Determinants of Health Care Use by Children in Rural Western North Carolina: Results From the Mountain Accessibility Project Survey
Objective. To assess determinants of health care visits among children in a 12-county region of western North Carolina representative of rural areas in the United States.
Methods. Households were randomly selected for surveys of household characteristics, health status, and health care use. Surveys were conducted June 1999 to January 2000 and were stratified for children younger than 5 years and 5 years and older. The number of health care visits in the year before the survey was used as the outcome measure. Weighted mean visits and associations of family demographic and child health variables with the number of visits were determined by ratio and multivariate survey regression methods.
Results. Among children who lived in rural Appalachian regions of North Carolina in 1999, 90% had either public or private insurance coverage. The mean number of visits per child was 5.7 (median: 2.6), and in each age group the number of visits in the previous year exceeded the recommended number of well-child visits. There were no apparent geographic access barriers to care in this population, in that increased distances to provider sites did not result in declining numbers of visits. For children younger than 5 years, the primary determinants of health care use during the previous year were age, insurance status, and household income. Infants had more visits than older, preschool children, and those with household incomes >$40 000 per year had 76% more visits than those with incomes <$20 000 per year. Children with public insurance, exclusively Medicaid in this population, had almost 4 times as many visits as uninsured children. Among the children and adolescents 5 through 17 years of age, health insurance status, household income, pain during the past month, and race were the primary determinants of health care use during the previous year. Those with public health insurance had 6 times more health care visits than uninsured children. Household incomes >$40 000 per year were associated with 2.5-fold increased health care visits, and those with household incomes between $20 000 and $40 000 per year had 2-fold increased health care visits, compared with those with household incomes <$20 000 per year. White children had almost twice as many visits in the past year as black children in this age group. Pain experienced during the past month, as perceived by the parent, also predicted the number of visits in the older age group.
Conclusions. This rural population seems to have reasonably good access to care overall. The key determinants of health care use among these rural children were similar to those found in urban and other populations in the United States and likely are universal: health insurance coverage, household income, and parent perceptions of their child’s pain. As in other populations, programs in rural areas that strengthen health insurance coverage and reduce poverty will have a direct impact on child health. Differential use of health care among white and black children, especially those 5 years and older, merits additional explanation.
An estimated 21% of children in the United States (approximately 16 million in 1996) live in rural areas.1 Larger proportions of children in the South and Midwest live in rural areas than in the Northeast and West. People who live in rural areas are health disadvantaged in several ways compared with their urban counterparts.2,3 These disadvantages include limited access to health care as a result of geographic barriers, such as time and distance to care sites, and availability of transportation. Even privately insured children in rural areas were less likely to have physician visits in the previous year than their urban/suburban counterparts in the 1990 National Health Interview Survey (NHIS).4 Health care for Americans who live in rural areas has been the subject of much policy discussion during the past 40 years, with the needs to increase access to services through public insurance programs and availability of services receiving the majority of the focus.2,5,6
In 1998, rural areas tended to have more family and general physicians but fewer physicians of all types than urban areas.7 Specialty health services for children and adults, both outpatient and inpatient, are less readily available to rural children than those who live in metropolitan areas. The end result is that rural-nonmetropolitan children are more likely to have unmet medical needs than children who live in metropolitan areas, although they may be at least as likely to have a regular source of care.1 Other child health status indicators such as use of dental care and teen smoking also are more negative in rural versus urban areas.
In the late 1990s, children who lived in nonmetropolitan (rural) areas of the United States had death rates, uninsured rates, and teenage pregnancy rates similar to those of inner-city metropolitan (urban) children, with both groups faring worse in these health measures as compared with children who lived in fringe counties of metropolitan areas (suburban).7 Immunization rates for rural children also have been found to be as low as those in urban populations.8 The most nonmetropolitan areas of the South have population poverty levels that are slightly higher than that of the most urban areas of the Northeast.7
Although much is known about the demographics and potential access to care issues of rural children, less is known about factors that affect their use of health care. A number of models to evaluate the most important determinants of the use of health care services have been developed,9–12 but these have been used primarily to evaluate health care use among urban populations. Models of health care use typically have evaluated the combined effects of health policy, population characteristics, characteristics of the delivery system, and customer satisfaction9,13 or the characteristics of availability, accessibility, accommodation, affordability, and acceptability to a particular population.14
Using the principles of these models, the Mountain Accessibility Project (MAP) was designed to explore hypotheses of health care use in rural areas through interview surveys of households in 12 counties in the Appalachian region of North Carolina. The data presented in this study used the household demographic and child questionnaire components of the MAP survey to assess determinants of the frequency of health care visits among rural children in this region.
Data for this analysis are based on surveys of 305 households with 1 or more children younger than 18 years collected as part of the MAP survey. Study interviews were conducted from June 24, 1999, through January 10, 2000, in 12 rural western North Carolina counties. A total of 1060 households were surveyed, with data collected from 1 adult and, where present, 1 child in each household. When >1 adult or child was present in the household, individuals were randomly selected for the survey. Data were collected by the Research Triangle Institute. Data collection procedures have been described in detail.15
The study region reflects the varied nature of rural areas in the United States (Fig 1). It includes very isolated communities (with long distances to even small, urban areas along winding roads that can be closed in inclement weather) along with other communities that are readily accessible to larger cities and amenities. There is increasing development with 4-lane highway access and the building of shopping malls and expensive summer homes. The degree of rurality of the 12 counties, as measured by Beale Code categories (combination of whether a county is in a metropolitan statistical area, size of any urban population, and adjacency of nonmetropolitan counties to metropolitan counties),16 encompasses the most rural 4 of the 10 categories, suggesting in aggregate good potential for generalizability of findings to other rural areas.
The survey addressed respondents’ health status (including presence of specific diseases and perceptions of pain and effects of health and illness on daily functions), health insurance coverage, medical care options, location of health care providers, use of home and traditional remedies, use of health care services, health prevention behavior, religious beliefs, location of daily activity venues, and degree of alienation. The child survey consisted of 2 components, 1 for children younger than 5 years and 1 for children 5 years and older, and included several scales from the Child Health Questionnaire.17 This study protocol was reviewed and approved by the Institutional Review Boards of the University of North Carolina at Chapel Hill, Wake Forest University School of Medicine, and the Research Triangle Institute.
Sample Selection and Weighting
A 3-stage sampling design was used, with the first stage dividing the 12-county region into 3788 geographic segments containing ≥15 single-dwelling units and then randomly selecting 265 segments. The second stage involved random selection of 1583 dwelling units to provide a final survey population, allowing for nonresponse, of approximately 100 black households and 960 other households. The final stage was random selection of 1 adult and 1 child (if present) from within a surveyed household. The household adult with the best knowledge of the selected child’s health care was interviewed for the child survey. Surveys were completed in 948 nonminority and 112 black households. The overall unweighted household response rate for all 12 counties was 83.8%.
Because sample members were selected with varying conditional probabilities at each stage (segments, dwellings, people), generation of unbiased estimates of population means and other analyses require that data from each respondent be weighted appropriately to reflect this variation in probability of selection. The overall unconditional probability of selection for each respondent was computed as the product of the conditional selection probabilities at each sampling stage with adjustments for nonresponse. The sampling weight was computed as the reciprocal of the resultant unconditional probability.
Geographic Access-to-Care Variables
Five variables that assessed aspects of geographic access to health care were assessed. Minimum distances from the dwelling to 1) the nearest practitioner and 2) the nearest health care facility were determined as separate variables, regardless of whether the family used this particular practitioner or facility. Additional spatial variables were determined on the basis of the locations of respondents’ homes and their work sites and where they went for shopping, visits, or recreational activities, as well as how often they went to these places. These data were entered into a geographic information system. The activity space of the adult respondents, measured in square kilometers, was defined as 1 standard deviational ellipse (SDE) of the geographic region in which respondents moved or traveled in the course of most of their daily activities. The SDE area is 1) determined by the maximum and minimum dispersion of a set of weighted data points, defined as time spent at specific places, from the mean center of the data points; and 2) contains approximately two thirds of the respondents’ activity data points.18 Two dichotomous variables were derived using this information: location of 1) the nearest provider site and 2) the nearest medical facility within or outside one SDE.
Outcomes Measures and Statistical Analysis
The primary outcome measure of child health care use was defined as total health care visits to specific health care providers (including doctors, nurses, physician assistants, chiropractors, and dentists) and to health care facilities (hospitals or public or private clinics) during the previous year. The outcome was the sum of counts from 2 questions corresponding to providers and facilities. Unusually large counts for each question were handled by truncating (before summing) the number of visits at 25. Truncation affected 2 children younger than 5 years and 4 children 5 years and older. Results were stratified into the younger than 5 years and 5 years and older age groups, according to the survey structure. Survey weights were used to account for unequal probability in the sampling scheme. Population means of health care visits were estimated using sampling weights with the PROC RATIO procedure in SUDAAN 8.0. (Research Triangle Institute, Research Triangle Park, NC).19 These means are equivalent to incidence densities because the exposure period of 1 year is the same for every child in the sample.20 The incidence density is the number of visits per unit time, in this case, 1 year. Pearson correlations were used to estimate relationships of continuous predictors with total number of visits.
Bivariate relationships between each potential predictor and total health care visits were assessed with survey log-linear regression (PROC LOGLINK in SUDAAN). Multivariate relationships were assessed with survey log-linear regression models that considered several explanatory variables simultaneously. An overview of the survey regression approach in the analysis of health surveys is provided elsewhere.21 Estimates of regression coefficients were adjusted for respondents’ sampling weights, and standard errors of regression coefficients were additionally adjusted for clustering of respondents within geographically based census segments. In this approach, exponentiation of the estimated regression coefficients gives the multivariate adjusted estimates of the ratio of incidence densities for 2 groups defined by a unit change in the covariate. Equivalently, these quantities are ratios of estimated population mean number of health care visits within 1 year between 2 groups.
Multivariate modeling was performed using in the initial models those variables with univariate P < .10 for total health care visits. Statistical significance for the final models was defined as P < .05. For the younger than 5 years group, a parsimonious model was sought (in consideration of the limited sample size) through use of a backward elimination procedure to discard variables that were not significant predictors of total health care visits. A backward elimination procedure was similarly used for the 5 years and older group.
Among the children younger than 5 years, 98% of visits were to medical care providers or sites and 2% were to dentists. Among children 5 years and older, 92% of visits were to medical providers or sites (11% of which were specified as being to an “eye doctor”) and 8% were to dentists. No visits to chiropractors were reported in either age group. The mean and median reported visits per child by age and gender in the year preceding the survey are provided in Table 1. Children younger than 2 years had more reported visits than older children (Table 1), as is typical. Reported health care use was generally higher for male children in the age groups younger than 2 years, 2 to 4 years, and 5 to 12 years and higher for girls in the 13 and older age group.
Approximately 90% of the children had health insurance coverage, with public insurance (Medicaid) accounting for half of coverage among those younger than 5 years but only 30% of coverage for those 5 years and older. Among the nonwhite children, 80% were black (8 of 10 younger than 5 years and 42 of 51 5 years and older). Other demographic characteristics available from the survey are listed in Table 2.
The overall health status of the children was reported as very good or excellent for 80% in the 2 age strata (Table 2). There were very few chronic illnesses present in the sample of children younger than 5 years, with only 5 of 69 having reported respiratory illnesses, which included asthma. Diseases and behavioral problems were more common among the children 5 years and older. One third of the children in both age groups were exposed to household tobacco smoke (Table 2).
Univariate Analysis of Determinants of Reported Health Care Use
Children Younger Than 5 Years
Age, race/ethnicity, birth weight, 5 household characteristics, 2 environmental exposures, 7 health-related measures, and 5 geographic access-to-care measures were evaluated for their potential impact on health care use. Results of categorical and continuous variables are presented in Tables 2 and Table 3, respectively. Household sizes >2 (vs household size = 2), yearly household income <$20 000 (vs higher income groups), public health insurance (vs uninsured), parent perception of the child’s overall health as good (vs very good or excellent), pain fairly often or often in the past month (vs a few times or none), pain moderate or severe in past month (vs mild or none), pain interfering some with activities in the past month (vs a little or none), and presence of a respiratory illness (vs none) were associated with increased numbers of health care visits in the year before the survey (P < .05; Table 2). The magnitudes (incidence density ratios) of these univariate associations were approximately in the 2- to 3-fold range.
Having the nearest primary care doctor farther away from the home than 1 SDE distance was associated with a 1.3-fold increase in the number of visits (P = .036; Table 2). Age was negatively correlated with number of visits, as were scores on the Global Health Item and Bodily Pain-Discomfort Scale (all P < .001; Table 3). Lower scores on the last 2 measures indicate poorer health of the child and increased pain, respectively.
Children 5 Years and Older
Age, race/ethnicity, birth weight, 5 household characteristics, 1 environmental exposure, 12 health-related measures, and 5 geographic access-to-care measures were evaluated for their potential impact on health care use. Being white (vs nonwhite), household size >2 (vs household size = 2), having public or private health insurance (vs uninsured), parent education less than high school (vs high school or beyond), and household exposure to tobacco smoke (vs no exposure) all were associated approximately with 2- to 3.5-fold increased numbers of reported health care visits in the previous year (P < .05; Table 2). Parent perception of the child’s overall health as fair or good (vs very good or excellent), pain often or fairly often in the past month (vs a few times or none), pain interfering with normal activities some or a little (vs none), presence of respiratory illness (vs none), presence of orthopedic problems (vs none), and presence of epilepsy (vs none) were associated approximately with 2- to 4.8-fold increased numbers of reported health care visits in the previous year (P < .05; Table 2).
Birth weight was negatively associated with reported visits (P = .005; Table 3). Scores indicating greater pain on the Bodily Pain-Discomfort Scale and greater emotional impact of the child’s illness on the parent on the Emotional Impact on Parent Item were associated with an increased number of reported visits (P < .001; Table 3). Of the 5 geographic access-to-care variables, only the minimum distance to the nearest health care facility had a significant association with number of reported visits, and this indicated increased visits with increasing distance from home to facility (correlation coefficient 0.13; P = .02; Table 3).
Multivariate Analysis of Determinants of Reported Health Care Use
Children Younger Than 5 Years
In the final multivariate model, age, household income, and health insurance were associated with the number of reported health care visits in the previous year (Table 4). Those with public insurance (Medicaid) had 3.8-fold more visits (95% confidence interval [CI]: 1.51–9.53) and those with private insurance had 1.6-fold more visits than uninsured children, although this latter result was not statistically significant. Children whose households had incomes >$40 000 had 1.76-fold more visits (95% CI: 1.08–2.86) than those with household incomes <$20 000. Those with household incomes between $20 000 and $40 000 had fewer visits than the lower income group, but this did not reach statistical significance in this sample.
In terms of the age of the child, incidence density ratio (IDR) of health care visits was 0.77 per each increasing year of age. A 4-year-old child therefore would have only 45.6% as many visits in the year before the survey than a 1-year-old child. The Bodily Pain-Discomfort Scale had an IDR of magnitude similar to that seen in the older age stratum (see below; Table 4). This result approached but did not reach statistical significance, possibly because of the small sample size of this age stratum.
Race/ethnicity (white, nonwhite), although not associated with number of visits in univariate analysis, was included in several initial models to assess for negative confounding in view of the 3-fold univariate difference by race in the older age group. In these adjusted models, there was a 2-fold increase in visits among whites relative to nonwhites, similar to that seen in the older age group (below), but this did not reach statistical significance.
Children 5 Years and Older
Thirteen of 19 variables that met multivariate inclusion criteria (univariate P < .10 or significance in the final model in children younger than 5 years) were used in the final analysis. Of 4 pain-related variables, the Bodily Pain-Discomfort Scale was selected as the single representation of the pain experienced by the child, as it was highly correlated with pain frequency in the past month and included the information captured in the pain intensity question. Orthopedic problems were excluded also because of high correlation with the Bodily Pain-Discomfort Scale. Epilepsy was excluded because of low frequency of occurrence (N = 10). Birth weight was “missing” from 20% of the sample, and its exclusion improved overall stability of modeled associations.
The final multivariate model had 4 variables with statistically significant associations with health care use: race/ethnicity, household income, health insurance, and bodily pain and discomfort (Table 4). White children had 1.97-fold more visits (95% CI: 1.18–3.27) than black children. Those with household incomes >$40 000 had 2.53-fold more visits (95% CI: 1.39–4.63), and those with incomes $20 000 to $40 000 had 2.07-fold more visits (95% CI: 1.15–3.75) than those with incomes <$20 000.
Health insurance had the largest influence on number of visits in the previous year: those with public insurance (Medicaid) had 6.08-fold more visits (95% CI: 2.83–13.03) than uninsured children. Privately insured children had 1.37-fold as many visits as uninsured children, but this did not reach statistical significance (95% CI: 0.70–2.69). There was no statistical difference between the health care use of the 2 insured groups.
The Bodily Pain and Discomfort Scale, which assessed both frequency and intensity of the child’s pain during the past 4 weeks, was associated with the number of visits in the previous year with an IDR of 0.86 (95% CI: 0.79–0.93) for every 10-point increase on the 100-point scale (higher scores = less pain). The scale assessed intensity of pain in 6 levels from none to very severe and frequency in 6 levels from none to every day or almost every day. A child with mild pain occurring a few times during the past 4 weeks (score: 60) would have only 55% as many visits in the past year as a child with severe pain occurring very often in the past four weeks (score: 20; Table 4).
Birth weight, which had a negative univariate correlation with the number of visits in the previous year (−0.32; P = .005; Table 3), was missing from 20% of the children in the 5 years and older age group and was excluded from the final multivariate analysis presented in Table 4. However, in exploratory multivariate models that included birth weight, the IDR was 0.96 for every 100-g increase in birth weight (95% CI: 0.93–1.00; P = .045). Therefore, a child with a 3000-g birth weight had on average only 54% of the health care visits of a child with a 1500-g birth weight in this age group.
The independent predictors of health care use in this study of rural children in North Carolina are similar to those found in multiple studies of general populations of children in the United States during the last 30 years. Among children who were younger than 5 years and living in rural Appalachian regions of North Carolina in 1999, the primary determinants of health care use during the previous year were age, insurance status, and household income. Infants had more visits than older, preschool children and those with household incomes >$40 000 per year had 76% more visits than those with incomes <$20 000 per year. Children with public insurance, exclusively Medicaid in this population, had almost 4 times as many visits, and those with private insurance had 1.6-fold as many visits as uninsured children.
Among the children and adolescents 5 through 17 years of age, health insurance status, household income, pain during the past month, and race were the primary determinants of health care use during the previous year. Those with public health insurance had 6 times more health care visits than uninsured children. Household incomes >$40 000 per year were associated with 2.5-fold increased health care visits, and those with household incomes between $20 000 and $40 000 per year had 2-fold increased health care visits, compared with those with household incomes <$20 000 per year. White children had almost twice as many visits in the past year as black children in this age group. Pain experienced during the past month, as perceived by the parent, also predicted the number of visits in the older age group.
It is widely recognized that uninsured children have fewer health care visits22–27 or more frequently unmet health needs27–29 than insured children. Approximately 10% of the children in this rural area of North Carolina were uninsured in 1999. This is similar to the findings of the national sample-based Medical Expenditure Panel Survey (MEPS) for 1996, when an estimated 9.0% of children younger than 5 years and 11.3% of children 5 to 17 years of age were uninsured,30 and MEPS 2000 data, which estimated that 13.9% of children were uninsured.31 The higher proportion of public insurance among younger children than teenagers in this rural area also was similar to the findings in the 1996 and 2000 MEPS studies. Childhood insurance coverage from public and private programs combined seems to be at least as good in this rural area as in the rest of the nation at a similar point in time.
It was striking in this rural population that, relative to uninsured children, those with public health insurance had greater health care use than those with private insurance in both age groups. The explanation for this is unclear, as the opposite trend is usually observed.5,31 This potentially is a credit to strong public health programs in North Carolina as well as widespread acceptance of Medicaid by private practitioners in the state. Having private insurance did not result in significantly more health care visits among children 5 years and older relative to uninsured children in this age group, but the lack of significance may be attributable to the small number of uninsured children (sample size). Alternatively, these children possibly were healthier overall than those with public insurance.
The association of higher annual incomes with increased health care use, independent of insurance status, has been documented in national studies.27–29,32–34 It is not clear from these studies or ours whether this is attributable to increased use for relatively minor health care needs in the higher income groups or underutilization with ongoing unmet needs in lower income groups. Higher income clearly allows more ready accession of care. In this rural population, children who were younger than 5 years and from households in the middle income group ($20 000 to $40 000 per year) had less health care use (although not statistically different) than those in the lower income group. This was in contrast to children 5 years and older: those in the middle income group had twice as many visits as those in the lower income group.
This age-stratum difference in impact of household income could be an artifact of small sample size in the younger age group and, relatedly, possible failure of the model to account fully for potential confounding effects of other variables on income. For example, children with household incomes <$20 000 per year had the most visits in both age strata (Table 2) but lower IDRs than at least 1 higher income group after multivariate analysis (Table 4). This change in apparent impact of household income is attributable in part to the presence of greater bodily pain among children in the lower income households (data not shown), which is adjusted for in the multivariate model.
Parent perception of the child’s pain during the past month, as measured by a Bodily Pain and Discomfort Scale, yielded similar associations with health care visits in the previous year in both age groups. The result did not reach statistical significance in the group that was younger than 5 years, likely because of the small sample size in this stratum. That apparent pain, regardless of cause, would be a driving concern for parents to seek care for their children is not surprising. Child pain has predicted health care use in other studies, explaining approximately 3% of the variance of health care use among 5- to 11-year-olds enrolled in a prepaid health maintenance organization in Maryland in the late 1980s.35 Among children with chronic rheumatic diseases, children who were more pessimistic (catastrophizing) in the descriptions of their pain also had parents who reported higher levels of personal pain than children with less pessimistic views of their pain.36
The 2-fold higher use of health care by white versus nonwhite (predominantly black) children in this study (statistically significant for older children but not those younger than 5 years, likely attributable to the small sample size) is in contrast to the findings in 1992 NHIS data, where aggregate data for all rural residents younger than 65 years showed somewhat less health care use than their urban counterparts but no difference in utilization between rural whites and blacks.34 It is not clear whether this is attributable to differential access to or differential use of care. Lower proportions of minority children have an identified source of regular health care relative to white children, independent of household income and insurance status.32,33 Black children also have been found to have fewer physician visits per year, even after controlling for health status, than white children.5,33,37–40 Among 10- to 17-year-olds in the 1988 NHIS study, health insurance increased access and usage measures for minority youths more than for white youths, but racial differences in health care visits persisted even after adjusting for health insurance, family income, and needs.38 It is not certain whether this decreased utilization is accompanied by increased perceived or real unmet health care needs.32 Nonfinancial methods of enabling more equitable use of health care by nonwhite children likely deserve additional study.38
Few recent data are available on overall health care use among rural children in the United States. The mean of 5.7 reported total health care visits per child during the year before the MAP survey in North Carolina in 1999 is greater than the mean of 3.1 health care visits per child (0–18 years of age) found for children who lived in rural areas in the 1980 National Medical Care Utilization and Expenditure Survey.5 Infants and younger children typically have more health care visits per year than older children,41 and this was seen in our study. Also, in each age stratum listed in Table 1, the mean and median numbers of visits per child exceeded the number of yearly well-child visits recommended by the American Academy of Pediatrics for that age group (average of 5 visits per year in the first 2 years of life and 1 per year thereafter).42
None of 5 variables assessing geographic distance to care or the availability of multiple vehicles in the household had any impact on number of visits in the previous year. These observations, along with the mean number of visits per child, suggest that health care is generally reasonably accessible to this rural pediatric population in western North Carolina, at least in terms of geographic and transportation issues. Results from analysis of MAP data for adults show that those who live in more rural areas are more willing to drive farther to care (data not shown).
Potential limitations of this study include reliance on parent recall of health care visits for their children without confirmation from their medical records. This approach has been used in many surveys of health care use, however, and review of medical records was beyond the scope of the project. The study sample was drawn from a single rural geographic area of the country, and the health-related behaviors of this population may not reflect those of rural areas in other parts of the United States. This area is home to Native American groups and a growing Latino population that were not able to be included in the study sample. The counties selected for this study are heterogeneous in their degree of rurality, but the population demographics otherwise seem to be similar to the aggregate rural United States.1,15 The proportion of uninsured children in this rural area also mirrors that of the country as a whole at a similar time period.
This study assessed predictors of use of all types of health care and not simply physician visits, so the results may not be directly comparable to some previous studies that focused only on these. However, the great majority of the visits to providers and facilities were for medical visits, and the similarities of the independent predictors in this study and others further suggests that our broader definition of health care visits does not diminish the generalizability of these observations.
Rural health and access of rural residents to health care has been a long but periodic focus of legislative bodies. In the late 1980s, there was recognition of the need to develop further profiles of rural residents and their health status, access to care, and utilization of care along the entire continuum of rurality.6 Our study is the first to examine specifically some of these factors for rural children. It is noteworthy that the key determinants of health care use among these children were similar to those found in urban and other populations in the United States and likely are universal: health insurance, household income, and parent perceptions of their children’s pain.
Although overall access to care in this rural population seems reasonably good at the height of an economic boom in the late 1990s, there still are problem areas that likely will require increasing attention in the years to come. Families that earn too much income to qualify for public health insurance programs but that either have employment that does not provide health insurance benefits or that earn too little to buy adequate private health insurance may be those in most jeopardy. Such families, perhaps especially those with younger children, should remain a priority in policy development and consideration for expansion of eligibility for State Child Health Insurance Programs. Rural areas, which often have a higher proportion of jobs without benefits and continue to lose jobs in the manufacturing sector that provide benefits, require ongoing monitoring for increases in numbers of uninsured households and other measures of health care use and access and unmet health care needs relative to urban and suburban areas of the country.
The issues of parent/household income, insurance coverage, and parent/maternal education are often intertwined. Higher percentages of children of parents with lower parent educational attainment are uninsured (MEPS 1996 data).30 More frequent use of physician services by the primary parent also has been associated with increased visits per year to physicians by their children.4,35,43 In a given study or population, 1 or more of these factors may be stronger predictors of health care use/nonuse than others,28,37,44–47 but all likely should be considered when developing policies to improve the health of children.
Policies for sustainable improvement in the utilization of child health care services in rural communities must address community development as well as individual behavior concerns. As in other settings, real or perceived racial/ethnic barriers to care need to be explored and addressed. Community development programs that increase educational and economic attainment will directly affect child health. The behaviors of more educated adults (eg, less tobacco use, greater use of health care) are positively associated with child health. More economically secure families (which is 1 outcome of increased educational attainment) are better able to have the financial resources to afford appropriate child health care. Communities with a better educated population likely will have industries/employers that provide benefits such as health insurance. Finally, differential use of health care among white and black children, especially those 5 years and older, merits additional exploration.
Support for this research was provided in part by grant R01 HS09624 from the Agency for Healthcare Quality and Research.
- ↵Clark SJ, Savitz LA, Randolph RK. Rural children’s health. In: Ricketts TC, ed. Rural Health in the United States. New York, NY: Oxford University Press; 1999:150–158
- ↵Gesler WM, Hartwell S, Ricketts TC, Rosenberg MS. Introduction. In: Gesler WM, Ricketts TC, eds. Health in Rural North America: The Geography of Health Care Services and Delivery. New Brunswick, NJ: Rutgers University Press; 1992:1–22
- ↵Ricketts TC. Preface. In: Ricketts TC, ed. Rural Health in the United States. New York, NY: Oxford University Press; 1999:vii–viii
- ↵Butler JA, Winter WD, Singer JD, Wenger M. Medical care use and expenditure among children and youth in the United States: analysis of a national probability sample. Pediatrics.1985;76 :495– 507
- ↵Eberhardt MS, Ingram DD, Makuc DM, et al. Urban and Rural Chartbook. Health, United States, 2001. Hyattsville, MD: National Center for Health Statistics; 2001
- ↵Aday LA, Andersen R. Development of Indices of Access to Medical Care. Ann Arbor, MI: Health Administration Press; 1974
- ↵Verheij R, editor. Urban-Rural Variations in Health Care. Utrecht, the Netherlands: NIVEL, Netherlands Institute of Primary Health Care; 1999
- ↵Ricketts TC, Savitz L. Access to Health Services. In: Ricketts TC, Gesler WM, Savitz L, Osborne D, eds. Geographic Methods for Health Services Research. Lanham, MD: University Press of America; 1994:91–119
- ↵Penchansky R. The Concept of Access: A Definition. Washington, DC: National Health Planning Information Center, Bureau of Health Planning and Resource Development, Department of Health, Education and Welfare; 1977 (Unpublished report; Contract No. HRA 230-75-0073)
- ↵Research Triangle Institute. Geographic Accessibility of Health Care in Rural Areas Project. Research Triangle Park, NC: Research Triangle Institute; 2000 (RTI Project 7438; Data Collection Report)
- ↵Butler MA, Beale CA. Rural-Urban Continuum Codes for Metropolitan and Nonmetropolitan Counties, 1993. Washington, DC: Agriculture and Rural Economy Division, Economic Research Service, United States Department of Agriculture, AGES 9425; 1994
- ↵Landgraf JM, Abetz L, Ware JE. The CHQ User’s Manual. Boston, MA: HealthAct; 1999
- ↵Research Triangle Institute. SUDAAN User’s Manual, Release 8.0. Research Triangle Park, NC: Research Triangle Institute; 2001
- ↵Lavange LM, Stearns SC, Lafata JE, Koch GG, Shah BV. Innovative strategies using SUDAAN for analysis of health surveys with complex samples. Stat Methods Med Res.1996;5 :311– 329
- ↵Wood DL, Hayward RA, Corey CR, Freeman HE, Shapiro MF. Access to medical care for children and adolescents in the United States. Pediatrics.1990;86 :666– 673
- Keane CR, Lave JR, Ricci EM, LaVallee CP. The impact of Children’s Health Insurance Program by age. Pediatrics.1999;104 :1051– 1058
- Holl JL, Szilagyi PG, Rodewald LE, et al. Evaluation of New York state’s Child Health Plus: access, utilization, quality of health care, and health status. Pediatrics.2000;105 :711– 718
- ↵Feinberg E, Swartz K, Zaslavsky A, Gardner J, Walker DK. Family income and the impact of a children’s health insurance program on reported need for health services and unmet health need. Pediatrics.2002;109(2) . Available at: http://www.pediatrics.org/cgi/content/full/109/2/e29
- ↵McCormick MC, Kass B, Elixhauser A, Thompson J, Simpson L. Annual report on access to and utilization of health care for children and youth in the United States–1999. Pediatrics.2000;219– 230
- ↵Simpson G, Bloom B, Cohen RA, Parsons PE. Access to health care. Part 1: children. Vital Health Stat 10.1997:1– 46
- ↵Newachek PW, Hughes DC, Stoddard JJ. Children’s access to primary care: differences by race, income, and insurance status. Pediatrics.1996;97 :26– 32
- ↵Schanberg LE, Anthony KK, Gil KM, Lefebvre JC, Kredich DW, Macharoni LM. Family pain history predicts child health status in children with chronic rheumatic disease. Pediatrics.2001;108(3) . Available at: http://www.pediatrics.org/cgi/content/full/108/3/e47
- ↵National Center for Health Statistics. Health, United States, 2001: Table 72. Health care visits to doctor’s offices, emergency departments, and home visits within the past 12 months, according to selected characteristics, United States, 1997–99. Available at: www.cdc.gov/nchs/products/pubs/pubd/hus. Accessed September 11, 2002
- ↵Committee on Practice and Ambulatory Medicine, American Academy of Pediatrics. Recommendations for preventive pediatric health care. Pediatrics.1995;96 :373– 374
- Newacheck PW, Halfon N. Access to ambulatory care services for economically disadvantaged children. Pediatrics.1986;78 :813– 819
- Copyright © 2003 by the American Academy of Pediatrics