The Journal of Emergency Medicine, Vol. 49, No. 3, pp. 347–354, 2015 Copyright Ó 2015 Elsevier Inc. Printed in the USA. All rights reserved 0736-4679/$ - see front matter
http://dx.doi.org/10.1016/j.jemermed.2015.01.034
Administration of Emergency Medicine
CAN NONURGENT EMERGENCY DEPARTMENT CARE COSTS BE REDUCED? EMPIRICAL EVIDENCE FROM A U.S. NATIONALLY REPRESENTATIVE SAMPLE Haichang Xin, PHD, Meredith L. Kilgore, PHD, Bisakha (Pia) Sen, PHD, and Justin Blackburn, PHD Department of Health Care Organization and Policy, School of Public Health, University of Alabama at Birmingham, Birmingham, Alabama Reprint Address: Haichang Xin, PHD, Department of Health Care Organization and Policy, School of Public Health, University of Alabama at Birmingham, 1665 University Boulevard, RPHB 427, Birmingham, AL 35294-0022
, Keywords—ambulatory care; nonurgent emergency department care costs; nationwide
, Abstract—Background: A well-functioning primary care system has the capacity to provide effective care for patients to avoid nonurgent emergency department (ED) use and related costs. Objective: This study examined how patients’ perceived deficiency in ambulatory care is associated with nonurgent ED care costs nationwide. Methods: This retrospective cohort study used data from the 2010–2011 Medical Expenditure Panel Survey. This study chose usual source of care, convenience of needed medical care, and patient evaluation of care quality as the main independent variables. The marginal effect following a multivariate logit model was employed to analyze the urgent vs. nonurgent ED care costs in 2011, after controlling for covariates in 2010. The endogeneity was accounted for by the time lag effect and controlling for education levels. Sample weights and variance were adjusted with the survey procedures to make results nationally representative. Results: Patient-perceived poor and intermediate levels of primary care quality had higher odds of nonurgent ED care costs (odds ratio [OR] = 2.22, p = 0.035, and OR = 2.05, p = 0.011, respectively) compared to high-quality care, with a marginal effect (at means) of 13.0% and 11.5% higher predicted probability of nonurgent ED care costs. Costs related to these ambulatory care quality deficiencies amounted to $229 million for private plans (95% confidence interval [CI] $100 million– $358 million), $58.5 million for public plans (95% CI $33.9 million–$83.1 million), and an overall of $379 million (95% CI $229 million–$529 million) nationally. Conclusions: These findings highlight the improvement in ambulatory care quality as the potential target area to effectively reduce nonurgent ED care costs. Ó 2015 Elsevier Inc.
INTRODUCTION From 1997 through 2007, emergency department (ED) visits in the United States increased by 23%, to a total of nearly 117 million visits per year (1). Among them, nearly half of these visits were for nonurgent medical care or were potentially preventable, leading to billions of dollars in potentially avoidable spending annually (2–7). It is estimated that nonurgent ED care can be $450 to $650 more expensive than care received in a physicians’ office. Urgent conditions that could be treated in physicians’ offices have $600 to $900 higher costs per ED visit than a physician visit (3). In the same report, a large insurer estimated that reducing these two types of ED visits by 5% would save between $6 million and $9 million, and a 25% reduction would save between $29 million and $43 million (3). Stakeholders including health systems, physicians, and payers have devised various interventions to discourage nonurgent ED visits, such as patient education, financial disincentives, encouragement of primary care physician (PCP) services on evenings and weekends, and an increase of PCP supply. Despite these efforts, nonurgent ED visits have continued to rise, warranting further examination of the underlying reason (8).
RECEIVED: 19 August 2014; FINAL SUBMISSION RECEIVED: 29 December 2014; ACCEPTED: 5 January 2015 347
348
H. Xin et al.
Persistent nonurgent ED use may represent a deficiency in the primary care or ambulatory care system. A well-functioning primary care system has the capacity to provide timely, adequate, and effective care for patients to avoid nonurgent ED use. One study found that higher levels of primary care capacity are associated with lower rates of ED utilization (9). Although that study examined Medicaid beneficiaries, its finding may apply to other populations as well. This current study examined deficiencies in primary care systems that are associated with nonurgent ED care costs. Donabedian’s structure-process-outcomes (SPO) model that measures health care quality and capacity served as the conceptual framework for this analysis. As guidance, the SPO model systematically identified and summarized ambulatory care system components that are associated with downstream ED use. Essentially, three domains of measures were identified, which reflect structure, process, and outcomes in ambulatory care, respectively. They are usual source of care, convenience of needed medical care, and patient evaluation of care quality. The first two domains of structure and process measures are within the health care access category. Based on this classification, this study specifically examined how access to and quality of ambulatory care is associated with nonurgent ED care costs, and to what extent these costs can be reduced if deficiencies in primary care systems could be improved. Precise identification of specific deficiencies in primary care systems will reveal the underlying reasons for nonurgent ED care costs. These deficiencies’ attributable cost magnitude will inform health policies on how improvements in specific areas of primary care systems can contribute to cost reduction for nonurgent ED care for insurance plans. To date, no studies have examined the association between deficiencies in primary care systems and nonurgent ED care costs reduction nationwide. This current study is the first to examine empirical evidence using a nationally representative sample, and the latest (2010–2011) that captures the most recent reforms and initiatives in ambulatory care systems and the population’s current ED utilization patterns. METHODS Data We used Medical Expenditure Panel Survey (MEPS) data for this study (10). The MEPS is a nationally representative survey of the noninstitutionalized civilian population of the United States and is designed to produce national estimates on the health care use, costs, sources of payment, and insurance coverage of these individuals.
Each new panel entails a series of five rounds of inperson interviews (11). This design, which covers two full calendar years from 2010 through 2011 used in this study, allows for tracking individuals’ preferences, health care utilization, and costs over time. Like many other national surveys, MEPS adopts a complex multistage, unequal probability, and cluster sampling study design (12). Because Hispanics, African Americans, Asians, and indigent populations have been oversampled to increase statistical power and improve the precision of estimates for specific subgroups, sample weights have been provided to calculate population estimates. Study Population This study used a retrospective cohort design. Individuals were included in this analysis if they were 18 years and older, had any ED visit in 2011, and had data from all five survey rounds. Outcome Variables Outcome measure was urgent vs. nonurgent ED care costs in 2011. The cost group status, instead of the cost amount itself, was of interest. The cost group status was adapted from the literature, rather than derived from cost values in these data. A study by Sarver et al. defined and measured nonurgent ED use (13). Specifically, a visit was considered to be urgent . if 1) it resulted in an admission; 2) the patient received an x-ray, magnetic resonance imaging, electrocardiography, electroencephalography, or any surgical procedure, and the patient reported the reason for the visit was an ‘accident or injury,’ diagnosis, or treatment, and, if it was an office or clinic visit, the visit was not the result of referral; or 3) the reason for the visit was an ‘accident or injury,’ diagnosis, or treatment, and the visit was within 3 days of the ‘accident or injury’ or onset of symptoms. The remaining visits were classified as nonurgent. (13)
We adopted the same approach to classify ED cost types, but made a minor revision by deleting the component ‘‘if it was an office or clinic visit, the visit was not the result of referral’’ within the second criterion to improve the construct validity of nonurgent ED use and its costs. This revision was made because this study focused on the nonurgent or urgent health care costs within the ED setting, instead of an ambulatory care setting. Independent Variables Based on the SPO model, three domains of measures were identified to reflect structure, process, and outcomes in ambulatory care, respectively: usual source of care, perceived convenience of needed medical care, and
Nonurgent Emergency Department Care, Costs Reduction
349
patient evaluation of care quality. Rather than identify multiple measures within each domain, a single comprehensive measure can be identified to fully represent each SPO domain. Specifically, the usual source of care variable that measured whether a patient had a regular source of care represented the structure domain. The perceived convenience variable that asked patients how difficult it was to get medical care when needed represented the process domain. Four dummy variables were created to reflect the four categories from the initial variable: never, sometimes, usually, and always. The first category was treated as the reference group. This single measure reflects all the components in this domain, for example, how difficult it is for patients to get an appointment at a clinic, patients’ waiting time on appointments with PCPs, the availability of physicians on weekends, and so on. The care quality variable in MEPS was introduced from the Consumer Assessment of Healthcare Providers and Systems (CAHPS), which was the patient rating of health care quality with a range from 0 to 10, with 10 being the highest quality. This single variable reflects timing, adequacy, effectiveness, and satisfaction in the outcomes domain. Based on this initial rating, we further created three dummy variables to indicate high, intermediate, and low levels of quality. The high-quality category contained ratings of 9 and 10, which is consistent with CAHPS criteria, whereas the low-quality category contained the ratings from 0 to 5. Ratings of 6 to 8 were combined to reflect an intermediate level of care quality. Each independent variable can sufficiently represent an overall domain of the SPO model without any omission or overlap. These independent variables were measured at the baseline (2010). Ambulatory care includes both physicians’ office visits and hospital outpatient care.
gland Journal of Medicine article indicated that the marginal cost for additional ED patients would produce economies of scale and reduce the average cost of ED visits (14). The ED’s significant fixed costs were covered by the additional revenue associated with nonurgent visits, therefore, the financial gain justified the social welfare of EDs by allowing them to be open and available at all times to consumers. To clarify this controversy, we generated three categories for the cost comparison variable: the more expensive ED group, the equal cost group, and the less expensive ED group. The more expensive ED group was the reference group. All predictors were treated as dummy variables because their effects on ED use may not vary in a linear way.
Control Variables Our analysis controlled for a range of factors in the baseline year 2010 known or likely to affect urgent vs. nonurgent ED care costs, including a person’s age, gender, race/ ethnicity, rural/urban location, and insurance status. Age was classified into three groups: 18–34 years, 35–64 years, and 65 years and older, with the youngest group as the reference. Insurance status was also classified into three groups: private, public, and uninsured, with the private group as the reference. Besides these covariates, we also created a variable for cost comparison between ED and ambulatory care in 2010 to shed light on the controversy between traditional belief and literature findings, because the traditional belief that ED use is more expensive than ambulatory care has been challenged by the literature. A widely cited 1996 New En-
Statistical Analysis There is a potential selection issue between nonurgent ED care costs and access to and quality of ambulatory care because conscientious patients may intentionally have low levels of nonurgent ED care costs and be more likely to use ambulatory care. Differences in individuals’ urgent and nonurgent ED care costs could reflect the combination of a causal effect of the access to and quality of ambulatory care and the effect of unmeasured characteristics that are correlated with primary care system capacity and urgency in ED care costs. Thus, the estimate for the impact of access to and quality of ambulatory care on urgent and nonurgent ED care costs could be biased. If this selection issue is not accounted for, it is likely to bias the estimated relationship between the primary care system capacity and the outcomes of interest. We attempted to attenuate this problem by including controls for education level and using the time lag effect in our regression model. Because conscientious and wellinformed patients with low level of nonurgent ED care costs may be more likely to have a usual source of care and more interaction with the ambulatory care system, we included education to account for the potential selection issue. The retrospective cohort study design was also chosen to reflect the time lag effect that accounted for the selection issue, because later urgent or nonurgent ED care costs in 2011 could not affect ambulatory care access and quality in an earlier period of 2010. We used logit models to analyze urgent and nonurgent ED care costs. The logit model is analyzed using maximum likelihood estimation. Among significant independent variables, marginal effects were further analyzed to indicate the change in magnitude of predicted probability, in addition to the odds ratio (OR), between different categories of the significant predictors, holding all other covariates at the mean level. For the cost reduction analysis, only significant independent variables were further examined, which reflected
350
H. Xin et al.
deficiencies in primary care systems. Specifically, compared to higher levels of care access and quality, lower levels were analyzed for their corresponding nonurgent ED care costs, overall and by plan types. Based on the identified reasons, we calculated the nonurgent ED care cost magnitude that could be reduced if deficiencies in primary care systems are improved nationwide. Typically, individuals within the survey’s sampling cluster (i.e., a county or a family) are more similar to one another than those in other clusters; as a result, the error term in urgent vs. nonurgent ED care costs for individuals within a cluster is correlated. Failure to consider this correlation at the cluster level likely results in an underestimation of variance, which will overestimate the significance of the estimates. Thus, survey procedures in STATA (version 12; StataCorp LP, College Station, TX) were used to conduct the analysis. They account for the complex weight and variance in the sampling design and yield nationally-representative results. RESULTS
Table 1. Study Sample Characteristics (N = 1287) Variable Age (years) 18–34 35–64 $65 Gender Male Female Race Non-Hispanic white Non-Hispanic African-American Hispanic Other Rural/urban location Urban Rural Education level (years) 0–8 9–12 13–17 Don’t know or refused Insurance coverage Private Public Uninsured
Number
%
383 635 269
29.8 49.3 20.9
495 792
38.5 61.5
624 323 259 81
48.5 25.1 20.1 6.3
1073 214
83.4 16.6
115 653 512 7
8.9 50.7 39.8 0.6
611 443 233
47.5 34.4 18.1
The Study Sample Characteristics The demographic and socioeconomic characteristics of the sample used in the analysis are presented in Table 1. The final study sample consisted of 1287 adults with any ED visit in 2011, which represented a weighted 29,463,684 people in the total population. Among them, 390 individuals (30.3%) had nonurgent ED use, representing a weighted 8.3 million people in the total population. The following groups made up the highest proportion of the study sample: middle-aged individuals (49.3%), females (61.5%), non-Hispanic Whites (48.5%), private enrollees (47.5%), those living in the urban area (83.4%), and with high school education (50.7%). The Multivariate Analysis Results The multivariate analysis results are presented in Table 2. In the multivariate logit model, usual source of care and perceived convenience of needed medical care were not associated with nonurgent ED care costs (p = 0.497 for the usual source of care status, p = 0.293, p = 0.240, and p = 0.472 for the set of convenience dummy variables). Patient-perceived poor and intermediate levels of primary care quality had higher odds of nonurgent ED care costs (OR = 2.22, p = 0.035, and OR 2.05, p = 0.011, respectively) compared to high-quality care, with a marginal effect (at means) of 13.0% (95% confidence interval [CI] 0.006, 0.27) and 11.5% (95% CI 0.03–0.20) higher predicted probability of nonurgent ED care costs. In
addition, white and black individuals (compared to other races) older than 65 years (compared to individuals aged between 18 and 34 years) also had lower odds of nonurgent ED care costs (OR = 0.28, p = 0.002, OR = 0.34, p = 0.018, and OR = 0.40, p = 0.011, respectively). Patients with public insurance (compared to private coverage) living in urban areas (compared to rural areas) had higher odds of nonurgent ED care costs (OR = 1.72, p = 0.031, and OR = 1.81, p = 0.041, respectively). The cost comparison between ED and ambulatory care was not associated with higher or lower odds of nonurgent ED care costs (p = 0.767 for the equal cost group, and p = 0.705 for the cheaper ED group, compared to the more expensive ED group). Similarly, education level was not associated with higher or lower odds of nonurgent ED care costs (p = 0.712 for the high school group, and p = 0.718 for the college group, compared to the group with
http://dx.doi.org/10.1016/j.jemermed.2015.01.034
Administration of Emergency Medicine
CAN NONURGENT EMERGENCY DEPARTMENT CARE COSTS BE REDUCED? EMPIRICAL EVIDENCE FROM A U.S. NATIONALLY REPRESENTATIVE SAMPLE Haichang Xin, PHD, Meredith L. Kilgore, PHD, Bisakha (Pia) Sen, PHD, and Justin Blackburn, PHD Department of Health Care Organization and Policy, School of Public Health, University of Alabama at Birmingham, Birmingham, Alabama Reprint Address: Haichang Xin, PHD, Department of Health Care Organization and Policy, School of Public Health, University of Alabama at Birmingham, 1665 University Boulevard, RPHB 427, Birmingham, AL 35294-0022
, Keywords—ambulatory care; nonurgent emergency department care costs; nationwide
, Abstract—Background: A well-functioning primary care system has the capacity to provide effective care for patients to avoid nonurgent emergency department (ED) use and related costs. Objective: This study examined how patients’ perceived deficiency in ambulatory care is associated with nonurgent ED care costs nationwide. Methods: This retrospective cohort study used data from the 2010–2011 Medical Expenditure Panel Survey. This study chose usual source of care, convenience of needed medical care, and patient evaluation of care quality as the main independent variables. The marginal effect following a multivariate logit model was employed to analyze the urgent vs. nonurgent ED care costs in 2011, after controlling for covariates in 2010. The endogeneity was accounted for by the time lag effect and controlling for education levels. Sample weights and variance were adjusted with the survey procedures to make results nationally representative. Results: Patient-perceived poor and intermediate levels of primary care quality had higher odds of nonurgent ED care costs (odds ratio [OR] = 2.22, p = 0.035, and OR = 2.05, p = 0.011, respectively) compared to high-quality care, with a marginal effect (at means) of 13.0% and 11.5% higher predicted probability of nonurgent ED care costs. Costs related to these ambulatory care quality deficiencies amounted to $229 million for private plans (95% confidence interval [CI] $100 million– $358 million), $58.5 million for public plans (95% CI $33.9 million–$83.1 million), and an overall of $379 million (95% CI $229 million–$529 million) nationally. Conclusions: These findings highlight the improvement in ambulatory care quality as the potential target area to effectively reduce nonurgent ED care costs. Ó 2015 Elsevier Inc.
INTRODUCTION From 1997 through 2007, emergency department (ED) visits in the United States increased by 23%, to a total of nearly 117 million visits per year (1). Among them, nearly half of these visits were for nonurgent medical care or were potentially preventable, leading to billions of dollars in potentially avoidable spending annually (2–7). It is estimated that nonurgent ED care can be $450 to $650 more expensive than care received in a physicians’ office. Urgent conditions that could be treated in physicians’ offices have $600 to $900 higher costs per ED visit than a physician visit (3). In the same report, a large insurer estimated that reducing these two types of ED visits by 5% would save between $6 million and $9 million, and a 25% reduction would save between $29 million and $43 million (3). Stakeholders including health systems, physicians, and payers have devised various interventions to discourage nonurgent ED visits, such as patient education, financial disincentives, encouragement of primary care physician (PCP) services on evenings and weekends, and an increase of PCP supply. Despite these efforts, nonurgent ED visits have continued to rise, warranting further examination of the underlying reason (8).
RECEIVED: 19 August 2014; FINAL SUBMISSION RECEIVED: 29 December 2014; ACCEPTED: 5 January 2015 347
348
H. Xin et al.
Persistent nonurgent ED use may represent a deficiency in the primary care or ambulatory care system. A well-functioning primary care system has the capacity to provide timely, adequate, and effective care for patients to avoid nonurgent ED use. One study found that higher levels of primary care capacity are associated with lower rates of ED utilization (9). Although that study examined Medicaid beneficiaries, its finding may apply to other populations as well. This current study examined deficiencies in primary care systems that are associated with nonurgent ED care costs. Donabedian’s structure-process-outcomes (SPO) model that measures health care quality and capacity served as the conceptual framework for this analysis. As guidance, the SPO model systematically identified and summarized ambulatory care system components that are associated with downstream ED use. Essentially, three domains of measures were identified, which reflect structure, process, and outcomes in ambulatory care, respectively. They are usual source of care, convenience of needed medical care, and patient evaluation of care quality. The first two domains of structure and process measures are within the health care access category. Based on this classification, this study specifically examined how access to and quality of ambulatory care is associated with nonurgent ED care costs, and to what extent these costs can be reduced if deficiencies in primary care systems could be improved. Precise identification of specific deficiencies in primary care systems will reveal the underlying reasons for nonurgent ED care costs. These deficiencies’ attributable cost magnitude will inform health policies on how improvements in specific areas of primary care systems can contribute to cost reduction for nonurgent ED care for insurance plans. To date, no studies have examined the association between deficiencies in primary care systems and nonurgent ED care costs reduction nationwide. This current study is the first to examine empirical evidence using a nationally representative sample, and the latest (2010–2011) that captures the most recent reforms and initiatives in ambulatory care systems and the population’s current ED utilization patterns. METHODS Data We used Medical Expenditure Panel Survey (MEPS) data for this study (10). The MEPS is a nationally representative survey of the noninstitutionalized civilian population of the United States and is designed to produce national estimates on the health care use, costs, sources of payment, and insurance coverage of these individuals.
Each new panel entails a series of five rounds of inperson interviews (11). This design, which covers two full calendar years from 2010 through 2011 used in this study, allows for tracking individuals’ preferences, health care utilization, and costs over time. Like many other national surveys, MEPS adopts a complex multistage, unequal probability, and cluster sampling study design (12). Because Hispanics, African Americans, Asians, and indigent populations have been oversampled to increase statistical power and improve the precision of estimates for specific subgroups, sample weights have been provided to calculate population estimates. Study Population This study used a retrospective cohort design. Individuals were included in this analysis if they were 18 years and older, had any ED visit in 2011, and had data from all five survey rounds. Outcome Variables Outcome measure was urgent vs. nonurgent ED care costs in 2011. The cost group status, instead of the cost amount itself, was of interest. The cost group status was adapted from the literature, rather than derived from cost values in these data. A study by Sarver et al. defined and measured nonurgent ED use (13). Specifically, a visit was considered to be urgent . if 1) it resulted in an admission; 2) the patient received an x-ray, magnetic resonance imaging, electrocardiography, electroencephalography, or any surgical procedure, and the patient reported the reason for the visit was an ‘accident or injury,’ diagnosis, or treatment, and, if it was an office or clinic visit, the visit was not the result of referral; or 3) the reason for the visit was an ‘accident or injury,’ diagnosis, or treatment, and the visit was within 3 days of the ‘accident or injury’ or onset of symptoms. The remaining visits were classified as nonurgent. (13)
We adopted the same approach to classify ED cost types, but made a minor revision by deleting the component ‘‘if it was an office or clinic visit, the visit was not the result of referral’’ within the second criterion to improve the construct validity of nonurgent ED use and its costs. This revision was made because this study focused on the nonurgent or urgent health care costs within the ED setting, instead of an ambulatory care setting. Independent Variables Based on the SPO model, three domains of measures were identified to reflect structure, process, and outcomes in ambulatory care, respectively: usual source of care, perceived convenience of needed medical care, and
Nonurgent Emergency Department Care, Costs Reduction
349
patient evaluation of care quality. Rather than identify multiple measures within each domain, a single comprehensive measure can be identified to fully represent each SPO domain. Specifically, the usual source of care variable that measured whether a patient had a regular source of care represented the structure domain. The perceived convenience variable that asked patients how difficult it was to get medical care when needed represented the process domain. Four dummy variables were created to reflect the four categories from the initial variable: never, sometimes, usually, and always. The first category was treated as the reference group. This single measure reflects all the components in this domain, for example, how difficult it is for patients to get an appointment at a clinic, patients’ waiting time on appointments with PCPs, the availability of physicians on weekends, and so on. The care quality variable in MEPS was introduced from the Consumer Assessment of Healthcare Providers and Systems (CAHPS), which was the patient rating of health care quality with a range from 0 to 10, with 10 being the highest quality. This single variable reflects timing, adequacy, effectiveness, and satisfaction in the outcomes domain. Based on this initial rating, we further created three dummy variables to indicate high, intermediate, and low levels of quality. The high-quality category contained ratings of 9 and 10, which is consistent with CAHPS criteria, whereas the low-quality category contained the ratings from 0 to 5. Ratings of 6 to 8 were combined to reflect an intermediate level of care quality. Each independent variable can sufficiently represent an overall domain of the SPO model without any omission or overlap. These independent variables were measured at the baseline (2010). Ambulatory care includes both physicians’ office visits and hospital outpatient care.
gland Journal of Medicine article indicated that the marginal cost for additional ED patients would produce economies of scale and reduce the average cost of ED visits (14). The ED’s significant fixed costs were covered by the additional revenue associated with nonurgent visits, therefore, the financial gain justified the social welfare of EDs by allowing them to be open and available at all times to consumers. To clarify this controversy, we generated three categories for the cost comparison variable: the more expensive ED group, the equal cost group, and the less expensive ED group. The more expensive ED group was the reference group. All predictors were treated as dummy variables because their effects on ED use may not vary in a linear way.
Control Variables Our analysis controlled for a range of factors in the baseline year 2010 known or likely to affect urgent vs. nonurgent ED care costs, including a person’s age, gender, race/ ethnicity, rural/urban location, and insurance status. Age was classified into three groups: 18–34 years, 35–64 years, and 65 years and older, with the youngest group as the reference. Insurance status was also classified into three groups: private, public, and uninsured, with the private group as the reference. Besides these covariates, we also created a variable for cost comparison between ED and ambulatory care in 2010 to shed light on the controversy between traditional belief and literature findings, because the traditional belief that ED use is more expensive than ambulatory care has been challenged by the literature. A widely cited 1996 New En-
Statistical Analysis There is a potential selection issue between nonurgent ED care costs and access to and quality of ambulatory care because conscientious patients may intentionally have low levels of nonurgent ED care costs and be more likely to use ambulatory care. Differences in individuals’ urgent and nonurgent ED care costs could reflect the combination of a causal effect of the access to and quality of ambulatory care and the effect of unmeasured characteristics that are correlated with primary care system capacity and urgency in ED care costs. Thus, the estimate for the impact of access to and quality of ambulatory care on urgent and nonurgent ED care costs could be biased. If this selection issue is not accounted for, it is likely to bias the estimated relationship between the primary care system capacity and the outcomes of interest. We attempted to attenuate this problem by including controls for education level and using the time lag effect in our regression model. Because conscientious and wellinformed patients with low level of nonurgent ED care costs may be more likely to have a usual source of care and more interaction with the ambulatory care system, we included education to account for the potential selection issue. The retrospective cohort study design was also chosen to reflect the time lag effect that accounted for the selection issue, because later urgent or nonurgent ED care costs in 2011 could not affect ambulatory care access and quality in an earlier period of 2010. We used logit models to analyze urgent and nonurgent ED care costs. The logit model is analyzed using maximum likelihood estimation. Among significant independent variables, marginal effects were further analyzed to indicate the change in magnitude of predicted probability, in addition to the odds ratio (OR), between different categories of the significant predictors, holding all other covariates at the mean level. For the cost reduction analysis, only significant independent variables were further examined, which reflected
350
H. Xin et al.
deficiencies in primary care systems. Specifically, compared to higher levels of care access and quality, lower levels were analyzed for their corresponding nonurgent ED care costs, overall and by plan types. Based on the identified reasons, we calculated the nonurgent ED care cost magnitude that could be reduced if deficiencies in primary care systems are improved nationwide. Typically, individuals within the survey’s sampling cluster (i.e., a county or a family) are more similar to one another than those in other clusters; as a result, the error term in urgent vs. nonurgent ED care costs for individuals within a cluster is correlated. Failure to consider this correlation at the cluster level likely results in an underestimation of variance, which will overestimate the significance of the estimates. Thus, survey procedures in STATA (version 12; StataCorp LP, College Station, TX) were used to conduct the analysis. They account for the complex weight and variance in the sampling design and yield nationally-representative results. RESULTS
Table 1. Study Sample Characteristics (N = 1287) Variable Age (years) 18–34 35–64 $65 Gender Male Female Race Non-Hispanic white Non-Hispanic African-American Hispanic Other Rural/urban location Urban Rural Education level (years) 0–8 9–12 13–17 Don’t know or refused Insurance coverage Private Public Uninsured
Number
%
383 635 269
29.8 49.3 20.9
495 792
38.5 61.5
624 323 259 81
48.5 25.1 20.1 6.3
1073 214
83.4 16.6
115 653 512 7
8.9 50.7 39.8 0.6
611 443 233
47.5 34.4 18.1
The Study Sample Characteristics The demographic and socioeconomic characteristics of the sample used in the analysis are presented in Table 1. The final study sample consisted of 1287 adults with any ED visit in 2011, which represented a weighted 29,463,684 people in the total population. Among them, 390 individuals (30.3%) had nonurgent ED use, representing a weighted 8.3 million people in the total population. The following groups made up the highest proportion of the study sample: middle-aged individuals (49.3%), females (61.5%), non-Hispanic Whites (48.5%), private enrollees (47.5%), those living in the urban area (83.4%), and with high school education (50.7%). The Multivariate Analysis Results The multivariate analysis results are presented in Table 2. In the multivariate logit model, usual source of care and perceived convenience of needed medical care were not associated with nonurgent ED care costs (p = 0.497 for the usual source of care status, p = 0.293, p = 0.240, and p = 0.472 for the set of convenience dummy variables). Patient-perceived poor and intermediate levels of primary care quality had higher odds of nonurgent ED care costs (OR = 2.22, p = 0.035, and OR 2.05, p = 0.011, respectively) compared to high-quality care, with a marginal effect (at means) of 13.0% (95% confidence interval [CI] 0.006, 0.27) and 11.5% (95% CI 0.03–0.20) higher predicted probability of nonurgent ED care costs. In
addition, white and black individuals (compared to other races) older than 65 years (compared to individuals aged between 18 and 34 years) also had lower odds of nonurgent ED care costs (OR = 0.28, p = 0.002, OR = 0.34, p = 0.018, and OR = 0.40, p = 0.011, respectively). Patients with public insurance (compared to private coverage) living in urban areas (compared to rural areas) had higher odds of nonurgent ED care costs (OR = 1.72, p = 0.031, and OR = 1.81, p = 0.041, respectively). The cost comparison between ED and ambulatory care was not associated with higher or lower odds of nonurgent ED care costs (p = 0.767 for the equal cost group, and p = 0.705 for the cheaper ED group, compared to the more expensive ED group). Similarly, education level was not associated with higher or lower odds of nonurgent ED care costs (p = 0.712 for the high school group, and p = 0.718 for the college group, compared to the group with
