OBJECTIVE. Our objective was to determine whether parent-reported, high-quality primary care was associated with decreased nonurgent pediatric emergency department utilization.
METHODS. A retrospective analysis of prospectively collected data for a cohort of children from the 2000–2001 and 2001–2002 Medical Expenditure Panel Survey panels was performed. Baseline parent-reported quality of primary care with respect to family-centeredness, timeliness, and realized access (a measure of the child's ability to receive necessary care and referrals) was assessed by using composite scores from the Consumer Assessment of Healthcare Providers and Systems survey. The primary outcomes were the numbers of subsequent nonurgent and urgent emergency department visits per child.
RESULTS. Of 8823 children included, 70.0% rated family-centeredness, 88.2% rated realized access, and 55.6% rated timeliness as high quality. After adjustment for demographic factors and health status, high-quality family-centeredness was associated with a 42% reduction in nonurgent emergency department visits for publicly insured children and a 49% reduction for children ≤2 years of age. Greater realized access was associated with a 44% reduction in nonurgent emergency department visits for children 3 to 11 years of age and a 56% reduction for children ≥12 years of age. Greater realized access was also associated with decreased nonurgent emergency department visits for publicly and privately insured children (37% and 35%, respectively). There was no significant association between timeliness and nonurgent emergency department utilization, nor was any quality-of-care domain associated with urgent emergency department utilization.
CONCLUSIONS. Parent-reported, high-quality family-centeredness and a high level of realized access to primary care were associated with decreased subsequent nonurgent emergency department visits for children. Parent reports of health care quality in these domains provide important complementary information on health care quality.
There are >25 million emergency department (ED) visits by children in the United States each year,1 with estimates of the proportion of these visits being for nonurgent conditions ranging from 37% to 60%.2–5 These nonurgent ED visits are associated with ED overcrowding,6–8 cost more than primary care provider visits,9–11 and fragment the care of children, leading to worse health outcomes.12,13 For urgent conditions, care should generally be sought in the ED; however, the large number of nonurgent ED visits is concerning, because such visits often are the result of decreased access to other sources of care.14–16
Evidence shows that there is an association between quality of primary care and ED utilization, with lower quality of care being associated with increased utilization. However, the quality of care evaluated to date focuses on processes of care,17 such as the timely receipt of immunizations or the prescribing of asthma controller medications.18–21 Although these process-of-care measures are important, we think that they fall short of a comprehensive assessment of health care quality, and we postulate that patient-reported quality domains might provide complementary information for assessment of health care quality.
Of the 6 quality-of-care domains (safety, effectiveness, patient- or family-centeredness, timeliness, efficiency, and equity) highlighted for improvement in the Institute of Medicine report Crossing the Quality Chasm,22 many are domains for which patient assessment is an important measure. The authors of the Institute of Medicine report thought that “patients' experiences should be the fundamental source of the definition of quality”23; however, patient experiences with care have largely been unexplored, in part because of the lack of means for measuring patient-reported quality of care for these domains.
The development of the Consumer Assessment of Healthcare Providers and Systems (CAHPS) (formerly the Consumer Assessment of Health Plans Study), with a parent-administered survey to evaluate quality of care for children, provides an instrument that assesses patient experiences with care validly.24 The CAHPS survey specifically addresses, from the patient or parent perspective, the family-centeredness, timeliness, and certain aspects of the effectiveness of care received from a primary care provider.25–28
The objective of the study was to determine the association between parent- reported quality of care in specific domains and subsequent ED utilization by children. We hypothesized that high-quality primary care, as indicated by parent-reported family-centeredness, timeliness, and effectiveness, would be associated with a decrease in subsequent nonurgent pediatric ED visits, without an associated decrease in urgent ED visits.
Study Design and Source of Data
This was a retrospective analysis of prospectively collected data for the cohort of children (<17 years of age) in the 2000 to 2001 (panel 5) and 2001 to 2002 (panel 6) Medical Expenditure Panel Survey (MEPS) panels. MEPS, a subsample of the National Health Interview Survey, is a nationally representative survey of the US civilian noninstitutionalized population conducted annually by the Agency for Healthcare Research and Quality. Data are collected on the specific health services that US residents use and how frequently they use them.29 MEPS uses a longitudinal system of 5 computer-assisted, in-person interviews over a 2-year period to collect household data, including demographic features, perceived health status, screener questions for children with special health care needs,30 insurance status, and ED, primary care, and hospital utilization.29 In 2000, MEPS added CAHPS questions to the second round of interview questions, which allowed analysis of parent-reported quality of care and subsequent child health care utilization in the same national database.31
All children for whom a caregiver completed the CAHPS questions (>96% response rate32) were eligible. Children were excluded if complete follow-up information was not available for the subsequent 3 rounds of interviews; this would occur if the child left the household, was institutionalized, died, or joined the military.
The primary outcomes were the numbers of parent-reported nonurgent and urgent ED visits per child. Visit urgency was determined on the basis of the parent-reported resources used during the visit. This method of assigning urgency was used previously in large database studies and is the database research method preferred by the Ambulatory Pediatric Association Pediatric Emergency Medicine Special Interest Group.3,33,34 Exactly as in a previous study, any visit that resulted in laboratory work, a radiograph, an electrocardiogram, an electroencephalogram, or admission to the hospital was considered urgent; all others were classified as nonurgent.33 The dates for all ED visits were recorded, and only those that occurred after the date of completion of the second interview (when the CAHPS responses were collected) were included in the analysis; this ensured that all ED visits occurred after completion of the CAHPS survey.
Assessment of Health Care Quality Domains
The quality of primary care received by the child was assessed with the CAHPS questions. The 9 quality-of-care questions were grouped into composite scores for the 3 quality domains, that is, (1) family-centeredness, (2) timeliness, and (3) “realized access” to care.35 Realized access is an aspect of health care effectiveness that addresses specifically whether families perceive problems in obtaining necessary care or referrals. The individual questions constituting each composite, and their Likert scales, are listed in Table 1.
The composite scores for each child were obtained by averaging the individual answered questions that constitute the quality-of-care composites, with higher scores corresponding to higher quality. These composite scores and question groupings were developed by the Agency for Healthcare Research and Quality, with 9 of the 10 original individual questions being added to the MEPS survey. To be consistent with previous research, the timeliness and family-centeredness composites, both scored on a 4-point Likert scale, were dichotomized, with scores of >3.5 indicating the highest-quality care and scores of ≤3.5 indicating lower-quality care.36 For realized access, scored on a 3-point Likert scale, scores were dichotomized as 3 vs <3.
Other patient/family characteristics extracted from the MEPS data set included age (categorized as 0–2 years, 3–11 years, or 12–17 years), gender, race/ethnicity (non-Latino white, Latino, non-Latino black, or other, which included Asian/Pacific Islander and Native American/Alaskan native), having a usual source of care (yes or no), having special health care needs (yes or no), interview language (English, Spanish, or both), and parent-reported child health status (excellent or very good versus good, fair, or poor). Multiple indicators of socioeconomic status were included, that is, combined annual family income categorized as a percentage of the federal poverty level and dichotomized as poor or nearly poor (<125% of the federal poverty level) versus higher than nearly poor (≥125% of the federal poverty level), health insurance (public, private, or none), and the highest level of parental educational attainment (dichotomized as less than high school graduate versus high school graduate or beyond).37
Two separate longitudinal cohorts of children were combined for the study, to provide sufficient power to allow for age and insurance subgroup analyses. The 2 cohorts included were the first 2 cohorts within MEPS to have CAHPS responses. Analyses were weighted to yield national estimates for civilian noninstitutionalized children in the United States. The “survey” commands of Stata 9.1 (Stata Corp, College Station, TX) were used for weighting the means and proportions and for the Poisson regression analyses, accounting for the nonrandom sampling strategy used in MEPS.31
Comparisons between the percentage of the population and the percentage of utilization by each demographic subgroup were made by using multiple-comparison-adjusted, survey-weighted tests. Generalized Poisson regression analyses, allowing for incorporation of an additional variance parameter to account for potential overdispersion, were used to assess the association between each composite and the number of subsequent nonurgent and urgent ED visits per child. Results from the regression analyses are presented as the incidence rate ratios (IRRs) for the number of urgent or nonurgent ED visits per child. An IRR of <1 indicates that higher quality is associated with fewer ED visits per child.
Each quality-of-care domain was analyzed in a multivariate, generalized, Poisson regression analysis with each measure of ED utilization. All covariates were entered into the model, grouped as described previously and as listed in Table 2. The groupings were derived by collapsing categories for a covariate when the relationships with ED utilization were similar. Because there were significant interactions between (1) insurance type and the composites and (2) age and the composites, stratified analyses were completed for insurance type (private insurance or public insurance) and age. The stratified regression results for children with no insurance are not presented because of small sample size.
There were 8823 children eligible for the study. Analysis of ED visits revealed that 1419 children (16.1%) made a total of 1786 ED visits during the study period; 768 (43%) of those visits were classified as nonurgent. The numbers of nonurgent visits ranged from 1 nonurgent visit for 500 children to 10 nonurgent visits for 1 child.
The characteristics of the children and the percentages of children with ≥1 urgent ED visit and ≥1 nonurgent ED visit according to characteristic are shown in Table 2. The youngest children, for example, while constituting only 17.3% of the population, accounted for 17.9% of children making urgent ED visits but 34.8% of children making nonurgent ED visits. Therefore, the youngest children were overrepresented with respect to children making nonurgent ED visits, whereas older children were underrepresented. Children with public insurance, while constituting only 25.0% of the population, were overrepresented with respect to both urgent and nonurgent ED use, and uninsured children were underrepresented with respect to urgent ED use. Other characteristics associated with a disproportionately large number of children with nonurgent ED visits included being male, being poor/nearly poor, having special health care needs, having worse parent-reported health status, having a usual source of care, and having a lower degree of family educational attainment.
Assessment of parent-reported, quality-of-care domains showed that 70% of parents rated the family-centeredness of their child's primary care as high quality, approximately one half rated timeliness as high quality, and almost 90% rated realized access as high quality (Table 3). For both family-centeredness and realized access, high-quality care was associated with a lower number of children with ≥1 nonurgent ED visit. For timeliness, the percentages of children with urgent and nonurgent visits were similar, with the percentage of children with urgent visits being significantly less than the percentage reporting high-quality care.
In univariate regression analyses, high-quality family-centered care (IRR: 0.64; 95% confidence interval [CI]: 0.50–0.82) and greater reports of realized access (IRR: 0.53; 95% CI: 0.37–0.78) were both associated with significant decreases in subsequent nonurgent ED visits per child. An IRR of 0.64 translates into a 36% reduction in nonurgent ED visits, with an IRR of 0.53 being equal to a 47% reduction in visits. There was no association between timeliness and nonurgent ED visits, nor were any of the 3 quality-of-care domains associated with decreased numbers of subsequent urgent ED visits (not shown).
To adjust for potential confounding by child and family characteristics, a multivariate, generalized, Poisson regression model was constructed. Because of significant interactions between the quality-of-care domains and both insurance type and age, the analysis was stratified according to those factors. Analysis of publicly insured children revealed that those with high-quality family-centeredness and realized access had 42% and 37% decreases in nonurgent ED visits per child, respectively (shown in Table 4 by IRRs with CIs not including 1). In contrast, high-quality timeliness was not associated with children making urgent or nonurgent visits. Among children with private insurance (Table 4), only high-quality realized access was associated with fewer nonurgent ED visits by children. High-quality care was not associated with decreased urgent ED utilization in any domain.
Analyses stratified according to age (Table 5) again showed no significant association between the specific quality-of-care domains and urgent ED visits. Among the youngest children, high-quality family-centered care was associated with a substantial decrease in nonurgent ED visits per child; for older children, greater realized access was associated with fewer nonurgent ED visits (Table 5).
In addition to the quality-of-care composites, 2 other factors, namely, being a child with special health care needs and worse parent-reported child health status, were associated consistently with ED utilization in all multivariate models. In the model for children with public insurance coverage, being a child with special health care needs was associated with increases in both urgent (IRR: 2.07; 95% CI: 1.51–2.86) and nonurgent (IRR: 1.73; 95% CI: 1.39–2.64) ED visits. Worse parent-reported child health status was associated only with an increase in nonurgent ED visits (IRR: 2.27; 95% CI: 1.32–3.93). These associations were similar in other models. None of the other characteristics, including gender, race/ethnicity, income, having a usual source of care, parental educational attainment, and interview language, showed consistent association with ED utilization.
The study findings indicated that parent-reported, high-quality, family-centered care and increased realized access were both associated with decreased numbers of subsequent nonurgent ED visits. For the 2 groups known from previous research to have disproportionately high ED utilization, that is, children ≤2 years of age and those with public insurance, high-quality family-centered care was associated with 40% to 50% reductions in subsequent nonurgent ED utilization, as evidenced by IRRs of 0.51 and 0.58, respectively. For older children and for children with private insurance, realized access (the ability to obtain referrals or other care that the doctor or family thought was necessary) was associated most strongly with decreased nonurgent ED visits. There were no statistically significant associations between any of the quality-of-care composites and urgent ED utilization, as evidenced by the CIs including 1.0. This lack of an association was expected, given that care for urgent conditions is less discretionary.
The association between high-quality family-centered care and nonurgent ED utilization for the youngest children and for publicly insured children is of interest. Children ≤2 years of age develop more acute illnesses and therefore access the health care system more frequently, for both primary care and ED visits, than do older children.38,39 When nonurgent conditions arise among these youngest children, parents who think that their children have access to high-quality family-centered care (primary care providers who respect what the parents have to say, who listen carefully, and who spend enough time with the children) may choose to have the children seen by that provider, rather than by a stranger in the ED. The content of the high-quality communication between the parents and the primary care providers may also play a role, with providers who practice family-centered care possibly explaining to families when ED care is warranted and when it is better to wait to be seen.
For children with public insurance, the issue of access to other sources of care becomes paramount. When transportation issues or financial difficulties make accessing the primary care provider more difficult, parents may use the ED for nonurgent conditions unless they perceive the loss of continuity with the primary care provider to be detrimental. The bond with the provider who delivers high-quality family-centered care may be strong enough to offset the short-term advantage of convenient ED care.
Nonurgent ED visits were less common for older children and children with private insurance than for younger children and children with public insurance. It seems that parental perceptions that the provider is not able to meet the needs of the child are more important in determining to seek ED care than the bond between the provider and the family. Whether this results from a provider actually not being able to meet the needs of the child or is simply the parent's perception of the provider's inability cannot be determined from this study. In either case, realized access was associated strongly with nonurgent utilization in these populations.
We had hypothesized that timely access to primary care would be associated with decreased nonurgent ED use, but we found no significant associations. Two explanations may be operative. First, the timeliness composite is composed of questions that assess care across many domains, including well-child care, sick care, and after-hours care, only some of which may be related to nonurgent ED utilization. Second, it may be that timely access alone is not sufficient to decrease ED utilization but the past interactions and content of previous discussions between the provider and the family dominate the family's decision to seek ED care.
The lack of an association between parent-reported quality of care and urgent ED visits also warrants exploration. As hypothesized, the quality of interactions with the primary care provider had no significant association with urgent ED visits. Studies that rely on a decrease in total ED utilization (approximately one half of which may not be discretionary) as the sole outcome may miss important reductions in nonurgent visits and may assume incorrectly that proposed interventions are not effective.
In addition to the quality domains evaluated, other covariates were associated with increased nonurgent ED visits. The findings that worse reported health status, having special health care needs, and having lower income were associated with increased ED visits are consistent with previous studies.40–42 The bivariate finding that having a usual source of care was associated with increased nonurgent ED visits is not as intuitively clear. One possible explanation is that children who lack insurance frequently lack a usual source of care; this population utilizes health care, including the ED, less frequently than do children with insurance.43 This explanation is supported by the finding that only items relating to an increased level of baseline illness (health status and special health care needs) remained significant after adjustment.
Certain study limitations should be noted. First, as with all MEPS data, the family reported the number of ED visits; there was no independent verification, which allowed for possible inaccurate reporting of data. The MEPS data, however, are collected prospectively, and the results are population based, with a large sample size. The 96% response rate for the CAHPS survey also supports the generalizability of the results. Second, the definition of urgency of ED visits was based on the resources used during the ED visit. Therefore, some visits might have been misclassified, especially those for conditions such as asthma exacerbation, for which therapeutic rather than diagnostic interventions would have resulted in classification as urgent. Classification based on resource utilization has been used in previous research, with criteria identical to those used in this study.3,33,34 Finally, we did not have a sufficient sample size for uninsured children to address adequately the association between quality of care and ED utilization in this population.
In the case of pediatric ED utilization, it is the parent who decides when and where to seek care for a child. Therefore, it is only by understanding the parent's perspective on the relationship between the family and the primary care provider that one can understand the decisions that families make when deciding where and when to seek care. The assessment of patient- or parent-reported quality of care, although not itself a comprehensive measure of health care quality, provides important complementary information about the quality of care that children receive. In addition, parent-reported quality of care in these specific domains may aid in the identification of children at increased risk for nonurgent ED visits and may allow for targeted interventions aimed at improving the care that such children receive.
This is the first study, to our knowledge, to show that parent-reported, high-quality family-centeredness and ability to receive needed care are associated with decreased nonurgent ED visits by children. We think that the addition of these parent-reported measures of child health care quality to the process or outcome measures of quality yields a more-comprehensive measure of the quality of care children that receive, at least with regard to risk factors for nonurgent ED utilization, and can aid in the identification of children and families receiving lower-quality care.
This work was supported in part by grant K08 HS015482-01A1 (to Dr Brousseau) from the Agency for Healthcare Research and Quality.
- Accepted January 22, 2007.
- Address correspondence to David C. Brousseau, MD, MS, Department of Pediatrics, Medical College of Wisconsin, CCC 550, 999 N 92nd St, Milwaukee, WI 53226. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵McCaig LF, Nawar EW. National Hospital Ambulatory Medical Care Survey: 2004 emergency department summary. Adv Data.2006;(372) :1– 30
- Guttman N, Nelson MS, Zimmerman DR. When the visit to the emergency department is medically nonurgent: provider ideologies and patient advice. Qual Health Res.2001;11 :161– 178
- ↵Christakis DA, Mell L, Koepsell TD, Zimmerman FJ, Connell FA. Association of lower continuity of care with greater risk of emergency department use and hospitalization in children. Pediatrics.2001;107 :524– 529
- Homer CJ, Szilagyi P, Rodewald L, et al. Does quality of care affect rates of hospitalization for childhood asthma? Pediatrics.1996;98 :18– 23
- Adams RJ, Fuhlbrigge A, Finkelstein JA, et al. Impact of inhaled antiinflammatory therapy on hospitalization and emergency department visits for children with asthma. Pediatrics.2001;107 :706– 711
- ↵Ritzwoller DP, Bridges CB, Shetterly S, Yamasaki K, Kolczak M, France EK. Effectiveness of the 2003–2004 influenza vaccine among children 6 months to 8 years of age, with 1 vs 2 doses. Pediatrics.2005;116 :153– 159
- ↵Institute of Medicine. Crossing the Quality Chasm: A New Health System for the Twenty-First Century. Washington, DC: National Academy Press; 2001
- ↵Berwick DM. A user's manual for the IOM's “Quality Chasm” report. Health Aff (Millwood).2002;21 :80– 90
- ↵Agency for Healthcare Research and Quality. National Healthcare Disparities Report: Prepublication Copy. Rockville, MD: US Department of Health and Human Services; 2003
- Mayer ML, Skinner AC, Slifkin RT. Unmet need for routine and specialty care: data from the National Survey of Children With Special Health Care Needs. Pediatrics.2004;113(2) . Available at: www.pediatrics.org/cgi/content/full/113/2/e109
- ↵Flores G, Olson L, Tomany-Korman S. Racial and ethnic disparities in early childhood health and health care. Pediatrics.2005;115(2) . Available at: www.pediatrics.org/cgi/content/full/115/2/e183
- ↵Agency for Healthcare Research and Quality. Questionnaire sections. Available at: http://meps.ahrq.gov/mepsweb/survey_comp/survey_results_ques_sections.jsp?Section=AC&Year1=2000&Submit1=Search. Accessed April 2, 2007
- ↵Agency for Healthcare Research and Quality. Weights and response rates for the self administered/parent administered questionnaire. Available at: http://meps.ahrq.gov/mepsweb/data_stats/download_data/pufs/h50/h50doc.shtml#C35WRSelfPareAdmini. Accessed April 2, 2007
- ↵Agency for Healthcare Research and Quality. Medical Expenditure Panel Survey. Available at: www.meps.ahrq.gov/PufFiles/H50/H50doc.pdf. Accessed September 3, 2006
- ↵Mistry RD, Hoffmann RG, Yauck JS, Brousseau DC. Association between parental and childhood emergency department utilization. Pediatrics.2005;115(2) . Available at: www.pediatrics.org/cgi/content/full/115/2/e147
- ↵McCaig L, Burt C. National Hospital Ambulatory Medical Care Survey: 2003 emergency department summary. Adv Data.2005;(358) :1– 40
- ↵Hing E, Cherry D, Woodwell D. National Ambulatory Medical Care Survey: 2004 summary. Adv Data.2006;(374) :1– 33
- ↵Fosarelli PD, DeAngelis C, Mellits ED. Health services use by children enrolled in a hospital-based primary care clinic: a longitudinal perspective. Pediatrics.1987;79 :196– 202
- ↵National Center for Health Statistics. Health, United States, 2005 With Chartbook on Trends in the Health of Americans. Hyattsville, MD: National Center for Health Statistics; 2005
- Copyright © 2007 by the American Academy of Pediatrics