Younger Dual-Eligibles Who Use Federally Qualified Health Centers Have More Preventable Emergency Department Visits, but Some Have Fewer Hospitalizations

Abstract

Objective: To determine whether younger dual-eligibles receiving care at federally qualified health centers (FQHCs) have lower rates of ambulatory care sensitive (ACS) hospitalization and emergency department (ED) visits. Data Sources: We used the 100% Medicare Part A and Part B institutional claims from 2007 to 2010 for dual-eligibles younger than 65 years, enrolled in traditional fee-for-service Medicare, who received care at an FQHC or lived in a primary care service area with an FQHC. Methods: Our cross-sectional analysis used negative binomial regressions to model ACS hospitalizations and ED visits as a function of prior year FQHC use. The model adjusted for beneficiary age, gender, race, and chronic diseases, as well as county fixed effects, time trends, and race-FQHC use interactions. Results: FQHC use is associated with a decrease in ACS hospitalization rates for whites (2.8 per 1000 persons), but an increase among blacks (2.5 per 1000 persons). FQHC use is also associated with an increase in ACS ED visits, from 27 to 33 more visits per 1000 persons per year, depending on patient race. Conclusions: ACS hospital use is higher for FQHC users than nonusers, but white FQHC users have fewer ACS hospitalizations. More research is needed to understand how this relationship varies within and between centers.

Introduction

Emergency department (ED) visit and hospitalization rates for ambulatory care sensitive conditions (ACSCs) are widely used indicators of primary care access, because outpatient treatment for these conditions often prevents the need for hospital-based care.^1-5 More than 10 million Americans dually eligible for Medicare and Medicaid have higher ED visit and hospitalization rates for ACSCs because of barriers to accessing primary care.^1,6-10 Younger individuals are an especially high-cost subset of the dual-eligible population, and tend to encounter more barriers to care than older adult dual-eligibles, including greater difficulty making physician appointments, more transportation and mobility challenges, and impediments related to highly prevalent mental disorders.^11-15 Consequently, policy makers hope to improve healthcare delivery and care coordination among younger dual-eligibles to realize better health outcomes and generate cost savings.

Federally qualified health centers (FQHCs), which provide primary care regardless of ability to pay and offer a variety of nonclinical enabling services to improve access to care, have been shown to manage ACSCs effectively.^5,16 Despite the challenges of caring for a predominantly low-income, racial/ethnic minority population with high uninsurance rates, FQHCs deliver high-quality patient-centered medical care that addresses both financial and nonfinancial barriers to care.^17-19 As a result, the care delivered by FQHCs results in lower rates of hospitalization and ED visits for ACSCs.^4,5,16,20-25 However, younger dual-eligibles’ FQHC use has not been studied. Thus, this study sought to determine whether younger dual-eligibles who visit an FQHC have lower rates of hospitalization and ED visits for ACSCs. Given FQHCs’ focus on increasing access to care in predominantly black and Hispanic neighborhoods, we hypothesized that racial and ethnic minorities would benefit disproportionately from visiting an FQHC.²⁶

Methods

Data Sources

We obtained information on dual eligibility, FQHC use, and hospital-based care using all Medicare Part A claims, all Medicare Part B institutional claims, and the Medicare enrollment file (including the chronic conditions segment) for 2007-2010. We restricted our analysis to individuals who were enrolled in Medicare Part A, Medicare Part B, and Medicaid for an equal number of months, who were never enrolled in Medicare Advantage during the year, and whose current reason for eligibility was disability. We excluded disabled individuals with end-stage renal disease, since they have a vastly different relationship with the health care delivery system than other disabled enrollees. Using methods described previously, we further limited our sample to dual-eligibles residing in a primary care service area (PCSA) with an FQHC, or who visited an FQHC at any point during the year.^26,27 Finally, we excluded any individuals with missing data, which represented just 3% of our sample.

Outcome Variables

Our outcomes of interest were the number of hospitalizations and ED visits for ACSCs. We identified all hospitalizations using Part A claims, and all ED visits using both Part A and Part B claims listing any revenue code 0450-0459 or 0981. Since we conducted analyses separately for each outcome, we defined ED visits resulting in hospitalizations as both ED visits and hospitalizations. We used Prevention Quality Indicator software available from the Agency for Healthcare Research and Quality to flag hospitalizations and ED visits for any of the 16 ACSCs with the exception of low birthweight.²⁸

Independent Variable

Our key independent variable was FQHC use. Using the Part B claims, we identified FQHC visits using type of bill code 731 or 771. We defined FQHC use as a binary measure, equal to 1 if a person had at least 1 visit to an FQHC annually. In a sensitivity analysis, we coded FQHC use as a count of visits, and this change did not meaningfully alter our results. To eliminate the potential for reverse-causality, we lagged the FQHC use variable by 1 year, using a beneficiary’s FQHC use in 2007-2009 to predict their hospitalizations and ED visits for ACSCs in 2008-2010. In a sensitivity analysis, we repeated our analyses with no lag, and our results were substantively unchanged.

Analysis

We constructed 2 negative binomial regression models at the individual level to examine the relationship between FQHC use and the number of hospitalizations (model 1) or ED visits (model 2) for an ACSC. We adjusted for beneficiary characteristics (age, sex, race) and health status (count of 26 chronic conditions), and included county-level fixed effects to control for unobserved state and county characteristics that could affect our outcomes of interest (eg, differences in Medicaid policy, health care supply, and local practice patterns).^29,30 We also interacted race and FQHC use, to determine if the relationship between FQHC use and care for ACSCs varies by race. Finally, we included a series of yearly dummy variables to capture time trends. To account for correlation between the observations of the same subject in multiple years, we used clustered standard errors (at the county level) in each model. The Brown University and University of Iowa institutional review boards approved this study.

Results

From 2007 to 2009, the number of younger dual-eligibles who visited an FQHC during the year grew from 170 934 to 213 663. The proportion of younger dual-eligibles who use FQHCs rose slightly as well, from 18.2% to 18.5%. And, among all dual-eligible FQHC users, the proportion who are younger has increased from 54.1% to 56.4% over the study period. Table 1 compares younger dual-eligibles using FQHCs with their geographically matched counterparts. Notably, FQHC users are less likely than nonusers to be male, live in a metropolitan area, or live in the South or Midwest regions of the United States. However, FQHC users appear to be in worse health than nonusers, based on their higher average count of chronic conditions. On average, FQHC users have more ED visits (both ACS and total), but an essentially equal number of hospitalizations (both ACS and total) as nonusers.

Table 1.

Descriptive Statistics for Disabled Dual-Eligible Federally Qualified Health Center (FQHC) Users and Geographically Matched Nonusers.^a

	FQHC Users	Nonusers
n (person-years)	573 330	2 576 675
Age, y, mean (SD)	48.60b (10.45)	48.20 (10.69)
% Male	44.77b	49.46
% Metropolitan	74.76b	85.20
Race, %
White	64.37	64.33
Black	24.80b	25.72
Other race	10.83b	9.94
Region, %
Northeast	21.33b	15.80
Midwest	20.31b	23.99
South	31.36b	36.59
West	27.01b	23.63
No. of chronic conditions,c mean (SD)	2.87b (2.37)	2.73 (2.55)
No. of annual all-cause hospitalizations, mean (SD)	0.45 (1.28)	0.45 (1.27)
No. of annual ambulatory care sensitive hospitalizations, mean (SD)	0.06 (0.39)	0.06 (0.39)
No. of annual emergency department visits, mean (SD)	1.77b (4.22)	1.30 (3.26)
No. of annual ambulatory care sensitive emergency department visits, mean (SD)	0.16b (0.73)	0.13 (0.65)

^aDifferences between FQHC users and nonusers were determined by 2-sample t tests for continuous variables and Pearson chi-square tests for categorical variables.

^bIndicates a statistically significant (P < .01) difference between groups.

^cIncludes a count of 26 different conditions from the Chronic Conditions Data Warehouse.

In Table 2, we present results of our negative binomial regression models of ACS hospitalizations and ED visits as a function of FQHC use. We find that any FQHC use is associated with 4.1% fewer ACS hospitalizations among white younger dual-eligibles. This represents a decrease of 2.8 ACS hospitalizations per thousand persons per year (see Table 3). Because of the interaction terms, determining the incremental effect of FQHC use among Blacks and all other races requires calculating the average of the conditional predicted values from the model.³¹ We find that—compared with nonusers—any FQHC use is associated with an increase of 2.5 ACS hospitalizations per thousand persons per year among blacks, and an increase of 1.6 ACS hospitalizations per thousand persons per year among all other races (see Table 3).

Table 2.

ACS Hospitalizations and ED Visits Among Disabled Dual-Eligibles.

Parameters	Incidence Rate Ratio (Standard Error)
	Hospitalizations	ED Visits
FQHC use	0.959** (0.013)	1.257*** (0.014)
Age	0.979*** (0.0006)	0.971*** (0.0004)
Male	1.071*** (0.012)	0.829*** (0.005)
Black	1.164*** (0.031)	1.249*** (0.024)
Other race	0.814*** (0.027)	0.850*** (0.022)
FQHC use × Black	1.071* (0.032)	0.930** (0.020)
FQHC use × Other race	1.076* (0.038)	1.011 (0.029)
Year 2009	0.947*** (0.008)	0.982** (0.005)
Year 2010	0.852*** (0.010)	0.937*** (0.007)
No. of chronic conditions	1.653*** (0.008)	1.472*** (0.006)
Constant	0.020*** (0.001)	0.132*** (0.003)
Person-year observations	3 150 005

Abbreviations: ACS, ambulatory care sensitive; ED, emergency department; FQHC, federally qualified health center.

^*P < .05; **P < .01; ***P < .001.

Table 3.

Marginal Effects of FQHC Use on Ambulatory Care Sensitive Hospital Use, by Race/Ethnicity.

	Predicted Rates of ACS Hospitalizations (per Person-Year)	Predicted Rates of ACS ED Visits (per Person-Year)
White
FQHC users	0.066	0.163
Nonusers	0.069	0.130
Black
FQHC users	0.089	0.202
Nonusers	0.087	0.173
Other
FQHC users	0.052	0.126
Nonusers	0.050	0.099

Abbreviations: ACS, ambulatory care sensitive; ED, emergency department; FQHC, federally qualified health center.

Among younger dual-eligibles, each additional year of age is associated with 2% fewer ACS hospitalizations, while males have, on average, 7.1% more ACS hospitalizations than females. Each additional chronic condition is associated with having 65% more ACS hospitalizations. Year indicator variables show a downward trend in ACS hospitalizations over the 3-year study period, with 14.8% fewer ACS hospitalizations expected in 2010 compared with 2008.

Also in Table 2, we find that white FQHC users have, on average, over 25% more ACS ED visits than white nonusers. This represents an increase of 33 ACS ED visits per thousand persons per year (see Table 3). Calculating the average of the conditional predicted values, we find that compared to nonusers, FQHC use is associated with an increase of 28 ACS ED visits per thousand persons per year among blacks, and an increase of 27 ACS ED visits per thousand persons per year among all other races (see Table 3).

Similar to its effect on ACS hospitalizations, age is associated with nearly 3% fewer ACS ED visits. However, the relationship between gender and ACS ED visits moved in the opposite direction from that observed for ACS hospitalizations. On average, men have 17% fewer ACS ED visits than women. The coefficients of the year and chronic conditions variables in modeling ACS ED visits are similar to those seen when modeling ACS hospitalizations. Each additional chronic condition is associated with 47% more ACS ED visits. Year indicator variables show a downward trend in ACS ED visits over the 3-year study period, with approximately 6% fewer ACS ED visits expected in 2010 compared to 2008.

Discussion

We examined the relationship between FQHC use and the likelihood of hospitalizations and ED visits for ACSCs among younger dual-eligibles, and found mixed evidence. Receipt of care at an FQHC was associated with a modest decrease in the expected number of ACS hospitalizations for whites, but an increase for blacks, with no difference among those of all other races. There was a stronger relationship between FQHC use and higher rates of ACS ED visits among all racial groups. Our results deviate from those of a prior study of FQHCs as a usual source of care among Medicaid-only beneficiaries, which found a reduction in potentially preventable hospital care.⁵

One key difference between hospitalizations and ED visits is the role of the provider in admitting patients to the hospital, versus the decision to present at the ED, which is typically made by patients. Thus, hospitalizations for ACSCs are a more consistent measure of patients whose condition truly requires acute care. Despite FQHC users’ greater number of ED visits for these same conditions, FQHCs may play an important role in reducing chronic disease burden through primary care. However, we also offer 3 alternate explanations for our results that suggest the need for further research into FQHCs, younger dual-eligibles, and potentially preventable hospital care.

First, younger dual-eligibles who use FQHCs may differ in important and unobserved ways from those who never use FQHCs. For example, some younger dual-eligibles may have better social support from family or friends, helping them navigate the health care system and receive appropriate primary care. Consequently, they might visit the ED for nonemergent care less often. Conversely, those who lack these social supports may find the enabling services offered by FQHCs essential to getting needed care, and may also find the ED easier to access.

Second, we did not have access to the carrier claims that would allow us to assign beneficiaries a usual source of care. If FQHC use only reduces ACS hospitalizations and ED visits when the FQHC is a beneficiary’s usual—or sole—source of care, we will not identify that here. This also prevented us from identifying subgroups of younger dual-eligibles that might have different patterns of hospital use (eg, those with substance abuse disorders or mental health conditions).³²

Finally, there may be heterogeneity among FQHCs that we cannot account for here. Younger dual-eligibles who visit high-performing FQHCs may have lower rates of ACS hospitalizations and ED visits, while those who visit poorly performing FQHCs have worse outcomes. By grouping all FQHC users together, we may be masking the relative contributions of different FQHCs to potentially preventable hospital care. Future research should consider collapsing beneficiary records to the FQHC-level and repeating our analyses, as well as identifying heterogeneous effects on the basis of disability type.

Ultimately, dual-eligibles are a focus for policy makers because of the potential to improve healthcare while lowering costs. Currently, health care for younger dual-eligibles is compromised by provider and payer fragmentation, and evidence suggests that payment and delivery reform must be approached simultaneously to be effective.^33-35 On the whole, we found that FQHC use is growing among dual-eligibles, and is growing more quickly among younger dual-eligibles than among dual-eligibles older than 65 years (data not shown). The upward trend in FQHC use among younger dual-eligibles provides an additional reason for providers, states, and CMS to consider FQHCs when designing and implementing care coordination initiatives and payment reforms.

Limitations

Our study has several limitations. First, because we use fee-for-service Medicare data, our results may not be generalizable to the Medicare managed care population. While 21.1% of all dual-eligible beneficiaries were enrolled in Medicare Advantage in 2009, this is less common in our sample, as just 15.8% of younger dual-eligibles were enrolled in Medicare Advantage in 2009 (authors’ analysis of claims data, results not shown). Second, we do not have access to the Medicare claims data that would allow us to determine whether a dual-eligible is institutionalized. Because institutionalized beneficiaries have higher rates of potentially-preventable hospitalization than community-dwelling beneficiaries, if FQHC use is negatively associated with being institutionalized, our results may be biased downward.³⁶ Fortunately, younger dual-eligibles have high rates of community residence. Finally, while we rely on established diagnoses to determine potentially-preventable causes of hospitalization and ED visits, claims data cannot be used to definitively conclude that any particular hospitalization or ED visit was actually preventable.

Conclusion

To our knowledge, this study is the first to demonstrate the relationship between FQHC use and potentially preventable hospitalizations and ED visits among younger dual-eligibles. We find that FQHC use is associated with a net decrease in ACS hospitalizations, with variation by race, and a net increase in ACS ED visits, irrespective of race. However, more research is needed to understand the extent to which FQHC use may reduce potentially preventable hospital-based care and how this relationship varies across centers.

Author Biographies

Andrew J. Potter is completing his doctorate in health management and policy at the University of Iowa. He has accepted a tenure-track position in political science at California State University Chico. His interests include Medicaid and long-term services and supports.

Amal N. Trivedi is an associate professor in health services, policy, and practice at Brown University. He is also a staff physician with the Providence Veterans Affairs Medical Center. His research focuses on quality of care and health disparities, with particular emphasis on the impact of patient and provider incentives on quality and equity of care.

Brad Wright is an assistant professor in health management and policy at the University of Iowa. His research interests include access to care for vulnerable populations, disparities in health and healthcare delivery, and health politics and policy. He has studied federally qualified health centers extensively.