Quality of Care in Veterans Affairs Health Care System In-Person and National TeleOncology Service-Delivered Care.

BACKGROUND
The Veterans Affairs Health Administration (VA) is a leader in the provision of telehealth-delivered care. All 10 million Veterans who are enrolled in VA healthcare are eligible to receive care via telehealth.1 Annually, the VA cares for approximately 50,000 incident and approximately 480,000 prevalent cancer cases (M. Kelley, personal communication, November 20, 2024). Overall, VA cancer care is consistent with clinical practice guidelines.3–8 Compared to those who receive care fragmented across both VA and private health care systems, health outcomes for Veterans treated entirely within the VA are often better.9,10
Some Veteran populations have historically had an increased likelihood of cancer care fragmentation. Approximately 38% of Veterans using the VA health care system reside in rural areas.11 Rural-dwelling Veterans with cancer have reduced access to specialized cancer care services. Veterans diagnosed with cancers that are not treated as frequently in the VA, such as certain gynecological cancers, may also have limited access to subspecialists who provide care close to a Veteran’s home. The VA may purchase care in a Veteran’s local community to facilitate timely access to care. However, with only 44% of rural U.S. communities having an oncologist, when the VA purchases care it still does not ensure that Veterans will receive needed care when they need it.12 To address this critical issue, the VA launched the National TeleOncology Service (NTO).1,13,14
NTO is one way that Veterans may receive care for their cancer. The NTO was initiated in 2020. The demand for NTO has been substantial. As of fiscal year 2021, approximately 13,300 unique Veterans used NTO services (V. Passero, personal communication,October 4, 2024).1 NTO has demonstrated that telehealth for oncology is acceptable to both patients and providers.15 Both Veterans and their NTO providers noted a relative advantage of NTO over in-person care is that it facilitates access to oncology specialists for Veterans who travel long distances. How the quality of VA telehealth-delivered cancer care compares to traditional in-person VA cancer care is unknown. Our objective was to understand quality of VA telehealth-deilvered cancer care for Veterans with cancer and how that compares to traditional in person care.

METHODS
Our analytic sample was constructed by conducting a series of data pulls from national electronic health record data via the VA’s Corporate Data Warehouse (CDW). We identified Veterans with an ICD-10 diagnostic code for any incident malignancy including colon cancer, lung cancer (small and non-small cell), and non-Hodgkin’s lympoma diagnosed from December 2016 to March 2021. The NTO is a hub and spoke model, where hub sites have specialized expertise. The hubs sites then work closely with designated spoke sites to provide needed care.1 We identified Veterans from three early adopting VA Medical Centers – Durham, NC; Omaha, NE; and Pittsburgh, PA and their associated spoke sites – that provided both cancer-related telehealth (TH) and traditional in-person (TR) care. Patients who did not have their primary or sole medical oncologist in the VA were excluded from this sample.
Once the sample was identified, two abstractors manually reviewed the VA’s national electronic health record (EHR) to verify telehealth use, demographic and clinical information (e.g., sex, cancer type) and to collect additional clinical and treatment characteristics (e.g., stage, treatment dates). Determination of telehealth use was a two-part process. First, using clinic stop codes, we initially categorized identified Veterans as being TH users or TR only users. Then, with this data available, the abstractors reviewed Veterans’ EHR to review, verify, and categorize Veterans as TH users if they had one or more primary or secondary clinic stop code associated with NTO in the previous year or telehealth visits were documented within the notes of the medical record. All chart abstracted data were entered into a Microsoft Forms database. Two primary abstractors (WR, DH) independently reviewed approximately 10% (n=118) of the same patient records. The study statistician (AJ), compared the abstractors’ responses for key variables and created a report detailing differences in reporting. Our goal was to achieve at least 95% agreement on 114 key variables. Upon initial abstraction, the two abstractors had 95% agreement on 46% of the patient records compared. The larger team met to review these differences, reconcile the reviewed records, and retrain on areas where there was poor concordance across abstractors.
Our abstraction team (WR, DH) used the EHR and manually collected data to calculate 25 measures based on the Quality Oncology Practice Initiative (QOPI).8,16 These QOPI measures were selected with input from the VA National Oncology Program and spanned multiple domains, specifically core measures (n=13), end of life and palliative care (n=10), and symptoms and toxicity management (n=2). We calculated measures consistent with the QOPI program of the American Society of Clinical Oncology (Appendix).17 We report the QOPI measures descriptively, overall and by telehealth user status. Differences in rates of at least 10% between the TR and TH groups were considered clinically significant. We also report t-tests and chi-squared tests to compare demographic characteristics and quality of care delivered by TH and TR. The Cochran Armitage test was used to test for trends over time in the prevalence of telehealth use. In addition to the main QOPI measures, we also calculated disease-specific measures for non-small cell lung cancer (n=10), small cell lung cancer (n=3), non-Hodgkin’s lymphoma (n=2), and colorectal cancer (n=13). Exact chi-square tests were performed on some of the cancer specific QOPI measures with small cell counts. All conducted analyses are reported; a p-value < 0.05 was considered statistically significant. We obtained institutional review board approval for this work.

RESULTS
We identified 972 Veterans with an incident malignancy during the study period. The population was predominately comprised of white (n=819, 85.8%) men (n=930, 95.7%) with a median age at the time of diagnosis of 70 (SD=8.9) years. The most identified cancers were non-small cell lung cancer (n=435, 45%) and colorectal cancer (n=200, 21%). Nearly half (n=427, 44%) of Veterans at the sampled NTO hub and affiliated spoke sites used telehealth (Table 1). Telehealth user vs non-telehealth users had similar mean age at the time of cancer diagnosis, ADI scores, sex, rurality, cancer type or hub site. There were statistically and clinically significant racial differences in telehealth use, with patients of non-white races to be lower telehealth users (n=373 of 819, 46% white; n=41 of 122, 33.6% black; n=8 of 13, 62% other race p=0.02). Over the years of diagnosis included in our study cohort, 2016 through 2021, there was a significant increase in telehealth use over time (36% TH use in 2016–2018, 47% in 2019, 65% in 2020–2021; p<0.0001; Figure 1).
Considering the quality of care provided between TH and TR care, overall, the TH outperformed TR for 12 measures, underperformed for 10 measures, and performed equally for two measures. There was one measure which was unevaluable for both groups because there was insufficient EHR data available. Only one QOPI measure showed both a statistically and clinically significant difference between TH and TR care. Appropriate tobacco use cessation support within the previous year was statistically higher in TH (85.3% vs 76.2% p=0.002).
Details of QOPI performances measures are reported in Figures 2, 3, and 4. In brief, there was nearly complete achievement of several core QOPI measures (Figure 2) for both TH and TR care including having a documented pathology report (TH 100%, TR 100%), having appropriate chemotherapy administered for Veterans with metastatic disease (TH 100%, TR 100%), having documentation of chemotherapy consent (TH 97%, TR 98%), and having pain addressed (TH 95%, TR 93%). There were modest, non-significant differences for several measures including documentation of advance directives (n=949 overall, TH 60%, TR 64%), documentation of a chemotherapy summary (n=368 overall, 83% TH, 75% TR), and documentation of an action plan to address a Veteran’s well-being (n=174 overall, 71% TH, 81% TR).
Performance for end of life and palliative care measures was comparable across TH and TR (Figure 2). Importantly, the two groups had similar rates of emergency department visits within 30 days of death (TH 43%, TR 48%), palliative care being offered (TH 83%, TR 86%), or enrollment in hospice (TH 85%, TR, 78%).
We assessed two measures in the domains of symptom and toxicity management (Figure 3). TH and TR performed similarly on both low and minimal risk (TH 24%, TR 16%) and medium and high risk of antiemetic therapy (TH 45%, TH 42%).
We also assessed disease-specific measures for non-Hodgkin’s lymphoma (n=135), non-small cell lung cancer (NSCLC; n=485), and small cell lung cancer (SCLC, n=72), and colorectal cancer (n=200). Due to small sample sizes, we present all results in the Appendix and present measures with sample sizes above ten in the text. For non-Hodgkin’s lymphoma, we assessed whether obinutuzumab, ofatumumab, or rituximab was administered when CD-20 antigen expression is negative or undocumented. A low number for use of CD-20 directed therapy indicates better quality care. All patients achieved this outcome (n=11; 100%) in both groups (TH n=4, TR n=7).
In the context of NSCLC, we assessed whether performance status was documented for patients with an initial AJCC stage IV or with distant metastatic disease. In our sample, 87% achieved this outcome (n=161 of 185) with no statistically significant difference between TH (92%, n=57 of 62) and TR (85%, n=104 of 123) users. We also assessed completion of molecular testing for patients with stage IV NSCLC with adenocarcinoma histology and found 84% achievement overall (n=92 of 110) and no difference between TH (87%, n=32 of 37) and TR (82%, n=60 of 73). We do not report findings from the SCLC cohort because all assessed indicators had a sample size fewer than ten Veterans.
In the context of colorectal cancer, the quality of cancer care was similar for TH and TR patients. Specifically, we assessed whether a carcinoembryonic antigen (CEA) was monitored within four months of curative surgical resection. Overall, approximately 83% (n=74 of 89) of patients achieved this outcome, with no statistically significant difference based on telehealth use (TH 84%, n=36 of 43; TR 83%, n=38 of 46). We also assessed whether adjuvant chemotherapy was recommended within four months of diagnosis for Veterans diagnosed with stage III colorectal cancer. Approximately 62% of patients overall achieved this measure (n=42 of 68), with similar rates between groups,TH (63%, n=17 of 27) and TR (61%, n=25 of 41). We assessed whether 12 or more lymph nodes were examined for resected colorectal cancer and determined that approximately 77% of patients achieved this outcome (n=61 of 79),. TH (81%, n=26 of 32) and TR (75%, n=35 of 47). Approximately 80% of patients overall (n=86 of 107), TH (n=41 of 51), and TR (n=45 of 56) achieved a colonoscopy before or within six months of curative colorectal resection or completed primary adjuvant chemotherapy. We assessed whether granulocyte-colony stimulating factor (GCSF) was administered to patients who received chemotherapy for metastatic colon cancer. For this measure, not achieving the outcome indicates better quality care. Overall, approximately 3% of patients achieved this outcome (n=1 of 38), TH (6%, n=1 of 18) and TR (0%, n=0 of 20). We assessed the percentage of colon cancer patients with PET or PET-CT ordered by practice after the completion of treatment with curative intent for colon cancer. Again, not achieving this outcome indicates better quality care for this measure. Approximately 9% (n=9 of 99) of veterans overall, 13% of TH (n=6 of 47) and 6% (n=3 of 52) achieved this measure. We also assessed microsatellite instability (MSI) status in colorectal cancer; achieving this outcome indicates better quality care for this measure. Sixty-two percent (n=114 of 184) of patients achieved this outcome, TH (55%, n=45 of 82) and TR (68%, n=69 or 102). Finally, we assessed use of transrectal ultrasound or pelvic MRI for patients diagnosed with non-metastatic rectal cancer. Slightly more than half (54%, n=25 of 46) of patients achieved this outcome, TH (63%, n=15 of 24) and TR (46%, n=10 of 22).
Upon identifying that there were few differences in care quality between TH and TR cancer care, we hypothesized that, even when a veteran receives care through TH, there could be differences in care quality between sites that have an oncologist on site (e.g, Altoona) and those that do not have an on-site oncologist available (e.g., Sioux Falls, Fayetteville, Clarksburg). We conducted a post-hoc sensitivity analysis to compare differences in care quality for all the evaluated QOPI core, end of life and symptom and toxicity management measures between sites with and without an on-site oncologist. We identified one statistically, but not clinically, significant core measure. Approximately 92% (n=726 of 788) of veterans from a site with an on-site oncologist, compared with approximately 88% (n=157 of 183) veterans from a site without an on-site oncologist had the veteran’s emotional well-being assessed by the second office visit (p=0.007). Approximately 46% (n=177 or 386) of patients from a site with an on-site oncologist received appropriate antiemetic therapy when receiving high or moderate-emetic risk antineoplastic agents, compared with 31% (n=27 or 84) veterans receiving care from a site without an on-site oncologist (p=0.02). Finally, the 13 end of life measures were similar between sites with and without an on-site oncologist in any of the 13 end of life measures (results not shown).

CONCLUSIONS
Telehealth has been used in the context of cancer care delivery in a variety of ways including for genetic counseling, behavior change interventions, symptom management, and palliative care.18 While the VA’s National TeleOncology (NTO) Service was conceived and implemented prior to the COVID-19 pandemic, the wider U.S. health care system’s interest in telehealth-delivered cancer was initially fueled by the COVID-19 pandemic. Telehealth use expanded across the U.S. nationally and has since stabilized as a complementary source of cancer care.19 Given the tremendous uptake of telehealth in oncology care settings, in early 2020, the National Cancer Institute took several steps to address meaningful evidence-based gaps in the use of telehealth across the cancer care continuum.18 The investments of both the VA and NCI will result in an enhanced understanding of the experiences, preferences, process, and outcomes associated with telehealth cancer care.
Our current work joins this new and growing body of literature addressing telehealth-delivered cancer care.14,18,19,22–24 In a cohort of early adopting telehealth sites, we discovered that the VA provides highly guideline-consistent cancer care. Our findings are aligned with existing literature regarding traditional, in-person care which demonstrates that the VA has a robust history of providing high quality cancer care, especially for Veterans receiving the bulk on their care within the VA health care system.3–8,10
Our study is among the first to consider quality of telehealth-delivered cancer care. A recent comparative effectiveness trial, conducted by Greer et al, of early palliative care delivered via telehealth versus in-person care for patients with advanced lung cancer found equivalent impacts of the two modes of care on patient-reported quality of life and that more caregivers attended telehealth vs in-person visits.24 While our study was conducted with a broader array of patients, our findings are consistent with those of Greer and colleagues in demonstrating equivalence of telehealth and traditional in-person care. Moreover, our study adds to the literature in several ways. First, our findings demonstrate similar quality for TH and TR-delivered cancer care. We also evaluated whether there were differences in care quality for people receiving care at sites with and without an on-site oncologist and found few statistically significant differences. To our knowledge the quality of care provided via TH and TR for patients living with cancer has not been evaluated with evidence- and consensus-based clinical practice guidelines. This finding is particularly relevant since telehealth use is on the rise across several clinical services. Specifically, in the U.S. nationally, telehealth use in oncology practices peaked at approximately 34% of health care visits in 2020 and stabilized at approximately 15%.23 In many areas of clinical care the U.S. is experiencing a consistent demand for telehealth-delivered care. There is evidence from our team’s prior work that Veterans and their cancer care team perceive relative advantages of telehealth delivered cancer care over traditional in-person care.15 This is important as telehealth-delivered cancer care has tremendous potential to bring cutting edge cancer care to people who might traditionally have limited access to subspecialists.
Beyond the question of quality, it is interesting to consider whether there are differences in resource use among clinic sites depending on whether there was an on-site oncologist. This is an important area for future research. For veterans receiving care at a clinic without a local oncologist and where the oncologist’s clinic was outside of a reasonable driving distance, all the cancer care was provided by an oncologist by telehealth either directly (by telehealth visits) or by collaboration with and supervision of an advanced practice provider. The telehealth oncologist’s direct care included an initial evaluation and treatment planning. Other components of care such as ordering treatment and testing, assessment of response and toxicity (i.e., pre-chemotherapy visits), and surveillance, could have been provided by either the telehealth oncologist or the on-site advanced practice provider; however, the oncologist would oversee care while veterans were undergoing treatment.
This study has several limitations. First, the timing of our analytic cohort overlaps with the COVID-19 pandemic, during which time there was increased demand for telehealth delivered cancer care and, for some services, decreased availability of traditional in-person care. We assert that this limitation is somewhat abated by the NTO pre-dating the pandemic and by VHA having a business model that is distinct from CMS reimbursement policies, which means that there was no financial incentive to favor one modality of care delivery over another. Second, for many measures our analyses were limited by small sample sizes. We have attempted to reduce this limitation by not presenting measures with a sample size below and by emphasizing clinically important differences.
Despite these limitations with our analysis, the VA is well positioned to be a pioneer in telehealth-delivered cancer care for several reasons. First, the national VA health care system has a robust history with providing telehealth care. Specifically, the VA is a high-volume provider of telehealth-delivered primary care and mental health care. The VA also has telehealth specific telehealth programs such as VA TeleSleep, TeleDermatology, National Telestroke Program, telehealth delivered genetic counseling, telehealth delivered smoking cessation, and Tele-ICU.25–32 The Tele-ICU program is a sterling example of VA’s telehealth capabilities; specialty providers are brought into intensive care unit rooms using video and can coordinate and communicate with other providers who are at a different facility providing hands-on care. The program reduced interhospital ICU transfers in meaningful ways.32 Second, the VA has its own financial model and is not subject to Centers for Medicare & Medicaid Services reimbursement policies. If the VA determines that telehealth-delivered cancer care is beneficial to Veterans with cancer, then it can establish and fund its own program design. Third, the VA is uniquely positioned to provide remote cancer care because its providers can practice care across state lines, which enables the VA to expand its reach to diverse and rural Veterans with cancer.33 The VA provides opportunities for clinicians providing remotely delivered care to communicate and coordinate care information through the national EHR, networking opportunities, and the VA’s national virtual tumor boards.34 This coordination and communication are essential for delivering high quality cancer care.
In our sample of diverse Veterans from early adopting NTO sites, we found few significant differences in which Veterans use telehealth for their cancer care. Future research should be conducted to examine potential racial differences in the reach of telehealth for cancer and to include additional years post-pandemic to determine whether these quality patterns are upheld as the NTO expands its service provision.

Supplementary Material
Appendix 1