for Childhood Internalizing and Externalizing Disorders Natalie Slopen, M.A., Garrett Fitzmaurice, Sc.D., David R. Williams, M.P.H., Ph.D., Stephen E. Gilman, Sc.D. Objective: This study investigated the associations of poverty and food insecurity over a 2-year period with internalizing and externalizing problems in a large, community-based sample. Method: A total of 2,810 children were interviewed between ages 4 and 14 years at baseline, and between ages 5 and 16 years at follow-up. Primary caregivers reported on household income, food insecurity, and were administered the Child Behavior Checklist, from which we derived indicators of clinically signicant internalizing and externalizing problems. Prevalence ratios for the associations of poverty and food insecurity with behavior problems were estimated. Results: At baseline, internalizing and externalizing problems were signif- icantly more prevalent among children who lived in poor households than in nonpoor households, and among children who lived in food insecure households than in food-secure households. In adjusted analyses, children from homes that were persistently food insecure were 1.47 (95% CI 1.12 to 1.94) times more likely to have internalizing problems and 2.01 (95% CI 1.21 to 3.35) times more likely to have externalizing problems compared with children from households that were never food insecure. Children from homes that moved from food secure to insecure were 1.78 (95% CI 1.07 to 2.94) times more likely to have externalizing problems at follow-up. Conclusions: Persistent food insecurity is associated with internalizing and externalizing problems, even after adjusting for sustained poverty and other potential confounders. These results implicate food insecurity as a novel risk factor for child mental well-being; if causal, this represents an important factor in the etiology of child psychopathology, and potentially a new avenue for prevention. J. Am. Acad. Child Adolesc. Psychiatry, 2010;49(5):444452. Key Words: Food insecurity, Poverty, Internalizing prob- lems, Externalizing problems, Prospective. P overty and food insecurity have adverse impacts on the course of child develop- ment, in part by raising the risk for care- giver depression 1-3 and for chronic iron de- ciency and vitamin deciencies. 4,5 According to the U.S. Census, 37.3 million persons in the United States (12.5%) were poor in 2007, as specied by the U.S. federal poverty guidelines, which take into account total household income and number of people residing in the household; for a family of four, this equates to an income below $21,203. 6 During this time, approximately 18.0% of children were poor, with black (or African American) and Hispanic children dispro- portionately affected. Food insecurity, dened as limited or uncertain availability of food because of inadequate resources, is one of many difcul- ties associated with poverty. 7 In a 2007 report, the U.S. Department of Agriculture estimated that 11% of U.S. households experienced food insecu- rity in 2005. Food insecurity was almost twice as common for households with children relative to households without children (15.6% vs. 8.5%), and 30% of single female-households were food insecure. 7 Like poverty, risk of food insecurity is also patterned by race/ethnicity. In the U.S., there is overlap in the populations of individuals meeting the denitions of poverty and food in- security; yet, individuals with food insecurity are This article is discussed in an editorial by Dr. Tamara Dubowitz on page 437. JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 444 www.jaacap.org not always poor, and poor individuals are not always food insecure. 7 For children, experiences of household poverty and/or food insecurity may have an impact on numerous domains of life, including family processes and parental well-being, cognitive and academic performance, and general health outcomes; each of these fac- tors may affect internalizing or externalizing problems. 8,9 There is limited prospective research that has considered the relationship among poverty, food insecurity, and childrens mental health. Prospec- tive study designs enable a rigorous test of the effect of these determinants on child mental health, and are most informative for examining causal inference; this is of substantial importance for researchers and clinicians who are interested in identifying optimal points for intervention. Several prospective studies on the relationship between poverty and child psychopathology suggest that chronic poverty is associated with an elevated risk of psychopathology. 10-13 For example, Bor et al. 10 found that chronic low family income (near or below the poverty line in Australia) was associated with an increased risk of internalizing, externalizing, and social, atten- tional, and thought problems at age 5 years. 10 Costello et al 11 found that an income supplemen- tation that moved families out of poverty was associated with a reduction in oppositional de- ant disorder and conduct disorder symptoms. After the income supplementation, the level of behavioral symptoms for children who were poor at baseline but no longer poor at follow-up fell to those of children who were never poor, whereas the level of symptoms among persis- tently poor children remained elevated in com- parison to the never-poor children. We are not aware of any prospective research on poverty and child mental health that has addressed the issue of food insecurity; however, cross-sectional studies suggest an independent link between food insecurity and higher levels of child behavioral/emotional problems. 2,14-19 For example, Whitaker et al. 2 analyzed data from the Fragile Families and Child Wellbeing Study and found a positive gradient in the prevalence of child behavior problems with increasing levels of food insecurity among children at age 3 years (i.e., 22.7% for those who were fully food inse- cure, 31.1% for marginally food insecure, and 36.7% for food insecure). Alaimo et al. 14 analyzed nationally representative adolescent data and found that food-insufcient adolescents were ap- proximately four times more likely to meet diag- nostic criteria for dysthymia and ve times more likely to have made a suicide attempt; in contrast, low family income (i.e., 130% of the poverty line) did not predict these outcomes. We were able to locate only one study that used a prospec- tive design to examine the impact of food inse- curity on child development, 20 although this study focused on social skills rather than inter- nalizing and externalizing symptoms. Although prospective research on food insecurity and child mental health is lacking, there has been some work in this area with adults. In a prospective study over 3 years, Hein et al 1 found that onset of food insufciency status was positively asso- ciated with onset of maternal major depression status while adjusting for many sociodemo- graphic characteristics, including income. The Project on Human Development in Chicago Neighborhoods (PHDCN) provides an opportu- nity to investigate the role of poverty and food insecurity on risk for childhood psychopathol- ogy. The present study used data over a 2 year period to examine the impact of trajectories of household poverty and food insecurity on change in child outcomes. We hypothesized that (a) persistent poverty and movement into pov- erty status are positively associated with inter- nalizing and externalizing problems, and (b) per- sistent food insecurity and movement into food insecure status are positively associated with internalizing and externalizing problems. METHOD Sample The PHDCN includes a prospective study of children and their families from 80 distinct neighborhoods over an 8-year period (19952002). A three-stage sampling design was used to enroll children and their primary caregivers into the study: rst, 80 neighborhood clus- ters from all neighborhoods in Chicago were selected based on a stratied probability sample of neighbor- hood clusters heterogeneous on race, ethnicity, and income; second, block groups within each neighbor- hood cluster were selected; and third, households with children 6 months post partum, aged 3, 6, 9, 12, 15, and 18 living in the selected block groups, were enrolled in the study. 21 The participation rate at the time of recruitment was 75%. The current study uses data from children aged 4 through 14 years collected be- tween 1997 and 2001; children and primary caregivers were interviewed twice, with an average period of 24 months between interviews. POVERTY, FOOD INSECURITY, AND BEHAVIOR JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 445 www.jaacap.org Internalizing and Externalizing Problems Internalizing and externalizing problems were measured using the Child Behavior Checklist (CBCL) for children 4 through 18 years of age. 22 This measure was adminis- tered to the primary caregiver at baseline and follow-up. The CBCL contains questions pertaining to the childs emotions and behavior in the past 6 months. The care- giver was asked to identify whether each statement has been very true, somewhat true, or not true. The internalizing subscale provides a rating for the extent to which the child has exhibited symptoms of anxiety, depression, or withdrawal. The externalizing subscale provides a rating of the extent to which the child has exhibited symptoms of aggression, hyperactivity, or non- compliance. Both scales have satisfactory validity and internal consistency. 22 We dened problem levels of internalizing and externalizing as scores at or above the 90 th percentile (T-score 64, based on published norms). High scores on the CBCL 1991/4-18 correspond closely to clinical diagnostic outcomes (i.e., anxiety, mood, dis- ruptive behavior, and conduct disorders) 23,24 dened by the Diagnostic and Statistical Manual of Mental Disor- der(DSM) III-R 25 Poverty Status Family poverty was dened according to the U.S. federal poverty threshold, which was originally de- signed to identify individuals and families with insuf- cient income to meet basic needs. The federal poverty threshold takes into account total household income and number of persons living in the household. We created a measure of income to needs for this study by calculating a ratio of household income over the federal poverty threshold; an income-to-needs ratio of less than 1 indicates that a family is below the poverty line. The federal poverty threshold has particular sa- lience for the current study, given that government benets such as welfare are linked to it. We dened family poverty over the course of the study based on poverty status at both time points (persistently poor, previously poor, newly poor, and never poor). Food Insecurity Food insecurity refers to limited or uncertain access to nutritionally adequate and safe foods due to lack of resources (p. 278). 26 The current gold-standard mea- sure of food insecurity is the 18-item U.S. Food Secu- rity Scale, which includes eight child-referenced items that make up the Childrens Food Security Scale. 7 Unfortunately, the PHDCN did not include a validated measure of food insecurity; only a single item was included, which may only partially capture the con- struct of interest. In this study, the primary caregiver provided information about household food insecurity status at baseline and follow-up in response to a single question, which referred to the past 6 months: Has there been a time when there was not enough money at home to buy food? (yes/no). This variable was strongly correlated with caregiver report of having to cut his/her meal size one or more times in the past 6 months (correlation 0.65, p .0001), which provides some assurance that our item reect the construct of interest. These two items were the only interview questions that measured access to food. We created a four-category variable to reect trajectory of food insecurity over time (persistently food insecure, previ- ously food-insure, newly food insecure, never food insecure). Demographic Characteristics We included several child and primary caregiver co- variates in the analyses. Child covariates included age at interview, gender, and race/ethnicity (black, white, Hispanic, and other). Household characteristics in- cluded the number of people residing in the house- hold, and caregiver cohabitation status over time (never partnered/married; newly partnered/married; previously partnered/married; always partnered/ married). Primary caregiver covariates included age at baseline and follow-up, education at baseline (high school degree vs. less than high school), major depres- sion at baseline (based on an adapted version of the Composite International Diagnostic Interview short form for DSM-IV 27 ), and alcohol problems at baseline (dened as a scoring 3 on the Short Michigan Alco- hol Screening Test 28 ). Statistical Analyses Regression analyses were used to relate internalizing and externalizing problems to patterns of poverty and food insecurity. Our rst set of models examined baseline associations among poverty, food insecurity, and problem status while adjusting for sociodemo- graphic characteristics of the child (age, gender, and race/ethnicity), household (cohabitation status, num- ber of persons in the household), and caregiver (age, depression, education, and alcohol problem). Our sec- ond set of models examined the effect of poverty and food insecurity trajectories on internalizing and exter- nalizing problems at follow-up, controlling for base- line problem status, and child, household and care- giver covariates. The prospective models for internalizing and externalizing problems followed the same model-building strategy: the rst two models estimated the separate effects of poverty and food insecurity, and the third model included poverty and food insecurity simultaneously. All analyses were based on Poisson regression models, which yield prev- alence ratios for the relation between the risk factors of interest and child internalizing and externalizing prob- lems. 29-31 Generalized estimating equations were used to obtain variance estimates that account for the nested structured of the data. SLOPEN et al. JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 446 www.jaacap.org Across the variables in our analysis, the percentage of participants with missing data ranged from 0% to 27%; more than half of our sample had at least one missing value, and the overall participation rate at wave 3 was 79.5%. Multiple imputation of missing data was performed using IVEware, 32 retaining all respondents who were enrolled in age cohorts 3, 6, and 9 at wave 1 (N 2,810). This strategy best preserved the original sample characteristics and reduced bias that could result from attrition. PROC MIANALYZE in SAS v.9.1 was used to combine the results from the 10 imputed datasets for all analyses. RESULTS The mean age of the child participants was 8.2 years at baseline, and 10.7 years at follow-up (Table 1). Female participants comprised slightly more than half of the sample (50.5%). Approxi- mately 32% of the sample was non-Hispanic black, 14% was non-Hispanic white, 42% was Hispanic, and 12% was coded as other race. The mean age of the primary caregivers in the sample was 35.6 years at baseline and 38.1 years at follow-up. The prevalence of poverty and food insecurity declined between baseline and follow- up: at baseline, 40% of families in the study were below the poverty line, and 23% of families were food insecure; at follow-up, the prevalences fell to 30% and 18%, respectively. Although poverty and food insecurity have some degree of overlap, these constructs are not entirely overlapping: at baseline, 52% of the sample was neither food insecure or poor, whereas 15% of our sample was both food insecure and poor; 8% was food inse- cure but not poor, whereas 25% was poor but not food insecure. The prevalence of internalizing and externaliz- ing problems was stable across waves, with ap- proximately 18% of children categorized as having internalizing problems, and 7% of children catego- rized as having externalizing problems at both time points. There was a relatively high degree of co- morbidity between internalizing and externalizing problems: at baseline, 27% of children with inter- nalizing problems also met criteria for externalizing problems; at follow-up, 25% of children with inter- TABLE 1 Characteristics of Study Participants (N 2,810) Characteristic Baseline Follow-up N % Mean (SE) N % Mean (SE) Below poverty line 1,135 40.38 856 30.48 Food insecure 643 22.90 516 18.36 Total household income ($ US) 2,810 31,1163 (481.11) 2810 37,669 (1101.54) Child characteristics Internalizing problems 508 18.07 482 17.16 Externalizing problems 213 7.57 191 6.79 Child age (years) 2,810 8.16 (0.05) 2810 10.67 (0.08) Gender Male 1,391 49.50 Female 1,419 50.50 Race Non-Hispanic black 897 31.92 Non-Hispanic white 388 13.82 Hispanic 1,195 42.54 Other 329 11.72 Primary caregiver characteristics Caregiver age, years 2,810 35.58 (0.31) 2810 38.09 (0.29) Less than high school education 1059 37.70 Caregiver depression 416 14.82 Caregiver alcohol problem 96 3.43 No live-in partner/spouse 915 32.57 907 32.28 Number in household 2810 5.34 (0.05) 2810 5.22 (0.14) Note: N values are based on multiply imputed data, and are shown rounded to full numbers. SE standard error for parameter estimates. POVERTY, FOOD INSECURITY, AND BEHAVIOR JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 447 www.jaacap.org nalizing problems also met criteria for externalizing problems. Internalizing and externalizing problems were signicantly more prevalent among children who lived in poor households as compared with nonpoor households (internalizing problems: 22.3% vs. 15.2%, p .0001; externalizing problems: 10.3% vs. 5.7%, p .0001), and among children who lived in food-insecure households as com- pared with food-secure households (internalizing problems: 27.6% vs. 15.4%, p .0001; externalizing problems: 10.6% vs. 6.7%, p 0.002). Children exposed to poverty and food insecurity at the same time had the greatest prevalence of internalizing problems (30.0%) and externalizing problems (12.3%) relative to all other groups. Before examining the prospective associations among poverty, food insecurity, and childhood internalizing and externalizing problems, we ex- amined the baseline associations, adjusted for child, household, and caregiver characteristics (child age, gender, race/ethnicity; caregiver age, cohabitation status, depression, alcohol prob- lems, education; and number in the household). In this analysis, poverty status was signicantly associated with internalizing problems: children from households below the poverty line were 1.32 (95% CI 1.08 to 1.62) times more likely to have internalizing problems, relative to children from households above the poverty line. Poverty status was also associated with externalizing problems (prevalence ratio [PR] 1.38, CI 0.99 to 1.92). Children in food insecure households were 1.49 (95% CI 1.24 to 1.79) times more likely to have internalizing problems, and 1.47 (95% CI 1.10 to 1.97) times more likely to have externalizing problems, relative to children in food secure households. Internalizing and Externalizing Problems Over the Follow-Up Period Internalizing Problems. In the prospective models adjusted for prior problem status and sociode- mographic characteristics (Table 2), persistent poverty was marginally associated with internal- izing problems at follow-up (PR 1.33, 95% CI 0.99 to 1.77). Movement into poverty status was not signicantly associated with internalizing problems, although the effect estimate was in the expected direction (PR 1.35, 95% CI 0.94 to 1.93). In contrast, there was a signicant associa- tion between persistent food insecurity and inter- nalizing problems: children from homes that were food insecure at baseline and follow-up (i.e., persistent food insecurity) were 1.56 (95% CI 1.20 to 2.02) times more likely to have inter- nalizing problems, relative to children from homes that were never food insecure. Children from homes that moved from food secure at baseline to food insecure at follow-up were 1.39 times more likely to have internalizing problems at follow up, although this association did not TABLE 2 Estimated Prevalence Ratios (and 95% Condence Intervals) for Internalizing Problems at Follow-up (N 2,810) a Number with Internalizing Problems % with Internalizing Problems in Each Category Model 1 Model 2 Model 3 Poverty Status Persistently poor 150 22.38 1.33 (0.99, 1.77) 1.21 (0.89, 1.65) Previously poor 88 18.94 1.13 (0.84, 1.51) 1.08 (0.80, 1.46) Newly poor 38 20.62 1.35 (0.94, 1.93) 1.25 (0.87, 1.81) Never poor 206 13.83 1.00 1.00 Food Insecurity Status Persistently food insecure 83 31.24 1.56 (1.20, 2.02)** 1.47 (1.12, 1.94)* Previously food insecure 81 21.26 1.22 (0.93, 1.59) 1.19 (0.90, 1.56) Newly food insecure 56 22.34 1.39 (0.98, 1.99) 1.37 (0.96, 1.96) Never food insecure 263 13.72 1.00 1.00 Note: N values are based on multiply imputed data, and are shown rounded to full numbers. a In addition to covariates presented, all models adjusted for child characteristics (gender, age, race, and internalizing problems at baseline), caregiver characteristics (age, education, cohabitation status, depression, and alcohol problem), and number people living in the household. Prevalence ratios should be interpreted as the increase or decrease in the probability of internalizing or externalizing problems, in comparison to the reference level. *p 0.05; **p 0.01. SLOPEN et al. JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 448 www.jaacap.org reach statistical signicance (95% CI 0.98 to 1.99). As shown in model 3, persistent food insecurity was a signicant predictor of internal- izing problems at follow-up, even when poverty trajectory is included in the model (PR 1.47, 95% CI 1.12 to 1.94). Externalizing Problems. Persistent poverty and movement into poverty status were not signi- cantly associated with externalizing problems at follow-up (Table 3). However, signicant associ- ations were observed for trajectories of house- hold food insecurity. Children from homes that were food insecure at baseline and follow-up (i.e., that experienced persistent food insecurity) were 2.11 (95% CI 1.30 to 3.45) times more likely to have externalizing problems at follow- up, relative to children from homes that were never food insecure. In addition, children from homes that moved from food secure at baseline to food insecure at follow-up were 1.80 (95% CI to 1.09 to 2.98) times more likely to have exter- nalizing problems at follow-up, relative to chil- dren from homes that were food secure at both times. The associations between persistent food insecurity and movement into food insecurity status with externalizing problems remained sig- nicant even when trajectory of poverty status was included in the model (model 3). Of note, a comparable set of analyses of the continuous scores for internalizing and externalizing symp- toms yielded similar results. Finally, to test the sensitivity of the poverty threshold used in the prospective analyses, we reanalyzed the data with the threshold for pov- erty status set to 130% of the federal poverty line (the eligibility standard for the federal Supple- mental Nutrition Assistance Program); in these analyses, the prevalence ratios were largely unchanged. DISCUSSION The present study used data from two waves of the PHDCN to investigate associations among poverty, food insecurity, and child mental health. Our ndings are of importance from a preven- tion standpoint, because both poverty and food insecurity are modiable conditions that can be addressed through social and economic interven- tions at the level of the community, state, or federal government. 33,34 The analyses of baseline data revealed that children from households be- low the poverty line were signicantly more likely to have internalizing problems, and that children from food insecure households were signicantly more likely to have internalizing and externalizing problems, independent of pov- erty status. TABLE 3 Estimated Prevalence Ratios (and 95% Condence Intervals) for Externalizing Problems at Follow-up (N 2,810) a Number With Externalizing Problems Percent With Externalizing Problems in Each Category Model 1 Model 2 Model 3 Poverty status Persistently poor 75 11.16 1.37 (0.91, 2.06) 1.17 (0.76, 1.80) Previously poor 26 5.52 0.88 (0.49, 1.58) 0.83 (0.47, 1.45) Newly poor 18 9.88 1.31 (0.70, 2.45) 1.21 (0.67, 2.18) Never poor 72 4.84 1.00 1.00 Food insecurity status Persistently food insecure 38 14.49 2.11 (1.30, 3.45)** 2.01 (1.21, 3.35)** Previously food insecure 31 8.20 1.52 (0.91, 2.54) 1.51 (0.88, 2.56) Newly food insecure 30 11.76 1.80 (1.09, 2.98)* 1.78 (1.07, 2.94)* Never food insecure 92 4.80 1.00 1.00 Note: N values are based on multiply imputed data, and are shown rounded to full numbers. a In addition to covariates presented, all models adjusted for child characteristics (gender, age, race/ethnicity, and externalizing problems at baseline), caregiver characteristics (age, education, cohabitation status, depression, and alcohol problem), and number of persons living in household. Prevalence ratios should be interpreted as the increase or decrease in the probability of internalizing or externalizing problems, in comparison to the reference level. *p 0.05; **p 0.01. POVERTY, FOOD INSECURITY, AND BEHAVIOR JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 449 www.jaacap.org In the prospective analysis, although the asso- ciations were in the expected direction, trajectory of poverty status did not signicantly predict internalizing or externalizing problems, control- ling for prior problem status, and other charac- teristics. This nding was contrary to our rst hypothesis and was inconsistent with prior re- search that has documented an association be- tween poverty trajectory and change in mental health. 11,13 These results may differ from previ- ous research because of study design: our study included only two waves of data, with approxi- mately 2 years between data collection, whereas other studies have had a longer observation period and/or more frequent assessment and have used symptom scores rather than diagnostic outcomes. 11,13,35 Our ndings for the trajectory of food insecurity provided support for our second hypothesisnamely, that children who lived in households that were food insecure at both time points had a greater likelihood of internalizing and externalizing problems at follow-up, inde- pendent of problem status at baseline. The strong and signicant associations between persistent food insecurity and internalizing and externaliz- ing problems after adjustment for poverty status indicates that the associations between food in- security and internalizing and externalizing problems are not confounded by poverty status. Our pattern of results raises several questions. In particular, it is noteworthy that household poverty is signicantly associated with internal- izing problems and marginally associated with externalizing problems in the cross-sectional analysis of the baseline data, but that persistent poverty or movement into poverty status was not a statistically signicant predictor of internaliz- ing or externalizing problems at follow-up. This inconsistency may be caused by unadjusted con- founding in the cross-sectional results that is accounted for in the prospective analyses by control of prior internalizing or externalizing problem status. Alternatively, it could also be linked to length of follow-up. It is also challeng- ing to reconcile our ndings that trajectory of food insecurity is predictive of internalizing and externalizing problems at follow-up, yet trajec- tory of poverty status is not. It is possible that food insecurity is a more proximal measure of stress load in the household, which may ulti- mately lead to the associations we observe. Fi- nally, it is puzzling that caregiver depression (measured at baseline) is a signicant predictor of internalizing and externalizing problems at baseline but not at follow-up; this may suggest that the effects of caregiver depression on child outcomes are largely contemporaneous in nature, rather than persisting over the long term. Unfor- tunately, we are unable to examine the associa- tion between caregiver depression at follow-up and behavior at follow-up because we do not have a usable measure of caregiver depression at this time. Several mechanisms may explain the associ- ations between food insecurity and mental health problems that we observed. Food and nutrition are important aspects of daily life; psychologically, routines involving food are important for the childs comfort and feelings of security. Moreover, all individuals in the household are potentially affected by food in- security; therefore mental and physical well- being of family members may be compromised, which may have an impact on familial efcacy for a wide range of activities. 9 In addition, food insecurity may have negative consequences for a childs nutrient balance and physical health, as well as cognitive and academic outcomes, which place children at risk for the develop- ment of behavior problems. 9,36 Each of these potential mechanisms represents an important direction for future research. The ndings for this study should be consid- ered in the context of several limitations. First, household food insecurity was measured using a single dichotomous question; this crude measure is not able to capture the full extent of food insecurity that a child has experienced, and does not measure child hunger experiences. However, within our data, this single-item measure is strongly correlated with caregiver report of cut- ting meal sizes, which indicates a lack of food available to members in the household. Evidence suggests that even slight forms of food insecurity and undernutrition can have lasting effects; 37 therefore we believe that the available measure is a useful, albeit imperfect, indicator. Second, our study is limited by incomplete data and loss to follow-up. We controlled for variables associated with attrition, which should prevent bias; in addition, we used multiple imputation to pre- serve the original sample composition. Third, the federal poverty line is not specic to region or urban/rural location; it is possible that the pov- erty line does not accurately represent the amount of money that a family requires to be SLOPEN et al. JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 450 www.jaacap.org self-sufcient in an urban area such as Chicago. As a result, some individuals categorized as nonpoor might meet criteria for poor if the threshold was adjusted for city-specic costs of living. However, as noted in the results section, our sensitivity analysis indicated that the results were largely unchanged when we altered the threshold for poverty status. Fourth, our mea- sures of internalizing and externalizing problems are limited to caregiver report; these ratings may be biased as the result of caregiver mental health. Although some researchers have hypothesized that depressed mothers report distorted percep- tions of childrens behavior, a review on this topic did not nd evidence to support this hy- pothesis. 38 Nevertheless, to rule out this possibil- ity, we repeated our analyses excluding children that had caregivers with major depression at baseline; our ndings remain largely the same. Finally, although we adjusted for a broad set of caregiver characteristics, it is important to con- sider the possibility that unmeasured factors re- lated to parenting characteristics may predict both food insecurity and psychological symp- toms in children. The above limitations highlight the need for additional prospective research that includes frequent and detailed measures of chil- dren and caregivers. 39 The present study has several implications for understanding the etiology of child psychiatric disorders, given that CBCL scores converge with DSM psychiatric diagnoses and reect substan- tial psychopathology. 23,24 This set of ndings contributes to our understanding about the etiol- ogy of childhood disorders, as it demonstrates that (a) continued exposure to household food insecurity is predictive of internalizing and ex- ternalizing problems, and (b) excess risk for internalizing and externalizing disorders is not discernible among children who lived in house- holds that were food insecure at baseline and then became food secure at follow-up. This sug- gests that interventions that alleviate household poverty and food insecurity among families ex- periencing these conditions may be effective in reducing the prevalence of childhood behavior problems. Our ndings illustrate the importance of screening children and caregivers about their access to food within the clinical setting, and developing an infrastructure to connect clients with nutritional resources (e.g., Supplemental Nutrition Assistance Program [SNAP], food pan- tries) when household food insecurity is present. Our results suggest that in addition to fullling a physical need, food has an important role in psychological well-being; therefore, food insecu- rity may be an important trait for clinicians to assess, as it may have a role in the development of behavioral problems. In conclusion, poverty and food insecurity are risk factors that can be targeted for intervention. An example of a successful poverty intervention is the Minnesota Family Investment Program. 34 This program has demonstrated that increased employment rates and small increases in income for working poor parents can reduce child pov- erty and lead to reductions in externalizing prob- lems among children. Research has also docu- mented that the Food Stamp Program (now referred to as SNAP) and the Special Supplement Nutrition Program for Women, Infants, and Chil- dren are able to increase nutrient intake among low income children. 39 Additional research is required to expand our ability to effectively pro- vide adequate nutrition to children and families that have poor access to high-quality foods, and to improve our understanding of adequate nu- trition in relation to child mental health. In summary, our ndings provide evidence for a prospective association between food insecurity and increased risk for internalizing and external- izing problems. This study highlights the impor- tance of efforts to develop and strengthen pro- grams that protect vulnerable families from the experiences and consequences of poverty and food insecurity. & Accepted February 16, 2010. Ms. Slopen and Drs. Fitzmaurice, Williams, and Gilman are with the Harvard School of Public Health. Dr. Williams is also with Harvard University. This work was supported by a doctoral research award from the Canadian Institutes of Health Research (Award 200610MDR- 168591-163832). Disclosure: Ms. Slopen and Drs. Fitzmaurice, Williams, and Gilman report no biomedical nancial interests or potential conicts of interest. Correspondence to Natalie Slopen: Department of Society, Human Development, and Health, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115; e-mail: nslopen@hsph. harvard.edu 0890-8567/10/2010 American Academy of Child and Adoles- cent Psychiatry DOI: 10.1016/j.jaac.2010.01.018 POVERTY, FOOD INSECURITY, AND BEHAVIOR JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 451 www.jaacap.org REFERENCES 1. Hein CM, Siefert K, Williams DR. Food insufciency and womens mental health: ndings from a 3-year panel of welfare recipients. Soc Sci Med. 2005;61:1971-1982. 2. Whitaker RC, Phillips SM, Orzol SM. Food insecurity and the risks of depression and anxiety in mothers and behavior problems in their preschool-aged children. Pediatrics. 2006;118:e859-e868. 3. Beardslee WR, Versage EM, Gladstone TRG. Children of affec- tively ill parents: a review of the past 10 years. J Am Acad Child Adolesc Psychiatry. 1998;37:1134-1141. 4. Heseker H, Kubler W, Pudel V, et al. Psychological disorders as early symptoms of a mild-to-moderate vitamin deciency. Ann N Y Acad Sci. 1992;669:352-357. 5. Lozoff B, Jimenez E, Hagen J, et al. Poorer behavioral and developmental outcome more than 10 years after treatment for iron deciency in infancy. Pediatrics. 2000;105:E51. 6. Bishaw A, Semega J. American Community Survey Reports, ACS-09, Income, Earnings, and Poverty Data from the 2007 American Community Survey. Washington, DC: U.S. Census Bureau, U.S. Government Printing Ofce; 2008. 7. Nord M, Andrews M, Carlson S. Measuring Food Security in the United States: Household Food Security in the United States, 2005. Washington, DC: USDA Economic Research Service. Eco- nomic Research Report No. 29; 2007. 8. Brooks-Gunn J, Duncan GJ. The effects of poverty on children. Future Child. 1997;7:55-71. 9. Ashiabi GS, ONeal KK. A framework for understanding the association between food insecurity and childrens developmen- tal outcomes. Child Develop Perspect. 2008;2:71-77. 10. Bor W, Najman JM, Andersen MJ, et al. The relationship between low family income and psychological disturbance in young children: an Australian longitudinal study. Aust N Z J Psychiatry. 1997;31:664-675. 11. Costello EJ, Compton SN, Keeler G, et al. Relationships between poverty and psychopathology: a natural experiment. J Am Med Assoc. 2003;290:2023-2029. 12. Duncan GJ, Brooks-Gunn J, Klebanov PK. Economic deprivation and early child development. Child Dev. 1994;65:296-318. 13. McLeod JD, Shanahan MJ. Trajectories of poverty and childrens mental health. J Health Soc Behav. 1996;37:207-220. 14. Alaimo K, Olson CM, Frongillo EA. Family food insufciency, but not low family income, is positively associated with dysthymia and suicide symptoms in adolescents. J Nutr. 2002;132:719-725. 15. Alaimo K, Olson CM, Frongillo EA, Jr.Food insufciency and American school-aged childrens cognitive, academic, and psy- chosocial development. Pediatrics. 2001;108:44-53. 16. Cook JT, Frank DA. Food security, poverty, and human develop- ment in the United States. Ann N Y Acad Sci. 2008;1136:193-209. 17. Murphy JM, Wehler CA, Pagano ME, et al. Relationship between hunger and psychosocial functioning in low-income American children. J Am Acad Child Adolesc Psychiatry. 1998;37:163-170. 18. Rose-Jacobs R, Black MM, Casey PH, et al. Household food insecurity: associations with at-risk infant and toddler develop- ment. Pediatrics. 2008;121:65-72. 19. Weinreb L, Wehler C, Perloff J, et al. Hunger: its impact on childrens health and mental health. Pediatrics. 2002;110:e41. 20. Jyoti DF, Frongillo EA, Jones SJ. Food insecurity affects school childrens academic performance, weight gain, and social skills. J Nutr. 2005;135:2831-2839. 21. Earls F, Buka SL. Project on Human Development in Chicago Neighborhoods. Washington, DC: National Institute of Justice; 1997. 22. Achenbach TM. Manual for the Child Behavior Checklist/4-18 and 1991 Prole. Burlington, VT: Department of Psychiatry, University of Vermont; 1991. 23. Kasius MC, Ferdinand RF, van den Berg H, et al. Associations between different diagnostic approaches for child and adolescent psychopathology. J Child Psychol Psychiatry. 1997;38:625-632. 24. Ferdinand RF. Validity of the CBCL/YSR DSM-IV scales anxiety problems and affective problems. J Anxiety Disord. 2008;22:126- 134. 25. American Psychiatric Association. Diagnostic and Statistical Man- ual of Mental Disorders: DSM-III-R. Washington, DC: American Psychiatric Association; 1987. 26. Kleinman RE, Murphy JM, Wieneke KM, et al. Use of a single- question screening tool to detect hunger in families attending a neighborhood health center. Ambul Pediatr. 2007;7:278-284. 27. Kessler RC, Andrews G, Mroczek D, Ustun B, Wittchen H. The World Health Organization Composite International Diagnostic Interview short-form (CIDI-SF). Int J Methods Psychiatr Res. 1998;7:171-185. 28. Selzer ML, Vinokur A, van Rooijen L. A self-administered Short Michigan Alcoholism Screening Test (SMAST). J Stud Alcohol. 1975;36:117-126. 29. McNutt LA, Wu C, Xue X, et al. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol. 2003;157:940-943. 30. Zou G. A modied Poisson regression approach to prospective studies with binary data. Am J Epidemiol. 2004;159:702-706. 31. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol. 2005;162:199- 200. 32. Raghunathan TE, Lepkowski JM, Van Hoewyk J, et al. A multi- variate technique for multiply imputing missing values using a sequence of regression models. Survey Methodol. 2001;27:85-95. 33. Siefert K, Finlayson TL, Williams DR, et al. Modiable risk and protective factors for depressive symptoms in low-income Afri- can American mothers. Am J Orthopsychiatry. 2007;77:113-123. 34. Gennetian LA, Miller C. Children and welfare reform: a view from an experimental welfare program in Minnesota. Child Dev. 2002;73:601-620. 35. Dearing E, McCartney K, Taylor BA. Change in family income- to-needs matters more for children with less. Child Dev. 2001;72: 1779-1793. 36. Skalicky A, Meyers AF, Adams WG, et al. Child food insecurity and iron deciency anemia in low-income infants and toddlers in the United States. Matern Child Health J. 2006;10:177-185. 37. Chilton M, Chyatte M, Breauz J. The negative effects of poverty and food insecurity on child development. Indian J Med Res. 2007;126:262-272. 38. Richters JE. Depressed mothers as informants about their chil- dren: a critical review of the evidence for distortion. Psychol Bull. 1992;112:485-499. 39. Rose D, Habicht JP, Devaney B. Household participation in the Food Stamp and WIC programs increases the nutrient intakes of preschool children. J Nutr. 1998;128:548-555. SLOPEN et al. JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY VOLUME 49 NUMBER 5 MAY 2010 452 www.jaacap.org