Provider-Reported Illness and Absence Due to Illness Among Children Attending Child-Care Homes and Centers in San Diego, Calif FREE

DESPITE THE significant public health^1- 2 and economic impact^3- 5 of child-care–associated illness in the United States, there are major gaps in our knowledge of this topic. There have been few prospective studies on the incidence of illness and absence due to illness among children in the United States. Although infants are more likely to be enrolled in child-care homes than child-care centers,⁶ few studies have compared illness or absence due to illness among children in various types of child-care homes. Studies based on information from parents or medical records have generally found that children in child-care centers had a greater incidence of illness than those in child-care homes.^7- 10 However, studies based on information from providers in Seattle, Wash, child-care facilities found that children in child-care homes had a greater incidence of illness but lower incidence of absence due to illness than children in centers.¹¹ Herein we describe the incidence of illness and absence due to illness among children from small and large child-care homes and present additional evidence that the incidence of illness among children in child-care homes may be greater than that among children in centers.

The San Diego County Health Department and San Diego State University, San Diego, Calif, and Centers for Disease Control and Prevention (CDC), Atlanta, Ga, collaborated in conducting active surveillance for illness and injuries in out-of-home child-care facilities in San Diego from July 6, 1992, through January 28, 1994. The purpose of this surveillance was, in part, to determine trends and patterns in the incidence of illness and absence due to illness among children attending out-of-home child-care facilities within the city of San Diego. We compare the incidence of child-care provider–reported illness episodes and absence due to illness among children in different types of child-care facilities. We also describe the frequency of illness symptoms reported by providers, report the association of these symptoms with absence from child-care facilities, and compare information from telephone interviews with parents with information from providers.

Our sampling and recruiting methods have been described elsewhere.¹² Facilities were eligible to participate if they were located within the city of San Diego and were licensed to care for children younger than 5 years. We obtained lists of all eligible facilities from a local resource and referral agency and from the state licensing bureau. Facilities were stratified by licensed capacity and randomly selected for recruitment. Because of the large number of small child-care homes in our area, we selected a 20% sample of small child-care homes, a 62% sample of large child-care homes, and a 45% sample of child-care centers.

We sent to selected facilities letters describing the project and informing directors that their facility had been selected for participation. Directors were contacted by telephone and invited to participate in the project. Project staff visited facilities whose directors expressed an interest in participating and discussed the project and data collection forms. Additional visits were made to train child-care staff, collect baseline information on facility characteristics, and address other questions or concerns. Facilities were asked to participate for at least 12 months. Facilities that left the project were replaced by newly recruited facilities. These were randomly selected from the same strata as those they replaced.

Continued participation was encouraged through weekly telephone contact with each participating facility. Periodic newsletters were also used to inform providers of disease trends, project results, and items of interest to the child-care community. Monthly raffles with modest prizes were held for providers who reported on time for that month. Holiday greeting cards and small tokens of appreciation were also sent to child-care providers.

Data were collected from July 6, 1992, through January 28, 1994. Data collection forms for each class or group of children were mailed to participating facilities each week. These forms were preprinted with the names of the facility, group, and children. Providers were asked to record each child's daily attendance as present, absent, not scheduled to attend, or no longer enrolled and to record any reported visits to health care providers. We also asked providers to obtain information from parents about illness episodes during periods children were not scheduled to attend and during unexplained absences. Providers recorded signs and symptoms they observed directly and diagnoses and illness information reported to them by parents. They selected 1 or more codes for each sign or symptom from a list of coded definitions. These were based, in part, on national criteria for excluding children from child care¹³ and were expressed in terms familiar to most laypersons. These signs and symptoms included specific conditions such as fever, diarrhea, vomiting, headache, rash, runny nose, sore throat, swollen glands, cough, earache, difficulty breathing, mouth sores, mouth sores with drooling, pinkeye, pinkeye with pus, jaundice, head lice, and worms, as well as nonspecific conditions such as irritability, decreased activity, and decreased appetite.

Providers were asked to mail completed data collection forms to the project office at the end of each week. On receipt, the project staff reviewed these forms for completeness and reports of illness among children. Weekly telephone calls were made to directors of each participating facility to encourage participation and reporting, discuss completed forms, clarify ambiguities, obtain missing information, and collect additional information about illnesses among children during the preceding week. Computer-assisted data entry programs were used in transferring information from data collection forms into data files.

Facilities were considered to be unavailable during recruiting if they had stopped providing child care or were planning to stop within the upcoming year, if the initial letters were returned undeliverable, or if staff were unable to make initial contact with directors after 3 telephone calls. We categorized child care facilities according to definitions established by the American Academy of Pediatrics and American Public Health Association¹³ and the California Department of Social Services, Sacramento.Child-care homes were facilities maintained in private residences. Small child-care homes provided care for 6 or fewer children, no more than 2 of whom were infants, whereas large child-care homes provided care for 7 to 12 children, no more than 4 of whom were infants. Child-care centers were facilities maintained outside a private residence that provided care for any number of children.

Children were considered absent due to illness if they were absent on a day they were scheduled to attend and were reported as having 1 or more signs or symptoms of illness. An illness episode was defined as a sequence of 1 or more days a child was scheduled to attend and was reported as having 1 or more signs or symptoms of illness. Anillness episode in a previously ill child was considerednew if it was preceded by at least 3 days with no reported signs or symptoms of illness. Illness episodes were categorized on the basis of reported symptoms. Categories were not mutually exclusive. For example, an episode involving an earache and diarrhea would be counted as both. A child-week was defined as a period when a child attended a facility at least 1 day during the week or was regularly enrolled but absent due to illness. A child-year was the equivalent of 52 child-weeks. The dates of entry and exit were the dates the facility first and last provided information on the child, respectively. The number of weeks of participation was determined by dividing the number of days between entry and exit by 7. Children's ages were their ages at entry into the project. Dates of birth were obtained from facility records. Information on sex, race, and ethnicity were obtained from child-care providers. Multiracial children were categorized as "other."

We collected data from August 2 through December 31, 1993. Each week we randomly selected 16 children who had been reported by providers as having fever, vomiting, diarrhea, conjunctivitis, medically attended illness, or a notifiable disease (provider-reported ill children). We also randomly selected 16 children who had not been reported as having illness signs or symptoms (provider-reported well children). These were matched, by facility, with provider-reported ill children. We contacted parents of the provider-reported well children by telephone and asked if their child had been ill during the preceding week. We included children with any illness in pilot studies but later restricted eligibility to the selected groups, as parents had difficulty recalling less severe symptoms. For this survey, a child was considered ill if reported ill by a parent or the provider. We compared the proportions of ill children reported by providers among the 3 types of child-care facilities.

Results were expressed as incidence density rates. Standardization of rates by age and race was by the direct method,¹⁴ using all children in the study as the standard population. All rates in our report are unadjusted. Rates in different types of facilities were compared using the incidence density ratio; 95% confidence intervals of the incidence density ratio were calculated using the Mantel-Haenszel method. The significance of differences in proportions was determined using the Pearson χ² statistic. The significance of differences in means was determined using analysis of variance.

A total of 2784 facilities (2205 small child-care homes, 397 large child-care homes, and 182 child-care centers) were eligible to participate. Child-care homes were less likely to be available to participate (47.6%) than centers (93% [76/82]; P<.001); 40.0% of directors of available small child-care homes, 37.8% of directors of available large child-care homes, and 57% (43/76) of directors of available child-care centers agreed to participate (P=.02). Facilities whose directors refused to participate were less likely to operate on a not-for-profit basis and had been in operation for slightly less time than those whose directors agreed to participate. They also tended to be in lower-income neighborhoods.

Illness information was collected from 64 small child-care homes, 58 large child-care homes, and 41 child-care centers. In addition, 30 small child-care homes, 29 large child-care homes, and 10 child-care centers withdrew before providing information for 12 months. These include 8 facilities that were administratively dropped because of poor compliance. Directors of 26 facilities said they withdrew because of the amount of time needed or their workload; 21 directors (all from child-care homes) said that they had quit providing child care or left the area. Child-care homes were slightly more likely to drop out than child-care centers (P=.03). Small and large child-care homes and child-care centers that left the project participated for an average of 30.0 weeks, 29.4 weeks, and 24.4 weeks, respectively, before withdrawing. Within each facility type, those that left were comparable with those that remained with respect to profit status, licensed capacity, and length of time licensed. The small and large child-care homes that withdrew reported information on fewer average children than those facilities that remained in the study (7.3 vs 11.0 children from small homes [P<.001]; 12.3 vs 17.4 children from large homes [P=.001]).

We collected 113,446 child-weeks of information regarding 5360 children (Table 1). These children participated for a total of 137,388 weeks. The average child participated for 25.6 weeks, and 13.5% participated for more than 1 year. The difference between the number of weeks of participation and the number of child-weeks of data reflects missing information and periods such as holidays and vacations when children were not in attendance. Children in child-care centers contributed fewer child-weeks of information per child (19.8 child-weeks per child) than those in large (24.4 child-weeks per child) or small child-care homes (25.6 child-weeks per child;P<.001). Children in small child-care homes were significantly younger than those in large homes or centers. Children in centers were more likely to be of Hispanic, Asian or Pacific Islander, or unknown race or ethnicity than those in child-care homes.

Child-care providers reported 14,474 illness episodes (6.6 episodes per child-year) resulting in a total of 8593 days of absence (3.9 days absent due to illness per child-year). Whereas the incidence of illness episodes decreased with age, the incidence of absence due to illness appeared to be relatively constant except in children older than 5 years (Table 2). The incidence of illness episodes was lowest in children who were Asian or Pacific Islanders, followed by African American, Hispanic, and white children. Hispanic children had the highest incidence of absence due to illness, followed by African American, Asian or Pacific Islander, and white children. These patterns remained after adjusting for type of child-care facility.

The greatest incidence of illness episodes was among children in small child-care homes, followed by children in large child-care homes, with the lowest incidence among children in child-care centers. However, children in child-care centers had a significantly greater incidence of absence due to illness than those in small or large child-care homes. Absence due to illness among children in small child-care homes was not significantly greater than that among children from large child-care homes. Differences in the incidence of illness episodes and absence due to illness by type of facility remained significant after adjusting by age and race or ethnicity.

Most illness episodes were associated with respiratory symptoms (runny nose and cough), although episodes associated with fever accounted for the greatest number of days absent (Table 3). Illness episodes with a rash had the most days of absence per episode. The incidence of diarrhea among children younger than 3 years in child-care centers was 0.57 episodes per child-year, whereas that among children in child-care homes was 0.71 episodes per child-year (incidence density ratio, 1.24; 95% confidence interval, 1.05-1.45).

A sample of 239 children reported ill by providers and 292 well children from 79 different facilities was selected to check the accuracy of illness reporting. These represented 62.6% of the illness episodes reported from August 2 through December 31, 1993. Parents of 139 children (58.2%) reported ill by providers and 176 (60.3%) reported well by providers participated in this study. Parents of children in child-care centers were less likely to participate (43.3%) than those in large (64.2%) or small child-care homes (76.5%; χ²₂=45.91;P<.001). Providers in small and large child-care homes reported 91.9% (57 children) and 87.9% (29 children) of the children reported ill by parents or providers from those facilities, respectively, whereas child-care center providers reported 84.1% (53 children). These differences were not statistically significant (χ²₂=1.80;P=.41).

This study confirms recent reports of a higher incidence of illness and lower incidence of absence due to illness among children in child-care homes than among those in child-care centers¹¹ and suggests that the risk for illness among children in small child-care homes may be greater than that among children in large child-care homes. The incidence of absence due to illness between both these groups did not differ significantly.

Our observations concerning illness among children in child-care homes and centers differ from those of many earlier reports, which suggested that the risk for illness among children in child-care homes was less than or, at most, comparable to that of children in child-care centers.^7- 10 Reasons for the difference between our findings and those of previous studies include differences in child-care practices and in sources of information. Most studies have focused on illness among children in a well-defined geographic area such as Houston, Tex¹⁰; Phoenix, Ariz⁸; Pittsburgh, Pa¹⁵; or Seattle.¹¹ Three reports have been based on nationally representative data from the National Health Interview Survey.^7,16- 17 However, parents in 1 study¹⁶ were asked to estimate the number of days their children were home ill in bed during the past year, and results were subject to problems with recall as well as to bias due to differences in exclusion practices. Marx et al¹⁷ reported that immediate group size, rather than type of child-care setting, was an important intervening variable in the parent reports of frequent ear infections in children younger than 12 months. Our results are similar to those from studies in Seattle.¹¹ Also, studies of children in Pittsburgh found that, during the second and third study years, the proportion of children in group care (comparable to our child-care homes) who experienced at least 6 infections per year was slightly greater than that among children in child-care centers.¹⁵ Extrapolation to other areas should be made with caution, as differences in licensing regulations and child-care practices may influence results. Also, our definition of age as age at entry into the study may have resulted in some misclassification, especially among children who were in the study for more than 1 year. However, these were in the minority; the definition was applied uniformly to all children in the study, and this practice is not likely to have influenced our major findings. Previous reports were all based on data collected from 1981⁸ to 1989.¹⁵ These reports generated concern about the health and safety of out-of-home child care. Much of this concern focused on child-care centers. Findings of the more recent studies,¹¹ as well as our report, suggest that programs and changes in child care resulting from this concern may have made a difference.

With respect to information sources, 3 studies in the United States, including ours, have used information from child-care providers to compare the risk for illness among children in child-care centers with that among children in child-care homes. One study focused only on diarrheal illness.⁸ Our report and another¹¹ considered all illnesses. All 3 reported greater incidence of illness among children in child-care homes than among those in child-care centers. Scandinavian studies using provider data have reported comparable or slightly elevated risks for absence due to illness among children in child-care centers.^18- 20 However, absence due to illness may be a poor measure of illness,¹¹ and providers in the 3-family systems in Scandinavia receive much more training and support than do those in most child-care homes in the United States. Other studies comparing the risk for illness among children in various forms of child-care facilities in the United States have been based on information from parents^7- 8,15 or from health care records.¹⁰ We have suggested that information from these sources may be biased.¹¹ Although underreporting by child-care providers has been suggested as a possible explanation for the decreased risk for illness among children in child-care centers compared with that among children in child-care homes,¹¹ results of our validation study indicate that it is unlikely to explain all the differences between groups in our study.

Whereas there does not appear to be a single criterion standard for information on mild illness among children in out-of-home child care, the issue is essentially one of incomplete counting. We were recently made aware of a technique that formally addresses undercounting and may be of use in future studies of mild illness among children in child-care facilities. This "capture-recapture" survey method was originally used by field biologists to determine the number of animals in a given area and has been used to monitor injuries.²¹ Information from provider logs and telephone interviews with randomly selected parents could be combined to give more accurate counts than could be obtained from either source alone. Problems with parent recall could be reduced by asking about illness immediately after the week surveyed rather than a broader time, such as 2 or more weeks. Prospective studies of 1-year duration or longer occasionally demonstrate decreases in illness rates during the second year.^22- 23 Use of capture-recapture methods could help determine the extent to which these decreases were due to actual reductions in morbidity or changes in reporting practices.

Our results suggest that children in small child-care homes have a higher incidence of illness than those in large child-care homes. Underreporting is unlikely to explain these differences, and they may be real. These differences are contrary to reports of a direct association between facility size and risk for illness.^8,24- 25 However, these reports involved child-care centers or mixed groups of facilities. Also, this association has been attributed to an increased likelihood of introduction of infections into a facility²⁶ and assumes an equal likelihood of transmission once a pathogen is introduced. It is possible that the additional provider required in large child-care homes may provide sufficient supervision to reduce transmission once a pathogen is introduced.

We attribute much of the difference in the risk for absence due to illness between children in child-care homes and centers to differences in exclusion policies and practices. Exclusion policies among large and small child-care homes were comparable. However, as a group, child-care homes in this study had much more lenient policies with respect to accepting mildly ill children at the beginning of the day and calling parents to pick up children who became ill during the day than did child-care centers (CDC, unpublished data, January 1994). This observation agrees with reports from child-care providers of exclusion practices during a massive outbreak of waterborne cryptosporidiosis in Milwaukee, Wis.²⁷ We observed differences in illness and absence due to illness between different racial and ethnic groups of children. To the best of our knowledge, these differences have not been reported previously and remain after controlling for type of child-care facility.

The frequent occurrence of respiratory symptoms (runny noses, coughs, and earaches) in children in child-care settings has been reported elsewhere.^11,15 Episodes involving runny noses had the lowest number of days of absence per episode, and routine exclusion of children with mild respiratory infections is not recommended.¹³ However, these illnesses were responsible for much of the absence experienced by children in these facilities. This reflects the high incidence of these conditions in our study population. Few studies have addressed prevention and control of respiratory infections in child-care facilities, and much more work is needed in this area. At present, our best strategy for reducing the economic impact of these illnesses probably involves educating providers in appropriate exclusion policies and the importance of good hygiene regarding respiratory secretions. Although less common than episodes of respiratory symptoms, episodes with rash resulted in almost 2 days of absence per episode, the highest of any sign or symptom we monitored.

We believe facilities in our study were relatively representative of those in the San Diego area. The high proportion of child-care homes that were unavailable to participate reflects the rapid turnover of these facilities and inaccuracies of licensing records. Whereas facilities are required to register before starting to provide child care, they are not ordinarily required to give notification if they stop. As a result, these facilities remain on the rolls until time for license renewal. The transience of child-care homes is also 1 reason why the dropout rate in child-care homes in our study was greater than that of the centers. Anecdotal information suggests that facilities in less affluent areas may have been less likely to participate than those in other areas and may have been underrepresented in our study group. This is unlikely to have influenced our major findings. Others have cautioned about broad extrapolations of results from studies in single geographic areas, and our study is no exception. However, the illness results from this study are comparable with those from Seattle¹¹ and Phoenix,⁸ suggesting that comparative data from providers will generally demonstrate a greater risk for illness among children in child-care homes than among those in centers. Absence data presented herein and from Seattle¹¹ and Milwaukee²⁷ suggest that, due to differences in exclusion policies and practices, child-care centers will generally have higher absence rates than homes. There are exceptions to this, and conditions may well change in time.

To the best of our knowledge, this is the largest prospective study of illness and injury in child-care facilities conducted to date. Our results demonstrate the need to address the methods in future studies in this field. This need becomes more evident in light of the increasing interest in out-of-home child care. Information from these types of studies will be used to develop policy that will affect the lives of millions of children in the United States and elsewhere. Although previous studies focused attention on child-care centers, our results suggest that this approach may need to be reconsidered. Our results also suggest that, as much of the economic impact of child-care–associated illness is due to time lost from work, training in exclusion may have as great an economic impact as training in hygiene.

Accepted for publication August 24, 1998.

This work was supported in part by cooperative agreement U50/CCU 907166-01 with the Centers for Disease Control and Prevention, Atlanta, Ga.

We acknowledge Michele Ginsberg, MD, for her assistance in the planning, development, and conduct of these studies; the cooperation and assistance of the staff at the many child-care homes and centers that participated in this project; and J. Shaw for editorial assistance in preparing this manuscript.

Reprints: Ralph L. Cordell, PhD, Centers for Disease Control and Prevention, MS A07, 1600 Clifton Rd, Atlanta, GA 30333 (e-mail: rzc4@cdc.gov).

Editor's Note: This large, prospective study of illness in child-care facilities challenges the view that larger centers have the greater likelihood of illness. So which vision is closest to 20/20?—Catherine D. DeAngelis, MD