Assessing the Relationship Between Well-Care Visit and Emergency Department Utilization Among Adolescents and Young Adults

Abstract

Purpose:

To investigate the association between adolescent and young adult (AYA) well-care visits and emergency department (ED) utilization.

Methods:

Vermont’s all-payer claims data were used to evaluate visits for 49,089 AYAs (aged 12–21 years) with a health-care claim from January 1 through December 31, 2018. We performed multiple logistic regression analyses to determine the association between well-care visits and ED utilization, investigating potential moderating effects of age, insurance type, and medical complexity.

Results:

Nearly half (49%) of AYAs who engaged with the health-care system did not attend a well-care visit in 2018. AYAs who did not attend a well-care visit had 24% greater odds (95% confidence interval [CI]: 1.19–1.30) of going to the ED at least once in 2018, controlling for age, sex, insurance type, and medical complexity. Older age, female sex, Medicaid insurance, and greater medical complexity independently predicted greater ED utilization in the adjusted model. In stratified analyses, late adolescents and young adults (aged 18–21 years) who did not attend a well-care visit had 47% greater odds (95% CI: 1.37 – 1.58) of ED visits, middle adolescents (aged 15–17 years) had 9% greater odds (95% CI: 1.01e1.18), and early adolescents (aged 12–14 years) had 16% greater odds (95% CI: 1.06 – 1.26).

Conclusions:

Not attending well-care visits is associated with greater ED utilization among AYAs engaged in health care. Focus on key quality performance metrics such as well-care visit attendance, especially for 18- to 21-year-olds during their transition to adult health care, may help reduce ED utilization.

Adolescents and young adults (AYAs) are known to underutilize preventive care services (e.g., well-care) [1–3]. Regular well-care visits are critical to reinforcing healthy behaviors, and AYAs are more likely to use more costly emergency department (ED) services for routine care [4–7]. In recognition of this, increasing AYA well-care visit attendance is a key quality performance measure for several accountable care organizations nationwide, including Vermont [8]. One meta-analysis showed health-care reform efforts have not increased well-care visits on a national level [9]; however, in Vermont, similar efforts have increased AYA well-care visits at the population level [10]. While increasing AYA well-care visit attendance overall is intended to help prevent risky behaviors and detect and promote selfmanagement of health conditions, it would also be important to know whether participating in preventive well-care is associated with less hospital utilization, such as ED visits.

A 2019 report from the Patient-Centered Primary Care Collaborative showed that increased primary care spending was associated with fewer ED visits [11]. Efforts by the Centers for Medicare and Medicaid Services to invest in comprehensive primary care initiates have shown modest reduction in ED visits, with no reduction in overall costs [12]. Despite mixed results within this growing field of literature, attention has largely been placed on increasing primary care spending generally and across the lifespan [11,12]. This study aims to address a gap in the literature by focusing on one foundational aspect of primary care for AYAs, the well-care visit.

There may be subpopulations of AYAs that benefit most from well-care to help them avoid ED visits. Among AYAs, older adolescents [2,4,5] and those with Medicaid insurance [2,13] are known to have greater ED utilization. There is also increasing focus on medical complexity as a driver of health-care utilization [14–16]. Multiple studies have shown increased utilization of hospital-based resources (i.e., ED, inpatient) [16–18] and outpatient services [14,19,20] among AYAs with medical complexity. Studies also suggest that AYAs with medical complexity may rely more on specialists than primary care providers [21], potentially depriving them of an annual well-care visit. It is unknown, however, how age, insurance type, and medical complexity may interact with AYA preventive and emergency health-care seeking behavior.

Our primary objective is to assess the association between well-care visits and ED utilization among AYAs engaged in the health-care system. Our secondary objective is to assess the impact of age, insurance type, and medical complexity on the association between well-care visits and ED utilization among AYAs. We hypothesize that (1) not attending a well-care visit will be associated with ED use and (2) the association will be strongest among AYAs who are older and those with Medicaid insurance and higher medical complexity.

Methods

Data source and population

We accessed the all-payer claims database, Vermont Health Care Uniform Reporting & Evaluation System (VHCURES), and included individuals in any health plan (e.g., commercial, Medicaid, or Medicare) in the state of Vermont from January 1 through December 31, 2018. VHCURES includes approximately 75% of the Vermont population and excludes uninsured individuals [22], those covered by federal insurance or military health-care programs (e.g., TRICARE), payers with fewer than 200 Vermont residents, and Employee Retirement Income Security Act self-insured payers who have not opted to continue submitting data to VHCURES after the 2016 Gobielle v. Liberty Mutual Supreme Court decision [23]. Our initial sample included AYAs (aged 12–21 years) eligible for insurance in 2018 (N = 61,443; Figure 1). We excluded AYAs with inconsistent sex across months of eligibility in 2018 (N = 104) rather than choosing a single sex; AYAs without at least one claim with an International Classification of Diseases, 10th Revision, Clinical Modification diagnosis code between 2016 and 2018 (N = 7,340); and AYAs without valid (i.e., not denied) medical claims paid as primary in 2018 (N = 4,910). Our final analytic sample included 49,089 AYAs.

Figure 1.

Final sample selection of adolescent and young adult population with insurance, sex, and medical claims information in 2018 in Vermont’s all-payer claims database.

Predictors and outcomes

Our main predictor was a binary variable for attending a well-care visit in 2018 (yes vs. no), using current procedural terminology codes for comprehensive and periodic preventive visits for new (99,381–99,385) or established (99,391–99,395) patients. Demographic covariates included age categorized into three groups (early adolescents, aged 12–14 years; middle adolescents, aged 15–17 years; and late AYAs, aged 18–21 years) [24] and having any Medicaid insurance during the year as binary (Medicaid vs. non-Medicaid), with the non-Medicaid category primarily commercially insured individuals and only .2% (N = 97) having Medicare. In addition, we controlled for sex (female vs. male) as a potential confounder and Pediatric Medical Complexity Algorithm version 3.1 defining three categories of medical complexity based on International Classification of Diseases, 10th Revision, Clinical Modification diagnosis codes across three years (2016–2018) of VHCURES claims [25,26]. We used the more conservative approach (requiring two or more claims for each body system identified) to group patients by increasing medical complexity: nonchronic (including healthy), noncomplex chronic (illness affecting one body system that was neither progressive nor malignant), and complex chronic (illness affecting ≥2 body systems or any progressive or malignant illness). Our outcome variable was a binary indicator for having one or more ED visits in 2018. Based on Centers for Medicare and Medicaid Services Research Data Assistance Center guidelines and recently published methodologic updates [27], we identified ED visits using place of service type [23], current procedural terminology codes (99,281–99,285, 99,291), Healthcare Common Procedure Coding System codes (G0378, G0380, G0381, G0382, and G0348), and revenue codes (450, 451, 452, 456, 459, 981, 762, and 760, if accompanied by G0378), for evaluation, management, observation, or critical care provided to a patient in the ED.

Statistical analyses

To examine our primary research question of whether not attending a well-care visit was associated with ED utilization, we conducted a multiple logistic regression analysis of our ED outcome on well-care visit, controlling for age, sex, insurance type, and medical complexity. For our secondary research question, we assessed whether the primary association was different within demographic subgroups (age and insurance) or across the three medical complexity groups. We included interactions between our main predictor (well-care visit attendance) and each of these covariates in our primary multiple logistic regression model. When significant interactions were present, we stratified into subgroups of the covariate and reran separate multiple logistic regressions. We used StataSE version 15 (StataCorp LLC, College Station, TX) statistical software for all analyses, and associations were considered significant if p < .05. The Institutional Review Board at the University of Vermont approved this study.

Results

Demographics and emergency department utilization

Table 1 shows the total number and percentage of AYAs in our overall sample and by ED utilization in 2018, categorized by age group, sex, insurance type, medical complexity, and well-care visit attendance. One quarter of AYAs who had a health-care claim in 2018 used the ED, with highest utilization among older AYAs, females, those with Medicaid, and those with higher medical complexity. Nearly half of AYAs in our sample did not attend a well-care visit in 2018. The unadjusted odds of ED use among AYAs who did not attend a well-care visit was 49% higher than that of those who did attend a well-care visit (odds ratio = 1.49, 95% confidence interval = 1.44, 1.56).

Table 1:

Sample sizes and percentages of covariates overall and by emergency department utilization

	Total sample, N (%)	Emergency department visit
		No (n)	Yes(n)	% Yes
Full sample	49,089(100.0)	36,767	12,322	25.1
Age group, years
12–14	15,584 (31.8)	12,506	3,078	19.8
15–17	15,399 (31.4)	11,620	3,779	24.5
18–21	18,106 (35.9)	12,641	5,465	30.2
Sex
Male	24,218 (49.3)	18,530	5,688	23.5
Female	24,871 (50.7)	18,237	6,634	26.7
Insurance
Medicaid	31,058 (63.3)	21,932	9,126	29.4
Non-Medicaid	18,031 (36.7)	14,835	3,196	17.7
Medical complexitya
Nonchronic/healthy	31,077 (63.3)	24,680	6,397	20.6
Noncomplex chronic	13,525 (27.6)	9,323	4,202	31.1
Complex chronic	4,487 (9.1)	2,764	1,723	38.4
Well-care visit
No	23,886 (48.7)	16,962	6,924	29.0
Yes	25,203 (51.3)	19,805	5,398	21.4

^aPediatric Medical Complexity Algorithm Version 3.1.

Covariates associated with emergency department utilization

Our main result from the multiple logistic regression model was that AYAs who did not attend a well-care visit in 2018 had 24% greater odds of visiting the ED in the same year than those who did attend a well-care visit, controlling for age, sex, insurance type, and medical complexity (Table 2). In addition, this adjusted model showed that older age, female sex, Medicaid insurance, and greater medical complexity independently predicted higher ED utilization (p < .0005). Late AYAs (aged 18–21 years) had 77% greater odds of going to the ED than early adolescents (aged 12–14 years), while middle adolescents (aged 15–17 years) had 30% greater odds of going to the ED than early adolescents. Female adolescents had 21% greater odds of visiting the ED than male adolescents, and those with Medicaid, compared to non-Medicaid, had 79% greater odds of using the ED. AYAs with noncomplex chronic and complex chronic disease had 61% and over 2-fold greater odds of going to the ED, respectively, than AYAs with no chronic illness including those who were otherwise healthy (Table 2).

Table 2:

The associations between emergency department visits and well-care attendance, age group, sex, insurance type, and medical complexity using multiple logistic regression

	aORa	95% Confidence interval
		Lower	Upper	p Value
Well-care visit (reference = yes)
No well-care visit	1.24	1.19	1.30	<.0005
Age group (reference = 12–14)
15–17	1.30	1.23	1.37	<.0005
18–21	1.77	1.68	1.87	<.0005
Sex (reference = male)
Female	1.21	1.16	1.26	<.0005
Insurance (reference = non-Medicaid)
Medicaid	1.79	1.71	1.88	<.0005
Medical complexityb
(reference = nonchronic/healthy) Noncomplex chronic	1.61	1.53	1.68	<.0005
Complex chronic	2.15	2.01	2.30	<.0005

aOR = adjusted odds ratio.

^aEach odds ratio is adjusted for the other covariates in the table.

^bPediatric Medical Complexity Algorithm Version 3.1.

Age, insurance, and medical complexity as moderators

Our interaction term between well-care visit and age group was significant when added to our multiple regression model, but neither of the interaction terms between well-care visit and insurance and between well-care visit and medical complexity were significant (p > .05). Thus, we report the results of our primary model stratified by age group only. Among late AYAs (aged 18–21 years), those who did not attend a well-care visit had 47% greater odds of going to the ED than those who did attend a well-care visit, controlling for sex, insurance type, and medical complexity (Table 3). Similarly, middle adolescents, aged 15–17 years, with no well-care visit had 9% greater odds of using the ED. Among early adolescents, aged 12–14 years, those with no well-care visit had 16% greater odds of going to the ED. Similar to the primary model results, associations between potential confounders and ED utilization persisted across the secondary models stratified by age group (Table 3).

Table 3:

The associations between emergency department visits and well-care attendance, sex, insurance type, and medical complexity, stratified by age group, using multiple logistic regressions

	12–14 Years	p Value	15–17 Years	p Value	18–21 Years	p Value
	aOR (95% CI)a		aOR (95% CI)a		aOR (95% CI)a
Well-care visit (reference = yes)
No well-care visit	1.16(1.06–1.26)	.001	1.09 (1.01–1.18)	.02	1.47 (1.37–1.58)	<.0005
Sex (reference = male)
Female	1.07 (.99–1.16)	.10	1.26 (1.17–1.36)	<.0005	1.29 (1.21–1.38)	<.0005
Insurance (reference = non-Medicaid)
Medicaid	1.84 (1.67–2.03)	<.0005	1.83 (1.68–2.00)	<.0005	1.75 (1.63–1.87)	<.0005
Medical complexityb (reference = non-chronic/healthy)
Noncomplex chronic	1.55 (1.42–1.70)	<.0005	1.65 (1.52–1.79)	<.0005	1.60 (1.49–1.73)	<.0005
Complex chronic	2.12 (1.86–2.42)	<.0005	2.22 (1.97–2.50)	<.0005	2.11 (1.90–2.35)	<.0005

aOR = adjusted odds ratio; CI = confidence interval..

^aEach odds ratio is adjusted for the other covariates in the table.

^bPediatric Medical Complexity Algorithm Version 3.1.

Discussion

In this population-based analysis of preventive and emergency care in Vermont, we found that among AYAs engaged with the health-care system, those who did not attend a well-care visit had greater odds of an ED visit, controlling for age, insurance, ex, and illness complexity. These findings were concerning given that nearly half (49%) of our AYA sample did not attend a well-care visit in 2018. Our result supports previous findings that AYAs had lower rates of well-care [28,29] and higher rates of ED use [2,6]. Furthermore, our finding that the association was strongest among late AYAs (aged 18–21 years) was also in line with literature showing that young adults more frequently used the ED [2,4].

This is the first study to our knowledge to examine these two health-care-seeking behaviors together and provide evidence of an association between not attending a well-care visit and greater ED use among AYAs. Although our study is the first to acknowledge this relationship, it is well documented that AYAs have continued to underutilize preventive services [30] and overutilize the ED [4,31] in spite of several attempts to address this at a national policy level. Passage of the Affordable Care Act in 2010 led to broader insurance coverage for AYAs because of extended eligibility for dependents through age 26 years, no cost-sharing for preventive services, and expanded Medicaid coverage and services [32]; however, it has thus far only yielded modest increases in primary care visits [29]. In addition, national guidelines established in 2017 specifically recommend an annual well-care visit for all AYAs [24]. Our findings provide timely evidence that despite efforts to reduce insurance barriers and promote AYA well-care visits with national guidelines, there remains an association between not attending a well-care visit and more costly ED use for AYAs. Our study also suggests that presumably the breadth of health-care maintenance and preventive care strategies provided through AYA well-care may in some way be related to ED visits, but further study is needed to elucidate such a relationship.

Although the etiology of this association is beyond the scope of the present study, it is important to consider possible mechanisms. Unmeasured factors such as a family or patient’s relationship with their primary care provider [4], education provided on appropriate use of emergency care [4,33], and the quality of risk reduction counseling received during an adolescent well-care visit [1,34] may affect and interact with the way AYAs seek health care. In the face of such complex factors, our data suggest that at the very least, it is appropriate for AYA well-care attendance to remain a key quality performance measure for pediatric practices and accountable care organizations to ensure continued focus on this important and vulnerable population.

Unsurprisingly, we found that the association between well-care visit and ED use was strongest among late AYAs, aged 18–21 years, suggesting potential systemic and/or developmental factors contributing to this association. Young adults are known to fare worse across a variety of health indicators than adolescents [1,28]. For example, young adults, whether healthy or medically complex, have a high risk of discontinuity of primary care during the period of transition from pediatric to adult health care [35]. Spatial distance to a primary care practice compared to that to an ED can affect resource use as well [36], which may contribute to preferential ED use over well-care as AYAs attend visits on their own, especially in a rural state such as Vermont. Overall, our results suggest that having a well-care visit may be paramount during the transition from pediatric to adult primary care, further warranting its designation as a key quality performance metric. In turn, pediatric-serving practices may focus limited outreach resources on bolstering well-care visit attendance, which may be especially impactful for late AYAs during the transition to adult health care.

While there was no evidence that the association between well-care visit and ED use was different based on insurance type or sex, both covariates independently predicted greater odds of going to the ED. Greater ED use among AYAs with Medicaid, a proxy for low socioeconomic status, supports prior literature that adolescents with lower socioeconomic status report greater disparities in access to health care [37]. It also suggests that AYAs with Medicaid may be a high-priority population when it comes to directing resources in the primary care setting as they are vulnerable based on age and insurance. Unexpectedly, female sex also predicted greater ED use. Literature from two decades ago on AYA ED utilization showed that males used the ED more than females [5,6]. However, more recent health-care utilization data show female AYAs use ambulatory and hospital services more frequently, after adjusting for pregnancy [7]. Female AYAs may also be a high-priority population when it comes to directing resources in the primary care setting as they are vulnerable based on age and sex. It is possible our findings underscore an ongoing transformation in ED utilization trends and may warrant further investigation.

To our surprise and contrary to our hypothesis, medical complexity did not moderate the association between well-care visit attendance and ED utilization, demonstrating the need to better characterize the association between types of ED visits and well-care visits among those with medical complexity. However, our results did show that medical complexity was an independent predictor of ED utilization, which was consistent with prior literature [14,38]. We know that well-care visits conducted by a primary care provider within a medical home are especially vital to AYAs with medical complexity, as they learn to manage chronic conditions and take increased responsibility for managing their own health care [15]. We also know that continuity of care and smooth transition of care are important for AYAs with medical complexity who are vulnerable because of their age and health status [39,40]. In spite of this, it is possible that AYAs with medical complexity have other protective factors that we did not account for such as more frequent health-care encounters overall [41] and/or specialists who provide more comprehensive care [42]. Either of these factors may lead to AYAs with medical complexity receiving preventive care outside of attending a well-care visit in primary care or in reducing the association of a well-care visit with ED utilization. In addition, there is emerging literature describing the relationship between children with medical complexity and ED visits [18]. Rurality, spatial proximity to well-care and ED providers, and frequency of health-care encounters may be other potential future directions aimed at better understanding health-care utilization among AYAs and those with medical complexity.

Limitations

There are several limitations of our study. The present study focused on health care seeking behavior outcomes among a sample of AYAs who had at least one health-care claim in 2018. Efforts to engage AYAs who are insurance eligible but do not seek health care (e.g., no claims for at least one year and a majority with no claims for three years) would likely look very different. We designed our study to have actionable results for practitioners seeking to make system changes around patients who seek health-care services within a relatively recent timeframe. Also, all-payer claims data do not include race or ethnicity information, nor do they include income information or uninsured AYA, thus, limiting our ability to account for socioeconomic status. In addition, results for a small, rural state such as Vermont may not be as generalizable to AYA populations in other states. However, our results may generalize to other rural states, and more broadly, these analyses may provide a useful framework that could be reproduced by other states with access to all-payer claims. These results are a first step in an ongoing investigation of the relationship between well-care and ED visits, and in this initial inquiry, we limited analyses to cross-sectional data, not assessing the temporal relationship between well-care and ED visits. However, our model retains primary care visits after the ED event, which may make our results more conservative (trending toward the null). Finally, we included all ED visits, as characterizing the nature and severity of the ED visit remains a complicated task and emerging area of investigation [33]. Measuring temporality, appropriate use of the ED, and other types of primary care visits would be crucial next steps in future longitudinal investigation of the relationship between AYA well-care visit attendance and ED utilization.

In conclusion, not attending AYA well-care visits is associated with greater ED utilization for all AYAs engaged with the health-care system, even after controlling for the known confounding effects of insurance type, sex, and medical complexity. This association is strongest among late AYAs. Thus, focus on meeting key quality performance metrics such as well-care visit attendance, particularly among 18- to 21-year-olds during the transition to adult health care, may help reduce ED utilization.

IMPLICATIONS AND CONTRIBUTION.

Results suggest that among adolescents engaged with the health-care system, not attending a well-care visit is associated with greater emergency department utilization, with the strongest association seen among 18- to 21-year-olds. Thus, continued emphasis on adolescent well-care visit attendance, a key quality performance metric, may help lower emergency department visits.