8) of studies used cross-sectional study designs precluding a causal assessment of associations between maternal depressive symptoms and feeding styles and practices [39,41,42,44-46]. Additional longitudinal studies are necessary to understand whether mothers’ depressive symptoms influence their feeding styles and practices. Furthermore, nearly all examined (6/8) studies used an array of self-reported questionnaires for assessments of maternal depressive symptoms and parental feeding practices (7/8), which is potentially problematic due to possible misclassification of depressive symptoms (exposure) and parental feeding practices (outcome). Finally, variability in the assessment of maternal depressive symptoms (eg, CES-D, DASS, BSI) and parental feeding practices (eg, Child Feeding Styles Questionnaire [CFSQ], CFQ, Family Mealtime Coding System [FMCS]) make it difficult to compare findings across studies and indicate that caution must be taken in drawing conclusive assertions.

Strengths of this review include the use of systematic criteria (ie, PRISMA) to identify and select studies and a quality assessment tool for the critical appraisals of studies. Nonetheless, this review may be incomplete given limitation to studies published in English. Another possible limitation of this review is the variability in the studies’ location. Multiple countries (United States, Australia, and England) were represented, which may limit cross-study comparisons. Finally, publication bias should also be considered, as should the fact that this review is limited to full-text studies published in English and may have excluded studies published in other formats and/or languages.

Additional research is needed to further examine the relationships between maternal depressive symptoms and nonresponsive parental feeding styles and practices. Specifically, longitudinal studies and additional studies including low-income and racial/ethnic minority populations at increased risk of depressive symptoms are needed. Future studies should explore the associations between maternal depressive symptoms, food insecurity and maternal feeding styles and practices. This is required especially due to documented evidence of greater prevalence of obesity among racial/ethnic minority populations [78]. In 2 examined studies [39,42], authors suggest that food insecurity may interact with maternal depressive symptoms such as stress to increase the risk of unintentionally detrimental feeding practices such as pressuring child to eat and/or restricting child food intake. Therefore, studies that assess the potential interactions of food security status and maternal depressive symptoms on maternal feeding styles and practices are needed. Moreover, given the inconsistencies in results across studies included in this review, future research should also consider the potential influence of additional factors such as mother’s socioeconomic status, acculturation level, social support, as well as contextual factors such as work strain, access to healthful foods, and so on. Finally, future research may benefit from examining differentials of depressive symptoms and parental feeding styles and practices according to the gender of the parent and the child.

In summary, studies identified and synthesized in this review provided mixed evidence for associations between maternal depressive symptoms and nonresponsive maternal feeding styles and practices. Nevertheless, given the high prevalence of maternal depressive symptoms among women of reproductive age [32-35], the indication from some studies of associations between maternal depressive symptoms and nonresponsive feeding styles and practices, and the fact that both maternal depressive symptoms and that nonresponsive feeding styles and practices are potentially modifiable, further understanding of these associations are likely to provide important insights for the development of interventions to prevent and control childhood obesity.