Objectives. 1) To examine age variation in unmet need/delayed care, access, utilization, and restricted activities attributable to lack of health insurance in children before they receive health insurance; and 2) to examine the effect of health insurance on these indicators within each age group of children (in years).
Methods. We use cohort data on children before and after receiving health insurance. The study population consists of 750 children, 0 through 19 years of age, newly enrolling in two children's health programs. The families of the newly enrolled children were interviewed at the time of their enrollment (baseline), and again at 6 months and 1 year after enrollment. The dependent variables measured included access to regular provider, utilization, unmet need or delayed health care, and restrictions on activities attributable to health insurance status. All these indicator variables were examined by age groups (0–5, 6–10, 11–14, and 15–19 years of age). χ2tests were performed to determine whether these dependent variables varied by age at baseline. Using logistic regression, odds ratios were calculated for baseline indicators by age group of child, adjusting for variables commonly found to be associated with health insurance status and utilization. Changes in indicator variables from before to after receiving health insurance within each age group were documented and tested using the McNemar test. A comparison group of families of children enrolling newly 12 months later were interviewed to identify any potential effects of trend.
Results. All ages of children saw statistically significant improvements in access, reduced unmet/delayed care, dental utilization, and childhood activities. Before obtaining health insurance, older children, compared with younger children, were more likely to have had unmet/delayed care, to have not received health care, to have low access, and to have had activities limited by their parents. This pattern held for all types of care except dental care. Age effects were strong and independent of covariates. After being covered by health insurance, the majority of the delayed care, low utilization, low access, and limited activities in the older age groups (11–14 and 15–19 years) was eliminated. Thus, as levels of unmet need, delayed care, and limitations in activities approached zero in all age groups by 1 year after receipt of health insurance, age variation in these variables was eliminated. By contrast, age variation in utilization remained detectable yet greatly reduced.
Conclusion. Health insurance will reduce unmet need, delayed care, and restricted childhood activities in all age groups. Health care professionals and policy makers also should be aware of the especially high health care delay, unmet need, and restricted activities experienced by uninsured older children. The new state children's health insurance programs offer the potential to eliminate these problems. Realization of this potential requires that enrollment criteria, outreach strategies, and delivery systems be effectively fashioned so that all ages of children are enrolled in health insurance.
- CI =
- 95% confidence interval
Under the Balanced Budget Act of 1997, the Congress established the State Children's Health Insurance Program, the most significant health insurance initiative for children since the enactment of the Medicaid program in 1965. Under this act, 24 billion dollars will be allocated to states over a 5-year period to provide health insurance to children who otherwise would be uninsured. The law gives the states considerable flexibility in how to insure the children and in the extent to which services (such as vision, hearing, and dental services) are covered. To date, the Federal government has approved 45 Kidcare programs. Recognizing the importance of this initiative, a number of private foundations and government agencies are funding evaluations of the program.
Some preliminary information on the likely impact of these new insurance programs can be gained through evaluations of earlier programs designed to cover uninsured children. We have been evaluating the impact on children of such a program. In the initial report of our evaluation, we described the impact of not having health insurance on the health status, utilization of services, and limitation of activity of uninsured children.1 In a second paper, we described the general impact of such a health insurance program on newly enrolled children.2 We found there was a significant increase in access to care, decrease in unmet need and/or delay in receiving services, more appropriate use of services, and less restriction of childhood activities after receipt of insurance coverage. However, we did not examine the impact of the program on children of different ages.
There are reasons to believe that the impact of such programs may vary by the age of the child. For instance, data from the 1977, 1987, and 1997 National Medical Expenditure Panel Surveys suggest that, among uninsured children, older children were less likely to have a regular source of care than were younger children.3 ,4 Thus, if health insurance led to equality in the probability of having a regular source of care across all ages of children, we would expect that a higher proportion of older children would benefit from such programs. Similarly, if uninsured children of older ages are more likely to have unmet need or restricted activities that can be eliminated through health insurance, then it would be especially important to target this group for coverage. However, there have been few studies that assess the impact of being uninsured on children in different age groups or the impact of obtaining coverage by the age of the child.
In this article, we report the effect that a children's health insurance program has on children of varying ages. Three separate, but interrelated, questions are addressed: 1) Before enrollment in the health insurance program, were there any differences in indicator variables across age groups?; 2) After being enrolled in the program for 1 year, was there any improvement in indicator variables within age groups?; and 3) After being enrolled in the health insurance program for 1 year, what was the magnitude of differences in indicators across age groups?
Four categories of dependent variables are measured: 1) having an usual source of care; 2) use of specific services; 3) unmet need or delay in obtaining services; and 4) childhood activities limited by parent because of the lack of health insurance.
The evaluation design has been described in detail elsewhere.1 ,2 In brief, we used a before–after design with a comparison group to study the effects of health insurance for uninsured children in Western Pennsylvania. The program studied is actually a blend of two programs: the Children's Health Insurance Program of Pennsylvania (called BlueCHIP in Western Pennsylvania) and the Caring Program for Children. Both Programs were administered by the Western Pennsylvania Caring Foundation for Children, an affiliate of Highmark Blue Cross Blue Shield. The two programs offered an identical comprehensive package of inpatient, outpatient (including dental and vision services), and preventive health care services to children. There was no copayment with the exception of a small copayment for outpatient drugs. Together they provided seamless coverage for uninsured children 0 through 18 years of age. The eligibility criteria varied by age of the child and family income.1 ,2
The Study Population
The names of 5864 uninsured children enrolling in the insurance programs between August and December 1995 were obtained from the Caring Foundation. We aggregated children to the family level and randomly selected 887 families to be recruited for telephone interviews by experienced interviewers. The families were contacted within 2 weeks after acceptance into either the BlueCHIP or the Caring Program. Of the 887 families, 783 (88.3%) agreed to participate and were interviewed. These families were contacted again after 6 months and 12 months; 659 families answered all three interviews. These 659 families had a total of 1031 children. Of the 1031 children at the end of the study year, ∼15% were covered by private health insurance, 7% were covered by Medicaid, 6% were uninsured, and 73% (750 children) were enrolled continuously, ie, still covered by either the BlueCHIP or Caring Programs. (We do not know whether the children who shifted to Medicaid did so because their family income decreased or because their family spent down to Medicaid because of health care utilization of another family member. We also do not know why some children lost their health insurance. We believe children who gained private insurance did so as a result of a change in their parents' income or work status.2) The 750 children who remained continuously enrolled are the subjects of this study. Given the high response and large sample size, this sample is representative of the total population of the Western Pennsylvania enrollees of the insurance program.
In this article, we include data from the baseline and 12-month follow-up interview. In an earlier paper, we reported all 6-month aggregate follow-up data.2 We showed that, with the exception of utilization variables, most indicators improved in the first 6 months after enrollment and then improved only slightly after that. However, utilization increased in the first 6 months and then somewhat decreased in the second 6 months. Thus, in the present article, we report the 6-month follow-up data only for health service utilization data.
The interviewers followed an almost identical survey instrument, with both fixed-response and open-ended questions, for both the baseline and follow-up interviews. The respondents, usually mothers, were asked questions about each child in the family. In addition to standard demographic information, several questions were included about access and use of health services, including whether the child had a usual source of medical and dental care, the number of visits the child made to different types of health care providers, and whether the child experienced unmet need or delayed care for six types of services. For all questions related to health services utilization and health status, the parents were asked to focus on the 6-month period before the interview. Parents also were asked about whether the child's health insurance status had any effect on usual childhood activities. During the baseline interview, the respondents were asked how long a child had been without health insurance.
The Comparison Group
A list of children who were enrolled in the BlueCHIP or the Caring Program between August and December 1996 (corresponding to 1 year after our 1995 cohort enrollment) was made available to the research team to serve as a comparison group. As for the cohort group, we aggregated children to the family level and randomly selected 371 families to be interviewed using the same telephone survey; 330 families (89%) who had 460 newly enrolled children agreed to participate. These children serve as a comparison group in that they enable us to assess whether changes observed in the study group were attributable to the insurance programs rather than to other underlying changes or influences in the environment. This design, a variation of a recurrent institutional cycle design, rules out a major threat to the internal validity of simple before–after evaluations, namely, the effects of a secular trend.2 ,5 ,6
The children were first grouped into age groups that are close to those used in recent American Academy of Pediatrics recommendations on frequency of health examinations. We grouped children into the following categories: 0 through 5, 6 to 10, 11 to 14, and ≥15 years of age.7 Although the American Academy of Pediatrics recommends that children 0 to 2 years of age should make more visits than children 3 to 5 years of age, we had too few children <2 years of age to justify creating a separate group; therefore, we merged this category into the 0- to 5-year stratum. These categories contain adequate sample sizes for each age group. We first performed the χ2 test to determine whether there were statistically significant differences in indicators across the age groups. In addition, we used logistic regressions in which the indicator variables were used as dependent variables and the independent variable was age group. Other child or family characteristics found to be associated with the indicator variables were included as covariates. These variables include months without health insurance coverage, parental working status, maternal education, the number of children in family, race, maternal education, parental employment status, urban–rural status, and health conditions of child (ear, nose, and throat infections, asthma, allergies, and overall health on a 4-point Likert scale). From the regression results, we calculated the odds ratios for our indicators by age group of child, controlling for the covariates. The odds ratios for each age group used the youngest age group as the reference category.
To determine the impact of the health insurance programs on the various indicators we: 1) examined the changes in indicators from before and after receiving health insurance within each age stratum, testing for statistical significance using the two-tailed McNemar test for within-subject changes; and 2) used a comparison group of children newly enrolling in the health insurance programs 1 year after the initial cohort to rule out the possibility that effects were attributable to a general secular trend.
The software package used for all calculations was the Statistical Package for the Social Sciences Version 7.5 (SPSS, Chicago, IL).8
Background Characteristics at Enrollment by Age Group
At the time of enrollment, younger children had been without insurance for a shorter period than had older children. Specifically, 8.6% of the youngest children (0–5 years of age) were uninsured for periods of ≥12 months, compared with ∼20% in each other age group (Table 1). The variation in time uninsured by age is statistically significant at the P< .005 level by the χ2 test. The only other statistically significant variation is in family size, with older children more likely to come from larger families as would be expected. The race of the respondent does not vary by age of child. The vast majority of this population is white, thus there is not statistical power to examine differences according to race.1 ,2Parental working status varies slightly with age, with the proportion with both parents working slightly decreasing with the age of the child, but the difference is not statistically significant. The majority of families have at least 1 working parent, for all age groups of children.
Reported ear infections, asthma, and allergies vary by age in the expected fashion. Ear infections are more common in the younger children, 34.5% of 0- to 5-year-olds had an ear infection in the 6 months before enrollment, 23.9%, 16.7%, and 14.1%, respectively, in the successive increasing age groups, significant at theP < .0005 level (Table 2). Asthma and allergies are slightly less common in the group of 0- to 5-year-old children, compared with the other age groups. The variation in allergies by age is statistically significant at the P < .005 level. The general health status of these newly enrolling children is good to excellent in ≥90% of the children in each age group.
Differences in the Indicator Variables at Enrollment
Regular Provider and Utilization Before Enrollment
The proportion of children who had a regular provider of health care and dental care varied markedly by age (Table 3). The pattern for dental providers is the opposite of that for regular health providers; relative to younger children, older children are more likely to lack a regular health provider but less likely to lack a regular dentist. (It is likely that very young children are not encouraged to see a dentist.) However, the age variation in having no regular provider is not statistically significant by the χ2 test. The differences across the age groups in not having a regular dentist are statistically significant at the P < .0005 level.
The odds ratios presented in Table 5indicate that the relationship between the age and indicator variables is strong. Controlling for covariates, children 11 to 14 years of age were 2.3 times as likely (95% confidence interval [CI]: 1.0, 5.4) than were children 0 to 5 years of age, to have no regular provider of care at baseline and ∼0.3 times as likely (CI: 0.2, 0.6) to lack a regular dentist, whereas children age 15 to 19 years of age were 2.7 times as likely (CI: 1.1, 6.5) to have no regular provider and 0.23 times as likely (CI: 0.1, 0.5) to lack a regular dentist. Note that the age difference in having a regular provider is statistically significant in this multivariate analysis, although it was not significant in the univariate analysis (Table 3).
Before enrollment in the program, the proportion of children who had no physician visit in the past 6 months varied greatly across the age groups; 21.6% of the children 0 to 5 years of age had no physician visit, compared with 53% of children 15 to 19 years of age (Table 3). There is a nonlinear pattern across the age groups in the proportion of children who had any dental visits; children <5 years of age have the fewest visits, whereas those in the 6- to 10-year age group show the greatest use of dental services at baseline. Consistent with this pattern, Table 5 shows that, controlling for covariates, children 11 to 14 years of age were 3.5 times as likely (CI: 2.1, 6.0) than were children 0 to 5 years of age to have had no visit to a physician, and 0.5 times as likely to have had no dental visits (CI: 0.3, 0.8) in the 6 months before enrollment. Similarly, children 15 to 19 years of age were 3.6 times as likely (CI: 2.0, 6.4) to have had no visits to a physician and 0.8 times as likely (CI: 0.4, 1.3) to have had no visits to a dentist.
Unmet Need and/or Delay
There was considerable unmet need and delayed care for each type of service in the 6 months before enrollment in the program (Table 4). Furthermore, older children experienced more unmet needs or delayed care than did younger children. The proportion of the children experiencing some unmet need and/or delayed care ranged across services from 5.0% for physician recommended care (essentially referrals to specialists or follow-up visits) to 43% for dental care. For each category of health care, the proportion of children experiencing unmet need or delayed care generally rises with each successive age group. The χ2 test indicates that, with the exception of emergency department care and prescription drugs, these trends are statistically significant at the P < .0005 level.
The odds ratios in Table 6 provide information on how the amount of unmet need or delayed care varied across children by age group after controlling for factors that have been demonstrated to influence the use of health care services. In general, these odds ratios are consistent with the data presented inTable 4. The 11- to 14- and the 15- to 19-year-old children are significantly more likely to have unmet need for care or delayed care in all categories except emergency care, recommended care, and prescriptions. For example, controlling for other factors based on reports of the mothers interviewed, children 11 to 14 years of age are 1.9 (CI: 1.1, 3.3) times more likely to experience unmet need or delay in receiving physician care than are very young children. Children 15 to 19 years of age are 3.2 (CI: 1.8, 6.0) as likely to experience unmet need/delay in care than are those in the youngest age group.
Limitations on Usual Activities
At the time of enrollment, parents were more likely to restrict the activities of their younger children than to restrict the activities of older children. During the interviews, participants were asked, “Has (child's name) health insurance status led you to limit (child's name) activities in any way?” Overall, as shown in Table 4, ∼12% of the children were restricted by their parents in their usual activities because of the lack of health insurance. As an example of such restrictions, several mothers indicated that they would not let their children ride bikes, because they were frightened of injuries and subsequent medical costs.2 There were marked differences across the age groups in the proportion of the children who were limited in this way. Almost 19% of children 15 to 19 years of age had activities limited by their parents because of lack of health insurance, whereas only 5.6% of the youngest children were so limited. The variation by age is statistically significant at theP < .005 level by the χ2 test. Parents were more than four times as likely to limit the activities of children >11 years of age than of children 0 to 5 years of age because of health insurance status (Table 6).
Change in Indicators After Enrollment
Most of the age differences had disappeared or were reduced markedly by 1 year after enrollment. Tables 3 and 4 also provide data on the indicators at 12 months as well as the statistical tests indicating whether the changes between baseline and 12 months within each age group were statistically significant. All access and utilization variables shown in Table 3 showed overall improvement. Most unmet need or delayed care variables, shown in Table 4, also showed overall improvement. In general, these indicators that improved statistically significantly for the overall group, also improved statistically significantly for each individual age group. The exception is visits to the medical doctor, although the overall proportion seeing a medical doctor increased, a statistically significant increase at 1 year was found only for the 11- to 14- and 15- to 19-year-old age groups. At the 6-month follow-up, however, the proportion of 0- to 5-year-olds without a visit to the medical doctor statistically significantly decreased to 13%. A very large proportion of 0- to 5-year-olds saw a physician in the first 6 months after enrollment.
The proportion of children who had a regular provider and who had a regular dentist both increased after the receipt of health insurance benefits; at the end of a year, nearly every child had a regular physician. The proportion of children who visited a dentist increased by nearly 25%. This large increase in the proportion of children who had a dental visit was observed for children in all age groups with the exception of children 11 to 14 years of age, in which the increase was 16%.
We found that there was a significant decrease in the amount of reported unmet need and/or delay for most types of health care in the cohort as a whole. Furthermore, a significant proportion of children in each age group reported a decrease in unmet need and delayed care for physician, dental, prescription drugs, and eye care. With the exception of children 0 to 5 years of age (who had almost no unmet need or delayed care at baseline), there was in each age group a significant drop in the proportion of children with unmet need/delayed care for eye care. The decreases in the proportion of children with reported unmet need and/or delayed care was marked especially in the older age groups. For example, at baseline, 37% of the children 15 to 19 years of age experienced some unmet need or delayed physician care in the past 6 months, whereas at 12 months, only 3.7% did so. Figure 1 illustrates how this variation by age at baseline vanishes after 12 months. Finally, the proportion of children whose activities were limited because of their health insurance status, as high as 18.5% in the 15- to 19-year-old group and 15.4% in the 11- to 14-year-old group, decreased to essentially zero in every age group after receipt of health insurance, as illustrated inFig 2.
Differences in the Indicator Variables at 12 Months Across Age Groups
The data presented in Table 3 indicate that many of the marked differences in the indicator variables across the age groups observed at baseline had been reduced after the children were covered by health insurance. Although there were still some statistically significant differences by age in such indicators as health care visits and having a regular provider, the absolute magnitude of the differences had decreased.
More dramatically, most age differences in reported unmet need or delay in health care had been eliminated by 12 months after receipt of health insurance. For example, at 12 months the following indicators did not vary significantly across the age categories: proportion of children with unmet need or delay in receiving physician services in the last 6 months (see Fig 1), proportion of children experiencing any unmet need or delay in the past 6 months, and the proportion of children whose activities were limited as a result of their health insurance status (see Fig 2). Unmet need for or delay in dental care was still statistically significantly different across age groups (P < .005), nevertheless, the absolute levels of such delay or unmet need had become low for all ages. The absolute difference in unmet need for dental care between the oldest children and the youngest age groups was 19% at baseline but decreased to 7.5% by the 12-month follow-up interview. With the exception of dental care, the 12-month follow-up levels of unmet need or delayed care for all health service categories approached zero within each age group.
Our findings indicate that: 1) before enrollment in a children's health insurance program there were significant differences in most indicators across the age groups; 2) after enrollment in the program for a year, there was a significant improvement in most indicators for the total group of children included in this study as well as within each age group of children; and 3) after enrollment in the program for a year, the differences in indicators across the age groups had decreased markedly and in some cases had disappeared.
The high levels of reported unmet need for health care and/or delayed care in both the 11- to 14- and the 15- to 19-year-old groups relative to the younger groups of children are remarkable. For instance, the parents reported that 37% of the children 15 to 19 years of age, compared with 18% of the children 0 to 5 years of age had experienced some unmet need and or delayed physician care in the 6 months before enrollment. Yet almost all the unmet need and delays in health care in the older age groups were eliminated through the provision of health insurance. Thus, although among uninsured children, the health care needs of younger children are attended to more frequently, this is much less the case among the insured. After receiving insurance, unmet need and delayed care is practically nonexistent in all age groups.
Older children were much more likely than were younger children to have their activities limited because of lack of health insurance. As described in an earlier paper, these restricted activities included being allowed to ride a bicycle or to participate in after school sports.2 Thus, the impact of lack of health insurance on the lives of older children extends considerably beyond foregone health care. Virtually all the activity limitations in the older age groups (and in all age groups) were eliminated through the provision of health insurance.
The improvements seen are not attributable to a general improvement in the health care delivery system over the 1-year study period. As reported in detail in an earlier publication, the comparison group of children enrolling in 1996, exactly 1 year after the enrollment of the 1995 study cohort, demonstrated levels of unmet need, utilization, and limited activities comparable to those experienced by the study children at baseline.2 Furthermore, there were no statistically significant differences between the comparison group indicators and the baseline findings within the age strata. Therefore, these findings can be ascribed to the health insurance programs rather than to some change in the health care environment.
Moreover, our findings of statistical significance should not be attributable to multiple comparisons for several reasons. First, we only looked within age groups for variables that showed overall significance before broken down by age (this includes most variables), the recommended methodology for post hoc subgroup testing. Most results within age groups were significant at a very small P value. Secondly, all overall statistical significance (not broken down by age) for the unmet or delayed care, and most overall results for utilization remained statistically significant at the P < .05 level after adjusting the overall results for the Bonferonni correction for multiple comparisons.2 ,9 Although 5% of test results might show significance attributable to multiple comparisons, not only did almost every variable show significance overall, but virtually every age group showed statistically significant improvement.
Sample sizes within age groups were adequate to detect improvements in the indicator variables within each age category. Within individual age groups, the 95% CIs for proportions range from about + or −1.3% (for proportions close to 1% or 99%) to + or −8.4% (for proportions close to 50% in the smallest group). At 1 year after enrollment, unmet need or delay for physician care, emergency department care, recommended care, eye care, prescriptions, and activities limited attributable to lack of health insurance were reported no higher than 5.9% in any age group and usually lower. Proportions this low are accurate within + or −3% or less. Thus, we have high statistical confidence in the study's findings of very low levels of unmet need and delayed care after receipt of insurance. The 1-year postenrollment levels for unmet need or delayed dental care were higher but still statistically significantly lower than at baseline.
A limitation of the study is that it focuses on children who enrolled in a health insurance program in a small section of the country. However, the study children and their families are very similar to those of uninsured children and their families elsewhere with respect to family size, family structure, and working status.10Another limitation is that the data are based on self-report information. However, the respondents were asked to provide information on access and use of services within the 6 months before enrollment, so it is unlikely that there would be any difference in the validity of the recall across the children of different ages or across the interview periods.
Children's health care has been a priority concern for many years. Public opinion polls have shown throughout the 1990s that the public wants children to be a top priority for government spending and considers access to health care more important than other key children's issues.11 Although the polls frame the questions and even the notion of public priorities, there clearly has been a mandate for expanding government support for children's health. Medicaid expansions have occurred since the late 1980s and early 1990s, even while other social programs have been cut severely. However, under Medicaid, older children still face more restrictive criteria with a lower income ceiling. The extent to which the recent State Children's Health Insurance Program (Title XXI) program will extend coverage depends primarily on the choices made by state policy makers, regarding eligibility criteria, benefits, outreach, and other aspects of their programs. There continue to be disparities and inequities in the health status of children regardless of health insurance coverage. Financing health insurance is only one step toward assuring adequate health care. Identification and implementation of strategies targeting adolescents also are needed. The youngest children will likely receive priority within many of the new state programs, but the decision makers should pursue strategies to ensure that all children from 0 to 18 years of age are actually enrolled in health insurance, considering the outstandingly high health care delay, unmet need, and limited activities experienced among older children who are not enrolled in health insurance.
The issues raised in this article go far beyond the mandate of this society to assure that children have adequate health care throughout their entire period of primary and secondary education. The data presented in this paper reveal a clear pattern of greatly reduced contact with the health care system by uninsured adolescents. This reduced contact leads to reduced opportunities to identify early and intervene promptly in a wide range of problems experienced by adolescents, such as teen pregnancy, substance abuse, depression/mental illness, persistent learning disorders, etc. In short, uninsured older children are quite likely to lack contact with the health care system until a health crisis occurs that sends them into a pattern of episodic care. It would seem that early treatment in all age groups for prevalent illnesses such as asthma, allergic response, eye conditions, dental abnormalities, and acute episodes would make sound economic as well as social sense.
This research was supported through a grant provided by the Western Pennsylvania Caring Foundation for Children, an affiliate of Highmark Blue Cross Blue Shield.
We thank Nancy Santangelo, Kathy Docherty, Susan Clark, Susan Edwards, Margaret Vescio, and Nancy Zimmerman for their contribution to the data collection for this study. We also thank the staff at the Caring Foundation for Children.
- Received November 13, 1998.
- Accepted April 6, 1999.
Reprint requests to (C.R.K.) 211 Parran Hall, 130 DeSoto St, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA 15261. E-mail: email@example.com
- Monheit AC,
- Cunningham PJ
- Weinick RM,
- Weigers ME,
- Cohen JE
- ↵Cook T, Campbell DT. Quasi-Experimentation. Chicago, IL: Rand McNally College Publishing Co; 1979
- ↵Campbell DT, Stanley J. Experimental and Quasi-Experimental Designs for Research. Chicago, IL: Rand McNally College Publishing Co; 1966
- ↵American Academy of Pediatrics. Appendix C. Parent and child guide to pediatric visits. In: Guidelines for Health Supervision III. 3rd ed. Elk Grove Village, IL: American Academy of Pediatrics; 1997:233–242
- ↵SPSS Inc. Statistical Package for the Social Sciences, Version 7.5. Chicago, IL: SPSS Inc; 1996
- ↵Edwards AL. Experimental Design in Psychological Research. 5th ed. New York, NY: Harper and Row Publishers; 1985
- Employee Benefit Research Institute
- Bales, SN
- Copyright © 1999 American Academy of Pediatrics