Medical Meccas: Which Hospital Is Best?

How good is a hospital? many criteria come into play--the reputation of the physicians, or how well equipped a facility is, or the quality of food and patient services. As health care becomes more a global enterprise, with patients going farther afield in pursuit of the best care, the need for an objective measure of hospital performance has become more pressing.

Fortunately there is a yardstick that applies equally well to the most luxurious tourist hospital in Thailand and the most basic NGO-run facility in Africa, and everything in between. Most countries are required by law to keep track of how many patients live and how many die. Once these mortality statistics are adjusted for factors such as the age of the patients, the diagnoses and the severity of the illnesses, they can provide a measure in black and white of how well a hospital is managing the performance of its physicians. Although the medical profession has put up some resistance to efforts to measure, and publicize, hospital performance in this way, in recent years doctors have been collecting these data on a growing list of hospitals around the world.

Doctors have long known that they are not exempt from the human proclivity to make mistakes now and then. But by and large the medical profession has been slow to face this fact in any formal way. That began to change 15 years ago, when the Harvard Medical study, produced by a team led by Prof. Howard Hiatt, found that about 10 percent of hospital deaths are related to avoidable adverse events. Much of the problem is that hospital managers simply don't pay enough attention to mortality rates. In 1999, the Bristol Royal Infirmary Inquiry in the U.K. investigated the reasons behind the unusually high number of pediatric-cardiac-surgery deaths at that hospital. One reason, the inquiry concluded, is that it was never clear who was responsible for monitoring death rates. Most Western countries have extensive databases of hospital records, usually going back for decades, but for some reason they haven't been used very much as a means of picking up problems and monitoring the effects of reforms. Steps taken at the Bristol Royal Infirmary, however, reduced the adjusted pediatric-cardiac-surgery death rate in that hospital by two thirds within a year, to one of the lowest rates in the country.

For several years, the Dr. Foster Unit at the Faculty of Medicine in Imperial College, London, has measured hospital standardized mortality ratios (HSMRs) to assess quality of care and to detect where problems may lie. We gather these data on hospitals in the United States, the U.K., Canada, the Netherlands and Sweden, and use them to monitor changes often on a monthly basis. We publish the U.K. data annually.

As doctors can be prickly at the prospect of hospital mortality data's being publicized, we haven't published the data gathered in the United States. Instead, we send the information to hospitals on request. Most U.S. hospitals have asked for their data. The HSMRs for 29 hospitals in Minnesota were published on a Web site in 2005. Sweden, the Netherlands and Canada have also started campaigns to reduce hospital deaths and are considering publishing their HSMRs. More than 3,000 U.S. hospitals participated in a campaign of the Institute for Healthcare Improvement (IHI) in Boston to reduce mortality rates.

Our experience has shown that mortality rates can be cut dramatically in short order. When we first published HSMRs in England in January 2001, Walsall Hospital had the highest rate. Administrators and clinicians responded by putting procedures in place that focused on mortality, and four years later had reduced deaths by 300 per year.

Doctors are naturally concerned about malpractice suits and about their reputations. Governments and other supervisory bodies may find it difficult to deal in matters of clinical performance with practicing clinicians, some of whom may be well known or practice at famous hospitals. But patients have a strong interest in making sure that problems affecting the quality of care are dealt with in an open and effective way.