JUNE 2015 The Skinny on Narrow Networks in Health Insurance Marketplace Plans In-Brief The Affordable Care Act (ACA) has prompted health plans to increase their use of “narrow networks” of providers as a cost containment strategy. These plans have proven popular on the ACA marketplace because they carry lower premiums. Yet consumers have little information to guide them on the tradeoff between lower premiums and network size when shopping among the various plans offered on the ACA marketplace. Regulators and policymakers also have little information on these networks. New federal requirements for updated, accurate provider directories create an opportunity to significantly improve consumers’ ability to make more informed health plan choices. The Leonard Davis Institute of Health Economics has assembled the first integrated dataset of physician networks for the plans offered on the ACA marketplace. This data brief uses this new resource to describe the breadth of the physician networks in plans sold on the state and federal marketplaces. The percent of physician networks that were classified as small or x-small came to 41% overall, 55% for HMO networks, and 25% for PPO networks. The most favorable cost sharing arrangements ACA provisions to create a fair and competitive reported that monthly premiums are more offered by a health plan apply only when marketplace – community rating, standardizing important than other factors in plan choice. enrollees use providers in the plan’s network. plans into tiers based on actuarial value of When that network offers a limited choice of cost sharing, and removing limits on annual or Insurers can use narrow networks to lower providers, it is referred to as a ‘narrow network’. lifetime benefits – left insurers with few options premiums in various ways. They can directly These narrow network plans are particularly for offering lower-cost plans. Narrow networks exclude high-cost providers from the network attractive to consumers who are willing to have become an important feature of premium and direct patients to high-value providers. trade off provider choice for lower premiums variation on the health insurance marketplace They can use the market power of networks and reduced out-of-pocket payments. Even as they remain one of the only remaining to negotiate lower reimbursement levels with though narrow networks have been around long pieces in the insurers’ cost-containment participating providers in exchange for greater before the Affordable Care Act (ACA), they have toolbox. The option of lower-cost plans in the volume, thereby keeping prices low. They can become an issue of increasing interest with ACA marketplace has proved to be important segment their network into tiers, with higher the implementation of the ACA and the Health for price-sensitive consumers as enrollees have cost-sharing for the higher tiers, resulting in Insurance Marketplace (see Figure 1). a de facto narrowing of the network for price- conscious consumers. All of these strategies are designed to control the costs of individual plans Figure 1. Public Interest in Narrow Networks as Tracked by Google Trends offered on the ACA marketplaces. 500 Narrow networks leave consumers vulnerable Normalized Relative Google Search Volume 450 to the financial burden of out-of-network care; 400 the challenge of navigating between in-network 350 providers increases as the network size October 1, 2013 decreases. As a result, network size, even as a 300 Health Insurance Marketplace Opens broad concept, is an important feature of a plan. 250 Yet surveys and other anecdotal reports suggest 200 March 23, 2010 November 15, 2014 that many consumers who selected narrow 150 Open Enrollment Affordable Care Act for 2015 coverage network plans largely on the basis of lower 100 signed into law premiums were unaware of the network size of 50 the plan they selected. Information on networks 0 specific to specialty or geography is mostly 2008 2009 2010 2011 2012 2013 2014 2015 non-existent. 1 JUNE 2015 To date, the only work summarizing the network sizes among require improving the information on the breadth of the networks the plans offered on the Health Insurance Marketplace has associated with the available plans. been limited to hospital networks. McKinsey & Co. recently categorized the network size of plans on the 2015 marketplaces by the proportion of participating hospitals in a rating area. WHAT WE DID They found 39% of networks in plans offered in the marketplace From the 2014 list of all 1,065 unique silver plans (and to be “narrow”, defined as a network with fewer than 70% of 6,690 unique plan / rating area combinations) sold in the hospitals in a rating area. This is a valuable though incomplete marketplaces for all 50 states plus DC as provided by HIX characterization of narrow networks, because it considers Compare, we identified 395 unique provider networks offered hospital participation only. It does not help consumers by 268 different issuers. We used the publicly available provider understand which physicians are part of the plan. directories on the issuer websites of individual marketplace- based insurance networks and plans to gather all providers in In this brief, we summarize network size using an integrated specified networks including data on provider characteristics and standardized list of physician providers across the provider such as specialty, name, gender, and geographic location. These directories of all networks tied to the silver plans sold on the data were collected in the fall of 2014. marketplaces in 2014. We describe the steps to create this integrated list and then we offer summaries of network size The provider lists from which these data were gathered were overall, by plan type, and by physician specialty. not uniform in their formats and coding. Thus we created a multi-stage cleaning process to integrate all lists into a list with The long-term goal of this project is to demonstrate how unified formats for names, addresses, and specialties. We the provider data can be integrated to build a tool that can converted specialties, listed in more than 6,000 ways, into 47 help those shopping for health insurance by improving the specialty groups. Given the preponderance of errors in these lists transparency of the network information for the market-based including duplicates, misspellings, typos, misclassifications, and plans. Preserving the opportunity for health plan consumers physicians who have relocated or retired, we confirmed unique to make an informed plan selection based on the tradeoff physicians by matching the information to national provider between the cost of their plan and the choice of providers will datasets using a set of algorithms that allow for variation in the About Plan Types and Cost Sharing The primary characteristics of the most common plan types may signal some network characteristics. The main plan types are Health Maintenance Organizations (HMOs), Preferred Provider Organizations (PPOs), Point-of-Service (POS) plan, or the newer Exclusive Provider Organization (EPO). The main distinguishing characteristics are provided in the table below. Figure 2. Characteristics of Main Insurance Plan Types PPO EPO POS HMO Primary care physician (PCP) Required? No No Yes, usually Yes Referrals required to see specialist? No No Yes, usually Yes Pre-authorization required? Yes, usually Yes, usually No, PCP referral No, PCP referral is enough is enough Out-of-network coverage? Yes No Yes No From these characteristics we see variation in the penalty for out-of-network care. HMOs and EPOs offer coverage exclusively in-network, which means that the beneficiary is responsible for 100% of care from out-of-network providers. In PPOs and POS plans, where there is coverage for out-of-network care, from the 2015 HIX Compare data we found that cost sharing is typically twice as high. In silver plans using coinsurance the average is 50% for out-of-network services compared to 25% for in-network services. In silver plans using copayments for in-network primary and specialty care, the copayments range from $22-$28 for primary care and $55-$60 for specialty care. In those plans, going out of network will result in coinsurance rates of 50%. Additionally, plans covering out-of-network services (PPOs and POS plans) have higher average deductibles out of network ($6,400-$6,500) than in-network ($2,700). While plan type is a strong signal as to the penalty for going out of network, plan type does not necessarily indicate network size. For example, while HMOs typically have smaller networks than PPOs, HMOs may have broad networks and PPOs may have narrow networks. Plan type is not a sufficient proxy for network size. 2 JUNE 2015 data. The first provider dataset we used was the National Plan Most networks offered on the marketplace are either PPOs or and Provider Enumeration System, better known as the National HMOs: 37% are PPOs, 43% are HMOs, 8% are EPOs, and 12% Provider Identifier (NPI) registry. This dataset allowed us to are POS plans. Within each plan type we categorize network distinguish non-physicians from the physicians and then exclude size. We find that the HMO and PPO distinction is meaningful, the non-physicians. We then applied the fuzzy match algorithms with 55% of HMOs having x-small or small networks, compared using the SK&A dataset of over 700,000 physicians that have to 25% of PPOs. EPOs and POS plans fall somewhere between location and specialty information regularly telephone verified. these two extremes with 37% and 39% of plans having x-small or small networks. The records that could not be matched to at least one of these datasets were assumed to be invalid records and were Figure 3. The Size of Physician Networks for Health Insurance excluded. We excluded physician locations that were not in Marketplace Plans: Overall and by Plan Types rating areas in states where plans were issued. We excluded 40 networks and 17 issuers where the data gathering process 100% 5% 4% 5% 11% Fraction of Networks by Network Size failed to gather complete data. Our analysis dataset consisted 20% 17% 28% of 450,232 physicians participating in plans issued by the 251 80% 24% 37% carriers across 355 networks where we were successful in 29% 22% X-Large gathering publicly available information on all physicians in these 60% Large 24% 28% networks. We also created a dataset of the 199,000 physicians 22% Medium from the SK&A data that were found to not be participating in 40% 26% Small 41% any marketplace network. X-small 30% 30% 30% 20% 17% QUANTIFYING NARROW PHYSICIAN NETWORKS 0 11% 8% 14% 7% 9% Overall PPO HMO EPO POS What determines a narrow network? The ACA provided a national standard for network adequacy, yet this definition has been difficult to put into practice. The definition states that marketplace When we sized networks within selected specialty groups we plans must maintain “a network that is sufficient in number and found 36% of primary care physician networks to be small or types of providers” so that “all services will be accessible without x-small. This is slightly lower than the 41% of small/x-small unreasonable delay.” This identifies three aspects of the network: networks overall, suggesting that some specialties have smaller size, representation of provider types, and geography. There is also networks. One such specialty is oncology, where 59% of the an element of uncertainty as to what should define “sufficient”. networks of oncologists are classified as at least small. The other In this brief we combine all of these elements in our quantification selected specialties listed are all even less likely than primary of physician networks. We estimate network size based on the care physicians to have small networks. fraction of providers in participating rating areas within a state that participate in the network. We look for representation by provider 120 Figure 4. The Size of Physician Networks for Selected Specialty Groups type by tracking particular specialty groups. We are uncertain as to the appropriate threshold for “sufficiency”, so we categorized 100% Fraction of Networks by Network Size 4% 21% network size into five groups using arbitrary cutoffs that might 31% 24% 19% 31% 80% provide meaningful information to consumers: x-small (less than 23% 18% X-Large 10%), small (10%-25%), medium (25%-40%), large (40%-60%), 23% 60% 24% Large 26% and x-large (more than 60%). 21% Medium 40% 22% 34% Small 22% 19% X-small WHAT WE FOUND 20% 22% 19% 17% 15% 25% 14% The distribution of provider networks seems to parallel the 0 8% 11% 7% distribution of hospital networks as described by McKinsey. Primary Care Physicians Internal Medicine Subspecialties Surgery Oncology Dermatology & Ophthalmology By our measures, 41% of networks are small or x-small: 11% of networks are x-small, meaning they include less than 10% of office-based practicing physicians in the area and another 30% Our measures of network size are based on the networks are small, including between 10% and 25% of physicians. At the that cover the entire portion of a state where that network is other end of the spectrum, 11% are x-large, which we define as attached to a plan offered on the marketplace. While these are networks including more than 60% of physicians. useful summary measures that can describe the full breadth 3 JUNE 2015 Figure 5. Physician Network Participation for Insurance Networks for Plans Offered in Health Insurance Marketplace in Atlanta Rating Area Estimated within Selected Geographic Areas ATLANTA ALPHARETTA SE ATLANTA INSURANCE NETWORK GEORGIA RATING AREA 30004+10 MI 30315+10 MI Alliant POS 26% 21% 23% 8% Ambetter HMO 13% 11% 11% 14% Humana HMO 5% 5% 12% 4% Kaiser HMO 11% 10% 8% 2% Humana POS 13% 5% 3% 3% Anthem HMO 24% 27% 37% 46% of the network across the state, these geographic areas are regulations require that plan directories be updated monthly, much larger than the area that would be relevant for any and available in machine-readable formats. This information can single consumer seeking a physician provider. This point is an form the basis for consumer-friendly decision tools that make important caveat to consider going forward as we develop tools the network distinctions clear and meaningful. to provide consumers better information regarding network size. We take the set of insurance networks that are part of plans Ultimately, these steps will improve health insurance coverage offered in the Atlanta rating area to demonstrate the difference and health care access by improving the implementation of in network size by the geographic area chosen (Figure 5). The narrow networks as a strategy for offering lower-cost plans on same network can have a different breadth when considered the marketplaces. Well-functioning narrow networks will survive state-wide, within an entire rating area, or over a smaller area. only if they are made more transparent to consumers and are While the differences are minor for some networks, Alliant POS regulated properly to ensure network adequacy. Our work has has a much smaller network around ZIP code 30315 than across implications for employer-sponsored insurance as well, where the entire Atlanta rating area, while Anthem HMO has a much plans are increasingly offering narrow network options. larger network around ZIP code 30315 than across the entire Atlanta rating area. This potential for variation in network size In the near term, the Leonard Davis Institute of Health within a rating area suggests that rating-area specific network Economics has created a standardized and integrated size information may not be sufficient for consumers. dataset of providers in networks that is of immediate benefit to researchers, policymakers, and regulators. Researchers, under data use restrictions, will be able to access raw provider POLICY IMPLICATIONS data to discover the underlying cost-quality tradeoff, as well as the actual value provided across various plan options. This Our findings demonstrate that networks can be measured along dataset will be available in the coming months from the Leonard various dimensions in a way that is useful to consumers and Davis Institute. A summary dataset by plan will soon be made regulators. These networks could be categorized into distinct, publicly available. The public dataset will provide policymakers, “T-shirt” sizes that make network information more transparent developers, and other interested parties detailed information to consumers shopping on the marketplaces. The summary on network size overall and by specialty for every silver plan information provided also demonstrates the usefulness of offered on the 2014 ACA marketplace. The summary information a standardized and integrated “Find a Doctor” database for on the scope of variation in the choices available to consumers comparing network size between plans that would far surpass on the marketplaces with respect to network size will enhance the utility of existing online plan directories. New federal transparency for consumers, regulators, and policymakers. 4 JUNE 2015 About the Authors This Data Brief was written by Dan Polsky, PhD and Janet Weiner, MPH. About The Leonard Davis Institute of Health Economics The Leonard Davis Institute of Health Economics (LDI) is the University of Pennsylvania’s center for research, policy analysis, and education on the medical, economic, and social issues that influence how health care is organized, financed, managed, and delivered. LDI, founded in 1967, is one of the first university programs to successfully cultivate collaborative multidisciplinary scholarship. It is a cooperative venture among Penn’s health professions, business, and communications schools (Medicine, Wharton, Nursing, Dental Medicine, Law School, and Annenberg School for Communication) and the Children’s Hospital of Philadelphia, with linkages to other Penn schools, including Arts & Sciences, Education, Social Policy and Practice, and Veterinary Medicine. About the Robert Wood Johnson Foundation For more than 40 years the Robert Wood Johnson Foundation has worked to improve health and health care. We are striving to build a national Culture of Health that will enable all to live longer, healthier lives now and for generations to come. For more information, visit www.rwjf.org. Follow the Foundation on Twitter at www.rwjf.org/twitter or on Facebook at www.rwjf.org/facebook. 5