PAP/NGC Programs Review
CAP Cytopathology Committee*
The Institute of Medicine raised the national awareness of medical errors and patient safety in 1999 by publishing "To Err Is Human: Building a Safer Health System," a landmark document that reported "stunningly high rates of medical errors—resulting in deaths, permanent disability, and unnecessary suffering" in U.S. hospitals. Since its publication, the IOM report has had the positive effects of raising concern for patient safety and increasing patient safety activities by researchers and clinical practitioners. Within the field of anatomic pathology, there has been an explosion of academic work examining sources and significance of diagnostic errors, methods for detecting and categorizing errors, and processes for reducing errors.
Of course, identifying, reporting, and proposing methods for reducing anatomic pathology errors are vital, important components of good patient care and quality assurance. However, because there is considerable disagreement about what constitutes an error, and because many "errors" are in fact not errors at all and actually represent the variability and variation of presentation of the underlying disease process, categorizing "errors" in an accurate and meaningful way is difficult. For example, if a patient undergoes a fine needle aspiration of a sclerotic lung nodule that yields only scant benign fibrous tissue, and the subsequent excisional biopsy demonstrates a sclerotic malignant tumor, some observers would categorize the negative FNA findings as an anatomic pathology error, because there is a cytologic-histologic discrepancy. On the other hand, many reasonable individuals would not consider this an error. There was no interpretive error by the cytopathologist: The absence of tumor cells in the FNA was a consequence of a sclerotic tumor that is not amenable to aspiration. In the face of an FNA result that did not adequately explain the suspicious clinical/radiologic findings, the patient underwent an appropriate excisional biopsy that yielded the correct diagnosis. An interested layperson might learn of the "error" rate of lung FNA and incorrectly conclude that some other procedure should be performed. Indeed the anatomic pathology error rate could be reduced to near zero by performing open biopsy on all patients with lung masses; however, the associated morbidity and mortality precludes this approach, of course. This global view of the care of the patient must be taken into consideration. The aggregate harm associated with different approaches of working up the patient must be considered.
Another example: If a patient has a Pap smear that shows a low-grade squamous intraepithelial lesion and the subsequent biopsy shows no lesion, has an anatomic pathology error occurred? Was the Pap smear interpretation a false-positive error by the cytopathologist that led to an unnecessary followup biopsy, causing psychological trauma to the patient and driving up health care costs? Or was the surgical biopsy interpretation a false-negative error by the surgical pathologist that delayed additional treatment, perhaps causing harm to the patient? Or are both the Pap smear and biopsy results correct, and the error lies with the gynecologist, who failed to biopsy the area of dysplasia identified by the Pap smear (sampling error)? Or perhaps no error occurred at all, and the absence of dysplastic cells reflects regression of the lesion in the time interval between the Pap smear and the biopsy, a well-documented part of the natural biology of low-grade lesions.
These examples illustrate the importance of explicitly distinguishing between discrepancies in interpretation between specimen types (sampling errors) and interpretive errors. If these examples do not provide a sufficient representation of the ambiguities associated with identifying anatomic pathology errors, one need look no further than the medicolegal arena, where expert witnesses for the plaintiff and defense can examine the same evidence and have opposite views of whether an error has occurred.
Another potentially unfortunate downside of articles that report error rates in anatomic pathology is the general public misunderstanding, and the mass media misrepresenting, the data. If medical experts cannot agree on what constitutes an error, or where the blame for the error lies, how can individuals without medical training critically evaluate the data fairly and accurately? Webster’s definitions of error include: "a usually ignorant or unintentional deviating from accuracy or truth" and "a product of mistake." Though articles in the anatomic pathology literature usually give an explicit operational definition of error, this definition is usually not reported in the mass media, and, therefore, the general public can be expected to use the lay definition of error when they read or hear these reports. Thus it can easily seem that anatomic pathology "errors" are due to pathologist incompetence instead of discrepancies due to sampling.
A recent article by Stephen S. Raab, MD, and colleagues provides highlights of some of the difficulties associated with: 1) identifying errors in anatomic pathology; 2) achieving consensus among pathologists about what constitutes a true error; 3) interpreting the significance of errors; and 4) reporting errors in a context that is not misunderstood by laypeople and members of the mass media. In an effort to characterize the frequency, cause, and impact of error in cancer diagnosis, Dr. Raab and colleagues reviewed tissue specimens, pathologists’ findings, and medical records from four institutions. They found that errors were relatively frequent and institution-dependent. Up to nine percent of gynecologic specimen pairs (Pap test and cervical biopsy) contained an error, and up to 12 percent of nongynecologic specimen pairs contained an error. Differences in institutional error rates were most likely related to variability in the methods of error reporting, though underlying quality improvement methods could have resulted in lower error frequencies at some institutions. The majority of errors were a result of suboptimal specimen collection, and the proportion of errors caused by pathologist misinterpretation reportedly ranged from five percent to 50 percent. When pathologists reviewed specimens from other institutions, they frequently disagreed with the assignment of error cause. Review of medical records showed that gynecologic errors resulted in harm in 45 percent of cases and non gyne co logic errors resulted in harm in 39 percent of cases. Harm most often consisted of delays in diagnosis and additional testing that drove up health care costs. These researchers say their study demonstrates not only a relatively high frequency of errors in anatomic pathology, but also a variability in clinical practice that impairs the ability to categorize and evaluate error causes. To that end, the authors conclude, "The standardization and uniform reporting of errors in cancer diagnosis is a first step in improving safety."
The findings of Dr. Raab and colleagues provide insights into many of the problems associated with identifying and reducing errors in anatomic pathology. Their study demonstrates that there can be considerable disagreement among experts about whether an error has even occurred, let alone what caused the error in question or whether any harm came to the patient. These investigators should be applauded for pointing out the difficulties and controversies associated with reporting errors, and for calling for standardization of error reporting so that error data can be interpreted meaningfully and accurately.
Ironically, the authors’ own findings may not be interpreted accurately by members of the lay media or informed patients who are interested in reading about anatomic pathology errors. Dr. Raab’s data clearly demonstrate that most errors are the result of inadequate or suboptimal specimen collection, and, in that sense, are not "anatomic pathology errors" at all; rather, they represent errors in sampling, which, in most instances, the pathologist does not control. Anyone who gives this article a superficial reading would conclude that errors are rampant in anatomic pathology and that many pathologists are incompetent. A more careful reader would conclude that anatomic pathology diagnoses are highly accurate, and that, when cytologic-histologic disagreement occurs, it is usually the result of inadequate sampling by the clinician or variability in disease biology. Thus, even this article, which sets forth the laudable goal of standardizing medical error terminology to improve communication and understanding, runs the risk of generating misunderstanding about how frequently errors occur and what the sources of those errors are.
Going forward, pathologists’ dual challenge will be to continue to improve
patient care and safety by identifying the causes of errors and standardizing
terminology, and to make sure these data are presented in a manner that
prevents misinterpretation by those outside the profession, and improper
use by the media or regulatory agencies.
1. Kohn LT, Corrigan JM, Donaldson MS, eds. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; 1999.
2. Raab SS, Grzybicki DM, Janosky JE, et al. Clinical impact and frequency
of anatomic pathology errors in cancer diagnoses. Cancer.
* Members of the CAP Cytopathology Committee are David C. Wilbur, MD,
chair; Joel S. Bentz, MD; Christine Noga Booth, MD; Karen M. Clary, MD;
Amy C. Clayton, MD; Camilla J. Cobb, MD; Terence J. Colgan, MD; Teresa M.
Darragh, MD; Barbara S. Ducatman, MD; Michael R. Henry, MD; Nicholas J.
Hruby, MD; Jonathan H. Hughes, MD, PhD; Gladwyn Leiman, MBBCh; Ann T. Moriarty,
MD; Marianne Unger Prey, MD; Mary R. Schwartz, MD; William D. Tench, MD;
Theresa M. Voytek, MD; Patricia G. Wasserman, MD; Nancy A. Young, MD; Maureen
F. Zakowski, MD; and Sarah E. Rollins, MD.