Obesity and overweight

Twenty-one studies investigated the relationship between HFI and child overweight/obesity and found mixed results^{(Reference Do, Bowen and Ksinan16,Reference Gundersen, Lohman and Garasky18,Reference Lohman, Stewart and Gundersen20,Reference Willis and Fitzpatrick22,Reference Bauer, MacLehose and Loth25,Reference Fleming, Kane and Meneveau34,Reference Fulay, Vercammen and Moran36,Reference Gundersen, Lohman and Eisenmann40,Reference Gundersen, Garasky and Lohman41,Reference Holben and Taylor44,Reference Hooper, Telke and Larson45,Reference Kirk, Kuhle and McIsaac51,Reference Kuku, Garasky and Gundersen53,Reference Martin and Ferris59,Reference Niemeier and Fitzpatrick66,Reference Parker, Widome and Nettleton68,Reference Poulsen, Bailey-Davis and Pollak69,Reference Robson, Lozano and Papas71,Reference Wirth, Palakshappa and Brown80–Reference Zhu, Mangini and Hayward82)} . Eight studies found no direct association between HFI and child overweight/obesity using various markers of weight status, including BMI, waist circumference, trunk fat mass, triceps skin folds, body fat percentage, metabolic syndrome, BMI z-scores and weight status^{(Reference Do, Bowen and Ksinan16,Reference Fulay, Vercammen and Moran36,Reference Gundersen, Lohman and Eisenmann40,Reference Gundersen, Garasky and Lohman41,Reference Martin and Ferris59,Reference Niemeier and Fitzpatrick66,Reference Parker, Widome and Nettleton68,Reference Robson, Lozano and Papas71)} . Eleven studies found a statistically significant association between HFI and greater odds of overweight/obesity OR range: 1·44–1·81 (95 % CI 1·13, 2·48)^{(Reference Bauer, MacLehose and Loth25,Reference Fleming, Kane and Meneveau34,Reference Holben and Taylor44,Reference Hooper, Telke and Larson45,Reference Kaur, Lamb and Ogden48,Reference Kirk, Kuhle and McIsaac51,Reference Kuku, Garasky and Gundersen53,Reference Poulsen, Bailey-Davis and Pollak69,Reference Wirth, Palakshappa and Brown80–Reference Zhu, Mangini and Hayward82)} . One study only found an association between greater BMI and HFI in adolescent boys^{(Reference Bauer, MacLehose and Loth25)}, while one study found an association in older children and not younger children (< 6 years)^{(Reference Kaur, Lamb and Ogden48)}. Two studies in this review did not report a statistically significant association between HFI and overweight. However, they did find a greater prevalence of obesity in food insecure populations^{(Reference Fleming, Kane and Meneveau34,Reference Hooper, Telke and Larson45)} . Only one study found an association between HFI and reduced odds of obesity^{(Reference Zhu, Mangini and Hayward82)}.

Studies found that the relationship between HFI and child overweight/obesity was determined by child age, child ethnicity, parental education, household income, maternal BMI and severity and timing of HFI^{(Reference Bauer, MacLehose and Loth25,Reference Holben and Taylor44,Reference Kirk, Kuhle and McIsaac51,Reference Kuku, Garasky and Gundersen53,Reference Niemeier and Fitzpatrick66,Reference Parker, Widome and Nettleton68,Reference Poulsen, Bailey-Davis and Pollak69,Reference Wirth, Palakshappa and Brown80–Reference Zhu, Mangini and Hayward82)} . For example, children who were food insecure, older, non-White, whose parents had lower educational attainment level and whose parents were obese were more likely to be overweight. One study found that food insecure girls were more likely to be overweight, compared with food-secure girls, but the same comparison was not drawn for boys^{(Reference Martin and Ferris59)}. Furthermore, persistent HFI and timing of HFI were also associated with both weight extremes. Children who experienced persistent HFI throughout their life from early childhood to early adolescence had higher odds of being overweight, while children who experienced a single exposure to HFI in early childhood were associated with underweight in early adolescence^{(Reference Kuku, Garasky and Gundersen53,Reference Wirth, Palakshappa and Brown80,Reference Zhu, Mangini and Hayward82)} .

Four studies explored the mechanism by which HFI may be associated with weight status^{(Reference Do, Bowen and Ksinan16,Reference Gundersen, Lohman and Garasky18,Reference Lohman, Stewart and Gundersen20,Reference Willis and Fitzpatrick22)} . Two studies exploring mechanisms found that an interaction between HFI and maternal stressors had an impact on weight status, for children aged 3–10 years old and adolescents aged 10–15 years old; however, results were mixed^{(Reference Gundersen, Lohman and Garasky18,Reference Lohman, Stewart and Gundersen20)} . Both studies did not find a direct association between HFI and child overweight. One study found that an interaction between HFI and maternal stressors amplified the probability of food-secure children being obese or overweight, compared with children living in food insecure households, whose mothers experienced similar stressor levels (P < 0·05)^{(Reference Gundersen, Lohman and Garasky18)}. In contrast, the second study found that maternal stressors enhanced the likelihood of living with overweight or obesity when an adolescent was exposed to HFI (P < 0·001)^{(Reference Lohman, Stewart and Gundersen20)}.

One study found that while HFI was directly associated with adolescent overweight (OR = 1·56; 95 % CI 1·13, 2·15; P < 0·05) and obesity (OR = 1·96; 95 % CI 1·40, 2·74; P < 0·05) an indirect relationship was not mediated by sleep duration (P = 0·23)^{(Reference Do, Bowen and Ksinan16)}. Additionally, another study found a direct and indirect relationship between HFI and child overweight, mediated by child psychosocial factors including depression and social status (P < 0·05)^{(Reference Willis and Fitzpatrick22)}.

Diet quality

Nineteen studies reported and found an association between HFI and diet quality in both children and adolescents, using various diet quality indicators^{(Reference Bruening, Lucio and Brennhofer26,Reference Canter, Roberts and Davis27,Reference Duke30–Reference Eicher-Miller, Mason and Weaver32,Reference Fram, Ritchie and Rosen35,Reference Gasser, Mensah and Kerr38,Reference King49–Reference Kral, Chittams and Moore52,Reference Landry, van den Berg and Asigbee54,Reference Molcho, Gabhainn and Kelly65,Reference Poulsen, Bailey-Davis and Pollak69,Reference Robson, Lozano and Papas71,Reference Soldavini and Ammerman72,Reference Tan, Laraia and Madsen74,Reference Widome, Neumark-Sztainer and Hannan79,Reference Yang, Sahota and Pickett81)} . Six studies found that food insecure children and adolescents were less likely to consume vegetables compared with their food-secure peers^{(Reference Canter, Roberts and Davis27,Reference Fram, Ritchie and Rosen35,Reference Kirkpatrick and Tarasuk50,Reference Landry, van den Berg and Asigbee54,Reference Molcho, Gabhainn and Kelly65,Reference Soldavini and Ammerman72)} . Three studies found that HFI was associated with reduced consumption of fruit^{(Reference Kirk, Kuhle and McIsaac51,Reference Molcho, Gabhainn and Kelly65,Reference Soldavini and Ammerman72)} . Four studies in older children (> 10 years) found no association between HFI and fruit and/or vegetable consumption^{(Reference Fram, Ritchie and Rosen35,Reference Kirkpatrick and Tarasuk50,Reference Poulsen, Bailey-Davis and Pollak69,Reference Robson, Lozano and Papas71)} . HFI was also associated with reduced healthy and unhealthy food availability in the household^{(Reference Widome, Neumark-Sztainer and Hannan79)}.

Five studies found that food insecure children and adolescents were less likely to consume dairy products than food-secure children/adolescents^{(Reference Dykstra, Davey and Fisher31,Reference Eicher-Miller, Mason and Weaver32,Reference Kirkpatrick and Tarasuk50,Reference Soldavini and Ammerman72)} . Studies also found that children and adolescents were likely to consume greater savoury and sweet snacks, sugar-sweetened beverages and fast foods than those who were food-secure^{(Reference Duke30,Reference Dykstra, Davey and Fisher31,Reference King49,Reference Kral, Chittams and Moore52,Reference Landry, van den Berg and Asigbee54,Reference Molcho, Gabhainn and Kelly65,Reference Soldavini and Ammerman72,Reference Yang, Sahota and Pickett81)} . Moreover, food insecure children and adolescents were more likely to have higher daily energy intake, with lower energy consumption from protein and wholegrains^{(Reference Fram, Ritchie and Rosen35,Reference Kirkpatrick and Tarasuk50,Reference Kirk, Kuhle and McIsaac51,Reference Landry, van den Berg and Asigbee54,Reference Tan, Laraia and Madsen74,Reference Widome, Neumark-Sztainer and Hannan79)} . Additionally, two studies found that HFI was associated with a reduced intake of vitamins A, B₆, B₁₂, thiamine, Fe, riboflavin, Mg, phosphorus and Zn^{(Reference Eicher-Miller, Mason and Weaver32,Reference Kirkpatrick and Tarasuk50)} . The relationship between HFI and diet quality differed by gender, ethnicity and child age^{(Reference Eicher-Miller, Mason and Weaver32,Reference Yang, Sahota and Pickett81)} .

Eating behaviours

Ten studies found that HFI was associated with poor eating habits in children and adolescents, characterised by reduced frequency of family meals, meal skipping, binge eating and overeating in the absence of hunger^{(Reference Appelhans, Waring and Schneider24,Reference Bruening, Lucio and Brennhofer26,Reference Duke30,Reference Fulkerson, Kubik and Story37,Reference Hooper, Telke and Larson45,Reference Kral, Chittams and Moore52,Reference Masler, Palakshappa and Skinner60,Reference Molcho, Gabhainn and Kelly65,Reference Robson, Lozano and Papas71,Reference Widome, Neumark-Sztainer and Hannan79)} . Maternal binge eating and breakfast skipping were associated with food insecure adolescents mimicking this behaviour^{(Reference Bruening, Lucio and Brennhofer26)}. A US study also found that food insecure children enrolled in a national school lunch programme were more likely to skip lunch than those enrolled in the programme^{(Reference Duke30)}. Two studies found that HFI was related to weight-control eating behaviour, including fasting, laxative use and increased weight loss attempts in children and adolescents^{(Reference Hooper, Telke and Larson45,Reference Masler, Palakshappa and Skinner60)} .

Dental cavities

Three studies found an association between HFI and dental caries, which varied depending on HFI severity and child age ^{(Reference Bahanan, Singhal and Zhao15,Reference Hill42,Reference Jackson and Testa47)} . The odds of dental caries were significantly greater (OR = 3·51; 95 % CI 1·71, 7·19; P < 0·05) in children who experienced severe HFI compared with fully food-secure adolescents. Older food insecure children had greater odds of dental caries^{(Reference Hill42)}. Children who experienced any HFI across all severities had poorer oral health compared with fully food-secure children^{(Reference Jackson and Testa47)}.

One study investigating mechanisms found that HFI was directly associated with untreated dental caries in children and adolescents (OR = 1·38; 95 % CI 1·11, 1·72; P < 0·01). In this study, diet quality was not found to be associated with untreated caries, so the authors did not perform a mediation analysis^{(Reference Bahanan, Singhal and Zhao15)}. However, HFI was associated with poorer diet quality in this study.

Prediabetes risk

Four studies found evidence demonstrating an association between HFI and increased prediabetes risk^{(Reference Duke29,Reference Lee, Scharf and Filipp55,Reference Marjerrison, Cummings and Glanville58,Reference Mendoza, Haaland and D’Agostino64)} . HFI was associated with greater HbA1c concentrations (> 9 %, which is high risk) in all studies. Studies reported racial disparities between HFI and prediabetes risk, with higher Hispanic and Black children having greater prediabetes risk (P < 0·001) than food insecure children of other races^{(Reference Lee, Scharf and Filipp55)}. Another study found higher odds of prediabetes risk among food insecure non-White Hispanic adolescents (adjusted ^{(Reference Mendoza, Haaland and D’Agostino64)} OR (aOR) = 2·83; 95 % CI 2·14, 3·73) compared with food insecure adolescents who were of Black race (aOR = 1·88; 95 % CI 1·12, 3·14) and Hispanic adolescents (aOR = 1·84; 95 % CI 1·14, 2·97)^{(Reference Duke29)}.

Hospitalisation risk

Two studies found that food insecure children had greater rates of hospitalisation than food-secure^{(Reference Marjerrison, Cummings and Glanville58)}. HFI was associated with hospitalisations (aOR = 3·66; 95 % CI 1·54, 8·66)^{(Reference Marjerrison, Cummings and Glanville58)} and emergency department visits (prevalence ratio = 2·95; 95 % CI 1·17, 7·45)^{(Reference Mendoza, Haaland and D’Agostino64)}.

Blood pressure

Five studies reported mixed evidence for the association of HFI with blood pressure^{(Reference Fulay, Vercammen and Moran36,Reference Holben and Taylor44,Reference Lee, Scharf and Filipp55,Reference Parker, Widome and Nettleton68,Reference South, Palakshappa and Brown73)} . Three studies found a small positive association, varying by HFI severity, gender, age, ethnicity and household income^{(Reference Lee, Scharf and Filipp55,Reference Parker, Widome and Nettleton68,Reference South, Palakshappa and Brown73)} . Two studies found little or no association^{(Reference Fulay, Vercammen and Moran36,Reference Holben and Taylor44)} .

Cholesterol, fasting glucose and other metabolic markers of health

Three studies found no association between HFI and various metabolic markers of health^{(Reference Fulay, Vercammen and Moran36,Reference Holben and Taylor44,Reference Parker, Widome and Nettleton68)} . One study found a significant difference only for marginally food insecure groups, where the odds of having elevated serum TAG (OR = 1·86 95 %; CI 1·14, 2·82), TAG/HDL-cholesterol (OR = 1·74; 95 % CI 1·11, 2·82) and apo B (OR = 1·98; 95 % CI 1·17, 3·36) were greater than those observed in food-secure groups. In this study, marginally food-secure females had greater odds than males of having low HDL-cholesterol (OR = 2·69; 95 % CI 1·14, 6·37)^{(Reference Tester, Laraia and Leung75)}

Asthma

Two studies found an association between HFI and asthma, which varied by race, household income and timing of HFI^{(Reference Mangini, Hayward and Dong56,Reference Mangini, Hayward and Zhu57)} . Food insecurity was associated with greater odds of asthma in non-Hispanic whites and Hispanics. However, odds were lower in non-Hispanic Black children^{(Reference Mangini, Hayward and Dong56)}. One study found that the timing of HFI was an important determinant in the association between HFI and asthma diagnosis^{(Reference Mangini, Hayward and Zhu57)}. For example, children who experienced HFI a year before entering kindergarten had 13 % greater odds of an asthma diagnosis in the third grade (OR = 1·13; 95 % CI 1·17, 1·20), while HFI experienced in the year prior to joining third grade was associated with 53 % greater odds of developing asthma in the third grade (95 % CI 1·51, 1·55)^{(Reference Mangini, Hayward and Zhu57)}.

Anaemia

One study found a positive association between HFI and anaemia. The odds of iron deficiency anaemia among food insecure adolescents were greater than food-secure adolescents (OR = 2·95; 95 % CI 1·18, 7·37; P = 0·02)^{(Reference Eicher-Miller, Mason and Weaver33)}.

Bone mass disparities

One US study found HFI to be associated with less bone mass, particularly in food insecure male children, who had significantly lower estimated total body (P = 0·05), trunk (P = 0·05), spine (P = 0·2), pelvis (P = 0·05) and left arm (P = 0·02) bone mineral content than food-secure males^{(Reference Eicher-Miller, Mason and Weaver32)}. Food insecure males consumed fewer dairy products, thus having lower calcium intake than recommended. HFI was more prevalent in non-Hispanic Black, Mexican American and other ethnic groups.

Physical activity

Two studies found that fully food-secure adolescents were more likely to participate in moderate-to-vigorous physical activity than food insecure adolescents (P < 0·02)^{(Reference Parker, Widome and Nettleton68,Reference To, Frongillo and Gallegos77)} .

Quality of life

Three studies found that food poverty or HFI were significantly associated with lower quality of life^{(Reference Kirk, Kuhle and McIsaac51)} and lower life satisfaction^{(Reference Molcho, Gabhainn and Kelly65,Reference Ovenell, Azevedo Da Silva and Elgar67)} in children and adolescents aged 10–17 years. Moderate-to-severe HFI was associated with lower health-related quality of life outcomes, particularly psychosocial outcomes (P < 0·05)^{(Reference Kirk, Kuhle and McIsaac51)}.

Sleep

Two studies found a negative association between HFI and sleep (P < 0·001)^{(Reference King49,Reference Robson, Lozano and Papas71)} . In both studies, food insecure children and adolescents reported poor sleep quality. One study found that HFI was not associated with mean-centred sleep duration in adolescents^{(Reference Do, Bowen and Ksinan16)}.

Smoking, alcohol and substance abuse

HFI was associated with greater odds of cigarette smoking^{(Reference Robson, Lozano and Papas71)}, alcohol consumption^{(Reference Robson, Lozano and Papas71)}, opioid misuse and lifetime use of illicit drug use^{(Reference McLaughlin, Green and Alegría63,Reference Turner, Demissie and Sliwa78)} . The relationship between HFI and smoking, alcohol and substance abuse was dependent on age and ethnicity; for example, one study found that food insecure non-Hispanic Black and non-Hispanic white children of older age were more likely to partake in substance abuse behaviour^{(Reference Turner, Demissie and Sliwa78)}.

Mental health and behaviour

Fourteen studies found an association between HFI and detrimental mental health and behavioural difficulties^{(Reference Gee and Asim17,Reference Hatem, Lee and Zhao19,Reference Marçal21,Reference Altman, Ritchie and Frongillo23,Reference Godrich, Loewen and Blanchet39,Reference Hobbs and King43,Reference Jackson and Vaughn46,Reference Maynard, Perlman and Kirkpatrick61–Reference McLaughlin, Green and Alegría63,Reference Niemeier and Fitzpatrick66,Reference Ovenell, Azevedo Da Silva and Elgar67,Reference Ramsey, Giskes and Turrell70,Reference Tevie and Shaya76)} . Most studies explored the relationship in older children or adolescents, with only two studies^{(Reference Gee and Asim17,Reference Hobbs and King43)} reporting in children under 10 years. Four studies found an association between HFI and anxiety, which was significantly worse in females, older children and adolescents and worsened with HFI severity^{(Reference Maynard, Perlman and Kirkpatrick61,Reference McLaughlin, Green and Alegría63,Reference Ovenell, Azevedo Da Silva and Elgar67,Reference Tevie and Shaya76)} . Reports of anxiety were greater in children and adolescents who were not shielded from HFI by caregivers^{(Reference Ovenell, Azevedo Da Silva and Elgar67)}. Four studies found an association between HFI and increased depression/atypical emotional symptoms^{(Reference McIntyre, Williams and Lavorato62,Reference McLaughlin, Green and Alegría63,Reference Ovenell, Azevedo Da Silva and Elgar67,Reference Ramsey, Giskes and Turrell70)} . Furthermore, hunger was associated with greater odds of depression and suicidal ideation in adolescents (OR = 2·3; 95 % CI 1·2, 4·3)^{(Reference McIntyre, Williams and Lavorato62)}.

Four studies found a positive association between HFI and misconduct, behavioural difficulties and internalising and externalising symptoms in children and adolescents^{(Reference Hobbs and King43,Reference Jackson and Vaughn46,Reference McLaughlin, Green and Alegría63,Reference Ramsey, Giskes and Turrell70)} . Persistent HFI was associated with greater misconduct in adolescents (bullying/fighting/stealing/cheating/lying/misbehaving), and this was worse in food insecure males than females compared with their food-secure counterparts^{(Reference Jackson and Vaughn46)}. Additionally, three studies found that HFI was associated with poor self-esteem in older children and adolescents, and this was stronger in girls than in boys ^{(Reference Altman, Ritchie and Frongillo23,Reference Godrich, Loewen and Blanchet39,Reference Niemeier and Fitzpatrick66)} . One study found a positive association between HFI and body dissatisfaction in US adolescents across all races and BMI categories, which was stronger among those with African American race/ethnicity (P < 0·001)^{(Reference Altman, Ritchie and Frongillo23)}.

Three studies explored mediatory pathways between HFI and child/adolescent mental health and behavioural outcomes. One study found that HFI, reported by the parent in early childhood (aged 5 years), was directly associated with adolescent anxiety and depressive symptoms. The study found that both HFI and housing instability combined had an indirect impact on adolescent anxiety and depression via parenting stress and maternal depression, reported when the child was aged 9 years^{(Reference Hatem, Lee and Zhao19)}. A second study found that HFI, at age 5, was not directly associated with adolescent aggressive or depressive behaviour; however, it did have an indirect impact on aggressive behaviour and depressive behaviour via parenting stress reported when the child was aged 9 years^{(Reference Marçal21)}. A third study found that HFI was indirectly associated with lowered ability to pay attention, lower inhibitory control and greater externalising behaviours via parental aggravation in children aged 4–5 years^{(Reference Gee and Asim17)}.

Quality assessment of mechanism studies

Eight mechanism studies, all using data from the USA, in children and adolescents aged 3–17 years, were assessed by the QUIPS tool^{(Reference Bahanan, Singhal and Zhao15–Reference Gee and Asim17,Reference Hatem, Lee and Zhao19–Reference Willis and Fitzpatrick22)} . Low risk of bias was found in seven^{(Reference Bahanan, Singhal and Zhao15,Reference Do, Bowen and Ksinan16,Reference Gundersen, Lohman and Garasky18,Reference Lohman, Stewart and Gundersen20,Reference Marçal21)} out of eight studies, while moderate bias was found in one⁽¹²⁾. Quality assessments are located in the online supplementary material, Supplemental material.

Conceptual model of results

This review shows that the evidence relating HFI to health outcomes remains mixed. HFI has detrimental health impacts on child health outcomes for prediabetes risk, dental cavities, bone mass, asthma, anaemia, physical activity, quality of life, behaviour, diet quality and mental health. However, it is unclear whether or how HFI is related to child overweight/obesity, blood pressure and cholesterol/other biomarkers.

The associations and mechanisms found in this review are illustrated in Fig. 2. The green arrows represent an OR > 1, while the red arrows represent an OR < 1. Dashed lines represent mixed evidence.

Figure 2 Conceptual framework of review results of mechanisms and associations reported between household food insecurity (HFI) and child/adolescent health outcomes and parental mental health outcomes. Red arrows indicate OR > 1 between HFI and outcomes, green arrows indicate OR > 1 between HFI and outcomes and a thick dashed line indicates mixed evidence regarding an association between HFI and outcomes or between outcomes.