The Quest for Certainty: on Medical Errors
A 13 year old girl had a physical examination by the school sports doctor in order to be cleared to join the soccer team. The physician was alarmed to find a "distended uterus" and immediately referred her for a $300 ultrasound examination which proved to be completely normal. When queried later, the girl admitted to having a full bladder at the time of examination.
Comment: Not all patients require medical imaging, not even those with a mass. The most common lower abdominal mass is a full bladder.
When it comes to making decisions, the problem of uncertainty is inescapable. This is as true for doctors choosing a test, as for Governments deciding to go to war or the legal question of guilt. In the explicit rules of criminal justice in this country and Great Britain: A man is innocent until proven guilty. Or, "When in doubt, acquit." This has been expressed in the maxim, "Better a thousand guilty men go free, than one innocent man is convicted." Of interest is the legal system in France and many European countries, where the opposite rule holds: A person is guilty until proven innocent (!)
In most cases of medical binary decision-making, no matter what the evidence, there are two kinds of errors: Concluding a patent is sick when he is well, called an "error of the first kind," a Type I error, or concluding a patient is well when he is sick, "an error of the second kind, a Type II error. As physicians, we learn early in our training, that it is far worse to dismiss a sick patient than to retain a well one. (Just ask the malpractice litigators). Thus, "When in doubt, continue to suspect illness." Far better a Type I error, a false presumption of illness than a Type II error, a false presumption of health. This tradition finds expression in our delusional and sacred obsession with certainty, and the maxim: "Better to make a thousand patients sick than miss one cancer," But is this always the better of two possible decision worlds?
Almost 45 years ago, Garland summarized his findings in a study of over 14,700 chest X-rays for signs of tuberculosis. He found over 1200 (false) positive readings for tuberculosis which turned out to be negative (Type I error) and only 24 (false) negative readings which turned out to be clinically active (Type II error). Thus, there were 50 times as many mistaken diagnoses compared to missed diagnoses.
The Ruling out Problem and Clinical Judgment
A 62 year old diabetic man was seen ten days following coronary bypass surgery after he developed sudden shortness of breath associated with fever. A chest X-ray was non-diagnostic. Lung scans showed a low probability of clot to the lung (pulmonary embolism). The following day a pulmonary specialist felt that even though the possibility of embolism was "less than 10%," he could not rule it out" unless direct visualization of the lung arteries with contrast (pulmonary angiography) was done. During this invasive procedure the patient went into shock, cardiac arrest, and almost died. He was put on a ventilator. That evening, bacterial pneumonia was diagnosed.
Comment: Performance of a hazardous diagnostic procedure in a patient with low likelihood of disease led to a clinical disaster (Type I error). In this patient, however, the consultant, knowing the risk of pulmonary angiography, felt the risk of missing pulmonary embolism was even greater, and elected to perform the study (In retrospect, he was wrong. However, many physicians would have supported his decision to rule out a serious condition.)
The Cascade Effect
The problem of abnormal or incidental findings leading to an uncontrollable series of unforeseen events-fear of a Type II error-sometimes culminating in catastrophe, has been described as the cascade effect, and is illustrated in the following true story:
A 35 year old X-ray technician underwent an ultrasound examination of the abdomen for vague upper abdominal symptoms. The study was negative except for the incidental discovery of a small liver lesion. She was told it might be serious, and underwent a CT scan which was nondiagnostic. Finally a biopsy was suggested. She delayed having further studies for several more weeks, during which time she developed a serious depression, and consulted a psychiatrist in the belief she might have cancer. Ultimately she had an invasive arterial study of the liver (hepatic angiography) which showed an angioma, a common benign liver tumor seen in up to 2% of the population. Review of her initial ultrasound examination showed a typical pattern of a benign hemangioma.
Medical Cost: $8,500, including psychiatrist
Emotional Cost: Impossible to estimate
This patient had an incidentaloma so named for a tumor or mass, usually cystic, which is incidentally found on imaging studies, and which is almost never of clinical significance. These tumors are found frequently in various organs of normal people, especially in the kidney (50% of people over 50) and the adrenal or thyroid gland (10% of the population). Yet their discovery sometimes results in a series of risky diagnostic studies which may culminate in emotional if not financial and therapeutic disaster. Cascade fiascoes and catastrophes are frequently catalyzed by anxiety on the part of the physician or patient, but they are becoming increasingly more common in clinical practice because of the introduction of new and ever more spectacular technologies.
Another form of the cascade effect, gives an insight into the dark side of modern technology. Dr. Howard Spiro, writing about narrow-gauge surgery of the gall bladder ("lap choly") or laparoscopic cholecystectomy, reminds us that not all patients with gall stones will be benefited by the procedure. He discusses a tendency on the part of internists and surgeons to ascribe a variety of abdominal complaints to the incidental findings of gall stones on abdominal ultrasound. In the past, the patient could be managed expectantly, because elective removal of the gall bladder was not to be taken lightly. The imagined ease of lap choly" compared to traditional open gall bladder surgery is leading to more and more unnecessary operations on the gall bladder with unforeseen consequences. Common bile duct and other operative injures, uncommon in the past, now occur with 2-4 times the frequency.
Postscript: The German poet Otto Hartleben was feeling quite ill and consulted a physician who, after a thorough examination, prescribed complete abstention from smoking and drinking. Hartleben picked up his hat and coat and started for the door. The doctor called after him, "My advice, Herr Hartleben, will cost you three marks." "But I'm not taking it," retorted Hartleben,
Martin F. Sturman, MD, FACP
Copyright 2004, Mathemedics, Inc.
EasyDiagnosis is an automated online subscription service that analyses existing medical symptoms and predicts likely causes and conditions. Click here to find out more about this unique service and click here to subscribe.