Abstract
Objective
To understand differences in financial performance, quality performance, supplemental benefits provision, and enrollee composition between integrated and non‐integrated plans in the Medicare Advantage (MA) program.
Data Sources
We used data from the Center for Medicare and Medicaid Services for 2015–2017. We included 156 integrated MA plans (31 unique contracts) and 2096 non‐integrated MA plans (392 unique contracts).
Study Design
We estimated linear probably models for financial performance, quality performance, supplemental benefits provision, and enrollee composition with state fixed effects and contract random effects. We adjusted for county‐level market structure‐related factors, cost‐related factors, and demand‐related factors. Our primary independent variable was an indicator of plan‐provider integration.
Principal Findings
Integrated MA plans were associated with $19.4 (95% CI: 9.2, 29.7) and $16.6 (95% CI: 10.3, 22.9) higher Part C and Part D monthly premiums, but were associated with higher star quality ratings. There were no significant differences in revenues and plan payments per enrollee between integrated and non‐integrated MA plans. Integrated MA plans were associated with $40.5 (95% CI: −54.0, −26.9) lower non‐claims costs than non‐integrated MA plans. There was limited evidence that integrated MA plans provided more generous supplemental benefits than non‐integrated MA plans. Enrollment rates in integrated MA plans were particularly low among socially marginalized groups (3.4 [95% CI: −5.9, −1.0], 4.7 [95% CI: −8.5, −0.9], and 4.4 [95% CI: −6.4, −2.4] percentage points lower among non‐Hispanic Black, Medicare–Medicaid dual eligible, and the disabled).
Conclusions
Our findings suggest that integrated MA plans may achieve higher efficiency and quality, but these benefits may not be experienced by all beneficiaries due to disparities in enrollment. As these models continue to spread, it is critical to develop policies to ensure that MA enrollees have equal access to integrated plans.
Keywords: disparity, enrollee composition, financial performance, integration, Medicare Advantage, quality performance, supplemental benefits
What is known on this topic
Around a fifth of Medicare Advantage plans are vertically integrated with health systems.
Prior research found that integrated MA plans are associated with higher premiums and quality.
However, less is known about their financial and clinical performance, supplemental benefits provision, and enrollee composition.
What this study adds
We find that integrated plans were associated with $19.4 and $16.6 higher monthly premiums for Part C and D, but were also associated with higher star quality ratings.
While there were no differences in revenues and plan payments per enrollee, integrated plans were associated with $40.5 lower non‐claims costs than non‐integrated plans, suggesting operational efficiency.
Integrated plans did not necessarily provide more supplemental benefits than non‐integrated plans, but enrollment rates in integrated plans were comparatively low among socially marginalized groups.
1. INTRODUCTION
In a competitive health care environment, health systems in the United States (US) are frequently looking for ways to align financial incentives and to improve efficiency. To address the challenges of fragmentation and to improve profits, health systems are increasingly integrating their care. While many health systems engage in horizontal integration (in which providers acquire other providers), vertical integration of health care delivery and financing (in which providers and insurers offer their own health plan) is growing and has recently become a prominent feature of the US health care system. 1 , 2 , 3 , 4
While vertical integration has the potential to improve efficiency and quality, there are concerns about its adverse consequences. Proponents argue that the vertical integration of providers and insurers can lead to administrative and clinical integration, possibly achieving high efficiency and quality. 5 However, opponents suspect that profit motives are the driving force behind vertical integration 6 and believe that the vertical integration of providers and insurers may increase health care prices without necessarily improving quality. Evidence suggests that improvements in efficiency or quality due to vertical integration are difficult to detect. 7 , 8 , 9 , 10 However, there is evidence that health care prices and spending increased after providers become integrated with hospitals and/or health systems. 11 , 12 , 13 , 14 Vertical integration is becoming a prominent feature of the Medicare Advantage (MA) program, the alternative to traditional fee‐for‐service Medicare run by private insurers. As MA plans are paid on a capitated basis and receive additional bonus payments based on quality performance, there may be an increased incentive to provide integrated care across care settings in order to maintain or improve their enrollees' health. These capitated payments may also be a means by which health systems can more effectively take on the risks of care for their patient panels. 15 As such, health systems may create their own MA plans for their enrollees, or MA plans may acquire their own health care facilities. However, there is a paucity of literature on integrated plans in the MA program, partly due to limited data. Prior studies suggest that compared to non‐integrated MA plans, integrated MA plans provided higher quality care, but charge higher premiums. 16 Also, integrated MA plans did not provide more generous benefits than non‐integrated MA plans. 16 However, integrated MA plans acquired more diagnoses than non‐integrated MA plans, but there were no differences in mortality, 17 suggesting evidence of upcoding.
In this study, we aimed to comprehensively understand differences between integrated and non‐integrated plans in the MA program. Specifically, our main analyses were conducted to compare differences in financial performance, quality performance, supplemental benefits provision, and enrollee composition.
2. METHODS
2.1. Data sources
Our data came from multiple sources for 2015–2017. First, we used the MA Landscape data to identify MA plans. Second, we used the MA Medical Loss Ratio (MLR) Public Use File to obtain information on contract‐level financial performance. Since 2014, the Affordable Care Act (ACA) has required insurers to submit annual MLR reports to the Centers for Medicare and Medicaid Services (CMS). As insurers report the MLR reports at the contract level, the revenues and costs reported in the data are based on the revenues and costs for all plans offered under the same contract. Third, we used the MA Star Ratings data to obtain information on contract‐level quality performance. CMS has implemented a 5‐star quality rating system to encourage insurers to improve care quality while helping enrollees compare and choose high‐quality MA plans. The data report quality at different levels of aggregation using over 40 different quality measures. Fourth, we used the Plan Benefit Package (PBP) data to obtain information on plan‐specific provision of supplemental benefits. The PBP provides information on the set of benefits that an MA plan offers. The data are submitted to CMS for benefit analysis, marketing, and beneficiary communication purposes. Recently, CMS has used the data to review and approve all benefits yearly to ensure that MA plans do not discriminate against enrollees with poor health or those who incur financial losses. Fifth, we used the Medicare Master Beneficiary Summary File to obtain information on plan‐specific enrollment composition by demographic and socioeconomic status. Finally, we used the American Community Survey and the Area Health Resource File to obtain county‐level characteristics. We linked contract‐level information using unique contract identifiers, plan‐level information using a combination of contract and plan identifiers, and county‐level information using area identifiers.
2.2. Study population
Using the MA Landscape data, we first identified MA plans. We excluded MA plans without Part D benefits as they are different from MA plans offering Part D benefits. We excluded special needs plans as they serve different populations (Medicaid‐Medicare dual eligible and institutionalized beneficiaries) than typical MA plans. We excluded MA contracts with only a single plan or MA contracts with fewer than 2400 member months as the MLR data do not report their information. Finally, we excluded MA plans with missing information. Our unit of analysis is a plan‐year. However, it is notable that MA plans are provided across multi‐county services areas, and thus it is proper to aggregate market factors across the counties in which plans operate. Following prior research, 15 , 16 we aggregated all county‐level market variables into plan‐level variables through enrollment weighting. As plan characteristics are constant across counties in the service area, these variables were not recalculated at the aggregated level.
2.3. Integration status
To identify integrated plans in the MA program, we followed prior literature 17 and searched through the full listing of MA contracts to flag those that might have been associated with a health system between 2015 and 2017. We checked each plan's website, and publicly available news and regulatory documents to classify if a plan was owned by a health care system, or if the plan owned a health care system. We validated our list of integrated systems against similar lists created by the Agency for Health care Research and Quality and others. 18 While MA plans could integrate with both hospital systems and outpatient providers, we focused our analysis on systems that included at least one hospital.
2.4. Outcomes
We included four types of outcomes: financial performance, quality performance, supplemental benefits provision, and enrollee composition.
First, we constructed continuous variables measuring the following five types of financial performance at the contract level: adjusted MLR, revenues, claims costs, non‐claims costs, and quality improvement expenses. We measured adjusted MLR as the proportion of revenue spent on enrollee care including quality improvement activities. The ACA required MA contracts to maintain an MLR of 85% or more, indicating that 85% of premium revenues be used for enrollee care. For revenues, we included plan payments from CMS, Part C premiums, Part D premiums, rebates for Part B premium reduction, and rebates for Part D premium reduction. For claims costs, we included claims costs, costs for claims incurred in each year, and costs for liability and reserves for claims. For non‐claims costs, we included non‐claims costs, cost containment expenses not included in quality improvement expenses, all other claims adjustment expenses, direct sales, salaries, and benefits, agent and broker fees and commissions, and other general and administrative expenses. For quality improvement expenses, we included quality improvement expenses, costs for improvement in health outcomes, costs for activities to prevent hospital admission, costs for patient safety and medical errors, costs for wellness and health promotion activities, and costs for information technology expenses related to improving quality. All monetary outcomes were measured per month per member and were adjusted for inflation and presented in 2017 dollars using the gross domestic product.
Second, we constructed binary variables indicating whether MA contracts had 4‐star ratings or above and 3.5‐star ratings or below. We examined quality performance overall as well as by the following domain: screenings, tests, and vaccines, managing chronic conditions, member experience with health plan, and member complaints, and changes in the health plan's performance. For a list of the quality measures reported by CMS, see Table A.
Third, we constructed binary variables indicating of whether MA plans provided the following 18 types of supplemental benefits: preventive dental services, comprehensive dental services, eye exams, eyewear, hearing exams, hearing aids, health education, nutritional/dietary benefit, additional sessions of smoking and tobacco cessation counseling, fitness benefit, enhanced disease management, telemonitoring services, bathroom safety devices, counseling services, in‐home safety assessment, personal emergency response system, post‐discharge in‐home medication reconciliation, and readmission prevention.
Finally, we constructed continuous variables measuring the share of enrollee composition by the following demographic and socioeconomic status: female, race/ethnicity (non‐Hispanic White, non‐Hispanic Black, non‐Hispanic Asian/Pacific Islander, Hispanic, non‐Hispanic American Indian/ Alaskan Native, or others), Medicare–Medicaid dual eligibility, and presence of disability as determined by the Social Security Administration.
2.5. Statistical analysis
First, we reported unadjusted outcomes by integration status. Then, we followed prior research and estimated adjusted differences in outcomes. 15 , 16 For each outcome, we conducted a linear probability model while adjusting for the following county‐level characteristics: market structure‐related factors (percentage of other MA parent organizations that were integrated, MA penetration, MA Herfindahl–Hirschman Index, MA benchmarks), cost‐related factors (fee‐for‐service Medicare standardized per capita costs, average fee‐for‐service Medicare beneficiary risk score, percentage of those older than 65 years old, percentage of urban or rural residents, and number of hospital beds, medical doctors, specialist physicians, and primary care physicians per 1000 residents), and demand‐related factors (percentage of residents with incomes below poverty, median household income, percentages of adult residents with at least a high school diploma and with at least 4 years of college, and percentages of workers in manufacturing and in construction). We included state and year fixed effects. As a MA contract may offer multiple plans, we followed prior research and included contract random effects, 15 , 16 assuming that an independent variable has a fixed relationship with the outcome variable across all observations, but that this fixed effect may vary from one observation to another. As there may be a trade‐off between financial and quality performance, we conducted the analysis described above by limiting outcomes to financial performance and additionally adjusting for quality performance. Furthermore, we conducted several sensitivity analyses. First, findings for integrated plans may be driven by Kaiser Permanente, which may be an outlier due to unique geographic distribution and historical context in terms of integration. Thus, we performed the analysis described by excluding Kaiser Permanente from analysis. Second, integrated MA plans may be more likely to operate in counties with certain types of beneficiaries, possibly raising the need of adjusting for differences in county‐level demographic characteristics. Therefore, we performed the analysis by additionally controlling for county‐level demographic characteristics (female, race/ethnicity, Medicare–Medicaid dual eligibility, and presence of disability). As these variables were also used as an outcome, we conducted this analysis only for outcomes of financial performance, quality performance, and supplemental benefits provision. Third, contract‐level risk scores may be related to outcomes, especially for financial performance and quality performance, and thus we performed the analysis by additionally controlling for contract‐level risk scores. Finally, as plans that choose to and are able to vertically integrate are likely different than plans that do not integrate, these differences may affect outcomes of interest. To address potential selection bias, we followed prior research 16 and conducted the analysis described while accounting for a propensity score matching approach based on plans' propensity to be vertically integrated.
3. RESULTS
We included 156 integrated MA plans (31 unique contracts) and 2096 non‐integrated MA plans (392 unique contracts). We found that the number of integrated MA plans remained relatively constant over time (131, 138, and 140 integrated MA plans [30, 31, and 29 unique contracts] for 2015, 2016, and 2017). We also found that at least one integrated MA plan was offered in 20 states. Population characteristics are presented in Table B.
Our adjusted analysis shows that integrated MA plans were significantly associated with higher premiums (Table 1). Specifically, integrated MA plans were associated with $19.4 (95% CI: 9.1, 29.6) and $16.7 (95% CI: 10.4, 23.0) higher Part C and Part D monthly premiums. However, we do not find evidence that integrated MA plans were more likely to have higher adjusted MLR, revenues, and plan payments than non‐integrated MA plans. Integrated MA plans were associated with lower rebates for Part B and D premium reduction, but the magnitude of the difference was marginal.
TABLE 1.
Differences in financial performance between Medicare Advantage (MA) plans by integration status
| Outcomes | Unadjusted outcomes | Adjusted differences among integrated MA plans relative to non‐integrated MA plans (95% CI) | |
|---|---|---|---|
| Non‐integrated MA plans (N = 2096) | Integrated MA plans (N = 156) | ||
| Adjusted medical loss ratio, % | 91.4 (7.7) | 92.5 (5.4) | 1.6 (−0.9 to 4.2) |
| Revenues, $ PMPM | 950.7 (200.1) | 920.6 (106.0) | −31.3 (−82.1 to 19.5) |
| Plan payment | 834.0 (190.2) | 787.7 (91.1) | −34.8 (−77.2 to 7.6) |
| Part C premium | 19.4 (36.9) | 39.0 (22.5) | 19.4 (9.2–29.7)*** |
| Part D premium | 15.2 (18.6) | 35.9 (13.9) | 16.6 (10.3–22.9)*** |
| Rebate for Part B premium reduction | 1.3 (7.2) | 0.1 (0.2) | −0.8 (−1.4 to −0.2)* |
| Rebate for Part D premium reduction | 13.4 (13.5) | 9.6 (9.4) | −4.8 (−9.1 to −0.6)* |
| Claims costs, $ PMPM | 829.2 (182.4) | 833.2 (101.1) | 10.1 (−42.5 to 62.8) |
| Claims incurred | 817.5 (178.8) | 822.1 (100.7) | 14.0 (−38.6 to 66.6) |
| Liability and reserves for claims | 2.9 (10.5) | 1.2 (1.1) | −3.8 (−5.5 to −2.1)*** |
| Non‐claims costs, $ PMPM | 97.2 (59.6) | 69.9 (23.7) | −40.5 (−54.0 to −26.9)*** |
| Cost containment expenses not included in quality improvement expenses | 8.3 (6.8) | 7.6 (7.7) | −2.6 (−4.7 to −0.5)* |
| All other claims adjustment expenses | 10.6 (12.8) | 8.7 (6.4) | −2.0 (−7.0 to 3.0) |
| Direct sales, salaries, and benefits | 7.4 (12.4) | 5.8 (11.1) | −0.9 (−5.7 to 3.8) |
| Agent and broker fees and commissions | 12.3 (9.6) | 6.2 (4.5) | −7.1 (−9.6 to −4.5)*** |
| Other general and administrative expenses | 57.6 (49.1) | 41.3 (18.5) | −27.8 (−37.8 to −17.7)*** |
| Quality improvement expenses, $ PMPM | 15.7 (11.1) | 12.1 (7.9) | −7.6 (−10.8 to −4.5)*** |
| Improvement in health outcomes | 7.9 (7.0) | 6.0 (5.0) | −4.0 (−6.1 to −2.0)*** |
| Activities to prevent hospital admission | 1.8 (2.6) | 1.5 (1.5) | −1.1 (−1.6 to −0.6)*** |
| Patient safety and medical errors | 1.6 (1.9) | 0.8 (1.0) | −0.9 (−1.3 to −0.5)*** |
| Wellness and health promotion activities | 2.5 (2.7) | 2.7 (3.5) | −0.5 (−1.4 to 0.4) |
| IT expenses related to improving quality | 1.9 (2.9) | 1.0 (1.8) | −1.1 (−1.7 to −0.4)** |
Note: For each outcome, we reported the adjusted difference by conducting a linear probability model while adjusting for the following county‐level characteristics: market structure‐related factors (percentage of other MA parent organizations that were integrated, MA penetration, and MA Herfindahl–Hirschman Index), cost‐related factors (fee‐for‐service Medicare standardized per capita costs, average fee‐for‐service Medicare beneficiary risk score, percentage of those older than 65 years old, percentage of urban or rural residents, and number of hospital beds, medical doctors, specialist physicians, and primary care physicians per 1000 residents), and demand‐related factors (percentage of residents with incomes below poverty, median household income, percentages of adult residents with at least a high school diploma and with at least 4 years of college, and percentages of workers in manufacturing and in construction). We included state and year fixed effects. We also included contract random effects as a MA contract may offer multiple plans.
Abbreviations: CI, confidence interval; PMPM, per member per month.
p < 0.05;
p < 0.01;
p < 0.001.
We find that integrated MA plans were significantly associated with lower non‐claims costs and quality improvement expenses, but there were no or marginal differences in claims costs (Table 1). Specifically, integrated MA plans were associated with $40.5 (95% CI: −54.0, −26.9) lower non‐claims costs per member per month. This difference was attributable to other general and administrative expenses ($‐27.8 [95% CI: −37.8, −17.7]), agent and broker fees and commissions ($‐7.1 [95% CI: −9.6, −4.5]), and cost containment expenses not included in quality improvement expenses ($‐2.6 [95% CI: −4.7, −0.5]). Also, integrated MA plans were associated with $7.1 (95% CI: −9.6, −4.5) lower quality improvement expenses per member per month. This difference was driven by costs for improvement in health outcomes ($‐4.0 [95% CI: −6.1, −2.0]), activities to prevent hospital admission ($‐1.1 [95% CI: −1.6, −0.6]), information technology expenses related to improving quality ($‐1.1 [95% CI: −1.7, −0.4]), and patient safety and medical errors ($‐0.9 [95% CI: −1.3, −0.5]).
We find that integrated MA plans were significantly associated with higher quality ratings (Table 2). Specifically, integrated MA plans were associated with 34.9 (95% CI: 21.0, 48.7) higher percentage points in the likelihood of having 4 stars and above. This result was consistently found across all domains (27.6 [95% CI: 14.3, 40.8] percentage points for screenings, tests, and vaccines, 20.5 [95% CI: 5.2, 35.8] for managing chronic conditions, 40.7 [95% CI: 28.9, 52.5] for member experience with health plan, and 20.4 [95% CI: 8.3, 32.5] for member complaints and changes in the health plan's performance). Also, integrated MA plans were associated with 23.6. (95% CI: −36.0, −11.2) lower percentage points in the likelihood of having 3.5 stars and below. This result was consistently found across all domains except for managing chronic conditions (−15.2 [95% CI: −26.0, −4.3] percentage points for screenings, tests, and vaccines, −24.8 [95% CI: −33.4, −16.3] for member experience with health plan, and −9.9 [95% CI: −19.0, −0.7] for member complaints and changes in the health plan's performance).
TABLE 2.
Differences in quality of care between Medicare Advantage (MA) plans by integration status
| Outcomes | Unadjusted outcomes | Adjusted differences among integrated MA plans relative to non‐integrated MA plans (95% CI) | |
|---|---|---|---|
| Non‐integrated MA plans (N = 2096) | Integrated MA plans (N = 156) | ||
| Rating of 4 or more stars, % | |||
| Overall | 57.5 | 90.7 | 34.9 (21.0–48.7)*** |
| Screenings, tests, and vaccines | 60.4 | 82.2 | 27.6 (14.3–40.8)*** |
| Managing chronic conditions | 48.8 | 58.3 | 20.5 (5.2–35.8)** |
| Member experience with health plan | 55.2 | 97.1 | 40.7 (28.9–52.5)*** |
| Member complaints and changes in the plan's performance | 75.5 | 91.4 | 20.4 (8.3–32.5)** |
| Rating of less than 4 stars, % | |||
| Overall | 33.2 | 8.0 | −23.6 (−36.0 to −11.2)*** |
| Screenings, tests, and vaccines | 29.8 | 16.6 | −15.2 (−26.0 to −4.3)** |
| Managing chronic conditions | 41.4 | 40.3 | −8.7 (−23.2 to 5.8) |
| Member experience with health plan | 32.2 | 2.2 | −24.8 (−33.4 to −16.3)*** |
| Member complaints and changes in the plan's performance | 15.7 | 7.3 | −9.9 (−19.0 to −0.7)* |
Note: For each outcome, we reported the adjusted difference by conducting a linear probability model while adjusting for the following county‐level characteristics: market structure‐related factors (percentage of other MA parent organizations that were integrated, MA penetration, and MA Herfindahl–Hirschman Index), cost‐related factors (fee‐for‐service Medicare standardized per capita costs, average fee‐for‐service Medicare beneficiary risk score, percentage of those older than 65 years old, percentage of urban or rural residents, and number of hospital beds, medical doctors, specialist physicians, and primary care physicians per 1000 residents), and demand‐related factors (percentage of residents with incomes below poverty, median household income, percentages of adult residents with at least a high school diploma and with at least 4 years of college, and percentages of workers in manufacturing and in construction). We included state and year fixed effects. We also included contract random effects as a MA contract may offer multiple plans.
Abbreviations: CI, confidence interval.
p < 0.05;
p < 0.01;
p < 0.001.
It is notable that findings for financial performance remained similar even after adjusting for quality ratings (Table C). Specifically, integrated MA plans were associated with $18.2 (95% CI: 7.6, 28.8) and $14.8 (95% CI: 8.4, 21.2) higher Part C and Part D monthly premiums. Also, integrated MA plans were associated with $36.3 (95% CI: −49.7, −22.9) lower non‐claims costs per member per month.
We show limited evidence that integrated MA plans were likely to provide more supplemental benefits than non‐integrated MA plans (Table 3). Specifically, there were no significant differences in the provision of supplemental benefits for 13 types. However, integrated MA plans were associated with 15.5 (95% CI: 0.2, 30.9) and 11.0 (95% CI: 1.0, 21.0) higher percentage points in the likelihoods of providing supplemental benefits for enhanced disease management and readmission prevention. However, integrated MA plans were associated with 16.0 (95% CI: −27.5, −4.5), 10.1 (95% CI: −13.2, −7.1), and 1.6 (95% CI: −2.8, −0.4) lower percentage points in the likelihoods of providing supplemental benefits for eyewear, counseling services, and personal emergency response system were.
TABLE 3.
Differences in provision of supplemental benefits between Medicare Advantage (MA) plans by integration status
| Outcomes | Unadjusted outcomes | Adjusted differences among integrated MA plans relative to non‐integrated MA plans (95% CI) | |
|---|---|---|---|
| Non‐integrated MA plans (N = 2096) | Integrated MA plans (N = 156) | ||
| Preventive dental services | 73.7 | 85.2 | −1.4 (−14.4 to 11.5) |
| Comprehensive dental services | 47.2 | 40.5 | 10.7 (−4.2 to 25.5) |
| Eye exams | 95.6 | 90.9 | 3.0 (−5.8 to 11.9) |
| Eyewear | 68.1 | 80.6 | −16.0 (−27.5 to −4.5)** |
| Hearing exams | 70.0 | 82.6 | −11.0 (−24.9 to 2.9) |
| Hearing aids | 50.4 | 63.6 | 0.5 (−13.6 to 14.6) |
| Health education | 45.4 | 42.2 | 7.7 (−8.6 to 24.1) |
| Nutritional/dietary benefit | 8.8 | 28.9 | 14.0 (−0.6 to 28.7) |
| Additional sessions of smoking cessation counseling | 30.7 | 12.7 | −6.2 (−18.8 to 6.5) |
| Fitness benefit | 84.7 | 82.7 | 19.8 (−6.0 to 45.6) |
| Enhanced disease management | 9.1 | 37.5 | 15.5 (0.2–30.9)* |
| Telemonitoring services | 7.1 | 31.4 | 9.4 (−4.1 to 23.0) |
| Bathroom safety devices | 3.2 | 5.6 | 5.1 (−3.7 to 13.9) |
| Counseling services | 23.8 | 0.0 | −10.1 (−13.2 to −7.1)*** |
| In‐home safety assessment | 2.6 | 19.7 | 4.9 (−2.5 to 12.3) |
| Personal emergency response system | 1.0 | 0.0 | −1.6 (−2.8 to −0.4)* |
| Post‐discharge in‐home medication reconciliation | 5.1 | 6.5 | 11.6 (−0.7 to 23.9) |
| Readmission prevention | 0.9 | 7.7 | 11.0 (1.0–21.0)* |
Note: For each outcome, we reported the adjusted difference by conducting a linear probability model while adjusting for the following county‐level characteristics: market structure‐related factors (percentage of other MA parent organizations that were integrated, MA penetration, and MA Herfindahl–Hirschman Index), cost‐related factors (fee‐for‐service Medicare standardized per capita costs, average fee‐for‐service Medicare beneficiary risk score, percentage of those older than 65 years old, percentage of urban or rural residents, and number of hospital beds, medical doctors, specialist physicians, and primary care physicians per 1000 residents), and demand‐related factors (percentage of residents with incomes below poverty, median household income, percentages of adult residents with at least a high school diploma and with at least 4 years of college, and percentages of workers in manufacturing and in construction). We included state and year fixed effects. We also included contract random effects as a MA contract may offer multiple plans.
Abbreviation: CI, confidence interval.
p < 0.05;
p < 0.01;
p < 0.001.
We find that enrollment rates in integrated MA plans were particularly low among socially marginalized groups (Table 4). Specifically, integrated MA plans were associated with 6.8 (95% CI: 2.8, 10.9) higher percentage points in the adjusted rates of enrollment among non‐Hispanic White enrollees. However, integrated MA plans were associated with 3.4 (95% CI: −5.9, −1.0) lower percentage points in the adjusted rates of enrollment among non‐Hispanic Black. Furthermore, integrated MA plans were associated with 4.7 (95% CI: −8.5, −0.9) and 4.4 (95% CI: −6.4, −2.4) lower percentage points in the adjusted rates of enrollment among Medicare–Medicaid dual eligible and the disabled.
TABLE 4.
Differences in proportion of plan enrollment between Medicare Advantage (MA) plans by integration status
| Outcomes | Unadjusted outcomes | Adjusted differences among integrated MA plans relative to non‐integrated MA plans (95% CI) | |
|---|---|---|---|
| Non‐integrated MA plans (N = 2096) | Integrated MA plans (N = 156) | ||
| Female | 55.6 (4.6) | 55.9 (3.8) | 0.1 (−0.9 to 1.1) |
| Race/ethnicity | |||
| Non‐Hispanic White | 74.6 (22.0) | 91.7 (7.4) | 6.8 (2.8–10.9)*** |
| Non‐Hispanic Black | 10.8 (12.9) | 3.8 (5.1) | −3.4 (−5.9 to −1.0)** |
| Non‐Hispanic Asian/Pacific Islander | 3.2 (7.6) | 1.1 (2.5) | −1.4 (−2.8 to 0.1) |
| Hispanic | 9.3 (14.9) | 1.5 (2.9) | −2.0 (−4.8 to 0.7) |
| Non‐Hispanic American Indian/Alaska Native | 0.3 (0.6) | 0.1 (0.1) | −0.1 (−0.2 to −0.1)*** |
| Others | 1.9 (1.4) | 1.8 (1.0) | 0.2 (−0.1 to 0.5) |
| Medicare–Medicaid dual eligible | 14.1 (13.2) | 7.6 (4.9) | −4.7 (−8.5 to −0.9)* |
| Disabled | 14.5 (8.3) | 9.7 (5.4) | −4.4 (−6.4 to −2.4)*** |
Note: For each outcome, we reported the adjusted difference by conducting a linear probability model while adjusting for the following county‐level characteristics: market structure‐related factors (percentage of other MA parent organizations that were integrated, MA penetration, and MA Herfindahl–Hirschman Index), cost‐related factors (fee‐for‐service Medicare standardized per capita costs, average fee‐for‐service Medicare beneficiary risk score, percentage of those older than 65 years old, percentage of urban or rural residents, and number of hospital beds, medical doctors, specialist physicians, and primary care physicians per 1000 residents), and demand‐related factors (percentage of residents with incomes below poverty, median household income, percentages of adult residents with at least a high school diploma and with at least 4 years of college, and percentages of workers in manufacturing and in construction). We included state and year fixed effects. We also included contract random effects as a MA contract may offer multiple plans.
Abbreviation: CI, confidence interval.
p < 0.05;
p < 0.01;
p < 0.001.
These findings are robust to sensitivity analyses after excluding Kaiser Permanente, after additionally adjusting for county‐level demographic characteristics, after additionally adjusting for plan‐level risk scores, and after accounting for propensity score matching (Tables D–G). There were changes in findings for some outcomes after accounting for propensity score matching, but overall findings remained similar.
4. DISCUSSION
Our study presents five main findings. First, compared to non‐integrated MA plans, integrated MA plans charged higher premiums for Part C and Part D, but had higher star quality ratings. Second, there were no significant differences in revenues and plan payments per enrollee between integrated and non‐integrated MA plans. Third, integrated MA plans had lower non‐claims costs and quality improvement expenses than non‐integrated MA plans. Fourth, there was limited evidence that integrated MA plans provided more generous supplemental benefits than non‐integrated MA plans. Fifth, enrollment rates in integrated MA plans were particularly low among socially marginalized groups.
Our findings for higher premiums and quality ratings among integrated MA plans reflect a trade‐off between price and quality. Higher premiums among integrated MA plans may be partly explained by higher quality. Indeed, integrated MA plans had higher quality ratings than non‐integrated MA plans and this was consistently found in prior research. 15 , 16 , 19 However, we also found that integrated MA plans still had higher premiums even after accounting for quality ratings. This also aligns with prior research showing that about 70% of premium differences are not explained by differences in quality ratings. 16 There may be multiple explanations for this finding. First, higher premiums may be driven by increased market power, regardless of quality of care. Evidence suggests that health care prices increased after providers become integrated with hospitals and/or health systems. 11 , 12 , 13 , 14 Second, higher premiums may be related to quality performance not captured in the star rating system.
Our results provide suggestive evidence that integrated MA plans may not achieve higher profits than non‐integrated MA plans. We found that integrated MA plans had higher premiums than non‐integrated plans, but there were no significant differences in plan payments and revenues per enrollee. However, our findings should be interpreted with caution because there are discrepancies in findings from prior research. There is evidence of upcoding in MA plans to increase profits and this increases with vertical integration. 20 Also, integrated MA plans tended to acquire more diagnoses than non‐integrated MA plans, which could increase plan payments and revenues. 17 It is worth noting that our unadjusted analysis shows that integrated MA plans had lower revenues per enrollee than non‐integrated MA plans because plan payments were relatively low. As MA plans are paid on a capitated basis, lower plan payments may be attributable to enrollment of healthier enrollees. Furthermore, risk adjustment for plan payments is imperfect, leading to overpayments for healthier enrollees and underpayments for sicker enrollees. 21 These may indicate that indeed integrated MA plans produce profits, but this does not appear to happen on the MLR report.
Our findings for lower non‐claims costs among integrated MA plans indicate gains in administrative efficiency. Such improved efficiency may be achieved by multiple channels. First, integrated MA plans no longer face contracting or negotiating costs due to a combination of integration status and increased market power, leading to low transaction costs. Second, integration may avoid duplicative investments in information technology or quality improvement activities, leading to cost savings. However, we found that integrated MA plans had lower quality improvement expenses than non‐integrated plans. It is possible that integrated plans already established infrastructure to provide high quality of care, possibly leading to lower variable costs for quality improvement activities. Similarly, integrated MA plans may exert higher managerial efficiency, leading to higher quality at lower costs. However, we cannot rule out the possibility that integrated MA plans may strategically respond to the star rating system by only improving outcomes included in estimating star rating.
We find limited evidence that integrated MA plans were likely to provide more generous supplemental benefits than non‐integrated plans. Provision of supplemental benefits may be used to induce more enrollment, but our findings suggest that integrated MA plans may not use supplemental coverage as marketing strategies. This finding is consistent with prior research in that integrated MA plans did not provide generous benefits. 16 This suggests that MA enrollees in integrated plans may have little benefits from higher premiums. Notably, integrated plans were more likely to provide enhanced disease management and readmission prevention than non‐integrated plans. As these benefits may be effective at maintaining health while reducing preventable events, integrated plans are likely to adopt them as approaches to improving care coordination.
Our findings for low enrollment of socially marginalized groups in integrated MA plans raise concerns about disparities. Although we did not examine the driving factor for this finding, there may be two potential explanations. First, integrated MA plans may engage in patient selection. As payments to MA plans do not account for social risk factors, this may lead to systematic underpayments for socially marginalized enrollees creating an incentive to avoid enrolling socially marginalized enrollees. One way to achieve this is to avoid offering a plan in counties with a disproportionate share of socially marginalized groups. 22 Second, higher premiums in integrated MA plans may serve as barriers to enrollment. Evidence suggests that MA enrollees were more sensitive to financial dimensions of health insurance such as premiums. 23 , 24 Thus, higher premiums may induce socially marginalized enrollees not to choose integrated MA plans. This may be of concern if these integrated systems are capable of offering better coordination of care.
Our study has important policy implications. First, our findings provide new insights about integrated MA plans. Our estimates showed that about 20% of MA enrollees were enrolled in integrated MA plans. As integration of providers and insurers has been increased over time, 15 the number of MA enrollees in integrated MA plans is expected to increase. Moreover, the integration is highly heterogenous depending on the type of health care system the plans are integrated with. Thus, more resources are needed to understand how these arrangements are working. Furthermore, prior research suggests that integrated MA plans were associated with higher quality of care, 17 but there is still limited understanding on the underlying mechanism that may contribute to these quality differences. Thus, further research is warranted to examine whether integrated MA plans play a critical role in improving efficiency of care or health outcomes. Researchers and policymakers may benefit from increased public reporting on which plans are integrated and how they are integrated to facilitate future research on the subject. Furthermore, the disparity in enrollment in integrated systems is concerning as it may indicate that minority and low‐income beneficiaries may have less access to the potential benefits that these integrated systems may provide. Third, if these systems have been able to successfully reduce administrative costs as we find in this study, there may be potential for other MA plans to learn from how similar results can be achieved.
5. LIMITATIONS
Our study had several limitations. First, our method of identifying integrated contracts was subjective and potentially prone to error. We carefully identified integrated plans, but we may have unintentionally omitted other small health systems. Relatedly, we primarily focused on integrated systems that included a hospital system, which may differ than other types of outpatient and primary care integrated systems. Second, we relied on a cross‐sectional study design, and thus we could not be certain of the direction of the causal pathway of the association between integration status and outcomes of the interest. Relatedly, we could not account for selection of enrollees into integrated plans and selection of plans into an integrated model, which may affect our findings. Third, our outcomes are generally limited to indicators reported by plans or contracts and/or reported by the government and may not align with the constructs we intend to measure. For example, star ratings may not necessarily reflect true quality of care and financial data reported at the contract level may differ somewhat from financial performance of specific plans. Fourth, we only focused on MA plans with Part D benefits, and thus our findings are likely not generalizable to non‐drug plans, special need plans, or plans in the non‐MA program. Finally, our measures of financial performance are based on data reported by MA contracts, but we are unable to measure any differences in financial performance that may be occurring on the health system side of the balance sheet.
6. CONCLUSION
While integrated MA plans charged high premiums, we find evidence that integration of providers and insurers can lead to administrative and clinical integration, possibly achieving high efficiency and quality. However, integrated MA plans have important negative consequences by increasing disparities in enrollment. As these models continue to spread, it is critical to develop policies to ensure that MA enrollees have equal access to integrated plans.
Supporting information
Data S1. Supporting information.
ACKNOWLEDGMENT
None.
Park S, Langellier BA, Meyers DJ. Differences between integrated and non‐integrated plans in Medicare Advantage. Health Serv Res. 2023;58(3):560‐568. doi: 10.1111/1475-6773.14101
REFERENCES
- 1. Kocher R, Sahni NR. Hospitals' race to employ physicians — the logic behind a money‐losing proposition. N Engl J Med. 2011;364(19):1790‐1793. [DOI] [PubMed] [Google Scholar]
- 2. Capps C, Dranove D, Ody C. The effect of hospital acquisitions of physician practices on prices and spending. J Health Econ. 2018;59:139‐152. [DOI] [PubMed] [Google Scholar]
- 3. Neprash HT, Chernew ME, Hicks AL, Gibson T, McWilliams JM. Association of financial integration between physicians and hospitals with commercial health care prices. JAMA Intern Med. 2015;175(12):1932‐1939. [DOI] [PubMed] [Google Scholar]
- 4. Nikpay SS, Richards MR, Penson D. Hospital‐physician consolidation accelerated in the past decade in cardiology, oncology. Health Aff (Millwood). 2018;37(7):1123‐1127. [DOI] [PubMed] [Google Scholar]
- 5. Orszag P, Rekhi R. The economic case for vertical integration in health care. NEJM Catalyst. 2020;1(3). [Google Scholar]
- 6. Post B, Buchmueller T, Ryan AM. Vertical integration of hospitals and physicians: economic theory and empirical evidence on spending and quality. Med Care Res Rev. 2018;75(4):399‐433. [DOI] [PubMed] [Google Scholar]
- 7. Cuellar AE, Gertler PJ. Strategic integration of hospitals and physicians. J Health Econ. 2006;25(1):1‐28. [DOI] [PubMed] [Google Scholar]
- 8. Carlin CS, Dowd B, Feldman R. Changes in quality of health care delivery after vertical integration. Health Serv Res. 2015;50(4):1043‐1068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Scott KW, Orav EJ, Cutler DM, Jha AK. Changes in hospital‐physician affiliations in U.S. hospitals. Ann Intern Med. 2018;168(2):156‐157. [DOI] [PubMed] [Google Scholar]
- 10. Beaulieu ND, Dafny LS, Landon BE, Dalton JB, Kuye I, McWilliams JM. Changes in quality of care after hospital mergers and acquisitions. N Engl J Med. 2020;382(1):51‐59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Baker LC, Bundorf MK, Kessler DP. The effect of hospital/physician integration on hospital choice. J Health Econ. 2016;50:1‐8. [DOI] [PubMed] [Google Scholar]
- 12. Koch TG, Wendling BW, Wilson NE. How vertical integration affects the quantity and cost of care for Medicare beneficiaries. J Health Econ. 2017;52:19‐32. [DOI] [PubMed] [Google Scholar]
- 13. Baker LC, Bundorf MK, Kessler DP. Vertical integration: hospital ownership of physician practices is associated with higher prices and spending. Health Aff (Millwood). 2014;33(5):756‐763. [DOI] [PubMed] [Google Scholar]
- 14. Whaley CM, Zhao X, Richards M, Damberg CL. Higher Medicare spending on imaging and lab services after primary care physician group vertical integration. Health Aff (Millwood). 2021;40(5):702‐709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Johnson G, Lyon ZM, Frakt A. Provider‐offered Medicare Advantage plans: recent growth and care quality. Health Aff (Millwood). 2017;36(3):539‐547. [DOI] [PubMed] [Google Scholar]
- 16. Frakt AB, Pizer SD, Feldman R. Plan‐provider integration, premiums, and quality in the Medicare Advantage market. Health Serv Res. 2013;48(6 Pt 1):1996‐2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Meyers DJ, Mor V, Rahman M. Provider integrated Medicare Advantage plans are associated with differences in patterns of inpatient care. Health Aff (Millwood). 2020;39(5):843‐851. [DOI] [PubMed] [Google Scholar]
- 18. Agency for Healthcare Research and Quality. Compendium of U.S. Health Systems. Agency for Healthcare Research and Quality. https://www.ahrq.gov/chsp/data-resources/compendium.html.
- 19. Lyon ZM, Feyman Y, Johnson GM, Frakt AB. Quality of provider‐offered Medicare Advantage plans. BMJ Qual Saf. 2018;27(3):247‐250. [DOI] [PubMed] [Google Scholar]
- 20. Geruso M, Layton T. Upcoding: evidence from Medicare on squishy risk adjustment. J Polit Econ. 2020;12(3):984‐1026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Newhouse JP, Price M, McWilliams JM, Hsu J, McGuire TG. How much favorable selection is left in Medicare Advantage? Am J Health Econ. 2015;1(1):1‐26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Park S, Werner RM, Coe NB. Racial and ethnic disparities in access to and enrollment in high‐quality Medicare Advantage plans. Health Serv Res. 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Reid RO, Deb P, Howell BL, Conway PH, Shrank WH. The roles of cost and quality information in Medicare Advantage plan enrollment decisions: an observational study. J Gen Intern Med. 2016;31(2):234‐241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Park S, Langellier BA, Meyers DJ. Association of health insurance literacy with enrollment in traditional Medicare, Medicare Advantage, and plan characteristics within Medicare Advantage. JAMA Netw Open. 2022;5:e2146792. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Supporting information.
