Relationship Between Early Primary Care and Emergency Department Use in Early Infancy by the Medicaid Population FREE

PERSONS IN lower-income groups, especially poor families with young children, tend to use primary care at much lower rates than other groups.^{1 - 10} Failure to establish linkage with a primary care physician can have a negative influence on both the quality and cost of pediatric health care.¹¹ Families without a primary care physician rely much more heavily on hospital emergency departments (EDs) for infant medical care and, therefore, tend to make more nonurgent visits to emergency facilities.^{12 - 13} The patterns of use of both primary care and emergency facilities can be established as early as the first year of life.¹²

Establishing a stable and consistent relationship with a primary care physician can provide families with continuity of care, advice, and support, and education on infant care.¹⁴ This concept, known as a medical home, attempts to address the complete needs of the patient.^{15 - 16} In contrast with primary care, ED visits usually focus on the presenting report, resulting in a lack of comprehensive assessment and preventive care, both of which are particularly important for young children.^{9 ,17}

The use of the ED as a site of routine care has been noted in several studies and has been related to maternal perception of infant health and maternal and social variables as well as barriers to primary care use, such as lack of a medical home, transportation needs, and child care issues.^{1 - 2 ,18 - 36} In previous studies at a single urban hospital in Ohio, we noted that up to 40% of all newborns discharged who were uninsured or receiving Medicaid used the ED for health care needs at least once in the first 3 months^{37 - 38} of life and 14% visited the ED before their first primary care visit.³⁷

This study was undertaken to extend this research to a larger population. We examined the use, characteristics, and timing of the first primary care visit in relation to ED use by healthy infants in the Medicaid population throughout Ohio. We hypothesized that early linkage to primary care would be associated with decreased ED use.

We conducted a retrospective cohort study using Medicaid claims data linked to vital statistics files from the State of Ohio for fiscal years 1992 through 1998. This combined data set provided information on sociodemographic characteristics of infants and mothers, along with date of birth, date of discharge from the hospital, dates of primary care and ED visits, sites where health care was provided, and diagnoses and procedures performed during the birth hospitalization. This study was approved by the institutional review board; informed consent was not required.

The source population for this study was all births in Ohio from July 1, 1991 through June 30, 1998. From this larger group, a subset of infants and their mothers were selected who were Ohio Medicaid recipients, had a valid Medicaid birth claim, and were enrolled in the Medicaid program for at least 91 continuous days after the infant's birth. Infants born to mothers enrolled in Medicaid health maintenance organization plans (approximately 6% of the population) were excluded from the analysis owing to incomplete reporting. To isolate the relationship between primary care and ED use and to eliminate confusion related to predisposing illness, we restricted the study sample to healthy full-term neonates. Healthy full-term neonates in this group were identified by the following 3 criteria: diagnosis related group 391 (normal newborn), birth weight of 2000 g or more, and gestational age of 37 weeks or longer. These criteria have been used in other studies of postneonatal care.³⁹

The primary outcome of interest was the occurrence of an ED visit within 91 days of birth. We chose to examine ED use within 91 days of birth because, in this data set, more than half of the infants who had any ED visit by 1 year of age had their first visit within 91 days of birth. Emergency department visits were identified by examining the Clinical Procedural Terminology (CPT) codes in the outpatient-physician claims. The CPT codes used to identify an ED visit were categorized as evaluation and management—ED services.⁴⁰

The primary independent variable studied was the establishment of primary care linkage. Primary care linkage was defined as the occurrence of a visit to a primary care physician within 21 days of the neonate's birth. Primary care visits were selected by examining the CPT codes for each outpatient office visit. The visit was considered primary care if a CPT code in the categories of (1) office visit and evaluation, (2) preventive medicine, or (3) health check visit was present. The office visit and evaluation category consisted of CPT codes under the major heading of Evaluation and Management and the subheadings of Office or other outpatient services, Office or other outpatient consultation, and Confirmatory consultation.⁴⁰ Preventive medicine and health check visits consisted of CPT codes under the major heading of Evaluation and Management and the subheading of Preventive medicine services.

Twenty-one days was chosen to allow for variation in the date of the first visit due to weekends and holidays. This was also verified by the peak of the first primary care visit that occurred around 15 days. The site of a primary care visit was classified as occurring at a community-based or hospital-based clinic location based on the "provider type" code in the Medicaid claim. If the provider type code was "general hospital," the visit was classified as hospital-based.

Confounders that might influence the relationship between primary care linkage and ED use were examined. These included maternal age, marital status, educational level, race, parity, and prenatal care visits. Newborn characteristics examined included year of birth, birth weight, gestational age, delivery route, and length of hospital stay. Jaundice was defined as the diagnosis of jaundice at discharge from the hospital, determined by International Classification of Diseases, Ninth Revision, Clinical Modification code. Year of birth was examined to address changes over time. To address rural vs urban influence, area of residence was stratified by whether the maternal county of residence was within a major metropolitan area of Ohio. Major metropolitan areas were obtained from US Bureau of the Census data.⁴¹ Regional variations in postdischarge health care and ED use over time were determined for the 6 perinatal service-education regions of the State of Ohio (ie, region 1, southwest; region 2, west central; region 3, northwest; region 4, southeast; region 5, northeast; and region 6, east central).⁴²

Bivariate analyses were performed to identify variables associated with hospital-based primary care visits and ED use in early infancy. χ² Tests were used for analysis of categorical variables and t tests were performed for normally distributed continuous variables. Wilcoxon rank sum tests were used for nonnormally distributed data. Multivariate analysis for ED use within 91 days of birth was performed using a logistic regression model. Confounders were entered into the model if they were significant at a P = .05 level on bivariate analysis or if their inclusion changed the coefficient of the primary variable. Birth year was added to the model as a series of indicator variables for each year (1992-1998) with 1991 as the reference value. To examine possible interactions between maternal age and race and maternal age and marital status interaction terms for these variables were entered into the model. Regression diagnostics were used to assure a stable model. Odds ratios and associated 95% confidence limits (CLs) were calculated for the likelihood of visiting the ED within 91 days of birth. All statistical analyses were performed using SAS Statistical Software, Version 8.1 (SAS Institute Inc, Cary, NC).

There were 1 069 693 births recorded in Ohio between July 1, 1991, and June 30, 1998. Of these, 254 074 infants and their mothers were identified as Medicaid recipients 91 days after birth in fee-for-service plans. From that group, 151 464 healthy full-term infants (diagnosis related group, 391; gestational age, ≥37 weeks; and birth weight, ≥2000 g) formed the final study cohort. The general characteristics of the patient population are given in Table 1.

Overall, only 53% of newborns made their first primary care visit within 21 days of birth and only 78% of infants in this population made at least 1 primary care visit within 3 months of the neonates' birth (Figure 1). Most early primary care visits (within 21 days) occurred at community clinics; only 15% occurred at a primary care clinic located within a hospital (Table 1). The proportion of newborns who received a primary care visit within 21 days of birth varied significantly between the 6 state perinatal regions, ranging from 43% to 58% (P<.001).

Mothers who took their infants to hospital-based clinics for primary care clinics for more than 50% of their primary care visits were significantly more likely to be younger, primiparous, and to live in a major metropolitan area of Ohio. They were less likely to be white, married, or to have completed high school, and they were likely to have had significantly fewer prenatal care visits during their pregnancy. They had a significantly shorter mean length of stay at birth and were more likely to have had a home care visit within 21 days of birth.

Of the 151 464 infants examined, over one quarter (Figure 1) had at least 1 ED visit within 3 months of the neonate's birth and 9% had more than 1 visit during this time. The mean age of the infants at the first ED visit was 39.7 days (Table 1). Infants who were taken to a hospital-based primary care clinic for most of their primary care visits were significantly more likely to use the ED. Emergeny department use between the 6 state perinatal regions varied from 25% to 31% (P<.001).

Infants who were taken to the ED within 91 days of birth were more likely to have had a primary care or home care visit within 21 days of birth and were more likely to have gone to hospital-based centers for most of their primary care visits (Table 2). Mothers of infants taken to the ED were significantly more likely to be younger and primiparous and less likely to be married or to have completed high school, but had significantly more prenatal care visits. They were significantly less likely to live in a major metropolitan county of Ohio. Race and a diagnosis of jaundice while in the birth hospital were not statistically different between the 2 groups.

The proportion of infants who had at least 1 ED visit within 91 days of birth increased each year through 1995 and then began to decrease (Table 3), returning to near 1991 baseline values by 1998 (P<.001). In addition, significant regional variation was seen in ED use (P<.001).

Even after adjusting for maternal and infant risk variables, primary care use remained a significant factor affecting ED usage in this age group. Contrary to our expectations, however, having a primary care visit within 21 days of the neonate's birth not only did not decrease ED use but was actually associated with a 16% increase in the likelihood of ED use. When the primary care visit occurred at a hospital-based clinic, there was a similar, but even stronger, independent effect on ED, the likelihood of which increased by almost a third (Table 4).

The maternal and infant factors most significantly associated with an increased risk of ED use within 91 days of the neonate's birth included less maternal education (no high school diploma), singleton birth, birth via cesarean section, and use of prenatal care. Compared with birth in 1991, birth in 1992 through 1996 was also significantly associated with an increased risk of ED use. Emergency department use decreased to baseline values in the later years. Factors associated with a decrease in the risk of ED use included primiparity, longer gestational age, and heavier birth weight. The interaction between increasing maternal age and nonwhite race and the interaction between increasing maternal age and single marital status were both associated with an increase in the odds of ED use. Residence in a nonmetropolitan area was associated with an increased risk of newborn ED use within 91 days of birth. Compared with perinatal region 1, residence in regions 2, 3, 4, and 6 were associated with increased ED use.

To address the possibility that the primary care and ED visits were part of the same care episode, the analysis was also run after excluding all patients with an ED visit prior to 30 days after the neonate's birth, allowing for a 9-day gap between the primary care and ED visit. The result was only a very small change in the adjusted odds ratio for a primary care visit (from 1.170 to 1.165).

We found that only half of the infants in this population had a documented early linkage to primary care. Almost one third of the infants had at least 1 ED visit in early infancy and almost 10% had more than 1 ED visit. Contrary to our expectations, infants who were taken to the ED within 91 days of birth were likely to have had more prenatal care visits and were more likely to have had early primary care linkage. They were more likely to have gone to hospital-based primary care clinics for most of their visits. Even after adjusting for maternal, infant, and residency characteristics, early primary care linkage did not result in a decreased risk of ED use in early infancy.

Several studies have shown that children insured through the Medicaid program use hospital EDs more frequently than privately insured children, particularly for nonurgent reasons.^{5 ,36 ,43} In a study similar to ours, Sharma et al⁴⁴ followed almost 73 000 infants in Missouri from their birth in 1995 through their first birthday and documented similar trends. In their study, Medicaid coverage was the most important predictor of ED use. Other factors associated with ED use included self-pay, black race, rural residency, presence of birth defects, and a nursery stay of longer than 2 days. The Medicaid patients in their study, however, had similar rates of ED use in both urban and rural areas. For our population of Medicaid patients, residence in a nonmetropolitan area was associated with an increased risk of newborn ED use.

Emergency department use increased during the first few years of the study, appeared to peak in 1995, and then began to decrease, returning to baseline levels by 1997. These observations may be related to changes in the timing of discharge of newborns occurring at this time.³⁹

Although many studies have examined the issue, there are no conclusive answers as to why parents routinely seek nonurgent ED care for their children. Reported possibilities include health care beliefs, convenience, socioeconomic issues, role modeling, overestimation of the severity of the illness, failure to understand how and when to access the health care system, and availability of a primary care physician.^{4 ,34 ,45 - 49}

Conventional wisdom states that improvements in primary care services will result in decreases in ED use.⁵⁰ The benefits of improved access to primary care have been suggested in some previous studies. Several have shown a decrease in ED use after establishment of community health centers^{10 ,13 ,51} and others have shown that patients with a regular source of care are less likely to use the ED.²² Gill and Diamond⁵² reported a decrease in ED use simply as a result of referral to a primary care physician. However, many families with a regular source of care continue to exhibit an uncoordinated pattern of seeking care. Glotzer et al⁵³ reported that prior approval requirements did not visibly change how patients used the ED and Gadomski et al⁴³ found that ED visits were not significantly reduced after institution of a Maryland program that involved assignment to a primary care physician with gatekeeping responsibility.

We have previously shown that women with poor prenatal care are less likely to seek ED care for their young infants.⁵⁴ In the current study, mothers who took their infants to the ED also used other sources of health care. They seemed to split their children's health care between a primary care site and the ED or to use the ED as a backup to their usual source of care. Habenstreit⁵⁵ speculated that some patients who have a regular source of care may prefer to use the ED for unexpected illnesses and, since EDs are open all hours, they may serve as the designated acute-care facility when the clinic or physician office is closed.⁵⁶ Even a nonurgent ED visit may be appropriate at a time when no other alternate health care is available.

All studies using administrative data, such as ours, are limited by the accuracy and completeness of the unaudited claims data.^{57 - 58} In addition, studies using large data sets, such as ours, often detect differences that are statistically significant but not clinically relevant. Despite this caution, large differences were observed for 5 important characteristics that appear to be significant for ED use in early infancy. These include an early primary care visit, primary care at a hospital-based clinic, maternal education, maternal age, and birth year.

Unfortunately, our data set did not provide information on the time or urgency of the ED visits. We also do not know how many mothers contacted their physician before going to the ED. A recent study of full-term neonates discharged from a single newborn nursery in Ohio found that one third of all ED visits were made when the primary care physicians' offices were open and 58% of these ED visits were determined to be nonurgent.⁵⁹ Only 15% of all visits to the ED were referred by a primary care physician. Twenty-nine percent of physician-referred visits were determined to be nonurgent while 64% of the self-referred visits were nonurgent. Our results are further limited by being restricted to fee-for-service Medicaid patients from a single state, although a recent study of Medicaid infants in an urban tertiary care center found no difference in ED usage patterns between managed care and fee-for-service enrollees.⁶⁰

In this high-risk population, having a primary care visit within 21 days of the neonate's birth and having the primary care visit at a hospital-based clinic were not associated with a decreased risk of ED use within 91 days of birth. In fact ED use increased in this population even after adjusting for possible confounders. This implies that simply having a primary care visit does not influence ED use. Alternatively, mothers seeking care may not perceive the ED and the primary care clinic as representing different health care systems, but rather make their choice based on convenience. The results of this study suggest that the focus should shift to measuring and improving the quality of the primary care services, especially around access, but based on the perceptions of the users of the services themselves. The quality and effectiveness of primary care services also needs to be monitored carefully by payers of care to ensure that they meet the needs of users. Alternatively, the ED could be seen as a reasonable, significant and integral part of the system of providing care for the Medicaid population. Models focusing on delivering primary care in this setting should be tested.

Accepted for publication March 19, 2002.

This study was supported in part by a Medical Technical Assistance and Policy Program grant from the Ohio Department of Human Services Bureau of Medicaid Policy, Columbus (Dr Kotagal).

The use of the ED as a site of routine care has been noted in several studies and has been related to maternal perception of infant health and maternal and social variables as well as barriers to primary care use, such as lack of a medical home, transportation needs, and child care issues. In previous studies at a single urban hospital in Ohio, we noted that up to 40% of all newborns discharged who were uninsured or receiving Medicaid used the ED for health care needs at least once in the first 3 months and 14% visited the ED before their first primary care visit.

The present study was undertaken to extend this research to a larger population. We hypothesized that early linkage to primary care would be associated with decreased ED use. However, contrary to our expectations, infants in this population who were taken to the ED within 91 days of birth were likely to have had more prenatal care visits and were more likely to have had early primary care linkage. They were more likely to have gone to hospital-based primary care clinics for the majority of their visits. Even after adjusting for maternal, infant, and residency characteristics, early primary care linkage did not result in a decreased risk of ED use in early infancy.

Corresponding author and reprints: Uma R. Kotagal, MBBS, MSc, Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229-3039 (e-mail: Uma.Kotagal@chmcc.org).