Injury Prevention 1997; 3: 267 - 271

Injuries among children in home and out-of-homecare

J B Kotch, V M Dufort, P Stewart, J Fieberg, M McMurray, S O'Brien, E M Ngui,M Brennan

AbstractAs the number of children receiving carein out-of-home settings increases in theUnited States, the risk of injury in suchsettings has become the subject of intenseresearch.

Objectives-This study examined the re-lative safety of out-of-home care com-pared with care in a child's own home.

Methods-This community based pro-spective cohort study of 656 families inthree adjacent counties in the Piedmontregion ofNorth Carolina characterizes thepatterns and rates of injuries amongchildren less than 5 years of age in threechild care settings, home care (HC),center based care (CBC), and other out-of-home care (OOHC). Information aboutminor and severe injuries was obtainedfrom parents using monthly telephoneinterviews over a one year period. Statis-tical modeling designed to handle unba-lanced data with correlated observationswas used as the primary tool for analysis.Results-Rate of minor injuries was high-est in CBC, foilowed by HC, and thenOOHC. However, these differences forOOHC may have been due to reportingbiases and errors in rate estimates. Therewere no significant differences in severeinjury rates among the three settings.Conclusions-The risk of serious injuryamong children under 5 in CBC is notdifferent from that of children in HC orOOHC despite the fact that the risk ofminor injury is higher.(Injury Prevention 1997; 3: 267 - 271)

Keywords: child care; home care; out-of-home care;center based day care.

As the number of working mothers with youngchildren continues to increase world wide, asignificant proportion of infants (under 1 yearof age), toddlers (ages 1 - 3), and preschoolers(ages 3-6) attend out-of-home care. In theUnited States, nearly 60% of mothers withchildren under 6 years of age work either full orpart time.' In addition, a recent national surveyfound that 40% of mothers with infants and52% with toddlers are working either full orpart time.2 Another study found that 60% ofchildren 0 to 5 were attending out-of-homecare regularly (V Cain, DA Dawson; paperpresented at American Public Health Associa-tion meeting, Chicago, 1989). Furthermore,

the proportion of infants in child care centershas quadrupled between 1976 and 1990, andthe proportion of toddlers has almost doubled.2The large cohort of children receiving out-of-home care has led researchers to investigate therisk of infectious diseases and injury.34

Since out-of-home care for young children isa growing trend, and injury in this age group isa significant cause of morbidity and mortality,the safety in these settings has been of concern.Some investigators have looked at the epide-miology of injuries at child care centers.5-"Others have studied hazards or the adherenceto safety regulations in child care centers orfamily child care homes.12-14 Still others havesought to obtain a comparison of the rate ofinjuries in out-of-home care compared withcare in a child's own home in an attempt toanswer the question 'Is out-of-home care assafe as own-home care?'15- 7 Two of thesestudies suggested that center based care wassafer than home care,8-'5 whereas two studiesprovide evidence that the rates of child careand home care injuries are not significantlydifferent." 18The objective of this study was to compare

injuries in three child care settings: (1) homecare (HC), defined as care provided in a child'sown home; (2) center based care (CBC),defined as child care services provided by aninstitution to six or more unrelated preschoolchildren for four or more hours per day,including organized child care centers, nurseryschools, Head Start centers, and preschools;and (3) other out-of-home care (OOHC),which included family child care homes (childcare centers that provide services to fewer thansix unrelated preschool children), babysitters,and any informal kind of child care provided ina place other than a child's own home.

Study design and methodsSTUDY DESIGN AND POPULATIONThis community based prospective studyobtained injury information from a sample offamilies with children less than 5 years of ageliving in three adjacent counties in the Pied-mont area of North Carolina. Households withtelephones and at least one resident childunder the age of 5 years were selected usingrandom digit dialing. For the initial screeninginterview, at least four attempts were made atevery sampled number to complete the inter-view. The calls were staggered over times of theday and days of the week to maximize thechances of making a contact. A total of 807families was identified.Of the original 807 eligible families, only 80

University of NorthCarolina at ChapelHill: Department ofMaternal and ChildHealthJB KotchS O'BrienEM NguiM Brennan

Departments ofMaternal and ChildHealth andEpidemiologyVM Dufort

Department ofBiostatisticsP StewartJ Fieberg

Frank PorterGraham ChildDevelopment CenterM McMurray

Correspondence to:Dr Jonathan B Kotch,Department of Maternaland Child Health, CB#7400 Rosenau Hall,University of NorthCarolina, Chapel Hill, NC27599-7400, USA.

267

on July 10, 2021 by guest. Protected by copyright.

http://injuryprevention.bmj.com

/Inj P

rev: first published as 10.1136/ip.3.4.267 on 1 Decem

ber 1997. Dow

nloaded from

http://injuryprevention.bmj.com/

Kotch, Dufort, Stewart, Fieberg, McMurray, O'Brien, Ngui, Brennan

(10%) refused to participate. Thirty three (4%)phones were no longer in service at the time ofthe first call, four families were not interviewedbecause of language or hearing difficulties, and46 (6%) could not be reached. A further 27(3%) were deemed ineligible at the time of thefirst interview contact because of either childdisability or children no longer within theeligible age range.

Recruitment of 63 more families was at-tempted in December 1994. Five of therandomly selected numbers were businessnumbers. Of the remaining 58 families, 47agreed to participate, increasing the potentialsample to 664. Another eight families werelost, leaving 656 subjects who completed oneor more of the monthly data collection calls.

DATA COLLECTIONAn initial parent/guardian interview instrumentwas designed to establish baseline demographicand child care data. A second interview instru-ment was designed for use once a month for thenext 12 months. Each month, a calendar wassent to respondents to record all injuries and toremind each of them of the coming phone call.The survey instruments were augmented byquestions from the National Health InterviewSurvey and its 1988 Child Health Supplement(CHS).'9 The injury questions used for thestudy conformed to the CHS. For eachreported injury, information was collected onseverity, type, cause, body part affected, carerequired, intentionality, day, time, and place ofoccurrence. Information on types of child careand hours ofcare was collected at the beginningof the study and updated whenever changesoccurred. Information was also collected onage, gender, and race of the child, caretaker'seducation, housing density, and travel time toand from place of care.

DATA VALIDATIONThe name of place of care was checked againststate regulatory agency lists of registered childcare facilities. Some discrepancies were noticedbetween reported place of injury and place ofcare. Whenever possible, these differenceswere resolved through follow up phone calls.If the discrepancy was not resolvable, the casewas dropped from the analysis. As well, 57injuries with 'questionable' dates that were notresolvable had to be dropped from the analysis.Body part injured, type, and cause of injury

were coded directly by the interviewers from apredetermined list. If injury information couldnot be coded, narrative information wasrecorded.

CALCULATION OF EXPOSURE AND OUTCOMEInjury rates were calculated for the time periodbetween 6:30 am and 5:29 pm, Mondaythrough Friday. These rates were calculatedby dividing the number of injuries in aparticular setting by the number of hours ofexposure. Hours of exposure for OOHC andCBC were obtained from the monthly tele-

phone interviews. Hours spent at home werecalculated as the time not engaged in eitherout-of-home child care or other activities of thehome. The time spent in other activitiesoutside of the home was estimated by callinga random sample of study parents at the end ofthe study. Parents reported that, when theirchildren were not in child care, over 75% oftheir time was spent at home.The total number of child care exposure

hours between interviews was determined fromthe total number of weeks between completedinterviews and the average hours of exposureper week. Injury rates were then calculated bydividing the number of injuries reported for agiven setting by the number of hours ofexposure in that setting since the previousinterview.

For all analyses, gender, age, householddensity (people per room) and primary care-taker's education (usually mother: years ofschool completed) were examined as potentialconfounders.

Injuries were classified as either severe orminor. Severe injuries were defined as anyinjury for which a health care provider wasvisited and/or the child's normal activity waslimited for four or more hours. Minor injuiresincluded any injury that required only first aidwithout any direct (non-telephone) assistancefrom a health care provider, and the child'snormal activity resumed in fewer than fourhours.

STATISTICAL ANALYSISLog linear statistical modeling, designed tohandle unbalanced data with correlated ob-servations, was used as the primary analytictool.

In addressing the research question, theregression equation included care setting,severity, sex, child's age, caretaker's years ofcompleted education, and housing density(persons per room in the child's home). Thismodel was used to obtain sex specific estimatesof the mean rates of severe and minor injury foreach of the three settings, with adjustment tothe average levels of age, education, and housingdensity. Caretaker's education and housingdensity were chosen as indicators of socio-economic status.

ResultsPROFILES OF CHILD CARE HOURSWhen the children were classified by child carestatus at the beginning of the study, 35% (226)had only HC, while 65% (429) had some out-of-home care arrangement. Of these 429children, 25% (108) spent fewer than 20 hoursa week in out-of-home care settings comparedwith 75% (321) who spent 20 or more hours aweek in such settings. The models used in thisstudy allowed for changes in child care statusthroughout the study period, and used timespent in the specified setting as the denomi-nator.

While only about one third of all childrenwere cared for exclusively in their own homes,

268

on July 10, 2021 by guest. Protected by copyright.

http://injuryprevention.bmj.com

/Inj P

rev: first published as 10.1136/ip.3.4.267 on 1 Decem

ber 1997. Dow

nloaded from

http://injuryprevention.bmj.com/

Injuries among children in home and out-of-home care

HC accounted for over 65% of the totaldaytime hours of exposure per week. Afterajusting for time spent traveling and inactivities other than child care, HC accountedfor 59% of the total exposure time, followedby CBC with 29%, and OOHC with 13%.

AGE AND SEXThree age groups were used in the analysis:< 1(101), 1 - < 3 (278), and > 3 years (277).Males and females were nearly equally repre-sented in the totals and within each caresetting.

RACEThe racial distribution of the total sample(70% white/23% African American/7% other)was approximately representative of the racialdistribution of North Carolina.20 However,there were fewer white children in child caresettings (63%) than there were among childrencared for exclusively at home (77%).

CAUSE OF INJURYAltogether 944 of the 971 reported injuries hadcomplete information on cause and place ofoccurrence. The most common cause in allchild care settings was falls (51%). Beingstruck by a blunt or sharp object was thesecond leading cause (32%), followed byhuman bites (8%). The distribution of injurycauses among care settings followed the samepattern; however, human bites were moreprevalent in CBC (61 of 72 bites), and therewere fewer injuries caused by sharp or piercingobjects in CBC.

TYPE OF INJURYCuts and scrapes were the most frequent injury(65%), followed by bruises (12%), swelling/bumps (12%), burns/reddened skin (3%), andsuperficial foreign bodies (2%). Incidents thatwere classified as 'no injury' included swallow-ing coins and choking on food and resulted inno visible damage. The distribution of injurytypes across care settings followed the samepatttern, although burns tended to occur morefrequently in HC settings (26 of 33 injuries).

BODY PART INJUREDOfthe 970 injuries where body part and place ofinjury were identified, 44% involved the headand neck, 24% the upper extremities, 28% thelower extremities, and 4% the torso. Head andneck injuries were relatively more common inOOHC and CBC than in HC, whereas injuriesto the lower extremities were more common inHC than in OOHC or CBC. Because of lownumbers, the distribution of OOHC injuriesshould be interpreted with caution.

CRUDE INJURY RATESA total of 971 injuries were reported asoccurring in all child care settings over the

study year, of which severe injuries comprised6%. Among the severe injuries, about 66%occurred at the children's own homes com-pared with 32% in CBC settings and none inOOHC. None of the reported injuries requiredhospitalization. Among the minor injuries,approximately 54% took place in the home,compared with 42% in CBC and 4% inOOHC settings. In all three settings, the crudeinjury rates among boys were higher than forgirls.A similar pattern of crude injury rates was

observed for severe injuries in HC and CBC.Rates among boys in HC were 5.4 injuries per100 000 child hours compared with 3.2 in-juries per 100 000 child hours among girls. InCBC the corresponding rates for boys were 6.0and 2.9 for girls.

INJURY RATES FROM THE LOG LINEAR MODELMinor injuriesFor minor injuries, the adjusted rates werehigher in CBC (64 injuries per 100 000 hours)compared with HC (45 injuries per 100 000hours) and OOHC (14 injuries per 100 000hours) (table). These differences were statisti-cally significant for all comparisons, even afteradjusting for multiple comparisons (p

Kotch, Dufort, Stewart, Fieberg, McMurray, O'Brien, Ngui, Brennan

cally significant results. Assuming a differenceon the order of 1.4 times the rate of injuriesbetween any two child care settings (as wasseen with the minor injuries between HC andCBC, for instance), one would have to analyze200 severe injuries to get 95% probability ofobtaining a significant result. At the rate of 1.5severe injuries per 100 000 child hours, thatwould mean following up more than 6600children over the course of one year. Thisestimate does not take into account stratifica-tion by age or controlling for covariates in themodels, which greatly increase the sample sizerequired. Although minor injuries by them-selves may not present significant healthconsequences, they may serve as sentinels forsevere injuries.Most of the work done previously on out-of-

home care injuries has only examined severeinjuries, ignoring minor injuries. Overpeck andKotch have shown that limiting countableinjury to those that are medically attendedmay miss up to 40% of injuries that haveimportant consequences, such as activity lim-itations.2' Furthermore, including minor in-juries would provide more completeinformation about the epidemiology of childcare injuries. Human bites, for example, are atype of injury whose true incidence may not berevealed if only severe injuries are studied."822Minor injuries can also serve as a marker ofpotentially hazardous situations.

Although more than one study has come tothe conclusion that out-of-home care is as safeand possibly safer than HC,8 15 the validity ofthese studies is limited by selection biases andstudy design. In addition, the relative safety ofchild care centers versus family child carehomes and informal child care arrangementshas received little attention. Family child carehomes are a distinctly different setting fromlarger, organized child care centers. Turnerand coworkers compared child care centersand family child care homes in NorthCarolina.23 They observed that injury rateswere twice as high in child care homes than inchild care centers, although more severeinjuries occurred at child care centers. How-ever, these conclusions are based on only 66injuries, of which 11 occurred in child carehomes. Gunn and associates also reported ahigher incidence of injuries for family childcare homes as an incidental finding. 16 Thisfinding was not tested for statistical signifi-cance and comprised only 24 injuries.While this study failed to find any statisti-

cally significant differences between severeinjury rates among the different child caresettings, this does not rule out the possibilitythat there were such differences. If the ratedifferences were close to 1.4, as with the minorinjuries, this study would not have had thepower to reveal these differences. However, itis not likely that severe injuries would havebeen lower in child care centers than inchildren's own homes, as earlier research hassuggested.

This study was the first to employ sophisti-cated statistical modeling to control for thenon-independence of multiple injuries. Failure

to do so could lead to an overestimation of theeffect of care setting on injury rates. Thesemodels also allowed controlling for potentialconfounders, that is, to take account ofinfluences of socioeconomic status, parent'seducation, child's age, and sex.Another strength of this study was that it was

able to capitalize on the information obtainedby identifying minor injuries. In other words,asking about minor injuries may have im-proved parents' recall of major injuries. Carewas taken to account for time spent outside ofthe different care settings, so the number ofexposure hours in own home care was notinflated. Failure to account for time spentoutside of the specific care settings would haveled to an underestimation of the actual rate ofhome injuries.One possible limitation of this study stems

from the way injuries were identified. Becauseparents were the sole source of this informa-tion, it is conceivable that minor injuriesoccurring outside of the home would beunder-reported. Validation of parental reportswith center records was not carried out.However, steps were taken to reduce theprobability of recall biases. On two occasionseach month, parents were reminded of thestudy. Parents were sent calendars each monthon which they were asked to record all injuries.They were also contacted monthly for a brieftelephone interview. While shorter interviewintervals may have improved the capture ofminor injuries, recall bias should not differen-tially affect home versus out-of-home injuryrates.

In the case of CBC, the higher rates ofreported minor injuries would lead us toconclude that under-reporting did not happen.However, the remarkably low injury rates forOOHC suggests that some injuries in thissetting may have been overlooked. Given thehigher level of organization in CBC, the factthat these centers are more closely regulated,and the imminent introduction of mandatorychild care injury reporting systems in NorthCarolina, it is possible that CBC centers wouldalready be more aware of, and responsive to, atracking system, thus reporting even minorinjuries to parents. In contrast, the OOHCsettings, given that they are largely unaffectedby the reporting regulations, would be lesslikely to record or recall injuries, or to reportthese injuries to the parents. A final possiblereason for the low rates in OOHC may be dueto the relatively low total exposure timeincreasing the unreliability of injury rateestimates and to less exposure to the largerpieces of playground equipment associatedwith institutional style child care.

In support of the argument that this studyreliably increased the capture of serious in-juries by using parental reports of all injuries,the distribution of injury causes is similar tothat of studies using other methods of injuryidentification. In this study, falls were theleading cause of injury, which is consistentwith other studies.9 17 Injuries caused by hu-man bites were more common in CBC than inother child care settings, and these findings are

270

on July 10, 2021 by guest. Protected by copyright.

http://injuryprevention.bmj.com

/Inj P

rev: first published as 10.1136/ip.3.4.267 on 1 Decem

ber 1997. Dow

nloaded from

http://injuryprevention.bmj.com/

Injuries among children in home and out-of-home care

similar to those of Kotch et al, who foundhuman bites to be the second most commoninjury cause in child care centers."8The distribution of injury types was also

consistent with other studies. The leadinginjury type was scrapes/cuts, followed byswelling/bumps, bruises, and bums. Cuts andbruises accounted for over 40% ofHC injuriesand 25% of injuries in out-of-home care in thestudy ofGunn et al,'6 and most bums occurredonly in HC, findings similar to ours.

However, when comparing rates of severeinjuries, differences become apparent betweenthis and other studies. Gunn and associates, forexample, found the rates of injury in out-of-home care to be 1.69 injuries per 100 000 childhours.'6 A study of medically attended injuriesin child care settings found rates of 1.56injuries per 100 000 child hours.24 In thisstudy, the risk of injury in CBC was 3.46injuries per 100 000 child hours for severeinjuries. This twofold higher rate of injurieswas certainly due, to a certain extent, to themore liberal definition of severe injuries. Astudy assessing injuries arriving at the emer-gency department would certainly miss anyinjury that had medical attention elsewhere, orwhich resulted in activity limitations but nomedical visits.

While the sampling strategy employed in thisstudy ensured a reasonable representation ofeligible families, it is limited by the exclusion offamilies without telephones. Poorer families,who couldn't afford telephones, would also beless likely to have access to child care outside ofthe home. However, because child injury ratesincrease as income decreases, this bias wouldhave led to relative underreporting of out-of-home injuries. In any event, since neithercaretaker's education nor housing density,both indicators of socioeconomic status, weresignificantly associated with injury rates in thisstudy, it is unlikely that including familieswithout telephones would have influenced thestudy's conclusions.

Implications for preventionThere is a need to pay continuing attention tothe safety of young children, both in child careand HC. Bums are a special concem forchildren at home, while falls from heightsneeds to be a focus for child care providers,particularly in organized child care centerswhere there is more and bigger playgroundequipment. Training of child care providersshould emphasize playground safety andknowledge of peak times of injury (afternoonsand late momings).Our report, along with other recent work such

as Kopjar and Wickizer's study, shows thatcomplacency with respect to child care safetyissues should be discouraged.'7 Safety in childcare is an ongoing concen. Health care provi-ders should be trained to follow up and keeprecords of child care injuries, and reliable childcare injury surveillance systems are needed.Training in injury prevention (in addition tofirst aid and cardiopulmonary resuscitation)should be required for all child care workers.

As changes come about in child careregulations, further research can focus on theeffects of changes in regulations on the typesand causes of injuries in different care settings.More detailed studies may also help inidentifying previously unknown or unsus-pected hazards. Finally, there is a need forevaluation of the impact of the regulation ofsafety hazards in child care and of other injurycontrol interventions on the frequency andseverity of injuries in preschool children.

Our thanks to Susan Alber, Betsy Crites, Laura Fish, MarciaHerman-Giddens, Jin-Whan Jung, Veronica Lawrence, LaurenWatterson, and Rick Williams.

This study was supported in part by grant #MCJ-370363from the Maternal and Child Health Research Program,Maternal and Health Bureau, Health Resources and ServicesAdministration, Public Health Service, United States Depart-ment of Health and Human Services.

1 Deitch SR, ed. Health in day care: a manual for healthprofessionals. Elk Grove Village, IL: American Academy ofPediatrics, 1987.

2 Willer B, Hofferth SI, Kisher EE, et al. The demand andsupply of child care in 1990: joint findings from thenational child care survey, 1990, and a profile of childcare settings. Washington, DC: National Association forthe Education of Young Children, 1990.

3 Haskins R, Kotch JB. Day care and illness: evidence, costand public policy. Pediatrics 1986; 77 (suppl 6) 951-82.

4 Kotch JB, Bryant D. Effects of day care on the health anddevelopment ofchildren. Curr Opin Pediatr 1990; 2: 883-94.

5 Landman P, Landman G. Accidental injuries in children inday care centers. Am J Dis Child 1987; 141: 292- 3.

6 Elardo R, Solomons HC, Snider BC. An analysis ofaccidents at a day care center. Am Jf Orthopsychiatty1987; 57: 60-5.

7 Chang A, Lugg MM, Nebedum A. Injuries among pre-school children enrolled in daycare centers. Pediatrics1989; 83: 272-7.

8 Sacks JJ, Smith DJ, Kaplan KM, Lambert DA, Sattin RW,Sikes RK. The epidemiology of injuries in Atlanta day-care centers.JAMA 1989; 262: 1641-5.

9 Briss PA, Sacks nJ, Addiss DG, Kresnow M, O'Neil J. Anationwide study of the risk of injury associated with daycare center attendance. Pediatrics 1994; 93: 364-8.

10 Alkon A, Genevro JL, Kaiser PJ, Tschann JM, Chesney M,Boyce T. Injuries in child-care centers: rates, severity, andetiology. Pediatrics 1994; 94: 1043-6.

11 Sellstrom E, Bremberg S, Chang A. Injuries in Swedish day-care centers. Pediatrics 1994; 94: 1033-6.

12 O'Connor MA, Boyle WE, O'Conner GT, Letellier R. Self-reported safety practices in child care facilities. Am PrlevMed 1992; 8: 14-8.

13 Wasserman RC, Dameron DO, Brozicevic MM, AronsonRA. Injury hazards in home day care. J Pediatr 1989; 114:591-3.

14 Browning KS, Runyan CW, Kotch JB. A statewide survey ofhazards in child care centers. Injury Prevention 1996; 2:202-7.

15 Rivara FP, DiGuiseppi C, Thompson RS, Calonge N. Riskof injury to children less than 5 years of age in day careversus home care settings. Pediatrics 1989; 84: 1011-6.

16 Gunn WJ, Pinsky PF, Sacks nI, Schonberger LB. Injuriesand poisoning in out-of-home child care and home care.Am JfDis Child 1991; 145: 779- 81.

17 Kopjar B, Wickizer T. How safe are day care centers? Daycare versus home injuries among children in Norway.Pediatrics 1996; 97: 43-7.

18 Kotch JB, Loda F, Harms T, Clifford RM, McMurray MA.Parents' report of home and out-of-home injuries amongchildren attending child care centers. Early Child Devel-opment and Care 1993; 96: 183-3.

19 US Department of Health and Human Services, NationalCenter for Health Statistics. National health interviewsurvey: child health supplement. US Department of Healthand Human Services, National Center for HealthStatistics [producer], 1987. Ann Arbor, MI: Inter-university Consortium for Political and Social Research[distributor], 1988.

20 US Bureau of the Census. Census ofpopulation and housing.Washington, DC: US Department of Commerce, 1990.

21 Overpeck MD, Kotch JB. The effect ofUS children's accessto care on medical attention for injuries. Am J PublicHealth 1995; 85: 402-4.

22 Garrard J, Leland N, Smith DK. Epidemiology of humanbites to children in a day-care centre. Am J7 Dis Child1988; 142: 643-50.

23 Tumer WT, Snow CW, Poteat GM. Accidental injuriesamong children in daycare centers and family day carehomes: briefreport. Children's Health Care 1993; 22:73- 9.

24 Rivara FP, Sacks nJ. Preventing injuries and improving theenvironment: injuries in child day care: an overview.Pediatrics 1994; 94: 1031-3.

271

on July 10, 2021 by guest. Protected by copyright.

http://injuryprevention.bmj.com

/Inj P

rev: first published as 10.1136/ip.3.4.267 on 1 Decem

ber 1997. Dow

nloaded from

http://injuryprevention.bmj.com/