Objective. The objective of this study was to examine the association between continuity of care (COC) and emergency department (ED) utilization during infancy, when the bonds between a provider and a family are first forming and health care utilization is at it highest levels.
Methods. A retrospective birth cohort was conducted of children who were born between December 1, 1999, and April 30, 2000; health care use during the first 25 months of life was analyzed. A COC score was calculated from well-child visits in the first 7 months of life. Subsequent ED utilization, for both a 12-month and an 18-month period, was evaluated through 2 measures: 1) ED reliance (the fraction of all visits that occurred in the ED) and 2) number of ED visits. Spearman rank correlation coefficients (rs) between COC and ED utilization were calculated. Partial rank correlations were calculated controlling for Medicaid status, income, gender, and the total number of health care visits.
Results. A total of 181 children were included in the study; 96 (53%) were male, and 48 (27%) were covered by Medicaid. COC scores ranged from 0 to 1, with a median of 1. COC scores were negatively correlated with both ED reliance (rs = −.214) and number of ED visits (rs = −.215) with 12 months of follow-up. The negative correlation was even stronger at 18 months of follow-up (ED reliance: rs = −.247; number of ED visits: rs = −.242), and this relationship remained significant with partial rank correlations.
Conclusions. Improved COC in infancy is associated with a decrease in subsequent ED utilization. Interventions aimed at increasing the continuity of early well-child visits may decrease ED utilization.
Continuity of care (COC) with a primary care provider has been shown to have many positive effects for the health care of children. Children with higher levels of continuity are more likely to receive measles-mumps-rubella vaccination in a timely manner,1 and individuals who have diabetes and increased continuity are less likely to have diabetic ketoacidosis.2 Provider continuity has also been shown to extend through a mother-child unit, with the children of mothers who received their care with the same provider who delivered the prenatal care having higher rates of immunizations than those followed at different sites.3
The effects of COC on pediatric emergency department (ED) utilization have been analyzed with conflicting results. An initial study showed no decreased ED utilization based on COC in a resident continuity clinic,4 but a subsequent study showed that increased COC resulted in decreased ED utilization and fewer hospitalizations for a large cohort of children.5 The children in this study averaged slightly over 5 years of age, and only 3% were covered by Medicaid.
No study to date has evaluated ED utilization on the basis of COC in the infant period, when scheduled interactions with a primary care provider are most frequent and ED utilization is at its highest levels.6,7 Patterns of ED utilization have previously been shown to be established as early as the first year of life,8 increasing the importance of analyzing the relationship during infancy. We hypothesized that increased COC with a primary care provider during infancy is associated with decreased subsequent ED utilization during early childhood.
This study used claims data for a retrospective birth cohort to analyze the association between infant COC and ED utilization during subsequent follow-up periods of 12 and 18 months.
Children who were born between December 1, 1999, and April 30, 2000, were eligible for inclusion when they received their primary care at a Children’s Medical Group site and were continuously enrolled in UnitedHealthcare of Wisconsin. “Continuously enrolled” was defined as no gap in insurance coverage of >1 month throughout the 7-month initial period and the 12-month follow-up period. UnitedHealthcare of Wisconsin is one of the largest insurers in Southeastern Wisconsin and provides both commercial and Medicaid coverage for children across Wisconsin. Children’s Medical Group is a group of affiliated pediatric practices located in the greater Milwaukee metropolitan area. No practices have financial incentives with regard to ED utilization. There were 30 pediatricians practicing at 11 different Children’s Medical Group clinic sites during the time of the study. Data were analyzed for encounters from December 1, 1999, until June 30, 2002. The study was reviewed and classified as exempt by the Institutional Review Board.
Continuity of Care
A child’s COC score was calculated on the basis of the primary care providers seen for well-child visits during the first 7 months of life. Seven months was chosen as the cutoff to allow for the occurrence of the American Academy of Pediatrics recommended 6-month well-child visit between the sixth and seventh months of life. We used only well-child visits in the COC score because of the belief that the bond between a provider and a family takes time to develop; therefore, visits to another provider during the first 7 months of life may not reflect the bond between child and provider as those visits may have occurred before even a single visit with the primary care provider.
Several measures of continuity have been described previously; we used the COC score. This score was first described by Bice and Boxerman9 but has subsequently been used to assess the effect of childhood continuity on ED utilization and hospitalizations by Christakis et al.5 The score is based on both the number of health care visits and the number of different individual providers seen for those visits. A score of 0 reflects a child’s seeing a different provider at each visit, and perfect score of 1 reflects the child’s seeing the same individual provider for all visits. The exact details of the scoring system have been described previously.9 As the stability of the COC score is substantially stronger with an increased number of encounters, only children with 3 or more well-child visits were included in the analysis.
Main Outcome Measures
The primary outcome measure was ED utilization in the subsequent 12 months (ie, from the 8th through the 19th months of life). Two measures of ED utilization were calculated: 1) the number of ED visits and 2) “ED reliance,” defined as the percentage of all health care visits that occur in the ED.10 ED visits were determined by Current Procedural Terminology (CPT) coding from a UnitedHealthcare administrative data set. CPT codes for urgent care visits were separated from primary care provider visits by a combination of provider code and location code and were included with ED visits for the analysis. CPT codes for both well-child and sick-child visits to a primary care provider were included in the analysis as visits to a primary care physician. For all analyses of CPT codes, only professional claims from the physician (HCFA-1500 claims) were used. All visits, both to the ED and to the primary care provider, were coded to a particular physician, not to a practice or a group. The initial 12-month follow-up period was extended to 18 months, and the analysis was repeated to evaluate the extent to which continuity continues to predict ED utilization as the child ages.
To account for the skewed nature of the data, Spearman rank correlation coefficients (rs) were used to examine the relationship between the COC score and both the number of ED visits and ED reliance. The relationship between COC and both measures of ED utilization was further analyzed by partial rank correlation to control for the effects of income level (as estimated by median income by zip code of residence), gender, Medicaid versus commercial insurance status (serving as a marker for both income level and the existence of a copay for those with commercial insurance), and the total number of visits made to the health care system. “Total visits” has previously been used as a proxy for severity of illness or propensity to seek care,5 both of which would be associated with increased ED utilization. In the final analysis, only total visits and Medicaid status remained significantly related to ED utilization and were included in the model; income was strongly correlated with Medicaid status, and the effect of gender was no longer significant. Although seemingly a component of ED reliance, the total number of visits to the health care system was used in the adjusted analysis for both ED utilization outcomes to separate children whose ED reliance scores were similar but had different utilization patterns. For example, a child with 1 of 2 health care visits to the ED could be distinguished from a child with 4 of 8 visits to the ED.
Children were further classified by whether they had a chronic illness. Chronic illness was defined as at least 2 visits (to either the primary care physician or the ED) during the first 19 months of life with an International Classification of Diseases, Ninth Revision diagnosis of any of the following: 1) wheezing/asthma, 2) sickle cell disease, 3) seizure, 4) diabetes, or 5) cerebral palsy. When this threshold was not met, the child was classified as having no chronic illness.
A total of 181 children fulfilled the eligibility criteria. Sixteen children were excluded from an initial 197 children secondary to a lack of continuous enrollment. Of the 181 children, 96 (53%) were male and 48 (27%) were covered by Medicaid. Only 4 (2%) children were classified as having a chronic illness. During the 12-month follow-up period, 61 (34%) children used the ED at least once, for a total of 117 ED visits, yielding an overall rate of ED use of 64.6 visits/100 children per year. The characteristics of the entire cohort and the ED users are shown in Table 1. ED users were more likely to be covered by Medicaid, had a slightly higher percentage of boys, had fewer children who saw the same provider for all well-child visits, and had higher health care utilization.
Analysis of primary care site utilization during the 12-month follow-up period revealed a total of 930 sick-child visits to primary care provider sites by 157 (87%) children and a total of 480 well-child visits by 173 children. Therefore, of all recorded visits, 117 (11.2%) of 1047 of all sick visits and 117 (7.7%) of 1527 of all physician visits occurred in the ED.
Evaluation of the association between COC and the measures of ED utilization for the 12-month follow-up period revealed significant negative correlations between increasing COC and subsequent ED reliance (rs = −.214, P < .01) and the subsequent number of ED visits (rs = −.215, P < .01). The association between COC and both measures of ED utilization remained significant after controlling for Medicaid status and the total number of health care visits (Table 2).
The analyses were repeated on the same cohort with 18 months of follow-up to examine for persistence of the association. Of the 120 children who did not use the ED during the initial 12-month follow-up period, only 8 used the ED over the next 6 months. Analysis of the relationship between COC and ED utilization for the 18-month follow-up period showed a stronger negative correlation between initial COC and ED reliance (rs = −.248, P < .001) and the number of ED visits (rs = −.242, P = .001). After adjustment, the rank correlation coefficients showed a stronger association than those in the adjusted 12-month analysis, and they remained significantly negatively correlated with COC (Table 2). The addition of an indicator of chronic illness for the 4 children who were classified as having a chronic illness had no effect on the model.
The results of this study suggest that increased COC with a single primary care provider during early infancy is associated with decreased ED utilization during the first 2 years of life. This association was seen in the analyses of both ED reliance and the number of ED visits during the follow-up periods. Previous studies have shown an association between COC and ED utilization, but this is the first study to evaluate this effect during infancy, when the bond between a family and a pediatrician is first being formed and ED utilization is at its highest level.
The majority of continuity of care scores were 1 for both the entire cohort and the subset of children who used the ED during the follow-up period. This reported level of continuity is higher than in the previous cohort study evaluating COC and pediatric ED utilization,5 presumably as a result of the relatively short period of time between infant appointments and the ability to predict when future appointments are needed. However, even with the high levels of continuity for this cohort, which would make an association more difficult to find, a significant negative correlation was found with subsequent ED utilization.
ED utilization for this cohort was consistent with previously published reports of ED utilization by young children. Our ED utilization rate of 65 visits/100 child-years was slightly less than national rates of 74 visits/100 child-years for children younger than 3 years.11 The proportion of all health care visits that occurred in the ED (8%) is slightly less than the 10% reported previously.12 These lower numbers could be attributable to the presence of a primary care provider and insurance coverage for all children. Even with these restrictions, however, COC remained a significant predictor of ED utilization.
Levels of chronic illness for our study were low, as would be expected for an analysis of children this young. The use of only 5 International Classification of Diseases, Ninth Revision classifications may have also slightly limited the number. Controlling for the presence of chronic illness had no effect on the relationship between COC and ED utilization.
The primary limitation of our study is the potential that other factors contributed to the ED utilization that was found in our cohort. Information on race/ethnicity was not available in the data set, and some previous studies have found race/ethnicity to be a risk factor for ED utilization.13–15 Other studies have disputed an independent effect of race, stating that the effect of race/ethnicity is no longer present after controlling for insurance status and income. We were able to control for both of these factors, but residual confounding may still exist. In addition, although our COC score began with the child’s birth, it is possible that a previous child from the same family may have partially determined the utilization pattern of a child in our cohort. Information about family size and composition was not available from the data set.
Additional limitations in this study are related to the analysis of existing data sets. We used only HCFA-1500 claims for our visit extraction. We therefore would have missed visits that incurred no physician charge. Although we believe this number to be low, it is still a potential area of concern. In addition, we do not have information about the practice characteristics that may affect ED utilization.
In summary, we conclude that increased COC with a primary care provider was associated with decreased ED utilization in the first 2 years of life. Future efforts aimed at ensuring that families have access not only to a site of care but also to a specific primary provider may decrease ED utilization by young children.
This study was funded by a grant from the Children’s Hospital of Wisconsin Foundation and was done in collaboration with Tom Dunigan, MD.
- Received March 24, 2003.
- Accepted July 28, 2003.
- Reprint requests to (D.C.B.) Children’s Hospital of Wisconsin, MS 677, 9000 W Wisconsin Ave, Milwaukee, WI 53226. E-mail:
- ↵Christakis DA, Wright JA, Koepsell TD, Emerson S, Connell FA. Is greater continuity of care associated with less emergency department utilization? Pediatrics.1999;103(suppl) :738– 742
- ↵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
- ↵McCaig LF, Burt CW. National Hospital Ambulatory Medical Care Survey: 1999 Emergency Department Summary. Adv Data.2001;(320) :1– 34
- ↵Feigelman S, Duggan AK, Bazell CM, Baumgardner RA, Mellits ED, DeAngelis C. Correlates of emergency room utilization in the first year of life. Clin Pediatr (Phila).1990;29 :698– 705
- ↵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
- ↵Alessandrini E, Shaw K, Bilker W, Perry K, Baker D, Schwarz D. Effects of Medicaid managed care on health care use: infant emergency department and ambulatory services. Pediatrics.2001;108 :103– 109
- ↵McCaig L, Ly N. National Hospital Ambulatory Medical Care Survey: 2000 emergency department summary. Adv Data.2002;(327) :1– 27
- Copyright © 2004 by the American Academy of Pediatrics