Psychology

A Longitudinal Analysis of the Efficacy of Environmental Interventions on Asthma-Related Quality of Life and Symptoms Among Children in Urban Public Housing

Description
A Longitudinal Analysis of the Efficacy of Environmental Interventions on Asthma-Related Quality of Life and Symptoms Among Children in Urban Public Housing
Categories
Published
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
   Journal of Asthma , 43:335–343, 2006Copyright  C   2006 Taylor & Francis Group, LLCISSN: 0277-0903 print / 1532-4303 onlineDOI: 10.1080/02770900600701408 ORIGINAL ARTICLE A Longitudinal Analysis of the Efficacy of Environmental Interventionson Asthma-Related Quality of Life and Symptoms Among Childrenin Urban Public Housing J. E. C LOUGHERTY , 1 , ∗ J. I. L EVY , 1 H. P. H YNES , 2 AND  J. D. S PENGLER 1 1  Harvard School of Public Health, Department of Environmental Health, Boston, Massachussetts, USA 2  Boston University School of Public Health, Department of Environmental Health, Boston, Massachussetts, USA In an environmental intervention study in public housing, we examined monthly Juniper Paediatric Asthma Quality of Life (QOL) Questionnairesfor 51 children. Longitudinal analysis and spline models were used to identify time periods with significant improvements in QOL to inform judgments about causality. We found significant improvements in QOL, with moderate improvements before environmental interventions, increasedrates of improvement immediately after, and reduced rates more than 5 months post-intervention. Effect modification analyses identified high-risk subpopulations and emphasized the importance of environmental, social, and economic conditions. Our results demonstrate the value of longitudinaltechniques in evaluating the benefits of environmental interventions for asthma. Keywords  childhood asthma, quality of life, symptoms, environmental interventions, indoor air I NTRODUCTION Severeasthmadisproportionatelyaffectsminoritychildrenin lower-income urban communities in the U.S. where com-plex environmental and social exposures may play a role inthe etiology and exacerbation of a highly multifactorial dis-ease. Asthma currently accounts for approximately 14.7 mil-lion missed school days and 11.8 million missed workdays,1.9 million emergency department visits, and 196,000 pedi-atric hospital admission per year (1). The dramatic increasein asthma prevalence over the past two decades (2) and thedisproportionate impact of severe asthma on inner-city chil-dren (3) indicate that genetics alone cannot be responsiblefor asthma etiology.A number of environmental, economic, and social factorshave been related to asthma exacerbation and may also af-fectbaselineimmunefunctionandresponsivenesstoenviron-mental triggers. Environmental exposures include allergenssuch as pollen (4), pets, dust mites, cockroach and rodentantigen (3), ambient air pollution including vehicle exhaust(5), household size, crowding and exposure to viral illness,environmental tobacco smoke (ETS) (6), and weather (7).Cockroachandmouseantigen,moreprevalentinlow-incomeurbanhousing,canactastriggersforasthmaandotheratopicdisease (8), and pesticides, used to combat infestation, mayinduceimmunedysregulation(9,10)andallergicdisease.So-cioeconomic factors including poverty (both individual andneighborhoodlevel),poornutritionandhousingstock,accessto healthcare and pharmacies, and regimen compliance caninfluence asthma patterns. In addition, sociobehavioral and ∗ Corresponding author: Jane E. Clougherty, Harvard School of Public Health, Department of Environmental Health, Landmark Cen-ter, Room 404-West, P.O. Box 15677, Boston, MA 02215; E-mail: jcloughe@hsph.harvard.edu indoorenvironmentalfactorsincludinganxiety,exercise,andhousehold smoking habits may play a role, and psychologi-cal stress (11, 12) and exposure to violence have been shownto directly alter immune function and inflammation (13–15).Theimmuneeffectsofsocialandpsychologicalstressorsmaycontribute to increased susceptibility within the same neigh-borhoods where physical environmental exposures may beelevated. The marked geographic and social patterning of asthma severity closely mirrors the distribution of a numberof these social and environmental exposures and may resultfrom multiplicative effects among them.Afewrecentstudieshaveexploredtheefficacyofallergen-reducing environmental interventions to alleviate childhoodasthma, and most have found some reduction in symptoms(16, 17). One study of preventative interventions (HEPAvacuuming, dust mite covers, carpet removal, and chemi-cal dust mite control) found reduced symptoms at one year(18),althoughonemeta-analysisfoundnoeffectofdustmitecontrol on symptoms (19). The National Cooperative Inner-City Asthma Study (NCICAS) found long-term reductionsinreportedsymptomsandsymptom-days,disruptionofcare-taker’s plans, lost sleep for child and caregiver, and missedschool days with behavioral interventions to reduce expo-sures to ETS, cockroach, mouse, mold, and pets (20). Someprior work examined varying symptom improvement rateswith interventions; NCICAS identified higher improvementrates during the first 2 months after initiating home inter-ventions. This study relied on 2-week reports of symptoms,unplanned doctors’ visits, and other endpoints, and detectedgreater improvements in the intervention group, during boththe intervention and follow-up year. In contrast, our studyuses formal longitudinal analysis and dense temporal datacollection to examine trends in quality of life and symptomimprovementandtounderstandinterventionefficacywithouta control group. 335    J   A  s   t   h  m  a   D  o  w  n   l  o  a   d  e   d   f  r  o  m    i  n   f  o  r  m  a   h  e  a   l   t   h  c  a  r  e .  c  o  m    b  y   U  n   i  v  e  r  s   i   t  y   O   f   P   i   t   t  s   b  u  r  g   h  o  n   0   5   /   2   8   /   1   3   F  o  r  p  e  r  s  o  n  a   l  u  s  e  o  n   l  y .  336  J. E. CLOUGHERTY ET AL. Few studies have examined whether in-home environ-mental interventions may differentially benefit some asth-matic individuals. NCICAS indicated several economic andsocial factors that may interfere with asthma managementand thereby modify intervention efficacy, including limitedasthma problem-solving skills, multiple asthma managers,child and adult adjustment problems, and higher life stress(11). However, NCICAS did not explicitly explore effectmodification in intervention compliance or efficacy by so-cial, medical, economic, or environmental factors.In general, intervention studies have not yet identifiedthe most effective environmental interventions for asthma,largely because indoor interventions tend to be bundled(i.e., integrated pest management includes intensive clean-ing and incorporates aspects of education and support). Fur-ther,moststudieshavefocusedondirectexposurereductionsfrom interventions but have not explicitly considered con-text, as we have done in this multifactorial investigation. Fi-nally,fewstudiesusedlongitudinalapproachestoinvestigatetime trends in health improvements around interventions,and no previous studies, to our knowledge, have exploredeffect modification of housing interventions for asthma indetail.The Boston Healthy Public Housing Initiative (HPHI), acollaborative community-based participatory research study,examined whether environmental interventions may be ef-fectively introduced into this setting to alleviate symptomsofpre-existingchildhoodasthma.Specifically,HPHIinvesti-gates whether in-home environmental interventions (includ-ing replacement of mattresses, industrial cleaning, and in-tegrated pest management) can reduce asthma exacerbationamong children 4 to 17 years of age living in Boston pub-lic housing and what factors may modify the efficacy of interventions.This paper examines the efficacy of a set of targeted in-home environmental interventions, using dense longitudinaldata collection and statistical methods to capture time trendsin respiratory symptom experience. Additionally, we exam-ine key factors in the social, physical, and medical environ-ment that may modify the efficacy of the interventions, po-tentially providing insight about both causality and meansof more effectively allocating intervention resources, frombothacostandhealthperspective,foroptimalenvironmentalimprovement and asthma alleviation.M ATERIALS AND  M ETHODS Between April 2002 and January 2003, 78 asthmatic chil-drenbetween4and17yearsofagewererecruitedfromthreepublic housing developments in Boston. Of those, 58 chil-dren in 44 households remained in the study to receive en-vironmental interventions, and 51 were retained through exitquestionnaires in August 2003. This analysis examines datafrom 520 Juniper Paediatric Asthma Quality of Life Ques-tionnaires (PAQLQ) from 51 participants, with 3 months ormore data capture per child before intervention, and 5 to 9months per child post-intervention. Recruiting efforts andenrollment criteria are detailed elsewhere (21).Given rolling recruitment and the inherent seasonalityof asthma, we chose to implement interventions in twophases. During fall 2002, 13 households in one development(Franklin Hill) received four environmental interventions:      Replacing the child’s mattress  with a microfiber mattressto prevent dust mite migration.      Industrial cleaning  to remove accumulated in-home aller-gens, and using caulking and expandable spray foam incracks to reduce pest entry.      Integrated pest management (IPM)  to reduce infestationwith minimal pesticide application. IPM visits included(1) initial assessment and trap placement, (2) recovery andcountingoftraps,(3)minimal,targetedpesticidetreatmentusing bubbletraps and gels, and (4) additional treatmentsas necessary.      In-home education about IPM  . IPM relies on participantcleaning behaviors. Residents were trained in basic pestandpesticideconceptsandmeasurestominimizeavailablefood, water, and shelter for pests. Storage bins, coveredgarbage cans, mops, and brooms were provided. ResidentswereencouragedtosubmitworkorderstotheBostonHous-ing Authority for holes in walls, torn window screens, andother repairs to reduce pest entry.During spring 2003, the remaining 31 households from 3developments received similar interventions. Radiators werecleaned in all homes, although industrial cleaning was moreintensive in homes with visible signs of roach infestation. Alimited asthma education/case management effort was pro-videdtoallparticipantsthroughoutthestudy,whereinaCom-munity Health Nurse created asthma action plans with thefamilies and provided peak flow meters. The nurse facili-tated some linkages with health care providers for childreninacutedistressbeforeenvironmentalinterventionsandeval-uated medications monthly to assess changes in respiratoryhealth and adherence to medication regimens. CommunityHealth Advocates (CHAs), members of the developments orsurroundingneighborhoodswhoweretrainedinbasicasthmaadvocacy and interviewing techniques, provided ongoingsupport and some asthma education. This limited case man-agement approach was designed to ensure that all childrenhadaccesstoroughlysimilarbaselinehealthcare,enablingusto better isolate the influence of environmental interventions.  Measures Standardizedhealthoutcomesquestionnaireswereadmin-istered monthly by the same CHA matched with particularfamilies, starting at intake and continuing through the post-intervention periods. Questionnaires were directly adminis-tered to older children, and caretakers responded for chil-dren under 8 years of age. Questionnaires included PAQLQfor both child and caregiver (22), questions on symptomfrequency that would allow for asthma severity classifica-tion according to National Heart, Lung, and Blood Insti-tute (NHLBI) guidelines, unplanned ED/clinic visits, missedschool/work days, and medication use changes. Here, we fo-cusonthechild’sPAQLQresults,whichprovidedthegreatestvariability in responses and broad representation of impactsof asthma on the children’s lives.The PAQLQ is a 23-item questionnaire that evaluatesasthma quality of life in three domains: Symptoms ( n  = 10),ActivityLimitation( n  = 5),andEmotionalFunction( n  = 8)(23).Thescoreforeachdomainistheaverageofalltheitemswithin it, and the symptom and activity limitation subscalesare reverse-scored. Scores range from 1 to 7, with 7 as the    J   A  s   t   h  m  a   D  o  w  n   l  o  a   d  e   d   f  r  o  m    i  n   f  o  r  m  a   h  e  a   l   t   h  c  a  r  e .  c  o  m    b  y   U  n   i  v  e  r  s   i   t  y   O   f   P   i   t   t  s   b  u  r  g   h  o  n   0   5   /   2   8   /   1   3   F  o  r  p  e  r  s  o  n  a   l  u  s  e  o  n   l  y .  LONGITUDINAL ANALYSIS OF INDOOR ENVIRONMENTAL INTERVENTIONS 337optimal score for any item or domain. A Total PAQLQ scoreis calculated as the mean score across all three domains, andthis total is used here as the key longitudinal outcome. In ad-dition, we repeat all significant results for the Total PAQLQscore for the Symptom domain.The PAQLQ is validated as both an evaluative (within-subject variation) and discriminative (between-subject varia-tion) instrument. The evaluative properties of the instrument,of most interest to our study, have been validated with paired t   tests of the sensitivity of the PAQLQ instrument to detectasthma changes due to treatment and natural fluctuations inasthmaovertime,suchasseasonality(  p  <  0 . 001),andtodif-ferentiate these groups from those who remained stable overtime (  p  <  0 . 0001). Intraclass correlations (ICC) also heldamong groups who remained stable over time (ICC = 0 . 95)(22). The scale has been validated for childhood popula-tions in several countries (24) and used to examine effectsof poverty, social support, age (25), obesity (26), and medi-cation changes (27) on asthma QOL, as well as the effect of QOL and emotional function on emergency department vis-its, school absenteeism, and doctors’ visits. An improvementof0.5pointsonthisscale,maintainingaconstantmedicationregimen, is shown to be clinically significant (22).With the exception of environmental sampling covariatesand allergy status, all effect modifiers were drawn from en-rollment questionnaires and are summarized elsewhere (21).This questionnaire, administered to the primary caretaker,included questions about symptom frequency, self-reportedallergies, perceived indoor environmental exposures, and so-cialstressindicatorsforthecaregiver:thefour-itemperceivedstress scale (PSS) (28), self-reported fear of violence, andthe My Child’s ETV scale (29). Environmental sampling forcockroach antigens, pesticides, and nitrogen dioxide (NO 2 )was performed in each home before and after environmentalinterventions. Allergy testing was performed once using theprick puncture method. Methods and results for environmen-tal sampling are detailed elsewhere (30, 31).  Analytic Methods Mixed models in SAS version 9.1 were used to examinePAQLQ changes over time and compare improvements inmultipletimeintervals(pre-intervention,upto5monthspost-intervention, greater than 5 months post-intervention). Knotpoints at intervention and 5 months post-intervention wereselected by AIC fit.The longitudinal model to examine overall improvementsover the course of the study is:Y ij  = β 0 + β 1 ∗ Time  j + e ij  (1)Thegenerallongitudinalmodeltoexamineoutcomesinmul-tiple periods is:Y ij  = β 0 + β 1 ∗ Time  j + β 2 ∗ Post ∗ Time  j + β 3 ∗ (GT5 MonthsPost)  j ∗ Time  j + e ij  (2)The general longitudinal model to examine effect modifica-tion is:Y ij  = β 0 + β 1 ∗ Group i + β 2 ∗ Time  j + β 3 Group i ∗ Time  j + e ij .  (3)WhereY ij  istotalPAQLQscore(continuous,wherei = hous-ingunitandj = time),Groupisbinaryforeacheffectmodifierexcept housing development, which is categorical. Time iscontinuous,centeredaroundthedateofinitialhomeinterven-tion (intensive cleaning); group differences in the interceptindicate differences at time of intervention. Post is binary,indicating whether each measurement occurs before or afterintervention, and GT5 Months is binary, indicating whetherthe measure occurred more than 5 months after the initialintervention date. Errors are assumed normally distributed.Reportedresultsreflecttherandominterceptmodelonly,withrandom slopes by household for exposure correlation withinseven sibling pairs in our study. Models were repeated withrandom slopes to confirm directionality of results, althoughsample size limits significance. Also due to power limita-tions, the effect modification model in Equation 3 excludesknot points, although again we evaluate the sensitivity of ourfindings to inclusion of these knot points.To evaluate effect modification, or whether interventionsweremorebeneficialforsomechildrenthanothers,providingsome insight towards causality, we considered the followingpotential modifiers individually:     Demographic variables : child’s age, sex, race/ethnicity,housing development     Individualhealthriskfactors: allergies to cockroach anddust mite antigen, NHLBI asthma severity classificationat enrollment, overweight, admitted to Neonatal IntensiveCare Unit (NICU) at birth, respirator at birth, under 5pounds birthweight, eczema     Socialfactors: perceivedstress,fearofviolence,restrictedoutdoor play due to parental fear of violence, social capital     Medical care indicators:  having a doctor to call otherthan the emergency department (ED), properly medicatedat study intake     Environmentalfactors: seasonofintervention,timespentaround smokers, smokers living in home, kitchen NO 2 concentrations, measured cockroach antigen reductions(kitchen Bla g 1 and Bla g 2), pesticide reductions (di-azinon,chlorpyrifos,permethrin,cyfluthrin,cypermethrin,and esfenvalerate).Reductions in kitchen concentrations of Bla g 1, Bla g 2,andsixpesticideanalytesweredichotomizedatthemediantoindicate relative reductions in the highest-concentration areaof the home. Reductions were calculated by comparing pre-intervention and average post-intervention concentrations.HomesweredichotomizedatmediankitchenNO 2  concentra-tions during heating season (median = 51 ppb). Medicationadequacy at baseline was assessed by physician review of reported prescriptions and asthma severity, as determined bysymptom frequency and spirometry.Effect modifiers were explored individually. In each case,the cohort was dichotomized by the modifier, and, due tosample size considerations, slopes were initially assumedconstant throughout intervention. We examined effects bygroup (indicating overall difference at time of intervention)and a group-by-time interaction term (indicating differingaverage improvement rates). Statistically significant modi-fiers were then examined against the PAQLQ Symptom sub-scale to assess objective health improvement. Lastly, relative    J   A  s   t   h  m  a   D  o  w  n   l  o  a   d  e   d   f  r  o  m    i  n   f  o  r  m  a   h  e  a   l   t   h  c  a  r  e .  c  o  m    b  y   U  n   i  v  e  r  s   i   t  y   O   f   P   i   t   t  s   b  u  r  g   h  o  n   0   5   /   2   8   /   1   3   F  o  r  p  e  r  s  o  n  a   l  u  s  e  o  n   l  y .  338  J. E. CLOUGHERTY ET AL. T ABLE  1.—Baseline demographic characteristics of asthmatic children in theenvironmental intervention cohort. West Franklin WashingtonBroadway Hill Beech Total Number of children 21 23 7 51Age: Percentage age 8 or older 67% 74% 86% 73%Sex: % Female 57% 48% 43% 51%Race/Ethnicity (%) ∗ Hispanic 81% 57% 86% 71%African-American 14% 43% 14% 27%Overweight 52% 70% 71% 57%Positive allergy test: Cockroach 52% 69% 50% 58%Dust mite 67% 62% 33% 60%(  Der f   1 or  Der p  1) ∗ Hispanic status and race were asked separately—total may be greater than 100%. improvement between the groups during each time period(pre-, early post-, late post-intervention) was compared foreach significant effect modifier, although power was limitedin this analysis. For this analysis, Equation 2 was appliedseparately to outcomes in each time period. Lastly, Equa-tion 1 was applied separately to each effect modifier group toexamine differences in slopes between time periods, withingroups.R ESULTS Amajorityofthechildrenwereeightyearsorolder(73%);and most were Hispanic (71%); 57% were overweight at in-take (BMI greater than 95th percentile by age). There was avery high prevalence of allergies in all three developments;over80%wereallergictooneormoreallergens,andapprox-imately 60% had positive skin prick tests to cockroach ordust mite. Demographic characteristics by development aredetailed in Table 1 and in our enrollment analysis (21).At enrollment, the mean Total PAQLQ score was 4.9(SD = 1.4),andtheSymptom,ActivityLimitation,andEmo-tional Function subscale means were 4.7 (SD = 1.5), 4.5(SD = 1.5), and 5.3 (SD = 1.4), respectively, across all de-velopments(Table2).Activitylimitationscoreswereslightlylower than for other domains, and scores in West Broadwaywere slightly lower than other developments (note: lowersymptom scores indicate more severe symptoms and lowerquality of life). A wide range of scores were indicated: from1.4 to 7 in Total PAQLQ.Overall, the participants showed significant improvementsin PAQLQ over the course of study in models withoutknot points. The overall improvement rate across all partici-pants and periods was 0.06 points per month (  p  <  0 . 0001).Improvements were comparable across the three PAQLQdomains.In models with knot points at intervention and 5 monthspost-intervention, we find significant improvement rates T ABLE  2.—Mean and standard deviation for PAQLQ domains at the start of thestudy, by development. Activity EmotionalSymptoms Limitation Function Total West Broadway 4.29 (1.65) 4.21 (1.81) 4.89 (1.62) 4.48 (1.58)Franklin Hill 5.09 (1.24) 4.75 (1.29) 5.51 (1.26) 5.17 (1.17)Washington Beech 4.59 (1.67) 4.51 (1.21) 5.70 (1.03) 4.95 (1.12)Total 4.69 (1.50) 4.50 (1.52) 5.28 (1.41) 4.86 (1.36)T ABLE  3.—Longitudinal analysis of PAQLQ scores with knot points placed atthe date of intervention and five months post-intervention. Estimate Intercept 5 . 14 (  p  < . 001)Slope pre-Intervention 0 . 09 (  p = . 07)Slope 0–5 months after intervention 0 . 16 (  p = . 39)Slope 5 to 9 months after intervention 0 . 04 (  p = . 001)  Note:  Slopes reported are actual slopes for each section of the curve,  p -values arereported for the difference between each subsequent slope. before interventions. Average improvement rates increase inthe 5 months following intervention, although this changein slope is not statistically significant (Table 3). More than5 months after interventions, improvements continue but ata significantly lesser rate. In only the first 5 months af-ter intervention, children showed average improvements of 0.80 points, greater than the 0.5-point improvement deter-mined to be clinically significant on the Juniper scale. Overthe 13 months of follow-up in this spline analysis, withperiod-specific improvement rates, the overall predicted im-provement is 1.32 points.Although the spline analysis provides some insight aboutthe relative influence of indoor environmental interventionsversus other factors (i.e., participation effect, social support,or case management), an evaluation of effect modificationmay address this issue in greater detail. Within our longi-tudinal model without knot points (Equation 2), we foundthat, at the point of intervention, boys had PAQLQ scoresthat were 0.64 points higher on average than girls but madecomparatively less improvement (Table 4). Significantly lessimprovement was also seen for overweight children, thosewhose caretakers reported high fear of violence in their com-munities, and children spending more time around smokers.Children allergic to cockroach also made less improvement(  p = 0 . 06).FactorsthatindicatedsignificantlylowerPAQLQat intervention included higher perceived stress by the care-giver, residence at Washington Beech, and age younger than8years.Thiseffectofagemaybeconfoundedbecauseparents T ABLE  4.—Statistically significant effect modification results for 51 children inlongitudinal analysis without knot points. Difference in intercept Difference in(at intervention) overall slope Demographics:Sex (male) (n = 25 male/26 female) 0 . 639 ∗ − 0 . 057 ∗∗ DevelopmentWest Broadway (n = 21, reference)Franklin Hill (n = 23) 0.070 0.033Washington Beech (n = 7)  − 0 . 886 ∗∗ − 0 . 003Age less than 8 years  − 0 . 826 ∗ − 0 . 007(yes = 14, no = 37)Individual Risk Factors:Positive roach allergy skin prick test(yes = 23, no = 17)  − 0 . 388  − 0 . 045 ∗ Overweight (yes = 29, no = 22)  − 0 . 082  − 0 . 050 ∗∗ Social Factors:High perceived stress (yes = 24, no = 26)  − 0 . 920 ∗∗ − 0 . 014Fear of violence (yes = 23, no = 26) 0.173  − 0 . 047 ∗∗ Environmental Factors:Time spent around smokers (yes = 24, 0.316  − 0 . 062 ∗∗ no = 24)Pesticides: Cyfluthrin (yes = 18, no = 14) 0.415  − 0 . 053 ∗∗ Significant:  p  <  0 . 1 ∗∗ Highly significant:  p  < . 05.    J   A  s   t   h  m  a   D  o  w  n   l  o  a   d  e   d   f  r  o  m    i  n   f  o  r  m  a   h  e  a   l   t   h  c  a  r  e .  c  o  m    b  y   U  n   i  v  e  r  s   i   t  y   O   f   P   i   t   t  s   b  u  r  g   h  o  n   0   5   /   2   8   /   1   3   F  o  r  p  e  r  s  o  n  a   l  u  s  e  o  n   l  y .  LONGITUDINAL ANALYSIS OF INDOOR ENVIRONMENTAL INTERVENTIONS 339 F IGURE  1.—Longitudinal spline model results for effect modifiers shown to be significant under non-spline analysis. answeredquestionnairesforchildrenyoungerthan8yearsof age.The largest overall between-groups difference at interven-tion was that children of families reporting higher levels of perceived stress had PAQLQ scores that were 0.92 pointslower at intervention (  p  <  0 . 01) and symptom scores thatwere 0.94 points lower, on average (  p  <  0 . 01).We explored additional analyses to better understand dif-ferences in improvement rates across time periods for dif-ferent groups, although our statistical power was limitedfor these assessments. First, we conducted a two-groupsspline analysis, applying Equation 2 by group. Our findingsgenerally corroborated the effect modification analyses inTable 4, with evidence that overweight children, those whospent time around smokers, and those with a high fear of violence had the most notable attenuation of benefits post-intervention(Figure1A–F),althoughnoneofthedifferencesin slope, including those for housing development, were sta-tistically significant. Alternatively, we can apply Equation 3within each of the defined time intervals to capture effectmodification in a model with no constraints on intercepts.In this model, the only statistically significant modifier wasperceived stress; children in higher-stress families made lessimprovement during both the early-post and late-post peri-ods. However, the limited number of observations per childper time interval impairs interpretability of this finding.No effect modification was seen for any medical care in-dicators, including access to medical care (having a doctorto call other than the ED), or pre-existing medical condition(eczema, low birthweight, being in NICU at birth, properlymedicated). No effect modification was found by cockroachreductions, NO 2 , or symptom severity. The only significant    J   A  s   t   h  m  a   D  o  w  n   l  o  a   d  e   d   f  r  o  m    i  n   f  o  r  m  a   h  e  a   l   t   h  c  a  r  e .  c  o  m    b  y   U  n   i  v  e  r  s   i   t  y   O   f   P   i   t   t  s   b  u  r  g   h  o  n   0   5   /   2   8   /   1   3   F  o  r  p  e  r  s  o  n  a   l  u  s  e  o  n   l  y .
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks