Are There Disparities in Emergency Care for Uninsured, Medicaid, and Privately Insured Patients?

Transcription

1 ACAD EMERG MED d November 2003, Vol. 10, No. 11 d Are There Disparities in Emergency Care for Uninsured, Medicaid, and Privately Insured Patients? Abstract CharleneBabcockIrvin,MD,JamesM.Fox,MD,BradleySmude,MD Objectives: To determine if there are any differences in proportion of high-acuity care and low-acuity care provided to uninsured, Medicaid-insured, and privately insured emergency department (ED) patients. Methods: This was a retrospective, observational study using physician level of service provided as a marker for acuity. The study used computerized billing data ( ) from an urban, teaching, Level I trauma center with 75,000 visits per year. All uninsured and Medicaid patients (age groups: pediatric, \18 years; adult, years) were compared by physician level of service billed to Blue Cross Blue Shield (BCBS) patients and analyzed using chi-square. Low-acuity care was defined by CPT codes and High-acuity care was defined by CPT codes and Results: There were 152,379 total ED visits, with 13.2% BCBS (5,273 pediatric, 14,951 adult), 29.6% Medicaid (20,578 pediatric, 24,511 adult), and 8.1% uninsured (1,879 pediatric, 10,405 adult) patients. The percent of pediatric BCBS, Medicaid, and uninsured patients receiving low-acuity care was 30%, 35.7%, and 35.8% (p \ 0.001), respectively; and for highacuity care, it was 7.8%, 6.1%, and 6.8% (p \ 0.001), respectively. The proportion of adults within these groupings was 13.7%, 13.2%, and 17.9% (p \ 0.001) for low-acuity care, and 28.5%, 22.9%, and 16.7% (p \ 0.001) for highacuity care, respectively. Conclusions: Whereas there were some statistically discerned differences between insurance groupings for proportionate receipt of low-acuity care and high-acuity care among both the pediatric and adult populations, the magnitude of most differences noted was not large, and may not reflect important differences in health care need or ED use based on insurance. Key words: emergency department acuity; uninsured; Medicaid; insured patients. ACADEMIC EMERGENCY MEDICINE 2003; 10: The assumption that the emergency department (ED) is occasionally inappropriately used for nonurgent problems is not new. 1 3 This assumption is based on previous studies that have used a retrospective, discharge/admission diagnosis driven definition of nonurgent, 4 or studies that have used a subjective, health care professional rated assessment of urgency at triage as a marker for nonurgent. 5 7 These studies have revealed dramatically different estimates of nonurgent ED care that range from 5% to 82%. 3,8-9 Because this care is considered by many to be more cost-effectively delivered in the primary care setting, 10 there have been attempts to identify populations with higher rates of inappropriate use of the ED. Some have suggested that redirecting nonurgent care out of the ED may provide some cost savings, and may also help decrease problems with ED overcrowding. 2,3 Populations identified as ED overusers for nonurgent From St. John Hospital and Medical Center, Detroit, MI (CBI, JMF, BS); and Wayne State University School of Medicine, Detroit, MI (CBI, JMF). Received February 28, 2003; revisions received April 15, May 1, and 10, 2003; accepted May 14, Presented at the SAEM Mid-Atlantic regional meeting, Washington, DC, March 2003; the SAEM Western regional meeting, Scottsdale, Arizona, April 2003; and the SAEM New York regional meeting, New York, NY, April Correspondence and reprint requests: Charlene Babcock Irvin, MD, Miller Ct., Chesterfield, MI cbi@123.net. doi: /s (03) care include Medicaid patients in general, 5,6 and Medicaid-insured children specifically. 11 In addition to issues regarding ED nonurgent use by Medicaid patients, there is concern that uninsured patients delay care, and thus may suffer increased serious or even life-threatening conditions Some studies suggest that uninsured patients are also more likely to be hospitalized. 15 Other studies suggest that lack of insurance does not lead to increased nonurgent ED care. 16 Comparisons of ED high-acuity care by differently insured populations suffer from the same problems as studies of nonurgent care, namely, the problems with the definition of what constitutes an urgent ED visit. One method to compare the urgencies or acuities of ED treatment by differently insured patients may be to use the distribution of the levels of service provided. Using the level of service provided takes into account the presenting complaint, the severity of the symptoms, the complexity of data to be reviewed, the risks of morbidity or mortality, and the patient management in the ED. This may help address more objectively the issues of whether uninsured patients are more likely to present for more serious conditions and Medicaid patients present for more nonurgent conditions, when compared with privately insured patients. This information may be useful to policymakers as they initiate specific programs to address disparities in health care.

2 1272 Irvin et al. d ED CARE OF UNINSURED, MEDICAID, AND INSURED PATIENTS The purpose of this study was to compare the distribution of physician level of service provided to Medicaid-insured, uninsured, and privately insured patients. The null hypothesis is that there will be no difference based on insurance in the proportions of patients provided high- and low-acuity levels of service. METHODS Study Design. A retrospective, observational study of computerized reimbursement data from a large, urban, teaching, Level 1 trauma center (75,000 patient visits per year) was performed. This study was considered exempt from informed consent by our institutional review board, because it was a review of previously collected billing information without patient identifiers. Study Setting and Population. The data collected included physician level of service provided to all patients, uninsured patients, Medicaid patients, and privately insured patients. Blue Cross Blue Shield (BCBS) was chosen as the privately insured patient population studied because it represented the largest privately insured group at this institution. In our state, BCBS does not provide automobile-related health insurance or worker s compensation related health insurance. These two categories of insurance are somewhat unique because the patient may have insurance for that particular ED visit, but not for other health issues. Using only BCBS excluded these partially insured patients from our evaluation because they represented neither privately insured, uninsured, nor Medicaid patients, and allowed us to compare Medicaid and uninsured with a purely private insurance group. The data for specific insurance groups were analyzed based on age of patient (\18 years, years) and level of service provided. Patients older than age 64 were excluded, because the majority of these patients have Medicare insurance with or without additional private insurance. This study only evaluated Medicaid, uninsured, and privately insured patients. Study Protocol. The physician level of service provided was assigned by specifically trained billing coders who are required to substantiate the level of service assigned based on documentation included in the physician record. Only one level of service was assigned to each patient. The level of service was assigned before the initiation of this study, and the billing coders had no knowledge of this study at any time. Measures. The levels of services recorded included and (critical care). The total number of patient visits in each insurance category, for each level of service, was extracted electronically from a large billing database for both age groups. These data then were manually input into an Excel spreadsheet for analysis. No other demographic information was extracted except the physician level of service billed. Generally, the two lowest levels of physician services, and 99282, involve minimal decision making, no blood testing or ancillary testing, and no or minimal risk of morbidity or mortality. 17 Both of these levels of service were included in the low-acuity level of service. The two highest levels of service, and 99291, involve complex decision making, the interpretation of numerous tests, and patients with a high risk of morbidity or mortality. 17 Both of these levels of service were included in the highacuity level of service. Table 1 lists the published criteria and published examples of the different levels of service. 17 The time period analyzed included all of 2000 and Data were collected in February 2003 to allow more than a year for any insurance discrepancies to be resolved (i.e., cases initially recorded as uninsured and after billing process determined to be insured during the time period of the ED visit). Data Analysis. Data were analyzed using chi-square analysis. RESULTS There were 152,379 patient visits evaluated during the study period (41,951 pediatric and 110,428 adult patient visits). BCBS represented 13.2% of the patient visits (12.6% of the pediatric and 13.5% of the adult patient visits). Medicaid patient visits represented 29.6% of the patient visits (49.1% of the pediatric and 22.2% of the adult patient visits). Uninsured patient visits accounted for 8.0% of the patients treated (4.5% of the pediatric and 9.4% of the adult patient visits). Table 2 describes the different levels of service provided to all patients, BCBS, Medicaid, and uninsured patients. Figure 1 shows the distribution of pediatric patients in the different levels of service by insurance status. Approximately 33.9% of all pediatric patients received the low-acuity levels of service. For the different insurance categories, 30% of BCBS, 35.7% of Medicaid, and 35.8% of uninsured pediatric patients received the low-acuity levels of service (p \ 0.001). The proportion of pediatric patients provided the lowacuity level of service was 5.7% (95% CI ¼ 4.3% to 7.1%, p \ 0.001) greater for Medicaid patients and 5.8% (95% CI ¼ 3.3% to 8.3%, p \ 0.001) greater for uninsured patients when compared with BCBS. The proportion of pediatric patients in the high-acuity levels of service for different insurance categories was 7.8% for BCBS, 6.1% for Medicaid, and 6.8% for uninsured (p \ 0.001). The proportion of pediatric Medicaid patients provided the high-acuity level of

3 ACAD EMERG MED d November 2003, Vol. 10, No. 11 d TABLE 1. Published Definitions Describing the Different Levels of Service and Supporting Examples ED visit (evaluation and management) requires these three key components: 1. Problem-focused history 2. Problem-focused examination 3. Straightforward medical decision making The presenting problem(s) are usually self-limited or minor 1. Suture removal from a well-healed, uncomplicated laceration 2. Patient visit for tetanus toxoid immunization 3. Patient visit for several uncomplicated insect bites ED visit (evaluation and management) requires these three key components: 1. Expanded problem-focused history 2. Expanded problem-focused examination 3. Medical decision making of low complexity The presenting problem(s) are usually of low to moderate severity year-old patient presents with painful sunburn with blister formation on the back. 2. Child presenting with impetigo localized to the face 3. Patient with a rash on both legs after exposure to poison ivy ED visit (evaluation and management) requires these three key components: 1. Comprehensive history 2. Comprehensive examination 3. Medical decision making of high complexity The presenting problem(s) are usually of high severity and pose an immediate, significant threat to life or physiologic function. 1. Patient with a complicated overdose requiring aggressive management 2. Patient with a new onset of rapid heart rate requiring intravenous medications 3. Patient exhibiting active, upper gastrointestinal bleeding A critical illness or injury acutely impairs one or more vital organ systems such that there is a high probability of imminent or life threatening deterioration in the patient s condition. Critical care requires high complexity decision making to access, manipulate, and support vital systems functions to treat single or multiple vital organ system failure and/or to prevent further life threatening deterioration of the patient s condition year-old with acute respiratory failure from asthma year-old who sustained a liver laceration, cerebral hematoma, flailed chest, and pulmonary contusion after being struck by an automobile year-old woman who, following a hysterectomy, suffered a cardiac arrest associated with a pulmonary embolus service was 1.7% (95% CI ¼ 0.8% to 2.4%, p \ 0.001) less than BCBS patients. The proportion of pediatric uninsured patients provided the high-acuity level of service was not statistically different from BCBS patients (1.0% difference, 95% CI ¼ÿ0.3% to 2.3%, p \ 0.15). Figure 2 shows the distribution of adult patients in the different levels of service by insurance status. For the different insurance categories, 13.8% of BCBS, 13.2% of Medicaid, and 17.9% of uninsured patients received the low-acuity levels of service (p \ 0.001). The proportion of adult Medicaid patients in the lowacuity levels of service was not statistically different from BCBS (0.5% difference, 95% CI ¼ÿ0.2% to 1.2%, p ¼ 0.13), but the proportion of uninsured adult patients in the low-acuity levels of service was 4.1% (95% CI ¼ 3.2% to 5.1%, p \ 0.001) more than BCBS. For the different insurance categories, 28.4% of the BCBS patients, 22.9% of the Medicaid patients, and 16.7% of the uninsured adult patients were in the high-acuity levels of service (p \ 0.001). The proportion of Medicaid adult patients in the high-acuity level of service was 5.5% (95% CI ¼ 4.6% to 6.5%, p \ 0.001) less than BCBC. The proportion of uninsured adult patients in the high-acuity level of service was 11.7% (95% CI ¼ 10.7% to 12.7%, p \ 0.001) less than BCBS. DISCUSSION The purpose of this study was to evaluate the distribution of care provided to Medicaid and uninsured patients compared with privately insured patients. If Medicaid patients sought a substantially larger amount of nonurgent care in the ED, then the distribution of Medicaid patients would reflect a higher proportion of low-acuity care compared with privately insured patients. This method also could be

4 1274 Irvin et al. d ED CARE OF UNINSURED, MEDICAID, AND INSURED PATIENTS TABLE 2. Number of Patients Evaluated in the Different Levels of Service by Insurance and Patient Age Group BCBS % Medicaid % Uninsured % Pediatric 99, ,282 1, , ,283 2, , , , , , , Total 5,273 20,578 1,879 Adult (18 65 years) 99, ,282 1, , , ,283 5, , , ,284 3, , , ,285 4, , , , Total 14,951 24,511 10,405 Pediatric and adult 99, ,282 3, , , ,283 7, , , ,284 4, , , ,285 4, , , , Total 20,224 45,089 12,284 used to evaluate the distribution of uninsured care in the ED and determine if uninsured patients have a more substantial proportion of high-acuity ED visits. The findings from this study support the fact that Medicaid children receive a higher proportion of lowacuity ED care compared with privately insured patients, although the magnitude of this difference is only approximately 6%. This study also revealed that there is no difference in the proportion of low-acuity care provided to Medicaid adults compared with privately insured adults. Finally, contrary to our hypothesis, uninsured patients did not receive more high-acuity care compared with privately insured patients. In this study, there was a statistically significant difference in the proportion of Medicaid pediatric patients in the low-acuity levels of service compared with privately insured patients. The reasons for this finding are unclear, but this study did attempt to compare the magnitude of the nonurgent Medicaid children s care with privately insured pediatric patients. This magnitude is not as large as one might have predicted, considering that Medicaid ED visits occur three times as often as privately insured patients per 100 population. 18 Although Medicaid pediatric patients may come to the ED more frequently for care, the distribution of low-acuity care received is only slightly more (6%) than that of the privately insured pediatric patients. This suggests that not all of the additional Medicaid pediatric visits are for nonurgent (low-acuity) reasons. Indeed, the data presented in Figure 1 for pediatric patients show the Figure 1. Proportion of pediatric emergency department visits in the different levels of service. BCBS ¼ Blue Cross Blue Shield.

5 ACAD EMERG MED d November 2003, Vol. 10, No. 11 d Figure 2. Proportion of adult patients (aged18 64 years) in the different levels of service. BCBS ¼ Blue Cross Blue Shield. similarities of the proportions in the different levels of service provided. This is an interesting finding; its implications are not clear. The analysis of the Medicaid adult patients yields different results. Here, contrary to previous studies, 6 Medicaid adult patients were not more likely to receive low-acuity emergency care compared with privately insured patients. The reason behind more low-acuity care for Medicaid children but not for Medicaid adults is unclear. One potential explanation is that the urgency was so minor that the adults were willing to forgo care for themselves or seek care at a later time for these minor complaints, but did not carry the same disregard for potentially nonurgent complaints when it came to their children. Again, considering the increased volume of Medicaid patients visiting the ED, the findings of this study suggest that for adults, the additional visits are not all for nonurgent reasons. As specific programs with substantial expense are initiated to decrease nonurgent visits by Medicaid patients, 19 understanding the findings of this study may be helpful, because only a small, additional proportion of pediatric (and no additional adult) Medicaid patients received lowacuity care. Analysis of the proportion of uninsured patients in the different levels of service is also interesting. The proportion of uninsured children evaluated in the low-acuity levels of service was increased only 6% the same magnitude as the Medicaid children when compared with privately insured patients. The adult uninsured patients also had a slightly higher (4% higher compared with BCBS) proportion of low-acuity care compared with privately insured adult patients. According to the National Hospital Ambulatory Medical Care Survey (NHAMCS), the uninsured seek ED care at a higher rate (35/100 population per year), and privately insured patients seek ED care at a lower rate (20/100 population per year), 18 so it is somewhat surprising that the proportion of low-acuity care for uninsured children and adults is only slightly more than privately insured patients. This suggests that not all of the additional visits from uninsured patients are for low-acuity problems. There are two possible explanations for the finding that the magnitude of nonurgent care provided to the uninsured is only slightly more than privately insured patients (6% more for children and 4% more for adults). It may be that because they lack any other access to care, uninsured patients seek ED care for any problem, even very minor problems. A chronic rash may be considered a very minor ED problem, and one that would easily be considered nonurgent. However, the individual with no other place to find out what it was and how to treat it eventually may present to the ED for this nonurgent problem. Another explanation for the slightly increased proportion of nonurgent care provided to uninsured may be that the services given to uninsured patients were less for all complaints, thereby shifting the distribution toward the nonurgent levels of service. This has been suggested by a previous study that found that uninsured patients receive less care in the ED 20 and are less likely to be admitted. 21 However, the proportion of nonurgent care provided to the uninsured is only slightly greater than that of privately insured patients; if this is the case, it does not appear to be occurring on a large scale. Further investigation is needed to determine the reasons for the slight increase of nonurgent care provided to uninsured patients. Analysis of the high-acuity care provided to uninsured patients is more challenging. This study arbitrarily combined the two highest categories to describe high-acuity care. For critical care to be assigned, the physician must specifically document the amount of time of critical care provided. Improper documentation of this time may result in a critical case assigned the lower category of For this reason, and because there were only a small number of patients in the critical care category, we combined the two high-acuity codes to represent high-acuity care. However, if the data are analyzed using only the critical care (99291) category to describe high-acuity care, both uninsured pediatric and

6 1276 Irvin et al. d ED CARE OF UNINSURED, MEDICAID, AND INSURED PATIENTS adult patients had a statistically significant higher proportion of critical care compared with those with insurance. By combining critical care (99291) with to describe the high-acuity group, this finding in adults reversed and the uninsured patients had statistically significant less high-acuity care; for pediatrics, there was no statistical difference from privately insured patients. Because the conclusion can be opposite based on an arbitrary grouping to describe high-acuity care, further study is needed before absolute conclusions can be drawn regarding the direction of the real difference with regards to adult uninsured high-acuity care. Again, although statistical significance may be present, the magnitude of the difference is small, no matter how the data are grouped. LIMITATIONS There are several limitations in this study. First, this study used a new method to assess acuity of ED visits, namely, the physician level of service provided to the patient. Second, the two lowest levels of service collectively represented the low-acuity population, and the two highest levels of service collectively represented the high-acuity population. Also, no demographic or other information was collected on the patients, making it impossible to control for intrinsic patient characteristics that may have contributed to the results, independent of insurance status. Finally, this study relied on computerized billing data that were dependent on accurate coding and appropriate physician documentation, and gathered data from only one hospital. This study sought to estimate the proportion of low- and high-acuity care provided to Medicaid and uninsured patients, and compare this with privately insured populations using a different marker for the urgency of the ED visit than previously published studies. One large study evaluating the magnitude and urgency of ED visits by different populations is the NHAMCS. This survey instrument uses a subjective, health care provider estimate at triage of how quickly patients should be seen as an indicator for urgency of ED visit (emergent ¼ \15 minutes, urgent ¼ minutes, semiurgent ¼ 1 2 hours, and nonurgent ¼ 2 24 hours). 22 Other studies have assigned urgency based on discharge diagnosis. These different and discrepant methods may account for the widely variable estimates of nonurgent ED visits ranging from 5% to 82%. 3,23 Whereas using the physician level of service provided to the patient is a new method to estimate acuity, it may be more objective than previously used methods. Using the level of service provided by the physician as a marker for ED acuity, as opposed to a single data point such as chief complaint or final diagnosis, may be a more objective means to evaluate different populations urgency (or nonurgency) for ED care. For example, infants with a discharge diagnosis of otitis media can look substantially different at triage, ranging from lethargic (from dehydration/fever) to playful. The discharge diagnosis will be the same, but the acuity of presentation can vary substantially. Using only triage assessment to assign ED urgency also has the potential for errors, because it does not take into account the actual diagnosis of the patient visit. Using the level of service as a marker for urgency may help resolve these discrepancies, because it takes into account the presenting complaint, the severity of the symptoms, the complexity of data to be reviewed, the risks of morbidity or mortality, and the patient management in the ED. 17 We arbitrarily grouped the two lowest levels of service together to define low-acuity ED care, and the two highest levels of service together to define highacuity ED care for purposes of this study. Similar analysis can be performed on only the highest and lowest levels of care. We chose not to analyze the highest and lowest levels of service, because the number of patients in these categories was small. Also, if the physician did not specifically document critical care time (even if the patient would otherwise have qualified for critical care reimbursement), it would not be categorized as such. Therefore, bundling levels and together would account for all critical care patients and other seriously ill patients. Some experts consider bundling groups 99281, 99282, and together to define nonurgent care. 24 Performing an analysis using this grouping yields no difference in the trends (Medicaid and uninsured patients still have higher proportions in this low-acuity group when compared with BCBS). Whereas our study bundled the levels of service as stated, there are many other potential ways to analyze the data. Another limitation of this study is the reliance on the reimbursement coding company for the accuracy of level of service assignment and accurate physician documentation to support the level of service assignment. The physician level of service component of the ED evaluation is coded by one company that codes approximately 200,000 charts per year. Coders are required to have training in emergency physician coding, and when they start their position, they receive an orientation period of approximately three months, during which every single chart coded is reviewed by a senior coder for one month, and then a representative sample is reviewed for the next two months. Charts coded by this company are sent twice a year to two independent, nationally recognized coding companies for audits. During the study period, there was 90% or greater agreement with other national coding companies and the company coding the charts used in this study. Whereas this study s validity is dependent on accurate reimburse-

7 ACAD EMERG MED d November 2003, Vol. 10, No. 11 d ment coding, these safeguards help ensure that there is little likelihood that significant discrepancies exist in the data based on poorly coded charts. Additionally, the possibility exists that some physicians may have been poor at documentation, and therefore all their patients charts may be lacking the appropriate documentation to accurately code the charts. This is a potential limitation, but if it existed, it should similarly affect all the groups, and would likely be minimized based on the large number of patients evaluated in this study. Additionally, at this institution, a template-driven charting system was in place for the study period to help with uniformity of documentation. Finally, this study was based on only one hospital s experience. Including other hospitals may yield more generalizable data. However, two years of data were used so that a substantial amount of patient data was available to review. CONCLUSIONS Using level of service provided by the physician as a marker for acuity of ED visit, we found that children with Medicaid insurance and uninsured children received a slightly higher proportion of low-acuity care when compared with privately insured children. There was no difference in the proportion of Medicaid adults who received low-acuity care compared with privately insured patients. We also found that uninsured patients did not receive a higher proportion of high-acuity ED care when compared with privately insured patients. Although this study identified statistically different proportions of levels of service provided to differently insured populations, the magnitude of the differences was not large. This raises questions regarding the existence of substantial disparities in ED care provided to differently insured populations. Further study in this area would be beneficial. References 1. Rubin MA, Binnin MJ. Utilization of the emergency department by patients with minor complaints. J Emerg Med. 1995; 13: Phelps K, Taylor C, Kimmel S, Nagel R, Klein W, Puczynski S. Factors associated with emergency department utilization for non-urgent pediatric problems. Arch Fam Med. 2000; 9: Gill JM. Non-urgent use of the emergency department: appropriate or not? Ann Emerg Med. 1994; 24: Sarver JH, Cydulka RK, Baker DW. Usual source of care and non-urgent emergency department use. Acad Emerg Med. 2002; 9: Cunningham PJ, Clancy CM, Cohen JW, Wilets M. The use of hospital emergency departments for non-urgent health problems: a national perspective. Med Care Res Rev. 1995; 52: Liu T, Sayre MR, Carleton SC. Emergency medical care: types, trends, and factors related to non-urgent visits. Acad Emerg Med. 1999; 6: Mitchell TA. Non-urgent emergency department visits whose definition? Ann Emerg Med. 1994; 24: Fong C. The influence of insurance status on non-urgent pediatric visits to the emergency department. Acad Emerg Med. 1999; 6: Richardson LD, Hwang U. Access to care: a review of the emergency medicine literature. Acad Emerg Med. 2001; 8: Baker LC, Baker LS. Excess cost of emergency department visits for non-urgent care. Health Aff (Millwood). 1994; 13: DeAngelis C, Fosarelli P, Duggan AK. Use of the emergency department by children enrolled in a primary care clinic. Pediatr Emerg Care. 1985; 1(2): Weissman JS, Stern R, Fielding SL, Epstein AM. Delayed access to health care: risk factors, reasons, and consequences. Ann Intern Med. 1991; 114: Baker DW, Shapiro MF, Schur CL. Health insurance and access to care for symptomatic conditions. Arch Intern Med. 2000; 160: Kellermann AL. Coverage matters: insurance and health care. Ann Emerg Med. 2002; 40: Weissman JS, Gatsonis C, Epstein AM. Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland. JAMA. 1992; 268: Peterson LA, Burstin HR, O Neil AC, Orav EJ, Brennan TA. Non-urgent emergency department visits: the effect of having a regular doctor. Med Care. 1998; 36: Clinical Procedural Terminology Chicago: American Medical Association, 2001; pp 18 21, Emergency departments an essential access point to care. Am Hosp Assoc Trend Watch. 2001; 3: Piehl MD, Clemens CJ, Joines JD. Narrowing the gap : decreasing emergency department use by children enrolled in the Medicaid program by improving access to primary care. Arch Pediatr Adolesc Med. 2000; 154: Jackson P. The impact of health insurance status on emergency room services. J Health Soc Policy. 2001; 14(1): Sox CM, Burstin HR, Edwards RA, O Neil AC, Brennan TA. Hospital admissions through the emergency department: does insurance status matter? Am J Med. 1998; 105: Advance Data from Vital and Health Statistics. April 22, 2002; 326:3, O Brien GM, Shapiro MJ, Woolard RW, O Sullivan PS, Stein MD. Inappropriate emergency department use: a comparison of three methodologies for identification. Acad Emerg Med. 1996; 3: Grossman LK, Rich LN, Johnson C. Decreasing non-urgent emergency department utilization by Medicaid children. Pediatrics. 1998; 102(1pt1):20 4.