Abstract
This study examined differences in health service utilization among anxious and non-anxious youth primary care patients. We further examined the moderating role of Hispanic ethnicity on the relation between anxiety and service utilization. Past 4-year health care utilization was examined in a group of 6962 American youth (51.10% male; 53.10% Hispanic). Youth with an anxiety disorder had significantly more medical visits over the prior 4-year period than youth without an anxiety disorder. Hispanic youth utilized health care services at higher rates than non-Hispanic youth; however, ethnicity did not moderate the relation between youth anxiety and health care utilization.
Elevated health care costs are common for individuals with anxiety. Estimates suggest that anxiety-related complaints account for more than 30 percent (US$46.6 billion) of total mental health expenditures in the United States, with nearly half of that figure attributed to repeated use of health care services (Lépine, 2002). This cost may be particularly burdensome for families with youth offspring (Cohen and Kirzinger, 2014; Cohen et al., 2012). It is well established that individuals with anxiety disorders, relative to those without such problems, tend to inappropriately utilize health care services (e.g. a panic attack leads to a costly trip to the emergency department; Lynch and Galbraith, 2003). Furthermore, anxiety problems are not often appropriately diagnosed in this context (Weissman, 1990), resulting in recurring utilization of expensive and inefficient means of symptom management (Lynch and Galbraith, 2003). This process may begin in childhood and adolescence. A recent study from the Netherlands found that families with anxious offspring spend 21 times as much on offspring health care as healthy families (Bodden et al., 2008), but no study to date has examined these processes among youth in the United States. This is problematic, particularly in light of evidence that late childhood and adolescence are considered a “core-risk” period for the development and onset of anxiety (Beesdo et al., 2009).
A critical process at play here could be misinterpretation of normal bodily sensations as dangerous, which then contributes to an array of negative consequences with significant developmental and societal costs (e.g. inappropriately seeking health care, school absenteeism; Ehlers, 1993). As an illustrative example, an anxious youth, compared to a non-anxious youth, may describe benign somatic experiences (e.g. gastrointestinal distress associated with menses) as relatively more upsetting, thereby eliciting differential parent responses (e.g. taking the child to the doctor, allowing him/her to miss school). For this reason, it may be critical to examine these factors in concert with the onset of puberty, as this developmental period is characterized by a host of new physical sensations. The physiological changes that characterize puberty (e.g. emergence of secondary sex characteristics) often begin between the ages of 9 and 12 years (Herman-Giddens, 2007). Therefore, this article selected a lower age bound of 8 years and an upper age bound of 17 years to capture some of the developmental changes that may occur as youth commence and proceed through pubertal development.
Another limitation of the current literature is that these issues have been examined in predominantly White population. Little work explores how health care utilization differs in US-residing Hispanic families with anxious youth versus Hispanic families with non-anxious youth. This a notable gap, given that Hispanics comprise the largest minority group in the United States (Humes et al., 2011) and appear to suffer significantly more from somatic and physiological symptoms of anxiety than Whites (Pina and Silverman, 2004). Additionally, given evidence that Hispanics express more stigmatizing attitudes toward mental illness than other ethnic groups (Corrigan and Watson, 2007), and that Hispanic adults tend to seek primary care services for mental health difficulties (Cabassa et al., 2006), it may be the case that Hispanic families are more inclined to seek physical health services than mental health services for their anxious offspring.
The proposed study addresses these gaps in the literature by examining health care utilization patterns among anxious youth presenting at a primary care clinic over a 1-year period. This study had two aims. The first was to examine differences in health service utilization among families with anxious and non-anxious youth offspring. It was hypothesized that youths with anxiety disorders would utilize health care services at higher rates than youths without anxiety disorders. The second aim was to examine differences between ethnic groups (i.e. Hispanic vs non-Hispanic) in health care utilization patterns and to examine whether ethnicity interacted with anxiety disorder status in its relation to health care utilization among children and adolescents. It was hypothesized that compared to non-Hispanic youth, Hispanic youth would utilize health care services more often. We made no predictions about the direction of an interaction between ethnicity and anxiety disorder status.
Method
Participants
Health care utilization frequency was examined in a group of 6962 children and adolescents between 8 and 17 years of age (51.10% male; 53.10% Hispanic). All unique youth patients who attended a primary care appointment at Federally Qualified Health Center (FQHC) in the southern United States during a 1-year period were included in the sample. The FQHC employs approximately 250 people and serves over 10,000 patients per year (half of whom are children, more than half of whom prefer to receive services in a language other than English, and over 90 percent of whom live at or below 200 percent of the Federal Poverty Level).
Demographics are presented in Table 1. In total, 142 youths (2.00% of youth) met criteria for a Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5) anxiety disorder, while 6820 did not. In terms of health insurance, 62.90 percent of participants utilized public insurance (i.e. Medicaid), 13.00 percent of participants utilized private insurance, and insurance information was not available for 24.10 percent of the sample. It is unclear whether this 24.10 percent of the sample did not have health insurance or whether their families declined to provide that information.
Descriptive statistics and bivariate associations among study variables.
SD: standard deviation.
p < .05, **p < .01, and ***p < .001.
Measures and procedures
Study procedures were approved by the University’s Institutional Review Board (IRB) and the executive director of the FQHC. Patients or caregivers sign consent forms, updated yearly, permitting the use of their medical data for program evaluation and research purposes. All data for this retrospective study were extracted from FQHC electronic medical records. Information about patient demographics (i.e. age, ethnicity, gender) was pulled from electronic medical records. The presence of a DSM-5 anxiety disorder was coded dichotomously from extracted International Statistical Classification of Diseases (ICD)-10 codes in the health care record. Health care providers entered these codes during visits over the previous 4 years. Health care utilization was operationalized as the number of (1) medical encounters and (2) behavioral health encounters (BHCs) over the prior 4 years. BHCs are visits that occur within the context of an integrated behavioral health care (IBHC) set up (IBHC; Blount, 1998). In this context, mental health professionals, referred to as BHCs, are available onsite at primary care facilities to provide mental health counseling to individuals who present to their primary care providers (PCPs) with mental health concerns. In the IBHC set up, a PCP refers patients with mental health concerns to BHCs using a warm handoff for a same-day, mental health consultation. In this study, behavioral health visits were entered as covariates, while number of medical encounters (excluding behavioral health visits) over the prior 4 years was the dependent variable. Health care utilization that occurred in systems outside of the FQHC was not captured.
Analytic approach
We began by computing descriptive statistics and bivariate correlations among study variables. To address the first aim, we first conducted an independent samples t-test comparing past-year medical service utilization in anxious and non-anxious youth. To further address the first aim and to evaluate the second aim, we conducted a hierarchical multiple regression predicting past-year medical service utilization from age, gender, and Hispanic ethnicity (step 1); the presence of a documented anxiety disorder (step 2); number of past-year behavioral health visits (step 3); and interactions between the demographic variables and a documented anxiety disorder (step 4). Step 2 allowed us to determine whether possible differences in service utilization rates among anxious and non-anxious youth would remain even after controlling for demographic covariates. Step 4 allowed us to examine whether Hispanic ethnicity interacted with the presence of a documented anxiety disorder to predict medical service utilization.
Results
Bivariate correlations between study variables and past 4-year medical service utilization are presented in Table 1. Point-biserial correlations revealed that anxious youth utilized behavioral health and medical services significantly more than did non-anxious youth. Hispanic youth utilized health care services at higher rates (M = 7.58, standard deviation (SD) = 6.27) than non-Hispanic youth (M = 5.81, SD = 5.88). Age was inversely related to past 4-year medical service use.
Consistent with our first hypothesis, we found that youth with an anxiety disorder had significantly more medical visits over the 4-year period (M = 9.70, SD = 7.20) than youth without an anxiety disorder (M = 6.69, SD = 6.12), t(145) = –4.94, p < .001. Levene’s test indicated unequal variances (F = 5.07, p = .024), so degrees of freedom were adjusted from 6960 to 145.
A hierarchical multiple regression analysis revealed a significant association between having a diagnosed anxiety disorder and increased medical service utilization, even after controlling for age, sex, and Hispanic ethnicity (Table 2, step 2). After controlling for behavioral health visits, a significant effect of anxiety disorder status on past four-year medical service utilization remained (step 3). Anxiety disorder status did not interact with any demographic variables to predict medical service utilization (step 4).
Hierarchical multiple regression analysis predicting past 4-year medical service utilization (N = 6962).
SE: standard error;.
p < .05, **p < .01, and ***p < .001.
Discussion
Anxiety disorders were relatively rare in this sample of pediatric primary care patients, with only 2 percent of youth having a documented DSM-5 anxiety disorder. Consistent with our hypotheses, the presence of an anxiety disorder was associated with elevated primary health care utilization. Although non-Hispanic youth had a slightly higher prevalence rate of anxiety disorders than Hispanic youth, Hispanics utilized primary health care services significantly more often than non-Hispanics.
These results are consistent with previous work from the Netherlands, which suggested that families with anxious offspring utilize health care services at higher rates than families without anxious offspring (Bodden et al., 2008). We found that anxious youths in the United States also utilize primary health services at higher rates than non-anxious youths. These findings are notable in light of evidence that elevated health care costs are particularly burdensome for families with youth offspring (Cohen and Kirzinger, 2014). We also bolstered prior work suggesting that Hispanics utilize primary care services significantly more often for mental health concerns than non-Hispanics (Cabassa et al., 2006). Results revealed that Hispanic youth utilize health care services at higher rates than non-Hispanic youth. Ethnicity and anxiety disorder did not interact in relation to health care utilization patterns, suggesting that anxiety disorder status and ethnicity make independent contributions to heath care utilization patterns.
A number of limitations merit mentioning. First, the rate of anxiety disorder diagnoses in this sample was relatively low. It may be that case that stigma related to anxiety disorder diagnoses among Hispanics may have contributed to lower rates of anxiety disorder diagnoses in medical records. The methodology also included a relatively conservative measure of anxiety disorders (i.e. providers identifying the presence of an anxiety disorder and recording it in the electronic medical record). In addition, the clinics where data collection took place employed no systematic attempts to screen and diagnose youth for anxiety disorders at the time these data were collected. It is possible that rates of anxiety disorders are higher in this sample than are indicated by the medical records. It is also possible that due to the relatively lower mean age of the same (12 years old), several youth may not have begun to report symptoms of anxiety to their medical provider. Critically, despite the low levels of recorded anxiety in this sample, youth with an anxiety disorder diagnosis utilized health care significantly more than youth without this diagnosis. The rates of anxiety in this sample are much smaller than rates in the larger youth population (Beesdo et al., 2009). To the extent some youth with undiagnosed anxiety disorders were part of the non-anxious youth group in this study, our results may be an underestimation of the degree to which anxious youth overutilize medical services compared to their non-anxious counterparts. It is also probable that in a more symptomatic cohort, the effects may have been larger than those in the current sample. Second, this assessment relied on a geographically limited sample of individuals residing the southern region of the United States. This sample was half Hispanic, and over half of the sample qualified for Medicaid, suggesting that their families may have had limited resources. Future work should endeavor to replicate these findings in more geographically and financially diverse samples. A third limitation is that it is possible that individuals included in this study may have attended health care services outside of the FQHC. Future studies should aim to control more rigorously for alternate forms of health care utilization. Fourth, it is likely that other co-occurring physical problems may contribute to rates of health care utilization (e.g. problems with asthma). Critically, when analyses were run controlling for the presence of a chronic illness, the same pattern of results and magnitude of the effect emerged—that is, the presence of an anxiety disorder still increased service utilization rates significantly in these youth and with the same standardized beta (.078) as when the analyses did not control for co-occurring medical problems
It also merits consideration that most youth are unable to attend the doctor without some assistance from their parents. It is therefore likely that parental anxiety could play a role here. Indeed, evidence suggests that parental anxiety vulnerability factors interact with offspring factors in predicting parental sick-role reinforcement behavior (e.g. taking child to the doctor or telling their children to go lay down in the presence of ambiguous physical sensations; Bilsky et al., 2018). Although, that possibility is outside the scope of the current article, it represents a valuable future direction for research. Finally, over half of the sample is currently receiving health insurance through Medicaid. This suggests that many of these families may lack resources for health care, which could make taking time off of work or school to attend the doctor prohibitive. Socioeconomic status may also be an important consideration in this regard.
Notwithstanding these limitations and considerations, this study suggests that screening and adequate treatment of youth anxiety disorders in primary care may be critical for decreasing unnecessary medical utilization and health care costs. Furthermore, in light of evidence that parental reinforcement of “sick role behavior” in the context of anxiety-relevant arousal is related to the development and maintenance of anxiety and related responses (Leen-Feldner et al., 2008), PCPs may benefit from providing parents with specific strategies to avoid reinforcing sick-role behavior. This may be helpful in reducing anxiety-related overutilization of health care services.
Footnotes
Acknowledgements
The authors wish to thank Kathy Grisham of Community Clinic at St. Francis House for partnering with them in the provision and evaluation of integrated care services and Bryan Smith for his technical assistance. Dr Bridges would like to disclose that she receives fees for consultation and supervision from Community Clinic at St. Francis House.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported in part by a National Science Foundation Graduate Research Fellowship Award under Grant No. 2015179871 awarded to S.A.B.