Factors associated with adherence to provider referrals for lung cancer screening with low dose computed tomography before and during COVID-19 pandemic

To the best of our knowledge, this is the first study to explore the comprehensive picture of challenges and supports surrounding LDCT-LCS from a multilevel, multifactor perspective in a large community healthcare system in the context of the COVID-19 pandemic. We found that those who received orders in Family Medicine or other specialties, had no major comorbidities, smoked at least 30 pack-years, and saw providers having more experience in LDCT-LCS ordering were more likely to complete an order, highlighting opportunities to improve LDCT-LCS processes and influence clinical practice at patient, provider, and systems levels.

In our data, orders for LDCT-LCS increased since the USPSTF recommendations, decreased substantially at the start of COVID-19 pandemic, then slowly went up, which is in correspondence with the ACS recommendations in April 2020 to postpone cancer screenings in order to prioritize urgent medical needs and reduce the risk of the spread of COVID-19 in health care settings [28]. However, while we found that the absolute number of LDCT-LCS orders declined dramatically at the beginning of the COVID-19 pandemic in 2020, the completion rate only slightly declined and remained at 50–60% throughout 2020. The patients receiving LDCT-LCS orders in 2019 and early 2020 might have postponed the screening test due to the COVID-19 pandemic, but a majority still completed LDCT-LCS within a year. Lang et al. found a similar trend in which institutional LDCT-LCS volume based on lung imaging data significantly decreased during the COVID-19 pandemic, followed by complete recovery of follow-up LDCT-LCS volume and more gradual recovery of annual and baseline LCS LDCT volume [29]. While many patients who did receive LDCT-LCS screening did complete it, there are still a large number who postponed their screenings. Healthcare systems should similarly examine surveillance data for cancer screening orders and completion rates to identify areas to enhance public communications, inform policy changes, and guide interventions as the cancer screenings that were missed or delayed because of pandemic could potentially lead to more patients with more advanced stages at diagnosis, poorer survival, and greater disease-related and treatment-related morbidity.

It is worth noting that before 2019, LCS-ineligible patients and patients with unknown eligibility received referrals for LDCT-LCS no less frequently than LCS-eligible patients, which may be the result of poor documentation of smoking history or that many clinicians and patients might not have been fully aware of the eligibility criteria for LDCT-LCS [30‐32]. Unlike other types of cancer screening that recommend universal screening in healthy populations of a certain age, LDCT-LCS targets older people who have smoked heavily long-term, determined using age, pack-years, and current smoking status. Likely exacerbating this at the provider level, identification of these LCS-eligible patients is hindered by the lack of complete and accurate information about smoking history at the point of care [33]. In our data, those with unknown pack years or quit years were separated as the unknown eligibility group, which enables us to reveal the issue of poor documentation of the smoking history among patients with LDCT-LCS orders. Recently in March 2021 USPSTF expanded screening eligibility for individuals at lower absolute risk for lung cancer by lowering screening start age to 50 years and smoking history to 20 pack-years. This change has increased the relative percentage of individuals eligible for screening compared with the 2013 criteria [34], and now includes more racial ethnic minorities and women [35]. With this expansion, millions more people who smoke will become eligible for the LDCT-LCS and the orders for LDCT-LCS among ineligible patients may likely further decline. If LCS-ineligible patients and patients with unknown eligibility continue to be referred for screening under the expanded guidelines, it may warrant further examination to understand whether this is driven by patients or providers and the reasons behind these referrals.

Among patients aged 55 to 80 years who received LDCT orders, the completion rates of eligible patients increased from 0% to 2013 to the highest monthly rate of 70% in 2019, indicating increased lung cancer screening awareness and acceptance in the 5 years since the USPSTF recommendations. Furthermore, LCS-eligible patients had a higher completion rate than the LCS-ineligible patients. Changes in organization-level workflows that may have played an important role in this include a Sutter-wide EPIC system change (“Smartset”) in April 2018, the Best Practice Advisory (BPA) for LDCT-LCS in place in April 2019, and implementation of a Navigator Program for LDCT-LCS in October 2019. The EHR-based LDCT-LCS multidisciplinary management system (“Smartset”) includes “structured tobacco history” allowing for identification of eligible patients, “order workflow” facilitating LCS orders for eligible patients during office visits, and “tracking workflow” enabling Lung Cancer Care Coordinators/Navigators to track qualifying patients through reporting worklists. These system-level changes may provide more coordinated, integrated care that leads to informed decision-making and improved patient adherence.

While we found that our completion rate among patients who had their first LDCT order in 2016–2020 was consistent with the adherence rates from previous studies [36, 37], we also found substantial variation across providers in completion of LDCT-LCS orders. When it comes to completion of LDCT-LCS orders, each clinic has their own processes, systems, and workflows, and there is no unified way to schedule, track, or ensure patients complete screenings or follow-ups. In our study, higher completion rates were found for more experienced providers who ordered more than the average number of LDCT-LCS; however, even after taking into account provider sex, profession, and prior experience, we observed large, unexplained variation across providers, suggesting the need for system-wide efforts to standardize LDCT-LCS practice [38].

With under a third of the variability in the LDCT completion rate accounted for by providers, this leaves a majority of the variability coming from patient factors or other unknown factors. At the patient level, we found that older age (78–80 years old), self-identifying as Black, having severe major comorbidities, having smoked fewer than 30 pack-years, being a new patient, and receiving LDCT-LCS orders in earlier years (i.e., 2016, 2017) or during the COVID-19 pandemic (i.e.,2020) were negative predictors of completing LDCT-LCS orders. The maximum age limit of 77 set by CMS for reimbursement of annual LDCT for lung cancer screening [39] may in part explain the lower completion rates among the oldest age group (78–80 years old), as well as the decreasing relative benefit of screening and early detection with increasing age and severity of comorbidities. Similarly, the previous eligibility criteria of at least 30 pack-years [39, 40] may explain the lower completion rates among those who smoked fewer than 30 pack-years and thus were considered ineligible for LDCT-LCS at the time of this study. Recently expanded USPSTF screening criteria are expected to improve completion rates among those who smoked between 20 and 29 pack-years and enable earlier detection of lung cancer and improved survival outcomes [41]. Consistent with a previous study examining the racial difference in lung cancer screening outcomes [36], we also found that Black patients had significantly lower odds of receiving LDCT compared to White patients even after controlling for other covariates. The changes in 2021 screening guidelines have been associated with a very small difference in lung cancer screening eligibility between Black and White individuals of − 12.7 percentage points in 2013 and − 12.2 percentage points in 2021 [42]. This finding along with the findings of our study suggest that accounting for factors beyond lowering the recommended ages and pack-years for LDCT-LCS is likely needed to significantly decrease disparities in access to lung cancer screening and treatment among racial and ethnic minority groups.

Additionally, we found that receiving the LDCT-LCS order in Family Medicine or other specialties (as opposed to the internal medicine department) and from physicians (as opposed to other providers) were positive predictors of completing LDCT-LCS orders. This is consistent with the findings of a recent study on a Safety-Net Medical System which also found that family practice providers ordered more LDCT-LCS than did other clinicians [38]. Further research is warranted to investigate the high completion rates for LDCT-LCS orders received in Family Medicine and other specialties. While internists typically diagnose and treat medical problems of greater complexity than family practitioners in both the office and hospital settings, family practitioners typically provide preventive medicine and more “well-patient” services in the office setting [43]. This likely results in a difference in provider’s capacity to provide support services following a LDCT-LCS order and continuity in the sense of overall responsibility to ensure patient’s completion of the order.

Our patient-level findings suggest that certain groups of patients – such as those who are older, racial ethnic minority, new to the system, have severe major comorbidities and receive LDCT-LCS order during COVID-19 pandemic may face additional barriers found in other studies like inconvenience, perceived smoking-related stigma, and distrust of the healthcare system which may lead to decreased LCS participation [44]. Health systems should consider focusing on improving these areas as they may potentially increase the likelihood that LDCT-LCS counseling and follow up will take place.

Several limitations to this study need to be acknowledged. We relied on structured data from billing, procedure, ordering, and administrative records, but many important constructs remain unexamined. For example, we lack information on specific factors influencing patient and provider decision-making including SDM weighing pros and cons and patients’ preferences, which may be in part available in unstructured, free-text EHR notes. Similarly, since the organization is not a health maintenance organization (HMO), it is difficult to know whether a patient completed the LDCT-LCS order outside the system. To address this limitation, we selected patients who had one or more office visits within one year after the LDCT order. If they had later office visits at Sutter Health, the care they received elsewhere prior to the visits will be available in our EHR system by the Care Everywhere function. Furthermore, using data from a single healthcare organization with a highly-insured patient population may limit the generalizability of our findings. However, Sutter’s population is about 10% of the state of California [45]. It reflects the diversity that makes up the state and includes low-income, rural, non-English speaking, and minority patients. One benefit of studying a single system is that its shared “infrastructure” allows us to control for access barriers and instead focus on variation across patients and providers in completion of LDCT-LCS orders within the healthcare system.