Quantifying the impact of the COVID‐19 pandemic on gastrointestinal cancer care delivery

Abstract Background and Aim This study quantifies how changes in healthcare utilization and delivery during the first months of the COVID‐19 pandemic have altered the presentation, treatment, and management of patients with gastrointestinal (GI) malignancies within an academic health system. Methods and results Patients diagnosed with a GI malignancy (ICD10: C15‐C26) who received medical care within the health system during the observation period (first 44 weeks of 2019 and 2020) were identified for a retrospective cohort study. Deidentified patient encounter parameters were collected for this observation period and separated into pre‐pandemic (weeks 1–10) and early pandemic (weeks 11–20) study periods. Difference‐in‐difference analyses adjusted for week‐specific and year‐specific effects quantified the impact of the COVID‐19 pandemic on care delivery between pre‐pandemic and early pandemic study periods in 2020. Across all GI malignancies, the COVID‐19 pandemic has been associated with a significant decline in the number of patients with new patient visits (NPVs) (p = 1.2 × 10−4), Radiology encounters (p = 1.9 × 10−7), Surgery encounters (p = 1.6 × 10−3), Radiation Oncology encounters (p = 4.1 × 10−3), and infusion visits (6.1 × 10−5). Subgroup analyses revealed cancer‐specific variations in changes to delivery. Patients with colorectal cancer (CRC) had the most significant decrease in NPVs (p = 7.1 × 10−5), which was significantly associated with a concomitant decrease in colonoscopies performed during the early pandemic period (r2 = 0.722, p = 2.1 × 10−10). Conclusions The COVID‐19 pandemic has been associated with significant disruptions to care delivery. While these effects were appreciated broadly across GI malignancies, CRC, diagnosed and managed by periodic screening, has been affected most acutely.


| INTRODUCTION
In responding to the COVID-19 pandemic, health care systems have been forced to redefine established processes of care delivery in order to balance local surges in COVID-19 patient volumes, infection risk, resource limitations, and the baseline needs of existing patient populations. This has particular relevance for cancer care as health systems balance benefits and risks of frequent clinical encounters to cancer patients, who are thought to be at increased risk of mortality and complications from COVID-19. 1,2 Although cancer patients traditionally have frequent points of contact with the healthcare system for clinical visits, imaging, procedures, and infusion sessions, this care delivery model has been significantly disrupted by the COVID-19 pandemic and implementation of infection prevention measures in hospitals. In the early pandemic, many national healthcare systems and professional societies limited or delayed screening procedures and elective surgeries. 3 However, many clinical decisions for cancer patients have been decided on an individual basis, and few studies quantify the degree of change and impact of these decisions on diagnosis, treatment, and outcomes. Furthermore, additional time and follow-up will be needed to fully understand exactly how COVID-19-related changes to care have affected long-term patient outcomes. This study seeks to quantify how the first months of the COVID-19 pandemic have affected the presentation, treatment, and management of patients with gastrointestinal (GI) malignancies in an academic medical center.

| PATIENTS AND METHODS
Patients with GI malignancies were identified by the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) principal diagnosis codes (ICD10: C15-C26). The electronic medical record was queried for deidentified encounter data for patients with a documented GI malignancy, including the weekly number of patients with new patient visits (NPVs) and/or specialty encounters during a 44-week observation period in 2019 and 2020. Specialty encounters were isolated by searching for either an encounter department (Radiology, Surgery, Radiation Oncology) or encounter type (infusion visit) of interest. The study period, which includes the first 20 weeks of 2020 (1/1-5/20), was split into two subdivisions, the prepandemic period (weeks 1-10, 1/1-3/17) and the early pandemic period (weeks 11-20, 3/18-5/19). Difference-in-difference analyses adjusted for week-specific and year-specific effects were used to compare encounter details between the early pandemic and pre-pandemic periods of 2020 with weeks 1-20 of 2019 as an additional control.
Within this framework, the difference-in-difference analysis quantified the impact of the COVID pandemic on health encounters by building a linear model (with encounter data from 2019 and the pre-pandemic period of 2020) to isolate the otherwise unexplained decline in the number of patients with encounters during the early pandemic period of 2020. When not otherwise specified, P-values reflect the probability that an observed change in the frequency of patients with an encounter during the early pandemic period could be ascribed to chance after accounting for week-specific and year-specific effects in a linear model. difference between cohorts on the basis of sex (χ 2 = 0.15, p = .93) nor race (χ 2 = 2.82, p = .99).
When analyzed collectively, the early pandemic period was associated with significant changes in care delivery for patients with a GI malignancy. A total of 49.8 fewer patients had NPVs in a given week (44.9% of the 2019 average) during the early pandemic period than would have otherwise been expected, representing a statistically significant decrease that could not be explained by week-specific or year-specific effects alone (p = 1.2 Â 10 À4 ; Figure 1, Table 2). This was seen across all parts of the health system, and among patients with a GI malignancy, the early pandemic period was associated with otherwise unexplained decreases in Radiology encounters (À175.  Table 2). These findings were all significant after correction for week-specific and year-specific effects (Radiology: p = 1.9 Â 10 À7 , Surgery: p = 1.6 Â 10 À3 , Radiation Oncology: p = 4.1 Â 10 À3 , infusion visits: p = 6.1 Â 10 À5 ).
Subgroup analyses revealed cancer-specific variation in the effect size of changes to care delivery during the early pandemic period.
Comparing the weekly count of patients with NPVs revealed independently significant declines most pronounced among patients diag-  week, 91.0% of the 2019 average, p = 5.5 Â 10 À12 ) ( Figure 2, Table 3). By the end of week 20, there were 6436 fewer patients who underwent colonoscopies in 2020 as compared to 2019 (Table 3). By Week 44, this cumulative deficit grew to 9659 patients despite increases in colonoscopy rates (Table 3)   The modeled deficit reflects the otherwise unexplained decrease in the number of patients with encounters per week in a difference-in-difference linear model accounting for week-specific and year-specific effects.
While overall decreases in NPVs and encounter types were seen in patients with all GI cancers, subgroup analyses by cancer type showed disease specific variation in treatment modalities and may have potential implications for the resurgence. For example, as pancreatic cancer typically presents with symptomatic or incidentally caught late-stage disease, pancreatic cancer presentation, and subsequent initial urgent treatment may not be as affected by the disruption of health care services from COVID-19. Similarly, no significant decreases in surgical department encounters were seen in esophageal and anal cancers, which also tend to present at higher stages and with a greater symptomatic burden. This is corroborated by the relatively low impact seen on surgical and radiation oncology encounters for patients diagnosed with pancreatic cancer. In our study, however, patients with pancreatic cancer were found to have significantly fewer radiology and infusion encounters than expected.
This trend is concerning as research suggests that the timing of chemotherapy administration affects pancreatic cancer outcomes. 6 Contributing factors likely include the temporary freeze of clinical trial enrollment at our large academic center during the pandemic and subsequent patient preference to undergo standard of care chemotherapy options closer to home and in the community.
Perhaps most worrisome is that some of the greatest decreases in NPVs and specialty encounters were seen in CRC, the only gastrointestinal cancer that is effectively and universally screened for, and as a result, can often be detected in earlier stages than other GI cancers.
These decreases were observed in the setting of the precipitous Altogether, these results suggest that all forms of routine surveillance scans and monitoring are also backlogged and may contribute to further diagnostic or therapeutic delays. As seen in Sozutek et al, a prospective study of 177 GI cancer patients who underwent elective surgery without delay from March -November 2020, the risk of delaying procedures may be significantly greater than the risk of contracting COVID-19 with proper isolation measures. 14 One limitation of our study is that the decreases seen in a single academic institution may not be generalizable to other sites. However, it is likely that due to the highly multidisciplinary nature of gastrointestinal cancer care, that these trends parallel those observed in similar academic health systems. In addition, our study's reliance upon our electronic health record's reporting tools, which were used to generate aggregate weekly metrics, precluded data reporting on an individual patient-level basis. Ongoing and future work at our institution will examine more detailed, patient-level trends, and trends in cancer staging will be interesting to examine as we move beyond the current pandemic.
It remains to be seen what the ultimate impact on patient survival and outcomes will be due to the COVID-19 pandemic, as many pertinent endpoints will take time to mature. It will also take time to fully determine the impact of delays in screening will have on factors such as patients' stage at presentation. Nonetheless, this study demonstrates a significant reduction in health encounters pertinent to the detection, diagnosis, treatment, and management of GI malignancy.
Most striking is the significant impact on CRC screening, potentially adversely affecting outcomes for a disease that is treatable and curable, especially in the early stages. Although it may be too early to ascertain the full impact of COVID on gastrointestinal cancer care and outcomes, one clear area of focus may be to increase the availability and access to colonoscopy screenings as hospital systems continue to adapt to the COVID-19 pandemic and expand health system capacities.