Current State of Cytologic-Histologic Correlation Implementation for North American and International Laboratories: Results of the College of American Pathologists Cytopathology Committee Laboratory Practices in Gynecologic Cytology Survey Open Access

A subset of the CAP Cytopathology Committee members and CAP technical staff, including a staff biostatistician, developed a survey and reviewed it for question clarity and validity. The final survey included 19 questions. The survey was delivered in the 2018-B Gynecologic Cytopathology PAP Education Program mailing to a total of 1545 laboratories. Data validity adjustments were applied to address skip sequence direction.

A 2-level institution location variable was created to define North American and international laboratories. North American laboratories included respondents from the United States, Canada, and Mexico. Institution location practice differences were evaluated with Pearson χ² tests and Fisher exact tests. Additional analyses were performed for the North American laboratories to determine practice differences regarding whether CHC is performed and how it is investigated and resolved. These analyses were performed with multivariate logistic regression models that were fit with 2 factors: institution type and institution size. Institution type was defined as a 5-level factor (hospital, clinic or regional/local independent laboratory, Veterans Administration/Department of Defense hospital [VA/DOD], university hospital/academic medical center, and national/corporate laboratory) and institution size was based on 2017 gynecologic test volume provided by respondents (≤1200 tests, 1201–5000 tests, 5001–25 000 tests, and >25 000 tests). A significance level of .05 was used for this analysis, and all analyses were performed with SAS 9.3 and 9.4 (SAS Institute, Cary, North Carolina).

The survey was delivered in the 2018-B Gynecologic Cytopathology PAP Education Program mailing to 1545 laboratories, and 701 returned the survey for a 45% response rate. There were 653 surveys that qualified for analysis; 48 surveys were excluded owing to duplication (11) and multiple missing responses (37) (Supplemental Table 1, see the supplemental digital content at https://meridian.allenpress.com/aplm in the January 2023 table of contents).

Most participants performed CHC on gynecologic cytology specimens with a worldwide rate of 87.0% (568 of 653), from which 87.6% (461 of 526) are from North America and 84.3% (107 of 127) are from international locations (Figure, A). The international laboratories were from 33 countries with more than one-third of the 127 laboratories located in Saudi Arabia (25 of 127; 19.7%) and the United Arab Emirates (19 of 127; 15.0%) (Supplemental Table 1). For the North American (United States, Canada, and Mexico) institutions performing CHC, we asked for the institution type (Figure, B). Of the respondents who performed CHC, most were from hospitals (54.3%; 244 of 449); the remaining participants were from clinic or regional/independent laboratories, VA/DOD, university hospital/academic medical centers, and national/corporate laboratories. Interrogation of gynecologic cytology volume and its association with CHC showed that institutions with the lowest volume (≤1200 tests) had the lowest CHC rate, 43.4% (36 of 83), compared to more than 95% for institutions with greater than 1200 specimens. Based on institution type, the rates at which CHC was performed varied from 82.9% to 95.7%, with the highest percentage in university hospital/academic medical centers (95.7%; 44 of 46) and VA/DOD (94.8%; 55 of 58) and the lower percentages in hospitals (86.2%; 244 of 283), clinic or regional/local independent laboratories (86.7%; 72 of 83), and national/corporate laboratories (82.9%; 34 of 41) (Table 1). There were also demographic characteristics significantly associated with how CHC discrepancies are investigated. VA/DOD laboratories had higher rates in using a second pathologist to review histology slides to find the reason for discordance. Only one of the small volume laboratories (2.8%; 1 of 36) investigated by dividing slides among several reviewers, although none of the institution types routinely investigated this way. There were no demographic factors significantly associated with who is responsible for the final CHC discordance resolution.

Most cases were identified by using a laboratory information system (LIS) to generate a list of patients for both cytology and histology specimens (79.9% [448 of 561]) (Table 2). In North America, most laboratories (83.8%; 383 of 457) used a LIS to generate patient lists for CHC. Significantly fewer international laboratories had access to an LIS (62.5%; 65 of 104) (P < .001). When investigating how cases were selected for CHC review, real-time concurrent CHC cases were mostly identified by the interpreting pathologist (60.2%; 240 of 399) in both North America and internationally (77.9%; 187 of 240 and 22.0%; 53 of 240, respectively). There was no difference in terms of using the selection criteria “diagnosis” or “specimen type” between the geographic locations. However, laboratories in North America used “body site” more frequently (22%; 71 of 322 versus 10.4%; 8 of 77 international) (P = .02).

More than 88.3% (498 of 564) of laboratories had access to CHC materials, which included cytology and histology final reports, diagnoses, and slides (Table 2). International laboratories had less access to CHC materials of all types (91.5%–94.3%; 420–433 of 459 for North American laboratories versus 73.3%–80.0%; 77–84 of 105 for international) (P < .001 for each comparison). The majority of North American laboratories had histologic diagnoses available for more than half of cytology abnormal interpretations and 32.2% (139 of 432) had histologic diagnoses for more than 90% of cytology abnormal results, significantly higher than international laboratory access (19.2%; 19 of 99) (P < .001).

CHC was performed mostly at the time the corresponding surgical pathology was reviewed for reporting (70.8%; 398 of 562) (Table 3). There were no statistically significant differences between North American and international laboratories regarding when CHC was performed. Multiple responses were allowed, and the most common CHC was real time/concurrently at the time the corresponding surgical pathology was reviewed for reporting: 71.1% (325 of 457) in North America and 69.5% (73 of 105) internationally. Several retrospective methods were listed by most North American laboratories such as when compiling statistics and when comparing final diagnosis from a report. There was no significant difference in time intervals used for selecting correlating cases. The most common response was within 6 months (37.5%; 208 of 554) with fewer choosing within 3 months (26.0%; 144 of 554), or 12 months (25.8%; 143 of 554). Investigations of discordances between histology and cytology specimens are usually performed at the time of histology interpretation/sign-out (67.7%; 382 of 564) or after compilation of cases to generate statistics (67.4%; 380 of 564). The most frequent time for the investigation of discordant specimens differed by the laboratory location (multiple responses were permitted). International laboratories predominantly investigated discordance at the time of histology interpretation/sign-out (75.7%; 81 of 107), whereas North American laboratories investigated after a compilation of cases to generate correlation statistics (70.5%; 322 of 457) (Table 3).

Almost half of laboratories investigated discordances with a single pathologist who reviews all discrepant histology and cytology slides (48.6%; 272 of 560) (Table 4). But how the discordant cases were investigated varied when the laboratory location was analyzed (multiple responses were permitted). Slightly more than half of the North American laboratories had a single pathologist review all discrepant histology and cytology slides to determine the reason for discordance (51.0%; 233 of 457 versus 37.9%; 39 of 103 international) (P = .02). Slightly more than half of international laboratories had a second pathologist review histology slides to determine the reason for discordance (50.5%; 52 of 103 versus 24.1%; 110 of 457 for North America) (P < .001). International laboratories were also more likely to have slides divided among several different reviewers to investigate discordance (18.4%; 19 of 103 versus 9.4%; 43 of 457 for North America) (P = .008). Both North American (40.5%; 185 of 457) and international (48.5%; 50 of 103) laboratories frequently had the original pathologist review their own cases.

The primary responsibility for a final CHC discordance resolution varied (multiple responses possible) but it most commonly fell on the pathologist responsible for the corresponding surgical pathology report in both North American and international laboratories (208 of 558; 37.3%). However, the distribution of responses between North American and international laboratories was statistically different with North American laboratories more frequently designating the cytopathology section director (143 of 452; 31.6% versus 17 of 106; 16% for international).

The primary reason for specimen discordance in both locations was sampling issues, overwhelmingly exceeding other causes (93.5%; 516 of 552) and affecting 95.6% (430 of 450) of discordant cases in North American laboratories and 84.3% (86 of 102) in international laboratories (Table 5). Laboratories were then asked to name additional reasons for discordance (multiple answers were allowed). The need for additional levels or processing and interpretation errors were common secondary reasons (52.5%; 242 of 461). International laboratories cited significantly more problems with process/preparation factors hindering interpretation, inappropriate/inadequate history, surgical specimens requiring reorientation, and inadequate or no sampling of the target lesion than North American laboratories.

To further understand how laboratories decided what paired cases were discordant, we asked which paired diagnoses were most common in discordant cases. There were discordant diagnoses that did not differ greatly between geographic locations; 80% or more agreed with the following: normal biopsy with HSIL Papanicolaou (Pap) test, low-grade squamous intraepithelial lesion (LSIL) biopsy with negative for intraepithelial lesion or malignancy (NILM) Pap test, and HSIL biopsy with NILM Pap test. Interestingly, statistically significant variations were found. North American laboratories were more likely to include discrepancies between a normal biopsy with HSIL Pap test (94.0%; 409 of 435) and HSIL biopsy with NILM Pap test (96.2%; 403 of 419). On the other hand, international laboratories were more likely to include discrepancies between LSIL biopsy with atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion (ASC-H) Pap test (21.7%; 20 of 92), LSIL biopsy with HSIL Pap test (58.7%; 54 of 92), and HSIL biopsy with LSIL Pap test (38.5%; 37 of 96).

We asked what metrics were chosen as monitors for discordant cases (Table 6, multiple responses possible). The more common monitors chosen were the total percentage of cases that correlated with subsequent biopsies (70.7%; 347 of 491), screening error rate by cytotechnologist (41.3%; 203 of 491), and interpretative error rate by cytotechnologist (40.3%; 198 of 491). There were no significant differences in selection between the locations for most monitors. North American laboratories were significantly more likely to monitor screening error rate by cytotechnologist (44.5%; 179 of 402) (P = .002) and interpretative error rate by cytotechnologist (45.5%; 183 of 402) (P < .001). Interestingly, international laboratories were significantly more likely to monitor the positive predictive value (PPV) of a positive gynecologic cytology (27.0%; 24 of 89) than North American laboratories (11.2%; 45 of 402) (P < .001).

Worldwide, data from the discordance metrics were predominantly summarized and shared in a quality improvement document (82.8%; 458 of 553) (Table 7). A major difference between the 2 locations is that North American laboratories also shared cytotechnologist results individually and as a summary for ongoing performance assessment or education more than half of the time (61.2% [275 of 449] and 52.1% [234 of 449]) (P < .001). CHC results were most frequently summarized monthly, and most commonly distributed to pathologists and laboratory quality committees. There was no statistical significance between the test volume and the frequency with which CHC results were summarized; however, it was noted that there was an increasing monthly calculation frequency with increased laboratory size (data not shown).

Remediation/retraining usually occurred when there was an error trend for pathologists (42.8%; 206 of 481) (Table 8). The 2 sites differed significantly for the other options. North American remediation/retraining occurred mostly for error trends for cytotechnologists (71.7%; 286 of 399) (P < .001). On the contrary, international laboratories did the bulk of remediation/retraining for every major discordance for cytotechnologists (58.5%; 48 of 82) (P = .003), followed closely by every major discordance for pathologists (46.3%; 38 of 82) (P < .001).

The majority of laboratories indicated they addressed discordances in the pathology report when current patient care or management would change (430 of 552; 77.9%) and when the histology specimen did not reflect a high-grade or positive cytology interpretation (303 of 552; 54.9%). Communication of discordances that change current patient care were more likely to be discussed directly with the patient's provider for North American (372 of 448; 83%) than international (72 of 103; 69.9%) laboratories (P = .002).

CHC was recommended by the CAP after CLIA 1988 as a standard practice for institutions performing gynecologic cytology and was incorporated into their LAP. Clarification on how best to perform gynecologic CHC was detailed in subsequent CAP-sponsored consensus guidelines,³  but universal penetration of these recommended processes has been difficult to capture. We designed this survey to investigate current CHC practices in North American and international laboratories. Of note, this survey is biased in favor of laboratories that are enrolled in CAP educational activities and may not reflect laboratory practices universally.

Review of microscopic slides during CHC allows for investigation of discordances, helps decrease interobserver variability, and provides invaluable learning opportunities.^4–7  Prior CAP surveys have shown that CHC practices are used in a very high proportion of institutions.³  This survey supports that more than 80% of CAP-participating laboratories worldwide practice CHC and includes non-US laboratories that are not federally mandated under CLIA to comply. However, review of the trends over time suggests that slightly fewer institutions now perform CHC. Our previous survey found that 480 of 512 respondents performed CHC for a rate of 93.8%.³  Our current survey identified only 461 of 526 laboratories performing CHC, a rate of 87.6%. This decrease may reflect slight differences in laboratories subscribing to CAP programs, but the factors underlying the decline are largely unknown at this time and largely beyond the scope of this survey. The time interval between the surveys is large (approximately 8 years), and the practice of cytology has changed during this period, complicating the assessment of this decline. Additional data are necessary to understand the reasons for this decrease in CHC, and ideally to reverse the trend.

Given the decline, we sought to determine if particular institution types had lower CHC rates and could account for the decrease in the total average. The survey identified certain North American laboratory types with lower rates: hospitals, clinics or regional/local independent laboratories, and national/corporate laboratories. These institutions made up more than three-fourths of the cohort performing CHC. On the other hand, about 95% of university hospitals/academic medical centers (44 of 46) and VA/DOD (55 of 58) perform CHC. Interrogation of volume load as a factor showed that the lowest CHC rate, 43.4% (36 of 83), was seen in institutions with the smallest annual volumes (≤1200 tests). This CHC rate is less than half the rate seen in institutions with higher volumes (>1200 tests). These findings seem to suggest that there are challenges in performing CHC when laboratories are either very large (national/corporate laboratories) or small (hospital/clinic/independent laboratories). Larger commercial laboratories, which may receive Pap tests from all over the country, may have less access to comparison biopsies and reports. Smaller laboratories may have insufficient resources (personnel, LISs) or too few cases to carry out elaborate CHC processes. The VA/DOD laboratories generally share standardized laboratory processes, one of which is prospective and retrospective case review, which may account for their high rates of CHC, and university laboratories often employ leaders in laboratory practice who understand the value of CHC.

CHC serves a significant quality assurance function, both for cytology and surgical pathology. Any drop in CHC performance rates suggests less robust quality assurance with possibly worsening clinical outcomes. The strikingly low rate of CHC among the smallest institutions suggests that these organizations may benefit from receiving assistance in establishing and carrying out a program of CHC.

Our survey found that international laboratories have slightly lower CHC rates, and some differences with North American laboratories, specifically in the ability to identify and access cases. Some international laboratories likely have multiple hurdles for CHC: decreased access to an LIS, CHC material, and histology specimens. One potential explanation may be institutional or governmental choices related to infrastructure and the financial allocation of resources. The United States, unlike many other countries, does not have universal health care. Therefore, US institutions may have increased autonomy for infrastructure and resource allocation, compared to countries where health care is overseen by the government. Despite these obstacles, however, international laboratories were able to maintain similar CHC rates when compared to North American laboratories. More importantly, a large number of all laboratories participating had access to an LIS and histology and cytology material.

For both North American and international laboratories, most case selection for CHCs occurred in real time, in other words, at biopsy sign-out. The rate in North American laboratories was 71.1% (325 of 457) and in international laboratories, 69.5% (73 of 105). Although multiple responses were allowed, most North American laboratories also performed retrospective CHC, including for statistical purposes. These numbers are comparable to, if not slightly better than, the previous rate of 67.3%.³  When retrospective CHC is performed, the time interval between Pap test sign-out and CHC is within 6 months after the abnormal Pap test finding for more than half the cases. One of the major purposes of CHC is to impact clinical care, and this is best accomplished in real time, allowing additional histologic sections and peer review to be easily performed without amending a report. Delays in correlation may result in unnecessary procedures (eg, with a false-positive HSIL Pap test result) or delay in diagnosis (eg, with a false-negative biopsy finding). Since longer time intervals are known to incur higher rates of discordance, it is important to note that more than one quarter of laboratories took up to 1 year to perform correlations following an abnormal Pap result.⁸  As long as CHC is performed within a reasonable time frame, correlation either at case sign-out or retrospectively is acceptable, and laboratories should tailor the processes to their specific needs and resource availability. Our question regarding when discordant investigations occurred revealed an interesting trend. North American laboratories frequently reported investigating cases in a batchlike manner after compilation of cases to generate correlation statistics and allow for identification of possible patterns of errors, while international laboratories did so less often. This may be related to availability of a supportive LIS or to LAP requirements to compile these statistics. Alternatively, it may be due to the fact that it is a federal mandate to do this in the United States, compared to the other countries where it is simply done for quality assurance.

The individual responsible for investigating CHC cases also differed significantly between North American and international laboratories. While multiple responses were often given, North American laboratories most frequently had a single pathologist review all discrepant histology and cytology slides to determine reasons for discordance (51.0%; 233 of 457), whereas international laboratories had a second pathologist review histology slides (50.5%; 52 of 103) and were more likely to have slides divided among several different reviewers to investigate discordance. The reasons for this are unclear but may be related to laboratory management framework, since in North America individual pathologists tend to be assigned to direct subsections of the laboratory, such as cytopathology, and may be held personally responsible for CHC review. Laboratory management by pathologists varies internationally, particularly in national health arenas, and duties may be shared more equitably among pathologists. Final discordance resolution was similar for both locations in that the pathologist responsible for the corresponding surgical pathology was commonly assigned to resolve any discordance. Additionally, North American laboratories were more likely to have others involved in resolution, including the cytopathology section director. This may be due, in part, to the paucity of cytotechnologists internationally and a difference in practice patterns. For example, in the United States, cytotechnologists are commonly colocated with pathologists as part of the cytopathology team, but in countries with centralized cervical cancer screening programs, cytology screeners/cytotechnologists may be remotely located. Overall, most locations worldwide had implemented multilayered review, which is the optimal approach, according to the consensus conference.

Once again, this survey identified sampling as the most common underlying cause of discordances, reported for 95.6% (430 of 450) of North American laboratories and 84.3% (86 of 102) of international laboratories. This finding has not changed in decades and accords with previous studies.^9,10  Unexpectedly, international laboratories had statistically more preanalytic clinical factors and more laboratory technical factors as the cause for discordance: inappropriate/inadequate history, surgical specimen requiring reorientation, and preparation factors hindering interpretation. Interpretive factors plus the need for additional sections and levels were common secondary reasons despite laboratory location. These are all recognized factors that affect CHC.¹¹ 

The cytologic-histologic diagnoses that are agreed upon as discordant were similar in the North American and international locations. Both recognized discordances for combinations of normal with high-grade diagnoses (normal biopsy with HSIL Pap test, HSIL biopsy with NILM). This is understandable, since these combinations mirror the minimal CLIA requirements for prompting CHC. Interestingly, international laboratories more frequently considered discordances with LSIL biopsy than those in North America. Since laboratories may select their own thresholds for CHC review, these different combinations may reflect regional or national CHC practices. Laboratories may independently define what constitutes a discordance, and in North America, some laboratories do not consider LSIL to HSIL as constituting a significant discordance. This tendency is reflected in our survey, where twice as many international laboratories considered LSIL Pap test/HSIL biopsy as a discrepancy. Sampling remains the primary cause of discordance, but few laboratories monitor sampling discordance rates by clinical providers: North America (5.7%; 23 of 402) and international (10.1%; 9 of 89). Although there are new guidelines for colposcopy practices, more studies are needed to gauge compliance to determine their contribution to discordances in CHC.^12–16 

Quality improvement metrics are standard practice in both North American and international locations. When selecting monitors for CHC statistics, laboratories were not very different. The most common metric is the percentage of Pap tests that correlate with biopsies from the same site. This metric is the simplest metric for determining a positive CHC and has the least interobserver variability.^17,18  Although this value is easily converted into the PPV, international laboratories were significantly more likely to take this step than North American laboratories (P < .001), and it is the preferred standard CHC metric proposed by the Gynecologic Cytopathology Quality Consensus Conference.³  For CHC purposes, PPV is defined as the number of matching correlating pairs divided by the number of total correlating pairs, with a published range of 71% to 94% by Jones and Novis⁹  and Raab et al.¹⁰ 

Two metrics were significantly less likely to be monitored internationally: the screening error rate by cytotechnologist and the interpretive error rate by cytotechnologist (P = .002 and P < .001, respectively). This may reflect a lack of cytotechnologists/cytology screening personnel internationally, lack of pathologist oversight of these personnel, or separation of screening services from pathologists, but the reasons are unclear. However, international laboratories did indicate that they performed remediation/retraining of cytotechnologists and were more likely to do so for every major discordance (58.5%; 48 of 82) than for error trends (36.6%; 30 of 82), whereas the reverse was true for North American laboratories. Remediation/retraining for every major discordance for international pathologists was also significantly (P < .001) more frequent. Communication of these metrics to appropriate personnel continues to be robust. What is clear from our survey is that about half of laboratories shared CHC results with cytotechnologists and three-fourths with pathologists. This is a positive trend, since CHC provides very useful self-regulating data and identifies difficult diagnostic dilemmas for both parties. In addition, discordances were likely to be discussed directly with the patient's provider and documented within the pathology report in most cases.

In summary, this survey has identified recent CHC practices in North American and international laboratories. Both geographic locations showed very high implementation of CHC standard practices, and practice patterns for both locations were surprisingly similar. The survey shows an overall decline of correlation rates in gynecologic specimens when compared to the prior survey. The underlying etiology for this small but notable decline is unclear. Unfortunately, this survey was not designed to determine reasons why laboratories do not perform CHC, and this will be an interesting topic for future studies. Upcoming changes in cytology screening and prevention practices, particularly with respect to HPV vaccination, HPV primary screening, and the diminishing cytotechnologist workforce may affect these continuing practice patterns and more importantly may affect the quality metrics of gynecologic cytopathology overall.^19–22