Diagnostic ‘errors’ in anatomical pathology: relevance to Australian laboratories

Study of amended reports to evaluate and improve surgical pathology processes

BACKGROUND

: Amended surgical pathology reports record defects in the process of transforming tissue specimens into diagnostic information.

OBJECTIVE

: Systematic study of amended reports tests 2 hypotheses: (a) that tracking amendment frequencies and the distribution of amendment types reveals relevant aspects of quality in surgical pathology's daily transformation of specimens into diagnoses and (b) that such tracking measures the effect, or lack of effect, of efforts to improve surgical pathology processes.

MATERIALS AND METHODS

: We applied a binary definition of altered reports as either amendments or addenda and a taxonomy of defects that caused amendments as misidentifications, specimen defects, misinterpretations, and report defects. During the introduction of a LEAN process improvement approach-the Henry Ford Productions System-we followed trends in amendment rates and defect fractions to (a) evaluate specific interventions, (b) sort case-by-case root causes of misidentifications, specimen defects, and misinterpretations, and (c) audit the ongoing accuracy of the classification of changed reports. LEAN is the management and production system of the Toyota Motor Corporation that promotes continuous improvement; it considers wasted resources expended for purposes other than creating value for end customers and targets such expenditures for elimination.

RESULTS

: Introduction of real-time editing of amendments saw annual amendment rates increase from 4.8/1000 to 10.1/1000 and then decrease in an incremental manner to 5.6/1000 as Henry Ford Productions System-specific interventions were introduced. Before introduction of HFPS interventions, about a fifth of the amendments were due to misidentifications, a 10th were due to specimen defects, a quarter due to misinterpretation, and almost half were due to report defects. During the period of the initial application of HFPS, the fraction of amendments due to misidentifications decreased as those due to report defects increased, in a statistically linked manner. As HFPS interventions took hold, misidentifications fell from 16% to 9%, specimen defect rates remained variable, ranging between 2% and 11%, and misinterpretations fell from 18% to 3%. Reciprocally, report defects rose from 64% to 83% of all amendment-causing defects. A case-by-case study of misidentifications, specimen defects, and misinterpretations found that (a) intervention at the specimen collection level had disappointingly little effect on patient misidentifications; (b) standardization of specimen accession and gross examination reduced only specimen defects surrounding ancillary testing; but (c) a double review of breast and prostate cases was associated with drastically reduced misinterpretation defects. Finally, audit of both amendments and addenda demonstrated that 10% of the so-called addenda actually qualified as amendments.

DISCUSSION

: Monitored by the consistent taxonomy, rates of amended reports first rose, then fell. Examining specific defect categories provided information for evaluating specific LEAN interventions. Tracking the downward trend of amendment rates seemed to document the overall success of surgical pathology quality improvement efforts. Process improvements modestly decreased fractions of misidentifications and markedly decreased misinterpretation fractions. Classification integrity requires real time, independent editing of both amendments (changed reports) and addenda (addition to reports).

Detection and classification of diagnostic discrepancies (errors) in surgical pathology

Detecting and classifying error in a surgical pathology (SP) practice is an important part of a comprehensive quality assurance program. There are a number of mechanisms to detect error, including secondary review, examination of amended reports, correlation studies (cytology-histology and frozen-final diagnosis correlation). These different detection methods are reviewed in this paper. Additionally, the most common methods for error classification are also reviewed, along with the benefits and limitations of each. Although there is presently no gold standard for detecting or classifying errors in SP, based on this review of the literature, it is clearly good practice to consistently apply a standard method. Most importantly, these data should be incorporated into quality assurance and quality improvement activities, such that departments strive to reduce errors, and to help improve overall quality in SP.

Molecular confirmation of pathological specimen integrity in Australasia

AIM

Investigations into 14 suspected pathology sample identification errors and mix-ups were performed, as a service for several public hospital and private laboratories, from 2005 to 2007.

METHODS

Analyses were performed with the forensic ABI Identifiler kit of 16 microsatellites (15 autosomal and amelogenin) on DNA from paraffin-embedded tissues or blood specimens and compared to independently verified (single or multiple) patient samples.

RESULTS

Of 23 unique patient specimens referred for sample integrity confirmation because of pathologist, clinician or patient concern, six (26.1%) were demonstrated to be discordant, indicating that specimen identification errors or mix-ups had occurred.

CONCLUSIONS

Due to their great sensitivity and high discrimination power, forensic identity multiplex systems using either microsatellites or single nucleotide polymorphisms (SNPs) can resolve concerns about pathology specimen identity and integrity.

Histologic, surgical, and imaging correlations of adnexal masses

OBJECTIVE

The purpose of this study was to quantify, categorize, and illustrate discrepancies between preoperative radiologic, surgical, and pathologic diagnoses and to assess the potential impact of discrepancies on clinical care.

METHODS

Adnexal masses reported by pathology during a 16-month period were included if prior imaging at our institution had been performed. Up to 3 sonographic, computed tomographic, and magnetic resonance imaging examinations were reviewed by a gynecologic sonographer and compared with the reported pathologic findings. Cases in which ambiguities were not resolved by consulting the surgical notes were reviewed by a gynecologic pathologist, who confirmed or modified the diagnosis and assigned a score to the pathology quality assurance issue: 0, no pathology quality assurance error; 1, differences in terminology; 2, discrepancy of diagnostic interpretation, benign findings; or 3, discrepancy of diagnostic interpretation, malignant findings.

RESULTS

Of 418 total masses, there was a discrepancy between imaging and pathology in 73 (17%) masses and 68 (21%) pathology reports. Twenty-five (6%) had pathology discrepancies resolved by correlation with the surgical notes alone (eg, torsion seen during surgery but not evident on pathologic examination). Histologic review was performed for 48 (11%) of 418 masses, with pathology errors identified in 34 (71%) of 48. Quality assurance scores were 0 (n = 14), 1 (n = 14), and 2 (n = 20), with no cases receiving a score of 3. Examples of pathology errors included gross (complex versus simple) and microscopic (neoplastic versus functional versus mesothelial) characterization of cysts, sizes of lesions not being described, characterization of fibrous lesions (cystadenofibroma versus cystadenoma), and lack of correlation with imaging (lesions not described).

CONCLUSIONS

This study illustrates the importance of imaging, surgical, and histologic correlation in assessing the diagnostic accuracy of sonography of adnexal masses.