A 37-Year-Old Man Trying to Choose a High-Quality Hospital: Title and subTitle BreakReview of Hospital Quality Indicators

DR REYNOLDS: Mr A is a 37-year-old executive who is looking for a new physician and health care system.

In November 2006, Mr A awakened with dull chest heaviness; after about 90 minutes, he called 911 and was brought to a local hospital, where he was diagnosed as having a myocardial infarction by electrocardiogram and enzyme measurement. Emergency cardiac catheterization revealed normal coronary arteries. Twenty-four hours later, he again experienced chest discomfort. Catheterization again showed normal coronary arteries.

Hematologic consultation revealed a slightly elevated dilute Russell viper venom time (dRVVT) test result on 2 occasions. Mr A's hematologist interpreted these results as suggestive of the presence of lupus anticoagulant and suggested anticoagulation with warfarin and, later, retesting of dRVVT values, with consideration of discontinuation of the warfarin if the levels decreased to normal.

Mr A therefore carries diagnoses of Prinzmetal angina and hypercoagulable state.

After discharge, Mr A initially followed up with the cardiologist who cared for him during his inpatient stay at the local hospital. However, Mr A ultimately decided to transfer his care to a larger center a longer distance away, seeking a cardiologist who would provide care with attention to the etiology of his myocardial infarction and who practiced in a “less chaotic” environment. He appreciated his new cardiologist's attention to his medication adverse effects and the system of care that included reminders about getting his international normalized ratio (INR) measured.

In the spring of 2009, Mr A's new cardiologist decided to leave the area and Mr A again found himself looking for the right physician and hospital for his health care.

Mr A's current medications include extended-release diltiazem, 180 mg/d; aspirin, 81 mg/d; nasal steroid spray; and warfarin, adjusted to an INR of 2 to 3.

Mr A lives with his wife and 2 children, 1 of whom has a congenital heart defect; he works and has commercial medical insurance. He does not smoke tobacco or use drugs; his alcohol use is limited.

On physical examination, he was normotensive with a normal cardiovascular examination; an electrocardiogram showed nonspecific ST-T wave changes and stress test results were normal aside from a left ventricular ejection fraction of 47%.

I was a mid-30-year-old who had standard problems. I had a general practitioner that I would go to for an every-5-year physical and various illnesses I would have.

All that changed when I unexpectedly had the heart attack, and at that time I had no cardiologist or anything. Obviously, I called 911 and went to the hospital that they took me to. A cardiologist on staff there was assigned to me.

The cardiologist I’m seeing now has transferred hospitals. As a result, I may technically be, at the moment, without an assigned cardiologist. So I am now at a point where I need to evaluate what it is I’m looking for in a doctor.

First, I have an interest in finding a hospital of a high caliber but that is also close enough that I could be transported there. [One] of the things that I really appreciated [from my cardiologist's practice was] a registered nurse who was under the cardiologist I was seeing was extraordinarily good about following up to remind me to get my PT [prothrombin] and INR scores. . . . I could easily go 3 or 4 months without realizing I had to take this test. So having her chase me down, just send me an e-mail once in a while, was fantastic.

I do have a primary care physician and I do like him. I’m also looking for a practitioner who is going to dive in a little more than the surface of the studies and try to really understand. But I will admit that I don't have a way to find out who that is other than trying different cardiologists over the years.

If I had to choose whether it was more important to find the right cardiologist or the right hospital, I think my first priority would be finding the right cardiologist because of the unusual nature of my case. But, even as I answer that question, I’m questioning it because hopefully the cardiologists go back and discuss with the whole department and bounce ideas off each other. So a strong department—as well as having strong nursing staff to support them, to follow up with me—would also be important.

In order to make my decision, I have talked to some friends who are doctors and who have some knowledge at this level. I’ve also looked on the Web but, unfortunately, what you wind up finding is the US News & World Report that ranks the top 100 cardiac hospitals, which only has some value. The reports are very superficial and they don't get to the questions I was asking.

The reason I have shrugged off the US News & World Report and other large distribution magazine “top-whatever” lists is that the methodology used is based on how much time the facility is willing to invest in filling out the form and returning it. The reports are also biased toward measurements such as success—number of deaths per admitted patient–type statistics. An excellent hospital may take on much tougher cases, but as a result they are going to have a higher death rate, and that isn't an adequate measurement of quality.

The other thing I do is actually Google some of the doctors' names and see what I can find about them and read their bios that they put up on hospital Web sites. But half of the profiles don't even exist or are empty templates. Others haven't been updated in a decade.

I’m such an unusual case that I often wonder: Do the general findings overall—since most people are going to go in with clogged arteries from lifestyle choices—do those outcomes impact me?

I would be very interested in how to improve this process of looking for a doctor. Getting information and evaluating it is tough. I want to know what the trends in health care quality are and I want to know what information patients can use.

How many individuals in the United States change care among health systems, hospitals, and/or clinicians each year and why? Should a patient choose a primary care physician, a specialist, or a health system? How can consumers/patients learn about the quality of care provided by hospitals, health systems, and physicians? How reliable and accurate is the public information and does it actually measure quality of care? What do you recommend for Mr A?

DR HOWELL: Mr A has a demanding road ahead. With 2 complex medical conditions, through no fault of his own he now finds himself bereft of the physician whom he identifies as primarily managing his care. Nearly all of us, at one time or another, have faced the challenge of finding a new health care system and a new physician, either for ourselves or for a close family member. Even for those of us whose careers lie within health care, it is not an easy task.

Tens of millions of individuals in the United States change health care plans, hospitals, and/or clinicians every year. Millions are forced to do so because of geographic relocation or loss of employment. About 38.6 million people move annually; of these, about 12.3 million move to a different county and about 1.2 million arrive from outside the United States.¹ So, about 14.5 million Americans are likely to change hospitals and physicians every year for this reason alone. Job loss also often brings with it a loss or at least a change of insurance. Since the current recession began in December 2007, the number of unemployed individuals has increased by more than 7 million.² More than half of those who are unemployed are also without insurance. Another quarter rely on public coverage.³

However, Mr A neither moved nor lost his insurance. Every year, millions of people with stable geography and employment also find themselves in this situation. The Community Tracking Study, a series of surveys of more than 37 000 nonelderly US residents, found that 13% of persons with private insurance changed their usual source of care during a single year. Even among those with stable health plans (like Mr A), 11% changed their usual source of care during the year.⁴ In a follow-up study 2 years later, the rate of switching was stable at 13%.⁵ Other large studies have found similar results. More than 14% of California's population had discontinuity in insurance status during 2003,⁶ and federal employees switch health plans at a rate of 12% per year.⁷

Quality of Health Care. It is important to understand what Mr A means when he says he is seeking a high-quality hospital and also to recognize that a single, unifying measure of health care quality may not exist. Donabedian's 1966 approach to understanding the quality of medical care⁸ sounds an important note of caution for Mr A today:

The assessment of quality must rest on a conceptual and operationalized definition of what the “quality of medical care” means. Many problems are present at this fundamental level, for the quality of care is a remarkably difficult notion to define.

Klein et al,⁹ 5 years earlier, had been still more forceful:

We found that a single, comprehensive way of measuring patient care was impossible. . . . In other words, it seems quite likely that there will never be a single comprehensive criterion by which to measure quality of patient care.

Although these concepts continue to confound answers to Mr A's questions, Donabedian defined 3 approaches to measuring quality that remain relevant today: structure, process, and outcome. Applied to Mr A's myocardial infarction, his hospital had structural elements required for high-quality modern cardiac care (eg, having a cardiac catheterization laboratory), they met process criteria for high-quality ST-elevation myocardial infarction care (eg, rapid cardiac catheterization), and the outcome was favorable (he is still able to live an active life). Most publicly available quality data can be thought of in 1 of these 3 categories.

Individuals Embedded in Systems. A major challenge in providing practical information that quantifies quality of care arises because physicians operate embedded in health care systems at many different levels. An individual physician may be part of a group practice, multiple hospitals, and several health plans simultaneously. These organizations, in turn, are embedded in regional and national health care systems. More than 50 years ago, Coleman et al¹⁰ showed that these networks are important: physicians' adoption of a new therapy was strongly influenced by their connections to and integration with other physicians. Physicians' practice is also meaningfully influenced by their practice organization,¹¹ health plan,¹² hospital,¹³ region,¹⁴ and country.¹⁵ So, the practice environment modulates both process and outcome of care provided by individual physicians, and Mr A will have to pay attention to both during his selection process.

Mr A should start by choosing a primary care physician embedded in a hospital or health system with markers of high quality. Care by a primary care physician is important in ongoing care, since Mr A is more than the sum of Prinzmetal angina and a (possible) hypercoagulable state. Epidemiologic evidence links higher rates of primary care with better outcomes. Shi¹⁶ found that as the number of primary care physicians in a state increased, cardiovascular and cancer death rates decreased and life expectancy improved. A systematic review found that for every additional primary care physician per 10 000 population, mortality rates decreased by an average of 49 per 100 000.¹⁷ Other studies have found concordant results for the relationship between primary care physician supply and reductions in breast,¹⁸ cervical,¹⁹ and colorectal cancer–related²⁰ mortality.

Why might this association exist? Quiz Ref IDPrimary care is extraordinarily complex. Providing recommended preventive care for an average primary care physician's panel may require more than 7 hours every day.²¹ Primary care physicians coordinate a broad array of care; the average primary care physician for Medicare patients works with 229 other physicians in 117 other practices.²² Although specialists are often the usual source of care for many patients,²³ they may provide less reliable preventive care than generalists.^{23 - 24} Finally, stability of care is associated with improved health outcomes. In a randomized trial of continuity of care, patients randomized to the higher-continuity group had fewer emergent admissions, shorter lengths of stay, and higher satisfaction than those with less continuity among outpatient general medicine physicians.²⁵ A later systematic review found that sustained continuity of care was associated with improved patient satisfaction, fewer hospitalizations and emergency department visits, and better preventive care.²⁶ This is far from saying that Mr A should receive only primary care. Rather, the weight of evidence suggests that for specific diseases and conditions, including cardiac disease,^{27 - 28} patients receive better care from specialists operating in their domain of expertise rather than solely from generalists. Mr A should therefore seek coordinated care between his specialists and primary care physician.

Unfortunately, Mr A will find little evidence-based guidance to help him choose an individual physician. In general, currently available performance measures do not provide the kind of reliable comparisons among physicians that Mr A might want. On the other hand, he can easily find out whether his physician is board certified from the American Board of Medical Specialties (eTable, a). Although studies are heterogeneous,²⁹ quality of care and patient outcomes are associated with physicians' board certification, licensing test results, and certification scores.^{30 - 31} Mr A can also check his physicians' licensure status, public disciplinary history, and sometimes other factors by querying his state medical board (at individual state Web sites or aggregated by Administrators in Medicine eTable, b).

In certain locales, Mr A can also review practice-level quality measures. The Agency for Healthcare Research and Quality (AHRQ) maintains a compendium of physician-, health plan–, and practice-level scorecards (eTable, c). As of this writing, the compendium includes 209 report cards, of which 35 provide physician- or practice-level information. For example, in Massachusetts, patients can review practice-level information about colorectal cancer screening, diabetes treatment, and other conditions (eTable, p). However, as Mr A himself points out, few of these measures apply to a 37-year-old man with his medical issues. Several other barriers limit these measures' usefulness. Quiz Ref IDFirst, physicians' practices are embedded in multiple other layers of the health care system, making physician-specific effects hard to tease out and requiring complex statistical analysis.³² Second, validated quality measures do not exist for all specialties.³³ Third, patient sample size may be insufficient to identify high- or low-performing physicians, particularly for small practices and for physicians who see a broad range of diagnoses, such as primary care physicians.^{33 - 34} In one study, individual physician profiling for diabetes could not distinguish quality from random variation even with complex case-mix adjustment.³⁵ Others have found that physician panel sizes are often too small to support comparative benchmarking.³⁶ Finally, for patients like Mr A with complex, uncommon diseases, the most important attribute of a physician may in fact be diagnostic reasoning and critical thinking. These complex sets of skills are challenging to practice, teach, and measure^{37 - 38} ; they are not well-captured in most existing quality metrics. These limitations may contribute to the fact that most internists believe that current quality measures are inaccurate and inadequately adjusted for patients' characteristics.³⁹ Some physician-level efforts have focused on higher-volume procedures such as coronary artery bypass graft surgery and percutaneous coronary intervention (eTable, m, n, and o). These may be useful to Mr A depending on which state he lives in. Although hotly debated^{32 ,40} and potentially complex to interpret, these reports do provide patients with operator-specific, risk-adjusted mortality outcomes. Finally, patients can post qualitative reviews of their experiences with clinicians,⁴¹ although no trials to date assess the relationship of these ratings to outcomes.

A number of sources provide information on hospitals' quality of care. AHRQ's compendium includes 78 health care report cards with hospital information as of this writing (eTable, c). Although most people, like Mr A, would prefer a single, simple quality measure for a hospital (like those often given to consumer purchases, wines, etc), a hospital's quality may vary depending on what care is required. Evaluating quality data differs for a specific procedure vs future general care.

Overall Evaluations of Hospital Quality . Several sources can help Mr A evaluate overall hospital quality. These rely on surveys, analyses of administrative data, or a combination of the 2. The best known, which Mr A himself mentions, is US News & World Report's “America's Best Hospitals.”⁴² This annual report aims to “help direct patients to centers that excel in the most difficult cases,” not to identify hospitals that provide the best routine care.⁴³ It integrates structural factors (such as membership in the Council of Teaching Hospitals) with a physician survey (assessing hospitals' reputations) and additional measures derived from sources such as the American Hospital Association Annual Survey and Medicare Provider Analysis and Review (MedPAR). The rankings are associated with clinical outcomes. In a large study of patients with acute myocardial infarction, admission to a top-ranked hospital was associated with a lower mortality rate compared with unranked hospitals. This appeared to be mediated by more reliable processes of care.⁴⁴ Quiz Ref IDHowever, although the average top-ranked hospital performs better than the average non–top-ranked hospital, there is considerable overlap. A study assessing top-ranked hospitals' process of care for acute myocardial infarction found that 31% scored below the average hospital in a group of 774 hospitals. Conversely, more than 40% of nonranked hospitals had better performance than the average top-ranked hospital in a cardiovascular composite measure.⁴⁵ Another study found that top-ranked hospitals had lower average risk-adjusted mortality rates after acute myocardial infarction but again found substantial overlap with unranked hospitals.⁴⁶ Another well-known “top hospital” list is published by the commercial service HealthGrades.⁴⁷ Ratings are based on MedPAR data. Similar to the US News rankings, HealthGrades' top-performing 5-star hospitals performed, on average, significantly better than 1-star hospitals for patients with acute myocardial infarction, but more than 90% of 1-star hospitals had a risk-adjusted mortality rate that was not statistically different from that of a 5-star hospital.⁴⁸ Thomson Reuters' 100 Top Hospitals uses MedPAR, the Medicare Cost Report, and other data to provide a composite ranking that attempts to balance risk-adjusted survival, complications, satisfaction, and financial performance.⁴⁹ When compared with nonranked hospitals, highly ranked hospitals had lower resource utilization but similar clinical outcomes.⁴⁴

Patient satisfaction and the experience of care is also a key dimension of quality health care.⁵⁰ Medicare's Hospital Consumer Assessment of Health Providers and Systems (HCAHPS) assesses communication with physicians and nurses as well as several other factors (eTable, d). Higher HCAHPS results are associated with better quality of care for acute myocardial infarction, congestive heart failure, pneumonia, and surgery.⁵¹ Finally, many online services enable patients to rate and review their hospitals and clinicians.⁴¹

What If Mr A Wants More Details? Mr A may also be interested in services that do not attempt to roll up hospital quality into a single measure. Four free sites supplying this kind of information are The Joint Commission's Quality Check (eTable, e), the Leapfrog Group (eTable, f), UCompareHealthCare (eTable, g), and the Centers for Medicare & Medicaid Services' (CMS’) Hospital Compare (eTable, d). The Joint Commission's Quality Check provides information about hospital accreditation and several process-of-care measures. On average, accredited hospitals and those with better process measures have better outcomes for cardiovascular care, though the magnitude of association may be small.^{52 - 53} The Leapfrog Group, a consortium of employers and purchasers, surveys the practices of more than 1200 hospitals annually.⁵⁴ Leapfrog provides a selection of structure, process, and outcome measures accessible to patients. However, studies of the validity of Leapfrog measures have been mixed. Hospitals that implemented Leapfrog-recommended practices generally have better processes of care and better outcomes,⁵⁵ but the Safe Practices Survey component of Leapfrog was not associated with risk-adjusted mortality.⁵⁶ UCompareHealthCare aggregates process and outcome measures from several sources and provides some risk adjustment.⁵⁷ While aggregation is promising, I found no external studies of the validity of this site, and a review of my own hospital found important differences between services that we offer (eg, computed tomography scanning) and those reported on the site. Finally, CMS’ Hospital Compare presents hospital-level process measures for a number of diseases, outcome measures for 3 conditions, and patient satisfaction information. Better performance on these measures is associated with a small but significant reduction in risk-adjusted mortality.⁵⁸

Specific Procedures and Conditions: The Volume-Outcome Hypothesis and Condition-Specific Reports. In general, if Mr A needs care for a specific condition or procedure, he should seek a hospital with substantial experience with that condition.Quiz Ref ID Most data suggest that hospitals with higher volumes have better outcomes for specific procedures and conditions, including some surgeries,^{59 - 60} pulmonary embolism,⁶¹ respiratory failure,⁶² some medical intensive care unit admissions,⁶³ correct emergent diagnosis of acute myocardial infarction,⁶⁴ and mammography.⁶⁵ Particularly relevant to Mr A, hospitals with high-volume invasive cardiac catheterization procedures appear to have better outcomes.⁶⁶ There are a few important caveats to this: volume-outcome relationships have not been documented for all procedures,⁶⁷ these relationships may change over time as quality improvement programs are implemented,^{68 - 69} the magnitude of effect is sometimes quite small,⁷⁰ and the etiology of the relationship is not fully understood.⁷¹

Mr A may want to check volume data available online. Leapfrog reports the annual volume of several procedures for each hospital, along with Leapfrog-determined benchmark volumes (eTable, f). Medicare also provides publicly available procedural volume data for Medicare patients (eTable, d). Mr A's state may also provide data. For example, Massachusetts publishes some hospital- and surgeon-specific procedural volumes (eTable, h) and New York publishes hospital-level volume statistics (eTable, i).

Mr A also has access to hospital-specific, risk-adjusted outcome data for a few conditions and procedures. CMS reports hospitals' risk-adjusted mortality rates for patients with acute myocardial infarction, heart failure, and pneumonia compared with the national average (eTable, d). Leapfrog also provides ranked survival predictions for pancreatic resection and esophageal resection⁷² and incorporates risk-adjusted mortality rates into their ratings for several cardiac procedures. Some states provide hospital-specific, risk-adjusted mortality rates for several procedures and conditions (eTable, j, k, and l). Other states publish hospital-specific, risk-adjusted mortality rates for coronary artery bypass graft surgery and percutaneous coronary intervention (eTable, m, n, and o). Importantly, risk adjustment has inherent limitations,⁷³ and evaluating whether hospitals are outliers is statistically complex.^{32 ,40}

Comparing the Approaches. Mr A might ask, “Do these sites all give me the same answer?” Unfortunately, even for common surgical procedures, different sites may provide different ordinal rankings for the same hospitals.⁷⁴ A recent comparison of several sites found that, even for similar-sized hospitals in a single metropolitan area, these sites provided inconsistent and often conflicting rankings.⁷⁵

It is frustrating to be unable to provide better evidence-based guidance to Mr A. Nonetheless, decisions have to be made, even under conditions of uncertainty, and not making a decision is its own decision.⁷⁶ First, I would advise Mr A to become familiar with a consumers' guide to approaching the uncertainties and complexities of the health care system, such as those by AHRQ (eTable, q) or the nonprofit Consumers Union (eTable, r). Second, he should discuss specialist and hospital selection with his primary care physician, with a particular focus on those that are geographically convenient, since this is important to him. At a minimum, he should assess whether his physicians are board-certified and he should review state medical board information, including licensure status and public disciplinary history (eTable, b). He should assess quality measures for the potential hospitals where his future cardiologist might work, seeking hospitals with consistently higher, broad-based quality ratings. Because there are no quality metrics for Mr A's specific combination of issues, he will need to do some additional detective work. The best surrogate measure that he is likely to find—unvalidated though it may be—is volume, and he should feel comfortable asking potential cardiologists, hematologists, and hospitals for an approximation of how many cases of each condition they see per year. He should prepare for the visits, preferably bringing copies of prior records and tests (eTable, q and r). Mr A would like e-mail reminders to help him monitor his warfarin and related laboratory tests; he should discuss these practice-level requirements early on, either over the phone when making appointments or at his first visit. In particular, warfarin is a high-risk medication, and systems with dedicated anticoagulation management programs may have superior outcomes.⁷⁷ Finally, after he meets his physicians, he should assess his comfort with the encounter. If things went well, he should let his physicians know what works for him, so that future encounters will proceed smoothly. If he is not satisfied, he should discuss his expectations with his physicians so that there is an opportunity to improve. If things do not work out, he might need to seek other physicians.

QUESTION: Isn't it the greatest failure of social justice and self-regulation for us to assume that patients should have to shop for high-quality medical care, knowing that the most vulnerable members of our society will be least able to succeed in applying this market-based approach?

DR HOWELL: Right now, a national debate rages over whether health care is a right, a privilege, or something in between. This debate won't be settled in time to help Mr A find a new cardiologist. So, I have tried to focus on the practical questions facing Mr A and millions like him: where can they look, right now, for guidance in trying to find high-quality care. Moreover, although social justice is a fundamental principle laid out in the Charter on Medical Professionalism, so are patient welfare and patient autonomy: not only must we physicians work to improve our practice, but “[p]hysicians must be honest with their patients and empower them to make informed decisions about their treatment.”⁷⁸ Providing transparent, accurate, reliable information about the quality and safety of our work seems to align with these principles. Moreover, improved quality can sometimes narrow disparities.⁷⁹ On the other hand, poorly designed quality reporting and pay-for-performance programs create a terrible risk of further widening already egregious disparities in health care.^{80 - 81}

QUESTION: I’m a believer in the Leapfrog-style evaluation system, but in my experience, few of our patients either know to pay attention to them or care about them. Do you know how many patients look at quality data now before choosing a physician, and do you see that increasing in the future?

DR HOWELL: Patients actually report great interest in information about quality of care.⁸² However, experimental evidence suggests that how quality information is presented materially affects whether patients use it or not.⁸³ In addition, consumer awareness of available quality data may be quite limited. For example, 5 years after the inception of a major guide to coronary artery bypass graft surgery, only 1% to 2% of cardiac surgery patients knew their hospital's or surgeon's rating.⁸⁴ By 2006, only 7% to 10% of Americans were using quality information to select a physician or hospital.⁸⁵ Not surprisingly, then, the empirical evidence on effectiveness of public reporting is mixed. Two recent systematic reviews concluded that there is limited evidence that the availability of information about quality affects consumers' choice of health care.^{83 ,86} However, the title of an accompanying editorial (“What Can We Say About the Impact of Public Reporting? Inconsistent Execution Yields Variable Results”)⁸⁷ points out the real subtlety in these data. The weight of evidence suggests that public reporting efforts have prompted hospitals to work on improvement and have had moderate effects on consumers' choices of health plans; in other areas, effects have been mixed or null.⁸⁷ In my own opinion, there are compelling reasons to believe that this is a problem of implementation rather than theory, but the evidence has not yet borne that out.

QUESTION: As a primary care physician, how should I be using quality data to inform my specialist referrals? We are typically encouraged to refer “in network,” but should I be looking geographically for the highest-volume, or highest-quality, specialist?

DR HOWELL: Most existing publicly available hospital and physician performance information has been targeted at patients, not physicians. Quiz Ref IDHowever, in a survey of Medicare patients who had undergone major surgery, a critical factor in selecting a surgeon or hospital was their referring physician's opinion. Patients used this source of information more than twice as often as recommendations by family or friends and 4 times more frequently than mortality data. Moreover, although they believed that performance data were important, in planning for future surgeries 95% wanted their physician to help them understand and interpret this kind of information.⁸⁸ Furthermore, patients reported that the physician was the main decision maker about whether to undergo surgery 31% of the time and was an equal decision maker in an additional 42% of cases.⁸⁹ Little interventional research has addressed how the provision of quality data affects primary care physicians' referral behavior. Previous observational studies have not found evidence that primary care physicians systematically incorporate quality data into referral decisions; rather, personal knowledge of the specialist and the quality of prior feedback were identified as the most important factors in selecting a particular specialist.⁹⁰ Together, these suggest that a promising opportunity may exist to direct quality and performance information toward referring physicians, in addition to patients.