Child Care and Common Communicable Illnesses: Title and subTitle BreakResults From the National Institute of Child Health and Human Development Study of Early Child Care FREE

AS THE NUMBER of infants who are cared for by someone other than their parents has increased, so too has concern about the effect of nonparental child care on childhood illness. Children who attend child care arrangements, especially those enrolled in child care centers, are exposed more often to common communicable illnesses. Previous research indicates that they experience more bouts of respiratory illness, otitis media, and diarrhea than children reared exclusively at home.^{1 - 17}

The explanations most commonly offered for increased rates of illness among children in child care point to elevated levels of exposure to pathogens carried by other children in child care settings. For example, the primary factor connected with enteric tract illness seems to be the number of other young children present in the child care arrangement. The most vulnerable time for transmitted infections is immediately after entry into a new child care arrangement.^{4 ,18} Studies of upper and lower respiratory tract infections also point to increased exposure to other children (and the viruses they carry) as the primary risk factor for elevated rates of illness among children in child care settings.^{3 ,5 ,14 ,19}

To our knowledge, there have been few prospective studies of communicable illness in children in relation to the many types of child care arrangements children now attend. Most studies have been cross-sectional, have examined only a single category of illness, have focused on only 1 or 2 types of child care arrangements, and have failed to control for other family and community factors known to be related to illness. A critical issue yet unresolved is whether the elevated rate of illness associated with child care is related to children's cognitive, language, and social development.

We reexamine associations between child care and common childhood infections, using data from approximately 1200 children living in 10 different sites in the United States who are participating in the National Institute of Child Health and Human Development (NICHD) Study of Early Child Care.²⁰ We investigate 3 key issues concerning the relationship between child care arrangements and 3 common childhood infections: enteric tract illness (GI), upper respiratory tract infection (URI), and ear infections or otitis media (OM). This study addresses (1) the pattern of these 3 common illnesses during the first 3 years of life for children in different types of nonmaternal care; (2) possible effects of child care experiences on the frequency of illnesses; and (3) the possible effect on later development in such areas as language, achievement, and social behavior. We hypothesize that rates of illness will be higher for children who have contact with large numbers of other children in child care. We also hypothesize that these increased rates of illness will not negatively affect other areas of children's development.

Families participating in this study were recruited through mothers' hospital visits that were made shortly after the birth of a child during the calendar year 1991. Families lived in the vicinity of Little Rock, Ark; Irvine, Calif; Lawrence, Kan; Boston, Mass; Philadelphia, Pa; Pittsburgh, Pa; Charlottesville, Va; Morganton, NC; Seattle, Wash; or Madison, Wis. Of the 8986 mothers who gave birth during the sampling period, 5416 (60%) met the eligibility criteria. The criteria were as follows: a healthy mother older than 18 years and conversant in English, a singleton child whose birth was normal and uncomplicated, a family living in a neighborhood not considered extremely unsafe and located less than 1 hour from the research site, and a family not planning to move. One hundred thirty mothers refused to be interviewed (1%) and 308 refused to be contacted again (3%). Of the 5416 eligible families, 3015 (56%) were selected using a conditional random sampling plan that ensured the recruited families reflected economic, educational, and ethnic diversity. Of the 3015 selected for participation, 1526 (51%) agreed to participate. The remaining 1489 families could not participate for a variety of reasons (60 babies remained hospitalized for 7 days postpartum, 91 families planned to move, 512 could not be contacted, 641 refused, and 185 had other reasons [most said that they just didn't have the time]). Of the 1526 families who agreed to participate, 1364 (89%) completed the initial data collection visit and gave signed consent when the child was 1 month old. Comparisons were made between the 51% of eligible families who agreed to participate and the 49% who did not on several key demographic and child variables. There were a number of small (effect sizes were always <0.25) but statistically significant differences. Mothers who agreed to participate were approximately a year older on average (28.0 years vs 27.0 years), slightly better educated (65% with more than a high school degree vs 50%), and less likely to be minority (19% vs 24%); however, the participating mothers were no more likely to be married. On average, the children were a little heavier at birth (3490 g vs 3393 g).

Of the 1364 families who participated in the study, 16.7% were living in poverty (defined as a family income-to-needs ratio of 1.0 or less), 10.2% had mothers with less than a high school education (68.7% had some education beyond high school), and 76.5% were 2-parent households. White, non-Hispanic children constituted 76.4% of the cohort, 12.7% were black non-Hispanic, and 6.1% were Hispanic.

We obtained information from the mother about the parents and the child using face-to-face interviews when the child was 1, 6, 15, 24, and 36 months of age and from telephone interviews done when the child was 3, 9, 12, 18, 21, 27, 30, and 33 months of age. Specifically, we obtained information about the family context, the type of child care used, changes in child care arrangements, the number of children in the child care arrangement, the amount of time the child spent in child care, and the child's health status and illnesses. All participating children came to the study site when they were 15, 24, and 36 months of age so that they could be observed in a variety of structured and unstructured settings and be given a battery of developmental tests.

Four pieces of demographic information were used: mother's level of education, family size, presence of the father or another adult partner in the home, and family income (the income-needs ratio).²¹ Child characteristics included ethnicity and sex. The Home Observation for Measurement of the Environment (HOME) Inventory was administered during a home visit when the child was 15 months of age (infant-toddler version); the early childhood version, at 36 months. HOME assesses the quantity and quality of stimulation and support available to the child in the home environment.²²

During the interviews held every 3 months, mothers were asked: "Since the last interview, has [child] had an ear infection? . . . a respiratory illness? . . . a gastrointestinal illness?"

Child care information includes (1) type of care (center care, a child care home, care by a relative, or care in the child's own home by a nonrelative); (2) the total number of other children in all the child's nonmaternal care arrangements during a particular 3-month interval; (3) stability of care (the number of child care arrangements started during each 3-month interval); and (4) hours in care (how many hours the child spent on average each week in all forms of child care).

Children aged 3 years were assessed with the Bracken Basic Concept Scales (BBCS) and the Reynell Developmental Language Scales (RDLS). The BBCS²³ consists of the diagnostic scale and 2 screening tests, and is designed to assess a child's knowledge of basic concepts necessary for school readiness. The RDLS²⁴ is composed of two 67-item scales: verbal comprehension and expressive language. When the children were aged 3 years, mothers completed the 99-item Child Behavioral Checklist (CBCL).²⁵ For the children in nonmaternal care, child care providers also completed the CBCL when the children were 3 years of age.

To determine whether the prevalence of each illness varied throughout time and whether the illness varied as a function of background characteristics, family circumstances, and child care experiences, longitudinal logistic regression analyses were conducted using the generalized estimating equation approach.²⁶ To adjust for the dependency in the data caused by repeated measures, separate intercepts were estimated for each child. Time-invariant predictors (between-subject variables) included site, mother's education, ethnicity, and sex. Time-varying predictors included several measures of family, child, or child care experiences that were obtained at each of the 12 assessment points. These include child's age, child's age squared, type of interview (telephone or in-person), whether mother had a spouse or other adult partner present in the household, household size, average hours of all forms of child care during that 3-month assessment, whether the child changed child care arrangements during that period, whether the child attended a child care center during that period, whether the child attended a child care home during that period, whether the child was cared for by the father or grandparents or in the child's own home during that period, and the total number of children and adults summed across the various arrangements. The other 2 time-varying predictors were income (for children aged 1, 6, 15, 24, and 36 months) and HOME scores (children aged 15 and 36 months). Logistic regression models also included several interaction terms: age × each child care variable, and age × household size.

A multiple regression analysis was used to determine the relationship between child care experiences and rates of illness during the first 3 years of life and the child's developmental status at age 3 years. For each illness, a series of 5 regression analyses was run, 1 for each of the five 3-year child outcomes: Bracken²³ school readiness, Reynell²⁴ verbal comprehension, Reynell²⁴ expressive language, CBCL²⁵ externalizing problems, and CBCL internalizing problems.²⁵ Each regression model included background, family, and child care factors as predictors. The time-invariant predictors included site, mother's education, ethnicity, and sex. The time-varying predictors were represented by the mean score for each family on each measure. Similarly, child care hours was represented as the average number of hours a child spent in all types of nonmaternal child care per week from birth through 3 years based on all 12 data collection points. There were 3 types of care variables: (1) number of assessment periods the child was in center care; (2) number of periods the child was in a child care home; and (3) number of periods the child was cared for by relatives or at home by nonrelatives.

To determine whether a child's history of illness in child care effects the relation between experiences in care and developmental outcomes, we conducted a second series of regression analyses. Specifically, we added several statistical interaction terms involving the proportion of time a child had a particular illness and each child care experience factor.

The proportions of children who experienced episodes of OM, URI, and GI during each 3-month period through age 3 years were examined for the following 4 groups of children: (1) those receiving exclusive maternal care; (2) those in child care centers; (3) those in family child care homes; and (4) those cared for by a relative or by a nonrelative in the child's own home. Longitudinal logistic regression analyses indicated that the incidence of ear infections rises in the first year of life and peaks by age 1 year, then gradually declines during the next 2 years (Figure 1). Respiratory tract illness follows the same basic pattern, although the decline after age 1 year is much less pronounced (Figure 2). The rate of GI also rises in the first year of life, followed by a slow decline (Figure 3).

Table 1 displays results from the longitudinal logistic regression analyses of each illness, as performed on the whole sample. The table displays odds ratios with accompanying 95% confidence limits for each aspect of child care experience examined, controlling for the family background and child characteristics described above.

With 2 marginal exceptions, the number of hours per week children spent in child care had little to do with their likelihood of contracting a communicable illness. For OM, hours per week in care during the first year of life was related to the likelihood of acquiring the illness. For GI, hours per week during the third year of life increased the likelihood of contracting the illness.

Starting a new child care arrangement during an assessment period was associated with lower rates of OM and URI.

The number of children present in nonmaternal child care arrangements was related to the frequency of URIs. The number of children was a significant predictor of both GIs and URIs. However, because the type of child care arrangement was controlled in the analysis, the findings may offer conservative estimates of the effect of number of children present in child care on the probability of acquiring an illness. Children in center care were exposed to an average of 10.3 other children at age 3 months, and 16.4 children at age 36 months. This contrasts with the average number of other children present in a child care home (2.3 children at age 3 months, and 3.8 children at age 36 months).

Compared with children in exclusive maternal care, those attending child care centers and family child care homes were more likely to acquire OM and URI (Figure 2 and Figure 3). When examining rates of OM in these 2 settings, the effect of attending child care arrangements on rates of OM was much more pronounced during the child's first and second years of life than during the third year. Results showed higher rates of OM in center care during the first year of life followed by a gradual decline, so that by age 3 years, the rate among children in the 4 types of care were no longer different. The pattern for URI was similar, but the findings were not quite as strong. Results also showed that when children were being cared for by a relative, the risk of GI was lower in the first year of life but higher in the third year of life.

As Table 1 presents, we also computed the relative odds (odds of illness if the child was in particular type of care divided by the odds of illness if the child was not in that type of care) for the 3 illnesses at 3 ages: 12, 24, and 36 months. For OM, the odds ratio associated with center care at age 12 months was 2.37, but that declined to 1.22 by age 36 months (children in center care were more than twice as likely as home-reared children to contract an ear infection at age 1 year, but only 22% more likely by age 3 years). Likewise, for URI, the odds ratio associated with center care was 1.92 at 12 months, declining to 1.36 at 36 months. For GI, none of the odds ratios was greater than 1.20.

The final issue addressed by this study was whether the increased illness rate of children in child care settings is associated with poorer developmental and behavioral outcomes at age 36 months. Results from the 15 regression analyses run on the whole sample (3 illness categories × 5 outcome variables) provide no evidence of an association between the frequency of acquiring a communicable illness and either language competence (not significant) or school readiness (not significant). By contrast, children who experienced higher rates of URI (β = 2.64, SE = 0.81, P<.001) and GI (β = 4.64, SE = 0.81. P<.001) were described by their mothers as having a modestly higher rate of internalizing problems. Similarly, children who experienced higher rates of OM (β = 3.11, SE = 1.13, P<.006), URI (β = 4.73, SE = 1.14, P<.001), and GI (β = 6.04, SE = 1.15, P<.001) were reported by their mothers as developing slightly more externalizing problems. For children in child care, we reran the analyses using teacher reports of behavior problems. Results showed no relationship between children's rates of illness and teacher reports of behavior problems. The difference in results, depending on whether mothers or teachers reported on behavior problems, may reflect a common reporter bias on the part of mothers, since mothers were the only reporters on childhood illness.

There was little evidence that child care factors interact with illness histories to affect the course of a child's development. For each type of illness, the regression analyses were rerun, adding a block of 4 statistical interaction terms (the proportion of 3-month assessment periods during which the illness occurred × [1] the average number of hours spent in child care, [2] the proportion of assessment periods spent in a child care center, [3] the proportion of periods spent in a child care home, and [4] the proportion of periods spent in relatives' care or in-home care). The only significant interaction involved mother-reported internalizing problems (hours in child care × the illness rate for URI); even this interaction was not significant when the child care provider's report of internalizing problems was analyzed.

Finally, because approximately 10% of the sample had ear ventilation tube placements prior to age 3 years, the analyses relating a child's history of ear infections to developmental outcomes were rerun adding 2 variables to the model: (1) whether the child had ear tube placement, and (2) the interaction between ear tube placement and the number of assessment periods during which a child had an ear infection. Neither of these variables was significant for any of the 5 outcomes examined.

Findings from the NICHD Study of Early Child Care provide significant new details about how children's experience in nonmaternal care relates to common childhood infections during the course of the first 3 years of life. Children who were in nonmaternal care experienced a higher incidence of ear infections, URI, and GI illnesses during the first 2 years of life. However, by age 3, illness rates for children in child care were no different than those reported for children reared exclusively at home. Despite the increased rates of illness for children in nonmaternal care in the first 2 years of life, there was no evidence that elevated rates of illness were associated with poorer language or school readiness skills or elevated rates of behavior problems in child care at age 36 months. However, children who experienced more bouts of GI and URI illnesses were described by their mothers as having more internalizing problems.

Results of this comprehensive, longitudinal study confirm findings from cross-sectional studies with respect to the prevalence of ear infections, GI illnesses, and upper respiratory illnesses.^{27 - 28} For example, males were slightly more likely to manifest an ear infection in early infancy, and African Americans reported fewer ear infections than did members of other ethnic groups.^{1 ,4 - 5 ,12 ,28 - 31} The NICHD study also confirms that the likelihood of children acquiring common communicable illnesses during infancy is related to the type of child care they receive and to the number of other children in that care environment. This confirms findings from other studies of GI and diarrheal illnesses^{4 ,29 ,32} and otitis media.^{2 ,5 - 6 ,11 - 12 ,15 ,33 - 37}

In contrast to reports stating that the likelihood of acquiring communicable illnesses increases with the amount of time spent in child care,²⁹ we found little evidence that the number of hours of care per week resulted in increased illness rates for any of the 3 diseases examined. This finding, which contradicts results from some earlier studies, may have emerged because most of the children in this study spent more than 20 hours per week on average in nonmaternal care throughout the first 3 years of life. That is, most were in care for a sufficient amount of time to permit exposure to the other children in the setting.

Rates of infection for each illness studied rose during the first year of life, peaked in the second year, and then gradually declined. These data suggest that spending time in nonmaternal care may accelerate immunological responses to the pathogens that cause these illnesses. Although this study does not provide specific evidence of increased resistance to communicable illnesses by age 3 years because of early entry into child care, it may be that the children with more extensive early child care experience will show lower rates of communicable illnesses during kindergarten and first grade than children with no prior child care experience.³⁸ Many home-reared children may be exposed to the pathogens that cause common communicable illnesses for the first time when they enter preschool or kindergarten, and they will not yet have developed an immumological resistance to such pathogens. It would be recommended to follow up the NICHD cohort until school entry.

Results clearly point to increased contagion as the primary reason for the frequency of each of the illnesses studied. Exposure to other children in nonmaternal child care arrangements increases the likelihood of contracting communicable illnesses, especially during infancy. Up to a threshold of perhaps 8 to 10 children, the greater the number of other children and the greater the amount of exposure, the greater the likelihood of contracting an illness. The one anomaly in our results pertains to the finding that children who changed child care more often did not manifest higher rates of illness.

When analyses were done within specific types of child care arrangements, the number of children present was significant only for URI and only for child care homes and care by relatives. This suggests that the pathogens connected to GI are so common and so virulent that exposure to even a very small number of other children is sufficient to increase the probability of contracting GI illnesses. This corresponds to findings from other studies that suggest that the risk for contracting a GI infection is greatest immediately after entering a new child care arrangement.⁴ Our findings regarding the relationship between number of children in child care and URI are reminiscent of findings by Paradise et al.¹² They found a "strong positive relationship (for days with middle ear effusion) to the degree of exposure to other children"¹² (p323) in arrangements with more than 5 children. The fact that the number of children present within each type of child care arrangement was not a factor associated with the rate of ear infections suggests that the number of children typically found in child care homes and relative care is too low to make a difference, and that the number of children typically found in child care centers is higher than the threshold (in effect, more in the range of 8-10 children). For children in child care, the number of children present in the child care arrangement seems to contribute less to the risk of GI infection with time. By contrast, the association between family size and rate of GI illness increased with age for children reared exclusively at home perhaps because older siblings bring GI infections into the household from exposures at school or elsewhere.

This study was the first to prospectively examine the question of whether increased rates of illness for children who attend child care has a negative effect on children's later developmental status. This issue was examined for a range of developmental outcomes in a large, highly diverse sample. In this sample, there was limited evidence that higher rates of illness were associated with poorer developmental outcomes. The only significant associations found were between rates of illness and mothers' reports of behavior problems. However, higher rates of illness were not associated with child care providers' reports of behavior problems for children in child care. This discrepancy may reflect increases in parenting stress associated with caring for sick children and perhaps greater fussiness or sullenness on the part of children while they are at home sick. Because many child care providers have policies restricting attendance for sick children, child care providers may be spared some of the brunt of children's illnesses. Although our results showed several small relationships between rates of illness and behavior problems, the absence of significant interaction effects between amount and type of child care and illness histories on child outcomes indicates that the effect of child care experience per se on infection is largely unrelated to children's behavior, to their language development, or to their school readiness. In effect, for children who attend child care there is little evidence that having a greater number of common communicable illnesses such as URI, GI, and OM during infancy significantly alters the normal progression of behavioral development.

Results from this study clarify the relationship among children's experience with nonmaternal care, children's history of communicable illness, and later developmental outcomes. The study points to a single mechanism, contagion, as the primary reason for the increased rate of illness for children in child care. Furthermore, the results indicate that a single period of life, the first 2 years, is the only developmental period for which increased exposure leads to greater illness among children who enter care early in life. Finally, the study indicates that increased early rates of illness for children who attend child care arrangements do not seem to have any other adverse developmental consequences. Continued analysis of the illness rates of children in this sample as they enter school will provide useful information regarding whether early exposure to the pathogens that cause common communicable illnesses leads to a lower risk for illness after school entry.

Some caution should be exercised in applying the results from this study to children from high-risk families. The exclusion criteria used to select families for the study, coupled with slightly higher rates of participation among children from high socioeconomic backgrounds, limits the generalizability of the findings. Previous studies have shown that children from low SES families are more likely to receive lower-quality child care and child care that is more sporadic, factors that may increase their risk of exposure to communicable illnesses.³⁹

Accepted for publication November 17, 2000.

This study was directed by a Steering Committee and supported by the NICHD through cooperative agreement U10, which calls for a scientific collaboration between the grantees and the NICHD staff.

University of California, San Diego: Mark Appelbaum, PhD. University of London, London, England: Jay Belsky, PhD. University of Washington, Seattle: Cathryn Booth, PhD. University of Arkansas, Little Rock: Robert Bradley, PhD. University of Pittsburgh, Pittsburgh, Pa: Celia Brownell, PhD. University of North Carolina, Chapel Hill: Margaret Burchinal, PhD. University of Arkansas for Medical Sciences, Little Rock: Bettye Caldwell, PhD. University of Pittsburgh: Susan Campbell, PhD. University of California, Irvine: Allison Clarke-Stewart, PhD. University of North Carolina: Martha Cox, PhD. NICHD, Bethesda, Md: Sarah Friedman, PhD. Temple University, Philadelphia, Pa: Kathryn Hirsh-Pasek, PhD. University of Texas, Austin: Aletha Huston, PhD. Research Triangle Institute, Cary, NC: Bonnie Knoke, MS. Wellesley College, Wellesley, Mass: Nancy Marshall, PhD. Harvard University, Cambridge, Mass: Kathleen McCartney, PhD. University of Kansas, Kansas City: Marion O'Brien, PhD. NICHD: Mary Overpeck, DrPH. University of Texas at Dallas: Margaret Tresh Owen, PhD. University of Virginia, Charlottesville: Robert Pianta, PhD. Georgetown University, Washington, DC: Deborah Phillips, PhD. Children's National Medical Center, Washington, DC: Peter Scheidt, MD. University of Washington, Seattle: Susan Spieker, PhD. University of Wisconsin, Madison: Deborah Lowe Vandell, PhD. Research Triangle Institute: Kathleen Wallner-Allen, PhD. Temple University: Marsha Weinraub, PhD.

Corresponding author and reprints: Robert H. Bradley, PhD, Center for Applied Studies in Education, University of Arkansas at Little Rock, 2801 S University Ave, Little Rock, AR 72204 (e-mail: rhbradley@ualr.edu).