Cancer treatment delays among cancer patients living with HIV during the COVID‐19 pandemic in the United States

Abstract Background The COVID‐19 pandemic led to care disruptions across the cancer continuum. It is unknown if immunosuppressed patients with cancer, who may be at higher risk for complications of SARS‐CoV‐2 infection, are disproportionately impacted. Thus, we aimed to compare delays in cancer treatment initiation between people living with HIV (PLWH) and cancer, the general cancer population (GCP), and patients with cancer and a history of solid organ transplant (SOT). Comparisons were made across the period 2 years preceding the pandemic versus the first year of the pandemic. Methods We used data from a real‐world electronic health record‐derived de‐identified database (2018–2021) comprised of US patients with cancer from 800 sites of care across the country. We included patients with 19 different cancer types. We calculated time to cancer treatment initiation (TTI) as the difference between the date of cancer diagnosis and the earliest date that cancer treatment was recorded. Results The sample included 181 PLWH, 65,073 GCP patients, and 195 patients with a SOT. Difference‐in‐difference regression models adjusted for age, sex, and presence of metastatic disease at cancer diagnosis revealed a significant increase in delayed TTI among PLWH compared to the GCP during COVID‐19 versus prior to COVID‐19, with delays increasing by approximately 1 month during the pandemic (DID: 32.6 days [8.9–56.3]; p = 0.007). The increase in TTI for PLWH was observed across treatment modalities, including surgery (DID: 55.1 [28.8–81.3], p < 0.001) and systemic therapy (DID: 30.4 [4.6–56.3], p = 0.021). Conclusions/Relevance PLWH experienced significant delays in cancer treatment initiation after diagnosis during the first year of COVID‐19, delays that may negatively impact cancer outcomes. These data warrant patient and provider attention as the pandemic continues to impact the US healthcare system.


| BACKGROUND
With the emergence and spread of the novel coronavirus pandemic (COVID- 19), the provision of cancer care was adversely impacted in the United States.Utilization of cancer prevention services decreased significantly due to a variety of factors, including office closures during lockdown, patient fear of contracting COVID-19 in the healthcare setting, or lack of transportation available to certain patients following widespread public transit closures. 1,2In April of 2020, routine screenings for breast and colorectal cancer had dropped by 89.2% and 84.5%, respectively, compared to April of 2019. 3Many health systems suspended routine cancer screening and diagnostic procedures due to low personal protective equipment inventory or lack of established safety protocols. 3This interruption in care also extended to cancer treatment; within the first month of the pandemic, the U.S. Food and Drug Administration announced that 26 oncology medications were in short supply. 4Furthermore, cancer surgeries were impacted by staff shortages and lack of operating room space or hospital beds, and COVID-related changes in guideline recommendations. 5,6IV care during COVID-19 was also adversely impacted, with decreases in HIV testing and an increase in loss to follow-up during HIV therapy.7,8 Similar to patients with cancer, people living with HIV (PLWH) may have avoided healthcare interaction due to concern for contracting the virus as a consequence of immunosuppression.9 Outpatient HIV clinic office hours were reduced; for example, in South Carolina, 82% of HIV clinics were partially interrupted or completely closed during the beginning of the COVID-19 outbreak.8 Furthermore, HIV service interruptions in that study were greatest in areas with higher percentages of uninsured patients, exacerbating a treatment access disparity that existed before COVID-19.8 Despite published data documenting the impacts of COVID-19 on cancer and HIV care separately, little is known about how COVID-19 impacted cancer treatment among PLWH.Cancer treatment and outcomes disparities in PLWH are well-documented in the pre-pandemic era, [10][11][12][13][14] so further interruptions to care in this vulnerable population during COVID-19 and its continued impact on the US healthcare system could have had serious implications for clinical outcomes.In this study, we compared time to cancer treatment initiation (TTI) between cancer patients with no known underlying immunodeficiencies, and cancer patients with underlying HIV infection, both during the 2 years prior to COVID-19 and during the first year of COVID-19.We utilized, an electronic health record-derived database from 800 sites of cancer care across the United States and hypothesized that delays in initiation of cancer treatment would be exacerbated for PLWH and cancer during the pandemic.

| Data source
We used de-identified electronic health record (EHR)derived data from the Flatiron Health Database, which is comprised of patients from approximately 280 cancer clinics (800 sites of cancer care) across the US.This longitudinal database includes both structured and unstructured patient-level, de-identified data obtained from technology-enabled chart abstraction. 15,16We included data from the following analytic cohorts in this study: Advanced Melanoma, Diffuse Large B-Cell Lymphoma, Early Breast Cancer, Advanced Endometrial Cancer, Follicular Lymphoma, Advanced Gastric Cancer, Hepatocellular Carcinoma, Advanced Renal Cell Carcinoma, Ovarian Cancer, Metastatic Pancreatic Adenocarcinoma, Metastatic Prostate Cancer, Small Cell Lung Cancer, Advanced Non-Small Cell Lung Cancer, Metastatic Breast Cancer, Acute Myeloid Leukemia, Chronic Lymphocytic Leukemia, Metastatic Colorectal Cancer, Multiple Myeloma, and Advanced Head and Neck Cancer.For patients with multiple cancer diagnoses, the first cancer diagnosis was used.

| Study groups
Three study groups were included in this analysis: people living with HIV (PLWH) and cancer, the general cancer population (GCP), defined as cancer patients without any underlying immunodeficiency, and cancer patients with a history of solid organ transplantation (SOT).We chose SOT patients as an additional, non-HIV comparison group because these patients are administered immunosuppressive medications to prevent transplant rejection, allowing us to determine if any observed treatment delays were specific to PLWH or were generally present in immunosuppressed patients with cancer.The following International The following ICD codes were used to select patients with SOT: D89.81, Z94, Z94.0, Z94.1, Z94.2, Z94.3, Z94.4,Z94.5, Z94.6, Z94.7, Z94.8, Z94.82, Z94.83, Z94.89, Z94.9.Patients with hematopoietic transplantation were excluded from the SOT group.Patients with inherited immune disorders were excluded from the study.Our inclusion and exclusion criteria are displayed in Figure 1.HIV status was extracted from ICD-10 codes with an associated date, which may or may not reflect HIV diagnosis date.
F I G U R E 1 Patient Inclusion and Exclusion Criteria.

| Primary outcome
The primary endpoint was time to treatment initiation (TTI).TTI was calculated as the difference in days between the date of the first cancer diagnosis in the real-world database and the earliest date listed for administration of any of the following cancer treatments: (cell-based treatment [CAR-T therapy, Provenge], radiation therapy, surgery, systemic therapy [chemotherapy, immunotherapy, androgen deprivation therapy], transplant, other localized treatment [cryotherapy, transarterial chemoembolization]).We allowed a 14-day window prior to diagnosis per oncologist-defined, rule based Line of Therapy, such that if a patient was listed as both diagnosed and treated within 2 weeks in the EHR, the treatment lag was set to zero.All time to treatment analyses reported were conducted among patients with available treatment data; 16.4%, 19.9%, and 13.3% of the GCP, PLWH, and SOT groups, respectively, did not have treatment information in the database, either due to no treatment administration or to lack of treatment documentation.

| Statistical analyses
The TTI was compared between study groups prior to COVID-19 and during COVID-19.The time period prior to COVID-19 was defined as a diagnosis date between January 1, 2018 and March 12, 2020.The time period during COVID-19 was defined as a diagnosis date between March 13, 2020, the day COVID-19 was declared a national emergency in the United States, and January 29, 2021.We used t-tests to compare median TTI between (1) the GCP and PLWH and cancer prior to COVID-19, (2) the GCP and SOT recipients prior to COVID-19, (3) the GCP and PLWH and cancer during COVID-19, and (4) the GCP and SOT recipients during COVID-19.A difference-in-differences analysis, which is a linear regression model that uses ordinary least squares to compare a change in the difference of an outcome between study groups over time, was performed to understand if differences in TTI between PLWH and the GCP changed across the pre-and during COVID-19 time periods.This adjusted analysis included patient age, sex, and extent of disease (presence versus absence of metastatic disease at cancer diagnosis) in the regression model.Based on sample size, we were able to conduct DID analyses both overall and specific to two treatment types (surgery and systemic therapy).All data management was conducted using R, 17 and statistical analyses were conducted using Stata V17.Flatiron Health datasets are de-identified using the Expert Determination method outlined in HIPAA section 164.514(b). 15,16Accordingly, this study was granted exemption by the Moffitt Cancer Center Scientific Review Board.

| RESULTS
Sociodemographic and clinical characteristics of patients with cancer are outlined in Table 1.The sample included 65,073 cancer patients in the GCP group, 181 PLWH, and 195 with a history of SOT.In the GCP group, the majority were male (51.5%), White (62.1%), over 65 years of age at the time of cancer diagnosis (62.7%), and the most common cancers were lung cancer (30.9%), cancers of the gastrointestinal (GI) tract (28.8%), and lymphomas (11.1%).These same patterns were seen in the SOT group.Among PLWH, although the most common cancer types were similar, the racial and age distribution differed, with 50.3% of PLWH and cancer identifying as either Black (35.4) or Hispanic (14.9%), and the majority being males (72.4%) aged 45-64 years at cancer diagnosis (63.0%).
Figure 2 summarizes TTI by demographics in all three study groups in the 2 years prior to COVID-19 in this set of cancer clinics.Overall, no statistically significant differences were observed in TTI between the GCP (median TTI 27.0 days) and PLWH (median TTI 27.0 days, p = 0.27), nor between the GCP and the SOT group (median 32.0 days, p = 0.39).A lack of significant differences in median TTI between PLWH and the GCP was consistent across most demographic and clinical groups.
Figure 3 summarizes the TTI across study groups during the first year of the COVID-19 pandemic.Overall, PLWH experienced statistically significant cancer treatment delays compared to the GCP (median TTI 41 days vs. 27 days, p = 0.0017).Females with HIV experienced pronounced cancer treatment delays, with a median TTI of 73.5 days, compared to 27.0 days for females in the GCP (p < 0.001).Statistically significant delays were also observed for smokers and obese PLWH, with median TTIs of 59.0 days and 65.0 days, respectively, compared to 30.0 and 31.0 days for smokers (p = 0.002) and obese patients from the GCP (p = 0.002).No difference was observed between the GCP and SOT recipients overall (median TTI 27 days vs. 20 days, p = 0.376).However, statistically significant cancer treatment delays were observed for SOT males compared to male GCP patients (median TTI 37.0 vs. 27.0,p = 0.005).In addition, both PLWH (median TTI 40 days) and SOT recipients (median TTI 40 days) ages 45-64 were more likely to experience cancer treatment delays compared to the GCP population aged 45-64 years (median TTI 27 days, p = 0.005 and 0.0005, respectively).This delay in treatment was not observed for patients age 65 and above.
Table 2 summarizes adjusted difference in difference regression estimates for changes in TTI between study groups prior to versus during COVID-19.After adjusting for age, sex, and metastatic disease, when compared to the GCP, we observed a statistically significant increase in TTI delays among PLWH and cancer during the COVID-19 pandemic.3).
Furthermore, due to the difference in number of subjects per cohort, a sensitivity analysis was conducted based on matched samples in the GCP.Cohorts were matched on age at diagnosis, sex, and year of diagnosis.The results of the sensitivity analysis revealed the same findings as the primary analysis, and are shown in Supplementary Tables S1-S4.

| DISCUSSION
This study demonstrated that during the COVID-19 pandemic, PLWH and cancer experienced significant delays in cancer treatment initiation compared to the general cancer population.We did not observe similar delays among cancer patients with a history of SOT, suggesting that the delay was specific to PLWH and not related to immunosuppression alone.This delay was observed for patients receiving both surgery and systemic therapy.These data warrant patient and provider attention, particularly as the pandemic continues to cause disruptions in healthcare access and cancer treatment utilization.Delays in cancer treatment initiation have substantial clinical relevance.9][20] For example, Ponholzer et al. reported that timely surgical resection in cT1-staged lung cancer patients had a 76.3% risk reduction of death 5 years after surgery when compared to those receiving surgery greater than 60 days after diagnosis. 21Delays in receipt of surgery reached an average 79 days during COVID-19 for PLWH in our analyses.Similar trends have been observed for delays in initiating systemic therapies.3][24] Therefore, although we did not analyze outcomes in this study due to limited available follow-up time postpandemic, the observed treatment delays among PLWH during COVID-19 are likely to precipitate adverse clinical outcomes in the coming years.
Cancer treatment delays during COVID-19 should also be considered in the context of previously established cancer treatment and outcome disparities for PLWH.PLWH are more likely to be diagnosed with cancer at more advanced stages. 25[13][14]25,26 More recently, data suggest that PLWH diagnosed with metastatic cancers, a population specifically recruited into the Flatiron database, are less likely to receive appropriate palliative care. 27Our data suggest that the already suboptimal provision of cancer care to PLWH has further deteriorated since the COVID-19 pandemic and requires urgent prioritization to avoid poor cancer outcomes in this vulnerable patient population.
To our knowledge, this was the first study examining differences in time between cancer diagnosis and cancer treatment for PLWH during the COVID-19 pandemic.We did not find a significant difference in TTI between PLWH and the GCP prior to COVID-19.This lack of a treatment disparity differs from previously published literature, [10][11][12][13][14] although prior studies primarily focused on receipt of any  28 Lower CD4+ counts are also associated with COVID-19 ICU admissions, mechanical ventilation, and death. 28Mortality rates for PLWH with COVID-19 were 9.4% overall and 51.5% among those admitted to an ICU. 29In summary, it is estimated that COVID-19 provided an 80% excess risk of death among HIV-infected individuals compared to those without diagnosed HIV/AIDS. 30While non-cancer, non-HIV medical co-morbidities may account for some of this excess risk, rather than HIV itself, PLWH and oncologists may have still been hesitant to bring immunosuppressed patients into healthcare settings, particularly during the first year of the pandemic (2020) when knowledge surrounding COVID prevention, spread, and treatment was more limited.
A major strength of this study is the uniqueness of the Flatiron Health database.Flatiron Health is an EHRderived, real-world clinical dataset with a substantial portion of cases with advanced disease, as well as documentation of multiple lines of cancer therapy.Flatiron Health de-identified data originated from approximately 280 US cancer clinics (~800 sites of care).Limitations of the study include the overall low number of patients with HIV and history of SOT diagnosed with cancer in the examined 1-year period after the emergence of COVID-19.Additionally, heterogeneity of the cancer types included in the study population made granular comparisons challenging.Finally, 16.4%, 19.9%, and 13.3% of the GCP, PLWH, and SOT groups, respectively, did not have treatment information in the database and thus were excluded from analysis.This may have been due to no receipt of treatment or lack of data in the medical record due to receipt of treatment elsewhere.Nonetheless, the percentage of missing data was comparable among the three groups and so was unlikely to appreciably impact our key findings.
T A B L E 3 Difference-in-difference analysis for time to cancer treatment initiation prior to and during COVID-19 by SOT status and treatment type.This is the first study to our knowledge that has examined delays in cancer treatment in PLWH during the COVID-19 pandemic.We observed that during COVID-19, delays in cancer treatment initiation for PLWH compared to the general cancer population were nearly 1 month greater.This delay was not observed among individuals with a history of SOT, suggesting a cancer treatment disparity specific to PLWH that warrants patient and provider attention.

65,073) Patients living with HIV and cancer (n = 181) Patients with cancer with a history of solid organ transplant (n = 195)
Sociodemographic and clinical characteristics of the general cancer population, patients living with HIV and cancer, and patients with cancer and a history of solid organ transplant, Flatiron Health (2018-2021).
T A B L E 1 was consistent when stratifying by patients who underwent surgery (DID: 55.1, 95% CI: 28.8-81.3,p < 0.001) and patients who received a systemic cancer therapy as their first treatment (DID: 30.4; 95% CI: 4.6-56.3,p = 0.021).We did not observe similar changes in the relative TTI between the SOT group and the GCP during COVID-19 (Table

to COVID-19 During COVID-19 Unadjusted DID p- value Adjusted DID a p- value HIV GCP Difference p- value HIV GCP Difference p- value
Difference-in-difference analysis for time to cancer treatment initiation prior to and during COVID-19 by HIV status and treatment type.treatmentamong PLWH rather than timely initiation of treatment.The large number of advanced and metastatic cancer patients included in the dataset (10 of the 19 cancer cohorts focused on advanced or metastatic disease) may have contributed to the similarity in TTI prior to COVID-19.As such, the observed delay in cancer treatment during COVID-19 among PLWH may have been even more extreme if we considered the full spectrum of cancer treatment centers across the U.S.The delay in cancer treatment initiation observed for PLWH and cancer during COVID-19 may reflect both patient and provider concerns about COVID-19 symptomatology and severity in the setting of existing immune dysfunction.Recent studies show that while PLWH had similar COVID-19 diagnosis rates as the general population, PLWH were more likely to be hospitalized and die from COVID-19, particularly those with elevated viral load.
T A B L E 2 cancer