The Anatomic Pathology Hospitalist Model : A Novel Approach to Frozen Section Practice in a Tertiary Care Center Open Access

Following approval by our institutional review board, the study cohort was extracted from our electronic pathology database using the following inclusion criteria: (1) intraoperative frozen section performed at our institution between July 1, 2021, and October 31, 2023; and (2) frozen-permanent diagnostic concordance documented prospectively (ie, at the time of final pathologic diagnosis) in the electronic pathology database. “Gross-only” intraoperative consultations (ie, no blocks frozen) were excluded, as were neuropathology and pediatric frozen sections, as the latter are performed almost exclusively by attending pathologists subspecializing in those disciplines.

Frozen sections were performed across 5 surgical sites, including 3 adjoining facilities on the main medical campus (the university hospital, cardiovascular center, and women’s hospital) and 2 off-site outpatient surgery centers (located 4 and 39 km [2.5 and 24 miles] away from the pathology department’s new off-campus location).

For each frozen section block, the following parameters were tabulated: pathology case identifier, date performed, surgical site (see preceding paragraph), frozen section pathologist, frozen section turnaround time (minutes between arrival in the frozen section laboratory and reporting of the frozen section diagnosis to the surgeon), permanent sign-out service (breast, general service [head and neck, bone and soft tissue, endocrine], gastrointestinal/pancreaticobiliary, gynecologic, genitourinary, and lung/thoracic), and frozen-permanent diagnostic concordance.

The academic rank and diagnostic subspecialization of each attending pathologist were documented. Each attending pathologist’s total days of frozen section versus non–frozen section service coverage were tabulated for an 18-month period before (July 1, 2018, through December 31, 2019; selected to avoid service disruptions from the COVID-19 pandemic) and after (January 1, 2022, through June 30, 2023) the implementation of the hospitalist program.

Each frozen section pathologist was assigned to 1 of 3 groups: hospitalist attending pathologist, nonhospitalist attending pathologist, or fellow. Hospitalist attending pathologists are faculty with clinical and academic emphasis in frozen section pathology, as outlined in the Introduction. Of 3 hospitalist attending pathologists, 2 were in this role for the entire study period, and the third from July 1 to October 31, 2023. All 3 filled this position immediately after completing fellowship training in gynecologic (n = 1), breast (n = 1), or general surgical pathology (n = 1). Nonhospitalist attending pathologists include all other surgical pathology faculty covering the frozen section service. In the 28-month study period, 22 faculty filled this role for a median of 24.5 (range, 2–28) months. Pathology fellows are senior trainees with graduated sign-out responsibility, who are credentialed as and have clinical privileges of junior faculty at time of hire. Each full-year (12-month) general surgical pathology fellow staffs 11 weeks of frozen section service, with graduated independent sign-out beginning in the third week. Subspecialty pathology fellows (from gastrointestinal, genitourinary, gynecologic, breast, and pulmonary pathology fellowships) may cover 2–4 weeks of frozen section service. In the 28-month study period, 16 fellows (13 full-year general surgical pathology fellows, and 3 subspecialty fellows [none of whom had previously completed the full-year surgical pathology fellowship]) covered the frozen section service for a median of 8 (range, 1–13) months. Only the frozen sections that were independently interpreted, reported, and signed by fellows were assigned to them for comparative analyses.

Between-group comparisons of categoric variables were performed with a Pearson χ² test. Between-group comparisons of continuous variables were performed with the Kruskal-Wallis test, with significant results further analyzed by a Dunn post hoc test. The paired t test was used to compare service distributions before versus after implementation of the hospitalist program. All P values are 2-tailed, with statistical significance defined by α = .05. The Holm method was used to correct for multiple comparisons, where appropriate, with Holm adjusted α values provided alongside P values. In these cases, P values less than the adjusted α were considered statistically significant.

Hospitalists and fellows performed, on average, significantly more frozen sections than nonhospitalists. In the 28-month study period, a total of 18 710 surgical pathology frozen section blocks were performed from 5143 surgeries across 5 surgical sites. Prospective frozen-permanent concordance data was documented for 18 313 blocks from 5003 surgeries, representing the final study cohort. Hospitalist attending pathologists performed 4744 (25.9%) of 18 313 blocks, nonhospitalist attending pathologists performed 7362 of 18 313 (40.2%), and pathology fellows performed 6207 of 18 313 (33.9%) (Figure 1). Of 2374 blocks frozen at 2 off-campus outpatient surgical centers, fellows performed 2 (<0.1%), hospitalist attending pathologists performed 710 of 2374 (30%), and nonhospital attending pathologists performed 1662 of 2374 (70%) (Supplemental Table 1, see supplemental digital content containing 5 tables at https://meridian.allenpress.com/aplm in the May 2025 table of contents.).

To account for changes in attending pathologists’ service obligations and the 12-month duration of the surgical pathology fellowship, the median number of frozen section blocks performed by hospitalist attending pathologists (median, 2174), nonhospitalist attending pathologists (median, 397), and fellows (median, 402) was normalized to the number of months each diagnostician was eligible for frozen section service. Hospitalist attending pathologists (median, 87.0 blocks per month) and fellows (median, 52.9 blocks per month) performed significantly more frozen section blocks per month than nonhospitalist attending pathologists (16.9 blocks per month; P = .002 and P < .001, respectively; adjusted α = .017) (Supplemental Table 2).

Among 13 nonhospitalist attending pathologists on faculty both before and after implementation of the AP hospitalist model, frozen section service days were significantly fewer after (656 of 2951 [22%] total service days, equating to an average of 2.8 frozen section service days per person per month) versus before implementation (857 of 2974 [30%] total service days, equating to an average of 3.7 frozen section service days per person per month; P = .005). In contrast, the hospitalist attending pathologists spent 343 of 494 total service days (69%) on frozen section service, equating to an average of 9.5 frozen section service days per hospitalist per month. Eleven of 13 nonhospitalist attending pathologists saw a decrease in the percentage of service days devoted to frozen section service (Figure 2), and the remaining 2 also covered fewer frozen section service days, offset only by an even greater decrease in days of non–frozen section service (likely attributable in both cases to increased administrative efforts). The nonhospitalist attending pathologists’ frozen section distribution between the main medical campus and off-campus surgical sites did not, on average, differ before (281 of 857 [32.8%] frozen section days at main campus) versus after implementation (218 of 656 [33.2%] frozen section days at the main campus; P = .93) (Supplemental Table 3).

Frozen-permanent concordance rates were marginally but significantly higher for hospitalist versus nonhospitalist attending pathologists. The frozen-permanent concordance rate was significantly higher for hospitalist attending pathologists (4701 of 4744 blocks, 99.1%) than for nonhospitalist attending pathologists (7259 of 7362 blocks, 98.6%; P = .02, adjusted α = .017). The concordance rate for fellows (6138 of 6207 blocks, 98.9%) did not differ significantly from the other 2 groups (see Figure 1). Among hospitalist and nonhospitalist attending pathologists, concordance rates did not differ between assistant (8850 of 8956 blocks; 98.8%), associate (1715 of 1733 blocks; 99.0%), and full professors (1395 of 1417 blocks; 98.4%) (P = .39). Concordance rates did not differ between full-year surgical pathology fellows (5771 of 5836 blocks; 98.9%) and subspecialty fellows (367 of 371 blocks; 98.9%) (Supplemental Table 2).

Frozen-permanent concordance rates were significantly lower for gynecologic and genitourinary frozen sections, but hospitalist attending pathologists outperformed in these areas. In a pooled analysis of all 3 practitioner groups, the concordance rate for gynecologic frozen sections (666 of 685 blocks, 97.2%) was significantly lower than for frozen sections from the general service (12 262 of 12 380 blocks, 99.0%; P < .001, adjusted α = .0033) and breast (2983 of 3018 blocks, 98.8%; P = .002, adjusted α = .0038). However, the concordance rate for gynecologic frozen sections was significantly higher for hospitalist attending pathologists (190 of 190 blocks, 100%) than for nonhospitalists (288 of 299 blocks, 96.3%; P = .008, adjusted α = .025) or fellows (188/196 blocks, 95.9%; P = .005, adjusted α = .017) (Figure 3, A and B).

Similarly, the concordance rate for genitourinary frozen sections (435 of 448 blocks, 97.1%) was significantly lower than for the general service (P < .001, adjusted α = .0036) and breast (P = .003, adjusted α = .0042) among all diagnosticians, but significantly higher for hospitalist attending pathologists (121 of 121 blocks, 100%) than for fellows (148 of 157 blocks, 94.3%; P = .007, adjusted α = .017).

The concordance rate for lung frozen sections appeared to be significantly higher for nonhospitalists (257 of 257 blocks, 100%) than for hospitalists (103 of 106 blocks, 97.2%; P = .007, adjusted α = .017) and fellows (450 of 460 blocks, 97.8%; P = .02, adjusted α = .025), though statistical evaluation was limited by the low numbers of discrepant cases in the first of these comparisons.

None of the other between-service comparisons were statistically significant.

Frozen-permanent concordance rates were significantly lower at the women’s hospital than at other sites, reflecting lower concordance among gynecologic frozen sections. Surgical subspecialties are unequally distributed between surgical sites (Supplemental Table 4). Mirroring its disproportionate share of gynecologic frozen sections, the women’s hospital showed significantly lower overall frozen-permanent concordance (764 of 784 blocks, 97.4%) than the university hospital (13 995 of 14 149 blocks, 98.9%; P < .001, adjusted α = .005), but the concordance rate for women’s hospital frozen sections was significantly higher for hospitalist attending pathologists (190 of 190 blocks, 100%) than for nonhospitalist attending pathologists (315 of 325 blocks, 96.9%; P = .02, adjusted α = .025) and fellows (269 of 279 blocks, 96.4%; P = .01, adjusted α = .017). No other between-site comparisons were significant after adjusting for multiple comparisons (see Supplemental Table 1).

Frozen-permanent concordance rates were not correlated with the subspecialty expertise of the frozen section pathologist. We hypothesized that frozen-permanent concordance rates might be higher within individual pathologists’ areas of subspecialty expertise. However, concordance rates did not differ between subspecialists, non-subspecialist attending pathologists, and fellows for breast, gynecologic, genitourinary, general service, or lung frozen sections (Supplemental Table 5). The concordance rate for gastrointestinal/pancreaticobiliary frozen sections was lower for subspecialists (89 of 94 blocks, 94.7%) than for fellows (290 of 292 blocks, 99.3%; P = .003, adjusted α = .017), though this counterintuitive result may reflect a bias toward involving subspecialty faculty in more challenging cases.

Turnaround times were not substantially different between hospitalists, fellows, and nonhospitalists. Across all frozen section blocks in the study period, turnaround time was significantly longer for hospitalists (median 32 minutes, mean 36.0 minutes) than for fellows (median 30 minutes, mean 33.3 minutes; P < .001, adjusted α = .025) and nonhospitalists (median 29 minutes, mean 31.7 minutes; P < .001, adjusted α = .017) (Figure 4, A). However, evaluation of cases comprising only a single frozen section block revealed the inverse: significantly shorter turnaround time for hospitalists (median 17 minutes, mean 19.6 minutes) than for nonhospitalists (median 19 minutes, mean 21.2 minutes; P = .01, adjusted α = .017), with no significant difference versus the fellows (median 18 minutes, mean 20.2 minutes; P = .27). These disparate results appear to reflect a difference in the number of frozen section blocks per case for each group: hospitalists froze on average significantly more blocks per case (mean 2.53) than nonhospitalists (mean 2.33; P < .001, adjusted α = .025) or fellows (mean 2.27; P < .001, adjusted α = .017) (Figure 4, B).

This retrospective interim analysis of our AP hospitalist program’s frozen section performance shows that surgical pathology faculty with an explicit frozen section emphasis can cover a disproportionate share of a high-volume, multisite frozen section service, significantly reducing frozen section service obligations for other faculty, while achieving diagnostic accuracy equal to or greater than nonhospitalist attending pathologists and incurring no clinically significant change in turnaround time. At our institution, this was attained with hospitalists directly out of fellowship. Our experience points to a model of subspecialization in general frozen section pathology, with clinical, educational, and academic implications in an evolving practice landscape.

During the study period, the average frozen section hospitalist covered 3.4 times as many frozen section service days and performed 5.5 times as many frozen section blocks as the average nonhospitalist attending pathologist. The hospitalists’ disproportionate frozen section coverage resulted in a ∼25% reduction in nonhospitalists’ frozen section coverage, falling from an average of 3.7 to 2.8 days per person per month after implementation of the AP hospitalist program. Total service days remained steady among nonhospitalists, indicating proportionally greater coverage of subspecialty sign-out services.

Our logistic and operational goals were attained without sacrificing diagnostic excellence, as the hospitalists’ frozen-permanent concordance rate exceeded that of the nonhospitalists. Concordance rates in both groups were at or above the 98.3%–98.6% concordance figures reported in 2 multi-institutional series totaling 228 153 frozen section blocks.^20,21  The basis for the small but statistically significant difference in diagnostic concordance between hospitalists and nonhospitalists is uncertain, and the durability and portability of this result remains to be seen, particularly given the small number of hospitalists studied here. However, it is possible that the hospitalists’ greater frozen section volume could result in improved adaptation to frozen section histopathology, rapport with technical staff and surgical colleagues, familiarity with surgical decision-making algorithms, formulation of surgically actionable frozen section diagnoses, or some combination of these factors. The hospitalists’ higher average number of blocks per case may also have played a role in reducing discrepancies resulting from undersampling.

Turnaround times were slightly but significantly longer for hospitalists than for nonhospitalists across all cases, but shorter for hospitalists on cases comprising only a single frozen section block. This seemingly contradictory finding may reflect a difference in the 2 groups’ case mix (ie, hospitalists performing a greater proportion of complex cases with multiple parts received from the operating room), as could be suggested by the hospitalists’ and nonhospitalists’ significantly different service distribution across frozen section sites (see Supplemental Table 1). Alternatively (or additionally), it could reflect the hospitalists’ tendency to freeze more tissue blocks in cases where this decision is left to the pathologist’s discretion. In either event, interviews with surgeons (data not shown) did not reveal dissatisfaction with or even awareness of these between-group differences, suggesting they are of little or no clinical significance.

What do our findings mean for our novel AP hospitalist model? Though the profile of the “ideal” AP hospitalist remains unresolved, the strong performance of our hospitalist team—all 3 of whom filled this position directly out of fellowship—suggests that junior faculty may be well suited to this role. Further, across hospitalists and nonhospitalists, we found no correlation between the frozen section pathologist’s seniority and frozen-permanent concordance, though published data on this count are conflicting.²²  High frozen-permanent concordance among our fellows may also support junior pathologists’ suitability for this role, though this figure may be inflated to some extent by the availability of faculty to assume responsibility for challenging cases. The optimal fellowship background (if any) for AP hospitalists remains uncertain. Published data on the correlation between frozen-permanent concordance and the frozen section pathologist’s diagnostic subspecialty are conflicting and limited by small cohorts,^22–25  but we found no such correlation in our study. On the other hand, the hospitalists’ relative strength in gynecologic frozen sections and weakness in lung may at least partly reflect the fellowship mix of our 3 hospitalists. While our data suggest that hospitalists need not necessarily have fellowship training in areas of high frozen section volume (eg, head and neck pathology, which accounts for the vast majority of our general frozen sections), it seems likely that the combined expertise of an AP hospitalist group with diverse subspecialty training could improve diagnostic as well as educational and academic outcomes.

The AP hospitalist model is still in its earliest phase, and a discussion of every opportunity and challenge is beyond the scope of this manuscript. However, we have much to learn from our hospitalist colleagues in other fields of general and specialized practice. Analogous to our counterparts in internal medicine hospitalist programs, we propose that AP hospitalists can fill multiple interrelated roles in the improvement of hospital-based pathology services, including quality assurance and quality improvement, education, and academic inquiry. Experience has shown that these missions are key to the success of other specialized hospitalist programs.²⁶ 

Like hospitalists in general and specialized medical fields, AP hospitalists are well positioned to play a key role in departmental and institutional quality assurance and quality improvement initiatives,¹²  including validation of novel protocols (eg, telepathology), resolution of frozen section–related issues with surgeons and technical staff, and prospective review of frozen-permanent discordances to permit timely resolution of disagreements. Additional areas of growth potential include implementation of intraoperative ancillary studies, and frozen section telepathology consultation for small or remote hospitals and surgical centers.²⁷ 

Regarding education, published data indicate that trainees give internal medicine hospitalists higher scores than nonhospitalists in educational aptitude, availability, and inpatient medicine expertise.^28–30  Our AP hospitalists have assumed an analogously outsized role in frozen section education for pathology residents, fellows, nonhospitalist pathology attending pathologists, and surgical colleagues, in the context of both routine frozen section sign-out and lecture- and case-based formats. Increased extramural frozen section educational programming is a growth priority for our AP hospitalist program.

Finally, AP hospitalists in academic centers are well positioned to increase academic output in the realm of frozen section pathology. In our practice, frequent interactions with surgical colleagues have prompted multiple ongoing studies, particularly around questions of frozen section utilization and efficacy. Academically successful internal medicine hospitalist programs (as defined by robust scholastic output and, to a more variable degree, grant funding) are typically longstanding (mean, 13 years) and large (mean, 36 faculty),³¹  suggesting that the development of academic AP hospitalist programs should be viewed as a long-term project.

In each of these areas, frozen section specialists serve as an identifiable point of contact for colleagues both inside and outside of the department, and as an intuitive collaborator in multidisciplinary or multi-institutional efforts.^32,33  Multi-institutional collaboration on professional standards and practice guidelines for frozen section pathology is a long-term goal of our program.

The AP hospitalist concept also faces challenges, and again we can learn from established hospitalist models in other fields.³⁴  For instance, completion of at least one subspecialty fellowship is the norm in pathology, and we must consider the proper mix of AP hospitalist and subspecialty service for pathologists wishing to practice in both arenas. In early days, internal medicine hospitalists were defined as those with at least 25% clinical effort in inpatient care.³⁴  This cutoff, though arbitrary, sets a precedent for pathologists to combine the AP hospitalist role with their subspecialty practice. Currently, all 3 of our AP hospitalists also practice subspecialty sign-out. Additional challenges include the financial sustainability of the model,^14,32,35,36  loss of frozen section expertise among nonhospitalist faculty,¹⁶  adequate opportunities for academic productivity and promotion among AP hospitalists in academic centers,³²  mitigation of burnout and creation of adequate compensation structures to reduce turnover,³⁶  and development of appropriate tools for evaluating AP hospital performance.^35,37 

The experience of hospitalist programs in other fields indicates that key determinants of AP hospitalist programs’ feasibility will include frozen section volume, number and location of surgical sites, anticipated growth in frozen section service requirements, availability and functionality of telepathology, feasibility of simultaneous cross-coverage of other services (including other hospital-based services, such as rapid on-site cytopathologic evaluation or autopsy), educational mandates, and compensation.^14,36,38  Variations in these parameters, combined with the flexibility of combining frozen section and subspecialty services in different proportions, suggest that the appropriate number of designated hospitalists may vary considerably even between comparably sized departments. Considering all factors, we anticipate the model will find its earliest application in large practices with mostly or wholly subspecialized sign-out and responsibility for at least some frozen section staffing away from the main pathology offices. Though the translation of these principles may look somewhat different, we anticipate that smaller practices could also benefit from concentrating particular frozen section expertise in one or more pathologists.

The success of the AP hospitalist model will be dependent on its acceptance by pathologists, surgeons and other clinical colleagues, hospital administrators, and payers. Future work in this area will include quantitative assessment of the model’s financial feasibility as well as qualitative assessment of the role’s acceptance by different stakeholders.

In summary, our data indicate that AP hospitalists with a clinical and academic emphasis in frozen section pathology can reduce logistic burdens on nonhospitalist faculty, while maintaining or even improving frozen section accuracy. Though certain elements of our model have likely been implemented at various institutions, this is, to our knowledge, the first formal codification of an AP hospitalist program. This dynamic model is intended to grow with the changing needs of pathologists, their clinical and surgical colleagues, and the patients and healthcare systems they serve. Frozen section pathology, like any discipline of surgical pathology, is a multifaceted practice area that demands clinical, educational, academic, and administrative excellence. A model that recognizes and interweaves these elements has the potential to substantially advance this field.

The authors wish to thank Threase Nickerson for her assistance in tabulating pathologists’ service distribution data.