Pediatric non-urgent emergency department visits and prior care-seeking at primary care

In 2016 the Centers for Disease Control and Prevention estimated that 40% of ED visits for children 4 years old and under received initial triage levels of semi- or nonurgent [1]. These ED visits are often a target for interventions, with the underlying goal of redirecting some of these visits to other sites of care, such as the child’s primary care office [2‐4]. Several studies have described associations between non-urgent ED use and child or family sociodemographic characteristics, such as age, race, income, insurance, and social support [5‐9]. For example, children with Medicaid coverage are twice as likely as privately insured children to visit the ED, and their ED visits are less likely to be assessed as urgent or emergent [10]. Race has been inconsistently associated with non-urgent ED visits, with a recent systematic review noting that four of nine studies found an association between Black race and non-urgent ED visits, while the remaining identified studies found no association [5]. Lower parental educational attainment and lower health literacy have also been associated with increased nonurgent ED visits [11, 12]. In addition to these sociodemographic factors, many dimensions of primary care access have been associated with non-urgent ED visit rates, including geographic access and perceived quality of care [13‐21].

In a recent systematic review, these sets of factors were synthesized into one model that highlighted potential causal pathway factors (e.g., perceptions of access, cost, convenience, quality, advice/referral from PCP) that prompt a patient’s decisions in whether to seek care and where to seek care (PCPs office or the ED) [5]. Unlike the previous studies mentioned, this model distinguished these causal pathway factors from associated sociodemographic factors (e.g., race, insurance) which the authors posited are likely associated with non-urgent ED visits only through their association with potential causal pathway factors [5].

Among the potential causal factors in this pathway, advice or referral from primary care warrants particular attention. After all, children referred to the ED by their primary care provider (PCP) have followed a different pathway to care than children whose families self-referred, and understanding their different perceptions of care and motives may guide different interventions for each group. In prior studies, between one-third and one-half of non-urgent pediatric ED visits and up to 80% of all pediatric ED visits were preceded by contact or attempted contact with the child’s primary care office [14, 22‐24]. How other potentially causal factors (e.g., perceptions of primary care access, costs, quality) influence the decision to attempt PCP contact prior to an ED visit are not known, but improved understanding of these heterogeneous routes to non-urgent ED visits may improve understanding of underlying sociodemographic variation in non-urgent ED visits as well.

Thus, to guide future interventions to reduce non-urgent ED visits, we sought to examine patterns of PCP contact prior to non-urgent ED visits through a survey administered to caregivers of children presenting with low acuity ED visits during hours when their PCP would be expected to be open. We examined prevalence of contact with their child’s PCPs office prior to the child’s non-urgent ED visit, and compared caregiver perceptions of their child’s PCP office and trust in their PCP as primary exposures in care seeking decisions. Our secondary analysis examined caregivers’ relative assessment of primary care versus emergency care among those who did and did not seek care from their PCP’s office prior to their non-urgent ED visit. We examined whether these perceptions and care preferences explained the variation in sociodemographic composition of the two groups.

We performed a cross-sectional survey of caregivers accompanying children to a pediatric ED during weekday business hours and triaged as low acuity. This study focuses on the association between care-seeking prior to presenting to the ED and respondent sociodemographic characteristics, primary care perceptions, and relative assessment of sites of care (ED vs PCP); a separate analysis examined non-response to health-related social need questions within the survey [25]. This study received ethical approval from our institution’s Institutional Review Board, who approved a waiver to document informed consent due to the study involving no more than minimal risk of harm.

One hundred sixty nine caregivers were approached, and 157 of these agreed to participate. After excluding those with missingness of key variables, 140 are included in the final analytic sample. Respondents were primarily between the ages of 25–44 years old, with 51% identifying as white and 29% as Black or African American (Table 1). Most respondents reported attending the ED with a child insured by Medicaid or CHIP (62%), and 40 % had completed college or post-college education.

Of respondents, 46% (64/140) reported that they had sought care from their child’s PCP office prior to presenting to the ED, while the remaining 54% reported they had not sought care from their child’s PCP office (Fig. 1). Of those who had sought care from their child’s PCP office, 25% (16/64) reported their child had a visit with their PCP, 69% (44/64) reported they talked with their child’s PCP office via telephone, and 6% (4/64) reported they attempted to contact the office but did not reach them. Of those who spoke with or saw their PCP’s office, 82% (49/60) were told to present to the emergency department, meaning 35% of the entire sample (49/140) had been instructed by their PCP’s office to present to the ED.

Caregivers reporting that they sought care from their child’s PCP office prior to their child’s ED visits differed significantly by child insurance status, caregiver race, and caregiver educational attainment (Table 1). Among caregivers of children triaged as low acuity in the ED, only 36% of those insured by Medicaid or CHIP had sought PCP care before presenting to the ED, compared to 68% of those insured by commercial insurance (p = 0.004). 62% of white caregivers presenting to the ED for a low acuity concern had sought PCP care prior to presenting to the ED compared to 24% of caregivers who identified as Black (p < 0.001). 77% of caregivers with a high school education/GED or less had not sought PCP care prior to presenting to the ED compared to 39% of those reporting at least a college education (p = 0.001).

Those who sought care from their PCP’s office prior to coming to the ED were significantly more likely to report higher median perception of primary care, with total score of 86 (IQR 71–92) for those who had sought PCP care and 76 (IQR 60–90) for those who had not sought PCP care (p = 0.01; Table 2). Significant variation was identified for three specific subscores: access, contextual knowledge, and coordination. For example, median access subscore was higher for those who had sought PCP care than those who had not (88 vs. 75, p < 0.001). Trust in provider did not significantly differ for caregivers who had and had not sought care from their child’s PCP office (p = 0.12).

Caregiver’ perceptions of travel time, wait time, and costs at their child’s PCP versus the ED were significantly associated with prior PCP care-seeking (Table 3). Among caregivers of children triaged as low acuity in the ED, 69% of those who reported travel time to their PCP was better than to the ED had sought prior PCP care compared to 18% of those who reported travel time to the ED was better (p < .001). Among respondents who reported wait time at their PCP was better than in the ED, 67% had sought prior PCP care compared to only 10% of those reporting ED wait time was better (p < .001). Assessment of quality of care between the two sites was not associated with prior care-seeking (p = 0.49).

In the first multivariable model including only sociodemographic variables (child insurance, parent race, and parent education; Table 4Model A), odds of having sought care from their child’s PCP office prior to their non-urgent ED visit were significantly lower for parents who had not entered college (OR 0.32, 95% CI 0.11–0.90) compared to those who had completed college and for parents identifying as Black or African American (OR 0.29, 95% CI 0.11–0.74) compared to parents identifying as white. When also including significant perception of primary care subscales (access, contextual knowledge, coordination, Table 4Model B), the model was not significantly improved (likelihood ratio (LR) test, p = 0.18), and sociodemographic associations maintained significance with similar magnitude and direction. When instead including significant relative assessments of site of care (travel, wait times, and cost estimations), the model was significantly improved (LR test p < 0.001). In this final model (Table 4, Model C), estimates of PCP wait time and travel time being better than ED wait time and travel time were significantly associated with odds of having sought care from child’s PCP office prior to their non-urgent ED visit (Travel: PCP better option, OR 3.46, 95% CI 1.34–8.92 compared to neutral response reference group; Wait time: PCP better option, OR 2.86, 95% CI 1.11–7.34 compared to neutral response reference group). In this final model accounting for parents’ relative assessments of care sites, odds ratios associated with Black parent race and lower educational attainment were reduced by 33 and 6%, respectively, and were no longer significantly associated with odds of having sought care from their child’s PCP office prior to their non-urgent ED visit.

One of the key findings from our cross-sectional survey of caregivers of children presenting for a non-urgent ED visit is that pediatric patients follow multiple different pathways when they present to the ED for visits that are triaged as low acuity. Consistent with prior estimates [22, 23], we found that nearly half of surveyed caregivers had sought advice from their PCP’s office before presenting to the ED. By identifying differences in caregivers’ perceptions of their primary care provider and relative preferences regarding the care they receive at their PCPs office versus the ED, our results suggest ways that prior primary care and ED experiences may influence decisions in where to seek care. Notably, primary care perceptions associated with seeking PCP care appear primarily related to convenience (access, care coordination), and contextual knowledge as opposed to quality, comprehensiveness, or communication. Our results also suggest that caregivers’ perceptions of their child’s PCP office may be less of a decisional factor than the relative assessment comparing potential time and financial costs at the PCP office versus the ED. This suggests that among the causal factors posited in prior conceptual models [5], further work should investigate the decision to seek advice/referral as a key decision point potentially influenced by other identified causal pathway factors (e.g., cost, convenience, access/availability, beliefs about knowledge and alternatives).

A second key finding is that while caregivers with lower educational attainment and caregivers identifying as Black or African American had lower odds of having sought PCP care before non-urgent ED visit, these associations were no longer significant when caregiver relative assessment of wait time, travel time, and cost were included. This lends support to the prior conceptual models [5, 28] that sociodemographic factors such as race, insurance status, and education are associated factors mediated through other specific causal factors in care seeking decisions (e.g., assessments of relative access, convenience, and cost). We note that these findings should be replicated in larger samples, but also that the significant likelihood ratio test (model C versus model A, p < 0.001) indicates that the loss of significance with race and educational attainment occurred in the context of an overall improvement in the model. In the United States, experiences of accessing and receiving care are consistently more difficult for individuals identified as Black or Hispanic as well as by educational background or insurer [31]. For example, studies in primary care and emergency department settings have reported that time spent waiting for care varies by race, education, and insurance [32‐34]. Implicit bias and structural racism can also shape experiences of care and geographic access to care [31]. These studies illustrate ways that families in the current US health care delivery systems may have health care experiences (e.g., travel time and wait time) that differ systematically by race. As a result, where a caregiver takes their child for non-urgent care may be driven by differential access to primary care across race, education level, and insurance type rather than these characteristics themselves.

Parents of sick children are seeking to make the best decisions for their children given the information and options they have. Thus, patterns of care-seeking deemed sub-optimal by health care professionals and systems should be viewed as symptoms of an imperfect system, rather than imperfect individual decision-making. With this mindset, analyses of non-urgent ED visits can be used to identify ways to change the system to make it easier for parents to obtain the care their child needs in lower cost settings. One possibility suggested by our data is to address travel time or wait time at primary care relative to the ED. One recent study found that implementing walk-in care within the medical home, which may alter wait time, reduced the proportion of care received in ED or urgent care settings for their patients [35]. Further opportunities to reduce travel time and wait time exist through school and childcare-based clinics and telehealth [36, 37] as well as more mundane interventions such as having adequate staffing and efficient scheduling and rooming workflows [38]. As specific examples, use of telemedicine to deliver school-based asthma care and childcare-based acute care have both reduced emergency department visits [36, 37]. Finally, our data suggest that for nearly half of non-urgent ED visits, parents are already contacting and being referred by their PCPs office. Since most caregivers who were recommended to present to the ED by their PCP office spoke with the PCP office over the telephone, future studies could focus on nurse triage systems at these PCP offices as an important alternative intervention point for reducing non-urgent ED visits. Indeed, prior work has demonstrated that physician involvement before nurse triage referral to the ED may reduce ED referrals by half [39]. Alternative strategies to reduce these PCP-referred low acuity ED visits may instead require primary care-ED partnered investigations of shared referral pathways, primary care resources, direct admission strategies, and access to specialty care [40, 41].

In conclusion, among caregivers of children presenting to the ED for non-urgent visits (triaged as low acuity), we found that (1) more positive parent perceptions of the convenience and contextual knowledge of their child’s primary care provider and (2) estimations of shorter travel times and wait times at primary care offices compared to the ED were both associated with increased likelihood of having sought care from the child’s PCP prior to a non-urgent ED visit. Differential primary care access by sociodemographic groups in the US may contribute to previously observed sociodemographic differences in non-urgent ED utilization.