Abstract
Introduction
In 2014, the Affordable Care Act (ACA) provided funding for states to expand Medicaid coverage to include citizens who earned up to 138% of the federal poverty line. We sought to ascertain whether physicians practicing in Medicaid expansion states reported an increase in Medicaid or newly insured patients with type 2 diabetes in their panels, compared to physicians practicing in non-expansion states.
Methods
We conducted a 55-question cross-sectional survey of 356 physicians providing outpatient care for adults with type 2 diabetes. We used adjusted multivariate logistic regression analyses to compare responses from physicians who practiced in expansion vs. non-expansion states regarding whether they observed an increase since 2014 in 1) the number of Medicaid or newly insured patients with diabetes and 2) the number of additional newly or previously diagnosed patients who were newly receiving care, in their panels, adjusting for physician, practice, and patient-level characteristics, weighted for the U.S. physician population.
Results
41% of eligible recipients responded. 64.2% of physicians who practice in an expansion state report an increase in Medicaid or newly insured patients with diabetes compared with 46.1% who practice in non-expansion states (p = 0.05; Table 2).
Conclusion
Compared with physicians who practice in non-expansion states, physicians who practice in Medicaid expansion states are more likely to report an increase in the number of Medicaid or newly insured patients with diabetes in their practice since 2014. The increased access associated with the Medicaid expansion may improve long-term outcomes for patients with type 2 diabetes.
Keywords: Medicaid, access to care, diabetes, Affordable Care Act
INTRODUCTION
Beginning in 2014, the Affordable Care Act (ACA) provided funding for states to expand Medicaid coverage to include citizens who earned up to 138% of the federal poverty line. Due to a 2012 Supreme Court ruling, however, the decision to expand Medicaid was left to individual states. Twenty-four states chose to expand Medicaid coverage and seven used waivers to expand coverage in ways that departed substantially from the ACA’s provisions. Understanding the impact of the Medicaid expansion on patients with diabetes is particularly important because diabetes has the highest level of personal health care spending in the U.S., and outcomes may improve with earlier diagnosis and better access to care. [1–3]
In 2009-2010, 23% of adults ages 18-64 with diagnosed diabetes were uninsured — roughly 3.5 million Americans. [4] Of these patients, 1.5 million met income qualifications for expanded Medicaid coverage. A recent study comparing insurance among adults age 18-64 with diabetes in 2016 versus 2009 found a statistically significant 5% increase in coverage.[5] The Medicaid expansion also created opportunities to identify diabetes among those unaware of their disease.[6] Rates of diabetes diagnosis increased by 5% in expansion states compared with non-expansion states, and studies using laboratory and prescription data have likewise suggested that new diabetes diagnoses in Medicaid expansion states increased compared to non-expansion states.[7–10]
To date, however, it is unclear how providers responded to the influx of patients with diabetes due to Medicaid expansion. Because of the program’s lower reimbursement rates, physicians have historically been less likely to accept Medicaid patients. [11] It is therefore possible that only a small number of physicians accepted new Medicaid patients with diabetes into their practices. Thus, we examined whether and to what degree providers in Medicaid expansion states increased the number of patients with type 2 diabetes in their panels compared to non-expansion states.
METHODS
Survey Administration
Beginning in May of 2016 and concluding in August of 2016, we mailed three waves of a 55-question survey about type 2 diabetes care practices to 1,200 U.S. endocrinologists and primary care physicians (PCPs). We obtained addresses and practice areas through the 2016 AMA Physician Masterfile, and included physicians from 49 states and Washington D.C. (by chance, Mississippi was not represented). We stratified recipients by state expansion status and physician specialty. We mailed surveys to physicians in the 24 states that participated in Medicaid expansion (including Washington D.C.) (N=524), the 7 states that received Medicaid Expansion Waivers (N=71), and the 21 states that did not expand Medicaid (N=606). Forty percent of our sample (N=480) identified their practice area in the AMA Masterfile as “endocrinology” or “diabetes”; the remainder (N=720) listed their practice area as primary care. Upon receiving completed surveys, we excluded respondents who stated that they did not provide longitudinal outpatient care for patients with type 2 diabetes from our eligible sample (N=182/588 received). We attempted to reduce non-response bias by conducting three mailed survey waves, including a $10 bill in the initial survey wave, and a postcard prior to the third wave. Previous studies have shown that these methods increase response rates. [12] This study was deemed exempt by the institutional review board at the University of Chicago Medicine.
Survey Development
The 55-question, 16-page survey gathered data on physician’s attitudes and observations in a variety of salient areas in diabetes care, including medication management and glycemic goal individualization. The data presented in this paper focus on three questions that sought to assess the impact of Medicaid expansion on the provision of diabetes care in the longitudinal outpatient setting. We included the complete survey as an online-only supplementary appendix.
Questions related to access to care were on page 11 of the 16-page survey. To assess changes in patient panels, we asked: 1) In the last two years, have you had an increase in the number of Medicaid or newly insured patients with diabetes in your practice? If “yes”, then in the last two years, 2) how many of these Medicaid or newly insured patients did you newly diagnose with diabetes?, and 3) how many of these Medicaid or newly insured patients were previously diagnosed with diabetes but not receiving care? The latter two questions had the following response options: 1-10, 11-25, 26-50, 51-100, 101-200, or >200 patients. To avoid biasing respondents, we did not mention the ACA in the survey. Because these survey questions were novel, they have not been externally validated.
Covariates
The primary covariate was practice state, categorized as an expansion, waiver, or non-expansion state based on 2016 status. The 24 expansion states included Alaska, California, Colorado, Connecticut, Delaware, Illinois, Kentucky, Louisiana, Maryland, Massachusetts, Minnesota, Nevada, New Jersey, New Mexico, New York, North Dakota, Ohio, Oregon, Pennsylvania, Rhode Island, Vermont, West Virginia, Washington state and Washington D.C. The 7 waiver states included Arizona, Arkansas, Iowa, Indiana, Michigan, Montana and New Hampshire. The remaining 20 states were considered non-expansion states.
Other covariates included age, gender, race/ethnicity, years in practice, specialty, professional activity, burnout, [13] practice setting, size, insurance mix, and patient age and race/ethnicity. Because of the limited sample size, post hoc, we dichotomized years in practice (≥20 years vs. <20 years); professional activity (research vs. no research); physician burnout (some or more vs. none); practice size (>1000 patients vs. ≤1000 patients); percent patients age >65 years, percent non-Hispanic white patients, and percent managed care (>50% vs. ≤50%); and setting type (hospital-based clinic vs. other).
Sample Size Calculation
In 2010, there were an estimated 209,000 primary care physicians in the U.S.[14] In 2011, there were an estimated 4,841 adult endocrinologists practicing in the U.S.[15] We estimated that we would need a sample of 383 primary care physicians and 356 endocrinologists to ensure that our responses were nationally representative within a margin of error of 5% and 95% confidence level. Expecting a response rate of 60%, we estimated that we needed to survey 613 primary care physicians and 570 endocrinologists, or at least 1183 physicians. Of the 588 surveys we received, we considered 232 to be ineligible, including 116 that were marked as return to sender. 41% of presumed eligible recipients responded (356/868). Following the administration of the survey, newly available data found that in 2016, 289,831 primary care physicians and 7,236 endocrinologists practiced in the U.S. [16]
Analysis and statistical methods
To calculate the response rate, we used the Council of American Survey Research Organizations formula, which adjusts for the estimated number of ineligible responses among non-respondents.[17] We used chi-square tests to examine associations between covariates and physicians reporting “yes” vs. “no”/“unsure” regarding an increase in the number of Medicaid or newly insured patients with diabetes in their practice. We also used multivariate logistic regressions to examine marginal effects for covariates of physicians reporting >10 vs. ≤10 new patients with either newly diagnosed diabetes or previously diagnosed diabetes newly receiving care. The adjusted models included covariates with p ≤ 0.10 in bivariate analyses. We performed multivariate regressions for the entire sample and for subgroup analyses for primary care physicians and endocrinologists. We stratified analyses by state expansion status and physician specialty; we computed poststratification weights to the 2016 U.S. physician population by the same measures. P-values <0.05 were considered statistically significant. We used the survey analysis procedures in SAS version 9.4 (SAS Institute Inc.) to perform analyses.
RESULTS
Our adjusted response rate was 41% (N=356) (Figure 1). Respondents and non-respondents did not differ by specialty type (p = 0.40), degree type (MD vs. DO, p = 0.49), or census region (p = 0.09). The response rate also did not differ between Medicaid expansion, waiver, and non-expansion states (p =0.47). Of the respondents, 142 (weighted percentage, 56.6%) were from expansion states, 25 (weighted percentage, 8.4%) were from waiver states, and 190 (weighted percentage, 35.0%) were from non-expansion states. Forty-six percent (n = 164) of respondents were endocrinologists; 193 (54%) were PCPs (Table 1).
Figure 1:
Respondent Flow Diagram
Table 1:
Physician and Practice Characteristics by Medicaid Expansion vs. Non-Expansion States
| Type of State | |||||||
|---|---|---|---|---|---|---|---|
| Overall | Non-expansion (N=188) | Expansion or Waiver (N=168) | |||||
| Characteristic | N | Weighted % (CI) | N | Weighted % (CI) | N | Weighted % (CI) | P-value |
| Provider characteristics | |||||||
| Female | 139 | 36.4 (28.0-44.8) | 69 | 34.9 (24.7-45.1) | 70 | 37.2 (25.6-48.8) | 0.77 |
| Age (years) | 0.79 | ||||||
| < 40 | 45 | 11.1 (5.7-16.4) | 24 | 7.9 (2.9-12.9) | 21 | 12.8 (5.0-20.7) | |
| 41-49 | 103 | 31.4 (23.4-39.4) | 58 | 33.1 (22.6-43.6) | 45 | 30.5 (19.5-41.5) | |
| 50-64 | 148 | 45.4 (36.4-54.4) | 82 | 46.4 (35.6-57.3) | 66 | 44.8 (32.2-57.4) | |
| ≥ 65 | 44 | 12.1 (7.2-17.0) | 22 | 12.6 (5.8-19.3) | 22 | 11.8 (5.2-18.4) | |
| Race/Ethnicity | 0.003 | ||||||
| Non-Hispanic White | 211 | 54.4 (45.7-63.2) | 110 | 55.7 (44.9-66.4) | 101 | 53.8 (41.7-65.8) | |
| Non-Hispanic Black | 19 | 6.5 (0.0-13.2) | 11 | 1.6 (0.0-3.6) | 8 | 9.2 (0.0-19.1) | |
| Hispanic/Latino | 32 | 9.9 (5.7-14.1) | 21 | 14.8 (7.4-22.3) | 11 | 7.2 (2.2-12.3) | |
| Asian | 73 | 23.4 (15.8-31.1) | 32 | 15.9 (8.2-23.6) | 41 | 27.5 (16.6-38.3) | |
| Other | 18 | 5.7 (2.1-9.3) | 14 | 12.0 (3.9-20.1) | 4 | 2.4 (0.0-5.6) | |
| Years in Practice | 0.93 | ||||||
| 1-9 | 97 | 18.0 (12.0-24.1) | 50 | 20.7 (12.2-29.2) | 47 | 16.6 (8.6-24.7) | |
| 10-19 | 79 | 23.0 (16.2-29.9) | 45 | 22.3 (13.3-31.3) | 34 | 23.4 (14.1-32.8) | |
| 20-29 | 97 | 33.0 (24.9-41.1) | 48 | 31.6 (21.1-42.1) | 49 | 33.7 (22.6-44.8) | |
| > 30 | 80 | 25.9 (17.6-34.2) | 45 | 25.4 (16.7-34.1) | 35 | 26.2 (14.4-38.0) | |
| Specialty | |||||||
| Endocrinology | 164 | 2.4 (2.2-2.7) | 87 | 2.2 (1.7-2.7) | 77 | 2.6 (1.9-3.3) | 0.45 |
| Family medicine | 101 | 48.2 (39.6-56.7) | 55 | 51.9 (41.2-62.6) | 46 | 46.2 (34.4-57.9) | 0.48 |
| Geriatrics | 15 | 5.5 (2.1-8.9) | 10 | 8.8 (2.2-15.4) | 5 | 3.8 (0.0-7.6) | 0.17 |
| Internal medicine | 103 | 48.2 (39.6-56.9) | 53 | 42.3 (31.7-53.0) | 50 | 51.4 (39.6-63.2) | 0.26 |
| Other | 11 | 4.6 (1.4-7.7) | 8 | 11.2 (2.8-19.5) | 3 | 1.0 (0.0-2.4) | <.001 |
| Burnout* | 0.22 | ||||||
| Some or more | 110 | 32.0 (24.0-40.0) | 52 | 26.2 (17.3-35.1) | 58 | 35.1 (23.9-46.4) | |
| None | 247 | 68.0 (60.0-76.0) | 138 | 73.8 (64.9-82.7) | 109 | 64.9 (53.6-76.1) | |
| Practice characteristics | |||||||
| Size (number of patients) | 0.57 | ||||||
| 1-500 | 79 | 24.9 (17.2-32.6) | 42 | 27.6 (17.6-37.7) | 37 | 23.4 (13.0-33.9) | |
| 501-1000 | 59 | 17.2 (9.1-25.3) | 26 | 11.6 (4.5-18.6) | 33 | 20.2 (8.5-31.8) | |
| 1001-2000 | 90 | 21.7 (15.0-28.4) | 50 | 22.8 (13.9-31.6) | 40 | 21.2 (12.1-30.3) | |
| > 2000 | 122 | 36.2 (28.1-44.3) | 69 | 38.0 (27.8-48.3) | 53 | 35.2 (24.2-46.2) | |
| Percent of Patients > 65 years old | 0.14 | ||||||
| 76-100% | 34 | 11.6 (6.8-16.4) | 21 | 16.7 (8.3-25.2) | 13 | 9.0 (3.3-14.7) | |
| 51-75% | 100 | 23.5 (16.6-30.4) | 56 | 29.6 (20.0-39.2) | 44 | 20.4 (11.2-29.6) | |
| 26-50% | 143 | 38.4 (29.5-47.2) | 73 | 30.0 (20.0-39.9) | 70 | 42.7 (30.5-54.8) | |
| 0-25% | 72 | 26.5 (19.0-34.1) | 36 | 23.8 (14.4-33.1) | 36 | 28.0 (17.6-38.3) | |
| Percent of non-Hispanic white patients | 0.71 | ||||||
| 76-100% | 86 | 26.5 (19.2-33.8) | 44 | 23.1 (13.5-32.6) | 42 | 28.3 (18.3-38.3) | |
| 51-75% | 108 | 25.7 (18.6-32.8) | 60 | 30.3 (20.7-39.9) | 48 | 23.3 (13.8-32.8) | |
| 26-50% | 108 | 28.3 (19.9-36.8) | 61 | 29.7 (20.0-39.4) | 47 | 27.7 (15.8-39.5) | |
| 0-25% | 46 | 19.4 (12.1-26.7) | 19 | 17.0 (8.6-25.3) | 27 | 20.7 (10.5-30.9) | |
| Location | |||||||
| Urban | 159 | 45.7 (37.0-54.4) | 80 | 44.5 (33.7-55.2) | 79 | 46.4 (34.4-58.4) | 0.81 |
| Rural | 149 | 36.9 (28.6-45.1) | 81 | 28.3 (19.2-37.4) | 68 | 41.5 (29.9-53.1) | 0.08 |
| Suburban | 63 | 22.2 (15.4-29.0) | 39 | 29.3 (19.5-39.0) | 24 | 18.4 (9.5-27.3) | 0.11 |
| Type of Setting | |||||||
| Private solo or group practice | 196 | 55.3 (46.7-64.0) | 109 | 56.2 (45.5-66.9) | 87 | 54.8 (42.9-66.8) | 0.87 |
| Freestanding clinic | 31 | 9.5 (5.0-13.9) | 22 | 14.3 (6.5-22.0) | 9 | 6.9 (1.5-12.2) | 0.11 |
| Hospital-based clinic | 98 | 19.3 (11.6-27.0) | 47 | 16.5 (8.8-24.3) | 51 | 20.9 (9.9-31.8) | 0.51 |
| Health maintenance organization or other prepaid practice | 18 | 6.8 (2.9-10.6) | 4 | 2.9 (0.0-6.2) | 14 | 8.9 (3.2-14.6) | 0.08 |
| Other | 30 | 14.4 (8.3-20.5) | 15 | 15.8 (6.9-24.8) | 15 | 13.6 (5.6-21.6) | 0.72 |
| Percent private insurance | 0.75 | ||||||
| 76-100% | 14 | 3.5 (0.7-6.2) | 5 | 3.3 (0.0-7.8) | 9 | 3.6 (0.1-7.0) | |
| 51-75% | 92 | 22.0 (15.0-28.9) | 48 | 19.0 (10.9-27.2) | 44 | 23.5 (13.8-33.2) | |
| 26-50% | 135 | 35.8 (27.3-44.3) | 73 | 33.4 (23.2-43.6) | 62 | 37.1 (25.3-49.0) | |
| 0-25% | 97 | 38.8 (29.8-47.8) | 54 | 44.4 (33.4-55.3) | 43 | 35.8 (23.2-48.4) | |
| Percent fee-for-service | 0.04 | ||||||
| 76-100% | 194 | 43.4 (35.0-51.7) | 117 | 50.3 (39.6-61.0) | 77 | 39.6 (28.1-51.0) | |
| 51-75% | 23 | 8.3 (4.1-12.4) | 8 | 4.7 (0.6-8.8) | 15 | 10.2 (4.2-16.2) | |
| 26-50% | 23 | 8.2 (3.4-13.0) | 5 | 1.4 (0.0-3.5) | 18 | 11.8 (4.6-19.1) | |
| 0-25% | 51 | 18.3 (11.9-24.6) | 26 | 23.3 (13.9-32.8) | 25 | 15.6 (7.3-23.8) | |
| Missing | 66 | 21.9 (13.9-30.0) | 34 | 20.2 (11.4-29.1) | 32 | 22.8 (11.5-34.2) | |
| Percent managed care | 0.68 | ||||||
| 76-100% | 53 | 17.6 (10.8-24.4) | 29 | 13.0 (5.8-20.1) | 24 | 20.0 (10.4-29.6) | |
| 51-75% | 66 | 17.8 (11.2-24.3) | 32 | 17.6 (8.7-26.5) | 34 | 17.9 (9.2-26.6) | |
| 26-50% | 91 | 30.5 (21.9-39.1) | 47 | 34.8 (24.2-45.5) | 44 | 28.3 (16.4-40.1) | |
| 0-25% | 117 | 34.1 (25.6-42.7) | 65 | 34.6 (24.0-45.2) | 52 | 33.9 (22.2-45.5) | |
| “In the last 2 years, have you had an increase in the number of Medicaid or newly insured patients with diabetes in your practice?” | 0.007 | ||||||
| Yes | 174 | 53.1 (44.7-61.6) | 78 | 39.9 (29.6-50.2) | 96 | 60.2 (48.7-71.7) | |
| No | 136 | 36.4 (28.3-44.6) | 80 | 39.5 (28.9-50.1) | 56 | 34.8 (23.7-45.9) | |
| Unsure | 47 | 10.5 (5.7-15.2) | 32 | 20.5 (11.7-29.3) | 15 | 5.0 (0.0-10.6) | |
| “In the last 2 years, how many of these Medicaid or newly insured patients did you newly diagnose with diabetes?”† | 0.04 | ||||||
| 0 | 183 | 46.9 (38.4-55.3) | 112 | 60.1 (49.8-70.4) | 71 | 39.8 (28.3-51.3) | |
| 1-10 | 73 | 18.7 (11.0-26.3) | 31 | 10.3 (4.6-16.1) | 42 | 23.2 (12.1-34.2) | |
| 11-25 | 46 | 14.3 (8.6-20.1) | 24 | 14.5 (7.1-21.9) | 22 | 14.2 (6.3-22.1) | |
| >25 | 51 | 19.3 (12.7-25.8) | 21 | 13.9 (6.7-21.1) | 30 | 22.1 (12.9-31.4) | |
| Missing | 4 | 0.9 (0.0-2.0) | 2 | 1.2 (0.0-3.5) | 2 | 0.7 (0.0-2.0) | |
| “In the last 2 years, how many of these Medicaid or newly insured patients were previously diagnosed with diabetes but not receiving care?”† | 0.04 | ||||||
| 0 | 183 | 46.9 (38.4-55.3) | 112 | 60.1 (49.8-70.4) | 71 | 39.8 (28.3-51.3) | |
| 1-10 | 65 | 20.8 (14.1-27.5) | 29 | 17.4 (9.7-25.2) | 36 | 22.6 (13.2-32.0) | |
| 11-25 | 57 | 20.9 (12.9-28.9) | 23 | 12.5 (5.5-19.4) | 34 | 25.4 (13.9-36.9) | |
| >25 | 50 | 11.0 (6.3-15.8) | 24 | 8.9 (3.3-14.4) | 26 | 12.2 (5.5-18.9) | |
| Missing | 2 | 0.4 (0.0-1.2) | 2 | 1.2 (0.0-3.5) | 0 | 0.0 (0.0-0.0) | |
Burnout: Some or more = “I am definitely burning out and have one or more symptoms of burnout, e.g. emotional exhaustion” or “The symptoms of burnout that I’m experiencing won’t go away. I think about work frustrations a lot” or “I feel completely burned out. I am at the point where I may need to seek help.”
None = “I enjoy my work. I have no symptoms of burnout” or “I am under stress and don’t always have as much energy as I did, but I don’t feel burned out.”
Only respondents who answered “Yes” to the question “In the last two years, have you had an increase in the number of Medicaid or newly insured patients with diabetes in your practice?” were asked to answer these questions. N’s represent the number of eligible patients stratified by non-expansion vs. expansion state. Response options included 1-10, 11-25, 26-50, 51-100, 101-200 or >200 patients.
Overall, 53.1% of physicians reported an increase in Medicaid or newly insured patients with diabetes (Table 1). A higher percentage of physicians in expansion or waiver states (60.2%) than non-expansion states (39.9%) reported such an increase (p = 0.01). Physicians from expansion and waiver states reported increases of 1-10 newly diagnosed patients in the previous two years (23.2%), 11-25 (14.2%) and > 25 (22.1%); physicians from non-expansion states reported fewer newly diagnosed patients (p = 0.04). Similarly, 22.6% of physicians from expansion and waiver states reported seeing 1-10 Medicaid or newly insured patients who had previously diagnosed diabetes that was untreated. One-quarter (25.4%) of physicians in the same states saw 11-25 previously diagnosed yet untreated patients.
Bivariate analyses of an increase in Medicaid or newly insured patients with diabetes by physician and practice characteristics revealed differences by state status, physician race/ethnicity, and urbanicity (Table 2). In adjusted multivariate logistic regression analyses, 64.2% of physicians who practice in an expansion state report an increase in Medicaid or newly insured patients with diabetes compared with 46.1% who practice in non-expansion states (p = 0.05; Table 2). Non-white physicians (69.6%) and those who practice in rural areas (68.3%) reported significant increases in the outcome measure compared with approximately 40% of white (p = 0.002) and physicians in urban and suburban areas (p = 0.004), respectively. The other physician and practice characteristics were not significantly associated with increase in Medicaid or newly insured patients with diabetes. In similar analyses stratified by PCPs and endocrinologists, the PCPs were nearly identical to the overall population of physicians (Table 3). However, approximately 60% of endocrinologists saw increases in these patients in larger practices (p = 0.003) and practices in urban (rather than rural) areas (p < .001). As shown in Tables 2 and 3, waiver states considered expansion states increased the difference in reported increases between expansion and non-expansion states, but the number of waiver states was too small to analyze as a separate group.
Table 2.
Percentage of physicians reporting increase in Medicaid or newly insured patients with type 2 diabetes.
| Unadjusted Weighted % (CI) | p-value | Adjusted Weighted % (CI) | p-value | |
|---|---|---|---|---|
| Expansion+waiver states | 0.01 | |||
| Yes | 60.2 (48.7-71.7) | --- | ||
| No | 39.9 (29.6-50.2) | --- | ||
| Expansion states | 0.06 | 0.05 | ||
| Yes | 60.0 (47.2-72.8) | 64.2 (50.2-76.1) | ||
| No | 44.1 (34.5-53.8) | 46.1 (33.0-59.7) | ||
| Physician characteristics | ||||
| Specialty | 0.26 | |||
| PCP | 53.3 (44.6-62.0) | --- | ||
| Endocrinologist | 46.1 (37.0-55.2) | --- | ||
| Gender | 0.25 | |||
| Female | 46.6 (32.2-61.0) | --- | ||
| Male | 56.8 (46.6-67.1) | --- | ||
| Age group | 0.76 | |||
| < 40 | 45.4 (19.5-71.3) | --- | ||
| 41-49 | 47.0 (32.1-61.8) | --- | ||
| 50-64 | 56.1 (42.2-70.0) | --- | ||
| ≥ 65 | 54.2 (33.8-74.6) | --- | ||
| Race/Ethnicity | 0.01 | 0.002 | ||
| Non-Hispanic White | 42.8 (32.4-53.2) | 40.1 (28.0-53.5) | ||
| Other race/ethnicity | 65.4 (52.7-78.1) | 69.6 (55.6-80.7) | ||
| Years in Practice | 0.54 | |||
| < 20 years | 49.3 (36.8-61.7) | --- | ||
| ≥ 20 years | 54.6 (42.9-66.3) | --- | ||
| Burnout | 0.18 | |||
| None | 49.2 (38.8-59.7) | --- | ||
| Some or more | 61.4 (47.3-75.5) | --- | ||
| Practice characteristics | ||||
| Size (number of patients) | 0.08 | 0.11 | ||
| ≤ 1000 | 44.8 (29.3-60.2) | 47.2 (33.3-61.6) | ||
| > 1000 | 61.6 (51.8-71.5) | 63.2 (48.9-75.4) | ||
| Percent of Patients > 65 years old | 0.10 | 0.54 | ||
| ≤ 50% | 59.0 (48.0-69.9) | 58.3 (44.8-70.6) | ||
| > 50% | 45.0 (32.1-57.9) | 52.3 (37.4-66.8) | ||
| Percent of non-Hispanic white patients | 0.33 | |||
| ≤ 50% | 50.9 (40.0-61.9) | --- | ||
| > 50% | 59.3 (46.2-72.4) | --- | ||
| Location | 0.04 | 0.004 | ||
| Rural | 61.6 (51.6-71.5) | 68.3 (55.9-78.6) | ||
| Urban/Suburban | 43.1 (29.0-57.2) | 41.6 (27.9-56.6) | ||
| Setting | 0.56 | |||
| Hospital-based clinic | 47.1 (23.8-70.4) | --- | ||
| Other | 54.6 (45.6-63.5) | --- | ||
| Percent private insurance | 0.89 | |||
| 76-100% | 48.3 (8.1-88.5) | --- | ||
| 51-75% | 55.2 (38.0-72.5) | --- | ||
| 26-50% | 59.3 (44.5-74.1) | --- | ||
| 0-25% | 52.0 (37.0-66.9) | --- | ||
| Percent fee-for-service | 0.10 | 0.26 | ||
| 76-100% | 51.3 (38.9-63.6) | 62.8 (40.9-80.4) | ||
| 51-75% | 67.7 (44.1-91.3) | 35.5 (20.6-54.0) | ||
| 26-50% | 74.3 (50.5-98.1) | 63.6 (34.4-85.3) | ||
| 0-25% | 34.0 (16.0-52.0) | 65.6 (34.6-87.3) | ||
| Unknown | 59.3 (39.6-78.9) | 48.5 (34.1-63.0) | ||
| Percent managed care | 0.92 | |||
| ≤ 50% | 54.4 (43.1-65.6) | --- | ||
| > 50% | 53.4 (39.0-67.8) | --- | ||
Adjusted models only included variables with bivariate p ≤ 0.10.
Table 3.
Percentage of physicians reporting an increase in Medicaid or newly insured patients with type 2 diabetes by specialty type.
| Primary Care | Endocrinology | |||||||
|---|---|---|---|---|---|---|---|---|
| Unadjusted Weighted % (CI) | p-value | Adjusted Weighted % (CI) | p-value | Unadjusted Weighted % (CI) | p-value | Adjusted Weighted % (CI) | p-value | |
| Expansion+waiver states | 0.01 | 0.30 | ||||||
| Yes | 60.5 (48.7-72.3) | --- | 48.8 (36.5-61.1) | --- | ||||
| No | 39.9 (29.4-50.4) | --- | 40.1 (29.3-51.0) | --- | ||||
| Expansion states | 0.06 | 0.06 | 0.15 | |||||
| Yes | 60.2 (47.1-73.4) | 64.3 (49.9-76.6) | 50.9 (38.0-63.9) | --- | ||||
| No | 44.3 (34.4-54.1) | 46.0 (32.5-60.0) | 38.4 (27.1-49.7) | --- | ||||
| Physician characteristics | ||||||||
| Gender | 0.25 | 0.54 | ||||||
| Female | 46.5 (31.6-61.4) | --- | 49.3 (35.9-62.8) | --- | ||||
| Male | 57.1 (46.7-67.6) | --- | 43.7 (31.4-55.9) | --- | ||||
| Age group | 0.75 | 0.90 | ||||||
| < 40 | 45.1 (18.3-71.8) | --- | 54.5 (31.5-77.5) | --- | ||||
| 41-49 | 47.0 (31.8-62.1) | --- | 46.6 (29.6-63.6) | --- | ||||
| 50-64 | 56.3 (42.1-70.6) | --- | 45.3 (30.7-59.9) | --- | ||||
| ≥ 65 | 54.6 (33.6-75.5) | --- | 41.6 (15.0-68.1) | --- | ||||
| Race/Ethnicity | 0.01 | 0.002 | 0.83 | |||||
| Non-Hispanic White | 42.7 (32.0-53.5) | 39.5 (27.1-53.3) | 45.8 (33.7-57.9) | --- | ||||
| Other race/ethnicity | 65.8 (52.9-78.7) | 70.2 (55.8-81.4) | 47.8 (33.7-61.9) | --- | ||||
| Years in Practice | 0.55 | 0.60 | ||||||
| < 20 years | 49.4 (36.6-62.3) | --- | 44.7 (32.3-57.1) | --- | ||||
| ≥ 20 years | 54.7 (42.8-66.6) | --- | 49.7 (35.6-63.8) | --- | ||||
| Burnout | 0.19 | 0.12 | ||||||
| None | 49.4 (38.8-60.1) | --- | 41.4 (30.6-52.2) | --- | ||||
| Some or more | 61.5 (47.0-76.0) | --- | 56.4 (40.7-72.0) | --- | ||||
| Practice characteristics | ||||||||
| Size (number of patients) | 0.08 | 0.13 | 0.05 | 0.003 | ||||
| ≤ 1000 | 45.0 (29.1-60.9) | 47.3 (32.9-62.2) | 35.8 (22.6-49.1) | 31.7 (20.4-45.6) | ||||
| > 1000 | 61.8 (51.7-71.9) | 63.0 (48.4-75.6) | 54.5 (42.4-66.6) | 59.5 (47.6-70.4) | ||||
| Percent of Patients > 65 years old | 0.10 | 0.56 | 0.59 | |||||
| ≤ 50% | 59.3 (48.1-70.5) | 58.3 (44.4-70.9) | 47.5 (36.0-58.9) | --- | ||||
| > 50% | 45.1 (31.9-58.2) | 52.3 (37.0-67.2) | 42.2 (26.9-57.6) | --- | ||||
| Percent of non-Hispanic white patients | 0.36 | 0.05 | 0.07 | |||||
| ≤ 50% | 51.3 (40.1-62.4) | --- | 37.7 (25.4-50.0) | 36.9 (25.1-50.4) | ||||
| > 50% | 59.4 (46.0-72.9) | --- | 56.2 (42.9-69.5) | 53.9 (40.7-66.6) | ||||
| Location | 0.03 | 0.003 | 0.003 | 0.0007 | ||||
| Rural | 62.2 (52.0-72.4) | 69.3 (56.5-79.7) | 33.5 (22.0-45.0) | 30.4 (19.8-43.6) | ||||
| Urban/Suburban | 42.7 (28.2-57.2) | 40.4 (26.7-55.9) | 58.6 (46.3-70.9) | 61.0 (48.6-72.1) | ||||
| Setting | 0.55 | 0.20 | ||||||
| Hospital-based clinic | 46.8 (22.1-71.4) | --- | 53.2 (38.0-68.4) | --- | ||||
| Other | 54.8 (45.7-63.9) | --- | 40.9 (29.9-51.9) | --- | ||||
| Percent private insurance | 0.88 | 0.47 | ||||||
| 76-100% | 47.7 (6.3-89.2) | --- | 67.6 (32.5-100) | --- | ||||
| 51-75% | 55.8 (38.0-73.6) | --- | 37.6 (21.6-53.6) | --- | ||||
| 26-50% | 59.6 (44.3-74.9) | --- | 51.6 (37.7-65.4) | --- | ||||
| 0-25% | 52.0 (36.8-67.3) | --- | 47.1 (24.8-69.4) | --- | ||||
| Percent fee-for-service | 0.10 | 0.26 | --- | |||||
| 76-100% | 51.6 (38.9-64.3) | 48.8 (33.8-64.0) | 41.3 (30.0-52.6) | --- | ||||
| 51-75% | 67.9 (44.2-91.7) | 66.1 (34.4-87.8) | 46.9 (0.0-96.8) | --- | ||||
| 26-50% | 73.8 (49.6-98.1) | 62.6 (33.4-84.8) | 100 (100-100) | --- | ||||
| 0-25% | 33.7 (15.3-52.0) | 34.9 (19.8-53.8) | 55.1 (27.4-82.9) | --- | ||||
| Unknown | 59.8 (39.8-79.8) | 63.5 (40.9-81.5) | 38.5 (17.8-59.2) | --- | ||||
| Percent managed care | 0.89 | 0.12 | ||||||
| ≤ 50% | 54.6 (43.1-66.2) | --- | 42.8 (30.7-55.0) | --- | ||||
| > 50% | 53.3 (38.5-68.1) | --- | 57.8 (43.2-72.3) | --- | ||||
Adjusted models only included ACA expansion state and variables that were significantly related (p≤0.05) to the outcome variable in unadjusted models.
--- Not available
DISCUSSION
Compared with physicians who practice in non-expansion states, physicians who practice in states that participated in the Medicaid expansion —including those who practice in waiver states — are more likely to report an increase in the number of Medicaid or newly insured patients with diabetes in their practice since 2014. These findings provide evidence that Medicaid expansion was associated with physicians empaneling new patients with diabetes and that this change in clinical practice was broadly enough applied, such that it was identifiable in this relatively small survey.
Our survey’s findings correspond to other assessments of health insurance’s impact on access to care for patients with diabetes. Analysis of the 2008 Oregon Health Insurance Experiment showed that increased Medicaid coverage was associated with increased rates of diabetes detection and medication use.[18] An early study of three states found that people with diabetes who resided in Kentucky, an expansion state, were more likely to have glucose-testing than in Texas, a non-expansion state, or Arkansas, which expanded private insurance.[7] Laboratory and pharmacy claims likewise suggest that newly identified diabetes based on HbA1c values >6.4%, ICD-9 diagnosis code of 250.x (diabetes), and diabetes drug prescriptions increased in Medicaid expansion states.[8,9] In addition, health insurance coverage expanded among patients with diabetes after the ACA, specifically among adults age 18-64 years.[5] Our study expands on this literature by suggesting that Medicaid expansion was associated with physicians increasing the number of patients in their practices with diabetes. Further, our study indicates that this increase was broadly distributed across practices, signaling that providers in expansion states were willing to empanel patients on Medicaid at an elastic rate. This may be significant in light of the historical reluctance of physicians to accept Medicaid’s lower reimbursement rates. [11]
Two other groups of physicians — non-white physicians and those who practice in rural areas — were significantly associated with increases in the number of Medicaid or newly insured patients with diabetes. Previous research has demonstrated that black and Hispanic physicians are more likely to care for patients of concordant racial and ethnic background, as well as greater numbers of uninsured and Medicaid patients. [19] Given the higher type 2 diabetes disease burden in communities of color, [20] we reasonably anticipated a disproportionate increase in the patient panels of non-white physicians. Similarly, patients who reside in rural areas are more likely to be on Medicaid, and the Medicaid expansion had a disproportionate impact on patients in rural areas, increasing coverage rates from 21% to 26%. [21]
For patients with diabetes, having health insurance coverage may be especially significant for their long-term health. Uninsured individuals with diabetes may be more than 2.5 times as likely to die as those with private insurance. [3] Earlier access to treatment may lead to earlier improvements in glycemic control, which could yield improved outcomes for decades. [18] Unfortunately, states with higher rates of undiagnosed diabetes were less likely to expand Medicaid, potentially compounding decades-long systemic disparities in diabetes care quality and serving as a negative confounder in our study. [4] Paradoxically, if efforts to repeal or weaken insurance market protections for patients with pre-existing conditions succeed, patients who have been diagnosed with diabetes because of Medicaid expansion may have greater restrictions for treatment than if Medicaid had never expanded, because these patients will now have a pre-existing medical condition.
Our response rate, while robust for a national physician survey, nevertheless introduces potential for non-response bias and was lower than we expected. However, because the survey investigated general diabetes practices, the influence of non-response bias on these specific results may be diminished. Further, respondents did not differ from non-respondents in terms of specialty type, degree type (DO vs. MD) or census region, though we did not have other specific demographics from non-respondents to analyze whether they differed from respondents otherwise. Although we intentionally asked about both Medicaid and newly insured patients in order to discern changes associated with Medicaid expansion, our findings may be confounded by nearly complete overlap between non-expansion states and states that opted for federally facilitated marketplaces (n=12) or federally facilitated marketplaces with the state conducting plan management (n=6). Also, similar to all surveys, our results are subject to recall bias; however, this form of bias would be unlikely to be substantively different between physicians practicing in expansion states vs. non-expansions states, but it could affect the effect size of our findings.
Conclusion
Compared with physicians who practice in non-expansion states, physicians who practice in states that participated in the Medicaid expansion —including those who practice in waiver states — are more likely to report an increase in the number of Medicaid or newly insured patients with diabetes in their practice since 2014. Given that early detection and regular treatment is associated with improved outcomes, the increased access associated with Medicaid expansion may improve long-term outcomes for patients with type 2 diabetes.
ACKNOWLEDGEMENTS
Contributors: N.L. is the guarantor of the work and takes responsibility for the work as a whole. N.L., E.S.H, R.M.S, M.T.Q, and A.G.N. designed the study and obtained funding. M.A.P., A.G.N. and N.L., contributed to the data collection. N.L. and S.A.H analyzed the data. M.A.P., N.L., S.A.H., E.S.H, R.M.S, M.T.Q, and A.G.N interpreted the data. M.A.P. drafted the manuscript and N.L., S.A.H., E.S.H, R.M.S, M.T.Q, and A.G.N critically reviewed/edited the manuscript.
Funding:
N.L. was supported by a NIDDK K23 DK092783 and is supported by the American Diabetes Association (1-18-JDF-037). E.S.H is supported by a NIDDK K24 DK105340. N.L., A.G.N., R.M.S., M.T.Q., and E.S.H. are members of the NIDDK Chicago Center for Diabetes Translation Research (CCDTR) at the University of Chicago (P30 DK092949). R.M.S. is supported by the American Diabetes Association (1-17-JDF-033) and by NIEHS (R01 ES028879). M.A.P. was supported by an Innovation Funding grant from the Pritzker School of Medicine. This work was also supported by a pilot award from the CCDTR and the University of Chicago Biological Sciences Division.
Footnotes
Prior Presentations: This study was presented in abstract form at the American Diabetes Association Annual Meeting 2017, San Diego, CA and previously published in abstract form in Diabetes 2017. 66:Supplement 1.
CONFLICT OF INTEREST
None of the authors have any potential conflicts of interest, financial or otherwise.
References
- 1.Dieleman JL, Baral R, Birger M et al. (2016). US Spending on Personal Health Care and Public Health, 1996-2013. JAMA, 316(24), 2627–2646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Colagiuri S, Cull CA, & Holman RR (2002). Are Lower Fasting Plasma Glucose Levels at Diagnosis of Type 2 Diabetes Associated With Improved Outcomes?: U.K. Prospective Diabetes Study 61. Diabetes Care, 25(8), 1410–1417. [DOI] [PubMed] [Google Scholar]
- 3.Bittoni MA, Wexler R, Spees CK, Clinton SK, & Taylor CA (2015). Lack of private health insurance is associated with higher mortality from cancer and other chronic diseases, poor diet quality, and inflammatory biomarkers in the United States. Preventive Medicine, 81, 420–426. [DOI] [PubMed] [Google Scholar]
- 4.Myerson R, & Laiteerapong N (2016). The Affordable Care Act and Diabetes Diagnosis and Care: Exploring the Potential Impacts. Current Diabetes Reports, 16(4) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Casagrande SS, Mcewen LN, & Herman WH (2018). Changes in Health Insurance Coverage Under the Affordable Care Act: A National Sample of U.S. Adults With Diabetes, 2009 and 2016. Diabetes Care, 41(5), 956–962. [DOI] [PubMed] [Google Scholar]
- 6.Centers for Disease Control and Prevention. National Diabetes Statistics Report, 2017. Atlanta, GA: Centers for Disease Control and Prevention, US Department of Health and Human Services; 2017. [Google Scholar]
- 7.Sommers BD, Blendon RJ, Orav EJ, & Epstein AM (2016). Changes in Utilization and Health Among Low-Income Adults After Medicaid Expansion or Expanded Private Insurance. JAMA Internal Medicine, 176(10), 1501. [DOI] [PubMed] [Google Scholar]
- 8.Kaufman HW, Chen Z, Fonseca VA, & Mcphaul MJ (2015). Surge in Newly Identified Diabetes Among Medicaid Patients in 2014 Within Medicaid Expansion States Under the Affordable Care Act. Diabetes Care, 38(5), 833–837. [DOI] [PubMed] [Google Scholar]
- 9.Ghosh A, Simon K, & Sommers BD. Three-Year Impacts Of The Affordable Care Act: Improved Medical Care And Health Among Low-Income Adults. Health Affairs 36 (6) 1119–1128. [DOI] [PubMed] [Google Scholar]
- 10.Wherry LR, & Miller S (2016). Early Coverage, Access, Utilization, and Health Effects Associated With the Affordable Care Act Medicaid Expansions. Annals of Internal Medicine, 164(12), 795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Hing E, Decker SL, & Jamoom E. Acceptance of New Patients with public and private insurance by office-based physicians: United States, 2013 NCHS data brief, no 195. Hyattsville, MD: National Center for Health Statistics; 2015. [PubMed] [Google Scholar]
- 12.Edwards P, Roberts I, Clarke M et al. (2003). Methods to increase response rates to postal questionnaires. Cochrane Database of Systematic Reviews [DOI] [PubMed] [Google Scholar]
- 13.Rohland BM, Kruse GR, & Rohrer JE (2004). Validation of a single-item measure of burnout against the Maslach Burnout Inventory among physicians. Stress and Health, 20(2), 75–79. [Google Scholar]
- 14.The Agency for Health Care Quality and Research. The Number of Practicing Primary Care Physicians in the United States. (2014, October 02). Retrieved from https://www.ahrq.gov/research/findings/factsheets/primary/pcwork1/index.html
- 15.The Lewin Group. Endocrine Clinical Workforce: Supply and Demand Projections. (2014, June). Retrieved from https://www.endocrine.org/-/media/endosociety/files/advocacy-and-outreach/other-documents/2014-06-white-paper--endocrinology-workforce.pdf?la=en
- 16.2017 State Physician Workforce Data Report. (2017, November). Retrieved from https://members.aamc.org/eweb/upload/2017StatePhysicianWorkforceDataReport.pdf
- 17.Council of American Survey Research Organizations. CASRO On the Definition of Response Rates. 1982. Retrieved from https://c.ymcdn.com/sites/www.casro.org/resource/resmgr/docs/casro_on_definitions_of_resp.pdf
- 18.Baicker K, Taubman SL, Allen HL (2013) et al. The Oregon Experiment — Effects of Medicaid on Clinical Outcomes. New England Journal of Medicine, 368(18), 1713–1722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Komaromy M, Grumbach K, Drake M at al. (1996). The Role of Black and Hispanic Physicians in Providing Health Care for Underserved Populations. New England Journal of Medicine, 334(20), 1305–1310. [DOI] [PubMed] [Google Scholar]
- 20.Centers for Disease Control and Prevention. National Diabetes Statistics Report, 2017. Atlanta, GA:, U.S. Dept of Health and Human Services; (2017). Retrieved from https://www.cdc.gov/diabetes/pdfs/data/statistics/national-diabetes-statistics-report.pdf [Google Scholar]
- 21.Artiga S, & Garfield R (2018, April 19). The Role of Medicaid in Rural America. The Kaiser Family Foundation. Retrieved from https://www.kff.org/medicaid/issue-brief/the-role-of-medicaid-in-rural-america/ [Google Scholar]