|Home | About | Journals | Submit | Contact Us | Français|
The relationship between health care insurance and quality of medical care remains incompletely studied. We sought to determine whether type of patient insurance is related to quality of care and subsequent outcomes for patients who arrive in the emergency department (ED) for acute asthma.
Using prospectively collected data from the Multicenter Airway Research Collaboration, we compared measures of quality of pre-ED care, acute severity, and short-term outcomes across 4 insurance categories: managed care, indemnity, Medicaid, and uninsured.
Emergency departments at 57 academic medical centers enrolled 1,019 adults with acute asthma.
Patients with managed care ranked first and uninsured patients ranked last on all 7 unadjusted quality measures. After controlling for covariates, uninsured patients had significantly lower quality of care than indemnity patients for 5 of 7 measures and had lower initial peak expiratory flow rates than indemnity insured patients. Patients with managed care insurance were more likely than indemnity-insured patients to identify a primary care physician and report using inhaled steroids in the month prior to arrival in the ED. Patients with Medicaid insurance were more likely than indemnity-insured patients to use the ED as their usual source of care for problems with asthma. We found no differences in patient outcomes among the insurance categories we studied.
Uninsured patients had consistently poorer quality of care and than insured patients. Despite differences in indicators of quality of care between types of insurance, we found no differences in short-term patient outcomes by type of insurance.
The relationship between type of health insurance and quality of care is complex and incompletely studied.1 Numerous studies have documented differences in care between those with health insurance and those without health insurance.2–4 Studies of the influence of insurance type on quality of care have generally found that patients with indemnity and managed care insurance have similar quality and outcomes.5–9 Few studies have made direct comparisons among both different types of insurance and those patients without any insurance.10
Insurance organizations have focused significant attention on the problem of poor quality and increased costs in the care of patients with asthma, at least in part because of evidence of low rates of use of efficacious medications11,12 and some limited evidence that systematic efforts to improve care can result in decreased acute care utilization.13–18 While attempts to improve quality and reduce the financial costs of asthma are now commonplace in managed care,19 there is little published evidence that these programs have been effective.20 Previous research on the relationship between the financing of asthma care and quality of care found little empirical evidence and therefore a significant need for further understanding of this topic.21
The increasing prevalence of asthma programs designed to decrease emergency department (ED) visits raises concerns about the potential for these programs to inappropriately delay care, resulting in greater morbidity.9 Do managed care mechanisms designed to decrease unnecessary ED utilization influence the quality of care and short-term outcomes of patients who do end up presenting to the ED with acute asthma?
We hypothesized that adults with acute asthma and no insurance would have worse quality of care and outcomes compared to patients with insurance. This hypothesis was based on conceptual models of the influence of insurance on quality of care.1,22 We further hypothesized that if managed care interventions were effectively increasing quality, then managed care patients who did end up in the ED would 1) have better performance on measures of pre-ED quality of care, 2) be as sick or sicker than other patients by measures of acute severity, and 3) have a high rate of admission and a low relapse rate compared to patients with other types of health insurance. These hypotheses stem from the position that patients should be well-managed prior to ED presentation, should only end up in the ED when they are sick enough to require the specialized services, but once hospitalized should not have higher lengths of stay or relapse rates. To investigate our hypotheses we compared quality measures and outcomes in the Multicenter Airway Research Collaboration (MARC) database.
This study combines data from 3 prospective inception cohort studies performed during 1997 to 1998 as part of MARC.23 Using a standardized protocol for data collection, investigators at 57 EDs in 21 states enrolled asthma patients presenting to their ED 24 hours per day during a median enrollment period of 2 weeks. The EDs included large urban medical centers with over 100,000 visits/year, as well as smaller urban hospital EDs with under 25,000 visits/year. The patients were well-distributed among the participating EDs, with the largest site contributing 11% of total subjects and each site contributing on average 2% of total subjects.
All patients were managed at the discretion of the treating physician. Inclusion criteria were physician diagnosis of acute asthma, age 18 to 54 years, and ability to give informed consent. Of 1,726 eligible patients, 1,275 (74%) were enrolled in the study. We excluded 256 patients because they were residents of Canada (92), insured through the U.S. military (83), or missing insurance data (81). Institutional Review Boards at each of the participating hospitals approved the study.
The ED interview assessed patients' demographic characteristics, asthma history, and details of their current asthma exacerbation. Data on ED management and disposition were obtained by chart review. Follow-up data were collected by telephone interview 2 weeks after presentation to the ED (response rate 70%) and included details of any urgent asthma visits, changes in medical management, and current asthma symptoms. All forms were reviewed by site investigators before submission to the MARC Coordinating Center in Boston, where they underwent further review by trained personnel and then double data entry.
The insurance categories used in the analysis, (managed care, indemnity, Medicaid, or uninsured) were established by the MARC site investigators from chart review. Managed care organization was defined as prepaid plans or plans who self-identified as managed care or health maintenance organization (staff model or preferred provider organization). Indemnity insurance was defined as fee-for-service indemnity type plans. Medicaid was defined as any Medicaid insurance product even for the rare individuals with a Medicaid managed care plan.
To test our 3 hypotheses, we created 3 different sets of dependent variables: quality measures, severity measures, and short-term outcomes.
We reasoned that patients more oriented to getting acute care in their doctors' office would have a lower ratio of urgent ED visits to urgent office visits. We asked patients for the number of visits for asthma in the past year to each of these locations and then combined the responses. We used the ratio of urgent ED visits in the past year to urgent office visits in the past year to determine the likelihood that non-ED sources of care for acute asthma flares were available to the patients. The percent with a physician's office visit in the past week were used to determine the likelihood of contacts with the patient's provider for the acute episode resulting in the identifying ED visit.
The percentage of patients on inhaled corticosteroids in the past month was used to measure the likelihood that the patient's treatment was consistent with National Asthma Education Promotion Program guidelines, and has been used in previous studies of quality of care for asthma.24 Finally, patients were asked if they had a “written set of instructions for what to do when their asthma gets worse.” Having a written action plan has been shown to predict lower ED utilization for asthma.25
We created a summary measure of overall quality based on the 7 items in our measurement set.26 This summary measure was created by assigning +1 or −1 for each question depending on whether the individual had a positive (+1) or negative (−1) response. Not being on an inhaled corticosteroid was counted as a −1 only if the patient had been admitted to an ED or hospital for asthma within the previous year. Otherwise, not being on an inhaled steroid was scored as 0 (i.e., not contributing either positively or negatively to an individual's quality score). We then summed the scores for each individual.
Acute severity on presentation was assessed in several ways from data collected during chart review. The percent of predicted peak flow on presentation was the primary measure. Peak expiratory flow rate (PEFR) is expressed as percentage of patient's predicted value, based on race, age, gender, and height.27 Changes in PEFR are expressed as the absolute change in percent predicted (e.g., an improvement from 40% predicted to 70% predicted would be expressed as a change of 30%). We also looked at the duration of symptoms prior to presentation, initial respiratory rate, and initial blood oxygen saturation as determined by pulse oximetry.
Outcomes were determined from chart review (admission, hospital length of stay) or follow-up telephone interview with patients (relapse or ongoing exacerbation). Relapse was defined as any urgent visit to an ED or clinic for worsening of asthma during the 2-week follow-up period. Continued asthma symptoms (ongoing exacerbation) at follow-up was defined as “severe symptoms” during the preceding 24 hours on any one of 2 questions (i.e., asthma symptoms “most of the time” or “severe” discomfort and distress due to their asthma) or asthma that was “about the same” or worse than at the time of their ED presentation. Copies of the data collection instrument are available upon request from the corresponding author.
We controlled for several covariates in our analyses. Subject age, gender, race/ethnicity, and education level were determined by self-report. For the purpose of this analysis, mutually exclusive categories of race/ethnicity were assigned: African American (non-Hispanic), white (non-Hispanic), Hispanic, and other. Median family income was estimated using home ZIP code.
We also controlled for some clinical characteristics of enrolled patients. We used self-reported age at asthma diagnosis and a history of intubation. We controlled for additional clinical variables that might confound the relationship between insurance and outcomes. These covariates, derived from models identifying significant predictors of hospital admission,28 included recent home nebulizer use, recent use of systemic steroids, recent use of asthma medication other than β-agonist or steroids, duration of symptoms, asthma severity during past 24 hours, initial respiratory rate, initial peak flow rate, final peak flow rate, and number of inhaled β-agonist treatments over the entire ED stay.
Data were presented as proportions, means (with standard deviation [SD]), or medians (with interquartile range [IQR]). Differences between insurance groups in quality, severity, and outcomes were first tested using Student's t test, Wilcoxon rank-sum test, and χ2 test, as appropriate. Variables that were significant on univariate testing at the P = .10 level or that were judged to have potential clinical significance were examined with multivariate logistic regression models, yielding odds ratios (ORs) and 95% confidence intervals (CIs). All multivariate analyses initially controlled for differences between individuals in age, race/ethnicity, education, household income, and severity of chronic disease as determined by history of hospitalization and intubation. Adjusting quality measures for patient demographic characteristics is controversial29; nonetheless, severity of illness and risk factors for asthma have been associated with socioeconomic status,30 and such factors are not distributed evenly between different types of insurance.31–33 We therefore presented our quality measures both with and without adjustment for race/ethnicity, education and estimated household income.
We examined whether having a primary care physician, use of the office for acute visits rather than the ED (access measures), and being on an inhaled corticosteroid prior to arrival in the ED (management measure) may have explained differences in outcomes between patients with different insurance by entering these variables into the final multivariate models. If these variables were significant independent predictors of outcomes and reduced the significance of the insurance group comparison, we interpreted this as indicating that differences between insurance groups were attributable, at least in part, to the differences in quality.
The possibility of period and site effects was examined by adjusting for period and site of enrollment in our models. Results presented are from multivariate models using generalized estimating equations to adjust for clustering of data at the level of the emergency departments. Odds ratios were converted to adjusted relative risks. All tests were 2-tailed and P < .05 was considered statistically significant.
Of the 1,019 patients included in the current analysis, 15% were insured through a managed care plan, 12% had an indemnity plan, 34% had Medicaid, and 39% were uninsured. Patients presenting to the ED with acute asthma were on average 36 years old and more likely to be female, high school graduates, and from a minority ethnic background. Our study population came from ZIP codes with a mean household income of approximately $28,000. With respect to their clinical history, subjects reported diagnosis of asthma during adolescence and most (66%) had been hospitalized at some point due to their asthma.
The uninsured subjects were more often younger black men compared to the insured groups (Table 1). Compared to the subjects with managed care and indemnity insurance, Medicaid subjects were more likely to be Hispanic, less likely to have completed high school, and had lower estimated median incomes. Managed care patients were least likely (23%) and Medicaid patients most likely (38%) to have been hospitalized within the past year for asthma.
Patients with managed care insurance were the most likely and uninsured patients the least likely to have an identified primary care physician. Managed care patients were the least likely and uninsured patients the most likely to report using the ED as their usual site of asthma care or their usual source of asthma prescriptions. Managed care patients appeared to be using their physician's office for the treatment of acute asthma more than patients with other types of insurance. This was reflected in the lower proportion of managed care patients who had a ratio <1 of ED visits to urgent office visits. Patients with managed care insurance were most likely to have taken an inhaled corticosteroid in the 4 weeks prior to arrival in the ED, whereas patients without insurance were least likely. The proportion of patients (30% to 40%) with a written asthma action plan did not differ significantly by insurance status. The summary quality score differed across insurance categories, with managed care patients having the highest score and uninsured patients having the lowest scores (P = .001).
Multivariate testing (Table 3) produced adjusted relative risks consistent with the univariate findings. When we removed patient demographics from the multivariate models, we found several additional differences in quality between insurance categories (Table 3). Uninsured patients were more likely than indemnity patients to use the ED than a physician's office for acute exacerbations of asthma, less likely to have visited their primary care physician in the week prior to being seen in the ED, and less likely to be taking inhaled corticosteroids. Medicaid patients were more likely than indemnity patients to use the ED as their usual source of prescriptions for asthma and were more likely to use the ED than a physician's office for acute exacerbations of asthma.
The adjusted summary quality score indicated that managed care patients had a higher overall pre-ED quality of care than indemnity patients (P = .03), although as noted above, the differences between managed care and indemnity on most measures were small. The 2 measures contributing most to the overall increased score for managed care included the increased likelihood of having a primary care doctor and the increased likelihood of having used inhaled corticosteroids in the 4 weeks prior to visiting the ED.
Acute severity as measured by the percentage of predicted PEFR was nearly identical for patients with any type of insurance (mean = 49% predicted), but uninsured patients had a lower PEFR (Table 2) that remained significantly different from indemnity-insured patients on multivariate testing (P = .01) (data not shown).
Short-term outcomes were not equivalent across insurance categories on univariate analysis (Table 2). Managed care patients were the most likely to be admitted and among the least likely to have a relapse leading to hospital admission or an ongoing exacerbation during the 2 weeks after ED discharge. After adjusting for the principle predictors of these outcomes (see METHODS) on multivariate testing, outcome differences between insured groups were no longer significantly different (Table 4). There were trends for lower rates of relapse and ongoing exacerbations in managed care and lower admission rates among the uninsured. The short-term outcomes did not change when patient demographic characteristics were removed from the multivariate models (data not shown).
We investigated the extent to which some of our quality indicators were related to outcomes in the multivariate models. While having an identified primary care physician was not a significant predictor of the 3 primary outcomes used in this study, being on an inhaled corticosteroid prior to presentation in the ED was associated with a lower likelihood of an ongoing exacerbation at 4 weeks after discharge (OR, 0.68; 95% CI, 0.45 to 1.02). Importantly, the inclusion of this quality indicator in the multivariate models decreased differences between insurance categories, indicating that this variable may mediate differences between insurance groups.
This multicenter investigation found differences in the pre-ED quality of care by type of insurance for adults with acute asthma. Overall, we found considerable opportunity for improvement in all insurance categories. Uninsured patients in our study consistently had the lowest quality of care. Although managed care patients appeared to have better overall performance on pre-ED quality of care measures, differences with other insured groups were generally small. We did not find differences in short-term outcomes between patients with different types of insurance.
Adults without health insurance presenting to an ED with acute asthma were more likely to be young black men. The persistent high rate of uninsurance among young men generally, and young minority men in particular, continues to be a problem in the United States.34 We found the uninsured population with asthma who visited the ED had poorer pre-ED quality of care by several indicators compared to those patients with insurance of any type. Specifically, our findings of lower likelihood of having a primary care doctor and increased likelihood of using the emergency department for problems and prescriptions were consistent with previous studies showing lower rates of access to primary care for uninsured patients compared to insured patients.35 The lower likelihood of uninsured patients with acute asthma having an inhaled corticosteroid may be due to inability to pay for medication, failure to have it prescribed, or both. Alternatively, uninsured patients may be less likely to have an indication for a controller medication, although this seems unlikely, given the comparability of chronic asthma characteristics between the uninsured and insured patients in our study population.
Patients without insurance may have a decreased likelihood of being admitted36 and our results support this assertion. The lower likelihood of admission is of particular concern, given our finding that uninsured adults were more acutely ill than patients with insurance, based on their initial PEFR. Despite the increased severity on presentation and decreased likelihood of admission, uninsured patients in our study were not more likely than insured patients to have an ongoing asthma exacerbation 2 weeks after their ED visit.
Overall, differences in quality between insurance types were small, and all measures indicated significant room for improvement. The increased performance of managed care compared to indemnity was mediated primarily by 2 of the 7 quality measures used in our study. Patients in managed care were more likely to have a primary care physician and more likely to be taking inhaled corticosteroids. Having an assigned primary care physician has been a hallmark of managed care. Although previous studies have suggested that the authorization function of the primary care physician may be of limited value,37,38 other studies have suggested that the coordination and continuity functions of an assigned primary care physician may benefit patients.39–42 Insurance coverage for medications may have contributed to our finding that managed care patients visiting an emergency department with acute asthma were more likely to be taking an inhaled corticosteroid. In addition, systems designed to identify and intervene with patients in need of controller medications (e.g., physician reminders or facilitated referrals to specialists43) may have contributed to the increased rates of inhaled corticosteroid use among the managed care patients.
As in previous studies,35,44 patients in our study with Medicaid insurance had overall quality of care similar to that of patients with indemnity insurance. Nonetheless, the patients with Medicaid insurance were more likely than indemnity-insured patients to use the ED for problems with asthma and medications. Although our data do not provide insight into the reasons for preferential use of the ED for acute asthma care among Medicaid recipients, several hypotheses are apparent including: 1) less available primary care,45 2) patient preferences, and 3) patient understanding of acute care resources. As with uninsured patients, the significance of some differences in quality measures was confounded by patient demographic characteristics, evidenced by the change in statistical significance when demographic characteristics were removed from models. Previous work has found patient demographic characteristics to be an important predictor of both quality of care in asthma24 and rates of appropriate hospitalization.46 One interpretation of our results of quality measures for Medicaid patients is that both the patient demographic characteristics of Medicaid patients and additional factors, possibly related to access to primary care, contributed to the propensity of Medicaid patients to seek care in the emergency room.
Consistent with previous studies,47 we found no significant differences in severity of acute presentation between the different groups of insured patients. This finding suggests that the management techniques designed to reduce inappropriate ED use for acute asthma used by the managed care organizations in this study do not appear to be creating a problem with increased morbidity. This interpretation was supported by a lack of difference in length of hospital stay. Previous work has shown that hospital length of stay can increase when care is inappropriately delayed.48 We were not able to determine if specific individuals were adversely affected by their type of insurance.
Short-term outcomes did not differ significantly between the insurance categories in our study. The important predictors of admission, length of stay, relapse, and ongoing exacerbation that we controlled for in our multivariate analyses were highly significant and overall had greater explanatory power than the patient's type of insurance. One limitation of this finding is the relatively low power of our study to detect differences in the less common outcomes. It is nonetheless interesting that our measure of quality management, being on an inhaled steroid prior to ED presentation, predicted a lower likelihood of having ongoing problems in short-term follow-up.
This study has several potential limitations. The EDs participating in MARC were not randomly chosen, and therefore our sample of patients may not be representative of all patients presenting with acute asthma to U.S. EDs. Nonetheless, the concentration of urban academic EDs resulted in high rates of Medicaid and uninsured patients as well as urban-dwelling minorities, precisely the group at greatest risk for morbidity and mortality from asthma. We did not collect detailed information on patient-physician contact (e.g., phone calls) prior to presentation to the ED. This may have resulted in an underestimate of amount of contact between patients and physicians prior to presentation to the ED, but was not likely to have introduced bias to our analysis. Our study also was limited to patients presenting to the ED, so we were unable to assess those who did not present to the ED, and cannot generalize our findings to the overall population of patients with asthma. Although our method of assigning insurance may be subject to misclassification, it was likely to have been more accurate than patient report.49 Our finding of improved performance on quality measures for patients in managed care may have underestimated differences between “managed” and “unmanaged” patients because differences in intensity of management within managed care may be greater than differences between managed care and indemnity insurance.
In summary, this study adds to the growing evidence indicating systematic deficiencies in the quality of care for patients with asthma. We have identified and described part of the relationship between the financing of asthma care and the quality of care received by adults presenting to the ED with asthma. Uninsured patients had lower quality of care than did insured patients. Higher quality of care for patients with managed care insurance was mediated largely by having an assigned primary care physician and having an inhaled corticosteroid in the period prior to arrival in the ED.
Dr. Ferris was supported by the Pediatric Scientist Development Program (AAP/HD-00850), and Dr. Camargo by Grant HL-03533 from the National Institutes of Health (Bethesda, Md). The Multicenter Airway Research Collaboration is supported by Grant HL-63253 from the National Institutes of Health, and by unrestricted grants from Glaxo Wellcome, Inc. (Research Triangle Park, NC) and Monaghan Medical Corporation (Syracuse, NY).