Development nd Vlidity of 2-Item Screen to Identify Fmilies t Risk for Food Insecurity WHAT S KNOWN ON THIS SUBJECT: Food insecurity (FI) in the United Sttes is public helth problem. FI mong young children is often invisible, becuse lthough young children who experience FI my experience negtive helth nd developmentl outcomes, their growth is often unffected. WHAT THIS STUDY ADDS: Providers need efficient methods for identifying young children in food-insecure households to ensure tht fmilies hve ccess to nutrition-relted services tht provide helthy food nd llevite cregiver stress. We present here brief, sensitive, specific, nd vlid FI screen. bstrct OBJECTIVES: To develop brief screen to identify fmilies t risk for food insecurity (FI) nd to evlute the sensitivity, specificity, nd convergent vlidity of the screen. PATIENTS AND METHODS: Cregivers of children (ge: birth through 3 yers) from 7 urbn medicl centers completed the US Deprtment of Agriculture 18-item Household Food Security Survey (HFSS), reports of child helth, hospitliztions in their lifetime, nd developmentl risk. Children were weighed nd mesured. An FI screen ws developed on the bsis of ffirmtive HFSS responses mong food-insecure fmilies. Sensitivity nd specificity were evluted. Convergent vlidity (the correspondence between the FI screen nd theoreticlly relted vribles) ws ssessed with logistic regression, djusted for covrites including study site; the cregivers rce/ethnicity, US-born versus immigrnt sttus, mritl sttus, eduction, nd employment; history of brestfeeding; child s gender; nd the child s low birth weight sttus. RESULTS: The smple included 30 098 fmilies, 23% of which were food insecure. HFSS questions 1 nd 2 were most frequently endorsed mong food-insecure fmilies (92.5% nd 81.9%, respectively). An ffirmtive response to either question 1 or 2 hd sensitivity of 97% nd specificity of 83% nd ws ssocited with incresed risk of reported poor/fir child helth (djusted odds rtio [OR]: 1.56; P.001), hospitliztions in their lifetime (OR: 1.17; P.001), nd developmentl risk (OR: 1.60; P.001). CONCLUSIONS: A 2-item FI screen ws sensitive, specific, nd vlid mong low-income fmilies with young children. The FI screen rpidly identifies households t risk for FI, enbling providers to trget services tht meliorte the helth nd developmentl consequences ssocited with FI. Peditrics 2010;126:e26 e32 AUTHORS: Erin R. Hger, PhD, Ann M. Quigg, MA,,b Mureen M. Blck, PhD, Shron M. Colemn, MS, MPH, c Timothy Heeren, PhD, c Ruth Rose-Jcobs, ScD, d John T. Cook, PhD, d Stephnie A. Ettinger de Cub, MPH, c Ptrick H. Csey, MD, e Mrin Chilton, PhD, f Din B. Cutts, MD, g Aln F. Meyers, MD, MPH, d nd Deborh A. Frnk, MD d Deprtment of Peditrics, University of Mrylnd School of Medicine, Bltimore, Mrylnd; b Deprtment of Psychology, University of Mrylnd Bltimore County, Bltimore, Mrylnd; c Dt Coordinting Center, Boston University School of Public Helth, Boston, Msschusetts; d Deprtment of Peditrics, Boston University School of Medicine, Boston, Msschusetts; e Deprtment of Peditrics, University of Arknss for Medicl Sciences, Little Rock, Arknss; f Deprtment of Helth Mngement nd Policy, Drexel University School of Public Helth, Phildelphi, Pennsylvni; nd g Deprtment of Peditrics, Hennepin County Medicl Center, Minnepolis, Minnesot KEY WORDS food insecurity, screening tools, nutrition, child development, hunger ABBREVIATIONS FI food insecurity HFSS Household Food Security Survey PEDS Prents Evlutions of Developmentl Sttus OR djusted odds rtio CI confidence intervl The uthors tke public responsibility for the content. All uthors certify tht they contributed substntilly to conception nd design or nlysis nd interprettion of the dt, drfting, or revision of content nd pprovl of the finl version. www.peditrics.org/cgi/doi/10.1542/peds.2009-3146 doi:10.1542/peds.2009-3146 Accepted for publiction Apr 5, 2010 Address correspondence to Erin R. Hger, PhD, Deprtment of Peditrics, University of Mrylnd School of Medicine, 737 W Lombrd St, Room 163, Bltimore, MD 21201. E-mil: ehger@peds.umrylnd.edu PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright 2010 by the Americn Acdemy of Peditrics FINANCIAL DISCLOSURE: The uthors hve indicted they hve no finncil reltionships relevnt to this rticle to disclose. e26
ARTICLES The US Deprtment of Agriculture hs reported tht 14.6% of US households in 2008 were food insecure, mening tht t some time during the yer they were unble to obtin dequte food becuse of constrined resources. 1 Adequte refers to the quntity or qulity of food for ll household members to mintin n ctive lifestyle t ll times. Among households with children, 21% were food insecure. 1 Blck or Hispnic households with single prents, young children, nd incomes below the federl poverty line re t incresed risk for food insecurity (FI). 2,3 Household FI is serious public helth concern, prticulrly for young children. Children in food-insecure households experience dverse helth nd development ttributble to the indequte qulity nd quntity of foods nd to overll fmily stress. 4 Our reserch group (Children s HelthWtch) found tht children younger thn 3 yers who live in food-insecure households hve 90% greter djusted odds of being in fir/poor helth (versus good/ excellent), 31% greter djusted odds of being hospitlized since birth, 5 nd 76% greter djusted odds of being t incresed developmentl risk compred with food-secure fmilies. 6 Cregivers with positive depression-screen results hve 2.69 times the odds of reporting FI compred with cregivers with negtive depression-screen results. 7 Providers need efficient methods for identifying young children in foodinsecure households to ensure tht fmilies hve ccess to nutritionrelted services tht provide helthy food nd llevite cregiver stress. Severl questionnires re vilble to identify food-insecure households. The 18-item US Household Food Security Scle (HFSS) is used by the Current Popultion Survey to monitor ntionl food-security sttus nnully. 2,8,9 Although the HFSS is widely used, it is time-consuming to dminister nd hs complex scoring lgorithm, which limits its use s clinicl tool. Severl shortened questionnires hve been published, including the HFSS Short Form ( 6-item version with excellent sensitivity nd good specificity) 9,10 nd recently developed single-item screen for hunger. 11 Although the hunger question hs cceptble sensitivity nd specificity for identifying fmilies t risk for hunger (ie, the senstion cused by involuntry lck of food), 11 the exclusive focus on hunger my miss food-insecure fmilies tht experience stress relted to uncertin ccess to enough food but not the physiologic senstion of hunger. Dt from the Community Childhood Hunger Identifiction Project 12,13 nd Children s HelthWtch 5,6 suggest tht negtive effects of FI ( t risk for hunger ) on child helth nd behvior re present before reching the threshold for hunger. 14 Dt collected by Children s Helth- Wtch provide unique opportunity to develop n FI screen, to test the sensitivity nd specificity of the screen ginst the HFSS, nd to test the convergent vlidity ginst negtive helth outcomes for cregivers nd children known to be ssocited with FI. The purpose of this study ws to develop brief screen to identify fmilies t risk for FI nd to exmine the sensitivity, specificity, nd convergent vlidity of the screen in multisite smple of low-income fmilies with young children. PATIENTS AND METHODS Prticipnts Dt were obtined from 30 098 cregivers interviewed in hospitl-bsed settings between 1998 nd 2005 s prt of Children s HelthWtch (formerly the Children s Sentinel Nutrition Assessment Progrm [C-SNAP]) in Bltimore, Mrylnd; Boston, Msschusetts; Little Rock, Arknss; Los Angeles, Cliforni; Minnepolis, Minnesot; Phildelphi, Pennsylvni; nd Wshington, DC. Institutionl review bord pprovl ws obtined from ech site. Trined interviewers surveyed cregivers who ccompnied children younger thn 36 months in cute/ primry cre clinics nd hospitl emergency deprtments during pek ptient flow times. Cregivers of criticlly ill or injured children were not pproched. Potentil respondents were excluded if they did not spek English, Spnish, or (in Minnepolis only) Somli, were not knowledgeble bout the child s household, lived out of stte, or did not provide informed consent. To ensure tht fmilies hd low income, nlyses were limited to fmilies who were uninsured or receiving public insurnce. Mesures All mesures re prt of the Children s HelthWtch survey instrument. Demogrphics Cregivers reported their ge, rce/ ethnicity, country of origin, mritl nd employment sttus, level of eduction, nd the child s ge nd gender. Food Insecurity The 18-item US HFSS, 2,8,9 which serves s the gold stndrd in the ssessment of household food security, ws used in this study. According to estblished procedures from the US Deprtment of Agriculture, households re clssified s food insecure if they endorse 3 ffirmtive responses to 18 totl questions. 2,9 Child Helth Outcomes Cregivers reported their child s birth weight, brestfeeding history, nd number of lifetime hospitliztions (excluding birth). Cregivers rted the PEDIATRICS Volume 126, Number 1, July 2010 e27
child s helth s excellent, good, fir, or poor, which yielded binry vrible (excellent/good versus fir/ poor). 15 Developmentl risk ws mesured by using the Prents Evlutions of Developmentl Sttus (PEDS), 10-item screen of prents concerns bout their children s development tht meets stndrds set by the Americn Acdemy of Peditrics for developmentl screening. 16 20 Developmentl risk ws defined by using published guidelines 21 s cregiver report of 1 or more developmentlly pproprite concerns. The PEDS smple ws restricted to children older thn 4 months, becuse the sensitivity nd specificity of the PEDS re better for children thn for infnts younger thn 4 months of ge. 22 The PEDS instrument ws incorported into the Children s HelthWtch survey instrument in 2004. Child Anthropometric Mesurements At the time of the interview, the child s weight nd length were mesured nd recorded by using equipment nd protocols stndrdized cross Children s HelthWtch sites. 6 Weight-for-length nd weight-for-ge z scores were clculted by using the 2000 US Centers for Disese Control nd Prevention ge- nd gender-specific reference vlues. 23 At risk for underweight ws defined s weight for ge t 5th percentile or weight for length t 10th percentile. Overweight ws defined s weight for length t 95th percentile for children younger thn 24 months nd BMI for ge t 85th percentile for children 24 months of ge or older. Cregiver Helth Outcomes Cregivers rted their physicl helth s excellent, good, fir, or poor, which yielded binry vrible (excellent/ good versus fir/poor). 15 Cregivers completed 3-item depression screen tht hs sensitivity of 100%, specificity of 88%, nd positive predictive vlue of 66% compred with the 8-item Rnd screening instrument. 24 Respondents with 2 or more positive responses were coded s hving positive depression-screen result. 24 FI-Screen Development Screen development includes considertion of sensitivity (the screen s bility to correctly identify food-insecure households), specificity (the screen s bility to correctly identify food-secure households), nd convergent vlidity (correspondence between the screen nd theoreticlly relted vribles). 25 27 We sought to develop n FI screen from the HFSS with 5 specific chrcteristics: (1) pplicble to fmilies with young children; (2) brief; (3) highly sensitive ( 90%); (4) specific ( 80%); nd (5) vlid (convergent vlidity). The prevlence of ffirmtive responses for ech item on the HFSS ws clculted for the totl smple nd for foodinsecure fmilies. Prevlence dt were used to generte sensitivity nd specificity tbles for combintions of 1 or 2 questions with the highest prevlence of ffirmtive responses mong food-insecure fmilies. Convergent vlidity ws exmined by using demogrphic nd helth informtion on smple of low-income fmilies cross 7 diverse US cities. Sttisticl Anlyses Anlyses were conducted by using SAS 9.1 (SAS Institute Inc, Cry, NC). Demogrphics nd Helth Outcomes Dt were exmined by using frequencies of demogrphic nd helthrelted vribles for children nd cregivers ccording to FI sttus bsed on the 18-item HFSS. 2 nlyses were conducted to determine differences in demogrphics nd helth outcomes ccording to FI sttus. Sensitivity nd Specificity A2 2tble of FI sttus bsed on HFSS criteri nd the FI screen ws generted. Sensitivity ws clculted s the number of food-insecure fmilies correctly identified by the FI screen divided by the number of food-insecure fmilies identified with HFSS criteri. Specificity ws clculted s the number of fmilies correctly identified by the FI screen s food secure divided by the number of food-secure fmilies identified with HFSS criteri. Convergent Vlidity Convergent vlidity ws tested by using 2 sets of logistic regressions to exmine ptterns of negtive helth outcomes by compring the FI screen with the HFSS. The independent vrible in ech set of models ws FI sttus (mesured by the FI screen or the HFSS). The dependent vribles were child s helth, number of hospitliztions in the child s life, child being t risk for underweight or overweight, child being t developmentl risk, cregiver s helth, nd cregiver s positive depression-screen result. All models were djusted for Children s Helth- Wtch site, cregivers rce/ethnicity, US-born versus immigrnt sttus, mritl sttus, eduction, employment, history of brestfeeding, child s gender, nd low birth weight. Covrites were chosen on the bsis of theoreticl nd bivrite ssocitions with both FI nd the outcomes. RESULTS Smple Of 41 669 cregivers pproched for recruitment, 37 805 (90.7%) were eligible to prticipte. Of them, 34 049 (90.1%) completed the interview. Eligibility criteri for this nlysis included complete dt for questions 1 nd 2 of the HFSS; 88% of the completed interviews collected between June 1998 e28
ARTICLES nd December 2008 were included in the finl nlyses (n 30 098). Tble 1 lists the smple ccording to FI sttus bsed on the HFSS; 23% were food insecure. Of the 7 Children s HelthWtch sites, Minnepolis hd the highest overll prevlence of FI, followed by the Boston site. Nerly 60% of the dt were collected from cregivers of child younger thn 12 months, nd there ws no difference in prevlence of FI sttus ccording to child s ge or gender. Compred with foodsecure households, higher proportion of children in food-insecure households were brestfed. A lower proportion of cregivers in foodinsecure households were younger thn 21 yers, born in the United Sttes, employed, nd hd high school diplom or college degree compred with cregivers in food-secure households. The mjority of cregivers interviewed were blck or Hispnic. A higher proportion of Hispnic cregivers compred with other ethnic groups were food insecure. Compred with cregivers in foodsecure households, cregivers in foodinsecure households were more likely to report their own helth s fir or poor, to hve positive depressionscreen result, nd to rte their child s helth s fir or poor (see Tble 1). In ddition, children in food-insecure households were more likely to be t developmentl risk nd to hve been hospitlized t lest once since birth. Compred with children in foodsecure households, fewer children in food-insecure households were t risk for underweight. No differences were found with respect to child overweight or low birth weight ccording to FI sttus. FI Screen Most respondents who lived in foodinsecure households nswered ffirmtively (often true or sometimes true TABLE 1 Smple Description According to FI Sttus (Determined by the 18-Item HFSS) (N 30 098 ) versus never true) to questions 1 nd 2 of the HFSS: 92.5% nd 81.9%, respectively. These questions sked (1) Within the pst 12 months we worried whether our food would run out before we got money to buy more nd (2) Within the pst 12 months the food we Food Secure (N 23 256), % Food Insecure (N 6842), % Site of dt collection.001 Bltimore 14.7 8.0 Boston 27.3 24.5 Little Rock 18.7 11.7 Minnepolis 21.9 40.6 Phildelphi 9.4 6.4 Los Angeles 5.9 5.3 Wshington, DC 2.0 3.6 Child predictor vribles Age.08 4 mo 26.3 26.7 4 12 mo 32.6 33.8 13 24 mo 25.9 25.3 25 36 mo 15.2 14.2 Gender.81 Femle 46.7 46.8 Mle 53.4 53.2 Low birth weight ( 2500 g) 14.2 13.8.41 Brestfed 50.9 66.1.001 Child outcome vribles At risk for underweight b 15.3 14.0.01 Overweight c 13.7 13.8.80 Child helth (fir/poor) 10.7 16.8.001 Number of lifetime hospitliztions 22.6 24.4.002 Developmentl risk d 12.4 18.0.001 Cregiver predictor vribles Birth mother 21 y of ge 21.7 14.5.001 Rce/ethnicity.001 Asin 1.5 0.9 Blck 56.6 43.5 Hispnic 26.0 45.4 White 14.9 9.3 Ntive Americn 1.0 0.9 Born in the United Sttes 72.2 47.6.001 Mrried/prtnered 37.7 44.3.001 Employed 41.8 32.9.001 Eduction.001 Some high school 33.6 43.6 High school grdute 41.0 36.2 College grdute 25.4 20.2 Cregiver outcome vribles Cregiver helth (fir/poor) 17.2 32.4.001 Cregiver positive depression-screen result 20.4 39.7.001 Limited to fmilies who were uninsured or receiving public insurnce. b At risk for underweight ws defined s weight for ge t 5th percentile or weight for height t 10th percentile. c Overweight ws defined s weight-for-length t 95th percentile for children younger thn 24 months nd BMI for ge t 85th percentile for children ged 24 months or older. If length dt were not vilble, weight for ge t 95th percentile ws used s proxy. d Developmentl risk ws determined by the PEDS ( 1 concern) only for children older thn 4 months, nd dt collection begn in 2004 (n 10 874). bought just didn t lst nd we didn t hve money to get more. Sensitivity nd Specificity Cross-tbultion tbles were generted for combintions of the first 2 questions of the HFSS to exmine sen- P PEDIATRICS Volume 126, Number 1, July 2010 e29
sitivity nd specificity. Four combintions were explored. An ffirmtive response to question 1 only or question 2 only of the HFSS provided sensitivity of 93% or 82% nd specificity of 85% or 95%, respectively. An ffirmtive response to both questions 1 nd 2 provided sensitivity of 78% nd specificity of 96%. An ffirmtive response to question 1 nd/or question 2 of the HFSS provided sensitivity of 97% nd specificity of 83% (Tble 2); therefore, these re the criteri tht comprise the FI screen. Risk for Negtive Helth Outcomes Adjusted logistic regression models were conducted by using both the HFSS nd the FI screen (seprtely) to exmine how FI sttus is relted to child nd cregiver helth outcomes while controlling for covrites (Tble 3). Compred with cregivers in foodsecure households, those in foodinsecure households (s mesured by the FI screen) were 1.56 times more likely to report their child s helth s fir or poor (djusted odds rtio [OR]: 1.56 [95% confidence intervl (CI): 1.44 1.68]; P.001), 1.99 times more likely to report their own helth s fir or poor (OR: 1.99 [95% CI: 1.86 2.13]; P.001), nd 2.76 times more likely to hve positive depression-screen result (OR: 2.76 [95% CI: 2.59 2.94]; P.001). Compred with those in foodsecure households, children from food-insecure households (s mesured by the FI screen) were 1.17 times more likely to hve hd hospitliztions in their lifetime (OR: 1.17 [95% CI: 1.10 1.24]; P.001) nd 1.6 times more likely to be t developmentl risk (OR: 1.60 [95% CI: 1.42 1.80]; P.001) (see Tble 3). These ssocitions re similr to, lthough slightly weker thn, the corresponding ssocitions with the 18-item HFSS, 5 7 which demonstrtes convergent vlidity of the FI screen s mesure of FI. Differences in child nthropometric indices (t risk for underweight or overweight) s detected by the 18-item TABLE 2 Cross-tbultion of Overlp Between the 18-Item HFSS nd the FI Screen in Identifying Food-Insecure Households Identified by the HFSS, n (%) Not Identified by the HFSS, n (%) Totl, n (%) Identified by the FI screen 6614 (97) 3977 (17) 10 591 (35) Not identified by the FI screen 228 (3) 19 279 (83) 19 507 (65) Totl 6842 (23) 23 256 (77) 30 098 (100) TABLE 3 Reltion Between FI Sttus on the HFSS nd on the 2-Item FI Screen With Child nd Cregiver Helth Outcomes (N 30 098) Food Secure HFSS FI Screen Food Insecure Food Food Insecure OR (95% CI) P Secure OR (95% CI) P Reported child helth (fir/poor) 1.0 1.73 (1.59 1.88).001 1.0 1.56 (1.44 1.68).001 Number of lifetime hospitliztions 1.0 1.19 (1.11 1.28).001 1.0 1.17 (1.10 1.24).001 At risk for underweight 1.0 0.96 (0.88 1.05).36 1.0 0.94 (0.87 1.01).09 Overweight 1.0 1.03 (0.94 1.12).56 1.0 0.98 (0.91 1.06).59 Developmentl risk 1.0 1.72 (1.51 1.97).001 1.0 1.60 (1.42 1.80).001 Cregiver helth (fir/poor) 1.0 2.29 (2.12 2.46).001 1.0 1.99 (1.86 2.13).001 Cregiver positive depression screen 1.0 3.13 (2.91 3.37).001 1.0 2.76 (2.59 2.94).001 Dt were djusted for site, rce/ethnicity, US-born mother versus immigrnt, mritl sttus, eduction, child gender, cregiver employment, brestfeeding, low birth weight, nd mternl ge. The smple ws limited to fmilies tht were uninsured or receiving public insurnce. Developmentl risk ws determined by the PEDS ( 1 concern) only for children older thn 4 months, nd dt collection begn in 2004 (n 10 874). HFSS or the FI screen were smll nd not sttisticlly significnt. To ssess whether the households identified s food insecure by the FI screen experienced risk despite clssifiction s food secure by the 18-item HFSS, nlyses were repeted mong those who were clssified s food secure on the bsis of the 18-item HFSS (N 23 256). The FI-screen results show ttenuted, but sttisticlly significnt, ssocitions with poor child nd cregiver helth outcomes. Cregivers clssified s food insecure by the FI screen but not the 18-item HFSS were 1.26 times more likely to report their child s helth s fir or poor (OR: 1.26 [95% CI: 1.12 1.40]; P.001), 1.41 times more likely to report their own helth s fir or poor (OR: 1.41 [95% CI: 1.28 1.56]; P.001), nd 1.88 times more likely to hve positive depression-screen result (OR: 1.88 [95% CI: 1.72 2.06]; P.001) compred with cregivers clssified s food secure by the FI screen. Children in these households were 1.11 times more likely to hve hd hospitliztions in their lifetime (OR: 1.11 [95% CI: 1.02 1.21]; P.001) nd 1.36 times more likely to be t developmentl risk (OR: 1.36 [95% CI: 1.15 1.61]; P.001) thn children identified s food secure by the FI screen in this subsmple of food-secure households (see Tble 4). DISCUSSION We used 2-item screen to identify fmilies of young children t risk for FI. The FI screen is brief, with high sensitivity, good specificity, nd convergent vlidity. A sensitivity of 97% indictes tht only 3% of fmilies who experienced FI were likely to be misclssified. With this highly sensitive screen, providers cn identify nerly ll children who lived in food-insecure fmilies. A specificity of 83% indictes tht 17% of fmilies who were food secure ccording to the HFSS were clssified e30
ARTICLES TABLE 4 Reltion Between FI Sttus on the 2-Item FI Screen nd Child nd Cregiver Helth Outcomes Among the Subset of Food-Secure Households on the HFSS (N 23 256) FI Screen Food Secure Food Insecure OR (95% CI) P Reported child helth (fir/poor) 1.0 1.26 (1.12 1.40).001 Number of lifetime hospitliztions 1.0 1.11 (1.02 1.21).01 At risk for underweight 1.0 0.90 (0.81 1.00).05 Overweight 1.0 0.95 (0.85 1.05).31 Developmentl risk 1.0 1.36 (1.15 1.61).001 Cregiver helth (fir/poor) 1.0 1.41 (1.28 1.56).001 Cregiver positive depression screen 1.0 1.88 (1.72 2.06).001 Dt were djusted for site, rce/ethnicity, US-born mother versus immigrnt, mritl sttus, eduction, child gender, cregiver employment, brestfeeding, low birth weight, nd mternl ge. The smple ws limited to fmilies tht were uninsured or receiving public insurnce. Developmentl risk ws determined by the PEDS ( 1 concern) only for children older thn 4 months, nd dt collection begn in 2004 (n 8497). s being t risk for FI by the screener. Results of 2 nlyses demonstrte tht households identified s t risk for FI were t incresed risk for dverse child nd cregiver helth outcomes compred with households identified s food secure by the FI screen. Regrdless of whether nlyses were conducted cross the entire smple or restricted to food-secure households bsed on the HFSS, households identified s t risk for FI by the FI screen were t incresed risk for negtive child nd cregiver helth outcomes, which suggests tht intervention, such s referrl to services, is wrrnted. In this smple, FI ws not ssocited with children s nthropometry, which suggests tht FI is often invisible mong young children becuse they my not pper undernourished (or overweight) yet still experience negtive helth nd developmentl outcomes. The FI screen cn be esily dministered in peditric offices, by clinicins or prctitioners working with young fmilies (ie, Deprtment of Socil Services, school systems, Supplementl Nutrition Progrm for Women, Infnts, nd Children [WIC], child cre progrms, etc), or by community groups to ssess individul nd community needs. The FI screen hs importnt clinicl implictions for ll prctitioners who work with very young children nd fmilies. By identifying interventions designed for fmilies identified s t risk for FI, prctitioners cn help fmilies identify resources. For exmple, in Bltimore, the City Helth Commissioner dvocted for widespred use of the FI screen nd developed Web site (www.hungryinbltimore.org) tht identifies resources including food bnks, food pntries, socil services, nd federlly funded nutrition progrms such s the Specil Nutrition Assessment Progrm (SNAP), formerly the Food Stmp Progrm, nd WIC. In ddition, the Minnesot Deprtment of Helth Fmily Home Visiting Progrm hs incorported the FI screen into their protocol long with referrls to finncil nd food resources. The FI screen is n efficient nd vlid wy to identify fmilies t risk for FI. For more comprehensive ssessment of FI, the 18-item HFSS should be dministered. There were limittions to this study. First, lthough the method used for identifying items to be included in the FI screen ws systemtic nd met the set criteri of this study, it ws not s precise s methods used in trditionl item-response theory, scientific pproch often used to crete shortened versions of lrge questionnires. Second, the dt used in these nlyses included lrge, multisite, clinicl smple of exclusively urbn, lowincome fmilies of very young children. Although there is strong link between poverty nd FI, the vlidity of the FI screen hs not been tested in popultion of vrying socioeconomic sttus, in rurl popultions, or in fmilies without young children. Further investigtions of the FI screen should be conducted in these popultions. Third, prticipnts responded to these questions s prt of lrger questionnire delivered by n interviewer with the ssurnce tht their responses would be confidentil nd not influence their medicl cre. It is not known to wht extent responses might hve differed if the questions were dministered in the context of clinicl interview by helth cre prctitioner. Additionl testing of the screen in clinicl settings is wrrnted. Finlly, s with ny self-report mesure, fmilies could hve intentionlly misrepresented themselves nd incorrectly reported FI. However, the sensitivity, specificity, nd convergent vlidity demonstrted suggest incresed vulnerbility mong children t risk for FI. To gurd ginst misrepresenttion, providers should incorporte the FI screen into other clinicl ssessments. CONCLUSIONS A 2-item FI screen for identifying fmilies t risk for FI ws developed nd demonstrted sensitivity, specificity, nd convergent vlidity. The FI screen quickly identifies households with young children t risk for FI, which enbles providers to trget services to meliorte the helth nd developmentl consequences ssocited with FI. ACKNOWLEDGMENT This reserch ws supported by unrestricted funding from the following sources: W.K. Kellogg Foundtion; MAZON: A Jewish Response to Hunger; PEDIATRICS Volume 126, Number 1, July 2010 e31
Gold Foundtion; Minnepolis Foundtion; Project Bred: The Wlk for Hunger; Sndpiper Foundtion; Anthony Spinzzol Foundtion; Dniel Pitino Foundtion; Cndle Foundtion; Wilson Foundtion; Abell Foundtion; Clneil Foundtion; Betrix Fox Auerbch donor-dvised fund of the Hrtford Foundtion (on the dvice of Jen Schiro Zvel nd Vnce Zvel); REFERENCES 1. Nord M, Andrews M, Crlson S. Household food security in the United Sttes, 2008. Avilble t: www.ers.usd.gov/ Publictions/ERR83. Accessed My 10, 2010 2. Nord M, Andrews M, Crlson S. Household food security in the United Sttes, 2007. Avilble t: www.ers.usd.gov/ Publictions/ERR66. Accessed My 10, 2010 3. Alimo K, Olson CM, Frongillo EA Jr, Briefel RR. Food insufficiency, fmily income, nd helth in US preschool nd school-ged children. Am J Public Helth. 2001;91(5): 781 786 4. Cook JT, Frnk DA. Food security, poverty, nd humn development in the United Sttes. Ann N Y Acd Sci. 2008;1136:193 209 5. Cook JT, Frnk DA, Berkowitz C, et l. Food insecurity is ssocited with dverse helth outcomes mong humn infnts nd toddlers. J Nutr. 2004;134(6):1432 1438 6. Rose-Jcobs R, Blck MM, Csey PH, et l. Household food insecurity: ssocitions with t-risk infnt nd toddler development. Peditrics. 2008;121(1):65 72 7. Csey P, Goolsby S, Berkowitz C, et l; Children s Sentinel Nutritionl Assessment Progrm Study Group. Mternl depression, chnging public ssistnce, food security, nd child helth sttus. Peditrics. 2004; 113(2):298 304 8. Nord M, Hopwood H. Recent dvnces provide improved tools for mesuring children s food security. J Nutr. 2007;137(3): 533 536 9. Bickel G, Nord M, Price C, Hmilton W, Cook J. Guide to Mesuring Household Food Security: Revised 2000. Alexndri, VA: US Deprtment of Agriculture, Food nd Nutrition Service; 2000. Avilble t: Susn Schiro nd Peter Mnus; Eos Foundtion; Endurnce Fund; Gryphon Fund; Shoffer Foundtion; Annie E. Csey Foundtion; nd nonymous donors. We would like to thnk Dr Joshu Shrfstein for his collbortion nd contribution to this project s Helth Commissioner for the City of Bltimore. We lso cknowledge Zhoyn www.fns.usd.gov/fsec/files/fsguide.pdf. Accessed My 4, 2010 10. Blumberg SJ, Bilostosky K, Hmilton WL, Briefel RR. The effectiveness of short form of the Household Food Security Scle. Am J Public Helth. 1999;89(8):1231 1234 11. Kleinmn RE, Murphy JM, Wieneke KM, Desmond MS, Schiff A, Gpinski JA. Use of single-question screening tool to detect hunger in fmilies ttending neighborhood helth center. Ambul Peditr. 2007; 7(4):278 284 12. Murphy JM, Wehler CA, Pgno ME, Little M, Kleinmn RE, Jellinek MS. Reltionship between hunger nd psychosocil functioning in low-income Americn children. JAm Acd Child Adolesc Psychitry. 1998;37(2): 163 170 13. Weinreb L, Wehler C, Perloff J, et l. Hunger: its impct on children s helth nd mentl helth. Peditrics. 2002;110(4). Avilble t: www.peditrics.org/cgi/content/full/110/ 4/e41 14. Kleinmn RE, Murphy JM, Little M, et l. Hunger in children in the United Sttes: potentil behviorl nd emotionl correltes. Peditrics. 1998;101(1). Avilble t: www.peditrics.org/cgi/content/full/101/ 1/e3 15. US Deprtment of Helth nd Humn Services. Ntionl Helth nd Nutrition Exmintion Survey III Dt Collection Forms. Hyttsville, MD: US Deprtment of Helth nd Humn Services; 1991 16. Glscoe FP. Evidence-bsed pproch to developmentl nd behviourl surveillnce using prents concerns. Child Cre Helth Dev. 2000;26(2):137 149 17. Glscoe FP, Oberklid F, Dworkin PH, Trimm F. Brief pproches to educting ptients nd Yng for excellent mngement of surveillnce nd interview dt nd SAS progrmming; nd Kthleen Brrett, MSE, Jennifer Breux, MPH, Joni Geppert, MPH, Ktherine Joyce, MPH, nd Tu Qun, MPH for excellent trining, scheduling, nd supervising of interview stff members nd for diligence in coding, clening, nd prepring questionnires for dt entry. prents in primry cre. Peditrics. 1998; 101(6). Avilble t: www.peditrics. org/cgi/content/full/101/6/e10 18. Hlfon N, Regldo M, Sreen H, et l. Assessing development in the peditric office. Peditrics. 2004;113(6 suppl):1926 1933 19. Developmentl surveillnce nd screening of infnts nd young children. Peditrics. 2001;108(1):192 196 20. Glscoe FP. Erly detection of developmentl nd behviorl problems. Peditr Rev. 2000;21(8):272 279; quiz 280 21. Voigt RG, Johnson SK, Mellon MW, et l. Reltionship between prenting stress nd concerns identified by developmentl screening nd their effects on prentl medicl cre-seeking behvior. Clin Peditr (Phil). 2009;48(4):362 368 22. Glscoe FP. Collborting With Prents: Using Prents Evlution of Developmentl Sttus to Detect nd Address Developmentl nd Behviorl Problems. Nshville, TN: Ellsworth & Vndemeer Press, Ltd; 1998 23. Kuczmrski RJ, Ogden CL, Grummer-Strwn LM, et l. CDC growth chrts: United Sttes. Adv Dt. 2000;(314):1 27 24. Kemper KJ, Bbonis TR. Screening for mternl depression in peditric clinics. Am J Dis Child. 1992;146(7):876 878 25. Gordis L. Epidemiology. 2nd ed. Phildelphi, PA: W.B. Sunders; 2000 26. Rosnow RL, Rosenthl R. Beginning Behviorl Reserch: A Conceptul Primer. 3rd ed. Upper Sddle River, NJ: Prentice Hll, Inc; 1999 27. Trochim WMK. The reserch methods knowledge bse [current s of October 20, 2006]. Avilble t: www. socilreserchmethodsnet/kb. Accessed My 10, 2010 e32