Assessing Barriers and Facilitators to Lung Cancer Screening: Initial Findings from a Patient Navigation Intervention

Abstract

Low-dose computed tomography-based lung cancer screening represents a complex clinical undertaking that could require multiple referrals, appointments, and time-intensive procedures. These steps may pose difficulties and raise concerns among patients, particularly minority, under-, and uninsured populations. The authors implemented patient navigation to identify and address these challenges. They conducted a pragmatic randomized controlled trial of telephone-based navigation for lung cancer screening in an integrated, urban safety-net health care system. Following standardized protocols, bilingual (Spanish and English) navigators educated, motivated, and empowered patients to traverse the health system. Navigators made systematic contact with patients, recording standardized call characteristics in a study-specific database. Call type, duration, and content were recorded. Univariable and multivariable multinomial logistic regression was performed to investigate associations between call characteristics and reported barriers. Among 225 patients (mean age 63 years, 46% female, 70% racial/ethnic minority) assigned navigation, a total of 559 barriers to screening were identified during 806 telephone calls. The most common barrier categories were personal (46%), provider (30%), and practical (17%). System (6%) and psychosocial (1%) barriers were described by English-speaking patients, but not by Spanish-speaking patients. Over the course of the lung cancer screening process, provider-related barriers decreased 80% (P = 0.008). The authors conclude that patients undergoing lung cancer screening frequently report personal and health care provider-related barriers to successful participation. Barrier types may differ among patient populations and over the course of the screening process. Further understanding of these concerns may increase screening uptake and adherence. Clinical Trial Registration number: (NCT02758054).

Introduction

Despite its proven mortality reduction, it is estimated that fewer than 5% of eligible individuals in the United States undergo low-dose computed tomography (LDCT)-based lung cancer screening.¹ Depending on imaging findings, the screening process may require multiple referrals, appointments, and time-intensive procedures such as positron emission tomography/computer tomography (PET-CT) or biopsy. Although adherence to annual screening exceeded 90% in the National Lung Screening Trial (NLST), the trial population was disproportionately White, educated, former rather than current smokers, and received care in the tightly controlled environment of a clinical trial.^2,3

Rates of adherence in real-world settings are likely to be far lower, as non-White race, lower education level, and active smoking have been associated with lower adherence to other cancer screening modalities such as mammography and colonoscopy.^4,5

With age- and smoking history-related eligibility for lung cancer screening recently broadened to include more at-risk individuals (particularly those from under-represented minorities),^6–10 understanding potential barriers to uptake and completion is critically important to improving real-world effectiveness. Within an urban, integrated safety-net health care system lung cancer screening program, the authors implemented telephone-based navigation to identify and address patient concerns. Among vulnerable populations, patient navigation may improve cancer screening rates, follow-up adherence, and patient satisfaction.¹¹

Through an algorithm-driven approach, navigators offered instrumental (task-oriented or logistic) support to assist patients with appointment scheduling, reminders, expenses, transportation, and other access to care and adherence issues.¹² They also provided emotional support, addressed patient–provider communication, and focused on quality of life. Initial findings on the number, type, and correlates of screening barriers reported to the navigation program are presented here.

Methods

Setting

Parkland Health (Parkland) is the integrated safety-net health system for Dallas County, Texas, providing care for more than 1 million under- and uninsured county residents through a central, 982-bed tertiary care hospital, specialty clinics, and 12 community-based primary care clinics.^13,14 These clinical sites provide critical outreach for Dallas County, which is the 9th largest and one of the most ethnically diverse counties in the country (40% Hispanic, 34% non-Hispanic White, and 21% African American).¹⁵ Parkland has an enterprise-wide electronic medical record (EMR) system (EPIC, Verona, WI, USA) that allows electronic tracking of patient characteristics and outcomes.

Likely reflecting more substantial smoking histories, more advanced-stage diagnoses, and higher lung cancer-related mortality among racial minorities and lower socioeconomic status individuals, Parkland patients experience a disproportionate lung cancer burden and could receive maximal benefit from an effective lung screening protocol.^16–18

Lung cancer screening program

LDCT for lung cancer screening is performed in the American College of Radiology (ACR)-accredited outpatient radiology facility located on the main Parkland campus, which is located 22 miles (35.4 km) from the furthest Parkland community-based primary care clinic. An EPIC order set incorporating patient eligibility criteria, documentation of counseling and shared decision making, and triggering Lung-RADS result reporting was implemented in 2017. In general, the process is decentralized, with primary care and specialty clinicians ordering LDCT from across Parkland sites. Screening results are communicated to the referring physician.

Providers have the option of referring patients with radiographic abnormalities suspicious for lung cancer to a lung diagnostics clinic on the main Parkland campus staffed by pulmonary medicine physicians. Patients diagnosed with lung cancer are referred to thoracic surgery, radiation oncology, and/or medical oncology clinics. In recent years, LDCT for lung cancer screening has been ordered for more than 500 patients annually, of whom the majority are racial/ethnic minorities.^19,20

Screening and recruitment of participants

A pragmatic randomized controlled trial was conducted to test the impact of telephone-based patient navigation on completion of the lung cancer screening process in a county-integrated health system.²¹ The primary endpoint for the pragmatic trial was completion of the first 3 consecutive steps in the screening algorithm. Per protocol, for low suspicion CT scans, the endpoint would include 3 yearly chest CT scans over a 2-year period, whereas for indeterminate or high suspicion scans, the endpoint would include initial CT scan, short-interval repeat CT scan, PET, biopsy, and possibly treatment.²¹ Analyses are underway to be reported elsewhere (Bhalla et al, American Society for Clinical Oncology, June 2023).

Study recruitment, intervention delivery, and study data collection were conducted through a study-specific interactive research database incorporating weekly EMR data of patients who received a referral order (above) for LDCT screening as part of routine care within the Parkland system. Specifically, half of patients were randomized to receive navigation, with the other half receiving usual care (ie, the same referral process without navigation intervention). Bilingual study staff used the database interface to access patient participants' contact and appointment information, as well as to administer research telephone surveys and interviews.

Navigation intervention and data collection

Research assistants, bilingual in English and Spanish, with a background in community outreach and engagement, were trained using an abbreviated curriculum (∼20 hours) focused on motivational interviewing, adapted from George Washington University Cancer Institute.²² Navigators also received supplemental training on specific patient resources within the Parkland system and motivational techniques and behavioral aspects of lung cancer screening, including smoking cessation and modalities.

Navigators made systematic contact with patients through 4 types of scripted telephone calls, as follows: (1) initial intake, (2) reminder (2–3 days before LDCT or other screening-related appointment), (3) follow-up (3–7 days after scheduled appointment), and (4) outreach (triggered by subsequent screening or diagnostic orders) (Fig. 1). Research navigators did not access the patient EMR directly but carried out the navigation protocol using the interactive study database.

FIG. 1.

Navigation process and calls. *Timing and nature of subsequent steps (eg, LDCT, diagnostic CT, PET-CT, biopsy) depend on Lung-RADS result of preceding LDCT. LDCT, low-dose computed tomography; PET-CT, positron emission tomography/computed tomography.

Navigators tracked assigned patients and documented the first 3 participant-reported barriers per call. During these calls, they used motivational interviewing techniques^23,24 to ask about the patient's status in the care continuum and assess interest in smoking cessation, and to empower the patient to connect with Parkland system resources or direct further questions back to their primary care team. Based on the types of barriers reported in earlier studies in mammography and colorectal cancer screening programs in safety-net populations,^25,26 navigators categorized barriers as practical, psychosocial, provider, system, or personal. Barrier types were recorded according to call type in the study database. Here the authors describe and analyze barriers raised in patients' first year of the screening process.

Statistical analysis

Calls (intake, follow-up, outreach, and reminder) and barriers (practical, psychosocial, provider, system, and personal) were categorized according to type. Call duration was a continuous variable measured in minutes and seconds. In univariable analysis, the team evaluated the association between barrier type and call type, separately for English-speaking patients and for Spanish-speaking patients, by Fisher's exact test due to low cell counts (<5). In multivariable analysis, the team assessed the association between barrier type and call type by multinomial logistic regression, controlling for language and call duration.

Results

In total, 225 patients were assigned to receive navigation services. Mean age was 63 years, 46% were female, and 70% were racial/ethnic minorities. Additional demographic characteristics are listed in Table 1. Overall, 1851 calls were attempted, and 858 calls (46%) were completed. Median call duration was as follows: intake (18 minutes 12 seconds English, 18 minutes 04 seconds Spanish), reminder (5 minutes 00 second English, 6 minutes 02 seconds Spanish), follow-up (11 minutes 45 seconds English; 11 minutes 22 seconds Spanish), and outreach (12 minutes 48 seconds English, 20 minutes 47 seconds Spanish). Across all call types, Spanish language calls were ∼30% longer than English language calls.

Table 1.

Characteristics of 225 Patients Who Received Navigation Services for Low-Dose Computed Tomography-Based Lung Cancer Screening

Characteristic	Mean ± SD or count (%)
Age	63 ± 5
Gender
Female	104 (46)
Male	121 (54)
Race/ethnicity
Black	124 (55)
Hispanic	33 (15)
White (Non-Hispanic)	67 (30)
Other	1 (0)
Primary language
Spanish	15 (7)
English	210 (93)
Insurance
Charity/self-pay	204 (91)
Commercial/Medicare/Medicaid	21 (9)
Marital status
Married/common law	37 (16)
Single/divorced/widowed/other	188 (84)
Household income
<$35,000/year	174 (87)
≥$35,000/year	9 (4)
Not reported	42 (19)
Education
High school graduate or less	141 (71)
Posthigh school education	57 (29)

SD, standard deviation.

Navigators documented 617 reported barriers (Table 2). The most common categories were personal (48%), followed by provider (28%) and practical (17%). The types of barriers reported by patients varied according to the screening process step. For instance, specific examples of barriers according to category are given in Table 3.

Table 2.

Barrier Category According to Call Type

Call type	N	Barrier category count (%)					Total
		Personal	Psychosocial	Practical	Provider	System
Intake	193	91 (35.3)	2 (0.7)	30 (11.6)	115 (44.6)	20 (7.8)	258
Reminder	299	97 (63.4)	3 (1.9)	35 (22.9)	13 (8.5)	5 (3.3)	153
Follow-up	282	80 (62.5)	2 (1.6)	25 (19.5)	17 (13.3)	4 (3.1)	128
Outreach	84	27 (34.7)	3 (3.8)	13 (16.6)	28 (35.9)	7 (9.0)	78
Total	858	295	10	103	173	36	617

Table 3.

Selected Examples of Patient-Related Barriers by Category

Personal	Psychosocial	Practical	Provider	System
Patient received results but did not understand next steps	Patient had anxiety regarding LDCT procedure	Patient inability to pay for LDCT procedure or copay	Patient was unaware of tobacco cessation referral	Patient identified limited availability for appointments
Patient unaware of how to reschedule LDCT appointment	Patient was dealing with other pressing medical issues	Patient lack of reliable transportation to LDCT appointment	Patient was unaware of LDCT order	Patient identified concerns with lack of calls being returned
Patient lacked time to attend appointment	Patient was dealing with other family concerns	Patient inability to pay transportation or parking costs	Patient did not understand how to obtain LDCT results	Patient identified long wait times as a concern
Patient did not understand the purpose of a follow-up test	Patient expressed that the process of quitting smoking is too difficult	Patient lack of insurance to cover LDCT	Patient did not understand the purpose of the LDCT order	Patient identified provider being changed without approval
Patient did not understand how to interpret the LDCT results	Patient did not want to know of a cancer diagnosis	Patient insurance did not cover medication prescribed	Patient did not recall conversation with provider for LDCT order	Patient identified delays in receiving results from LDCT
Patient insurance did not cover medication prescribed	Patient did not want to know of a cancer diagnosis	Patient did not recall conversation with provider for LDCT order	Patient identified delays in receiving results from LDCT	Patient did not understand how to interpret the LDCT results

LDCT, low-dose computed tomography.

Patient-reported barriers varied by screening process step (Table 2). Overall, the number of barriers recorded decreased substantially throughout the process, with 258 barriers recorded during intake calls and 78 barriers recorded during outreach calls. Provider-related barriers (eg, failure to recall a clinician discussion about screening) accounted for 45% of all barriers noted during intake calls, but only 8% of barriers in reminder calls and 13% of barriers for follow-up calls. Conversely, representation of system-related barriers (eg, wait times, delays) was relatively stable throughout the navigation process, ranging 3%–9% across call types.

Personal barriers (eg, lack of transportation, challenges paying for transportation/parking) accounted for 48% of all reported barriers and decreased over the course of navigation calls. Psychosocial barriers, such as anxiety related to the screening test and/or anticipated results, accounted for only 2% of reported barriers and were the only barrier category that did not clearly decrease during the navigation process.

Reported barriers according to language spoken are displayed in Table 4 (English) and Table 5 (Spanish). For English-speaking patients, there was a statistically significant association between call type and barrier type (P < 0.001), with provider, system, and personal barriers reported most commonly during intake calls. For Spanish-speaking patients, the team observed a significant association between call type and barrier type (P = 0.04) with personal and provider barriers reported most commonly during intake calls. Notably, no psychosocial or system barriers were reported by Spanish-speaking patients. Two Spanish calls (13%) and 27 English calls (15%) had no recorded barriers.

Table 4.

English Language Calls

Call type	Barrier category count (%)					Total
	Personal	Psychosocial	Practical	Provider	System
Intake	79 (33.7)	2 (0.9)	28 (12.0)	105 (44.9)	20 (8.5)	234
Reminder	93 (64.6)	3 (2.1)	32 (22.2)	11 (7.6)	5 (3.5)	144
Follow-up	70 (61.4)	2 (1.8)	22 (19.3)	16 (14.0)	4 (3.5)	114
Outreach	25 (35.2)	3 (4.2)	13 (18.3)	23 (32.4)	7 (9.9)	71
Total	267	10	95	155	36	563

Fisher's exact test was used to assess the association between barrier category and call type (P < 0.001).

Table 5.

Spanish Language Calls

Call type	Barrier category count (%)					Total
	Personal	Psychosocial	Practical	Provider	System
Intake	12 (50.0)	0	2 (8.3)	10 (41.7)	0	24
Follow-up	10 (71.4)	0	3 (21.4)	1 (7.2)	0	14
Reminder	4 (44.5)	0	3 (33.3)	2 (22.2)	0	9
Outreach	2 (28.6)	0	0	5 (71.4)	0	7
Total	28	0	8	18	0	54

Fisher's exact test was used to assess the association between barrier category (practical/provider/personal) and call type (P = 0.004).

In multivariable analysis controlling for language and call duration, call type was significantly associated with barrier category. Taking intake calls as the referent group, the authors found that the odds of a patient identifying a provider barrier (vs. personal barrier) decreased by 85% (odds ratio [OR] 0.15, P < 0.001) over the course of the screening process. For outreach calls, the likelihood that a patient reported a provider barrier decreased by 8% (OR 0.92, P < 0.001) for outreach calls and by 88% (OR 0.12, P < 0.001) for reminder calls.

Discussion

Lung cancer screening represents a complex process. Unlike Pap smears for cervical cancer screening, LDCT cannot be performed in the outpatient office setting. Unlike mammography for breast cancer screening—for which the same radiologists interpreting the imaging studies may order and perform same-day subsequent evaluations such as ultrasonography or biopsy—diagnostic steps such as PET-CT or lung biopsy generally require LDCT result viewing and order placement by the clinician who ordered the LDCT, a process that may take several weeks.

Unlike colonoscopy for colorectal cancer screening, during which polypectomy may both diagnose and treat precancerous lesions, in lung cancer screening diagnostic steps are never therapeutic. Even if a patient undergoes surgical excisional biopsy for diagnosis, if cancer is confirmed, a completion lobectomy and lymph node dissection are generally recommended. Given these considerations, it is critical to study the real-world considerations of lung cancer screening, especially in minority and underserved populations. These individuals may face particular challenges in completing screening process steps, while at the same time facing greatest risk of and worst outcomes from a lung cancer diagnosis.^16,27,28

Therefore, to understand and address potential barriers to lung cancer screening, the team implemented a telephone-based navigation program within an integrated urban safety-net screening program. The navigated cohort was highly diverse, with 70% under-represented minorities. Furthermore, fewer than 20% were married, and fewer than one-third had received education beyond high school. These characteristics clearly distinguish this group from LDCT clinical trial populations and are known to be associated with lower likelihood of screening adherence.^29–31

Call completion rate while low (46%) was comparable with routine telephone reminders made in the course of routine care in this safety-net population. Indeed, the team sought to mediate limited access due to lack of insurance—a primary social determinant of health—by testing protocolized navigation in an integrated system that provides financial assistance for all county residents.

Barriers to successful completion of lung cancer screening steps were reported throughout the screening process, with the greatest number identified initially and then decreasing thereafter. At first contact—which occurred after LDCT order placement but before the imaging study was performed—the most commonly reported barriers reflected communication gaps. These included no recalled discussion of the purpose of or plans for lung cancer screening. Whether this perception reflects an actual lack of discussion or failure to understand or remember a discussion that occurred is not known.

Regardless, one can envision a number of reasons for this observation. Over time, primary care physicians have had less time to address more requirements during clinic visits.^32,33 Given the age and smoking history of LDCT-eligible individuals, clinicians are likely to have numerous other matters to discuss as well, including other recommended screening procedures and management of comorbidities.¹⁹ Furthermore, in this study, most patient participants had limited education and were not married.

The former characteristic has been associated with decreased understanding of and participation in medical care, whereas the latter could imply a greater likelihood of attending medical appointments alone, which in turn may decrease recollection of provider discussions.^34,35 Recent studies have reported inconsistent quality of provider–patient shared decision-making discussions, noting specifically that non-White and less educated patients scored lower on knowledge, but also that most patients did not report discussion of potential harms rather than potential benefits.^36–38 This study's findings among safety-net patients highlight the need for further work to optimize how LDCT-eligible patients engaged in screening receipt and follow-up

At subsequent points in the screening process, practical and personal concerns reflecting broader social determinants of health accounted for most reported barriers. These included low economic status, for example, difficulty paying, finding time, or securing transportation for scheduled procedures, as well as health literacy, for example, not understanding LDCT results or the details of next steps. Once a subsequent screening step was ordered, psychosocial concerns such as anxiety regarding test results or the need to detail with more pressing medical issues became more common. However, in general, psychosocial concerns represented under 5% of reported barriers, a smaller proportion than anticipated based on reports from other cancer screening modalities.^25,26

The team observed some differences in reported barriers according to language spoken; although the number of Spanish-speaking participants was very small, these differences should be considered exploratory. Spanish-speaking individuals were less likely to report barriers. Specifically, they did not report psychosocial barriers at any time point. Nor did they report any system-related barriers (eg, long wait times for LDCT, difficulty contacting the clinical facility). The authors consider these potential differences in light of cultural influences on communication.

One potential explanation for these findings is strong family and community support, which underlie the so-termed Hispanic paradox³⁹ and might help patients complete steps in the screening process. Alternatively, these trends could reflect differences in expectations or differences in barrier reporting or recording rather than true differences in experience. However, Spanish-language calls averaged about one-third longer than English-language conversations, and our navigators were fluent in Spanish and had passed a stringent proficiency examination before starting.

Interestingly, Spanish-language outreach calls were particularly lengthy, averaging almost twice as long as English-language outreach calls, even though fewer barrier categories were recorded for the Spanish calls. Parkland Health has a long-standing commitment to serving the large number of county residents that prefer to receive services in Spanish. As part of sustainability plans for their standard-of-care services, the system has incorporated resources for bilingual navigation capacity in the lung cancer screening program.

Although LDCT as a procedure is distinct from other cancer screening modalities, this study's findings echo results from studies in these other processes. In a survey of ∼9000 underserved women referred for mammography, the most common concerns were cost, procedure-related pain, and the potential for bad news.²⁵ For colorectal cancer screening, patients report anxiety about the procedure or outcome, financial challenges, and logistic challenges including transportation and time.²⁶ Similarly, reported barriers to cervical cancer screening include structural barriers (cost, transportation, lack of time), fear of finding cancer, and anticipation of pain.^40,41

Limitations of this study include a single-center setting and an insufficient number of Spanish-speaking patients to power detailed analyses of this population. In addition, because navigators did not document other patient characteristics with specific calls, the authors cannot determine the association between reported barriers, other demographic characteristics, or completion of screening steps. Study strengths include a navigation database allowing capture and coding of call details, the assessment of barriers longitudinally, and a highly diverse patient population.

Furthermore, the limited education and lower socioeconomic status of the patient population allowed the team to study those individuals who may face the greatest challenges to accessing screening, prevention, and treatment services.^18,28 At the same time, they may have greater risk for and worse outcomes from lung cancer.⁴²

In conclusion, patients report numerous and diverse barriers throughout the lung cancer screening process. Early on, these center around gaps in communication with providers. Later, patients became more concerned with their own ability to adhere to screening steps. Navigation might address these and other concerns by offering an opportunity to revisit and clarify earlier conversations, as well as provide a setting to empower and problem solve.

Authors' Contributions

Prof. Lee contributed to conceptualization, funding acquisition, investigation, methodology, resources, supervision, roles/writing—original draft, and roles/writing—review and editing. Ms. Jessica Lee was involved in project administration, supervision, roles/writing—original draft, and roles/writing—review and editing. Prof. Zhu carried out formal analysis, investigation, methodology, supervision, and roles/writing—review and editing. Prof. Chen and Ms Patel were involved in data curation, formal analysis, investigation, and roles/writing—review and editing. Ms Wahid took charge of data curation, formal analysis, investigation, roles/writing—original draft, and roles/writing—review and editing.

Prof. Hamann was in charge of conceptualization, funding acquisition, investigation, methodology, supervision, and roles/writing—review and editing. Dr. Bhalla, Dr. Cardenas, Dr. Johnson, and Dr. Natchimuthu were in charge of investigation and roles/writing—review and editing. Dr. Santini took care of project administration, supervision, and roles/writing—review and editing. Dr. Gerber carried out conceptualization, funding acquisition, investigation, resources, supervision, roles/writing—original draft, and roles/writing—review and editing.

Human Subjects Approval

Protocol was approved by the UT Southwestern Medical Center Institutional Review Board (STU#122015-046) and by the Parkland Office of Research.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Research was supported by the Cancer Prevention Research Institute of Texas (CPRIT; RP160030 to Prof. Hamann, Prof. Lee, and Dr. Gerber; and PP190052 to Dr. Gerber and Prof. Lee). Additional support was received through the Harold C. Simmons Comprehensive Cancer Center (5P30 CA142543), the UT Southwestern Center for Patient-Centered Outcomes Research (R24 HS022418), and a Mary Kay Ash Foundation International Postdoctoral Scholarship (to Dr. Cardenas).