SEARCH

SEARCH BY CITATION

Keywords:

  • healthcare reform;
  • healthcare policy;
  • public policy

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

Lack of healthcare insurance is considered one of the drivers of the rise in emergency department (ED) use in the United States. Using survey-based, individual-level data, we compare ED use after the 2006 Massachusetts health insurance reform with ED use before the reform both in Massachusetts and nearby states. We find that the reform increased the insurance rate significantly by 5.29 percentage points. We do not find a statistically significant effect on ED visits but do not have enough power to rule out potentially relevant effects. It seems that policy makers hoping that insurance reform will dramatically decrease ED overcrowding are likely to be disappointed.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

From 1996 to 2006 in the United States, the number of emergency department (ED) visits increased by 32 percent while the number of EDs declined by 5 percent (Pitts, Niska, Xu, & Burt, 2008). Many EDs across the country are now experiencing overcrowding (Kellermann, 2006).

Overcrowding is often attributed to ED use by uninsured individuals. Policy makers have argued that the uninsured rely on EDs to receive care because the Emergency Medical Treatment and Active Labor Act (EMTALA) requires EDs to treat patients regardless of their ability to pay (General Accounting Office, 2001; Olshaker & Rathlev, 2006; Pane, Farner, & Salness, 1991). It is also perceived that growth in the number of the uninsured over the past decade has led to increases in ED visits and overcrowding (General Accounting Office, 2001).

If the uninsured are an important driver of emergency-department crowding, increasing the share of the population with health insurance should reduce ED visits. This hypothesis is particularly relevant given the 2010 Affordable Care Act, which seeks to reduce the share of U.S. residents without health insurance. Is this reform likely to reduce ED crowding?

The 2006 Massachusetts (MA) healthcare reform provides a natural experiment to test this hypothesis. This study investigates whether Massachusetts' healthcare reform increased health insurance in the state and whether this increase reduced ED use compared with nearby states.1

Three studies have previously analyzed the effect of the Massachusetts healthcare reform on ED visits. Miller (2012) exploited the variation in insurance rate across counties in Massachusetts before reform to estimate the effect of the reform on ED use. The author found that the Massachusetts reform reduced ED use by between 5 and 8 percent. Her data came from the Acute Hospital Case Mix Database, which allowed her to investigate the effect of the reform by type and time of visits. She found that ED visits for problems that could be treated in a physician's office declined while she did not find any impact on non-preventable visits. Also, she did not find any impact on ED visits outside office hours. Because she aggregated the data at the county level, she cannot observe individuals' characteristics.

Kolstad and Kowalski (2012) used the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) to investigate the impact of the Massachusetts reform on hospitalizations and ED visits. They found that hospital admissions originating from EDs declined after the reform. Unlike survey data that relies on individual responses which is usually top-coded, the authors observed actual hospital visits. In addition, they examined the effect of the reform on hospital costs. The hospital-level data, however, included only those ED visits that led to inpatient admission. Unfortunately, this sample-selection bias cuts against the hypothesis that the uninsured use emergency departments for primary care. Under this hypothesis, we would expect the uninsured to disproportionately use the ED for non-urgent care—exactly the sort of visit not observed in the Kolstad and Kowalski data. While the data we use has limitations, we believe it is more relevant to this question.

In addition, Chen, Scheffler, and Chandra (2011) compared trends in ED use in Massachusetts and two neighboring states (Vermont and New Hampshire) and found that the reform had not changed the trend of ED use in Massachusetts relative to the control group.

Using a difference-in-differences methodology, we compare the effect of going from 2003 to 2007 in Massachusetts and in comparison states. Unlike the above studies, we use survey-based, individual-level data, which allows us to control for individual differences. We also observe both total ED visits and ED visits that did not result into inpatient admission.

Background

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

Overcrowding in EDs has become a nationwide problem. A survey of hospital ED directors showed that ED crowding is a problem in almost all states and leads to long waits and ambulance diversion (Olshaker & Rathlev, 2006).

Wilper et al. (2008) found that the median wait time to see an ED physician increased from 22 min in 1997 to 30 min in 2004 (Wilper et al., 2008). A 2009 U.S. Government Accountability Office (GAO) report indicated that on average, patients needing emergency care were waiting longer than the recommended maximum wait time. Generally, at the time of ED arrival, a triage nurse categorizes patients into five groups based on their acuity level: immediate (should be seen by a physician in less than 1 min), emergent (should be seen in less than 15 min), urgent (can wait for 1 h), semi-urgent (can wait for 2 h), and non-urgent (can wait for 24 h). In 2006, patients who needed immediate care waited 28 min to be seen by a physician and those categorized into emergent care waited for 37 min (Government Accountability Office, 2009).

In addition, overcrowding leads to adverse clinical outcomes (Wharam et al., 2007). For example, Chalfin et al. (2007) showed that patients who were boarded in EDs for more than 6 h (those who were admitted to the facility but were held in the ED, instead of being transferred to the ICU) had a 5 percent higher mortality rate than those who were boarded for less than 6 h. Approximately 2 percent of patients admitted from the ED to the ICU were boarded in the ED for more than 6 h.

Moreover, many medical services are more expensive to provide in EDs than in other settings (Bamezai, Melnick, & Nawathe, 2005). A comparison in five Massachusetts hospitals between patients who visited EDs and those admitted from other departments indicates that patients admitted to EDs had higher costs when adjusted for disease severity (Stern, Weissman, & Epstein, 1991).

One of the most cited causes of ED crowding is the uninsured. It is often argued that uninsured individuals do not have access to primary care and they rely on EDs to receive medical care. Several studies of individual EDs attributed crowding to uninsured and underinsured individuals. Analyzing 1,000 patients who walked into the ED of the University of California Irvine Medical Center, Pane et al. (1991) showed that patients with low income, no insurance, or under-insurance were more likely to use the ED. Stern et al. (1991) compared inpatients admitted from the ED with other inpatients in five different hospitals in Massachusetts and found that the uninsured, Medicaid recipients, the elderly, and low-socioeconomic individuals were more likely to be admitted through EDs.

These older, hospital-based studies do not differentiate between the uninsured and the underinsured, but are the empirical basis for the “the uninsured use EDs” claim. Drawing on national samples, more recent studies have argued that the uninsured are not responsible for the growing use of EDs. Weber, Showstack, Hunt, Colby, and Callaham (2005) showed that uninsured individuals were not more likely to visit an ED. Cunningham (2006) found that communities with high ED use had fewer uninsured residents than low ED-use communities.

Also, it is difficult to argue that the uninsured have caused the recent increases in ED visits given that their share of the ED population has been stable (Weber et al., 2008) and is proportionate to the share of the entire population. Between 1996 and 2003, the uninsured accounted for 15 percent of the population, 14 percent of ED visits, and 12 percent of ED expenditures (Peppe, Mays, Chang, Becker, & DiJulio, 2007).

We add to this literature by studying an exogenous shock to insurance levels. We exploit a natural experiment on insurance by investigating the effect of providing near universal health coverage on ED visits in Massachusetts. Recent studies on the effect of Massachusetts reform used either hospital-level data (Kolstad & Kowalski, 2012) or county-level data (Miller, 2012). We will investigate the impact of the reform by examining survey-based, individual-level data that allows us to control for individual characteristics. In addition, we observe both total ED visits and outpatient ED visits (ED visits that did not result into inpatient admission), whereas Kolstad and Kowalski (2012) examined only those visits resulting in inpatient admission. We find that the reform increased the health insurance coverage significantly by 5.29 percentage points while we do not find a statistically significant impact on ED visits.

Data and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

Sample

We use the fourth and fifth rounds of the Community Tracking Study (CTS) Household Survey, conducted by the Center for Studying Health System Change from 2003 to 2004 and 2007 to 2008.

The survey asked a “family informant” about basic demographic and insurance information for all household members. Each adult in the household (including the informant) then answered questions on detailed health topics including ED visits for the past year. Up to eight persons per household were included in the survey. Parents answered questions about their children.

Individuals in 60 communities—stratified by region, community size, and metropolitan/non-metropolitan—were interviewed by telephone in the first four rounds of the surveys. Households in these communities were selected by random-digit dialing. To reach individuals without telephones, interviewers visited 12 metropolitan areas.2

The CTS changed its design for the 2007–2008 survey to a national-sample design. In this round, households were randomly selected from 48 states in the United States by random-digit dialing. To maintain as much comparability as possible between the rounds, we restrict our 2007–2008 data to include only individuals sampled from the same 5-digit FIPS codes included in the 2003–2004 sample. Our sample includes 10,213 individuals (8,961 observations in 2003 and 1,252 observations in 2007).

Measures

We examine the effect of the reform on four dependent variables. The first dependent variable is whether the respondent has health insurance. The second dependent variable is total ED visits in the last year. This variable is top-coded at 7 for confidentiality reasons and is based on the respondent's answer to the question, “During the past 12 months, how many times have you gone to a hospital emergency room?” Due to the top coding, we are unable to see reform effects on high-frequency ED users. Figure 1 shows the overall distribution of ED visits. The third dependent variable is outpatient ED visits. This variable is also a categorical variable and top-coded at 7 (see Fig. 2).

image

Figure 1. Distribution of Total ED Visits.

Source: Community Tracking Study Household Survey, rounds 4 and 5.

Download figure to PowerPoint

image

Figure 2. Distribution of Outpatient ED Visits.

Source: Community Tracking Study Household Survey, rounds 4 and 5.

Download figure to PowerPoint

We also check the effect of the reform on whether the respondent reported any visits to the ED in the previous 12 months. This binary variable is our fourth dependent variable that is constructed from the total ED visit variable.

Control variables include age, sex, race, general health status, and log of family income. General health status was constructed by asking individuals, “In general, would you say your health is excellent, very good, good, fair or poor?” We dichotomized this variable into good (excellent, very good, good) and poor (fair or poor).

Analysis

We use a difference-in-differences methodology. For both insurance status and ED use, we compare the effect of going from 2003 to 2007 in MA and in a comparable group of neighboring states (Connecticut, New York, New Jersey, and Maine) that were included in the 2003 survey.3

Our dependent variables are not continuous. As a result, we need to use nonlinear models to analyze the data. For the categorical measures (the number of total ED visits and the number of outpatient ED visits in the last year), we use the negative binomial model; for the binary measures (having insurance and having at least one ED visit in the last year), we use the logit model. According to Norton, Wang, and Ai (2004), in nonlinear models, the difference-in-difference estimate is not captured by the coefficient-of-interaction term. In order to estimate the effect of the reform on health insurance coverage (ED visit), we need to find the predicted probabilities of having health insurance (ED visit) for each individual in our sample, then find the average of these predicted probabilities and calculate the difference-in-difference estimate.

To determine the reform effect on health insurance coverage, we begin by running the following logit model. inline image is a binary variable for insurance status for person i in state s at time t. The variable inline image equals 1 for residents in Massachusetts. inline image is 0 for observations in 2003–2004 and 1 for observations in 2007–2008. inline image equals 1 for observations in MA in 2007–2008 and 0 otherwise. Xist is a vector of control variables that include age, sex, race, general health status, and log of family income,

  • display math

where F(•) indicates the logit distribution.

We then use the logit results to find the predicted probability of having insurance for each individual under different scenarios (in Massachusetts after the reform, in Massachusetts before the reform, in the control group after the reform, and in the control group before the reform) holding all control variables at their sample means. We than calculate the average of these predicted probabilities for the four scenarios. Finally, we calculate the difference-in-difference estimate by computing how much the predicted coverage changed from pre-reform to post-reform in Massachusetts relative to the control group. The following formula gives the difference-in-difference (DD) estimate:

  • display math

To estimate the effect of the reform on ED use, we replace the dependent variable with the categorical variables (“total ED visits” and “outpatient ED visits”) and change to a negative binomial model. We then follow the same three steps and use similar independent variables for each measure to calculate the difference-in-difference estimates.

Sample weights provided by the Community Tracking Study Household Survey are used to adjust for unequal-selection probabilities (Strouse et al., 2009). Standard errors account for the complex sample design.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

Data for this study include 2,446 observations in Massachusetts and 7,767 observations in the comparison states. Individuals in Massachusetts and in the control group were similar across age, sex, general health condition, and having access to the usual sources of care (see Table 1). In terms of race, Massachusetts had more white respondents and fewer African-American and Hispanic respondents. The mean of family income was also higher in Massachusetts.

Table 1. Observable Characteristics of Massachusetts and Comparison States
CharacteristicsIndividuals in Massachusetts, N = 2,446Individuals in the Control Group, N = 7,767All Individuals, N = 10,213
Source: Community Tracking Study Household Survey, rounds 4 and 5. SDs in parentheses.
Age37.67 (22.62)36.77 (22.26)36.98 (22.34)
Sex
Female0.55 (0.50)0.53 (0.50)0.54 (0.50)
Race
White0.82 (0.38)0.60 (0.49)0.65 (0.48)
Black0.04 (0.19)0.17 (0.38)0.14 (0.35)
Other0.07 (0.26)0.07 (0.26)0.07 (0.26)
Hispanic0.07 (0.26)0.15 (0.36)0.13 (0.34)
Has Usual Source of Care
Yes0.91 (0.28)0.87 (0.34)0.88 (0.33)
General Health Condition
Good0.89 (0.31)0.86 (0.34)0.87 (0.34)
Family Income74,751 (45,000)64,489 (46,000)66,851 (46,000)

Figure 3 shows the percentage of people with different numbers of ED visits before and after the reform in both Massachusetts and the comparison group. The percent of individuals with one ED visit increased slightly after the reform in Massachusetts, but in general, the number of ED visits did not change dramatically.

image

Figure 3. Distribution of Total ED Visits in Massachusetts and in the Control Group, Before and After the Reform.

Source: Community Tracking Study Household Survey, rounds 4 and 5.

Download figure to PowerPoint

Now we turn to the regression results. For each measure, we report the predicted value for Massachusetts and the comparison states, before and after the reform, holding all control variables at their sample means. All confidence intervals (CI) are reported at the 95 percent level. All tests are two-tailed.

Relative to the control group, the percentage of insured individuals in Massachusetts increased significantly by 5.29 percentage points (CI 0.10 to 10.49 points). See Table 2.

Table 2. Change in the Predicted Percentage of Insured Individuals
 Before the Reform (2003–2004)After the Reform (2007–2008)Difference
  • **

    Significant at 0.05,

  • *

    Significant at 0.10.

Source: Authors' analysis of the Community Tracking Study Household Survey, rounds 4 and 5. SEs in parentheses. The computations hold control variables fixed at their sample means.
Massachusetts95.92%98.25%2.33%
[94.01, 97.84]**[96.84, 99.66]**[0.00, 4.67]*
(0.010)(0.007)(0.012)
Control Group91.99%89.03%−2.96%
[89.35, 94.64]**[85.32, 92.74]**[−7.69, 1.76]
(0.013)(0.019)(0.024)
Difference3.93%9.22%5.29%
[0.71, 7.15]**[5.25, 13.20]**[0.00, 10.49]**
(0.016)(0.020)(0.026)

Table 3 shows the predicted number of ED visits calculated by the negative binomial model. Relative to the control group, the predicted number of ED visits per respondent in Massachusetts declined insignificantly by 0.018 or 18 ED visits per 1000 (CI −0.198 to 0.162), a 4.6 percent decline.

Table 3. Change in the Predicted Number of ED Visits
 Before the Reform (2003–2004)After the Reform (2007–2008)Difference
  • **

    Significant at 0.05.

Source: Authors' analysis of the Community Tracking Study Household Survey, rounds 4 and 5. SEs in parentheses. The computations hold control variables fixed at their sample means.
Massachusetts0.3900.4130.024
[0.312, 0.467]**[0.303, 0.524]**[−0.111, 0.158]
(0.039)(0.056)(0.069)
Control Group0.3560.3990.042
[0.296, 0.417]**[0.299, 0.497]**[−0.079, 0.164]
(0.031)(0.051)(0.062)
Difference0.0330.015−0.018
[−0.066, 0.132][−0.133, 0.163][−0.198, 0.162]
(0.050)(0.075)(0.092)

Given that the insurance rate increased by 5.29 percentage points, each percentage point of additional insurance rate may have had an effect of −0.0034 visits per respondent (−0.018/5.29), with a potential range of −0.037 (−0.198/5.29) to 0.031 (0.162/5.29) visits.

As Table 4 shows, relative to the control group, the predicted number of outpatient ED visits per respondent in Massachusetts declined insignificantly by 0.026 (CI −0.184 to 0.132).

Table 4. Change in the Predicted Number of Outpatient ED Visits
 Before the Reform (2003–2004)After the Reform (2007–2008)Difference
  • **

    Significant at 0.05.

Source: Authors' analysis of the Community Tracking Study Household Survey, rounds 4 and 5. SEs in parentheses. The computations hold control variables fixed at their sample means.
Massachusetts0.3150.3200.005
[0.244, 0.386]**[0.216, 0.425]**[−0.121, 0.131]
(0.036)(0.053)(0.064)
Control Group0.2680.3000.031
[0.214, 0.323]**[0.225, 0.374]**[−0.064, 0.127]
(0.028)(0.038)(0.049)
Difference0.0470.021−0.026
[−0.044, 0.138][−0.107, 0.149][−0.184, 0.132]
(0.046)(0.065)(0.081)

Finally, after the reform, the percentages of individuals with at least one ED visit in the past year increased insignificantly in Massachusetts by 3.80 percentage points (CI −4.87 to 12.47 points) relative to the control group (see Table 5). This suggests that the reform did not change the percentage of individuals with at least one ED visit. Given the low precision of our estimates, we do not want to read too much into the difference in sign between our estimates for ED use and our estimates for the percentage of people using EDs. However, this result supports the general pattern of the other results in that reform did not have a significant effect on ED use in the short term.

Table 5. Change in the Percentage of Individuals With at Least One ED Visit
 Before the Reform (2003–2004)After the Reform (2007–2008)Difference
  • **

    Significant at 0.05.

Source: Authors' analysis of the Community Tracking Study Household Survey, rounds 4 and 5. SEs in parentheses. The computations hold control variables fixed at their sample means.
Massachusetts23.7%27.8%4.2 points
[20.3, 27.03]**[21.9, 33.8]**[−2.7, 11.1]
(1.7%)(3.0%)(3.5)
Control Group23.4%23.7%0.4 points
[19.7, 27.0]**[19.9, 27.6]**[−4.9, 5.7]
(1.9%)(1.9%)(2.7)
Difference0.3 points4.1 points3.8 points
[−4.7, 5.3][−3.0, 11.2][−4.9, 12.5]
(2.5)(3.6)(4.4)

As a robustness check, in only one specification, we include all states in our sample and estimate the effect of the reform in Massachusetts in comparison to all states. Table 6 shows the predicted number of ED visits when all states are used as the comparison group.

Table 6. Change in the Predicted Number of ED Visits, Relative to All Other States
 Before the Reform (2003–2004)After the Reform (2007–2008)Difference
  • **

    Significant at 0.05.

Source: Authors' analysis of the Community Tracking Study Household Survey, rounds 4 and 5. SEs in parentheses. The computations hold control variables fixed at their sample means. This table reports a robustness check comparing Massachusetts to all other states, not only neighboring states.
Massachusetts0.4040.4190.015
[0.318, 0.490]**[0.310, 0.528]**[−0.123, 0.154]
(0.044)(0.055)(0.071)
Control Group0.3420.3550.013
[0.324, 0.359]**[0.310, 0.528]**[−0.022, 0.048]
(0.009)(0.016)(0.018)
Difference0.0620.0640.002
[−0.026, 0.151][−0.484, 0.177][−0.141, 0.145]
(0.045)(0.057)(0.073)

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

Using survey-based, individual-level data, we investigated the impact of the Massachusetts healthcare reform on ED use. We found that ED visits by residents of Massachusetts have not declined significantly after the reform, although the reform has increased the number of the insured significantly. This finding is consistent with studies that have analyzed nationally representative samples of individuals in contrast to studies on individual hospitals. It also matches the result of Chen et al. (2011) that showed the trend of ED use in Massachusetts has not changed since the 2006 reform.

Not finding a significant change in ED visits after the reform may imply three major points. First, the benefit of the reform (providing access to primary care for a larger population) was quickly negated by another driver of ED visits (reducing out-of-pocket costs for the previously uninsured). Second, insurance status is not a major determinant of ED use. A survey of ED users in California indicates that convenience and a positive attitude toward EDs play a major role in choosing EDs (California Healthcare Foundation, 2006). The insured ED users in California believe that they receive better diagnostic testing, easier access to specialists, and get higher quality care in EDs. Third, people continue with the type of provider that they are used to (EDs) and it takes time for them to switch to another type of provider (physician offices). Our analysis investigated just the short-term effect of the reform. The reform may reduce the ED visits in the future.

In addition, Bertrand, Duflo, and Mullainathan (2004) argue that when the dependent variable is serially correlated, the conventional difference-in-difference models may underestimate standard deviations and lead to a significant result when there is not really an effect. The authors showed that collapsing the data into before and after periods can address this problem when the number of states is small. Consistent with this study, our data set has only two periods (pre- and post-reform) and the number of states in our sample is small.

Miller (2012) finds a reduction between 0.023 and 0.037 visits per resident (or a 5 and 8 percent decline in ED use). The author interprets these declines as large. Kolstad and Kowalski (2012) also find a 5 percent decline in inpatient admission originating from EDs. While our data do not give us enough power to definitively pin down the short-term effect of the Massachusetts reform on ED use, our point estimate (−4.6 percent) is close to the other two studies' estimate (−5 percent) although ours is not significant.

Considering the fact that ED visits are increasing in the United States by 3.2 percent each year (Pitts et al., 2008), a decline on the order of 5 percent is considered a moderate decrease. Policy makers hoping that insurance reform would “bend the curve” in ED use are likely to be disappointed. We interpret our finding, and the relevant literature in general, as suggesting that the most important drivers of ED use lie elsewhere.

Limitations

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies

The last round of the Community Tracking Study Household Survey was conducted from 2007 to 2008. As a result, we do not have data on ED visits after 2008. The reform may be affecting ED visits with a lag. The standard limitations of survey data apply to this study as well, such as non-response and recall bias, although these biases should be partially mitigated by the survey's sampling methods and weighting. In addition, the data on ED visits is top-coded at 7 for confidentiality reasons and we are unable to see the effect of the reform on frequent ED users. Approximately 1 percent of our sample had 7 or more ED visits.

The biggest limitation is the relatively small number of respondents in Massachusetts after the reform (318 observations), which greatly weakens the precision of our results. We view our point estimates as suggestive, but do not have the precision to draw strong conclusions.

Furthermore, there are clearly substantial issues of external validity if we try to use the Massachusetts results to anticipate the effects of national health insurance reform, especially given the already high levels of insurance in the state before the 2006 reform.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies
  • Bamezai, Anil, Glenn Melnick, and Amar Nawathe. 2005. “ The Cost of an Emergency Department Visit and Its Relationship to Emergency Department Volume.” Annals of Emergency Medicine 45 (5): 48390.
  • Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan. 2004. “ How Much Should We Trust Differences-In-Differences Estimates?The Quarterly Journal of Economics 119 (1): 24975. doi: 10.1162/003355304772839588.
  • California Healthcare Foundation. 2006. Overuse of Emergency Departments Among Insured Californians [Online]. http://www.chcf.org/∼/media/MEDIA%20LIBRARY%20Files/PDF/E/PDF%20EDOveruse.pdf. Accessed April 2012.
  • Chalfin, Donald B., Stephen Trzeciak, Antonios Likourezos, Brigitte M. Baumann, and R. Phillip Dellinger, and DELAY-ED study group. 2007. “ Impact of Delayed Transfer of Critically Ill Patients From the Emergency Department to the Intensive Care Unit.” Critical Care Medicine 35 (6): 147783.
  • Chen, Christopher, Gabriel Scheffler, and Amitabh Chandra. 2011. “ Massachusetts' Health Care Reform and Emergency Department Utilization.” New England Journal of Medicine 365 (12): e25. doi: 10.1056/NEJMp1109273.
  • Cunningham, Peter J. 2006. “ What Accounts for Differences in the Use of Hospital Emergency Departments Across U.S. Communities?Health Affairs 25 (5): w32436. doi: 10.1377/hlthaff.25.w324.
  • General Accounting Office. 2001. EMTALA Implementation and Enforcement Issues. GAO-01-747.
  • Government Accountability Office. 2009. Hospital Emergency Departments: Crowding Continues to Occur, and Some Patients Wait Longer than Recommended Time Frames. GAO-09-347.
  • Gruber, Jonathan. 2010. “ Massachusetts Points the Way to Successful Health Care Reform.” Journal of Policy Analysis and Management 30 (1): 18492. doi: 10.1002/pam.20551.
  • Holtz-Eakin, Douglas. 2010. “ Does Massachusetts's Health Care Reform Point to Success With National Reform?Journal of Policy Analysis and Management 30 (1): 17884. doi: 10.1002/pam.20553.
  • Kellermann, Arthur L. 2006. “ Crisis in the Emergency Department.” The New England Journal of Medicine 355 (13): 13003. doi: 10.1056/NEJMp068194.
  • Kolstad, Jonathan T., and Amanda E. Kowalski. 2012. “ The Impact of Health Care Reform on Hospital and Preventive Care: Evidence From Massachusetts.” Journal of Public Economics 96 (11–12): 90929. http://dx.doi.org/10.1016/j.jpubeco.2012.07.003.
  • Miller, Sarah. 2012. “ The Effect of Insurance on Emergency Room Visits: An Analysis of the 2006 Massachusetts Health Reform.” Journal of Public Economics 96 (11–12): 893908. http://dx.doi.org/10.1016/j.jpubeco.2012.07.004.
  • Olshaker, Jonathan S., and Niels K. Rathlev. 2006. “ Emergency Department Overcrowding and Ambulance Diversion: The Impact and Potential Solutions of Extended Boarding of Admitted Patients in the Emergency Department.” Journal of Emergency Medicine 30 (3): 35156.
  • Pane, Gregg A., Michael C. Farner, and Kym A. Salness. 1991. “ Health Care Access Problems of Medically Indigent Emergency Department Walk-In Patients.” Annals of Emergency Medicine 20 (7): 73033.
  • Peppe, Elizabeth M., Jim W. Mays, Holen C. Chang, Eric Becker, and Bianca DiJulio. 2007. Characteristics of Frequent Emergency Department Users. Menlo Park, CA: Kaiser Family Foundation.
  • Pitts, Stephen R., Richard W. Niska, Jianmin Xu, and Catharine W. Burt. 2008. “ National Hospital Ambulatory Medical Care Survey: 2006 Emergency Department Summary.” Advanced Data 7: 140.
  • Stern, Robert S., Joel S. Weissman, and Arnold M. Epstein. 1991. “ The Emergency Department as a Pathway to Admission for Poor and High-Cost Patients.” The Journal of the American Medical Association 266 (16): 223843. doi: 10.1001/jama.1991.03470160070034.
  • Strouse, Richard, Barbara Carlson, John Hall, and Karen Cybulski. 2009. Community Tracking Study. Household Survey Methodology Report 2007 (Round Five). Technical Publication No. 72. Washington, DC: Center for Studying Health System Change [Online]. http://www.hschange.com/CONTENT/1060/1060.pdf. Accessed January 2011.
  • Weber, Ellen J., Jonathan A. Showstack, Kelly A. Hunt, David C. Colby, and Michael L. Callaham. 2005. “ Does Lack of a Usual Source of Care or Health Insurance Increase the Likelihood of an Emergency Department Visit? Results of a National Population-Based Study.” Annals of Emergency Medicine 45 (1): 412.
  • Weber, Ellen J., Jonathan A. Showstack, Kelly A. Hunt, David C. Colby, Barbara Grimes, Peter Bacchetti, and Michael L. Callaham. 2008. “ Are the Uninsured Responsible for the Increase in Emergency Department Visits in the United States?Annals of Emergency Medicine 52 (2): 10815.
  • Wharam, J. Frank, Bruce E. Landon, Alison A. Galbraith, Ken P. Kleinman, Stephen B. Soumerai, and Dennis Ross-Degnan. 2007. “ Emergency Department Use and Subsequent Hospitalizations Among Members of a High-Deductible Health Plan.” The Journal of the American Medical Association 297 (10): 1093102. doi: 10.1001/jama.297.10.1093.
  • Wilper, Andrew P., Steffie Woolhandler, Karen E. Lasser, Danny McCormick, Sarah L. Cutrona, David H. Bor, and David U. Himmelstein. 2008. “ Waits To See An Emergency Department Physician: U.S. Trends And Predictors, 1997–2004.” Health Affairs 27 (2): w8495. doi: 10.1377/hlthaff.27.2.w84.
Notes
  1. 1

    See Gruber (2010) and Holtz-Eakin (2010) for discussions of extrapolating from the Massachusetts experience to the United States as a whole.

  2. 2

    The surveys were conducted in English and Spanish. Due to language barriers, only 0.3 percent of households were not interviewed. The response rates for the 2003 and 2007 surveys were 56.5 and 43.5 percent, respectively (Strouse, Carlson, Hall, & Cybulski, 2009).

  3. 3

    New Hampshire, Rhode Island, and Vermont were not sampled in 2003–2004 thus we cannot include them in our control group. New Jersey is a more distant neighbor but very similar to Massachusetts in observable characteristics.

Biographies

  1. Top of page
  2. Abstract
  3. Introduction
  4. Background
  5. Data and Methods
  6. Results
  7. Discussion
  8. Limitations
  9. References
  10. Biographies
  • Niyousha Hosseinichimeh got her Ph.D. in Public Administration and Policy from SUNY at Albany in 2012. She is currently a post-doctoral associate at Virginia Tech, where she works on modeling of major depressive disorder.

  • Stephen Weinberg (Ph.D. economics, Harvard 2007) is MPA Director at SUNY Albany's Rockefeller College of Public Affairs and Policy, where he teaches health policy and statistics.