Owning up: tests that were not done were reported as normal (1); And tests that were done were not reported at all.

A physician's mother falls and is briefly unconscious. She is evaluated and sent for a CT scan, which she refuses.

[ILLUSTRATION OMITTED]

Unfortunately, neither the family nor the attending physician is notified; indeed, the attending dictates "normal CT scan" in the discharge summary. The woman later sees a neurologist who does not check the actual scan, believing the family's statement that it was normal.

The woman and her husband travel out of state, where a second fall, causing a hip fracture with subsequent delirium, eventually leads to a difficult transfer home. At the original hospital, another CT scan (actually, the first) is performed, and only the technologist's remarkable recall of the patient's prior refusal to have the first scan is the reason the error is discovered.

The actual scan shows a small subdural hematoma and significant unilateral atrophy that of course could not be compared with a prior scan. However, it probably did not affect the ultimate course.

Your thoughts?

1. It ultimately didn't matter, so move on. These things happen.

2. You are kidding.

3. Why did this happen?

The errors:

* There was an inadequate system to ensure both the attending and the family knew when a test was not performed.

* The information was in the ED record but was not seen by others. Again, because something is charted does not mean it is read.

* No system was present to ensure that tests that were done were reported.

* There was an assumption that the family's recall of a test result was correct. (How often do we actually view the films or the lab report?)

* The attending's altering of the chart, not the error itself, almost led to a lawsuit. Patients and families usually sue for three reasons: to get information, because they are angry, or to ensure that the error won't re-occur. (2)

Develop a process where it is known what tests have been ordered, whether the results are seen by the referring physician and whether the patient has been notified. Such a system would prevent the Bi-Rads 4 or 5 mammogram that is not followed up for months.

James Reason, in his book, Human Error (3) lists three organizational responses to error:

1. Denial

Suppression -- punish the reporter, expunge the report

Encapsulation -- deny the validity of the report

Example: One group of night shift ICU nurses destroyed rhythm strips because they got yelled at when they called a cardiologist. The solution was to destroy the evidence.

2. Repair

Public relations -- the reports are public knowledge, but the significance is downplayed or denied

Example: A hospital CEO dealt with a major newspaper's article on its errors by stating that the hospital had "good people" and was "safe." He gave no data.

Local repair -- the problem is admitted and fixed without looking at a larger issue

Example: A nursing home dealt with a Class IV decubitus by assigning a wound care nurse to that patient each shift. Nothing, however, was done to fix the underlying system.

3. Reform

Dissemination -- global action is taken on the problem

Reorganization -- action leads to significant change to the underlying system

Error prevention

Changing the attitude toward error is one way to tackle the problem. In one of the few surveys about physician attitude: (4)

* 60 percent of the respondents said they could function well when fatigued.

* 67 percent felt that their personal problems would not affect the quality of their work.

* 25 percent were not encouraged to report safety concerns in their institutions.

* 33 percent of ICU physicians didn't acknowledge they made errors.

In the airline industry, changing pilot attitudes led to changing pilot behavior. Here are four attitudes that are highly correlated with pilot--and likely physician--performance:

1. Attitude toward error

3. Attitude toward teamwork

4. Attitude toward hierarchy in job function

5. Attitude toward stress

Under stress, thought processes and breadth of attention narrow, lessening the scope of possible solutions. Therefore, aviation relies on crew resource management to utilize the collective knowledge of the crew in order to deal with crises. Health care might consider adapting the idea.

What can we learn? If we accept our fallibility and learn from errors, we could design better systems. We should change the hierarchy to "I'm responsible, but I have people who can help me if matters get really difficult or unusual."

What we can do today to reduce errors

1. Deal with fatigue -- Good sleep habits are important, including uninterrupted sleep and 15-30 minute naps, especially in the afternoon or after midnight.

2. Control interruptions -- Designate specific times or people to handle interruptions whenever possible.

4. Control information overload and read back information. -- Use written memos, simple words, few numbers, and ask for any verbal instruction to be read back, including signing out cases.

5. Occasionally get help, a fresh look -- Personal problems, stress, pre-occupation and hurry happen to all of us and can increases the likelihood of error. Fresh looks by others are useful in these instances.

6. Periodically ask how your judgment or assumptions might be wrong.

7. When evaluating an error, look for a system fault.

8. Change the QA process to one that looks for ways to learn, not to judge or assign responsibility.

9. Count what is important to count -- If you don't like to count things, get somebody to help and tell them what to count.

10. Prevent errors before they happen -- If you create a new system, define the tasks, what incorrect actions people can do, what might result, and how these actions could be prevented.

11. Develop a confidential, voluntary, protected, good faith reporting system for errors and near errors, regardless of outcome -- We need a safe way to learn from one another.

We know much about the causes and consequences of error in other industries. It is time to apply that information to medical care. We will all be better for it.

References:

1. Smith MS. "We don't deserve tort reform." Medical Economics. October 11, 2003.

2. Berman, S. "Reporting outcomes and other issues in patient safety: An Interview with Albert Wu." Joint Commission Journal on Quality Improvement, 28(4), 2002.

3. Reason, J. Human Error, Cambridge University Press, Cambridge, 1990.

4. Mather H and Elkeles R. "Attitudes of consultant physicians to the Calman proposals: a questionnaire survey." BMJ, Oct 1995, (www.bmj.com)

5. Focus on Patient Safety, National Patient Safety Foundation, Vol. 4(1), 2001.

6. Merry A and Smith AM. Errors, Medicine, and the Law, Cambridge University Press, Cambridge, 2001.

RELATED ARTICLE: TYPES OF ERRORS (3,5)

1. Procedural -- skill-based errors, such as mistakes, slips and lapses require appropriately designed systems and back-ups to ensure errors do not propagate.

2. Proficiency -- knowledge-based errors are due to lack of skill or knowledge. More common in novices, these lessen with education.

3. Communication errors -- these occur between physicians or between physicians and other medical workers.

4. Decision errors -- rule-based errors occur when a chosen course of action--an inappropriately applied rule--unnecessarily increases risk.

5. Intentional non-compliance errors -- violation of procedures or where countermeasures are not employed. These require discipline or sanctions.

[ILLUSTRATION OMITTED]

RELATED ARTICLE: When to blame

We should distinguish between errors and issues that are morally wrong. Here are five levels: (6)

1. Poor outcome but no rules broken and reasonable actions taken. Example: Anaphylaxis when there was no known prior drug allergy. This is neither a fault nor an error.

2. Unintentionally falls short of a textbook or theoretical norm but expected on a statistical basis.

Example: A radiologist misses a breast cancer, which is later seen. Is that good? No. Can it happen to a competent physician occasionally? Certainly. There may be error here, but there is no moral issue. The real question is how often should we expect a person to miss this type of tumor (or give the wrong drug)? Unfortunately, courts generally aren't interested in the statistical question.

3. Falls short of how things are usually done. The distinction between this level and the prior one is not always clear and depends upon how the data are gathered, the statistical tests applied, and the analysis of those tests. It is not a moral issue unless the person has had a long history of clearly falling short in this area and no changes have been made.

4. Reckless: A person knowingly takes risks that should not be taken, although the reason for doing so is ostensibly for the good of a patient.

5. Deliberate or purposeful.

[ILLUSTRATION OMITTED]

By Michael S. Smith, MD, MS

Michael S. Smith, MD, MS, a statistician, wants to help people in the medical community use statistics to make better, faster and easier decisions. He is self-employed and may be reached at 520-410-7917 or [e-mail].