The authors hypothesized that individuals with intellectual and developmental disabilities (IDDs) are more likely to have an emergency department (ED) admission for nontraumatic dental conditions (NTDCs). The authors analyzed 2009 U.S. National Emergency Department Sample data and ran logistic regression models for children ages 3–17 years and adults age 18 years or older. The prevalence of NTDC-related ED admissions was 0.8% for children and 2.0% for adults. Children with IDDs were at increased odds of NTDC-related ED admission, but this difference was not statistically significant (odds ratio [OR] = 1.06; 95% confidence interval [CI] = 0.91, 1.23). Adults with IDDs had significantly lower odds of an ED admission for NTDCs (OR = 0.49; 95% CI = 0.44, 0.54). Children with IDDs are not at increased odds of NTDC-related ED admissions, whereas adults with IDDs are at significantly reduced odds.
Use of the emergency department (ED) for management of nontraumatic dental conditions (NTDCs) is a growing problem in the United States (Lee, Lewis, Saltzman, & Starks, 2012; Wall, 2012), especially among vulnerable population subgroups such as individuals without health insurance, Medicaid enrollees, and racial/ethnic minorities (Hong et al., 2011; Lewis, Lynch, & Johnston, 2003; McCormick, Abubaker, Laskin, Gonzales, & Garland, 2013; Nalliah, Allareddy, Elangovan, Karimbux, & Allareddy, 2010; Okunseri et al., 2013). It is common for state legislatures to eliminate Medicaid dental benefits to balance budgets, which leads to greater unmet dental care needs and subsequent ED admissions (Cohen, Manski, & Hooper, 1996; Wallace, Carlson, Mosen, Snyder, & Wright, 2011). Individuals with poor access to office-based dental care services commonly use the ED for management of dental pain and infection (Dorfman, Kastner, & Vinci, 2001; Patel, Miner, & Miner, 2012). Most studies have reported individual-level risk factors associated with ED admission for NTDCs (e.g., age, gender, race/ethnicity, income, health insurance status and type; Cohen et al., 1996; Hong et al., 2011; Lee et al., 2012; Lewis et al., 2003; McCormick et al., 2013; Nalliah et al., 2010; Okunseri et al., 2013; Wall, 2012; Wallace et al., 2011). Despite evidence suggesting that individuals with intellectual and developmental disabilities (IDDs) are at risk for dental disease and poor access to timely dental care (Chi et al., 2011; Morgan et al., 2012), no published studies to date have examined NTDC-related ED admissions for individuals with IDDs.
It is important to understand NTDC-related ED admissions for individuals with IDDs for three main reasons. First, individuals with IDDs are a vulnerable population subgroup whose dental needs are more efficiently managed within office-based dental homes. For instance, NTDC-related ED admissions are costly, with per patient charges ranging from $381 to $526 (Nagarkar, Kumar, & Moss, 2012; Nalliah et al., 2010). Second, most NTDCs in the ED are managed nondefinitively with analgesics or antibiotics, which do not address the underlying cause of dental disease and can lead to repeat ED visits (Davis, Deinard, & Maïga, 2010; McCormick et al., 2013; Pajewski & Okunseri, 2012). Thus, management of NTDCs in the ED is an inefficient use of scarce health care resources. Third, there is a dearth of studies on ED use for patients with IDD and no data on NTDC-related ED admission rates.
Although there are limitations associated with adopting diagnosis-based approaches to identify individuals with IDDs in secondary datasets (e.g., underrepresentation of individuals with IDDs), the availability of such datasets enables researchers to begin addressing critical knowledge gaps that limit progress in the field. The knowledge generated is a starting point that will make it possible to design future population-based studies that address limitations with existing datasets, develop evidence-based polices and interventions, evaluate outcomes associated with efforts aimed at reducing NTDC-related ED admissions, and improve the oral and systemic health of individuals with IDDs. The goal of this study is to analyze a nationally representative dataset to test the hypothesis that among ED utilizers in the U.S., individuals with IDDs are more likely to be admitted to the ED for a nontraumatic dental condition than individuals without IDDs.
Materials and Methods
Study Design, Participants, and Data
This study was a cross-sectional analysis of 2009 National Emergency Department Sample (NEDS) data. NEDS, a publicly available dataset, is the largest all-payer ED dataset in the United States and contains information on over 29 million ED admissions from 29 states (Healthcare Care Utilization Project, 2009). Our analyses focused on individuals ages 3 years and older, all of whom were admitted to a hospital ED in 2009. We excluded participants under age 3 years because IDDs, our main predictor variable, are typically diagnosed after a child's third birthday (Pinto-Martin, Dunkle, Earls, Fliedner, & Landes, 2005). The final study sample consisted of 26,791,871 individuals ages 3 years and older who were admitted to an ED in 2009 (n = 4,325,309 children ages 3–17 years and n = 22,466,562 adults ages 18 years and older). This study was exempted from human subjects approval by the University of Washington Institutional Review Board.
The outcome was whether the ED admission was for an NTDC (no/yes). The etiology of NTDCs is preventable and includes diseases such as dental caries (tooth decay) and periodontitis (gum disease). We defined NTDCs using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes from any of the 15 diagnosis fields in the NEDS. The specific ICD-9-CM codes, which have been published previously (Okunseri et al., 2008), included 521 (diseases of hard tissues of teeth), 522 (diseases of pulp and periapical tissues), 523 (gingival and periodontal diseases), 525.3 (retained dental root), 525.9 (unspecified disorder of the teeth and supporting structures), and 873.63 (internal structures of the mouth, without broken tooth).
Main predictor variable
The main predictor variable was IDD status (no/yes). We adopted a diagnosis-based approach of identifying IDDs developed previously by a team of physicians, clinical psychologists, and dentists based on relevant ICD-9-CM codes present in any of the 15 NEDS diagnosis fields (Chi et al., 2010; Chi et al., 2012). IDDs are associated with lifelong, nonacquired cognitive deficits or impairments. The codes included 299 (autism), 317–319 (mental retardation), 343 (cerebral palsy), 741 (spina bifida), 758 (Down syndrome), 759.5 (Tuberous sclerosis and Bourneville's diseases), 759.83 (Fragile X syndrome), and 760.71 (fetal alcohol syndrome).
We used the Behavioral Model for Vulnerable Population (Gelberg, Andersen, & Leake, 2000) to select relevant covariates for our explanatory model. There were two predisposing variables: age (modeled as a categorical variable) and sex (male/female). The three enabling variables were health insurance type (private or health maintenance organization/Medicaid or Medicare/self-pay/no charge/other), median community income (quartiles), and size of community (a six-level categorical variable ranging from large central metro to noncore). The two need variables were IDD status (defined previously) and number of chronic conditions (0, 1, 2, or more, measured by aggregating the total number of chronic health conditions as indicated in the 15 NEDS chronic condition indicators).
After generating descriptive statistics separately for children and adults, we evaluated the bivariate relationships between model covariates and our outcome measure using unadjusted odds ratios (ORs; α = .05). We generated two multiple variable logistic regression models to test the hypotheses that, among ED admissions, children and adults with IDDs would be more likely to be admitted to the ED for an NTDC than those without IDDs. All analyses were adjusted for hospital clusters and strata and weighted by patient discharge weights to derive nationally representative ORs. All statistical analyses were completed using SAS Version 9.3 for Windows and the SURVEYLOGISTIC procedure (SAS Institute, Cary, NC).
Characteristics of Study Participants
About 0.7% of children and 0.4% of adults with an ED admission had an IDD (see Table 1). Nearly one in five children (18.3%) had one or more chronic conditions. In terms of health insurance, 45.5% were publicly insured (Medicaid or Medicare), 41.6% were privately insured, and 8.5% were self-pay (uninsured). For adults, 20.7% had one chronic condition, and 35.8% had two or more chronic conditions. In terms of health insurance, 43.5% were publicly insured (Medicaid or Medicare), 31.2% were privately insured, and 18.8% were self-pay (uninsured). The prevalence of NTDC-related ED admissions was 0.8% for children and 2.0% for adults (see Table 1). There was a bimodal distribution, with an initial peak prevalence at age 6 years (0.9%) and a second peak at age 27 years (4.4%; see Figure 1).
Unadjusted Regression Models
In the bivariate analyses, having an IDD, for both children and adults, was associated with significantly lower odds of NTDC-related ED admission (see Table 2). Older children (ages 6–17 years) had significantly greater odds of an NTDC-related ED admission compared with younger children (ages 3–5 years), whereas older adults (ages 50 years and older) had significantly lower odds of an NTDC-related ED admission compared with adults ages 18–49 years. Publicly insured, self-pay, and uninsured individuals had a significantly greater odds of an ED admission related to NTDCs compared with those with private insurance. There was also an income gradient present, with individuals living in lower income communities at greater odds of ED admission for NTDCs. Children and adults with chronic health conditions had significantly lower odds of NTDC-related ED admission.
Covariate-Adjusted Regression Models
In the covariate-adjusted logistic regression models, having an IDD was associated with NTDC-related ED admissions for children and adults, but the relationship was statistically significant only for adults (OR = 0.49; 95% confidence interval [CI] = 0.44, 0.54; p < .0001) and not for children (OR = 1.06; 95% CI = 0.91, 1.23; p > .05; see Table 3). Whereas an income-related gradient was present for children in the odds of an NTDC-related ED admission, there was no gradient for adults. The other findings were consistent with results from the unadjusted regression models.
This is the first known study that examined NTDC-related ED admissions for individuals with IDDs. We hypothesized that, among U.S. patients admitted to the ED, those with IDDs would be more likely to have a NTDC-related ED admission than those without IDDs. Based on data from the 2009 NEDS, we found that there was no significant difference in ED admissions for NTDCs by IDD status for children. Adults with IDDs were significantly less likely to use the ED for NTDCs than adults without IDDs. In addition, other factors from our conceptual model (e.g., male sex, not having private health insurance, living in a lower income or rural community, and not having a chronic health condition) were associated with significantly greater odds of having an NTDC-related ED admission.
Our first finding was that children with IDDs were not significantly more likely to be admitted to the ED for NTDCs than children without IDDs. Previous work has indicated that children with developmental disabilities are two to three times as likely to have an ED visit as children without developmental disabilities (Boulet, Boyle, & Schieve, 2009). There are no comparable studies from the literature, but we have two plausible explanations for our findings that should be evaluated through additional research. First, although young children with IDDs have difficulties with initiating use of preventive dental care (Chi et al., 2011) and establishing a dental home, they may not have greater difficulties than children without IDDs in obtaining subsequent preventive dental care (Chi et al., 2010; Chi et al., 2012). Although patients with IDDs are at greater risk for tooth decay (Morgan et al., 2012), no difference in NTDC-related ED admissions may mean that children with IDDs have a regular place to go for treatment of NTDCs. This underscores the importance of ensuring routine preventive dental care for all children with IDDs. Second, children with IDDs may have impaired peripheral nociceptors (Symons, 2011) or difficulties communicating dental pain (Breau & Burkitt, 2009), which could lead to reduced NTDC-related ED admissions even though underlying symptoms are present. Future research should examine the extent to which underreporting of dental pain that could lead to reduced ED admissions is a problem for children with IDDs.
Our second finding was that adults with IDDs were significantly less likely to be admitted to the ED for NTDCs than adults without IDDs. There are no studies from the dental literature to which we can directly compare our findings. However, two studies suggested that general ED use is higher in adults with IDDs (Lunsky et al., 2011; Rasch, Gulley, & Chan, 2013) and two studies have focused on ED use in adults with IDDs (Lunsky & Elserafi, 2011; Venkat et al., 2011). There are three main explanations for our finding. First, a large proportion of adults with IDDs live in group homes (Morgan et al., 2012) and may have access to palliative dental care that helps prevent NTDCs. However, this explanation is unlikely given previous work that has identified dental care as a notable gap among adults with IDDs (Parish, Moss, & Richman, 2008). A more important factor in lower odds of NTDC-related ED visits for adults with IDDs may be having a health care advocate who can reinforce positive oral health behaviors. Second, lower NTDC-related ED use for adults with IDDs may actually indicate barriers to dental care that stem from an inability to express pain or discomfort, transportation problems, or general neglect. Previous work has identified neglect and poor care as common forms of mistreatment of individuals with IDDs (Oktay & Tompkins, 2004). Third, compared with national edentulism (i.e., total tooth loss) rates (Dye et al., 2007), nearly three times as many adults with IDDs are edentulous (Morgan et al., 2012), which may reduce the odds of NTDCs and concomitant NTDC-related ED admissions. This latter explanation is consistent with NTDC-related ED admissions rates presented in Figure 1, in which adult rates peak in the late 20s and trend downward with older age. Despite this trend, it is important to note that NTDC-related ED admissions rates for older adults are not zero. We ran post hoc interaction models between IDD status and age and found that, compared with adults without IDDs ages 18–21 years, adults with IDDs ages 65–75 years had significantly greater odds of NTDC-related ED admissions (OR = 1.91; p = .02).
Collectively, these findings suggest that despite overall lower odds of NTDC-related ED admissions for adults with IDDs, there are subgroups of adults with IDDs who are at increased risk for ED admissions. Future research should continue to identify these subgroups, with the ultimate goal of developing interventions that reduce ED admissions by promoting oral health behaviors such as regular toothbrushing with fluoride toothpaste, decreased frequency of carbohydrate intake, and regular dental examinations and treatment when necessary.
Covariates from all three domains (e.g., predisposing, enabling, and need) of the Behavioral Model for Vulnerable Population were significantly associated with NTDC-related ED admissions. In terms of predisposing factors, male sex was associated with significantly greater odds of an ED admission for children and adults. This is consistent with a previous study on inpatient hospitalization (Chi & Masterson, 2013), but inconsistent with results presented from an ED utilization study based on data from the National Hospital Ambulatory Medical Care Survey (Lee et al., 2012). Older children had greater odds, whereas older adults had significantly lower odds, of NTDC-related ED admissions. The reasons for these trends for children are unknown. Future work should further clarify the types of teeth (primary versus permanent) that are implicated in pediatric NTDCs, which has implications for the development of training programs for ED health providers on how to diagnosis and manage NTDCs in children.
In terms of enabling factors from our model, not having private health insurance and living in a lower income community were significantly associated with greater odds of NTDC-related ED use for both children and adults. These findings are not surprising given that socioeconomic vulnerability is one of the most consistent correlates of ED use for NTDCs (Hong et al., 2011; Lewis et al., 2003; McCormick et al., 2013; Nalliah et al., 2010; Okunseri et al., 2013). In addition, although community size was not significantly associated with ED use for children, adults in smaller communities had significantly greater odds of ED use than adults living in metropolitan areas. These findings suggest heterogeneity in how community-level enabling factors influence NTDC-related ED admissions. Community-based interventions in rural areas might focus on reducing ED admissions among adults, whereas intervention targeting socioeconomically vulnerable individuals should focus on both children and adults.
In terms of the other need factor in our model, we found that, for both children and adults, having a chronic condition was associated with significantly lower odds of NTDC-related ED admission. This finding was inconsistent with our original hypothesis but runs parallel to a study from Canada on factors related to ED use (Moineddin, Meaney, Agha, Zagorski, & Glazier, 2011). Future investigations should identify subgroups of individuals with chronic conditions at increased risk for NTDC-related admissions.
Beyond the predisposing, enabling, and need variables that we examined in this study, there is a need to identify the behavioral and social factors related to ED use for individuals with IDD. This could be accomplished through primary data collection methods, which would lead to information needed to develop strategies that prevent ED use for NTDCs.
There are four main study limitations. First, because the NEDS data were cross-sectional, our findings are associations rather than causal relationships. Our hypotheses should be investigated further with other national ED datasets (Owens et al., 2010) and with prospective studies. Second, our measure of IDD was diagnosis-based. If the ED staff missed diagnoses, conditions were miscoded, or there was underreporting by caregivers, there is a possibility of classification bias for our main predictor variable. The prevalence of IDD in our study (0.4%–0.7%) is lower than rates reported in previous studies (Chi et al., 2010; Maulik, Mascarenhas, Mathers, Dua, & Saxena, 2011; Murphy, Yeargin-Allsopp, Decouflé, & Drews, 1995), which suggest that a strictly diagnosis-based IDD case-finding protocol underidentifies IDD. Future investigations should adopt diagnosis-based and noncategorical protocols whenever possible to identify IDDs. Third, we assume coding accuracy of NTDCs, but as with our IDD measure, there is a possibility of underidentification. Validation studies could be conducted to evaluate the extent to which IDDs are undercoded during ED visits. Fourth, our models did not include behavioral and social factors that may be important mediators or moderators of the relationship between IDD and NTDC-related ED admissions. For instance, oral health–related behaviors such as access to preventive dental care, topical fluoride exposure, and diet may be important factors to consider when modeling ED admissions related to NTDCs. Future research should further clarify the role of behaviors and social context as the next step in developing interventions and policies aimed at reducing NTDC-related ED admissions. In particular, there is a need for mixed-methods research, with an emphasis on qualitative research, on NTDC-related ED admissions to understand the complex factors that contribute to ED use.
Despite these limitations, our study has clinical, policy, and research significance. In terms of clinical significance, our findings support efforts to ensure that publicly insured and uninsured patients have access to regular preventive and restorative dental care provided in an office-based setting. Private practice dental offices and community health centers should be urged to have after-hours policies that do not simply refer patients to the ED in cases of dental emergencies. In addition, ED physicians, nurses, and staff must be adequately trained to manage patients with NTDCs. Patient management includes the use of appropriate referral networks to ensure that patients receive definitive treatment and do not return to the ED with the same problems. In terms of policy significance, private health insurance plans, including state Medicaid programs, could implement pay-for-performance measures that incentivize dentists for managing all NTDCs within a non-ED setting. There is also the need to reinstate dental coverage for adults within state Medicaid programs that have eliminated dental benefits, which would help reduce ED admissions for NTDCs and ED crowding (Institute of Medicine, 2007; Sun et al., 2013; Wallace et al., 2011), and provide dental benefits for all Medicare beneficiaries (Manski et al., 2010). More broadly, federal health care reform that incorporates dental care would help to address the limitations associated with Medicaid and the dental safety net (Bailit & D'Adamo, 2012). In terms of research significance, this is only the first study to examine NTDC-related ED admissions with an emphasis on individuals with IDDs. Additional research is needed to assess the behavioral and social determinants of ED use, the underlying mechanisms that help prevent ED admissions for patients with IDDs, and the extent to which repeat ED admissions are a problem for this vulnerable population. Qualitative methods should be used in conjunction with quantitative studies to uncover important factors related to NTDC-related ED admissions.
In conclusion, we found that children with IDDs were not more likely to be admitted to the ED than children without IDDs, whereas adults with IDDs were significantly less likely to have an ED admission for NTDCs. Other factors were associated with significantly greater odds of NTDCs for both children and adults, including male sex, nonprivate health insurance, living in a lower income community, and not having a chronic health condition. ED use for NTDCs has become a major public health problem. Management of NTDCs within ED settings is costly, inefficient, and contributes to ED crowding. Because most NTDCs are preventable, new strategies are needed that focus on improving oral health–related behaviors and ensuring proper access to timely and appropriate dental care services in high-risk individuals. Such strategies are likely to reduce NTDC-related ED admissions and improve the oral health of all vulnerable individuals.
This research was funded by the National Institute of Dental and Craniofacial Research (NIDCR) and National Institutes of Health (NIH) Grant Numbers K08DE020856 and TL1TR000422 and Health Resources and Services Administration (HRSA) Grant Number R40MC26198.
Donald L. Chi, Erin E. Masterson, and Jacqueline J. Wong, University of Washington.