Filling the need for trusted information on national health issues…

Geographic Focus

This study examined Medicare Advantage plans available in 2015 in 20 counties: Allegheny County, PA; Clark County, NV; Cook County, IL; Cuyahoga County, OH; Davidson County, TN; Douglas County, NE; Erie County, NY; Fulton County, GA; Harris County, TX; Jefferson County, AL; King County, WA; Los Angeles County, CA; Mecklenburg County, NC; Miami-Dade County, FL; Milwaukee County, WI; Multnomah County, OR; New Haven County, CT; Pima County, AZ; Queens County, NY; and Salt Lake County, UT. The county is the smallest area, in general, that a Medicare Advantage plan must cover. Counties vary greatly in size and may not be the best metric to assess the health care market of particular locales, but an analysis at the county level provided the most complete set of data available for this type of analysis, as well as a reasonable snapshot of the health care market accessible to beneficiaries in that region.

Sources and Analysis of Data

We categorized networks into one of three sizes based on the share of physicians in the county that were included in the directory: broad (70% or more of the physicians), medium (30-69% of the physicians), and narrow (less than 30% of the physicians). The analysis included the 26 specialties that Medicare Advantage plans are required to include in their networks in 2015, and excluded all other specialties. For example, the analysis does not include pediatricians, diagnostic radiologists, anesthesiologists, or emergency medicine specialists.

The provider directories were collected from the websites of the insurers offering the plans during the Medicare Open Enrollment Period in order to best mimic what Medicare beneficiaries would access when selecting a plan. We focus our analysis on HMOs and local PPOs because the other types of Medicare Advantage plans either do not have networks (e.g., some private fee-for-service plans), have networks that are structured to cover areas larger than a county (e.g., regional PPOs), are paid in unique ways that influence the selection of providers available to beneficiaries (e.g., cost plans), or are not available for general enrollment (Special Needs Plans and employer group waiver plans). In 19 of 20 counties, both HMOs and local PPOs were available in 2015, with the exception of Los Angeles County, which only had Medicare Advantage HMOs. Of the 422 plans eligible to be included the analysis in the 20 counties, we were able to download the provider directories for 391 plans, including 292 HMOs and 99 local PPOs. These 391 plans accounted for 94 percent of the Medicare Advantage enrollment in the 20 counties. The provider directories missing from the analysis (31 out of 422 plans) either could not be downloaded and saved in a readable format, or were not available on the insurer’s website or directly from the insurer.

The directories were converted to machine-readable format and the text in the directories were matched against a national census of physicians maintained by SK&A using a text-matching method.1 The text-matching program searched the directories for the names of every doctor practicing in each county.

We converted the directories for each plan to a machine-readable format to conduct this analysis. Many PDF were saved as PDFs without text, requiring the files to be recreated in order to be read. Once converted to a machine-readable format, we matched the text in the directories against a national census of physicians maintained by SK&A using a text-matching method. SK&A Information Services Inc., maintains a national database of physicians, which they continuously update through a telephone verification process, and this analysis used an SK&A database prepared on January 4, 2016. The text-matching program searched the directories for the names of every doctor practicing in each county.

Using the R statistical software program, we used a publicly available PDF text converter to load each directory as a text file (.txt).2 The analysis excluded the following doctors:

  • Doctors who work exclusively for the Veterans Administrations (about a percent of all physicians).
  • Doctors whose listed specialty in SK&A was not a required category for Medicare Advantage plan networks.

SK&A assigns physicians up to two specialties. If a doctor’s primary classification was pediatrics or a select number of other specialties, they were excluded. The largest specialties not included in the analysis are pediatricians, diagnostic radiologists, anesthesiologists, emergency medicine specialists and optometrists. For the remaining doctors, we used the primary classification to categorize them into one of the 26 classifications established by CMS. For the remaining providers who were neither excluded nor classified, we turned to the secondary specialty and either grouped them with the 26 categories or dropped them. Nationally the largest specialty was primary care (32%) followed by gynecology OB/GYN (7%), cardiology (6%) and orthopedic surgery (6%). Geriatricians were defined separately; any doctor included in the analysis was grouped into a specialty as well as a geriatrician. Therefore, a geriatrician could be listed as both a primary care specialist, as well as a geriatrician. About one percent of physicians are geriatricians. In sum, more than 25% of doctors were excluded from the analysis.

The boundaries of counties do not always align with the zip codes available in the SK&A data. To be conservative, we included doctors who have at least one practice location in a zip code, which is at least partially included in one of the 20 counties. Our text-matching method searched for a string match of all the doctors in a given county. For each doctor’s name within the county, we flagged every occurrence where the last name was within three words of the first name. This method would identify both “Johnathan Hancock” and “Hancock, Johnathan” as well as cases in which the directory included a middle name or title such as “Johnathan Adam Hancock”, “Hancock IV, Johnathan Adam” or “Hancock, (Surgeon), Johnathan Adam”.

In addition, we searched last names and up to nine nicknames for every doctors in the county. For example, we would successfully identify “Johnathan Hancock” as either “John Hancock” or “Johnny Hancock”. We selected 435 formal names, which often have nicknames; for example Matt as a nickname for Matthew or Bill, Billy, Will, Willie, Willy as nicknames for William. Thirty-eight percent of physicians in the 20 counties had at least one nickname – about 0.64 percent had six or more nicknames (Elizabeth, Katherine, Margaret).

To reduce match errors with hyphenated last names – we eliminated punctuation – so regardless of whether a directory listed Kareem Abdul-Jabbar as “Kareem Abdul-Jabbar”, “Kareem Abdul Jabbar”, or “Kareem AbdulJabbar” we would successfully identify the doctor as practicing.

To test the validity of the text-matching program, 20 names were drawn at random from each of the 20 counties (400 names in total) using the SK&A dataset, and then a person searched for those names in the directories by hand, in the same fashion that an enrollee would look for their doctor. For 377 out of 400 names (94%), the person and the text-matching program agreed. Of the names that did not match, 3% were false-negatives (text-matching program did not count the physician in the directory when it should have done so) and 3% were false-positives, indicating that the totals were 99.8% accurate. The most important form of error may be the accuracy of the provider directory – our text analysis only identifies whether a provider is listed in the directory, not whether that provider is taking new patient or even still practicing in the service area. The error in our text analysis method may work to both overestimate and underestimate the percentage of doctors.

Some examples of potential false-positives – or cases in which out method would over estimate the number of doctors may include:

  • Cases in which the directory contains the name of someone who is not an eligible physician and there is an eligible physician by the same name practicing in the county but is not included in the network. For example, a plan does not include a primary care doctor named “Marie Curie” but does include an optometrist by the same name “Marie Curie”.
  • While most directories are structured as list of physicians followed by contact information it is possible that there are at least some occurrences of false-positives through the proximity of names. For example if a directory included the sentence, “Thomas Jefferson and John Adam were fierce political rivals” and there happened to be a physicians in the county named “John Jefferson” we would incorrectly identify him as a physician in the plan.
  • If a county included multiple doctors with the same name, and a plan only included one of these doctors in its network – we would incorrectly identify them all as participating. Only 502 instances occurred where there were duplicate names (out of 66,679 physicians).

Alternatively, there are cases in which our text analysis may undercount the percentage of doctors participating in the network. Some examples of true-negatives may include cases in which a directory lists a physician under a different name than the one listed in the SK&A directory. This may include cases in which the physician’s name is spelled differently in the directory and the SK&A database or the doctor uses only one of their two hyphenated last names.

Some of this study’s findings were aggregated by county and across counties by weighting each plan’s results by its enrollment using the Centers for Medicare and Medicaid Services Medicare Advantage Enrollment files for March 2015. Data on plan enrollment was made available by CMS for March 2015 (”State/County/Contract/Plan (CPSC) Enrollment Data”). No enrollment data is available for plans with fewer than 10 enrollees (7% of the plans in our sample).3 Categorization of the plans as Preferred Provider Organizations (PPO) and Health Maintenance Organization (HMO) was made by CMS.4

Tables

Table 1. Characteristics of Counties Included in the Analysis, 2015
County Largest city Number of Plans Available in County (Universe) Number of Plans Successfully Evaluated Medicare Advantage Enrollment Evaluated Medicare Advantage Penetration Rate Percent of Enrollment in County Evaluated Counties’ Enrollment As Percent of National Medicare Advantage Market
All Counties 422 391 1,512,624 40% 94% 13.9%
Allegheny, PA Pittsburgh 22 22 100,832 62% 100% 0.9%
Clark, NV Las Vegas 11 11 98,767 38% 100% 0.8%
Cook, IL Chicago 19 18 51,404 17% 64% 0.7%
Cuyahoga, OH Cleveland 26 20 18,231 37% 40% 0.4%
Davidson, TN Nashville 15 15 27,624 42% 100% 0.2%
Douglas, NE Omaha 13 13 15,501 23% 100% 0.1%
Erie, NY Buffalo 25 25 74,043 56% 100% 0.6%
Fulton, GA Atlanta 17 15 21,933 35% 96% 0.2%
Harris, TX Houston 31 28 122,838 39% 100% 1.1%
Jefferson, AL Birmingham 12 12 35,862 42% 100% 0.3%
King, WA Seattle 26 23 73,976 34% 98% 0.7%
Los Angeles, CA Los Angeles 35 34 379,715 43% 99% 3.3%
Mecklenburg, NC Charlotte 15 15 26,535 31% 100% 0.2%
Miami-Dade, FL Miami 31 29 204,503 62% 100% 1.8%
Milwaukee, WI Milwaukee 6 5 31,697 41% 98% 0.3%
Multnomah, OR Portland 30 30 43,941 58% 100% 0.4%
New Haven, CT New Haven 16 13 32,475 28% 94% 0.3%
Pima, AZ Tucson 13 13 55,754 46% 100% 0.5%
Queens, NY New York City 46 37 52,739 43% 62% 0.7%
Salt Lake, UT Salt Lake City 13 13 44,254 41% 100% 0.4%
SOURCE: Kaiser Family Foundation analysis of CMS Medicare Advantage enrollment and landscape files for 2015.
Table 2. Distribution of Medicare Advantage Plans’ Physician Networks Versus Plan Enrollment, 2015
  Distribution of Medicare Advantage Plans Distribution of Medicare Advantage Enrollees
County Narrow
(Less than 30%)
Medium
(30-69%)
Broad
(At Least 70%)
Narrow
(Less than 30%)
Medium
(30-69%)
Broad
(At Least 70%)
All Counties 27% 52% 21% 35% 43% 22%
Allegheny, PA 0% 5% 95% 0% 0% 100%
Clark, NV 45% 55% 0% 91% 9% 0%
Cook, IL 39% 61% 0% 40% 60% 0%
Cuyahoga, OH 35% 45% 20% 3% 70% 27%
Davidson, TN 0% 73% 27% 0% 74% 26%
Douglas, NE 0% 69% 31% 0% 24% 76%
Erie, NY 16% 28% 56% 0% 3% 96%
Fulton, GA 27% 73% 0% 20% 80% 0%
Harris, TX 61% 39% 0% 78% 22% 0%
Jefferson, AL 0% 83% 17% 0% 71% 29%
King, WA 22% 70% 9% 14% 72% 14%
Los Angeles, CA 53% 47% 0% 51% 49% 0%
Mecklenburg, NC 0% 0% 100% 0% 0% 100%
Miami-Dade, FL 66% 34% 0% 46% 54% 0%
Milwaukee, WI 0% 40% 60% 0% 7% 93%
Multnomah, OR 20% 80% 0% 26% 74% 0%
New Haven, CT 0% 62% 38% 0% 49% 51%
Pima, AZ 31% 69% 0% 15% 85% 0%
Queens, NY 27% 68% 5% 2% 75% 23%
Salt Lake, UT 0% 54% 46% 0% 38% 62%
SOURCE: Kaiser Family Foundation analysis of Medicare Advantage plans’ physician networks in 20 counties, 2017.
Table 3. Distribution Across Plans of the Share of Physicians Included in Medicare Advantage Plans’ Networks, 2015
  Number of Plans Average Minimum 25th Percentile Median 75th Percentile Maximum
All Counties 391 46% 1% 28% 44% 64% 87%
Allegheny, PA 22 77% 30% 72% 74% 87% 87%
Clark, NV 11 41% 21% 25% 53% 53% 60%
Cook, IL 18 32% 17% 28% 33% 35% 54%
Cuyahoga, OH 20 41% 7% 21% 40% 50% 77%
Davidson, TN 15 62% 37% 56% 65% 68% 77%
Douglas, NE 13 58% 43% 44% 58% 75% 79%
Erie, NY 25 60% 4% 50% 80% 81% 83%
Fulton, GA 15 39% 11% 29% 40% 54% 59%
Harris, TX 28 33% 17% 23% 28% 43% 59%
Jefferson, AL 12 54% 42% 43% 52% 55% 82%
King, WA 23 41% 14% 32% 42% 46% 76%
Los Angeles, CA 34 28% 1% 18% 28% 42% 46%
Mecklenburg, NC 15 73% 70% 70% 70% 77% 82%
Miami-Dade, FL 29 25% 4% 17% 26% 33% 46%
Milwaukee, WI 5 67% 42% 52% 79% 81% 81%
Multnomah, OR 30 42% 9% 37% 40% 60% 65%
New Haven, CT 13 65% 53% 54% 66% 73% 73%
Pima, AZ 13 43% 22% 28% 46% 55% 66%
Queens, NY 37 40% 6% 26% 39% 58% 71%
Salt Lake, UT 13 66% 34% 59% 63% 82% 83%
NOTE: All percentages are not weighted by the number of enrollees in each plan.
SOURCE: Kaiser Family Foundation analysis of Medicare Advantage plans’ physician networks in 20 counties, 2017.
Table 4. Average Share of Physicians Included in Medicare Advantage Plans’ Networks, by County and Specialty, 2015
Specialty All Counties Allegheny  Clark Cook Cuyahoga Davidson Douglas
All Specialties 46% 77% 41% 32% 41% 62% 58%
Primary Care 42% 77% 32% 30% 40% 47% 51%
Geriatricians 43% 88% 38% 37% 46% 68% 76%
Allergy and Immunology 51% 75% 41% 33% 43% 67% 78%
Cardiology 54% 86% 52% 43% 41% 82% 65%
Cardiothoracic Surgery 55% 86% 29% 46% 37% 83% 53%
Dermatology 48% 81% 48% 26% 48% 65% 85%
Endocrinology 45% 85% 36% 28% 33% 66% 63%
ENT/ Otolaryngology 53% 90% 64% 32% 47% 79% 63%
Gastroenterology 56% 83% 59% 37% 49% 82% 74%
General Surgery 51% 77% 44% 31% 40% 67% 61%
Gynecology, OB/GYN 46% 75% 51% 30% 40% 53% 66%
Infectious Diseases 45% 82% 52% 46% 43% 50% 53%
Nephrology 52% 74% 74% 39% 38% 80% 59%
Neurology 46% 75% 35% 33% 35% 74% 61%
Neurosurgery 50% 80% 69% 34% 42% 90% 51%
Medical, Surgical Oncology 47% 83% 51% 34% 49% 77% 67%
Radiation/ Radiation Oncology 52% 47% 71% 43% 56% 88% 38%
Ophthalmology 59% 88% 57% 39% 51% 73% 74%
Orthopedic Surgery 49% 82% 37% 32% 48% 64% 66%
Physiatry, Rehabilitative Medicine 38% 67% 33% 24% 29% 61% 38%
Plastic Surgery 35% 79% 10% 27% 35% 39% 55%
Podiatry 48% 77% 42% 31% 38% 63% 69%
Psychiatry 23% 45% 9% 13% 25% 43% 32%
Pulmonology 53% 88% 56% 43% 41% 71% 62%
Rheumatology 44% 72% 45% 28% 42% 65% 58%
Urology 57% 88% 54% 43% 44% 78% 64%
Vascular Surgery 55% 93% 24% 32% 49% 61% 71%
Erie Fulton  Harris Jefferson King Los Angeles Mecklenburg
All Specialties 60% 39% 33% 54% 41% 28% 73%
Primary Care 57% 32% 28% 50% 39% 32% 71%
Geriatricians 43% 30% 32% 35% 33% 29% 83%
Allergy and Immunology 71% 44% 27% 35% 49% 28% 73%
Cardiology 70% 50% 40% 59% 53% 34% 88%
Cardiothoracic Surgery 68% 53% 45% 77% 51% 40% 79%
Dermatology 60% 39% 31% 46% 42% 17% 78%
Endocrinology 63% 39% 28% 44% 42% 24% 74%
ENT/ Otolaryngology 62% 44% 27% 51% 46% 27% 87%
Gastroenterology 68% 52% 37% 68% 52% 32% 85%
General Surgery 67% 45% 40% 68% 45% 31% 77%
Gynecology, OB/GYN 57% 27% 30% 60% 35% 26% 84%
Infectious Diseases 60% 49% 36% 51% 47% 26% 56%
Nephrology 61% 48% 43% 41% 46% 39% 60%
Neurology 62% 49% 34% 38% 49% 24% 70%
Neurosurgery 73% 41% 31% 66% 38% 35% 69%
Medical, Surgical Oncology 55% 54% 23% 35% 48% 26% 77%
Radiation/ Radiation Oncology 73% 59% 42% 83% 38% 30% 90%
Ophthalmology 70% 55% 54% 77% 51% 34% 81%
Orthopedic Surgery 59% 43% 36% 59% 43% 24% 79%
Physiatry, Rehabilitative Medicine 53% 31% 25% 50% 40% 18% 73%
Plastic Surgery 54% 25% 20% 30% 22% 15% 39%
Podiatry 68% 46% 36% 80% 42% 28% 69%
Psychiatry 36% 18% 25% 34% 22% 7% 30%
Pulmonology 62% 52% 38% 68% 52% 33% 76%
Rheumatology 54% 31% 28% 42% 52% 24% 60%
Urology 66% 51% 51% 86% 46% 32% 79%
Vascular Surgery 66% 57% 50% 84% 51% 38% 66%
Miami-Dade Milwaukee Multnomah New Haven Pima Queens  Salt Lake
All Specialties 25% 67% 42% 65% 43% 40% 66%
Primary Care 20% 65% 35% 54% 41% 37% 60%
Geriatricians 18% 50% 29% 75% 36% 34% 59%
Allergy and Immunology 53% 73% 50% 64% 54% 43% 69%
Cardiology 31% 74% 44% 82% 53% 50% 71%
Cardiothoracic Surgery 35% 90% 54% 79% 51% 43% 74%
Dermatology 24% 63% 48% 79% 44% 44% 68%
Endocrinology 21% 64% 46% 73% 34% 40% 66%
ENT/ Otolaryngology 34% 69% 60% 77% 52% 50% 78%
Gastroenterology 44% 73% 61% 79% 48% 50% 72%
General Surgery 33% 70% 54% 79% 51% 45% 67%
Gynecology, OB/GYN 23% 74% 51% 66% 44% 41% 67%
Infectious Diseases 27% 62% 30% 45% 59% 34% 55%
Nephrology 37% 79% 56% 73% 60% 38% 67%
Neurology 31% 64% 32% 69% 41% 44% 66%
Neurosurgery 23% 61% 40% 69% 56% 37% 80%
Medical, Surgical Oncology 24% 71% 34% 68% 39% 46% 72%
Radiation/ Radiation Oncology 40% 66% 45% 74% 67% 38% 49%
Ophthalmology 31% 73% 61% 87% 53% 55% 82%
Orthopedic Surgery 27% 77% 44% 83% 53% 39% 73%
Physiatry, Rehabilitative Medicine 10% 55% 31% 61% 29% 43% 60%
Plastic Surgery 11% 65% 41% 65% 31% 32% 55%
Podiatry 20% 68% 52% 84% 46% 31% 70%
Psychiatry 10% 41% 23% 26% 7% 20% 37%
Pulmonology 47% 80% 55% 70% 28% 49% 56%
Rheumatology 38% 54% 44% 66% 50% 35% 52%
Urology 44% 77% 57% 88% 50% 54% 71%
Vascular Surgery 28% 64% 46% 71% 67% 56% 84%
NOTE: All percentages not weighted by the number of enrollees in each plan.
SOURCE: Kaiser Family Foundation analysis of Medicare Advantage plans’ physician networks in 20 counties, 2017.
Table 5. Average Share of Physicians Included in Medicare Advantage HMO and PPO Networks, By County, 2015
County Number of Plans Size of Provider Networks, by Plan Type
Total HMO PPO Overall HMO PPO
All Counties 391 292 99 46% 42% 57%
Allegheny, PA 22 13 9 77% 80% 73%
Clark, NV 11 6 5 41% 30% 56%
Cook, IL 18 12 6 32% 29% 37%
Cuyahoga, OH 20 15 5 41% 33% 65%
Davidson, TN 15 11 4 62% 57% 75%
Douglas, NE 13 8 5 58% 58% 59%
Erie, NY 25 18 7 60% 65% 47%
Fulton, GA 15 10 5 39% 34% 48%
Harris, TX 28 19 9 33% 27% 44%
Jefferson, AL 12 9 3 54% 48% 71%
King, WA 23 18 5 41% 37% 58%
Los Angeles, CA 34 34 0 28% 28% N/A
Mecklenburg, NC 15 9 6 73% 74% 72%
Miami-Dade, FL 29 26 3 25% 24% 30%
Milwaukee, WI 5 3 2 67% 71% 61%
Multnomah, OR 30 17 13 43% 40% 46%
New Haven, CT 13 11 2 65% 65% 66%
Pima, AZ 13 12 1 43% 41% 66%
Queens, NY 37 33 4 40% 38% 60%
Salt Lake, UT 13 8 5 66% 60% 74%
NOTE: No Medicare Advantage PPOs were offered in Los Angeles County in 2015. All percentages are not weighted by the number of enrollees in each plan.
SOURCE: Kaiser Family Foundation analysis of Medicare Advantage plans’ physician networks in 20 counties, 2017.
Report

The Henry J. Kaiser Family Foundation Headquarters: 2400 Sand Hill Road, Menlo Park, CA 94025 | Phone 650-854-9400
Washington Offices and Barbara Jordan Conference Center: 1330 G Street, NW, Washington, DC 20005 | Phone 202-347-5270

www.kff.org | Email Alerts: kff.org/email | facebook.com/KaiserFamilyFoundation | twitter.com/KaiserFamFound

Filling the need for trusted information on national health issues, the Kaiser Family Foundation is a nonprofit organization based in Menlo Park, California.