CAP Today is the trade news journal of the College of American Pathologists. On a regular basis, CAP Today includes Q & A and general education sections for CAP physician members. The following is a list of questions for which the answers reference published studies performed by Dr. Novis and his coworkers. December 2005 Q. Is there a benchmark or community standard for the percentage of stat tests relative to total workload? Our overseas military hospital would probably be most comparable to a small community hospital.
A. Stats vary according to the institutional services and, as such, are a barometer of those services, rather than a target to be achieved. I am unaware of any published general benchmarks, although specific articles appear occasionally regarding turnaround time, which may have data regarding stats embedded in them, such as for troponin measurements (Novis DA, Jones BA, Dale JC, et al. Arch Pathol Lab Med. 2004;128: 158-164).
I believe staff should review the CAP Laboratory Accreditation Program standards regarding TAT and ensure that the needs of the medical staff are being met. If there is a perception that too many stats are being ordered (although that is not implied in the question), the medical director should perhaps review lab TAT in general or with respect to particular tests to ascertain if process improvement is needed. Alternatively, if there are individual abusers among the medical staff who regularly order stats inappropriately, then it is the medical director’s responsibility to attempt to change these ordering patterns and, thereby, assure appropriate laboratory use and resource consumption.
Q. Our laboratory maintains cytology/histology correlation in a separate file and keeps track of any discrepancies noted. We send out a letter to clinicians for all abnormal cytology cases without correlation asking for documentation of followup. We have several questions about these correlations: Should a comment on correlation be included in our pathology reports? When should requests for followup be made, and what should be done with this information?
A. CLIA 88 mandates that there is laboratory comparison of clinical information, when available, with cytology reports and comparison of all gynecologic cytology reports with a diagnosis of high-grade squamous intraepithelial lesion (HSIL), adenocarcinoma, or other malignant neoplasms with the histopathology report, if available in the laboratory (either on-site or in storage), and determination of the causes of any discrepancies.1 These requirements are reflected in cytopathology checklist questions CYP.01900, CYP.07543, CYP.07556, and CYP.07569.2
The method for documenting the cytohistologic correlation results is not specified and is left to each individual laboratorys discretion. Communication of cytology and biopsy correlation results to clinicians is key and provides critical information for optimal patient management.6 Cytohistologic correlation for individual patients can be documented in biopsy reports, via phone calls, or in letters, and, in more general terms, correlation statistics can be discussed in interdepartmental committees or conferences.
Evaluation of cytohistologic correlations is also an important part of a laboratorys quality improvement program.10 The definition of what constitutes diagnostic discrepancy should be established, and it should be recognized that perfect correlation is not realistic. The 1996 CAP Q-Probes study5 of 22,429 paired cervicovaginal cytology and biopsy specimens reported a discrepancy rate of 16.5 percent with a Pap sensitivity of 89 percent and specificity of 65 percent. The majority of discrepancies in this study were due to sampling differences rather than screening or interpretive errors. Because both the Pap test and colposcopic biopsy are subject to sampling errors, reasons for discrepancies should be pursued when the biopsy is negative, as it may not always represent the gold standard. Negative cytology cases should also be reviewed, if available, when the biopsy is positive. Peer or multidisciplinary review or both of noncorrelating specimens may be helpful in achieving consensus. Regular summary and evaluation of results can identify trends and improvements.
There is no requirement for correlating biopsies with lesser abnormalities, for correlating subsequent cytology with previous biopsies, or for correlating concurrent biopsies. However, these cytohistologic correlations are recorded in many laboratories and may be a useful part of the quality improvement program. Laboratories are required to document the number of cases that do have histologic correlation in the annual laboratory cytology statistical report.
In a 1997 article by Andrew Renshaw, MD,7 the optimal time for correlation of cytology and subsequent biopsies was found to be between 60 to 100 days, during which time the correlation of the Pap test and the subsequent biopsies was the highest. Biopsies performed over 100 days were less likely to correlate with the initial Pap. The difference may be explained by regression of lesions over time.
When followup material is not available in the laboratory, documentation of followup correspondence or reports, telephone calls, or requests for information, whether as separate letters or in the histology report, must be maintained and kept for two years. In cases without biopsy followup, other studies such as human papillomavirus testing, repeat Pap tests, and colposcopic examination findings may provide useful information, especially in cases of HSIL, glandular abnormalities, and carcinoma.
5. Jones BA, Novis DA. Cervical biopsy-cytology correlation. A College of American Pathologists Q-Probes study of 22439 correlations in 348 laboratories. Arch Pathol Lab Med. 1996;120:523531.
Is there a right time for cyto/histo correlation in gyn cytology?
Cytologic-histologic correlation is an important component of any quality improvement program in cytology. A documented effort must be made to obtain and review follow-up histologic reports or material that is available within the laboratory when high-grade squamous intraepithelial lesion or malignant findings are identified in gynecologic cytology.1 There is no specific requirement to obtain correlation for any gynecologic cytology specimen in the absence of HSIL, and there is no specific requirement that histologic findings be correlated with cytologic findings, though many laboratories do make these correlations and the results certainly can be a component of a quality improvement program.2-5
The time period over which these correlations should be made is not specified. Data do suggest, however, that the optimal period for examination may be 60 to 100 days. In a study involving 419 low-grade squamous intraepithelial lesion and 277 HSIL smears, Renshaw, et al.6 correlated the rate of subsequent biopsy and the rate of correlation with that biopsy over a period of one year. In this study, 811 biopsies were performed. While biopsies that correlated with the initial cytologic finding could be identified as late as one year after the initial cytology, the highest rate of confirmation was obtained in biopsies performed within 60 days, and fully 78 percent of all correlating biopsies were obtained within the first 100 days. The chance of finding a correlating smear decreased after that time. In other words, biopsies performed more than 100 days after the initial biopsy were less likely to correlate with the initial cytologic finding.
One explanation for the increased number of discrepancies was regression of the lesions. After 100 days, there is a greater likelihood of regression, which leads to an increase in the number of perceived false-positive cytology results when, in fact, a number of them are actually true positives. Limiting correlations to 100 days after the cytologic specimen was obtained is a reasonable way to limit the impact of false-positive correlations on the quality improvement program and the cytologic staff, while at the same time obtaining the majority of all biopsies for which correlation is available.
More controversial is whether cytologic specimens should be taken at the same time as the biopsy and correlated with it. Some literature suggests that cytologic specimens taken at that time have a higher likelihood of being false-negativesthat is, the cytologic specimen is more likely to not sample the lesion found in the biopsy.7 In the study by Renshaw, et al.,6 this was not found to be the case, and indeed cytologic specimens obtained at the time of biopsy were more likely to correlate with the results of biopsy than cytologic specimens taken at any subsequent time.
No requirement specifies that concurrent Pap tests need to be correlated with the biopsy since these cytology specimens were not the reason for obtaining the biopsy. Technically concurrent biopsies are not a followup to the cytology. In the interests of patient care, however, HSIL or malignancy identified on the cytology specimen with a concurrent negative or low-grade biopsy result should be reconciled. Furthermore, the subsequent histologic specimens must be correlated. It appears that the optimal biopsies to correlate are those obtained within 60 to 100 days after the Pap test.
Jones BA, Novis DA. Cervical Biopsy-Cytology Correlation. Arch Pathol Lab Med. 1996;120:523-531.
Fresh frozen plasma and platelet utilization
The authors of this study reported normative rates of expiration and wastage for units of fresh frozen plasma and platelets. Participants in the CAP Q-Probes laboratory quality improvement program collected data retrospectively on the number of units of FFP and PLTs that expired or were wasted due to mishandling. The participants also completed questionnaires describing their hospitals’ and blood banks’ laboratory and transfusion practices. The studies covered 1,639 public and private institutions and included data submitted on 8,981,796 units of FFP and PLTs. The aggregate combined FFP and PLT expiration rates ranged from 5.8 to 6.4 percent, and aggregate combined FFP and PLT wastage rates ranged from 2.0 to 2.5 percent. Among the top-performing participants (at the 90th percentile and above), FFP and PLT expiration rates were 0.6 percent or lower and FFP and PLT wastage rates were 0.5 percent or lower. Among the worst-performing participants (at the 10th percentile and below), expiration rates were 13.8 percent or higher and wastage rates were 6.8 percent or higher. The authors were unable to associate selected hospital characteristics or blood bank practices with lower rates of FFP and PLT utilization. They concluded that it is possible for hospital blood bank personnel to achieve FFP and PLT expiration and wastage rates of less than one percent.
Novis DA, Renner S, Friedberg RC, et al. Quality indicators of fresh frozen plasma and platelet utilization. Arch Pathol Lab Med. 2002;126:527-532.
Reprints: Dr. David A. Novis, For reprints, contact Dr. Novis at davidnovis.com.
Q. New requirement for nongyn TAT
A. The 2002 CAP Laboratory Accreditation Program checklist contains a new question related to turnaround time of nongynecologic cytology, or NGC, cases:
0YP.06532 Phase I: Are 90 percent of reports on routine nongynecologic cytology cases completed within two working days of receipt by the laboratory performing the evaluation?
This question was added to the checklist to underscore the importance of turnaround time as a measure of laboratory service quality. In a 2000 Q-Probe authored by Bruce A. Jones, MD, and David A. Novis, MD (QP08), the factors influencing TAT for 16,925 NGC specimens from 180 laboratories were analyzed. The authors found that 50 percent of participating laboratories had a mean TAT of 2.1 days or less from specimen collection to final report sign-off. The factors that delayed TAT included the use of reference laboratories for screening, lack of timely transcription, difficulty obtaining adequate specimen information from the submitting physician, and pulling old slides/tissue blocks for review or performing special stains, or both.
The CAP believes that a goal of two working days TAT for routine NGC specimens is reasonable. Documentation can consist of continuous monitoring of data or periodic auditing of reports. Longer times may be allowed for specimens requiring special processing or staining (for example, immunohistochemistry), provided these special classes of specimens are documented so that the inspector can evaluate their appropriateness.
For laboratories that are finding it difficult to meet the CAP TAT guidelines, the 2000 Q-Probe study makes recommendations for improving overall TAT. They are as follows: minimize the use of reference laboratories; educate the submitting physician’s office staff or change requisitions to expedite the gathering of important information, or both; reevaluate general laboratory workflow and transcription services; and continuously monitor TAT.
Nongynecologic cytology plays an important role in diagnosing and managing patients, many of whom may be acutely ill. The new CAP guideline emphasizes the importance of NGC turnaround time for patient care and clinical decision-making in today’s competitive, customer-service-oriented health care systems. Of course, the quality of diagno-sis should never be compromised for the sake of TAT.
Q. I have a question about correlation between Pap testsboth regular and ThinPrepand biopsy. What percentage is the benchmark? Some physicians are not satisfied with our service, and they seem to expect 100 percent correlation.
A. The percentage of cytology-biopsy discrepancies depends on the definition of discrepancy and the methods used to track discrepancies. One discrepancy definition offered in the CAP Quality Improvement Manual in Anatomic Pathology is a difference in interpretation that would have an impact on patient management decisions.1 Another definition is a two-step interpretive difference, for example low-grade squamous intraepithelial lesion on biopsy versus squamous cancer on the Pap. Excluding certain specimen types, for example endocervical curettings, will also have an impact on the discrepancy rates. Finally, the time interval and number of specimens considered per patient (single versus multiple cytology-histology combinations) will also affect the calculation.
A discordant Pap-biopsy combination, as defined by Joste et al, is one “in which one of the specimens is reported as a significant squamous or glandular lesion and the other specimen is reported as within normal limits.”2 This definition excluded atypical squamous cells of undetermined significance and biopsies lacking the transformation zone. In their 14-month study of 56,497 cervical smears, 2.8 percent (1,582) were followed by cervical biopsy. Of 1,582 paired samples, 175 cases (11 percent) were identified as discrepant. (This group represents 0.3 percent of all smears reviewed.) In a vast majority (93.2 percent) both cytologic and histologic diagnoses were confirmed and the discrepancies were classified as sampling errors. Only 3.4 percent of cases were found to have correctable (interpretive or screening) errors. Tritz et al also found an 11 percent discrepancy rate, with the majority representing sampling issues, although the definition of discrepancy involved a two-step difference in interpretation.3
Jones and Novis reported results of 12 months’ followup of 16,132 cervical smears from 306 laboratories as part of a CAP Q-Probes evaluation.4 They found that 18 percent of patients with low-grade squamous intraepithelial lesion on cytology had high-grade squamous intraepithelial lesion, or HSIL, on followup biopsy. Only 67 percent of patients with LSIL on cytology had LSIL on biopsy, and 86.5 percent had any abnormal biopsy. Of those patients with HSIL on smear, 15.5 percent had LSIL on corresponding biopsy, 75.5 percent had HSIL on biopsy, and 93.5 percent had an abnormal biopsy.4 Similar to the American experience, the United Kingdom’s screening program ranges between 65 and 85 percent concordance for biopsy-proven HSIL after HSIL on cervical smear.5
Brown et al evaluated 48 discrepant cases of HSIL on cervical smears with corresponding biopsies revealing LSIL.6 Biopsy specimens were tested and typed for HPV with molecular techniques. Thirty-seven cases were positive for HPV DNA: two for low-risk HPV types, 17 for high-risk types, and 18 for types of unknown oncogenicity. The prevalence of high-risk HPV was significantly higher in LSIL biopsies with a history of HSIL smears.6
Some cytology-histology discrepancy data have also been reported using liquid-based cytology. For example, Diaz-Rosario and Kabawat reported that 20.9 percent of HSIL ThinPreps and 26.8 percent of LSIL ThinPreps were followed by negative biopsies.7
It is unrealistic to expect 100 percent correlation between cervical cytology and cervical biopsies, and an open discussion with concerned clinicians is recommended. Cervical cytology is appropriately used as a screening test, which means that some specificity will be sacrificed for increased sensitivity, while the colposcopically guided cervical biopsy is recommended as a confirmatory test. Both tests are subject to sampling error. Although the cervical biopsy is often considered the gold standard, not all lesions will be fully characterized on an initial colposcopy, and a lesion that is small or deep in the glands may not be sampled. Some lesions will regress in the interval between the Pap test and the colposcopy. In some cases, the cervical smear may better represent the pathology of the cervix than the biopsy.2-6 Appropriate treatment and followup should then be dictated by a combination of clinical, cytology, and biopsy data. In addition, the pathologist’s advice or report comments may be extremely helpful.
4. Jones BA, Novis DA. Follow-up of abnormal gynecologic cytology. A College of American Pathologists Q-Probes study of 16,132 cases from 306 laboratories. Arch Pathol Lab Med. 2000;124:665-671.
Sidestepping common deficiencies
A top deficiency from the anatomic pathology checklist comes from this recently revised question, 08:1182, on frozen section turnaround time: “Are at least 90 percent of frozen section interpretations rendered within 20 minutes of specimen arrival in the frozen section area?”
The new guideline is based on a Q-Probe study of frozen section turnaround time published in the Archives of Pathology & Laboratory Medicine (Novis DA, et al. 1997;121: 559-567). It requires specimens to be prepared, analyzed, interpreted, and reported within 20 minutes. Previously, frozen section slides had to be ready for a pathologist to analyze within 15 minutes.
“A lot of labs just didn’t realize that it changed or they’re not tracking their turnaround time, so they can’t say whether they’re hitting that [target] or not,” Dr. Ruhlen says.
Complicated cases that require multiple frozen sections, however, aren’t expected to meet this new standard. One example is a skin lesion with multiple margins that requires several frozen specimens for a complete interpretation. “Clearly it would be ridiculous to say you have to do them all in 20 minutes when that’s often just impossible,” Dr. Ruhlen says.
CAP TODAY: http://www.cap.org/apps/cap.portal?_nfpb=true&cntvwrPtlt_actionOverride=%2Fportlets%2FcontentViewer%2Fshow&_windowLabel=cntvwrPtlt&cntvwrPtlt%7BactionForm.contentReference%7D=cap_today%2Fcap_today_index.html&_state=maximized&_pageLabel=cntvwr