Metropolitan Cincinnati residents have traditionally had among the highest health care costs in the United States, yet little evidence exists that people are getting their money's worth, especially in terms of preventive and primary care. On measures of misuse of care—such as emergency department (ED) visits or hospital admissions for conditions that should be managed in primary care settings, such as asthma—Cincinnati's rates are higher than rates in the state of Ohio or nationwide. Cincinnati also has higher rates of preventable mortality (Radley and Commonwealth Fund, 2012).
Recognizing that high health care spending was not resulting in a healthy population, community leaders began to prioritize local health care reform long before it became a national priority. Recently, large employers, health plans, and health care providers in the Cincinnati area joined with community organizations in a renewed effort to simultaneously lower costs and increase quality. Several factors unique to Cincinnati have spurred this initiative:
- the presence of several large employers (including General Electric [GE], Procter and Gamble, and The Kroger Co.) desiring to keep their employees healthy while controlling their health care costs
- changes to the health care infrastructure, including the consolidation of some hospitals and health care systems, resulting in a reduction in the number of players
- a long history of actively convening organizations comprising the business and health care communities (exemplified by the Health Collaborative, the Greater Cincinnati Health Council, and HealthBridge, which combined in 2012).
In 2009, GE's Healthy Communities Initiative in Cincinnati built on and revitalized this successful collaboration among employers, health plans, providers, and community organizations, helping them win a number of grants. These awards included funding to develop patient-centered medical homes (PCMHs), funding from the Office of the National Coordinator for Health Information Technology to expand electronic health records (also a focus of PCMHs), and an award from the Centers for Medicare and Medicaid Services (CMS) Comprehensive Primary Care (CPC) Initiative to develop innovative models for controlling Medicare, Medicaid, and commercial health care spending. Buoyed by this support, the collaboration designed and implemented a comprehensive intervention, the Healthy Communities Initiative, to improve health care delivery in the Cincinnati metropolitan area.
The overarching goal of the Healthy Communities Initiative was based on the Institute for Healthcare Improvement's (IHI's) “Triple Aim,” which calls for (1) improving the health of populations, (2) improving the patient experience of care, and (3) reducing the cost of care. Such an approach targets all levels of the health system and reflects the complex nature of the current health care environment. The stakeholders for the initiative included large employers, health plans, health systems and providers, and community and government organizations.
To achieve the Triple Aim, the stakeholders focused on five strategic priorities:
- coordinated primary care focused on transforming local practices into PCMHs, a health care delivery model with the goal of delivering comprehensive, coordinated, patient-centered, accessible care with an emphasis on evidence-based quality and safety
- health information exchanges to support communication, clinical decisionmaking, and coordinated care by making individual patient information available to a wide range of health service providers
- quality improvement focused on two common chronic conditions: childhood asthma and adult type II diabetes (Both of these conditions, prevalent in the Cincinnati population, can be controlled through evidence-based processes in ambulatory settings. Failure to follow those standards can lead to costly exacerbations, as well as avoidable ED use and hospitalizations, also called ambulatory care sensitive admissions.)
- public reporting and consumer engagement through a website to publicly report quality measures, which is thought to improve care quality by empowering patients to choose higher-quality care providers and, in turn, spurring providers to improve care delivery
- payment innovations to create aligned incentives for providers, patients, and health plans so that they follow best practices and use resources prudently.
We conducted a rigorous evaluation of the initiative's impact during its first three years. The goal of our analysis was to assess the effect of the Healthy Communities Initiatives in Cincinnati on the Triple Aim of better health, better care, and lower cost for the first three years of this ongoing intervention.
We utilized three sources of public and private data to compare health and behavioral risk factors, employment status, and health care utilization for the Cincinnati metropolitan area with 15 other major metropolitan statistical areas (MSAs) with similarly sized populations.* Measures for the “better health” dimension were derived from the Selected Metropolitan Area Risk Trends (SMART) data of the Behavioral Risk Factor Surveillance System (BRFSS), which contain information on health risk factors including obesity, smoking, and alcohol consumption (Centers for Disease Control and Prevention, 2012), and the Current Population Survey (CPS), which has data on time missed from work due to illness (U.S. Census Bureau and the Bureau of Labor Statistics, 2012).
To measure outcomes related to the “better care” dimension of the Triple Aim, we used measures from two sources: the Healthcare Effectiveness Data and Information Set (HEDIS), which is used to benchmark health plans and is maintained by the National Committee for Quality Assurance (NCQA), an agency that accredits health plans; and the Prevention Quality Indicators developed by the Agency for Healthcare Research and Quality (AHRQ), an agency of the U.S. Department of Health and Human Services. Both used the Truven MarketScan Research Database, a national database of health plan claims.
To measure outcomes related to the “lower costs” dimension, we examined health care costs on a per-member-per-month (PMPM) basis drawing from inpatient, outpatient, ED, and prescription drug claims in the MarketScan database.
Our analysis estimated the differential changes in our measures between Cincinnati and the reference cities during the first three intervention years (2010–2012) compared with a baseline period of 2006–2009. We controlled for individual- and market-level factors to make the reference cities and their residents comparable to the Cincinnati market.
Summary of Findings
Compared to the 15 reference cities, Cincinnati residents were more likely to be non-Hispanic white, less likely to have completed college, and more likely to participate in a high-deductible health plan (HDHP). At baseline, Cincinnati residents had a smaller number of hours missed from work due to illness in the last week, a lower self-reported health status, more office-based primary care visits, more ED visits, more prescription drug fills, and larger total health care costs, and were more likely to be obese and to binge drink. However, the prevalence of chronic conditions among Cincinnati residents was similar to the prevalence among residents in the reference cities. The remainder of this section summarizes the most salient findings. The findings are summarized in Table 1.
Table 1. Intervention Effects on Health, Quality of Care, Health Care Utilization, and Health Care Cost
|Domain||Outcome Metrics||Intervention Years|
|Health||Self-reported health status||—||—||—|
|Productivity||Any missed work due to illness||—||—||↓|
|Work hours missed due to illness||—||—||↓|
|Access to Primary Care||Adults' access to preventive/ambulatory health services||—||—||↓|
|Children's and adolescents' access to primary care practitioners||↓||↓||↓|
|Chronic care||Percentage of patients using angiotensin receptor blockers (ARBs)/angiotensin-converting enzyme (ACE) inhibitors who receive appropriate monitoring||—||↓||↓|
|Percentage of patients on diuretics receiving appropriate monitoring||—||↓||↓|
|Percentage of asthma patients receiving appropriate medications||↓||↓||—|
|Percentage of diabetic patients receiving hemoglobin A1c testing||—||↓||—|
|Percentage of diabetic patients receiving low-density lipoprotein cholesterol testing||—||—||—|
|Percentage of patients with lower back pain without imaging within 28 days of the diagnosis||—||—||—|
|Preventable admissions and ED visits||Ambulatory care sensitive inpatient admissions||—||—||↓|
|Inpatient readmissions within 30 days of discharge||↓||—||—|
|Potentially avoidable ED visits||—||—||—|
|Outpatient||Office-based primary care visits||↓||↓||↓|
|Outpatient PMPM cost||—||—||—|
|Prescription drug||Prescription drug fills||↓||↓||↓|
|Prescription drug PMPM cost||—||↓||↓|
|Emergency care||ED visits||↑||↑||↑|
|ED PMPM cost||—||↑||—|
|Inpatient care||Inpatient admissions||—||—||—|
|Inpatient PMPM cost||—||—||—|
|Total cost||Total PMPM Cost||—||—||—|
NOTE: The table shows the changes in Cincinnati relative to the reference cities. — indicates no statistically significant findings. ↑ represents a statistically significant increase in Cincinnati relative to the reference cities (p≤0.05). Compared to the reference cities, an outcome metric may increase more or decline less in Cincinnati. ↓ represents a statistically significant decrease in Cincinnati relative to the reference cities (p≤0.05). Compared to the reference cities, an outcome metric may increase less or decline more in Cincinnati.
The Healthy Communities Initiative in Cincinnati was associated with improved employee productivity. Over the course of the intervention, the percentage of people in Cincinnati who responded that they had missed work dropped by about 1 percentage point, while it remained almost constant in the reference cities. In 2012, we found a significant decline in the likelihood of being absent from work, which translated to an estimated 7,281 fewer Cincinnati employees calling in sick over the course of the year. In addition, there was a trend toward a decrease in the mean number of hours missed per person per year over the course of the intervention. For 2012, the difference amounted to about 140,000 working hours or about 70 full-time-equivalent (FTE) employees. Nonetheless, the intervention was not linked to significant improvements in residents' health and health behaviors.
We found improvements in preventable hospital admissions and readmissions that point to better care coordination, particularly for higher-risk patients. Ambulatory care sensitive admissions decreased from 4.73 per 1,000 residents (0.55 more than the reference cities) during the baseline period, to 3.72 per 1,000 in 2012—a significant decrease. However, we found that adherence to evidence-based recommendations for chronic care management decreased in Cincinnati compared with the reference cities.
At the same time, use of primary and outpatient care decreased in Cincinnati compared with the reference cities. During the baseline years, Cincinnati averaged 1,714 outpatient visits per 1,000 member years, about 36 visits per year more than the reference cities. The number of office-based primary care visits declined significantly in all years of the intervention, though the decline reflected fewer than five visits per 1,000 member years in 2010–2011. By 2012, we estimated that Cincinnati residents had 136 fewer office-based primary care visits per 1,000 member years than the reference cities.
During this period, use of prescription drugs decreased also. In 2009, Cincinnati residents had an estimated 5,234 prescriptions per 1,000 residents, 108 more prescriptions per year than statistically similar residents of the reference cities. By 2011, we estimated 510 fewer prescriptions per 1,000 residents in Cincinnati. In other words, a Cincinnati resident used about 0.5 fewer prescriptions per year after the second year of the intervention.
Use of ED services increased during the intervention. Utilization increased significantly in Cincinnati relative to the reference cities during the three intervention years. Cincinnati began the intervention with approximately 161 ED visits per 1,000 member years, or seven more ED visits per 1,000 member years than the reference cities. That difference rose to 13 more visits per 1,000 member years by 2011 before dropping to only ten more visits in 2012. The differences between Cincinnati and the reference cities changed significantly from the baseline difference during the intervention years.
Among the changes in utilization patterns, only the change in prescription drug use led to lower prescription costs, but no statistically significant changes in overall health care costs were observed.
The absence of a strong effect of the intervention on health status indicators, such as body weight and smoking rates, is not surprising. First, it may take more than three years to observe significant changes in health behaviors and status. In addition, the intervention did not explicitly target health-related behaviors; rather, it largely focused on improving medical care. Even well-resourced and personalized interventions, such as workplace wellness programs, are known to have only a limited effect on health-related behaviors, such as smoking cessation and physical activities (Mattke et al., 2013). Thus, it is to be expected that improvements in health are unlikely to materialize as a “side effect” of the interventions. However, we found a differential decrease in illness-related work loss in Cincinnati. While instances of sick leave declined in both Cincinnati and the 15 reference markets after 2009 (presumably as a consequence of the recession), the trend was much stronger in Cincinnati and the difference reached statistical significance in 2012, the third year of the intervention. Because the trend is adjusted for differences in age structure and burden of disease between Cincinnati and the reference markets, we interpret it as an early sign of improved health of the workforce.
We found that the Healthy Communities Initiative in Cincinnati was significantly associated with reductions in hospital readmissions and ambulatory care sensitive hospital admissions, suggesting improved care coordination and better post-discharge management for higher-risk patients. During the intervention period, the Greater Cincinnati Health Council's Accountable Care Transformation group engaged 21 hospitals in reducing readmissions and improving care coordination after a hospital admission, which could contribute to the decline in hospital readmissions. Additionally, the 2012 introduction of the HealthBridge alert system is likely to have contributed to this result, as it notifies primary care providers if one of their patients has been admitted to the hospital or visited the ED, and thus facilitates planning and management of care transitions. As mentioned above, by October 2012, 87 sites were running and 26,000 alerts had been sent (U.S. Department of Health and Human Services and Office of the National Coordinator for Health Information Technology, November 2012).
At the same time, the observed decline in use of primary and outpatient care and of prescription drugs is counterintuitive, as is the lower adherence to evidence-based recommendations for chronic care management and increased ED use. If the intervention worked as expected, we would observe increased use of primary care and prescriptions, improved chronic disease management, and fewer inpatient admissions and ED visits.
It should be emphasized that a finding of no effect would not have been entirely surprising, as the intervention is still in its early stage. While formally in its third year at the time of the evaluation, many of the more fundamental changes are just beginning to take effect. Better access to high-quality primary care through the promulgation of the PCMH concept is one of the cornerstones of the initiative. As of mid-2013, 84 practices in the Cincinnati MSA had obtained NCQA Level 3 PCMH recognition, representing about 24 percent of the primary care providers in Cincinnati. Practices that have only recently obtained their PCMH designation may not have reached their full potential. A recent study of a medical home pilot in Pennsylvania did not detect significant effects on utilization or cost of care, and only a limited effect on quality of care over a three-year intervention (Friedberg et al., 2014). Similarly, a new alert system of the HealthBridge health information exchange was implemented in 2012.
Yet finding opposite trends from what we expected is surprising, although it can likely be explained by the wide adoption of HDHPs in Cincinnati relative to the reference markets during these first three years of the Healthy Communities Initiative. The share of HDHPs in Cincinnati more than doubled from 13.4 percent in 2009 to 28.5 percent in 2012, but only increased from 7.5 percent to 10.8 percent in the reference cities during the same time frame. As HDHPs shift responsibility for health care costs to plan members, they have profound impacts on utilization patterns. Prior literature suggests that HDHPs reduce health care spending by 5 to 14 percent on average, although the effect varies across employers (Bundorf, 2012). Cost savings from HDHPs are primarily because of reductions in prescription cost and outpatient cost. HDHPs have no consistent effect on inpatient admissions; they have modest to no reductions in preventive service use when they are exempted from the deductible and significant reductions when they are not. Consumers in HDHPs may indiscriminately reduce utilization.
To account for the adoption of HDHPs in our analysis, we included an indicator for HDHP in all the models and also conducted sensitivity analyses using the individuals who were always in an HDHP or never in an HDHP during 2006–2012 under the assumption that these individuals did not experience a significant change in cost-sharing arrangements and thus would not likely change their care-seeking behavior. However, the results in those subsamples were largely similar to those in the overall population. One possible explanation is that the rapid adoption of HDHPs in the Cincinnati market affected not only the individuals in HDHPs but also those not in them. This hypothesis is consistent with prior research showing that providers tend to orient their practice patterns to the average or modal insurance coverage in their catchment areas (Glied and Zivin, 2002; Hu and Reuben, 2002; Landon, 2004). A caveat, however, is that this body of literature is primarily based on the experience in managed care, which typically imposes financial incentives on providers, whereas in this case, HDHPs primarily influence patient care-seeking behaviors. It is also possible that other unobserved factors, such as new provider payment arrangements, led to the similarity in findings between individuals who never enrolled in an HDHP and those in the full sample.
Changes in use of care did not translate into changes in overall cost of care during the first three years of the intervention. This finding can be explained by the fact that payment innovations—which, based on prior research, are the most likely instrument to reduce overall cost—were implemented only recently in Cincinnati. In the Cincinnati-Dayton region, 75 practices have joined the Centers for Medicaid and Medicare Services' CPC Initiative, in which payers offer bonus payments to primary care providers who effectively coordinate patient care. In addition, Cincinnati's Mercy Health System is involved in a national payment reform pilot with the CMS Medicare Shared Savings Program, but the initiative became operational only in the fall of 2012.
Limitations of the Analysis
This analysis had several limitations. First and foremost, unobservable differences between Cincinnati and the 15 reference markets, as well as their respective residents, may have influenced our results. The CPS and BRFSS data did not allow us to track individuals over time, and it might have reduced our ability to detect the intervention effect. In addition, we were not able to control for the changes in the health care delivery systems of other markets during the study period. This challenge could have led to an underestimation of the effect of the interventions implemented in Cincinnati. Further, no data were available on nontraditional forms of care delivery in PCMHs, such as phone consultations or electronic exchanges between providers and patients. It is likely that nontraditional services substitute for in-person physician visits. If this is true, office-based primary care visits are not a good measure for quality improvement resulting from PCMHs because a decrease in office-based primary care visits might not reflect an actual decline in access to primary care. Moreover, we were not able to tease out the potential selection in HDHP participation. GE, as one of the largest employers in Cincinnati, required all salaried employees to join an HDHP in 2010 and required all production employees to do the same in 2012. But we are not sure whether an HDHP plan was the only option among other employers in Cincinnati or employers in the reference cities. This potential selection bias may have confounded our sensitivity analyses for those always or never enrolled in an HDHP plan.
We conducted the first quantitative evaluation on the Healthy Communities Initiative in Cincinnati for its first three intervention years. Overall, our findings were largely inconclusive because of the concomitant marketwide shift to HDHPs and the early stage of the intervention.
Transitions to HDHPs are known to affect care-seeking behaviors profoundly—particularly immediately after a change in benefit design, as plan members adapt to the new incentives. As the share of HDHP plans more than doubled in Cincinnati during our analysis period, it may have concealed intervention effects. As the level of HDHP penetration in Cincinnati stabilizes—and, thus, the effect of switching to HDHPs on utilizations and costs decreases—analyzing additional years of data will allow the effect of the intervention to be disentangled from the effect of benefit design changes.
As key components of the intervention—such as payment redesign, PCMHs, and the HealthBridge alert notification—were still being fully implemented during the period of analysis, the intervention will not have been able to take full effect. We did find some encouraging signs that better care coordination bears fruit, such as less illness-related work loss and fewer avoidable hospital admissions and readmissions. These early impacts suggest that the initiative may succeed in improving care, lowering cost, and improving health status if given sufficient time. Therefore, a future evaluation of the Healthy Communities Initiative in Cincinnati will be able to assess a more mature program, leverage more data, and result in more conclusive findings.
Bundorf, M. K. (2012). Consumer-directed health plans: Do they deliver? POLICY 1:6.
Centers for Disease Control and Prevention (2012). Behavioral risk factor surveillance system survey data. Atlanta: U.S. Department of Health and Human Services.
Friedberg, M. W., Schneider, E. C., Rosenthal, M. B., Volpp, K. G., and Werner, R. M. (2014). Association between participation in a multipayer medical home intervention and changes in quality, utilization, and costs of care. JAMA 311(8), 815–825.
Glied, S., and Zivin, J. G. (2002). How do doctors behave when some (but not all) of their patients are in managed care? Journal of Health Economics 21(2), 337–353.
Hu, P., and Reuben, D. B. (2002). Effects of managed care on the length of time that elderly patients spend with physicians during ambulatory visits: National Ambulatory Medical Care Survey. Medical Care 40(7), 606–613.
Landon, B. E. (2004). Commentary on “Penetrating the 'Black Box': Financial incentives for enhancing the quality of physician services,” by Douglas A. Conrad and Jon B. Christianson. Medical Care Research and Review 61(3 suppl), 69S–75S.
Mattke, S., Liu, H., Caloyeras, J., Huang, C. Y., Van Busum, K. R., Khodyakov, D., et al. (2013). Workplace wellness programs study. Santa Monica, Calif.: RAND Corporation, RR-254-DOL. As of August 6, 2014:
Radley, D. C., and Commonwealth Fund (2012). Rising to the challenge: Results from a scorecard on local health system performance, 2012. Commonwealth Fund.
U.S. Census Bureau and the Bureau of Labor Statistics (2012). Current populations survey.
U.S. Department of Health and Human Services and Office of the National Coordinator for Health Information Technology (November 2012). Fact Sheet: Greater Cincinnati Beacon Collaboration.
* The 15 MSAs include Charlotte-Gastonia-Rock Hill, N.C.-S.C.; Cleveland-Elyria-Mentor, Ohio; Columbus, Ohio; Denver-Aurora-Broomfield, Col.; Jacksonville, FL; Kansas City, Mo.-Kan.; Las Vegas-Paradise, Nev.; Memphis, Tenn.-Miss.-Ark.; Nashville-Davidson-Murfreesboro-Franklin, Tenn.; Orlando-Kissimmee-Sanford, Fla.; Portland-Vancouver-Hillsboro, Ore.-Wash.; Providence-New Bedford-Fall River, R.I.-Mass.; San Antonio-New Braunfels, Texas; San Jose-Sunnyvale-Santa Clara, Calif.; and St. Louis, Mo.-Ill.
This research was conducted by RAND Health Advisory Services, a part of RAND Health.