Table notes:  

1. "Heart Failure" or "Weak Heart" refers to weak pumping because of muscle deterioration, stiffness, leaking valves, etc. Medical terms for this include heart failure, congestive heart failure, cor pulmonale; or cardiomyopathy. It is not the same as a heart attack or heart stopping.
2. A Congressional agency, MedPAC, confirms that the penalty per excess readmission [Col A] = "Payment rate for the initial DRG [Col B] ... × 1 / national readmission rate for the condition  [Col D] " (p.99). So if readmission rates fall, penalties rise proportionately, and the only way hospitals as a group can reduce penalties is to reduce their variation around the national average or reduce the number of admissions  Click for other details and sources and timing of penalties.
3. Penalty for Knee or Hip Replacement is large ($250,000), since the initial payment is big, and readmissions are rare, so the multiplier is big. It applies to elective replacements, not those done for broken bones. 
4. Updated September 2017.

​Those hospitals have large penalties because of a combination of their large size and the patients they treat, who need extra care. They do NOT deserve the penalties. They are penalized for giving thorough care.

$6,250,000 FL Florida Hospital Orlando
$3,940,000 MA Southcoast Hospital Group, Inc Fall River
$3,430,000 CT Yale-New Haven Hospital New Haven - They think the penalties are a good idea and calculate them for Medicare.
$3,010,000 FL St Lucie Medical Center Port St. Lucie
$2,630,000 FL North Florida Regional Medical Center Gainesville
$2,610,000 FL Lakeland Regional Medical Center Lakeland
$2,290,000 NJ St Joseph's Regional Medical Center Paterson
$2,270,000 MS Forrest General Hospital Hattiesburg
$2,190,000 MA Beth Israel Deaconess Medical Center Boston
$2,080,000 PA Thomas Jefferson University Hospital Philadelphia
$1,780,000 MI Beaumont Hospital, Royal Oak Royal Oak
$1,730,000 NJ Virtua West Jersey Hospitals Berlin Berlin
$1,710,000 NY Vassar Brothers Medical Center Poughkeepsie
$1,660,000 NY Montefiore Medical Center Bronx
$1,610,000 VA Reston Hospital Center Reston
$1,610,000 IL Presence Saint Joseph Medical Center Joliet
$1,600,000 MA Lahey Hospital & Medical Center, Burlington Burlington
$1,550,000 NY Orange Regional Medical Center Middletown
$1,540,000 NY Long Island Jewish Medical Center New Hyde Park
$1,540,000 NJ Robert Wood Johnson University Hospital New Brunswick
$1,530,000 MI Beaumont Hospital - Dearborn Dearborn
$1,490,000 NY North Shore University Hospital Manhasset
$1,440,000 FL Lee Memorial Hospital Fort Myers
$1,430,000 TX Memorial Hermann Hospital System Houston
$1,420,000 MO North Kansas City Hospital North Kansas City
$1,420,000 SC Grand Strand Regional Medical Center Myrtle Beach
$1,340,000 TN Baptist Memorial Hospital Memphis
$1,290,000 NY Good Samaritan Hospital Medical Center West Islip
$1,290,000 CA Oroville Hospital Oroville
$1,280,000 PA Pinnacle Health Hospitals Harrisburg
$1,250,000 FL Leesburg Regional Medical Center Leesburg
$1,240,000 NJ Kennedy University Hospital - Stratford Div Stratford
$1,190,000 NJ Jfk Medical Ctr - Anthony M. Yelencsics Community Edison
$1,190,000 MO Centerpoint Medical Center Independence
$1,180,000 VA Cjw Medical Center Richmond
$1,170,000 NY Mount Sinai Beth Israel/Petrie Campus New York
$1,170,000 VA Chesapeake General Hospital Chesapeake
$1,170,000 FL Baptist Medical Center Jacksonville Jacksonville
$1,160,000 CA Regional Medical Center Of San Jose San Jose
$1,150,000 IL Advocate Christ Hospital & Medical Center Oak Lawn
$1,140,000 AL D C H Regional Medical Center Tuscaloosa
$1,140,000 DE Bayhealth - Kent General Hospital Dover
$1,130,000 NY New York-Presbyterian Brooklyn Methodist Hospital Brooklyn
$1,130,000 MI Mclaren Flint Flint
$1,120,000 AZ Northwest Medical Center Tucson
$1,120,000 FL Lawnwood Regional Medical Center & Heart Institute Fort Pierce
$1,100,000 NJ Cooper University Hospital Camden
$1,090,000 DE Christiana Care Health Services, Inc. Newark
$1,090,000 NY Huntington Hospital Huntington
$1,070,000 MA Lowell General Hospital Lowell
$1,070,000 MA South Shore Hospital South Weymouth
$1,070,000 MO Mercy Hospital Springfield Springfield
$1,070,000 MI Edward W Sparrow Hospital Lansing
$1,050,000 NV Sunrise Hospital And Medical Center Las Vegas
$1,050,000 PA Aria Health Philadelphia
$1,040,000 TX East Texas Medical Center Tyler
$1,030,000 IL Franciscan Health Olympia & Chicago Heights Olympia Fields
$1,000,000 NY Maimonides Medical Center Brooklyn

If we want legitimate patients treated, how can we penalize their hospitals? Faced with the level of penalties being imposed, hospitals cannot afford to treat many seniors. There are also incentives against treatment in some of the other ratings of hospitals.

Measuring and rewarding medical providers can backfire and reduce quality by reducing motivation (see a very good, broad article on these effects). Studies find that doctors avoid treating risky patients, when there is public reporting of outcomes. 75% of interventional cardiologists decided not to treat a patient due, at least in part, to protecting the doctor's or hospital's success rate. 74%  sometimes or often delay treatment to see if the patient dies first. 52% worry their superiors won't support a decision to treat risky patients.

Because of the readmission penalties, all hospitals try to be below average on readmissions, which makes the average get smaller (8% smaller in 2013; goal is 20% smaller, p.292). Faced with moving targets, hospitals cannot afford these penalties. They need to prevent as many readmissions as possible, often by emergency treatment without hospital admission, or brief admissions for observations instead of full treatment, or treating fewer patients for these conditions in the first place. If a risky patient is not admitted, s/he can't be readmitted.

The American College of Surgeons has warned Medicare about "the potential that these hospitals will decrease their care for such patients, thereby creating an access issue."

The latest data and several studies show that readmissions prevent deaths, so penalties are deadly. The American Hospital Association reported in Trendwatch September 2011, "mortality is inversely related to readmissions."

Dr. Kripalani of Vanderbilt University asks, "which would we rather have -- a hospital readmission or a death?"

Doctors Krumholz, Lin and colleagues in the Journal of the American Medical Association Feb.13, 2013 reported a 17% correlation between higher readmissions and lower deaths among heart failure patients. These are the same Yale authors who develop Medicare's readmission data, yet their own hospital cannot avoid readmissions. Yale-New Haven Hospital did 253 hip and knee replacements and will pay a quarter of that revenue as a readmission penalty.

Doctors Gorodeski, Starling and Blackstone of the Cleveland Clinic showed with a graph in the New England Journal of Medicine July 15, 2010 that hospitals with higher readmissions after heart failure treatment had significantly fewer deaths among the patients.

Hospitals are disclosing the financial risks of penalties in bond disclosures (p.25).

Evaluations have shown limited results.

Researchers at Columbia and Yale found that even an extra day of hospital treatment for pneumonia or heart attack saves thousands of lives (Table H). So reducing access to hospital treatment will be deadly.

Table H. Lives Saved by More Hospital Treatment

Puerto Rico and Maryland are exempt from readmission penalties, and are still covered by Medicare. Links for Puerto Rico are below.

Foreign hospitals and doctors have their own payment systems for their residents, but still have prices for US travelers. They are not covered by Medicare, so hospital stays abroad do not incur readmission penalties, and do not count as costs in Accountable Care Organizations. Some US and international private insurance pays for foreign care. In any case costs are often much less than in the US, and specialized hospitals can give better care. Some of the best offer luxury as well as ordinary care.

Kiplinger (2017) says the book Patients beyond Borders, and its website have reviews of hospitals. I have not seen the book, but the website shows the size of each hospital, and often the total patients treated, though not the number of times doctors do any particular procedure, nor other measures of quality. You may have to ask.

NA=US Average not available to calculate percent. See link for costs abroad

Insurance premiums for Medicare and Medigap policies are described below.

The Wall Street Journal has a good free guide to looking for health care costs, no matter if you pay with Medicare, other insurance, or cash. Finding costs before you get treatment is hard. Dr. David Belk explains how about cost of office visits, including higher pay from HMOs than from fee-for-service.

Doctors' fees under Medicare are in the Specialists tab above. They show what Medicare pays, and the list price for each procedure from each provider. Medicare costs include the total paid by Medicare, supplemental insurance and patients.

Each patient's cost depends on his or her insurance. For those without insurance, the Medicare level is a starting point for negotiation. In order to know total costs, patients can ask the doctor's office whether an anesthesiologist, assistant surgeon or hospital fee will be needed.

Anesthesiologist fees are in the Specialists tab above.

Hospital fees for the most common 100 diagnoses are in hospital spreadsheets from Medicare, and are mapped nicely at ClearHealthCosts. 

For example the data files show that surgeon costs for knee replacement are typically around $1,500, assistant surgeon $300, anesthesiologist $200, and hospital costs (for "major joint replacement or reattachment of lower extremity") averaged $14,000 if there were no major comorbidities and complications (MCC), or $23,000 if there were.

For a few procedures (primarily imaging, tests, counseling, dental extractions or implants, cosmetic procedures), ZendyHealth gives (free) a range of local prices within a radius you choose. They offer you a doctor based on how much you want to pay ($49 referral fee). You cannot use insurance with the doctor, but Zendy helps you submit a claim to your insurance company, so your cost counts against the deductible. For these and other procedures they offer a free consultation. You have to pay their legal bills if there's a problem ("indemnify"), and accept arbitration. You have no choice of provider, and see the name assigned to you only after you have paid the referral fee. For example different MRI centers have different strength magnets, and you are likely to get the cheapest, weakest magnets, which give less precise images. If you have time to search the Specialists tab above, you can find the lowest price providers and negotiate directly.

Costs for treatments in North Carolina are available from Blue Cross/Blue Shield of North Carolina, based on their patients and their contracts with providers: bcbsnc.com/content/providersearch/treatments. These have actual costs for a treatment episode, including hospital and doctors. Very easy to access. The free system compares all providers within any radius of a zip code, up to the whole state. You can sort by cost, name or distance. However there are only 1,200 procedures, no info on how often each doctor does the procedure, voluminous output with typically 3 providers per screen, not downloadable, only North Carolina, no procedure codes, so it is hard to be sure what each item covers,  no lab costs or drug costs. Their data come from one year, but they don't say which year.

Doctors' incomes derive from the payments above and the volume each doctor does. Average incomes (after expenses)  by specialty range from $200,000 per year for Public Health and Pediatrics to $500,000 for Plastic Surgery and Orthopedics, with wide variation. Concierge (2% of doctors) and cash-only (5%) doctors earn slightly more than average, comparable to other self-employed doctors. 13% are direct primary care doctors, generally subscription-based but lower fees than concierge. Three quarters do not charge for no-shows, whatever their policies may say. 70% see patients for 45 hours per week or less, but they spend 10 hours or more on paperwork and administration.  A fifth to a quarter of most specialties would not choose medicine again if they had the chance. Wealth averages $1-2 million, depending on specialty and age. A quarter of doctors have over $1 million by the time they are 35, and two thirds do by the time they are 50.

​Most doctors at hospitals work for large groups (TeamHealth, Schumacher) which contract to provide hospitalists, radiologists, emergency doctors, etc. Some companies provide doctors to hundreds of hospitals (Envision + Amsurg). Hospital doctors earn $200,000 - $400,000 per year. About half feel fairly compensated. Only a quarter "regularly" discuss the cost of treatment with patients. Over three quarters would choose medicine again and the same specialty.

Doctors identified 25 procedures and tests which may be wasteful, since they usually have low value for patients. They were given to 25% to 42% of Medicare beneficiaries each year, depending on definitions, but they only cost 0.6% to 2.7% of total Medicare costs.

In the United States, 

• 13% of medical spending goes on people who die within a year Mt.Sinai
• 25% of Medicare spending goes on people who die within a year NBER

This is because most Medicare spending goes on sick people (!), especially very sick people (!!), and some die unexpectedly (!). However very little spending goes on those who are likely to die soon.

• 5% of Medicare spending goes on people likely to die within a year (50% or more chance) NBER
• About half the people with high costs got better after an expensive treatment: heart attack, cancer treatment, etc. Mt.Sinai
• About 40% of the people with high costs have chronic conditions, and death is unpredictable Mt.Sinai
• "[R]eliably predict who will die and therefore would not benefit from receiving intensive care... turns out to be extremely hard to do. In a recent article in Science, researchers used a sophisticated machine-learning prediction tool to identify patients who are most likely to die, and found that there is no group of people for whom death is easily predictable." The Science article supplement table S2 shows that only 1% of Medicare enrollees had more than 50-50 chance of death within a year (46.6%-53.4% chance). So half of these did die, representing 0.5% of Medicare enrollees. The average death rate among the other 99% of enrollees was 4.6%, representing 4.5% of enrollees. So a total of 5% died, and most would not be predicted to die within a year. Similar data for hospitalized Medicare patients are in table S5.

Vox quotes economists that the US health care prices per item are abnormally high. So other countries get more health care for less cost:
"When you’re paying the highest prices in the world for basic services, for scans and drugs, it will undoubtedly be a struggle to provide all citizens with health care...

• Americans go to the doctor four times each year.
• Dutch people go to the doctor, on average, eight times each year.
• Germans make 9.9 annual doctor trips.
• Japanese residents clock in an impressive 12.8 doctor visits each year — more than three times the frequency of their American counterparts...

When Americans do go to the doctor, we tend to have less face time or interaction with our providers.

• The average hospital stay, for example, is 5.4 days in the United States.
• This puts us roughly in line with New Zealand and Norway (5.2- and 5.8-day averages, respectively) and with much shorter stays than Canadians (7.5 days) or Germans (7.8 days).

The real culprit in the United States is not that we go to the doctor too much. The culprit is that whenever we do go to the doctor, we pay an extraordinary amount."

A 2018 JAMA article compares insurance systems, and a 2008 JAMA article shows the same cost issues in 2004-6, with US doctors' salaries double the level in other countries.

​DailyKos has more detail on the range of costs.

• "America’s economic competitors discovered years ago and still share today... Whether negotiated directly or through a national association of insurers, the government sets the prices for prescription drugs, tests, treatments, hospital stays, and pretty much everything else...
• Economics, after all, is the study of the allocation of scarce resources... In the face of the infinite “wants” for healthy citizens, financially secure families, well-compensated practitioners, and strong profits for private companies of all stripes, societies must choose how and why to distribute discomfort and dissatisfaction to some or all of the constituents."

​​And DailyKos quotes the Commonwealth Fund comparisons:

 Related Topics:
Hospitals' Financial Data
Medicare Payment Formula for Hospitals

Patients' premiums depend on income and what coverage they want.

The government pays a lot for people at all income levels. Medicare Part B (doctors) and Part D (drugs), are not paid by the payroll tax, and are paid by premiums and government aid. (Part A, hospitals, is paid by the payroll tax.) Currently the Part B premium is $105 per month per person, and the cost is 4 times as much, $420 per month, so taxpayers pay a 75% subsidy. Premiums go up with income and subsidy is reduced, in several bands of income, but even the highest income participants get 20% subsidy.

The current premium is about 2% of income (red line above). It is

• 1.5% to 2.5% of income between $101,000-$537,000 for couples
• 1.5% to 2.5% of income between $50,000-$270,000 for individuals.

People below that pay more; people at the top pay less (in percentage). The advantage of bands is that premiums generally do not change for small changes of income, simplifying administration a little.

The Bipartisan Policy Center recommends starting bands at lower incomes (p.59 of full report), which result in higher premiums (and lower subsidies - green dashes above):

• 2.3% to 4.1% of income between $62,000-$349,000 for couples
• 1.7% to 3.1% of income between $41,000-$233,000 for individuals

People below that pay more; people at the top pay less. The plan would raise $6 billion per year.

Kaiser summarizes a variety of 2014 Budget proposals involving 15% increases in the premiums paid by high income participants, starting the first band lower, and slowly lowering all bands by not adjusting for inflation for several years (red dots above). Premiums would be:

• 1.6% to 2.7% of income between $93,000-$576,000 for couples
• 1.6% to 2.7% of income between $47,000-$288,000 for individuals
  

People below that pay more; people at the top pay less.

A Tucson blogger recommends charging 5% of income, up to the full cost (purple line above). Dots show bands of income, where people pay

• 4.25% to 5.75% of income between $44,000-$237,000 for couples
  
• 4.25% to 5.75% of income between $22,000-$118,000 for individuals
  

People below that pay more; people at the top pay less. She does not estimate the savings. A spreadsheet here estimates the savings from this proposal at $19 billion per year, using IRS counts of people by income, and consistent assumptions.

This option charges low income people the current $105, since Medicaid already pays the premium for most of them. Dropping the premium to 5% for low income people would cost Medicare more, but save an equivalent amount in Medicaid assistance, so the $19 billion overall savings would remain. It is far more than the $1.5 billion saved by the readmission penalty. Incomes can be adjusted for cost of living (purchasing power parity) by using US government locality pay. AARP presents arguments for and against basing premiums on income.

In the spreadsheet you can try different percentages and bands. A 3% charge could have bands of income where people pay

• 2.5% to 3.5% of income between $72,000-$402,000 for couples
  
• 2.5% to 3.5% of income between $36,000-$201,000 for individuals

People below that pay more; people at the top pay less. The plan would raise $6 billion per year.

The graphs show subsidies people would receive from various proposals. The current Medicare subsidy is large, even at incomes well over $100,000. The government does not subsidize food or housing for people at those incomes. The highest income limit on Food Stamps is $15,000 for one person, $20,000 for two; in subsidized housing it is $55,000 for one, $63,000 for two (Honolulu). Housing tax benefits do go to higher incomes, but people still have to pay the basic cost themselves. Why does the government make such large direct payments for health insurance for people with incomes over $100,000?

ACOs are groups of health providers who get waivers of rules on competition and kickbacks. They also may get paid more if they reduce spending on their Medicare patients, and sometimes patients of other insurers.

Patients do not sign up. Medicare tracks 9 million patients who get most of their primary care from an ACO. Medicare shelters ACOs from antitrust rules. When an insurer saves money on ACO patients, the ACO can keep half the savings, but when the insurer pays more for ACO patients, the ACO often keeps all the extra income.  There were 32 million patients overall in ACOs, Medicare plus private, on 3/31/2017, 10% of the US population. A recent JAMA article maintains the myth that ACOs' goal is cutting costs, rather than reducing competition, and reducing care for expensive patients.

Medicare provides a current list of ACOs updated through 2017.

• When you reach the list, you can get it as a spreadsheet by clicking the pale blue "Export" button in the upper right corner. I did this and added other information: aco2017.xls
• Before Medicare provided that list, I had an earlier list of ACOs (or spreadsheet), updated to 2015, based on Medicare press releases here. It includes some quality measures from here, but is not as up to date as Medicare's list.

• Medicare describes ACOs here, describes related efforts here, and reports some quality measures here.

Health providers who join an ACO have incentives to refer to affiliated specialists so payments are kept within the group. Quality measures slightly discourage hospital admissions (pp.13-15) and readmissions (p.11), but if the doctors' practice is owned by a hospital, extra income for the hospital more than makes up for any quality measures. They face no loss from patients' deaths or from diseases getting worse.

Patients are poorly informed about which doctors are in an ACO, and about their doctors' incentives. Medicare took three years to  answer my request for some of this information under the Freedom of Information Act.

• A doctor in California, Dr. Weinmann, has written several times about hospitals' and ACOs' power over doctors, "if Physician A has ordered diagnostic testing and treatment averaging $25,000 per patient  while Physician B has made the hospital or foundation pay $50,000 per patient, the physician who made his plan pay twice as much for patient care is less likely to be offered a renewal contract."
• A retired urologist in Minnesota, Dr Geist, has written that ACO doctors are “double agents playing the dual role of caregiver and insurance underwriter"
• ACOs can monitor their referrals to outside specialists and choose specialists based on costs, outcomes and patient satisfaction. Patients can go elsewhere but most follow the referral, thinking that quality is the main or only criterion.

According to Medicare, patients who don't want a doctor with ACO incentives and pressures can "seek care from another provider." If they want to keep their doctor, they can avoid cuts by dropping Medicare part B doctor coverage and using other insurance, or they can accept that the doctor has incentives to recommend less expensive treatments and hospice. They can get second opinions from doctors who have not joined an ACO, who will have less incentive for cost cutting.

Harvard's Dr Herzlinger says, 

• "20 percent of patients generate 80 percent of the costs... if I were a [heart failure patient], I would see a heart failure specialist. If I were an advanced type II diabetic, I would go to the Joslin Diabetes Center, not a primary care provider."
• Patients need "provider organizations that deliver highly specialized care for a certain group of patients, such as those who have diabetes, cancer, or congestive heart failure. You need specialists for that. They are the opposite of organizations, such as ACOs, that do everything for everyone."

Dr. Prince, CEO of Beacon ACO in New York, presciently said before Beacon became an ACO, "If they’re going to put the risk back on to the ACO and onto the physician, it’s going to be more difficult and we could start self-selecting which patients we want to include in our ACO." High cost patients can get more care outside an ACO.

61% of doctors plan to stay out of ACOs. 24% were members in January 2014, with another 10% planning to join in 2014. 91% of kidney dialysis patients are not in ACOs, so Medicare is setting up other doctor groups (End Stage Renal Disease Seamless Care Organizations - ESCO) just to cut costs for dialysis patients. "Members must place their fiduciary duty to the ESCO before the interests of any ESCO participant."

Medicare patients who use an ACO doctor cannot opt out as long as they have Medicare Part B, though they can opt out of letting that doctor see their Medicare claims from other doctors. (42 CFR 425.708). ACOs can maximize their inexpensive patients by reminding them to come in for checkups. 

• "it is important to keep our healthiest (least expensive) patients active and engaged. Reaching out to these patients allows a practice to generate revenue and offer preventive or wellness opportunities to patients. In addition to the benefits to the practice, there are organizational benefits as well. Maintaining the attribution of these lower cost patients to our population pool acts to depress our overall spend while we deliver high quality care."

ACOs are largely sheltered from restrictions on antitrust, kickbacks, and referrals to financially related providers 

• Medicare and Justice have given ACOs waivers.
• The reason many hospitals are at the center of ACOs may be these waivers of referral and antitrust rules, resulting in higher, not lower costs. 
• The government says, "The multiplicity of waivers is intended to afford flexibility to ACOs in varying circumstances." 
• If they were not sheltered, violations could cost hundreds of millions in lost income and penalties (Drakeford v. Tuomey), so the ACO shelter is supremely valuable.
• The government is also trying to loosen privacy on substance abuse, to ease the efforts of ACOs to share information on patients.

An April 2011 editorial in the New England Journal of Medicine (NEJM) said most medical groups spend money to set up an ACO, and cannot profit from it, unless they cut costs 20% or more. The large number of ACOs, all signing up after the editorial was published, shows managers want the kickbacks, referrals and antitrust waivers described above, or expect to cut spending on your care at least 20%, neither of which is desirable.

A July 2015 study in Annals of Family Medicine found that primary care doctors received very little pay for quality, whether they were in ACOs or not:

• A quarter got all their compensation based on the services they billed ("productivity").
• A quarter got all their compensation from straight salary (no word how it was set)
• The rest had a mix. ACO doctors' compensation averaged 49% from salary, 46% based on billable services, 3% based on "quality", and 1.5% other, almost the same as non-ACO doctors.
• The study had a 50% response, and excluded solo practices. 

​
There is a revolving door between ACOs and the federal Department of Health and Human Services. 

• The so-called "most influential" physician executive in the country moved from senior manager at Geisinger (part of Keystone ACO), to director of Medicare's Innovation Center, to CEO of CHE Trinity Health, which is part of 5 ACOs and is forming 11 more, and is carefully navigating anti-trust rules in Michigan.
• Another executive moved from manager of a NY program using commercial electronic health records, to federal coordinator of health records, to a company setting up and managing ACOs. He took a key executive with him from the federal government. (Under his leadership as federal coordinator, electronic records expanded rapidly in hospitals and doctors' offices, without adequate security.)

​HOSPITAL POWER AND COST

A January 2012 discussion reported in the Wall Street Journal included a former Medicare administrator, Scully, saying ACOs would not work and would be dominated by locally powerful hospitals forming ACOs. "The biggest flaw with ACOs is that they are driving more power to hospitals—not to doctors. Very scary, and I am a hospital guy. The goal of ACOs was to organize doctors to focus more on patients and keep the patients out of hospitals. Instead, doctors are selling practices to hospitals in droves."  A Virginia professor, Goldsmith, said ACOs had been tried 2005-2010 and failed, the same pattern noted by NEJM.

The professor, Goldsmith, also has written, "Hospitals and systems that became powerful in the marketplace through mergers and acquisitions aggressively shifted costs onto private insurers." Austin Frakt, a VA health economist and Boston professor writes, "If ACO formation proceeds with few checks... lower public-sector costs but higher private-sector premiums" are likely.

​In fact integrated health care delivery networks (IDNs) raise both public and private costs. A 2015 study found "there is growing evidence that hospital-physician integration has raised physician costs, hospital prices and per capita medical care spending... Diversification into more businesses is associated with negative operating performance. This is consistent with the management literature, which shows that diversification increases a firm’s size and complexity, in turn increasing its cost of coordination, information processing, and governance/monitoring."

​A 2017 report says, "Not only did sixty drug companies combine into ten, but hospitals, outpatient facilities, physician practices, labs, and other health care providers began merging vertically and horizontally into giant, integrated, corporate health care platforms that increasingly dominated the supply side of medicine in most of the country... Even nominally independent surgeons, for example, can’t stay in business if the only hospital in town won’t grant them admitting privileges, or if it grants “affiliated” surgical teams better terms... A full 40 percent of all hospital stays now occur in health care markets where a single entity controls all hospitals... not a single highly competitive hospital market remains in any region of the United States."

Hospitals are rapidly buying doctor practices, so the doctors become employees and refer patients to the hospitals, not necessarily saving money, but strengthening the hospital. 

• A November 2012 article in the New York Times, supported by hundreds of comments from doctors and nurses, reported on hospitals buying doctor groups, raising prices, pressuring doctors to maximize billings, then paying doctors bonuses if they discharge quickly, after the hospital earns its flat payment for the diagnosis.
• Only 53% of employed doctors think patient care is better from employed doctors than self-employed doctors. 
• Over half get pay based on their productivity or profits, rather than straight salary, and 39% of these are dissatisfied with their productivity targets. 
• 45% are expected to see 16 or more patients per day. 
• 49% are satisfied with their autonomy at work (Medscape report pp. 13, 16, 17, 20, 23). 
• 90% of hospitals hired temporary doctors in 2013, up from 74% in 2012, to fill in for doctors who quit or went on vacation.

Some doctors leave hospitals when the employment contract becomes unsatisfactory.

Medicare has lost money on ACOs every year. In each year some ACOs raise Medicare's costs, others get small bonuses. The most expensive ACOs look as if they save the most money, but their costs are still higher than less expensive ACOs, which do not get bonuses. A 2018 report analyzes costs, risk adjustment and alternatives.

An ACO gets a bonus payment of up to 50% of the cost savings from Medicare Parts A and B for its patients. By 2015, half the ACOs have raised Medicare's costs, an average of 3%. The other half of ACOs have cut costs, an average of 4%. Among all 392 ACOs, 30% cut costs enough to earn a bonus. The bonus received by these ACOs averages 3% of their costs. During 2014 there were 353 ACOs, and a quarter, 97, saved enough money to get bonuses, though not necessarily enough to pay back the set-up costs. During 2012, the initial year, there were 114 ACOs,  and a quarter, 29, saved enough money to get bonuses.

A rational ACO with professional management knows these 30% odds of getting a 3% bonus, which seems so small that it is unlikely to be the reason for forming an ACO. The other main benefit of an ACO is obtaining federal waivers.

Three ACOs have signed up for a version of the program where they can get up to 60% of the savings, but they share losses. One cut costs 17% (third biggest cut in the country), giving them a 12% share. One cut costs one percent, and the other raised costs half a percent, not enough to share. They have an incentive to avoid expensive treatments, which give them losses.

Theoretically bonuses can be large, so some ACOs may cut aggressively, like the one which cut costs 18%. Five small ACOs (average 7,000 patients) in Florida and Texas cut costs over 15%, averaging $1,000 per patient.. A cancer practice saved $1,000,000 per year per doctor, without even signing up for an ACO.

Detailed steps and definitions for calculating cost savings and risk adjustment issues are in another section.

Each year some doctors and patients in any ACO will chance to have above-average costs, reducing the bonus for everyone else. How will the ACO and peer pressure penalize these doctors and their sicker patients? The articulate Dr. Prince, CEO of Beacon ACO, says about their doctors, 

• "We’ll have data and see who the outliers are, and there are teeth in the agreement," he says... "If we do go down the road of the ACO, everyone needs to be rowing in the same direction. It won’t make sense to be a high utilizer and gaming the system."

The AMA has an article about the difficulties in distributing bonuses. Another AMA study found little connection between ACOs and expenses: 62% of ACO doctors' costs were on non-ACO patients. 68% of specialist visits by ACO patients were to non-ACO doctors. 33% of ACO patients were assigned to 2 different ACOs in 2 successive years. An AMA study as long ago as 2009 said, "the results from this study call into question the wisdom of pay-for-performance programs and quality reporting initiatives that focus on differentiating the value of care delivered to the Medicare population by primary care physicians."

Doctors are seeking ways to avoid these programs. A third are in ACOs. 61% plan to stay out of ACOs. 6% refuse insurance and 3% charge concierge subscriptions starting at $600 per adult per year and $120 per child, or more typically $1,500 per adult. 10% of Texas doctors do not take insurance, and instead charge for each visit, starting at $50 per visit.

Within each measure, the ACO gets only partial points if it is below the 90th percentile (p.67899, see graph below), so most will not get the full 50% (60%) of cost savings. Many of the screening standards are so easy that ACOs and other doctors will be clustered closely. The circulatory and hospital admission standards are the main "quality" measures where ACOs may distinguish themselves.

Penalties for readmissions after a nursing home stay will reduce the number of good nursing homes willing to accept risky patients, as well as deterring needed hospital treatment. More discussion and evidence are in the Nursing Home section.

Patient surveys have pros and cons. The surveys are here.

Electronic health records are problematic, since they have enabled vast breaches of medical privacy for 30,000,000 patients. Great systems are rare, though ideally they would show key information clearly in the way that each clinician needs it. Bad systems are not read by clinicians, are full of errors, generate erroneous prescriptions, and interrupt doctors when listening to patients.

Checking medications at an office visit is problematic, since hospital stays are the main cause of prescription changes, and the office visit is too late. Medicare says (p.19), "28% of chronic medications were canceled" during hospital stays, so immediate coordination is important and needs to be required. Office checks are also incomplete, since the patient rarely knows what medicines are given at the dentist, dermatologist, dialysis center, chemotherapy session, and other specialist locations.

Medicare has a ceiling to protect patients from doctors' excessive cost-cutting incentives. (pp.67935-6). They worry if the ceiling is adequate.

• The reward is limited to 10% of the baseline total Medicare spending on their patients (15%, if 2-sided). So an ACO gets the lesser of 10% of total Medicare spending or 50% of savings.
• Medicare originally proposed the reward ceiling at 7.5%, not 10%, to avoid "an overly large incentive such that an ACO may be encouraged to generate savings resulting from inappropriate limitations on necessary care" (p.67935). 

• The industry cited the NEJM editorial mentioned above, and said it needed a bigger ceiling than 7.5% in order to invest in setting up ACOs (pp.67935-6). Which means the industry expects to hit the ceiling. Medicare gave them the higher ceiling.
  1. ACOs which earn savings at the maximum 50% rate will hit the ceiling when they cut 20% of Medicare spending on their patients (50% of 20% is 10%)
  2. ACOs with median quality scores earn 35% of savings and will hit the ceiling when they cut 29% of Medicare spending (35% of 29% is 10%)
  3. ACOs willing to risk the lowest quality scores (30th-40th percentiles) earn 27% of savings and will hit the ceiling when they cut 36% of Medicare spending (27% of 36% is 10%)

• As noted above, industry comments to Medicare show they want to reach the ceiling, so they hope to cut costs the full 20%-36% allowed by the ceiling (30%-55%, if 2-sided).

• Medicare says, "Many health care researchers believe that the rate of unnecessary health care is more than... 10 percent" (p.67935). They did not say how many researchers believe the rate of unnecessary health care is 20% to 55%. Patients who think 20-55% of their own health care should be omitted can go to ACOs, knowing these doctors have incentives to reduce spending that much.

The first graph below shows what fraction of the theoretical bonus an ACO gets, depending on its quality ranking (p.67899). The second graph shows threshold savings ACOs must reach to earn any bonus (p.67928-9). Large ACOs need to save 2% to earn any bonus, while small ACOs need to save at least 3.9%, to avoid payments for random variation. (ACOs which risk losses need 2% savings regardless of size.)

A doctor with a $21 million grant from Medicare to achieve "lower costs with better outcomes" says "Significant improvements in cost and quality may not be felt until fee-for-service falls below 50% of provider reimbursement," which means a ceiling on bonuses not 10%, but over 33%, leaving 33% for Medicare and 33% in direct costs, which is far too little to pay for needed care.

From a doctor's point of view, doctors are subject to "whatever cost-savings techniques the ACOs use, e.g., not accepting doctors who have too many elderly patients  or patients with expensive chronic diseases. The days of searching out rare and unusual diseases to care for are over: these unfortunates will be obliged to find  whatever comfort is available under the nearest bus. If the ACO is well managed from a fiscal perspective, providing participants will share in  the savings  as a second source of income. Quietly, with as little fanfare as possible, physicians and hospitals will be encouraged to avoid the sickest, oldest, and most complicated patients."

Cancer doctors have been especially concerned about quality and cuts, since cancer represents 1% of patients and 10% of medical costs.

Dr. Cary Presant, chair of the Medical Oncology Association of Southern California, said, "The unspoken word is 'try and find a way to get these patients to not utilize these drugs, and consider whether this patient who is going to be a big expense should go into a hospice earlier rather than later.'" Groups can also discourage expensive patients by limiting their appointments or recommending palliative care.

A similar concern applies to Medicare Advantage (Part C) plans. Newell Warde, RI Medical Society director described a big Medicare Advantage plan dropping specialists who served expensive patients. "They look at your patient mix... They’re not just dumping doctors. They’re dumping patients. These may be expensive patients."

Case Western researcher, Anish Mehta, and Dr. Roger Macklis, a Cleveland Clinic cancer doctor write, "cancer-specific guidelines are not included in the quality measurements... Oncologists may feel pressured to curb the use of costly drugs and expensive procedures. New treatments from across all branches of oncology—from proton therapy to hyperthermic intraperitoneal chemotherapy to sipuleucel-T—will now reflect directly on the PCP... Pathway-driven medicine may lead to bare-bones cancer care, significantly reducing the universe of treatment options used by specialists."

A cancer lobbyist, Matt Brow, wrote, "There is another great risk that the ACO will not be held to delivering quality oncology care in any way, leading to the desire to see oncologists use the least costly type of therapy or no therapy at all.”

At the same time Medicare creates a shortage of cancer doctors by not training as many as the number retiring.

Aside from ACOs, insurance plans for non-Medicare patients are beginning to exclude specialized cancer treatment centers.

Informing Patients about ACOs

Special ACOs and Insurance Issues

Press Announcement

ACOs are entitled to exemptions from several laws, which for many years have promoted good care and prevented conflicts of interest:

1. Physician Self-Referral (Stark Law) prohibits doctors from referring Medicare patients to anyone who has a financial relationship with the doctor (some exceptions apply).
2. Federal Anti-Kickback Law prohibits payments for referrals under any Federal health care program. Safe harbors clarify what is and is not allowed.
3. Prohibition on Inducements to Beneficiaries prohibits gifts to Medicare or Medicaid patients to encourage them to use any particular provider (some exceptions apply)
4. Gainsharing Civil Monetary Penalty (CMP) has been amended so it no longer needs waivers in or out of an ACO. It now prohibits hospitals from paying doctors to limit medically necessary care for Medicare or Medicaid patients.

The waivers let hospitals and doctors in an ACO reward doctors for referring to certain specialists and hospitals, even if the doctor thinks the best referral would be somewhere else, as long as the ACO says they advance any "one enumerated purpose... we continue to define the purposes of the Shared Savings Program...

• promoting accountability for the quality, cost, and overall care...
• managing and coordinating care...
• encouraging investment in infrastructure and redesigned care processes for high quality and efficient service delivery for patients... we continue to interpret the purpose of “efficient service delivery” to include, among other things, appropriate reduction of costs to, or growth in expenditures of, the Medicare program, consistent with quality of care, physician medical judgment, and patient freedom of choice."

ANTI-TRUST

The Justice Department has issued policies protecting ACO members from anti-trust enforcement if they cooperate in an ACO with up to 30% of the local market.

A. SUPPRESSION OF PUBLIC INFORMATION

Many doctors, hospitals, therapists, etc. have joined Accountable Care Organizations (ACOs). These have important features which patients need to understand. However Medicare does not let ACOs tell patients about some features, and requires very unclear wording for the others. ACOs cannot write their own wording:

1. ACOs have waivers of the usual federal rules prohibiting payments for referrals and requiring free and open competition among medical providers. Patients are not told about the waivers at all.
2. ACOs may not list individual doctors participating in the ACO, if they belong as a group; only the group name can be listed, which patients may not recognize. It is thus hard to do a web search to see if your doctor belongs to an ACO. Patients need to ask the office whenever making an appointment, which seems impractical.
3. ACOs may not tell patients about contract terms between doctors and ACOs, "public disclosure shall not include the financial or economic terms of the arrangement.". Contracts also silence doctors if the ACO disapproves treatment, "contract language that physicians are precluded from discussing disagreements about compensation 'and other matters' with patients. The treating physician is told in contractual language that where 'disagreement cannot be resolved ... under no circumstances shall such disagreement be expressed to the Enrollee.' "
4. ACOs have a 30% chance of getting a small bonus if they cut spending on patient care. Medicare requires patients be told the focus is quality and coordination, "The ACO may share in any savings that result from providing you with high quality and more coordinated care."
5. Medicare and ACOs talk about quality care, without saying they ignore deaths, cures, or good or bad outcomes from most treatment (more in C.2, below).
   
6. Medicare reveals all of a patient's Medicare claims and diagnoses to each ACO the patient goes to. The patient can stop some but not all disclosures, as described in the following paragraphs:

Medicare has prepared a sign which is supposed to be posted at "a limited number of locations in each ACO" (Federal Register 11/2/11 p.67946) (42cfr425.312(a)). The sign virtually denies that Medicare discloses private information. It says:

• "You may not have to fill out as many medical forms that ask for the same information.
• Each of your doctors will not only know about the health issues they’ve treated, they will have a more complete picture of your health through talking with your other doctors."

Medicare has also prepared a notice which is supposed to be available in doctors' offices upon request, and the ACO may mail it to patients if the ACO chooses to. If not mailed, the only hint of its existence is on p.119 of Medicare and You Handbook, which pretends it has the same limited information as the sign, so people who can read the sign would have no reason to ask: "A poster with information about your doctor’s participation in an ACO will be displayed. At your request, the doctor will also give you this information in writing." This notice describes the loss of privacy. It is the only place which says patients cannot stop all disclosures, and that a patient's private information can go to several ACOs.. Patients have to ask for and hold onto this paper, since it was not on the web (until I got it by FOIA):

• "information will include things like dates and times you visited a doctor or hospital, your medical conditions, and a list of past and current prescriptions... 
• your information may also be shared with other ACOs in which any of your doctors or other healthcare providers participate.
• If you don’t want your information shared with these other [sic] ACOs, follow the instructions below...
• Even if you don’t want Medicare to share your personal information with us or with other ACOs for coordinating and improving the quality of your care, ... Medicare may share some of your personal health information with ACOs when measuring the quality of care given by healthcare providers at ACOs."

Medicare's FAQs, which are on many ACO websites, say the ACO will get metadata about your diagnoses, prescriptions and appointments:

• "Medicare will share information about your medical information [sic] with your doctor’s ACO, including medical conditions, prescriptions, and visits to the doctor...

• you may choose to have your name and other personal information removed from the information that Medicare shares with your doctor... [if not], your medical information will be shared automatically."

Medicare's website for patients does not say you can opt out (site for doctors does). It says:

• "doctors and hospitals will share information and coordinate your care...
• your doctors will get your medical information from Medicare to help them to know your medical history, including your medical conditions, prescriptions, and visits to the doctor

B. NOTICES AND SIGNS AT THE POINT OF CARE

A patient arriving for an appointment may or may not suddenly see a sign that the doctor is in an ACO, with minimal explanation. Medicare's theory is that the patient can "seek care from another provider". The patient is supposed to decide suddenly whether to see the doctor, based on minimal information, often sick and in pain, and maybe facing a cancellation fee.

• The ACO must "Notify beneficiaries at the point of care" and "Post signs in their facilities to notify beneficiaries" "that their ACO providers/suppliers are participating in the Shared Savings Program." (42cfr425.312(a)).

• "there will be a limited number of locations in each ACO in which the signs will need to be posted..."  (Federal Register 11/2/11 p.67946)
• The ACO must "Make available standardized written notices regarding participation in an ACO and, if applicable, data opt-out. Such written notices must be provided by the ACO participants in settings in which beneficiaries receive primary care services." (42cfr425.312(a)(3)).

On June 8, 2016 Medicare gave me these standardized written notices and signs, which must be displayed to patients when they get care. I had requested these copies under the Freedom of Information Act in July 2013. They include:

1. A sign to post (8.5 x 11 inches, 12-point type; such signs on the wall are very hard to read with bifocals)
2. A notice available for patients
3. A form for any patient to tell Medicare NOT to release certain information about the patient to the ACO
4. A form for any patient to change his/her option about releasing information
5. Instructions to ACOs about the notice, forms and signs
   
6. A small card which patients may show doctors, to ask them to share information directly with an ACO doctor
7. Instructions to patients about the information card
8. Instructions to ACOs about the information card
   

​Based on Medicare's rules, patients may have no effective notice. ACOs are permitted to mail notices to patients in advance. Otherwise patients will only learn about the ACO "when you visit the office. A poster with information about your doctor’s participation in an ACO will be displayed. At your request, the doctor will also give you this information in writing" (p.119 of Medicare and You Handbook, 2017, emphasis added). The poster/sign does not say that written information is available. The sign just says you can talk to your doctor or Medicare about it. Medicare in 2014 proposed making the sign more complex, with information on opting out of data disclosures (p.72789, 12/8/14), but the final rule (6/9/15) did not state whether they would change the sign, and disclosures are still not mentioned in the sign released in 2016.

1. "let your doctor know how you feel about sharing your medical information"
   
2. "Your doctor or primary care provider is participating in [ACO Name]"
3. "An ACO is a group of doctors, hospitals, and health care providers working together" [sign says they "voluntarily work together"]
4. [goal is] "high quality, more coordinated service and care."
5. "doctors and primary care providers to communicate more closely with your other health care providers"
6. "ACOs may share in the savings it achieves for the Medicare program"
7. "spending health care dollars more wisely"
8. "You Can Still Choose Any Doctor or Hospital"
9. "Your doctor may recommend that you see particular doctors or providers"
10. "Medicare plans to start sharing information with us about your care... dates and times you visited a doctor or hospital, your medical conditions, and a list of past and current prescriptions.  "
11. "shared only with people involved in giving you care."
12. "your information may also be shared with other ACOs in which any of your doctors or other healthcare providers participate"
13. "you can ask Medicare not to share information with us or with any other ACOs for care coordination and quality improvement purposes by doing one of the following"
14. "we need to get your decision by" [date, to prevent data sharing]
15. "you may choose to stop this information-sharing at any time in the future"
16. "Medicare will still use your information for some purposes, including certain financial calculations and determining the quality of care"
17. "Medicare may share some of your personal health information with ACOs when measuring the quality of care given by healthcare providers at ACOs."
18. "If you have questions or concerns, you can call [the ACO], make an appointment to see your doctor or primary care provider, or bring it up next time you’re in your doctor’s office."
19. "You also can call 1-800-MEDICARE and tell the representative you’re calling about ACOs, or visit www.medicare.gov/acos.html."

The main topics in these 19 points from Medicare are:

Revealing confidential medical information is the focus of items 1, 5, and 10-17.

The ACO's organizational structure, which may direct referrals to particular doctors, is the topic in items 3 and 9. However the material does not mention that doctors in ACOs can get kickbacks from  referrals and can refer where the ACO doctor has a financial interest. Medicare approved waivers of referral and kickback rules for ACOs. So Medicare patients in ACOs are no longer protected by normal rules against kickbacks and "self-referrals," where the doctor has a financial interest. Furthermore, the Justice Department has issued policies protecting ACOs from anti-trust enforcement, similarly left unmentioned in the written material.

Medicare mentions quality of care in items 4 and 7. Medicare has a narrow and changing concept of the quality measures it looks for. The patient is not given any link or place to see what Medicare means by quality.

The ACO's incentive to cut costs is mentioned in items 6 and 7, without mentioning:

• ACO's "share" in savings can be up to 50% in most ACOs
• A few ACOs can lose money when they give expensive care to patients, which is a further incentive to economize

Readers will decide for themselves whether Medicare's wording covers what patients need to know. On the other hand, ACOs cannot decide for themselves. Medicare requires ACOs to use Medicare wording when giving information about the ACO (42cfr425.310(c)(1)), so they keep tight rein on the public spin about ACOs. However non-ACO members are not constrained in what they publicize.

C. REQUIRED WEB PAGE

Medicare requires ACO websites to have a page or pages showing staff, quality measures and "Shared Savings/Losses" (Public Reporting Format), though the web page does not have to be linked from anywhere, and sometimes can only be found if you know to search for it. Some also have an older page about "How Shared Savings Are Distributed." The examples below are from the biggest ACO, but Medicare requires every ACO to have the same wording.

1. ACOs usually include group practices, and the required web page must list them by corporate name, like "Access Neurocare PC," not the individual doctors who work there.

2. Medicare and ACOs talk about quality care, but the quality standards do not measure deaths, cures, or good or bad outcomes from treatment. Patients will not realize how limited the quality measures are, since quality measures shown to patients must use Medicare's opaque wording, such as:

• CAHPS: Getting Timely Care, Appointments, and Information - 77.71 ...
• Risk Standardized, All Condition Readmission - 14.74 ...
• Angiotensin-Converting Enzyme (ACE) Inhibitor or Angiotensin Receptor Therapy - for patients with CAD and Diabetes or Left Ventricular Systolic (LVEF<40%) - 80.88
• CAHPS = Consumer Assessment of Healthcare Providers and Systems, PQI = Prevention Quality Indicator, LVSD = left ventricular systolic dysfunction, ACE = angiotensin-converting enzyme, ARB = angiotensin receptor blocker, CAD = coronary artery disease.

They do not say what the numbers mean, whether bigger numbers mean better or worse care (they vary), nor what the range is in different areas. . They also show other numbers labeled as "Mean performance rate (SSP-ACOs)" without saying SSP means a comparison to all Medicare ACOs (Shared Savings Program). ACOs are not allowed to clarify these labels.

3. On this same obscure web page Medicare requires ACOs to tell patients the how many millions of dollars are "Shared Savings/Losses" without definition, and without context on how this compares to total spending. For example:

• "Agreement period beginning 2012, Performance Year 2015: $33,537,591"

Medicare does not let ACOs tell patients that the ACO doctors can get paid more by using the cheapest treatment, rather than the one most likely to work quickly, and by referring patients to specialists who avoid expensive treatments.

Special Types of Accountable Care Organizations

About 30 ACOs are Pioneer ACOs (Phase 0 in list) with even stronger incentives to save money than the more common "Shared Savings." Differences include:

1. "The first two years of the Pioneer ACO Model are a shared savings payment arrangement with higher levels of savings and risk than in the Shared Savings Program.
2. "Starting in year three of the initiative, those organizations that have earned savings over the first two years will be eligible to move to a population-based payment arrangement and full risk arrangements that can continue through optional years four and five.
3. "Pioneer ACOs are required to develop similar outcomes-based payment arrangements with other payers by the end of the second year..." (Fact Sheet 9/12/2012 pp.6-7)  "50% of all revenues must be in ACO-like arrangement." This is significant only if Medicare less than 50% of their business. Medicare especially wants state Medicaid agencies to sign up (MedPAC 4/2013 p.11, details in Request for Application pp.13-14)
4. Pioneer ACOs do not have to exclude spending on patients above the 99th percentile, if they want to cut such spending among their patients. (Request for Application p.13)
   

Emphasis is added: cost-saving incentives are stronger in Pioneer ACOs than others, and non-Medicare patients will face the same pressures as Medicare patients for cheap care in these ACOs. The 30 Pioneer ACOs have 5 different payment structures (MedPAC 4/2013 p.10). Several were publicized initially, and ACOs could suggest others (Request for Application pp.8-10):

• Pioneer gets up to 50% of loss or gain, not to exceed 5% of baseline total Medicare cost
  

• Pioneer gets up to 60% of loss or gain, not to exceed 10% of baseline total Medicare cost
• Pioneer gets up to 70% of loss or gain, not to exceed 15% of baseline total Medicare cost

• Pioneer gets up to 75% of loss or gain, not to exceed 15% of baseline total Medicare cost
• Pioneer gets all of loss or gain, with flat payment per beneficiary per month, in years 3-5
• Actual percent up to 50%-75% depends on limited quality standards as in regular ACOs.
  

About 35 ACOs are Advance Payment ACOs ("a" in list) where Medicare lends money to start and operate the ACO.

• "If the ACO does not generate sufficient savings to repay the advance payments as of the first settlement for the Shared Savings Program, CMS will continue to offset shared savings in subsequent performance years and any future agreement periods, or pursue recoupment where appropriate." Advanced Payment ACO Model 1/10/2013 p.3
  

Once providers have received advance money, they will be very averse to giving it back, so pressure to generate savings will be even stronger than at the other Shared Savings ACOs, which do not have Advance Payments.

How Does Other Insurance Control Cost?

CIGNA health insurance has 66 ACOs, and Premier has 23. Blue Cross has them in New Jersey, and in an AQC program with 9 ACOs in Massachusetts. Aetna promotes the idea and has them in Arizona, San Diego, Maine, New Jersey (also here), Pennsylvania (also here), and Virginia (also here).

15% of people with health insurance from private employers have a flat fee per patient for each doctor, regardless of the amount of treatment provided (capitation).

Homeowners' insurance is like health insurance in that few people have big claims, and everyone wants low premiums. Companies frequently reword policies to avoid unexpected costs, and customers theoretically can compare policies, but often have gaps in coverage.

Many types of non-healthcare insurance have premiums based on experience or risk. These varying premiums (unlike Medicare) give incentives to reduce claims: car, workers' compensation, unemployment insurance, FDIC, etc.. The person buying the insurance thus chooses between higher risks, claims and premiums, or lower claims with lower premiums.

Some types of insurance collect property to reduce losses and provide a disincentive to claim: car, mortgage. Many claims are inherently unpleasant, so have incentives against overuse: sickness, death, car accidents, house damage, and theft.

Some insurance companies make it hard to collect: Social Security Disability Insurance initially  denies claims, and the disabled need to appeal. Car insurance is regularly rated by Consumer Reports on claims services. Private health insurance faces frequent complaints from patients and providers about difficulty getting payments. Long term care insurance has a risk of being similar, but there is too little experience to compare the insurers.

Moral hazard is the risk that an insured person or company will incur extra risks and losses because of having insurance. Banks are subject to moral hazard because of FDIC and bailout funds. However people rarely get sick or want more invasive tests voluntarily, so moral hazard rarely applies to health insurance.

How Do Other Entitlement Programs Control Costs?

Governments reduce budgets and cut services of entitlement programs overtly or covertly. Home health services and services for foster children are targeted and reduced.

Advocates limit cuts by asking voters or courts to insist on more money. Prisons are overcrowded until prisoners get courts to order improvements.

Medical care is subject to malpractice suits as an incentive against cost-cutting, though these are expensive with expert testimony, and many states restrict awards for suffering, so they are not as useful for elderly (Medicare) patients as they are for highly paid younger patients who can claim lost earnings. Class actions may be a way to aggregate enough awards to make legal enforcement worthwhile.

ERISA preempts many claims for damages against employer-provided health insurance, even HMOs, though not necessarily against independent doctors and hospitals (legal history to 2003).

Direct link to this page: foia.globe1234.com

• States have distinct rules for how you get information from state and local government, including schools, etc. (NFOIC, RCFP). State and local Courts have their own rules.
• MuckRock helps file and track FOIA requests to all levels of government in the US ($20 for up to 4 requests). They cooperate with FoiaMachine, free, which lets users track requests, while MuckRock offers a "full service" which submits requests, follows up, and posts results online.
• FOIA group is a commercial version ($125 or more per FOIA), and they were reported to have government contracts processing requests.
• Many other countries have FOIA sites using standard (Alaveteli) software, German and Austrian software or other (FOIAdvocates, FOIAnet). Most countries have better laws than the US.
• Using public government data is always faster than FOIA. Each US agency has an electronic "Reading Room" with its adjudication orders, policies, manuals, and commonly requested items. Politico used usaspending.gov to find the HHS Secretary's use of chartered jets. Oversight.garden has Inspector General reports. Agency websites have vast information.
• ​Individuals don't need a lawyer at any stage, even going to court, and organizations only need a lawyer in court. For hard cases many people value advice and representation by a lawyer, but if you don't want to spend that much, see the sections below on learning from similar cases and representing yourself in court.
• To find an expert lawyer, you can search for cases like yours in CourtListener, to see which private lawyers handle cases like yours, and see their work. 
• Success rates of lawyers: JusticeToolbox,
• Reviews by clients: Avvo (background)
• Ratings of lawyers by lawyers; most aren't rated: Martindale, BestLawyers, SuperLawyers 
• Some companies rate their outside counsel with Qualmet's brief questionnaires ($)
• ​Other search sites for lawyers: Findlaw, Nolo, Justia, Lawyers.com

• The public gets some information in 60% of FOIA requests, in a third of appeals at agencies, and in 60% of court cases. (spreadsheet)
  
• Fees  may cover searching and/or copying. The cost depends on whether they respond in time, and whether you will use it for business, informing the public, etc. (DOJ, and Muckrock)
• Model letters (NFOIC), and far more detail (which may not be helpful) at S.ai
  
• System to generate letters and track response (IFOIA from RCFP). Also MuckRock and FoiaMachine above
• Joint tracking and filing site for several agencies (FOIAOnline)
• You can often file on line, which is fast, but reaches the agency as plain text, with no control over the spacing or format. You can improve clarity by submitting a pdf, using certified mail if there is no way to submit it online or by fax. Email submissions are getting uncommon.
• Advice from Public Citizen, Unredacted, NSArchive, EFF
• Legal advice for reporters from SPJ and ASNE, discussed by CJR
• Requests under FOIA can specify a preference for email, fax or mail, to receive answers and handle questions from the agency. Contact addresses need to be kept up to date. If a decision takes time, the agency may ask if you are still interested, and close the file if you do not respond, though they have to reopen it if you respond soon after they told you to.
• FOIAmapper indexes the contents of federal record systems, and each agency's FOIA processing times, though they cannot always be up to date.
• FOIA.wiki has posts from the public on many Federal FOIA topics, including each agency 
• Ombudsman at the National Archives (OGIS) achieves action occasionally. They also have a compliance office which seeks voluntary compliance, and their advisory committee has minutes, transcripts, and livestreams.
• American Society of Access Professionals (ASAP, $50/yr) has introductory webinars ($50-100) and training conferences ($500-1,000) attended by government FOIA staff and frequent requesters (about half the income goes to the management company, Bostrum, and half to other program costs).

• The median time to answer simple requests is within 20 working days at 80-90% of federal agencies. Simple means under 500 or so pages from one location, but each agency has its own criteria.
  
• It will often be faster if you ask first for any records which may have already been released to others. These records have already been "pre-processed." If they've been released to several others, they should already be in the agency's online Reading Room. Agencies have internal indexes to check if they have released records covered by your request before. On the other hand if you already have those records, you can tell the agency just to send additional records, so you don't have to pay for duplicates.
  
• Here is the median decision time in 2016 for each agency which received over 300 requests that year. For parts of agencies (like FBI) and smaller agencies, see the annual reports, Table VII. A.

 
Will it be faster if you go to Court?

• If you don't get the information as fast as this table shows, you can go to court, with or without a lawyer, as discussed below. Many requesters get documents soon after filing suit, since the Justice Department assigns a lawyer to represent the government, and s/he takes a fresh look and often convinces the agency to release documents, rather than lose in court. There's an anonymous saying, "For a $400 filing fee I get a Justice Department lawyer who really knows the law as my FOIA officer!" You can find out about Rule 41(a), Voluntary Dismissal, if the case gets deeper than you can handle.
  
• When considering whether to give up, wait or sue, The New York Times says, "If requesters always shrug and walk away at that point, it means we are leaving it to FOIA bureaucrats to decide just how secret our government is going to be. That was never part of democracy’s plan. By suing regularly, we hope to achieve two things. We put agencies on notice that we will take them to court if our requests are not handled properly, and it gives us a shot at shaping the law through court decisions."
• A former State Department lawyer, now making FOIA requests for a nonprofit says, "where we distinguish ourselves is that by litigating, we solve one of the fundamental challenges that’s at the heart of the public records system, which is that whether through a lack of resources or obstruction, it’s very difficult to get information in a timely way.
  The lawsuit that we filed with the FCC was over net neutrality, and through that, we established ourselves as quite an advocate on that issue and interested in that issue. Since we’d already shown the FCC that we weren’t afraid to sue over net neutrality related documents, the FCC gave us the comment system crash records without a fight. That’s actually a really wonderful example of how our engagement in litigation identified us to the FCC as a requester who’s willing to go to court."
• It is possible to go to court if a request takes longer than (a) agency's median time, and (b) 20 working days. Courts expect first-in-first-out processing within each track (simple, complex, or expedited) (p.47, DOJ Litigation Considerations), so no request should take more than the median. NSArchive says, "it is productive to talk with the agency and wait a reasonable time for the agency to process the request." They do not say what is reasonable, so you can judge if the median is a reasonable time. DOJ guidance tells agencies themselves how to estimate reasonable times for completion, "Agencies that utilize multi-track processing can also consider the agency's average processing times for its various tracks. This information is readily available in the agency's Annual FOIA Report and on FOIA.gov.​"
• Some requests take much longer than the median, when requesters do not take the agency to court. 1% take over 14 months.
• Complex requests mean you want a lot of information, or from multiple locations, etc.  Median time to answer is within four months (84 working days) at three quarters of federal agencies. Agencies typically do not tell you if they put the request in the simple or complex track.
• Expedited requests mean you convinced them to process on a special track, usually because you need a lot of information for a deadline. This special track can take nearly as long or longer than complex requests. Each agency's regulations explain how to request expedited, if you think it will be helpful.
• Check the annual report. They're hard to follow, but table numbers seem standard: Table VII. A shows median days to decide at each part of an agency, such as FBI (in Justice) or Medicare (in HHS). Instructions tell agencies to count working days (p.8). I would generally feel comfortable waiting that many days, but going to court soon after, unless the agency convinces me a decision is imminent.
• Annual reports also show their approval rate: compare full and partial "grant" in table VI. B. (1) to number "processed" in table V. A. There's no detail, and this mostly reflects simple requests, so we don't know the approval rates of complex or expedited requests. It may be worth waiting longer for a simple request if the agency approves most of them, or if they reverse most denials on appeal (tables VI. A. and VI. B.) and decide appeals fast (table VI. C. (4)).
• Summary spreadsheet shows 2016 processing times and approval rates at cabinet departments and independent agencies.
• I've waited too long, but eventually I took them to court without a lawyer, and got the information quickly. 

• Searchable list of federal FOIA appeals in court, by agency, topic, judge, name, date, location, etc. (FoiaProject) has docket of each case and free copies of some documents.
  
• Searchable list of all federal cases. You can search on FOIA, name, agency, location, judge, etc. (CourtListener.com). Not as good a search as FoiaProject above, but it also has non-FOIA cases and free copies of many documents.
• All federal court documents are at Pacer, which is free for under 150 pages per calendar quarter, and 10 cents per page otherwise. Installing an app in Firefox or Chrome lets you add to CourtListener's free archive whenever you pay for documents from Pacer.
• Lists of FOIA appeals in court opened & closed (US Dept. of Justice) not as detailed as FoiaProject above, but more complete on results. ​
  
• State courts do not have good ways to search, but you can search state supreme court opinions and some private newsletters in CA, FL, IL, LA, MO, PA, TX and WV which report civil cases.
  
• The following graph summarizes final federal court outcomes on FOIA in fiscal year 2016. The public won some information by settlement or court decisions in at least 60% of court cases, and may have received some information even when the final court decision on the last information went fully to the government.

• The Federal Circuit Court for DC ruled in Payne v. US, 837 F.2d 486, 494 (D.C. Cir. 1988), that courts don't just order release of documents; they can stop an "impermissible practice" at any agency by declaratory and/or injunctive relief. A US District Court used this authority in 2011 to rule that agencies must provide estimated completion dates ( § 552(a)(7)(B) ).
  
• When a court orders legal fees, it can order investigation of FOIA staff who seemed arbitrary or capricious (p.125, DOJ Litigation Considerations) but not without ordering legal fees, so not in FOIA cases brought by individuals without a lawyer. The 6 times lawyers asked the courts for such investigations from 2013-2016, courts said no.
  

• Overview of issues when representing yourself: shorter from NOLO and PublicCounsel or longer from US District Court in N. California. There are also books.​
  
• The government usually settles. You can read my settlement negotiations, though details will vary in each case. ABA has strategic and ethical advice on settlements, and there are books.​
• Federal Practice Manual for Legal Aid Attorneys describes what to do in all stages of federal litigation, from drafting and filing the complaint, to trial practice and limitations on relief.
• Federal Rules of Civil Procedure online, or here, and printed and each district court's Local Rules.
• Filing any court case costs $400 (or free if you can't afford $400, Rule 24), plus a few dollars for copies and certified mail (4(i)). The government can charge more costs in some situations, under Rule 68, and Rule 54(d). The latter cost from 54(d), is "uncommon" (p.840) on FOIA.
  

• Annual and quarterly reports on activity at each agency (report instructions). If a request takes longer than the median shown there, it may be time to appeal to court. They're supposed to handle them first-in-first-out (p.47), so none should take too long.
• Statistics on requests, from FOIA annual reports: print or spreadsheet of any number of agencies and multiple years at a time (needs cookie to get spreadsheet, FOIA.gov)
• Overview of the US Freedom of Information Act
• Legal Treatise on FOIA cases
• Guidance to agencies on complying with FOIA
• Changes in June 2016 with law marked up with amendments

Besides the law, each federal agency publishes regulations (also called rules) to carry out the law. To find them try a search engine, after you change name_of_agency to the agency you want:

• name_of_agency "freedom of information" site:law.cornell.edu/cfr

There is another search at ecfr.gov, but it is not as complete. Agencies also have internal guidance, which MuckRock obtained by FOIAs. They focus on just one of the exemptions, but include guidance on other issues.

Like other agencies, the Department of Health and Human Services (HHS) has rules for FOIAs, and each major office in HHS also has rules. The overall HHS rules on FOIAs, adopted in 2016 are at:
https://www.federalregister.gov/documents/2016/10/28/2016-25684/freedom-of-information-regulations

Former rules at HHS were in effect for decades and are at:
https://www.gpo.gov/fdsys/pkg/CFR-2015-title45-vol1/pdf/CFR-2015-title45-vol1-part5.pdf

There were lots of changes. For example on appeals, after they deny a FOIA, an improvement in the new rules is:

• 90 days to appeal, up from 30 days (required by 2016 law)

Worse in the new rules:

• They give less detail about what should be in the appeal letter
• When they grant an appeal, now they'll "reprocess your request" which is pretty vague.  Previously their rules said they had to "send the records to you promptly or let you inspect them, or else we will explain the reason for any delay and the approximate date you will receive copies or be allowed to inspect the records."
• They have a new rule that they'll stop processing a FOIA appeal when a requester files a lawsuit. Lawsuits did not freeze processing under the old rules, and in fact HHS sometimes released records in response to a suit, without waiting for a judge to decide.

It is not clear how these HHS rules interact with separate FOIA rules at sub-agencies of HHS. For example the Centers for Medicare and Medicaid Services (CMS) has its own rules, which still say 30 days to appeal, and do not address the freeze in processing during a lawsuit. CMS rules are at
https://www.law.cornell.edu/cfr/text/42/part-401/subpart-B
and CMS also has policies at
http://www.cms.gov/Regulations-and-Guidance/Legislation/FOIA/Downloads/FOIAProcessingPolicyProceduresGuide-.pdf

Text of FOIA from Paul Burke submitted online July 15, 2013, Control # 071620137079
Denied Feb 25, 2015 in a letter from Medicare, then released June 8, 2016
Court case and settlement discussions

An ACO gets a bonus payment of up to 50% of the cost savings from Medicare Parts A and B for its patients. (If an ACO is willing to risk losses as well as bonuses (2-sided), they get up to 60% of the savings instead of 50% 42 CFR 425.606.

1. The baseline for savings is past years' spending, "Parts A and B fee-for-service expenditures for beneficiaries that would have been assigned to the ACO in any of the 3 most recent years" (42 CFR 425.602(a)) based on taxpayer numbers of doctors now in the ACO. The 3 years are not averaged equally; the last year gets 60% weight, middle year 30%, first year only 10%. 
2. The exclusion of retired doctors moves the baseline down, by excluding their typically older patients from the baseline.
3. The baseline omits patients who stop getting care from the ACO by choice or by death. (42 CFR 425.602(a)(8)) 
4. Excluding dead patients moves the baseline down again by omitting the typically high costs of the last year of life. Moving the baseline down means deeper cuts in current expenses are needed before the ACO sees savings.
5. The 3 base years stay the same for the entire "agreement period," (42 CFR 425.602(c)) typically 3 years.
6. The fixed baseline is updated by national rises in Medicare costs. (42 CFR 425.602(a))
7. The fixed baseline is adjusted for changes in patient mix. (42 CFR 425.602(a)) The adjustments are slight, explaining only 2% to 12% of the total variation actually caused by patient mix (p.65 table 3-22, "r-squared" of version 21). Adjustments are inherently ineffective, "You can’t get all the right variables on the page" said Berwick, former Medicare Administrator,
8. The comparison is Parts A and B fee-for-service costs for an ACO's patients each year. (42 CFR 425.604)
9. The cost for each patient is limited to the national 99th percentile of patient costs each year, to limit the effect of the most expensive patients (42 CFR 425.602(a)(4)). Doctors expect the limit will be about $100,000 per patient, so patients costing over $100,000 will be averaged as if they cost $100,000.
10. Patients are included in the ACO if they get more primary care from the ACO than from any other ACO or non-ACO sources (42 CFR 425.402). Patients who get no care in a year are not included.

Medicare requires ACOs to have at least 5,000 Medicare patients. With 260 Medicare patients per doctor, this means at least 20 primary care doctors in each ACO. Most ACOs are larger. However 3-8 doctors give better care, because they take more responsibility and have fewer managerial distractions than big practices (Kussin 2011, p.36). With at least 20 doctors, each doctor has little effect on the ACO's bonus, so the ACO needs internal incentives to motivate doctors, such as reviews of doctors, limits on expensive procedures, and rewarding individual doctors for cutting their costs. Much like an HMO.

Each hospital has a Patients Rights Advocate you can go to for complaints and help. People who use them wish they had gone sooner.

​Medicare has a complaint process, and supports advisors at each State Health Insurance Assistance Program (SHIP) for free personalized help.

​Patients and family can research treatment options with the resources at drugs.globe1234.com

Patients can get an outside professional advocate, often a senior nurse, to advise them about recommended and alternate treatments. Some will come to the hospital to listen to doctors and advise. Others can interact with patients and doctors on Skype or Google video chat. Cost can be $60-$200 per hour, cheaper than poor health care. As long as doctors and hospitals have hidden incentives to avoid care, patients need expert advocates. (The parallel is with courts, where most people would get a lawyer, in spite of the cost.)

National directories are at:

• aginglifecare.org
  
• advoconnection.com
• nahac.memberlodge.com
• patientadvocates.com (smaller membership)

1. Advice from Kiplinger and Huffington Post
2. aphadvocates.org/ sponsors the advoconnection national directory
3. healthadvocateprograms.com/ lists training programs for patient advocates
4. billadvocates.com links specifically to help on medical bills
5. healthcarebluebook.com lists prices for many procedures
6. Consumers Union has a "Safe Patient Project" to reduce medical errors, with many stories from patients

Free telephone help will not be as effective, but it is free for the very ill from Patient Advocate Foundation on skin, breast, cervix, ovaries, prostate, and colorectal cancer, narcolepsy, heart valves, Aetna claim problems, and dealing with other insurers.

The heavy penalties which Medicare puts on hospitals for readmitting patients, mean that doctors and hospitals may advise patients not to get treatment, if there is much chance of readmission, even when the treatment would help the patient.

So patients need to worry that advice from doctors, hospitals, or Accountable Care Organizations (ACOs) can be biased away from treatment.

Patients who want treatment can avoid being a subject of hospital penalties and hospital risk by:

1. Prevention, which includes diet (such as low salt and fat), exercise, immunizations, medicines, preventing falls. Doctor visits need to address blood pressure, cholesterol, swelling, apnea, and infections before they cause hospitalization. You can check your prescriptions for interactions at MediGuard.org
2. Avoiding admission to short term hospitals  by using prevention, urgent care centers, long term hospitals, outpatient surgery at a hospital or surgi-centers (though these can have worse outcomes, more risks, and more deaths) and other outpatient services. Patients can use Medicare's software to see if they have high readmission risk, to know if hospitals will be wary of them. Research is looking at more kinds of hospital services which can be given at home.
3. Dropping Part B of Medicare, if you have other insurance. "Medicare Part B is of limited value to someone already covered by a good health plan" (Center for the Study of Services, 2013). Most Medicare penalties and rankings are based on patients who have Part B, so if you drop it, your doctors know you will not affect their penalties. Medicare resists dropping Part B, and if you re-enroll later, they increase the Part B premium 10% for each year you lacked Part B, so it is a significant decision. The form to drop it is CMS-1763 (or copy); in the form part B is called "medical insurance," and part A is called "hospital insurance." Your right is established by 42CFR407.27. 14,000 people per year file this form to drop coverage (p.4 of Statement).
4. If you pay premiums for Part A Hospital Insurance you can drop it too, 42CFR406.28, but Part A is generally free. If you get Part A free, then the only way to drop it is the draconian step of refusing Social Security payments and repaying any Social Security and Part A benefits you have received. This rule was approved by the federal Circuit Court in DC in 2012 (Hall v. Sebelius); page 4 of the dissent (p.11 of the pdf file) analyzes this Social Security policy. The Supreme Court decided not to hear an appeal. Dropping part A involves the same form listed above for part B.
5. Federal retirees can opt out of Part A with fewer problems than most people. Many federal retirees lack Social Security and have federal health insurance. Even if they drop Medicare, they get charged the same as Medicare patients for hospital inpatient treatment, and for doctors who participate in Medicare (and up to 15% more by non-participating doctors). Outpatient hospital care and non-physician based care (e.g. PT, nursing homes) are not covered by this requirement for equal charges; regular Plan benefits apply. These costs are in each plan brochure, in a section titled, "When you are age 65 or over and you do not have Medicare." The limits were established by laws in the 1990s: Omnibus Budget Reconciliation Act of 1990~P.L.101-508 §7002, p.28; and of 1993~P.L.103-66 §11003, p.30.
6. Using a VA or military hospital if eligible. Members of Congress can use Walter Reed military hospital.
7. Using a specialist hospital (usually with better than average care). The following are exempt from readmission penalties: "rehabilitation hospitals and units; LTCHs [long term care hospitals]; psychiatric hospitals and units; children's hospitals; and cancer hospitals. Religious nonmedical health care institutions (RNHCIs)" 
8. Using a hospital in Puerto Rico (exempt) or Maryland (exempt, even famous hospitals like Johns Hopkins). Medicare's head office is in Maryland, so retirees and family members there do not face Medicare's readmission penalties.
9. Using a top hospital abroad; these are not covered by Medicare.
10. Using a Critical Access Hospital (small, rural, isolated, exempt, marked in "Combined list") if it has the skills you need 
11. In a US hospital, Medicare patients generally cannot pay privately: 42 U.S.C. § 1395cc(a)(1)(A)(i) says a hospital cannot charge or accept private payment "for items or services for which [an] individual is entitled to have payment made under [Medicare, Part A]" (p.16, note 9)
12. Getting a second opinion from doctors not affiliated with the hospital and not in an Accountable Care Organization
13. Arranging assisted living or caregivers and equipment at home so one can go home right after an outpatient operation and stay there successfully. Used hospital beds, bedside commodes, wheelchairs, walkers, alternating pressure pads, and caregivers are affordable. Lift chairs are expensive, but one that goes flat (A or B) will help you up and down from a sleeping position without you having to twist in and out of bed.
14. Long term design changes can be considered whenever remodeling, and there are financing options.
15. If staying overnight at a short term hospital, consider getting it called observation, rather than being admitted. Drugs will cost more, and it will not entitle you to covered nursing home care, but the stay will also not count toward a readmission penalty.
16. Joining a Medicare Advantage plan (may promote hospice aggressively, and limited appeal rights). They are paid a flat rate per member, so they need to minimize all hospital stays and other expensive care, but they have no specific incentive to avoid readmissions.
17. If there is any ambiguity in your principal diagnosis, you can try to convince the hospital to avoid having it be one of those that have penalties (Table A).
18. Lobbying Congress to repeal or reduce the penalty

Regardless of exemptions, it is possible doctors who primarily see ordinary Medicare patients will have a habit of minimal treatment, even for managed care and private insurance patients exempt from the penalties.

Also see lists of US Doctors:
Who offers long appointments?
Who treats you in both hospital and office?
Who has wide experience with procedures you need?
And data on Hospitals or Medicare and other patients

Doctors get many pressures which affect patient care. The Journal of the American Medical Association (JAMA) May 2 2017 summarizes a lot of research and has a proposal that doctors be paid by salary, to avoid:

• incentives to over-treat, when paid per procedure
• incentives to under-treat, when paid a flat rate per patient, or rewarded for cutting costs

They say salaries, without performance reviews for profit and loss, are, "the most promising solution... adopted by a variety of health systems, such as the Mayo Clinic, the Cleveland Clinic, and the Kaiser group". They ask the Medical Students Association to rate medical schools on how they pay their doctors. These students already rate medical schools on conflicts of interest with industry, though they allow industry gifts for student use. In the meantime patients can ask doctors at groups, medical schools or hospitals if they are paid more or less when they order more procedures.

Practices with 2 or more primary care doctors were analyzed in a July 2015 study in Annals of Family Medicine (632 practices surveyed 2012-13):

• A quarter of practices paid each doctor entirely based on the services they billed ("productivity" right side of the graph).
• A quarter of practices paid each doctor by straight salary. No word how it was set.
• The rest had a mix, shown in the graph. No word on how many doctors each type of practice covered.
  
• The study had a 50% response, and excluded solo practices. 
• Most solo practitioners receive income per procedure, and take home what is left after they pay staff, office and other costs (concierge doctors are exceptions).
• In the graph, the right side is light grey from top to bottom (100% salary). The left edge is dark gray (100% quality or miscellaneous payments). A big area in the left middle is medium grey from top to bottom (100% based on services billed). In other areas the amount of each color shows the amount of income received that way.

Doctors' average incomes (after expenses)  by specialty range from $200,000 per year for Public Health and Pediatrics to $500,000 for Plastic Surgery and Orthopedics, with wide variation. Concierge (2% of doctors) and cash-only (5%) doctors earn slightly more than average, comparable to other self-employed doctors. 13% are direct primary care doctors, generally subscription-based but lower fees than concierge. Three quarters do not charge for no-shows, whatever their policies may say. 70% see patients for 45 hours per week or less, but they spend 10 hours or more on paperwork and administration.  A fifth to a quarter of most specialties would not choose medicine again if they had the chance. Wealth averages $1-2 million, depending on specialty and age. A quarter of doctors have over $1 million by the time they are 35, and two thirds do by the time they are 50.

​Most doctors at hospitals work for large groups (TeamHealth, Schumacher) which contract to provide hospitalists, radiologists, emergency doctors, etc. Some companies provide doctors to hundreds of hospitals (Envision + Amsurg). Hospital doctors earn $200,000 - $400,000 per year. About half feel fairly compensated. Only a quarter "regularly" discuss the cost of treatment with patients. 

Payments from industry to doctors are discussed on the drugs page.

Patients can use published surveys to find doctors recommended by other doctors.

Consumer Checkbook asks doctors to recommend other doctors in big metro areas, and provides the counts for doctors recommended most often (list of counties). They also count patient recommendations for primary care doctors (in Boston, Chicago, DC, Delaware Valley, Puget Sound, San Francisco, and Twin Cities areas), so you need to read column headings carefully to be sure whether you are seeing ratings by doctors or by patients. In "more filters" you can sort by number of recommendations or distance from a zip code (if you selected "search by zip code"). 

"We regularly send surveys to all actively practicing physicians in the 53 largest metropolitan areas in the U.S. and ask them to tell us which one or two specialists in each of 35 or more different specialty fields they 'would consider most desirable for care of a loved one.' " ($28 online). Their research says that in general these doctors also:

• "Get much higher ratings than other doctors when we survey patients.
• Are much more likely than other doctors to be board certified.
• Are less likely than other doctors to have disciplinary actions filed against them with state medical boards.
• In surgical specialties for which we have good data on outcomes, have better results."

Many city magazines publish lists of:

• "Top Doctors" selected by editors at Castle Connolly ($2/month), or 
• "SuperDoctors" selected by editors at MSP Communications. 

Both sources start with nominations from doctors around the country and end with editors' review based on many other criteria. These are not simply doctors' votes like Checkbook.

A 1999 study found that doctors were more likely to be in such lists "if they trained in prestigious residencies (P<0.01) or fellowships (P<0.05), or if they had an academic appointment (P<0.05) or 15 or more years of experience (P<0.001)."

Referral services also select doctors they think are high quality, and you pay for the referral.

Malpractice and disciplinary cases (rare) are another place where some doctors are reviewed.

The best statistics are:

• Doctors' Experience with a procedure
• Average Time spent with patients

Less helpful are:

• Outcomes

Outcomes are described in the following paragraphs, but first a warning: All statistics on outcomes create pressure to "cream" or "cherry-pick," i.e., treat the healthiest patients and deny treatment to the sickest patients (saying, "you're not a good candidate for treatment" or having long delays even for urgent appointments). Deadly examples are in two recent articles in The Guardian and the New York Times. Measuring and rewarding doctors can also backfire and reduce quality by reducing motivation (see a very good, broad article on these effects).

Studies find that doctors avoid treating risky patients, when there is public reporting of outcomes. For example doctors who treat narrowing of the coronary arteries (for example to treat or prevent heart attacks, PCI-Percutaneous Coronary Intervention) in New York and Massachusetts have the death rates of their patients publicly reported, and these doctors avoid PCI on the riskiest patients, even if it might help the patient, since the higher death rate will hurt the doctor's or hospital's reputation.

• 75% have decided not to perform PCI in a patient due, at least in part, to protecting the doctor's or hospital's success rate
• 74%  sometimes or often delay PCI to see if the patient dies first.
• 95% say other doctors avoid doing PCIs on risky patients.
• 60% have been pressured by colleagues to avoid them.
• 52% worry their superiors won't support a decision to do PCI on a risky patient.

Groups which report outcomes often say they adjust for initial health to put all patients on a level field, but the adjustments are very weak. They have low explanatory power, and few variables: "the most assiduous work on risk adjustment has produced tools of only moderate power. The prospects for solving this problem with improved risk adjustment are not promising.[4],[5]"

Consumer Checkbook rates individual surgeons on death (within 90 days) and total bad outcomes, SurgeonRatings.org. They use Medicare patients 2009-12, and only report surgeons with results significantly above average. Unlike their hospital data, they unwisely exclude hospice patients from the surgeons' results (p.3). Few patients would have these surgeries if they were on hospice, so they must have gone on hospice after surgery. These deaths are no more or less attributable to surgery than any other deaths in 90 days. They acknowledge a random level of deaths in any population, unrelated to surgery, and exclude it with statistical tests. 

Surgeons' total bad outcomes for Checkbook include deaths within 90 days, atypically long hospital stays (indicating major complications), and readmissions within 90 days of initial hospital discharge. They seem to include all readmissions, even unrelated to the surgery, so they penalize surgeons who take on patients sicker than average. They say they adjust for patient riskiness, without details about which variables they use. They do list the "c statistic" for each adjustment, ranging from 0.626 to 0.913 (pp.14-39). The c statistic ha a scale of 0.5 to 1, where 0.5 means their equations do no better than chance, and 1 means their equations are perfect.  "Models are typically considered reasonable when the C-statistic is higher than 0.7 and strong when C exceeds 0.8 (Hosmer & Lemeshow, 2000; Hosmer & Lemeshow, 1989)." So some equations are little better than chance, and they still rate surgeons with them.  They cover (definitions on pp.9-13): 

ProPublica almost simultaneously with Checkbook in 2015 released death and complication rates for all surgeons with at least 20 surgeries during 2009-13, in the categories below. They count deaths during the same hospital stay as the operation, and wisely do not exclude hospice deaths (p.5). As "complications" they count readmissions within 30 days if these are for diagnoses considered likely to be related to the original surgery. These are 46% of all 30-day readmissions (p.6). They count a surgeon as having a "high" complication rate, based on their best single estimate, even if his/her confidence interval extends all the way into the "low" range.

ProPublica limits the data to elective surgeries, which usually involve healthy patients, but can include patients with other serious conditions, as long as these do not prevent the operation, conditions such as diabetes, dialysis, weak immune systems, etc. They say they adjust data based on the sickness of the patients, but they tried only one summary measure of all health conditions, and it had little effect (pp.10-11, with column heads defined on p.4). Age has the most effect, but they group it into 5-year categories, instead of using exact years of age (maybe because of data availability?). They do not provide summaries of their equations' power, but do acknowledge that their adjustments for surgeons' differing patients make only a "small difference" (p.13). For each equation they provide the standard deviation of the random effects (ranef sigma), which they interpret to mean most of the variation is among surgeons, not hospitals (p.15). They cover (definitions on App.1-3): 

1. Knee Replacement-Replace diseased knee joint with an artificial knee.
2. Hip Replacement-Replace diseased hip joint with an artificial hip joint.
3. Cervical (Neck) Spinal Fusion-The fusing of two or more vertebrae of the neck, using orthopedic devices to hold them in place.
4. Lumbar Spinal Fusion, Posterior Technique-The fusing of two or more vertebrae in the lower back, performed on the back portion of the spine.
5. Lumbar Spinal Fusion, Anterior Technique-The fusing of two or more vertebrae in the lower back, performed on the front portion of the spine.
6. Prostate Resection-The resection and removal of a portion of the prostate through the urethra.
7. Prostate Removal-The removal of the entire prostate gland via the open or laparoscopic or robotic method.
8. Gallbladder Removal, Laparoscopic-Minimally invasive gallbladder removal.

ProPublica lists some critiques and praise of their apprach. Note Rand's critique says ProPublica uses deaths within 30 days (pp.2, 6), but ProPublica confirms they use "deaths during the initial surgical stay", though twice they say ambiguously "deaths and readmissions to a hospital within 30 days"; and here.

Levy says the profession keeps secret a better scorecard, NSQIP.
​
Chest surgeons show 3-star ratings on about 500 hospitals and 500 group practices (typically the group of surgeons operating at a hospital), for

• Coronary artery bypass grafts (CABG) and
• Aortic valve replacements (AVR).

You can search by state, not by city. They define measures (deaths in 30 days, complications, best practice), but do not say what the cutoffs are for each star, nor how they adjust for patient sickness or where they get data.

Cardiologists show 4-star ratings for use of recommended drugs by a few hospitals after

• Implanting Cardiac Defibrillators, and
• Diagnostic Catheterization and Percutaneous Coronary Intervention (PCI/Angioplasty)

You can search by state, city or metro area. They report the numbers of each procedure, saying, "The number of ... procedures a site performs does not necessarily indicate higher quality, but it may be an indication of how experienced this site is with the procedure." They do not report outcomes, nor what the cutoffs are for each star, nor where they get data. They list 2,000 hospitals, but have counts and stars for only a few.

California has a rating system for doctor groups with HMO members:

• colorectal cancer screening of adults 50-75
• lower back pain patients who had X-ray, MRI, or CT scan within 28 days of the diagnosis (fewer is considered better)
• diabetes/blood: kidney function screenings, HA1c blood sugar testing, HbA1c <8.0%, blood pressure <140/90, cholesterol screenings, cholesterol LDL-C <100

New York shows deaths within 30 days after a procedure for individual heart surgeons and cardiologists. You can filter by name of doctor and/or hospital, and region of the state. They describe methods and definitions for the surgeon and cardiologist data. Another NY site has profiles of all doctors, but does not link to the death rates.

CMS Qualified Entities are allowed to use Medicare claims data, in order to provide quality measures on doctors, but it is not clear if any have yet released quality measures.

Patient reviews tell if a doctor speaks clearly and listens, as well as giving early warning of problems. Dr Wen of George Washington U and Dr Kosowsky of Harvard say doctors need to communicate well with patients to get information for a good diagnosis:

• "Choose someone who makes time to listen to you and answer your questions... and who engages you in a discussion of your diagnosis. By the same token, watch out for doctors who display signs of impatience, intolerance, condescension, or inflexibility" (p. 211). 

Wen's and Kosowsky's book goes on to describe in detail how patients need to prepare before seeing a doctor. 

Patient reviews tell you the style of different doctors, especially in the written comments. 

• The most widespread written comments are: RateMDs, AngiesList and Vitals. 
• There are fewer written comments at Wellness, Yelp and eHealthScores. EHealthScores shows amount of experience for each doctor, but their data are old, and you cannot search by amount of experience. As of January 2017, they still use 2012 Medicare data, while 2014 data are available on the Specialists tab.
• There are votes, but few or no written comments at HealthGrades, LifeScript and UCompareHealth. 
• There are some comments at Yahoo and Google, but patients' friends see those postings, so most patients do not want to discuss their medical situation frankly there.
• Checkbook has votes (often 10-70 per doctor) for many doctors in 11 metro areas (listed below), on communications, advice, accessibility, courtesy, etc. No written comments.

Most of the sites above allow one review or vote per email address, so the ballot box can be stuffed.

1. AngiesList gives more weight to reviews by paid subscribers, so stuffing would be expensive or impossible. AngiesList offers a national subscription which lets patients check doctors far and near. However doctors are told the real name of each reviewer on AngiesList, so there are few negative reviews.
2. Checkbook's ballot box is even harder to stuff. They send questionnaires to:

• subscribers of Consumer Reports and Consumer Checkbook, in 7 metro areas: Primary care doctors in: Boston, Chicago, Delaware Valley, Puget Sound, San Francisco Bay, Twin Cities, DC. $34 for 2 years of results.

• people insured by Aetna, Blue Cross/Blue Shield of Kansas City, BlueCross BlueShield of Tennessee, CIGNA HealthCare, and UnitedHealthcare, in 4 metro areas: Primary care doctors in Denver and Kansas City; Memphis also has cardiologists and obstetricians / gynecologists; New York has all these and gastroenterologists and orthopedists. Free results.

Some doctors and hospitals "aggressively combat negative social media posts, casting a pall over one of the few ways prospective patients can get unvarnished opinions of doctors... consumers need to know there can be consequences if they post factually incorrect information." Some doctors have required patients to sign contracts which prohibit negative reviews or let doctors remove negative reviews. The "Consumer Review Fairness Act of 2016" makes those clauses unenforceable. 3,000 doctors had these contracts available in 2011, though not all used them. These doctor-patient contracts are not mentioned by any of the review sites as a reason for removing reviews, but one assumes they do it, or doctors would not bother with the contracts. 

Before patients choose any doctor based on positive reviews, it is important to ask the doctor's office: Do patients have to sign a contract controlling patient reviews?

Rules about the reviews they keep are important. Here are rules at the 3 biggest sites:

RATEMDS 

Allowable reviews: They have few restrictions: "post only truthful, non-libelous, and relevant ratings and posts."

Removing reviews: "We remove ratings for a number of reasons, but it is usually due to one of our automated spam filters thinking there were multiple ratings coming from the same rater... We generally do not remove ratings. This site is for people to report on their experiences...

• [If doctor says:] "I have no choice but to sue you!
•  [Website answers:] "Think twice. The Communications Decency Act (the "CDA") is a complete bar to our liability for the statements of others on this website...
• [If doctor says:]"I'm going to sue you anyway. Taking down the review is cheaper than paying a lawyer.
•  [Website answers:] "Sounds like a Strategic Lawsuit Against Public Participation (SLAPP.) Twenty-eight states, including our home state, have passed strict anti-SLAPP laws, authorizing expedited motions to dismiss and giving rise to counterclaims seeking attorneys’ fees and liquidated damages. If you want to drag us into court, bring your checkbook with you, because you'll be paying our attorneys to defend us."

Are reviews anonymous? Possibly. They keep patients' names anonymous unless subpoenaed, but they let advertisers and social media companies track which pages you see, so those companies can see your IP address when you post reviews. You can be anonymous if you post your reviews from someone else's computer, like a library, and don't identify yourself on that computer (eg. logging in to Facebook or email). Using your computer at a coffee shop gets you a new IP address, but your computer probably has persistent identifiers which their advertisers can track.

Are searches anonymous? Only if you use someone else's computer. Their advertisers track IP address and every page visited, so they can see what you're searching for. To foil advertisers you can turn off cookies, but not necessarily beacons. Buttons for Facebook, Twitter, and Google+ on every page presumably report every page visit to those companies.

Multiple ratings and updates: They limit patients to one rating "for the same doctor from the same computer or user." "If you were logged into your account when you submitted the rating, you can edit your star rating in your profile. Comments are not editable, but you can leave a new comment... If you were not logged in when you submitted the rating, the only thing you can do is to try to remove it and then enter a new rating." 

Can patients report why they didn't go to a doctor? Yes.

Terms of Use: 100 words; nice and short, and they don't require patients to pay for RateMDs legal costs or accept arbitration. 

Privacy policy is also short, at 400 words.

ANGIESLIST

Allowable reviews: They have woolly restrictions: "accurate, truthful and complete in all respects" (10) and not offensive, harmful or distasteful, among many other criteria (16). 

Removing reviews: They pose the question and answer it several different ways: "Angie's List may modify, adapt, or reject Your reviews if they do not conform with Angie's List's publication criteria, which may change from time to time at Angie's List's sole discretion" (10g). "we never remove reviews unless the member who posted the feedback contacts us to delete it." "If ever we question the legitimacy of a review, we take it down."

Are reviews anonymous? No. They reveal the author of each review to the doctor (10f), though not to other members. Writing any review waives privacy and lets doctors publicly post "Your private or confidential health information in response to Content You submit" (15). "Angie's List may in our sole discretion share your User Generated Content with others."

Are searches anonymous? No. They disclose personal information when "permitted by relevant law," and they have four tracking companies on their search pages (see a table at the bottom of this page), so those companies know what you're searching for. They allow themselves to "link the information we record using tracking technologies to Personal Information we collect."

Multiple ratings and updates: "You have the right to share your honest opinions at any time." 

Can patients report why they didn't go to a doctor? Yes. Reviews by people who checked out a doctor but did not go there are posted, but not weighted as heavily as people who did choose the doctor. So good or bad experiences before getting care can be read, though the doctor's average rating is not much affected.

Terms of use: 8,000 words. They do not require arbitration. They do require patients to pay AngiesList legal costs: "indemnify, defend and hold harmless Angie's List... from and against all losses, liabilities, expenses, damages, claims, demands and costs, including reasonable attorneys' fees and court costs due to or arising from: 

•  "(a) any violation of this Agreement by You;
•  "(b) the inaccurate or untruthful Content... or
•  "(c) any intentional or willful violation of any rights of another or harm You may have caused to another." (27). 

Even though you pay the costs, "Angie's List will have sole control of the defense of any such damage or claim" (27). 

Privacy policy has 3,000 words.

VITALS 

Allowable reviews: must be "true and accurate" and not offensive, harmful or "otherwise objectionable," and not "deemed confidential by any contract or policy," among many other criteria.

Removing reviews: They say they provide "The complete list of all reviews from patients just like you." However they'd suppress reviews that violate their lengthy criteria, so it is not really "all." They sell to doctors a service called: "Reputation Management." It "enables providers to append responses to specific reviews [and] Opportunities to encourage patients to write additional reviews." Vitals' CEO is quoted in Buzzfeed saying the site can remove an "outlier" negative review at the doctor's request.

Are reviews anonymous? Maybe. "Except as described in MDx’s Privacy Policy (the “Privacy Policy”), MDx will not be required to treat any of Your Submissions as confidential..." The Privacy Policy does not say that it protects your IP address. They let advertisers and social media companies track which pages you see, so those companies can see when you post reviews. You can be anonymous if you post your reviews from someone else's computer, like a library, and don't identify yourself on that computer (eg. logging in to Facebook or email). Using your computer at a coffee shop gets you a new IP address, but your computer probably has persistent identifiers which their advertisers can track.

Are searches anonymous?: No. The Privacy Policy does not say that it protects your IP address at all. For your name and email it is reasonably strict, though it allows "Developing new Products and Services, which... may be supported by third-party advertising... companies and networks may place or recognize a unique 'cookie' on your browser or within the code of the mobile application or use a 'web beacon.' " Buttons for Facebook, Twitter, Google+, Pinterest and LinkedIn on every page presumably report your page visits to those companies, so they can see what you're searching for. 

Multiple ratings and updates: "post only one (1) Submissions regarding the same Healthcare Provider, entity, procedure or subjectduring any thirty (30) day period."

Can patients report why they didn't go to a doctor? No. "By clicking Submit, I...  verify that I have received treatment from this doctor."

Terms of use: 6,000 words. They require patients to pay Vitals' legal costs and accept binding arbitration in Lyndhurst, New Jersey, "You... agree to defend (at MDx’s option and at Your sole expense), indemnify and hold MDx harmless from any damages, losses, costs, or expenses, including reasonable attorneys’ fees, which MDx may incur as a result of Your Submissions... Any controversy or dispute between You and MDx... shall be submitted to final and binding arbitration as the sole and exclusive remedy." 

Privacy policy has 3,000 words.

How Hospitals Can Manage Risk

Each penalty is much larger than the money earned from initially treating the patient. Caring for any Medicare patient frail enough to have a large readmission risk has become financially risky and an act of charity. Medicare even offers software to show which patients have high readmission risk, so hospitals can be very aware of them. In advising Medicare patients, Medicare now wants hospitals to consider financial risk to themselves, as well as medical risk to patients.

Hospitals can in theory address their risks with Medicare patients by: (A) improving care, (B) taking the riskiest patients out of the statistics system, or (C) serving more low risk patients.

(A) Improving care has several aspects:

• Inside their walls, hospitals have always provided care as well as they know how. Continuous improvement is always good, but usually limited, and often expensive.
• Outside their walls, hospitals have little role in the best ways to improve care, such as better patient compliance, frequent monitoring, adjustment of medications after discharge, diet, exercise. These are the role of outpatient doctors and general public education. Hospitals can encourage and provide reminders and feedback. A patient survey used at 120 hospitals covers very rudimentary matters, showing hospitals' limits.
• During discharge, hospitals can send patients to nursing homes rather than home, at higher cost to Medicare and/or the patients. If this redirection cuts readmissions, hospitals may pursue it regardless of cost to Medicare, or whether being home could benefit more patients. And hospitals do need to choose the most helpful nursing homes. The Advisory Board (a consultant group) is one of many places with discharge advice.
• Ignoring Medicare incentives. Following them uses skilled staff time, and Medicare is so focused on cost that the budget will be cut no matter what the hospital does.
• Telling patients about readmission penalties, to convince patients that the first 30 days after discharge have real risk of readmission, and motivate patients to work with health care providers to stay well

(B) Taking the riskiest patients out of the statistics can include:

• sending people home from the emergency room or into a long term hospital, rather than admitting them

• having patients for observation, rather than as inpatients, entailing higher patient costs than inpatient
• Medicare recommends advising patients and families to accept comfort care/symptom relief, and "do not resuscitate" (DNR), and hospice, so more patients die at home rather than coming back to the hospital

• advising postponement of treatment, hopefully to another hospital
• advising use of a Medicare Advantage plan (HMO or PPO; they do not face these penalties but have other cost limits)

• advising patients to drop Part B if they can get other insurance (penalties are based on patients with Parts A and B)
• not billing Medicare for frail patients for 30 days, absorbing the initial cost rather than the penalty

• reviewing the principal diagnosis for a complex patient to see if it can properly be an issue which is not one of the six categories subject to penalty (in Table A). Veterans Affairs Medical Centers change the "heart failure" diagnosis to "hypervolemia," too much water in the blood, which has no readmission tracking. (VA has no readmission penalties, but hospitals are tracked and bosses get bonuses based on results.) Theoretically there is no tolerance for improper coding
  
• educating Congress to repeal or reduce the penalty

(C) Serving more low risk patients will not save Medicare money, but can save the hospital money. It can include:

• building relationships with referring doctors
• marketing, such as direct mail to hikers and runners, with ads for the hospital's skill at knee replacement, thus recruiting a generally healthy group to improve the hospital's readmission rates for knee replacement. HCA seems to do this.

Whenever medically defensible, the hospital would earn more by avoiding risky Medicare patients in the listed diagnoses, since serving these patients raises the number of readmissions. Hospitals have large financial penalties for readmissions, though not for deaths.

National Partnership for Women & Families noticed the incentive problem when the penalty was first enacted in 2010, "models like hospital readmission penalties may create perverse incentives for providers to reduce or avoid providing care to complex patients who could be less profitable under these models" (p.6).

Health providers who join an Accountable Care Organization (ACO) have further reason to promote symptom relief and minimize treatment. They gain from cost savings, lose from readmissions (p.10) and even from admissions (pp.12-13), and have no loss from death. Patients will not be well informed. A list of ACOs is here.

Too much penny-pinching in government health care leads to separate systems with unequal care: "in the General Hospital, the government pays Siloam [an Indonesian hospital chain] a capped price per patient for a given condition; in Lippo Village most patients pay for themselves" (Economist 17 May 2014)

Map shows where hospitals are (or maps of doctors)
​Financial Data (below)
Quality measures and mixed incentives
Readmission penalties or xls (6MB)
Biggest penalties (methods)
Previous data: Readmission Penalties (August 2013, 3 MB xls)
List of Accountable Care Organizations (many include hospitals)
Other Medicare data
Medicare Costs, Premiums, and Alternatives

​A spreadsheet (5MB) shows each hospital's 2013 or 2014 financial statements, and many Medicare calculations at each hospital. It includes both Medicare and non-Medicare revenue and spending. It is compiled by Medicare to provide a context for Medicare spending in each hospital, and is called a "Medicare Cost Report"

The spreadsheet includes:

• Revenue from patients, investments, donations, and other
  
• Expenses for Salaries, Capital, major departments such as ICU, Operating room, Radiology, Emergency, Lab, Pharmacy, Administration, Building operation
  
• Chain membership, Ownership type, number of beds, number of Medicare patients, address, phone, latitude + longitude, date certified
• Medicare payments for Diagnostic groups, outliers, education,
• Several penalties: HAC, VBP, IQR, EHR, and readmissions.

The spreadsheet has brief labels; fuller explanations are in the original Medicare form and instructions. Chapter 40 of Medicare's manual has the form (R6P240f), including work sheets S (p.1),  A (p.22), E (p.84), and G (p.100). Chapter 40 also has the instructions (pr2_40, abbreviations are on pp.9-11), which can answer many questions about the entries on the form. The current format has been used since 2010, and other data are available back to 1995.

Another article shows helpful commands for the spreadsheet.

The source also has each hospital's occupancy rate for several departments: general, maternity, ICU, coronary care, burn, hospice, psychiatric, rehab, etc. These have not been put in this spreadsheet. If you would find the occupancy rates useful, please leave a comment below or send an email.

The original Medicare databases are available from 1995 to the present. They are far more complex than the spreadsheet, with 3 types of records, and millions of records, since every number and answer on each form has a separate record. For those who need it, a CMS documentation page has record  counts, a spreadsheet of hospitals covered, and  layouts. The Medicare database averages 3,000 numeric records and 600 alpha records per hospital each year. 65 key items are in the spreadsheet, and others can be available if needed.

Electronic Municipal Market Access (EMMA) has PDF copies of operating expense and audited financial statements for each hospital (or other facility), if it has outstanding tax-free bonds. Put hospital name in their search box, to list its past & present bonds. Click any bond which is still outstanding (on right), accept the disclosure, then click "Continuing Disclosure" to see annual and sometimes quarterly data. The data are similar to the spreadsheet above, but in PDF, often with more data from the past, but fewer hospitals.

"Summary of audit findings and federal awards" is an Audit Clearinghouse form a few pages long for each hospital showing checkoffs for any audit findings, and the amount of each federal grant spent during a year ("awards"); it does not cover Medicare or Medicaid, since these are exempt from the federal "single audit" rules. It also shows address, Employer ID number (EIN) and DUNS number. 

IRS form 990 for US nonprofit hospitals is available from AHCJ (search by name, place or person; just hospitals; other sources have many more non-profits, which can make hospitals harder to find), Open990 (2010 to present, downloadable spreadsheets) or Guidestar or Foundation Center or ProPublica or Charity Navigator, or CitizenAudit (full text search $350/yr). Form 990 shows total revenue and expense and highly paid staff (schedule J ) and contractors (VII-B). It also shows total received from Medicare and Medicaid in section VIII-2 "Revenue, Program Service," and several types of expenses in section IX, balance sheet in X, political spending in schedule C.

Canadian nonprofits (non-governmental organisations, NGOs) have financial information at donate2charities.ca

Column A. Each "Penalty" is the cost of Initial Treatment in Column B multiplied by the US ratio of admissions to readmissions (Column D). 

1. A Congressional agency, MedPAC, confirms that the penalty per excess readmission [Col A]= "Payment rate for the initial DRG [Col B] ... ×  [Col D] 1 / national readmission rate for the condition" (p.99). 
2. The law and Medicare's explanation have more complex wording, but are equivalent to this multiplication. An example is in this spreadsheet.
3. Congress told Medicare how much to penalize hospitals which have "excess" readmissions, and told Medicare to decide which treatments would face penalties. Medicare chose to apply penalties to the conditions in Table A (see timing). "Excess" readmissions means above the US average, adjusted for patient mix. 

Column B. The "Average Base Payment" is an average of Medicare's detailed payments, by diagnosis from October 2014 to September 2015 (FY 2015). There are different payment levels for patients with and without other unrelated illnesses, and the average payment here is weighted by the number of patients having each payment level.

Column C. "US Average Readmissions" above were updated in Hospital Compare 26 July 2017. 

1. Unscheduled readmissions of Medicare patients, for almost any reason at most Medicare and VA hospitals count in the readmission rate of the hospital where the initial visit happened.
2. Penalties are only charged at hospitals with 25 or more admissions for a condition
3. Some types of admissions are excluded, which exempts them from penalties.

Column D. "US Ratio of Admit to Readmit" is one divided by Column C.

1. American Hospital Association thinks penalties are so large that using Admit as the numerator must be a "technical error" (p.4), but they agree that is what the law says. 

Column E. "Each Condition" is the total of Medicare's counts of admissions, by diagnosis

Note F. Medicare does not provide as much detail as this table, but its data support an estimate of $52,000 average penalty per excess readmission in 2016, $36,000 in 2014 and $40,000 in 2013:

1. $227 million total penalties in 2014, divided by 6,300 excess readmissions per year, since there were 18,902 total penalties in the 3 base years they show
2. $280 million total penalties in 2013, divided by 7,000 excess readmissions per year--20,947 in 3 years shown
3. Difference  between 2013 and 2014 is budget inaccuracy, not real change
4. $420 million total penalties in FY2016 (p.2064), divided by 8,051 excess readmissions per year -- 24,153 in 3 years shown (at 2,666 hospitals, p.2064)

Note G. Medicare adjusts readmission rates for patient mix at each hospital, to level the playing field, but the adjustments have a very poor fit, explaining only 3-5% of the variation in readmissions.

Note H. MedPAC looked at the multipliers (shown in Column D), and recommended the law should change to: 

1. Make all the multipliers 1, not 4 to 21, so each penalty would be smaller, but 
2. Set the standard at fewer readmissions than the national average.
3. If MedPAC's recommendation is adopted, more hospitals and more readmissions would be penalized, and total penalties would be the same (p.101). 
4. With Congress polarized on the healthcare law, amendments seem unlikely.

Note I. Medicare has an interesting map of readmission rates by county and race. Choose readmissions under "measure."
​Note J. Penalties for each hospital depend on constants in Table 1, and factors for each hospital in the Impact file each year.

Patients come to hospitals with a range of conditions which might bring them back later. Medicare provides software to predict which are most likely to come back.

To see how hospitals are affected, here is a simple example of

• 20 patients with 15% chance of readmission, so 3 return, and
  
• 20 patients with 35% chance of readmission, so 7 return

If this hospital makes no special efforts, it would have a total of 10 readmissions. Assuming other US hospitals reduce total readmissions 20% as Medicare wants, the new national average and the penalty cutoff will drop to 8, and our hospital above will pay for 2 penalties over the new national average.

To reduce readmissions, a hospital needs to put effort where it has the best chance of reducing them. Working on the 20 patients who have 3 readmissions among them has little payoff. Working on the other 20, who have 7 readmissions among them, has a much better chance of avoiding some readmissions.

Part of the effort will be to cure patients well and refer them to good follow-up care. This has limited potential, because hospitals have little influence on patients after they leave.

An additional approach is to convince patients not to seek treatment any more, and hospice groups vocally support this direction. Every extra health problem raises the patients' odds of readmission and also raises their vulnerability to pressure for comfort care, "Do Not Resuscitate" orders and hospice. The line between appropriate advice and inappropriate pressure is easy to cross when hospitals have strong financial incentives.

This page first discusses Excel, then statistics in general near the bottom of the page.

This is a simple introduction to some Excel commands which are helpful with the large spreadsheets of medical data on the site. The commands here work in Excel 2010 and Excel 2003.

When you open an excel spreadsheet from the web, it may ask if you want to "enable editing." You'll need to say Yes to make any changes in your copy, or to "find all" entries of a certain type, or filter, or sort.

If the spreadsheet was slow to download, click File/Save As a couple times with new names, so you can get back to the original version without downloading again.

If you need extra help, many people who work in bookkeeping or finance are good with Excel spreadsheets, or you can search the web.

Basics. Most of you already know these:

• Find Records by pressing Ctrl and F at the same time, and type what you want into the search box. If you know which column it's in, Highlight that column by clicking the letter atop the column, before  you press Ctrl-F, and the search will be much faster.
• Rows start with 1, and Columns start with A. You can move around the spreadsheet with the arrow keys, the page down and page up keys, and the sliders on the right and bottom sides of the window, if they appear.
• Click a Cell and a little window just above the letters at the top of the columns shows you the number or formula in that cell, or Type something new to replace what's there.

Select Records (Filter), such as a State, or Specialty:

1. If small ▼Triangles already appear near the top of each column, skip to step 5. If you don't see triangles, start with step 2.
2. Click the letter at the top of the column showing State, or whichever column you want to select on.
3. Click Data. The Data menu is found alongside File, Home, Insert...
4. Click Filter (you may have to first click Sort-and-Filter). After you click Filter, a small ▼Triangle will appear near the top of the selected column(s).
5. Left-click the triangle, to see a little menu where you can de-select All, then Click the state(s) or specialty(ies) you want. Click OK, and only your state(s) or specialties will appear. You can get the others back, by clicking the little triangle again and selecting All. The little menu under the triangle has its own little slider to go up and down the list; All is at the top of the list.
6. The triangle may hide some of the column heading. You can remove the triangle by again clicking Data, then Filter.
7. To select a range of records, such as costs over $1,000, click Number Filters in the little menu under the triangle in the costs column, then click Greater Than, then fill in $1,000 ($ and comma are optional), then click OK. Number Filters can even give you Between 1,000-2,000, or Top 1 to 500, or Above or Below Average. Sometimes they're called Text Filters, with similar choices.
8. You can filter several columns, like female Pain doctors in Virginia and DC.

Sort in Any Order You Want:

• Choose cells to sort: Highlight any area by left-clicking the upper left corner, then shift-left-clicking the lower right corner, or Drag your mouse from one corner to another while holding the left key , or Click the letter atop any column to highlight a column, or Click the number left of any row to highlight a row
• Click Data at the top of the screen, then click Sort. Answer the questions to expand to all columns or rows. Then you will see a window of sorting choices.
• To keep headings at the top, be sure the box is clicked for "My data has headers"
• Normally it moves Rows up and down to put them in order. Click Options to move Columns right and left into order
• In the Sort By box, click the ▼ triangle, and choose a column (or row) to sort by
• If you want subsorting (such as by state and then by revenue within state), you may need to click Add Level and fill in the second item to sort by
• Click OK. A file over 100,000 records may take a few seconds to a minute to sort, depending on your computer's speed.

Undo a Sort or anything: press Ctrl and Z at the same time. You can do this repeatedly to back up to previous versions, sometimes as far as the last time you saved the file

Save Frequently, with new names, so you can go back to previous versions if you need to.

Calculate an Average or Sum: Suppose you want Average of Column T

• Insert a new row to hold the average, by right-clicking the Number to the left of a row, like row 2: On the menu which appears, left-click Insert
• Click a cell in the new row, probably in column T
• Type into that cell (don't forget the equals):  =Average(T3:T50000)

• Include the full range you want, which depends how big your spreadsheet is

• Press enter, the formula should disappear, and a number will appear.
• =Average( ) gives the average of numbers, including zeros, but ignoring blanks. It includes rows you hid with Filter. To do a lot of work on one state, you may want to delete all the others, so your averages or sums are just that state. Save Frequently.
• You can also use =Sum( ) or =Median( ) or many other functions (click Help)

Calculate a Ratio, or Difference Suppose you want the Ratio or Difference of Columns K and L:

• Insert a new column by right-clicking the Letter atop a column: On the menu which appears, left-click Insert
• Click a cell in the new column, on the row where you want the first ratio, such as row 2
• Type into that cell the ratio you want from that row (don't forget the equals): =K2/L2
• Press enter, the formula should disappear, and a number will appear.
• Then put your cursor over the lower right corner of the cell. Your cursor becomes a +; double-click it and blank cells below it will fill all the way to the bottom. Click some of those to see how the formula appears as it goes down the page.
• Formulas can use / * -  + ( ) and many other functions described in Help.
• If cells have too many decimal places, right-click the cell, column or row: On the menu which appears left-click Format cells: In the window of format choices which appear, click Number and the number of decimal places you want, then OK

Summary Tables (Pivot) can count or sum records in categories. For example if you want average revenue by state:

• Select state and revenue columns by left-clicking the letter of the left column, then shift-left-clicking the letter of the right column (every column in the range must have label in first row)
• Excel 2010: Click Insert (or in Excel 2003: click Data) then click PivotTable. Answer any questions, and it will open a new page.
• 2010: On the right side of the screen, drag a category, like State, down to Row Labels or Column Labels. (2003: click a category, then use ▼triangle to add the category to rows, columns, or data.)
• Move any variable, maybe State again, to Values (called Data in 2003). Be sure it appears there as "Count of.." Notice the summary table on the left side of the screen now shows the count of rows for each state
• 2010: Drag a numeric variable, like revenue, to Values. It will develop a little arrow, Left-click the little arrow to choose "Value Field Settings.." and then Sum, Count, Average, Max, Min, etc. (2003: Add a numeric variable, like revenue, to Data. See where it appears on the table as Count of. Right-click it: On the menu which appears, left-click Pivot Table Field and then Sum, Count, Average, Max, Min, etc.)
• The summary table on the left can give you the count of rows in each state, and the average and/or sum of as many variables as you wish, for each state.
• You could also do rows for each type of hospital or other category in your data
• 2010: If the menu of variables on the right side of the screen disappears, click anywhere in the table on the left side of the screen, and the right side will reappear. (2003: A Pivot Table bar floats on the page, with a button on the right end to hide or show the Field List.)
• 2010: If you want some states, but not others, move your cursor to the upper right corner where the state variable is listed. When you point at a variable there, a ▼triangle will appear to the right of the variable name. Click the triangle, then de-select as many states as you wish.

Advanced Formulas

• ^ means exponent so =2^3 means 2 cubed or 8, and 2^.1 means tenth root of 2
• You'll get warnings in cells where you divide by zero. You can avoid them by typing conditions: =IF(condition , result if condition is true , result if false): =IF(sum(L2)<>0,K2/L2," ")
• Excel often objects if you compare text to a number, but the sum of text is zero, which can be compared to a number as shown above. <> means "not equal to"
• When you copy a cell down a column, cell references generally change: K2/L2, K3/L3, K4/L4, etc. If some should not change (suppose L2 is a national average, and you want all the Ks as a ratio to that national average), then put in the first cell K2/L$2, and it will fill down as K3/L$2, K4'L$4, etc. The $ does not affect the value, and does not turn it into cash (which is done with Format Cell). The $ just says not to change 2. Similarly $L2 means to keep L, and $L$2 means to keep both L and 2.
• Besides clicking the bottom right of a cell, you can copy it by using Edit, or highlighting an area and using Ctrl-R to copy to the right, or Ctrl-D to copy down
  
  Why Excel
  This site uses Excel, since Google Sheets are limited to 2 million cells and Open Office is limited to 65,536 rows. The files here have hundreds of thousands of doctors with 20-60 cells for each, so they need the size of Excel or QuattroPro. These instructions cover Excel, since more people have it. You can do the same work in QuattroPro.
  

For subscribers, AMA has advice for using statistics with large data files, a checklist, a series discussing medical databases, and an article comparing "odds ratios," probabilities, and "relative risk ratios" (emphasis added): 

• when randomly selecting a card from a deck, the probability of selecting a spade is 13/52 = 25%.
• The odds of selecting a card with a spade are 25%/75% = 1:3.
• Clinicians usually are interested in knowing probabilities, whereas gamblers think in terms of odds...
• Differences between 2 different groups... can be compared using odds ratios [or]... relative risk ratio, which is the ratio of 2 probabilities...
• The odds ratio... can be interpreted as whether someone with the risk factor is more or less likely... to experience the outcome of interest... For example, an odds ratio for men of 2.0 could correspond to the situation in which the probability for some event is 1% for men and 0.5% for women. An odds ratio of 2.0 also could correspond to a probability of an event occurring 50% for men and 33% for women, or to a probability of 80% for men and 67% for women.
• Second, and less well known, the magnitude of the odds ratio... is scaled by an arbitrary factor (equal to the square root of the variance of the unexplained part of binary outcome)... [A]dding more independent explanatory variables to the model will increase the odds ratio of the variable of interest (eg, treatment) due to dividing by a smaller scaling factor.

There are excellent articles on statistical analysis of health data in the British Medical Journal, though it requires a subscription, which you may find at a university or hospital library.

Experience with Each Procedure is shown by the number of times per year each hospital billed insurance for the procedure. Hospitals have different skills in different fields. The best results for each procedure usually happen at hospitals with plenty of experience in that procedure. Or you can start with a different approach, which helps you find the most experienced doctors.

Medscape asked 8,500 US doctors in Dec 2016-March 2017 about the best hospitals. "Suppose you or someone in your family were just diagnosed with a complex or difficult case of (list below). Assuming that there are no barriers (eg, transportation cost) to treatment at the hospital you prefer, what hospital would you choose for treatment?"  They list the top 10 hospitals recommended for each condition. (Medscape is a subsidiary of WebMD and is supported by ads and by drug companies' payments in support of continuing medical education. They and their advertisers use cookies and web beacons to track users by IP address.)

• Breast Cancer: Oncologists recommended: Dana Farber-Boston, Mayo-Rochester, MD Anderson-Houston, Memorial Sloan Kettering-NYC, U of Michigan-Ann Arbor. All doctors recommended: Memorial Sloan Kettering-NYC, MD Anderson-Houston, Mayo-Rochester, Dana-Farber-Boston, Cleveland Clinic-Cleveland, Hopkins-Baltimore, Stanford Health Care-Stanford (CA), City of Hope Helford-Los Angeles, Duke-Durham, U of California-San Francisco, Brigham+Women's-Boston.
• Prostate Cancer: Oncologists recommended: Dana Farber-Boston, Hopkins-Baltimore, Mayo-Rochester, MD Anderson-Houston, Memorial Sloan Kettering-NYC. All doctors recommended: Memorial Sloan Kettering-NYC, MD Anderson-Houston, Mayo-Rochester, Hopkins-Baltimore, Cleveland Clinic-Cleveland, Dana-Farber-Boston, U of California-San Francisco, Massachusetts General-Boston, Duke-Durham, Stanford Health Care-Stanford (CA).
• Lung Cancer: Oncologists recommended: Dana Farber-Boston, Hopkins-Baltimore, MD Anderson-Houston, Mayo-Rochester, Memorial Sloan Kettering-NYC. All doctors recommended: MD Anderson-Houston, Memorial Sloan Kettering-NYC, Mayo-Rochester, Dana Farber-Boston, Hopkins-Baltimore, Cleveland Clinic-Cleveland, Duke-Durham, City of Hope Helford-Los Angeles, U of California-San Francisco, Stanford Health Care-Stanford (CA).
• ​Hip Replacement: Surgeons recommended: Cleveland Clinic-Cleveland, Hospital for Special Surgery-NYC, Mayo-Phoenix, Mayo-Rochester, New England Baptist Hospital-Boston. All doctors recommended: Hospital for Special Surgery-NYC, Mayo-Rochester, Cleveland Clinic-Cleveland, Rush Univ-Chicago, New England Baptist Hospital-Boston, Hopkins-Baltimore, Massachusetts General-Boston, Duke-Durham, U of California-San Francisco, Cedars-Sinai-Los Angeles.
• Cardiac Conditions: Cardiologists recommended: Brigham+Women's-Boston, Cedars-Sinai-Los Angeles,  Cleveland Clinic-Cleveland, Mayo-Rochester, Mount Sinai-NYC.  All doctors recommended: Cleveland Clinic-Cleveland, Mayo-Rochester, Presbyterian-NYC, Massachusetts General-Boston, NY, Hopkins-Baltimore,  Cedars-Sinai-Los Angeles, Brigham+Women's-Boston, Stanford Health Care-Stanford (CA), Duke-Durham, U of California-San Francisco, Mount Sinai-NYC, U of Pennsylvania-Philadelphia.
• ​Interventional Cardiac Surgery: Cardiologists recommended: Brigham+Women's-Boston, Cleveland Clinic-Cleveland, Hopkins-Baltimore, Mayo-Rochester, Stanford Health Care-Stanford (CA).  All doctors recommended: Cleveland Clinic-Cleveland, Mayo-Rochester, Presbyterian-NYC, Massachusetts General-Boston, NY, Hopkins-Baltimore, Stanford Health Care-Stanford (CA), Brigham+Women's-Boston, Cedars-Sinai-Los Angeles,  Duke-Durham, U of California-San Francisco, U of Pennsylvania-Philadelphia.
• Stroke: Cardiologists recommended: Brigham+Women's-Boston, Cedars-Sinai-Los Angeles, Cleveland Clinic-Cleveland, Mayo-Rochester,  All doctors recommended: Mayo-Rochester, Presbyterian-NYC, Massachusetts General-Boston, NY, Cleveland Clinic-Cleveland, Hopkins-Baltimore, U of California-San Francisco, Duke-Durham, Stanford Health Care-Stanford (CA), Northwestern-Chicago, Brigham+Women's-Boston. Cedars-Sinai-Los Angeles.
• Infectious Disease HIV and Infectious Disease specialists recommend: Cedars-Sinai-Los Angeles, Emory-Atlanta, Hopkins-Baltimore,  Massachusetts General-Boston, NY Presbyterian-NYC. All doctors recommended: Mayo-Rochester, Hopkins-Baltimore,  Massachusetts General-Boston, Cleveland Clinic-Cleveland, NY Presbyterian-NYC, Emory-Atlanta, U of California-San Francisco, Duke-Durham, U of Michigan-Ann Arbor, U Pennsylvania-Philadelphia.
• ​Multiple Sclerosis: Neurologists recommended: Brigham+Women's-Boston, Cleveland Clinic-Cleveland, Hopkins-Baltimore, Mayo-Rochester, U of Texas Southwestern-Dallas. All doctors recommended: Mayo-Rochester, Hopkins-Baltimore, Cleveland Clinic-Cleveland,  Massachusetts General-Boston, NY Presbyterian-NYC, U of California-San Francisco, Brigham+Women's-Boston, Duke-Durham, Stanford Health Care-Stanford (CA), U of Michigan-Ann Arbor.

There are older lists of hospitals worldwide. 

Medicare reports many issues as quality measures, with detailed definitions, and imposes financial penalties which are estimated in the hospitals' financial statements on this site:  HAC, VBP, IQR, EHR, and readmissions.  Unfortunately all these measures create incentives for hospitals to give less treatment to sicker patients, since sicker patients hurt the hospital's statistics and earn penalties.
Medicare reports the death rate within 30 days after a hospital stay. However their death rate excludes patients who go on hospice. This exclusion reduces the apparent death rate, and creates a strong incentive for hospitals and doctors to urge the sickest patients onto hospice, whether they have a real chance of cure or not, since deaths on hospice don't hurt the hospital's statistics. As with most medical incentives, medical staff who urge hospice do not reveal their incentives. (methods)

​Veterans' Hospitals have report cards from 2008-2013 on each hospital's size, volume, staffing, deaths, infections, readmissions, patient satisfaction, etc. Each hospital has an annual star rating, and quarterly spreadsheet (SAIL) from 2016 to 2018, but the explanations are opaque. Bosses get bonuses for good outcomes, and their death rates, like Medicare's exclude patients who go on hospice, so they press the sickest veterans to go on hospice, sometimes against their will.

US News and World Report does not show an exact death rate, but shows hospital death rates (within 30 days) in deciles for particular diseases.  They do not exclude hospice or palliative care, so they don't create an incentive for hospitals to hide deaths by referring to hospice. They generally omit transfers from other hospitals, to avoid high death rates in major hospitals which receive many transfers of the very ill. So they are trying not to penalize hospitals which serve the sickest patients. They adjust to reflect all patients, not just the elderly, and slightly adjust to reduce random variation in the smallest quarter of hospitals. (methods) 

Consumer Checkbook (subscription $22 for 2 years) shows hospitals' exact risk-adjusted death rates (within 90 days), for medical patients, surgical patients, and combined, based on 10 serious medical diagnoses and 14 surgical. They use Medicare patients from Oct 2009 - Sept 2012, and do not exclude hospice or palliative care patients, so hospitals can't hide deaths by referring to hospice. They also show a combined "adverse outcomes" rate for surgical patients, the total of deaths, atypically long stays, which indicate major complications, and readmissions within 90 days of initial hospital discharge. Penalizing readmissions does penalize care for the sickest patients, who are more likely to need another hospital stay within 3 months. (methods)

Ratings of doctors by Checkbook, ProPublica and others are discussed on another page.

Hospitalinspections.org counts and describes "deficiencies cited during complaint inspections at acute-care, critical access or psychiatric hospitals throughout the United States since Jan. 1, 2011. It does not include results of routine inspections or those of long-term care hospitals. It also does not include hospital responses to deficiencies cited during inspections." (emphasis added)

Healthgrades does not show an overall death rate, but reports hospital death rates for over 20 diagnoses. For non-surgical diagnoses, they ignore deaths of patients discharged to hospice. Starting in 2014 they no longer ignore deaths of patients who saw a palliative care doctor in the hospital. Through 2013, for half of diagnoses, they omitted patients from the death rate who saw a palliative care doctor in the hospital (methods: 2014 and 2013).

Healthgrades has a risk calculator for 6 common surgeries, which shows hospitals near you with low rates of death and complications, and it shows high-volume doctors at those hospitals. Their definition of high volume is a mix of high volume on the procedure you need along with total volume for all procedures. You can get more specific information on experience with these 6 and all other procedures on another page.

NICHE (Nurses Improving Care for Healthsystem Elders) no longer lists hospitals which have adopted strong programs in Geriatric nursing, so here is a spreadsheet of their old data. Many of these hospitals have Acute Care for the Elderly (ACE) units. They encourage walking, try not to interrupt sleep, and they minimize prescriptions. They encourage hospitals to help caregivers as well as patients. 

Transplant Centers are listed by volume and outcome for each organ, or volume, waiting lists etc. and regional success rates.

Commonwealth Fund brings together data, mostly from Medicare, and lets you compare hospitals in different areas. (methods)

Leap Frog Group asks hospitals how well they comply with certain quality standards, and has answers for about a quarter of hospitals. (methods)

​QualityCheck has limited information from the Joint Commission (which accredits hospitals and other health care organizations).

Dartmouth Atlas has multiple lists with unique information on hospitals:

• Types of health care used during last 2 years of life by Medicare patients who died in 2010. Also some earlier years. This shows use of doctors, hospitals, nursing homes, hospice, home health: average days and spending. (methods)
• Readmission rate, and use of doctors during 30 days after discharge from a hospital, for Medicare patients who were admitted in 2010. Also some earlier years (methods)

Truven, (subscription) formerly part of Thomson Reuters, now uses Medicare data on deaths and readmissions (methods) at hospitals.

Consumer Reports (subscription $7/month or $30/year) groups Medicare data on readmissions into categories, and shows many hospitals at once, so it may be easier to use though less precise than the Medicare site. (methods). They also have heart surgery data on hospitals, described in the next paragraph.

Heart surgeons show 3-star ratings on about 500 hospitals and 500 group practices (typically the group of surgeons operating at a hospital), for

• Coronary artery bypass grafts (CABG) and
• Aortic valve replacements (AVR).
• For 60 hospitals they rate surgery on congenital heart defects

Almost all have 2 stars. You can search by state, not by city. They define measures (deaths in 30 days, complications, best practice), but do not say what the cutoffs are for each star, nor how they adjust for patient sickness. They probably get data from hospitals, since they encourage hospitals to sign up. Consumer Reports has an excerpt of the data, with far less detail, and fewer hospitals, mostly above-average. Their methodology does not identify the cutoffs either.

Cardiologists have a spreadsheet of 4-star ratings for use of recommended drugs by 550 hospitals after

• Implanting Cardiac Defibrillators, and
• Diagnostic Catheterization and Percutaneous Coronary Intervention (PCI/Angioplasty)

Explanations are in a user guide, with appendices showing various codes. You can search by state, city or metro area. They report the numbers of each procedure, saying, "The number of ... procedures a site performs does not necessarily indicate higher quality, but it may be an indication of how experienced this site is with the procedure." They do not report outcomes.

California rates hospital quality on:

• Hip/knee replacments: Medicare 30-day unplanned readmissions, all ages' surgical site infections, and 8 Medicare patient complications which are: heart attack within 7 days, pneumonia-7 days, sepsis/shock-7 days, surgical site bleeding-30 days, pulmonary embolism-30 days, death-30 days, mechanical complications-90 days, joint/wound infections-90days
• COPD (Emphysema or Chronic Bronchitis): readmissions and non-hospice deaths
• Childbirth: cesareans (spelling), episiotomies, breastfeeding in hospital, vaginal birth after cesarean, cesarean infections.

The same California site as above, also rates doctor groups on:

• Cancer screening of adults 50-75
• Lower back pain patients who had X-ray, MRI, or CT scan within 28 days of the diagnosis (fewer is considered better)
• Diabetes/blood: kidney function screenings, HA1c blood sugar testing, HbA1c <8.0%, blood pressure <140/90, cholesterol screenings, cholesterol LDL-C <100
• Pediatric care: upper respiratory infections and immunizations

Luxury Suites are available at many hospitals for $250 to $2,500 extra per day.

Review article in 2010 covered similar information at that date. It recommended that hospital staff should learn how hospice and palliative care affect the ratings from each group, so each hospital can get ratings as high as possible.

The Center for Healthcare Quality and Payment Reform notes bad incentives in Medicare's payment programs:

• "Patients who are not receiving adequate preventive care will be excluded [by doctors, from getting care], and providers who take on care of these patients can be financially penalized.  
• "The patients most in need of care coordination will be excluded, and providers who provide coordination to complex patients may be financially penalized.  
• "Providers can be financially penalized for keeping their patients healthy...
• "[P]erverse incentive for a physician not to become involved with a patient who already incurred significant healthcare spending earlier in the year, even though these are the patients who may most need additional help...
• "Spending Measures Do Not Distinguish Appropriateness of Services ...
• "Risk Adjustment Systems Do Not Adequately Adjust for Patient Needs"

Medicare publishes death rates to help people choose safe hospitals. However they omit your death if you have been in hospice any time in the past year. It is in the hospital's interest (even if not the patient's) to promote hospice for at least a day per year to patients who they think are most likely to die. Hospice takes them out of the reported death rate. Hospitals cannot ethically suggest coming off hospice after a day, so the patient's treatment shifts to hospice.

Medical groups (ACO - Accountable Care Organizations) have a quality standard to avoid high hospital-wide readmissions (HWR) of their patients, but patients who die within 30 days of the first discharge are excluded from that standard (p.11 and p.53523). Hospitals were rated on the same measure starting October 2014. In either case when a patient is readmitted in less than 30 days, the group or hospital looks better if the patient dies within the same 30 days, so the readmission can be excluded.

Hospitals are rated on hospital-wide readmissions from all causes, and some procedures have more readmissions than others, especially among the elderly, so hospitals have an incentive to minimize these procedures.

Healthgrades death rates ignore your death if the hospital has sent a palliative care doctor to see you, or if you are discharged to hospice. Hospitals know that some consumers use Healthgrades to evaluate hospitals, so they have an incentive to promote palliative care and hospice. 

Consumer Checkbook hospital data, ProPublica surgeon data, and US News and World Report do not exclude hospice or palliative, so they provide a more complete picture, and less incentive to push patients into hospice.

The US Department of Justice prosecutes hospitals and doctors for billing Medicare for care outside Medicare guidelines, even if appropriate under other expert guidelines. The investigations chill the willingness of doctors to provide care.

Another page gives a wide range of information on doctors' quality.

Medicare has data on size and location of nursing homes, staffing, and summary comparisons.

Most nursing homes offer limited social structure for residents, with most interactions dominated by staff.

• The Eden network works to expand quality of life in nursing homes by varied and freer social interaction.
• Green Houses are licensed as skilled nursing homes, with home-like layouts for 10 to 12 elders with low staff/elder ratios.
• Board and care homes are similar, without the skilled nursing, and in older buildings.
• Assisted living is a far larger, more professional version of board and care, usually large enough to have many activities, still without skilled nursing. Some have semester-long college courses, like the arts. They can be combined with home health services to avoid or postpone going to a nursing home.
• Village to Village is a network of local groups where volunteers and paid staff help elders at home get transportation, health and wellness programs, home repairs, social and educational activities and trips.
• PACE is a Medicare/Medicaid version of Village to Village, helping people stay at home even if they are eligible for a nursing home, in some states. Also at pace4you.org
• Senior cohousing involves homes clustered around shared facilities, run by the residents.
• Life care communities include a range of care levels, usually independent living, assisted, and skilled nursing, not run by the residents. If prepaid, they act like a long term care insurance policy, with limited reserves for long expensive nursing.
• Living in some kind of group or institution prevents the isolation of many old people at home, when their friends have died or become immobile.
• Private associations include: PioneerNetwork, LeadingAge, formerly Association of Homes and services for the Aging, and Action Pact (consultant).
• Few nursing homes take patients on ventilators or with dementia below age 60, since Medicaid does not pay enough to cover the cost

Medicare plans to pay Accountable Care Organizations to reduce the number of patients who go from hospital to nursing home and back to hospital within 30 days. This penalty discourages doctors both from sending patients back to hospitals and from sending risky patients to nursing homes ('You'll be better at home...").

Medicare's justification repeatedly cites a 2010 article (with 2006 data) by Mor et al. for the statement that 78 percent of re-hospitalizations from nursing homes within 30 days were potentially avoidable. The article made that statement without evidence. It said,

• "Specifically, MedPAC [14] has found that five conditions—congestive heart failure (CHF), respiratory infection, urinary tract infection (UTI), sepsis, and electrolyte imbalance—for which rehospitalization is potentially avoidable account for 78% of all 30-day SNF rehospitalizations."

The MedPAC report cited (2006) said,

• "Kramer and colleagues at UCDHSC identified five conditions for which rehospitalization is potentially avoidable in nursing homes... These five conditions are congestive heart failure (CHF), respiratory infection, urinary tract infection (UTI), sepsis, and electrolyte imbalance. Not all hospitalizations for these conditions are preventable; however, rates of hospitalization for these conditions were significantly lower in facilities with higher nurse’s aide and licensed staff levels as well as in facilities with higher staff retention, after adjusting for facility case mix" (Kramer et al., 2001, emphasis added).

That 2001 study addresses the number of nursing home staff needed to achieve:

• "Quality measures related to hospital transfer for potentially avoidable causes (e.g. urinary tract infections, sepsis, electrolyte imbalance) for a short-stay sample of Medicare SNF admissions."

Medicare does not report how many rehospitalizations could be prevented, though it says 97% of nursing homes do not have enough staff.

Furthermore Medicare gives no hint of how many of the 78% potentially preventable readmissions can actually be prevented, nor why they penalize 100%, when many cannot be prevented.

The article Medicare cites, by Mor et al., focuses on how much states differed in 2006 in the rates of readmission to hospitals, from nursing homes (and other Medicare costs). The article says (again without evidence) the differences depend on "provider norms, practice patterns, bed availability, and presence and willingness to use hospice." They ignore the most direct explanation for variation in readmission rates, which is variation in health. The traditional broadest measure of health is life expectancy, which does explain 34% of the 2006 variation in readmissions from nursing homes. The reasons why some states have short life expectancies include a mix of environmental hazards, poverty, diets and other causes, all of which drive health problems too, and the average readmission rate in those states is higher. Including household income, along with life expectancy, would raise the explanatory power to 40%.

On financial incentives Mor et al. say "skeptics have raised a range of potential issues including the increased incentives for selection of the most profitable patients, withholding of patient care, upcoding and fraud, along with the technical difficulties of case-mix adjustment and quality measurement and monitoring" and they do not address these issues.

A 2013 study found that skilled nursing facilities (SNFs) respond to incentives. Starting in 2002 nursing homes were rated on quality of care for patients who stayed over 2 weeks, but not under 2 weeks and not re-hospitalizations, so they sent more of their sickest patients for discretionary re-hospitalizations (before 2 weeks). At the same time they had fewer non-discretionary hospitalizations, which the authors suggest means "that nursing homes were avoiding admitting the sickest patients in the uncertainty of a new policy." (p.348)

• "Selectively discharging patients prior to their 14th day of their SNF stay will bias report card scores; in the extreme, extremely good SNFs that avoid unnecessary readmissions to the hospital may appear to be of low quality, while low-quality SNFs that discharge sicker patients appear to be of high quality. While policymakers have expressed concern that almost half of SNF patients do not stay long enough to have a 14-day assessment and therefore will not be counted (Medicare Payment Advisory Commission, 2006), the manipulation of this margin adds a layer of bias to the problem of an already selected sample (p.350)
• it is uncertain whether the current SNF NHC [Nursing Home Compare] measures can induce broad quality improvement or whether they should be used by consumers to compare quality (p.350)
• Each system has to define a denominator for quality measurement, but there is generally room for gaming the denominator in one way or another (p.350)
• If gaming the denominator is a lower-cost response to the presence of public reporting than true improvements in quality, we should expect gaming of the denominator. (pp.350-1)
  

• Policy makers worry that providers will “game” the system by selecting patients of lower risk to make quality scores look better. If providers game the system, true quality improvement may not occur, and in the worst case net welfare may decrease as sicker individuals face reduced access to care (p.341)
• selecting healthier patients may be used as a lower-cost approach than true improvements in quality in many cases." (p.343)

Like the penalties for hospitals, Congress penalizes skilled nursing facilities (SNFs) which have above-average rehospitalizations, starting October 2018. A major result will be giving SNFs a strong incentive to decline admission to the frailest patients, who need them most. Starting October 2019, Medicare will stop paying SNFs based on services they provide to patients, but will pay based on how sick the patients are.

Even without the penalty, "Medically complex patients ... can be hard to place" (MedPAC 6/07 p. 208 ). Some SNFs are "selective about the SNF patients they admit" (3/12 p.195) choosing those with "lower severity of illness" (6/07 p.204). 

MedPAC (an arm of Congress) explicitly recommends "hospice use and the presence of advance directives" as methods to decrease rehospitalizations (3/12 p.195). They give an example, "25 facilities undertook early detection ... in-facility treatment ... and improved end-of-life care strategies (such as advance care planning and palliative care) ... savings (from fewer self-reported hospitalizations) range from 17 percent to 24 percent" (3/12 p.196)

MedPAC had recommended the nursing home penalties in their March 2012 Report

• "Congress should direct the Secretary to reduce payments to skilled nursing facilities with relatively high risk-adjusted rates of rehospitalization" (p.199). 
• This will lessen their current "incentive to rehospitalize high-cost patients as a way to shift costs they would otherwise incur onto hospitals" (p.194).

MedPAC reitereated the recommendation in March 2013

• "We have recommended readmission policies for hospitals (now in place) and SNFs [skilled nursing facilities], and we are working on similar policies for home health care and IRFs [inpatient rehabilitation facilities)" (p.153)

If you want to find which hospitals have the most experience with each procedure, you need to download a spreadsheet of codes used at hospitals. These Diagnostic Related Groups (DRG) cover major types of hospital treatment. (or click for experienced doctors):

1. Download and open the spreadsheet of codes used at hospitals, Globe1234.org/hosp.xlsx
2. After it opens, click "Enable editing" at the top, if it asks
3. Click the B at the top of column B to highlight the whole column of procedure names
4. Press Ctrl and F at the same time, or click Home/Find, or View/Find, or Menu/Find
5. Type a search word in the Find box, like knee, and click "Find All." Use up and down arrows to see all the choices in the file.
6. Note the codes and search before and after them for related work. For example a search on "knee" gives 10 codes between 466 and 489. Looking at this range you will see 469 and 470 are "Major Joint Replacement Or Reattachment Of Lower Extremity." That covers knee and hip replacements even though they don't use either word.
7. They use abbreviations such as: Exc-except, Mcc-major complications and comorbidities (defined here), Mv-mechanical Ventilation, Pdx-principal diagnosis, W-with, W/O-without
   
8. Copy down any codes you want

 
Download a detailed file of hospitals' experience, the Inpatient Charge Data at:
cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Inpatient2014.html

1. Download the latest Detailed Excel data on that page.
2. Double-click the downloaded zip file.
3. Double-click the xlsx file inside it.
4. Save it so you don't have to download it again: Click File/Save As and put it somewhere you can find it, like the desktop.
5. Near the top of column A, click the small triangle to see the choices in that column
6. Unclick "Select All" or click "Clear"
7. Click one or more categories you need, such as 469 and 470
8. Click OK if asked
9. At the top of Column i - Total Discharges, click the small triangle and click "Sort Largest to Smallest"
10. Now the highest volume hospitals are at the top
11. You can also see the average amount Medicare allowed for each procedure, and the average charges each hospital submitted for the procedure.  These costs do not include the doctor bills, which you would need to find on the Specialists page.

Penalties on heart failure, heart attacks and pneumonia have been paid since October 2012, based on readmissions back to July 2008. Penalties on emphysema, chronic bronchitis (COPD), knee and hip replacements will be paid starting October 2014, based on readmissions back to July 2010.

Medicare calculates new penalties each fiscal year:

Penalties charged in any one year are based on readmissions in three years ending 15 months earlier.

NIH says heart failure means a weak heart. They say an alternate name is congestive heart failure.

NIH says knee or hip arthroplasty (Medicare term) means replacement.

NIH says COPD means emphysema or chronic bronchitis.

NIH says AMI, acute myocardial infarction, means heart attack.

NIH defines comfort care as symptom relief at the end of life, along with mental and spiritual comfort for terminal patients, so when doctors mention comfort care, they do not mean cure.

NIH says palliative care includes many treatments at any stage of illness, but they immediately discuss advance directives, DNR, and refusal of life-sustaining treatment. They also consider palliative care part of the same research field as end of life. While they say palliative care does not mean intent to die or forgo treatment, the message is very mixed, which is why many doctors and HealthGrades ratings of hospitals think it signals less treatment. Often it causes drowsiness, so patients participate less in decisions.

Critical Access Hospitals, designated state by state, are generally rural with less than 25 beds, average stays under 4 days, and 35 miles from other hospitals. (42cfr485.601 to 647). They are marked in our "Combined list" of hospitals, and are identified by "13" in the middle of the hospital Id number. They get extra payments to support better care than small hospitals otherwise could give.

Accountable Care Organizations (ACO) are groups of health providers who get paid more if they lower Medicare costs for the patients they see and meet minimum quality standards, including reducing admissions and readmissions (pp.10-13). Readmissions do not count against them if the patient dies within 30 days of initial discharge, and deaths do not count at all. Patients do not sign up. Medicare tracks which patients get most of their care from the ACO, and then rewards the ACO if Medicare saves money on these patients.