Where does the data come from?

If you read a few articles on the topic of medical errors, you are likely to encounter a number: 98,000, an estimate of the number of patients that die each year in the U.S. due to medical errors. In this post I am going to explain a bit about how research on the prevalence of medical errors is performed and where that often-repeated number comes from.

I was glad to find a helpful overview of medical error research in chapter 2 of Tom Baker’s book, The Medical Malpractice Myth. Using those references, I gathered the original research papers. Here is what I found.

The California study, 1974

  • Funders: The California Medical Association and the California Hospital Association.
  • Purpose: To determine the frequency and severity of injuries which would need to be compensated if the state switched to a malpractice system that didn’t require litigation, but instead directly compensated injured patients. This type of system is sometimes called a “no-fault system”, and it was thought that this approach might be more economical.
  • Type of study and source of data: A hospital record review. The data comes from a random sample of over 20,000 patient records from 23 California hospitals.
  • Methods: The records were read and “potentially compensable events” (PCEs) were identified and reviewed. These PCEs were graded for severity, from minor temporary disability (less than 30 days) to death.
  • Results: The researchers found that 4.7% (nearly 1 out of 20 patients) admitted to these hospitals experienced a PCE. How severe were these injuries? About 80% were temporary disabilities lasting 30 days or less. But nearly 10% resulted in death.
  • Outcome: Because only a small fraction of the number of injured patients were taking hospitals and physicians to court, a no-fault system would not have provided a financial benefit to physicians and hospitals. And medical and hospital associations didn’t publicize the number of medical errors they were making.
  • Report: A technical summary was published in 1978 and can be downloaded here.

The New York study, 1984

  • Funders: The State of New York.
  • Purpose: To estimate the prevalence of injuries due to medical errors in New York hospitals, and the percent of those injuries due to negligence. Like California a decade earlier, New York was concerned about the cost of medical malpractice litigation and interested in the possibility that a no-fault system would be less costly.
  • Type of study and source of data: A hospital record review. The researchers gathered a random sample of over 31,000 patient charts from 51 hospitals in New York.
  • Methods: The charts were screened in a two-step process. Nurses first identified cases of possible medical injury (which they term “adverse event”), then these cases were independently reviewed by two physicians. Each physician first determined whether there was evidence for injury, and if so, the strength of evidence that the injury was caused by negligence.
  • Results: They determined that about 3.7% of hospitalizations resulted in injuries. About a quarter of these injuries were due to negligence. Of all injuries, about 13% led to death. Half of these deaths due to injuries were caused by negligence.
  • Outcome: The data showed that many more patients were injured than were filing malpractice claims, and New York did not implement a system for compensating all injured patients.
  • Report: You can read the academic abstract here.

The Utah and Colorado studies, 1992

  • Funders: The Robert Wood Johnston Foundation (a philanthropy devoted to public health).
  • Purpose: To determine if the results from the New York study applied to other areas of the country.
  • Type of study and source of data: A hospital record review of over 15,000 patient charts (5,000 in Utah and 10,000 in Colorado).
  • Methods: This was conducted by the same group from Harvard that did the New York study, and they used methods similar to their previous study.
  • Results: The results are similar to the New York study. They estimated that 2.9% of patients were injured while receiving care, and about a third of these injuries were due to negligence. Of all injuries, about 7% led to death. About a quarter of the deaths due to injuries were believed to be caused by negligence.
  • Report: You can read the academic abstract here.

An undisclosed hospital in Illinois, 1989-1990

  • Funders: The Robert Wood Johnson Foundation, the American Bar Foundation, and the MD Anderson Foundation.
  • Purpose: To understand the prevalence and types of adverse events in a hospital setting.
  • Type of study and source of data: An observational study, with data collected by analyzing discussions of adverse events at a single large urban hospital.
  • Methods: Over a period of nine months, four ethnographers attended rounds (when doctors visit patients in their rooms), case conferences, and other meetings. A total of 1047 patients were discussed by healthcare workers. The observers recorded information about all adverse events that were mentioned, but they did not ask questions.
  • Results: 46% of patients had at least one adverse event. The likelihood of experiencing an adverse event increased with seriousness of illness and length of hospital stay. 18% of patients had at least one serious adverse event that led to a longer hospital stay and increased cost to the patient (but only 1.2% of patients made a claim for compensation). For those adverse events in which a cause was identified, 38% were caused by an individual healthcare worker, 16% were due to interactions between individuals, and 10% were due to administrative decisions (such as inadequate staffing).
  • Report: You can read the academic abstract here.

So where does 98,000 come from?

Each of these studies used a very specific population of patients. But the results can be used to estimate the rate of adverse events and medical errors for similar populations of patients. For example, the California data can be used to estimate that approximately 140,000 people were injured badly enough by California hospitals in 1974 that they probably should have been compensated. And for all patients who entered a California hospital in 1974, a bit less than 2% of them died as a result of errors in their treatment.

The 98,000 estimate comes from applying the data on deaths due to medical errors from the New York study to the entire population of hospital patients in the United States. This statistic was a centerpiece of the To Err is Human report published by the Institute of Medicine in 1999.

Because this estimate of 98,000 is based on data from the New York study, it has some limitations. That study did not attempt to account for deaths due to errors that occurred long-term care facilities or during out-patient procedures. The researchers had no way of including deaths due to errors that were undetected by the hospital or that weren’t obvious from the medical records. For these reasons and others, this number of 98,000 has been criticized as being too low. (And remember it accounts only for deaths, not disabilities that did not result in death.)

So now what?

There is no question that far too many people experience medical errors in this country. In my opinion, it doesn’t matter if the true number of deaths is 98,000 or 200,000 or more. What matters is what is being done to reduce the number. Policy in healthcare should be based on research and ethics. Individual hospitals need to look to research and ethics, but so do state governments. I hope that my family’s legislative bill in Kansas not only brings the public’s attention to the problem of medical errors, but also starts a conversation about how to hold the healthcare system accountable for reducing these errors.

So what number is important number to me? “1”. The is the number of errors that I know were made in my father’s care on March 30, 2012.


Thoughtful comments are always welcomed.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s