Lung cancer screening in rural primary care practices in Colorado: time for a more team-based approach?

Abstract

Background

Despite lung cancer being a leading cause of death in the United States and lung cancer screening (LCS) being a recommended service, many patients eligible for screening do not receive it. Research is needed to understand the challenges with implementing LCS in different settings. This study investigated multiple practice members and patient perspectives impacting rural primary care practices related to LCS uptake by eligible patients.

Methods

This qualitative study involved primary care practice members in multiple roles (clinicians n = 9, clinical staff n = 12 and administrators n = 5) and their patients (n = 19) from 9 practices including federally qualified and rural health centers (n = 3), health system owned (n = 4) and private practices (n = 2). Interviews were conducted regarding the importance of and ability to complete the steps that may result in a patient receiving LCS. Data were analyzed using a thematic analysis with immersion crystallization then organized using the RE-AIM implementation science framework to illuminate and organize implementation issues.

Results

Although all groups endorsed the importance of LCS, all also struggled with implementation challenges. Since assessing smoking history is part of the process to identify eligibility for LCS, we asked about these processes. We found that smoking assessment and assistance (including referral to services) were routine in the practices, but other steps in the LCS portion of determining eligibility and offering LCS were not. Lack of knowledge about screening and coverage, patient stigma, and resistance and practical considerations such as distance to LCS testing facilities complicated completion of LCS compared to screening for other types of cancer.

Conclusions

Limited uptake of LCS results from a range of multiple interacting factors that cumulatively affect consistency and quality of implementation at the practice level. Future research should consider team-based approaches to conduct of LCS eligibility and shared decision making.

Background

Lung cancer is the leading cause of cancer-related death in both men and women in the United States (U.S.), averaging around 150,000 deaths per year [1,2,3]. The American Cancer Society estimated the number of new cases in 2020 was 228,820 and estimated the number of deaths at 135,720 [4]. Rural areas are disproportionately affected by smoking and incidences of lung cancer in comparison to urban areas [5]. Americans living in rural areas are more likely to die from lung cancer than their urban counterparts with a 20% higher lung cancer mortality rate [6]. The U.S. Preventive Services Task Force (USPSTF) recommends patients ages 50–80 with a 20 pack-years smoking history who currently smoke or quit within the past 15 years receive low-dose computed tomography (LDCT) lung cancer screening (LCS) annually [7, 8].

In the United States, primary care practices have not widely implemented these guidelines, especially in comparison to other evidence-based screening guidelines for other cancers [9, 10]. Of the LCS eligible Americans, only 5% to 6% have been screened as of 2020 [11, 12]. While utilization varies significantly across states, people who are uninsured are less likely to undergo LCS [13]. LCS is higher among those with chronic respiratory conditions, who were divorced, separated, or widowed, who had previous cancer diagnoses, and aged 65 to 74 [13].

Various barriers to LCS exist for rural patients, clinicians, and health care facilities [14]. Rural patients who are most in need of LCS tend to have lower levels of formal education, are less trusting of doctors and health care, have inadequate insurance coverage, and face geographic access barriers [15, 16]. Rural patients also are more likely to have limited access to primary care physicians who address the LCS recommendation and referral, and to specialty care [3, 10]. LCS via LDCT in the U.S. is a covered test by Medicare and some Medicaid programs (CMS); however, Medicaid in many states does not cover it, and some state Medicaid programs do not cover it as well as some private insurances in the U.S [17]. Around 50% of those eligible for LDCT per the recommendations of USPSTF are uninsured or Medicaid insured [18, 19].

Compared with clinicians of other specialties, primary care clinicians serve an important role of ensuring appropriate screenings are recommended to eligible patients [20]. Studies of primary care clinicians have reported barriers including lack of knowledge of the current guidelines (e.g., incorrectly ordering chest X-rays over LDCT scans), implicit biases based on sex, race, ethnicity, and smoking history that hinder recommendations for LCS to patients, and lack of time, shared decision-making tools, or ability to facilitate an effective conversation on LCS [10, 21]. Disparities in access to both information and screening facilities is more severe for rural than urban patients [10].

Despite all that is known about issues related to LCS, we still need to understand more about how perspectives of different key players (e.g., physicians, other staff, patients) may influence different aspects of implementation as it relates to the context of rural primary care. With many processes in primary care, especially around cancer screenings, clinical and administrative staff play a key role in identifying eligible patients, supporting the clinician in shared decision making about the screening choice, and coordinating screening activities with providers of services and reimbursement with payers. Thus, their perspective is an important one to capture. Likewise, understanding the patient’s perspective adds insight on why patients may or may not agree to the LDCT or follow through with test completion.

We have previously used the RE-AIM framework and more recently its contextual expansion to PRISM in our implementation research [22] and determined that it might provide unique insights into different aspects of implementation, especially across the various roles of important members in the system. RE-AIM is an acronym for reach and effectiveness at the patient level, and adoption, implementation and maintenance at the setting level [23, 24]. Examining multiple key players’ views through a RE-AIM lens might reveal more about how LCS eligibility and referral for services can be conducted in rural primary care. Thus, the purpose of this paper is to understand issues related to performance of LCS activities in rural primary care practices in Colorado. We assessed the conduct of various aspects of LCS including patient identification, tracking, shared decision making, use of decision aids, smoking cessation advice and counseling, referral and follow-up from the perspectives of clinicians, practice staff, and patients and using the RE-AIM model to elucidate these issues in the hopes of overcoming potential barriers to LCS implementation in practice.

Methods

We conducted this qualitative study as part of the Colorado Implementation Science Center for Cancer Control (https://coisc3.org); the study was approved by the Colorado Multiple Institutional Review Board (COMIRB) for research with human subjects (COMIRB #: 19–1706; date: April 27, 2020). We used methods in accordance with relevant guidelines and regulations. Recruitment occurred June – November, 2020; interviews occurred July – December, 2020. Analysis occurred throughout the interview period and ended June, 2021.

Participants and recruitment

The goal was to elicit viewpoints from multiple important partners; thus, we asked to participate clinicians (physicians, nurse practitioners, physician assistants), clinical staff (nurses, care managers, medical assistants), and administrative staff (practice managers, front desk staff), generally three to four participants per practice. Additionally, practices recruited three to five patients who met specific inclusion criteria. Patients were to have currently or previously smoked cigarettes and be 50–80 years of age (as part of the qualification for LCS eligibility).

To recruit practices, study staff worked with the directors and staff of the State Networks of Ambulatory Practices and Partners (SNOCAP) and their member practice-based research networks (PBRNs): High Plains Research Network (HPRN), Colorado Research Network (CaReNet) and Partners Engaged in Achieving Change in Health Network (PEACHnet). In addition to rural location, purposeful selection also included a mix of ownership (federally qualified health centers versus privately owned versus health system owned), geographic location (across Colorado, U.S.) and practice size. Rural location was defined as located in a county with a rural or frontier designation or providing care for a significant number of patients residing in rural areas. We contacted 28 practices, and nine participated; many declined due to stressors from the COVID-19 pandemic but more specific information about declines was not available. All participants provided verbal informed consent per the protocol approved by the institutional review board. Recruitment continued with practices while data were analyzed until sufficient thematic saturation was achieved.

Instruments and data collection

Interviews were conducted using a semi-structured interview guide developed by the research team to explore current practices for LCS in this setting [25]. The interview contained two parts: 1) exploration of participants’ values in a general sense – what is important to them and what brings meaning to their lives and work (reported elsewhere [26]), and 2) how the practice approached and conducted LCS and smoking cessation, including values and priority assigned to different types of cancer screening (this paper). The goal was to explore general values as well as values and importance as applied to a specific health care delivery topic (which was LCS). This second portion contained questions to explore if and how clinicians and practices currently assess smoking status, provide assistance with smoking cessation, understand LCS screening guidelines, conduct shared decision making for the LCS decision, and refer and coordinate LCS for eligible patients. We included smoking use and status because it is part of the recommendations for LCS. Interviewers used depth questioning to clarify details of the process including who did what, when, for what patients.

We also used a pre-developed work process flow diagram (see Fig. 1) to depict how LCS eligibility, screening, decision making, referral and treatment coordination (if needed), and follow-up might happen. Practice participants commented on their processes in relation to this diagram, if they did/did not do certain steps, how they ordered LDCT, and what influenced their processes. In an online shared document interviewers annotated the flow diagram (i.e., made changes in real time to the document as well as comments).

Fig. 1

Lung cancer screening clinical work flow diagram

For the patient interviews, a flow diagram was not used because they were not privy to the practice’s processes for this aspect of care. Patients were asked if they had participated in or had been recommended to have LCS, if they had been asked about their smoking status and were encouraged to quit smoking, and if they received assistance with smoking cessation. The interviewers explored patients’ perspectives on these topics including importance, confidence, and barriers associated with participation.

Interviews lasted approximately one hour each and were conducted either by one or both of the qualitative analysts (R Gomes or JSH). Interviewers took extensive notes during the interviews and completed a summary immediately following the interview. Due to the COVID-19 pandemic, all interviews were done either by video or telephone. Interviews were recorded and professionally transcribed verbatim. Each participant was compensated with a $100 gift card for completion of the interview.

Data analysis

Two qualitative analysts (R Gomes and JSH) coded and analyzed the data. ATLAS.ti version 8 (ATLAS.ti GmbH, Berlin, Germany) was utilized for data management and coding purposes. No specific qualitative philosophical frameworks were used and instead followed a thematic analysis perspective. In general, we used an immersion crystallization approach to examine the data across multiple passes and from multiple perspectives to triangulate across the researchers completing the work, the question/code categories, the respondent roles and the key features of the responses [27]. In addition, we developed and used a manual code book to code the 49 transcripts. Quotation reports were created and reviewed in discussion to identify key themes and illustrate process descriptions. Coding and analysis was inductive except for the application of the RE-AIM dimensions application to the data.

Part of the coding was driven by the RE-AIM model to capture how the LCS discussions described processes likely to influence the reach and effectiveness of their efforts as well as the degree of adoption, implementation, and maintenance at the practice level. Table 1 describes the RE-AIM dimensions and how they were defined for this study. As they reviewed the quotation reports, the analysts created a summary table to identify how different participant groups reported on the RE-AIM dimensions for the process of LCS identification and facilitation. Additionally, the analysts reviewed the annotated workflow process diagrams to glean any additional insights not found in the transcripts to add to the understanding of process elements informing RE-AIM outcomes. For purposes of this analysis, discussion of smoking status, recommendations to quit smoking, and smoking cessation were considered both part of the LCS primary care process, as well as a specific set process for assistance with stopping tobacco use. Thus, we included in the analysis pertinent statements relating to smoking and possible cessation processes. The analysts iteratively shared results with the larger research team for review and consultation. We also reviewed the relevant literature in consideration of the findings to corroborate themes as well as to consider the data from multiple perspectives.

Table 1 RE-AIM dimensions and definitions for this study

Results

Table 2 provides the descriptive characteristics of the nine practices that participated. Thirty-two patients were recruited and 23 completed an interview. Most were female (n = 19; 82.6%), all were White race (predominant in rural Colorado), and some reported Hispanic ethnicity (n = 4; 17.3%). Patients ranged in age from 53–74 (mean of 64.3) years. About half currently smoked (n = 13; 56.6%) with the rest reported a history of smoking. Among individuals who previously smoked (n = 10), about half had quit within the past year (n = 6; 26%). About half (n = 12; 51.7%) recalled being told about LCS (with about half of those deciding to undergo LCS). All were eligible for LCS based on recruitment guidelines for LCS age and smoking history eligibility.

Table 2 Practice characteristics

Thematic results by role groups and RE-AIM dimensions

In addition to examining thematic results overall, we inspected responses by respondent group. Table 3 outlines the major thematic elements by each RE-AIM dimension. Concordance and discordance across roles is highlighted. Table 4 includes salient quotations from participants that highlight main themes from the different groups.

Table 3 Concordance and discordance of perspectives on Lung Cancer Screening (LCS)^a by Role: Practice Member (Clinicians, Clinical Staff, Administrators) and patient in rural primary care using RE-AIM Dimensions

Table 4 Illustrative quotations demonstrating themes by RE-AIM domain

Reach

Two main factors affected reach. One is the offering of LCS to patients, and the second is patients’ decision to get LCS and then complete the LDCT. We found that although the risk assessment and smoking cessation aspects of the flow diagram (Fig. 1, columns 1 and 3) were reported as happening routinely, the screening eligibility and shared decision making (columns 2 and 4) were not. Considering the rows that depict which team member conducts the activities, it is the clinicians who have often not yet routinized these actions, suggesting that smoking cessation processes are established protocols for staff whereas the LCS components are not. All groups reported that LCS was offered less consistently to patients than smoking status assessment and cessation assistance. Many respondents spoke to the desire to establish systematic processes for LCS eligibility such as requesting EMR prompts, reminders and templates such that the clinican could be ready engage in shared decision making with the patient. For the second factor (patient agreement and follow-through), represented in the actions needed in columns 4 and 5, (shared decision making and screening respectively in Fig. 1), all parties identified issues such as perceived lack of insurance coverage, hassles experienced by patients, and resistance to screening by some patients. The patients provided the most robust explanations of their resistance and the factors that influenced their unwillingness to have LCS or to quit smoking.

Effectiveness

For clinicians and clinical staff, LCS is consistently stated as very important, and as important as other cancers or health issues; however, these roles also reported having few patients with lung cancer, which made it appear less relevant than other cancers. In contrast, all types of practice members types emphasized the importance of smoking cessation for preventing lung cancer. Their frustration was in finding patients interested in quitting smoking and successfully encouraging those patients to consider it. The few patients who had completed LCS thought it was important and effective in preventing death from LCS; however, many others had not completed LCS and stated reasons for skepticism such as fatalism, stigma, and money making by health care entities, as well as just not wanting the hassle of another medical intervention. Some patients expressed that they should quit smoking (indicating effectiveness in improving health), but doubted they could or ever would.

Adoption

Factors affecting adoption are those that influence whether the clinicians and clinical team offer smoking cessation and LCS. Smoking cessation was generally offered routinely as reported by most practice members of all types. There was less consistency across groups on LCS because only the clinicians knew how to conduct this process; other groups did not know about what happens or why. This highlighted that while screening for smoking was a routinized process for staff, asking about eligibility for or interest in LCS was not. It was largely the role of the clinician to determine eligibility and offer LCS. There was expressed openness to some parts of the process being systematized by staff. Clinicians relayed that they completed the discussion utilizing principles of shared decision making, but when we asked about a typical conversation, they did not describe processes of shared decision making, and none used any shared decision-making tools or aids. Patients both did and did not recall being offered smoking cessation or LCS. The clinicians shared how patient response could make them feel less willing to offer LCS – when patients demonstrated difficulty or resistance, the clinicians were less likely to want to discuss LCS with future patients as it was perceived as a hassle. The benefit/burden of spending their time with LCS shared decision making was not always perceived as worthwhile (“a lot of burden”).

Implementation

For implementation, we sought to understand factors influencing what would make LCS or smoking cessation go better or worse. In general, there were more implementation challenges with shared decision making with LCS than for smoking cessation, which has been performed routinely for much longer, resulting in established and effective workflows. Patients tended to focus more on how the clinician or clinical team brought up smoking and whether they communicated in a way that made the patients want to quit smoking or get screened (for LCS), rather than on the process. However, clinical team members focused more on how they could institute processes to make asking and offering help more consistent, although some clinicians did also acknowledge that how the subject was brought up can affect patient receptiveness. All practice member groups endorsed that there were more problems with implementing LCS than smoking cessation because eligibility, understanding reimbursement or insurance coverage, referral coordination, and ensuring LDCT was performed is more difficult than smoking cessation. In summary, many of the steps necessary to conduct shared decision making for LCS were complicated and contingent on previous steps as well as largely reliant upon the clinician to remember to do them.

Maintenance

Maintenance in terms of helping patients to maintain tobacco cessation and continue to get annual LCS was discussed less often. There were also clearly implementation challenges with maintaining annual shared decision making for LCS at the practice level as comments were made about needing reminder systems for annual discussions and follow-up to check if patients did get the LCS that was recommended to them.

Summary

Overall, we found that shared decision making for LCS often does not happen due to intersecting and perhaps interdependent factors. If just one step in the process does not happen, LCS does not happen. This includes: 1) identifying eligible patients – often performed by a clinician and not involving other staff and/or automated procedures except to identify smoking status, 2) the clinician using shared decision making to offer LCS as per the CMS mandate for payment to ensure reimbursement, 3) the patient deciding to undergo LCS, 4) office staff scheduling and coordinating LCS, 5) insurance approval leading to scheduling LDCT at another facility, 6) the patient traveling to and attending the appointment, 7) the clinician receiving and reviewing the results, and 8) the clinician or other health care professionals conducting any necessary follow-up. Like our workflow diagram in Fig. 1, there are multiple places where these steps may not occur. It became clear that with LCS, as opposed to smoking cessation, much of the work is in the clinician’s workflow rather than the staff.

Cumulatively the above issues create a cascade effect to produce low rates of LCS. The RE-AIM analysis sheds light on why each of these steps might not happen or be implemented with quality. We used a separate flow diagram (Fig. 2) of RE-AIM to consider the typical cascade of events that results in low effectiveness of an intervention [28]. From our data in this study on the issue of LCS implementation, we found consistency with this figure. Starting from the top left of the diagram: 1) Adoption – clinicians lack the knowledge/time to properly discuss and initiate shared decision making for LCS, 2) Implementation – clinical teams and their workflows are not systematized to get LCS eligibility or shared decision making prompted for the clinician or for others on the team to do those tasks or to follow through with referrals, 3) Reach – as a result patients are not offered LCS consistently or in a way that is compelling and feasible, so they do not go, which affects 4) Effectiveness – reduces the impact and benefits of LCS for patients. Last, 5) Maintenance—some patients believe it is not something they should do or it is a hassle, and it falls off the radar, then this reinforces the practice not doing it. Considering the interconnectedness of these steps (Using RE-AIM, process diagrams or related approaches) and understanding how decisions further upstream or earlier in this cascade can affect results downstream may be useful for future interventions aimed at improving LCS in rural settings.

Fig. 2

Cascade of events influencing RE-AIM outcomes

Discussion

Although previous studies have examined the challenges of consistently conducting shared decision making for LCS in rural primary care, this study is unique in its use of qualitative methods, an implementation framework (RE-AIM) for examining implementation issues specifically, and assessment of multiple participants in the primary care setting including patients in order to triangulate viewpoints. Key new findings for the understanding of LCS implementation include the lack of systematic processes that integrate clinic staff for shared decision making for LCS (as compared to smoking cessation and other cancer screenings [29]), which is further complicated by contingencies based on patient responses and preferences and stigma around the relationship of smoking to lung cancer. A main recommendation is further study to investigate if systematizing the process and involving more practice team members results in better uptake of LCS.

Some of our results replicate those found in prior research about LCS in rural settings [30, 31]. These include lack of geographic access to LDCT screening programs and limited or ineffective clinician-patient communication [19, 32]. Although previous studies assessing patients’ perceptions of cancer screening found that fatalistic belief systems contribute to patients’ declining LCS [33], this factor seemed especially strong among patients in our study.

Further research is needed to investigate what might be done to overcome this challenge, but it may be due in part to the combination of long term smoking, stigma regarding smoking [34, 35], sociocultural beliefs about lung cancer having a poor prognosis [36, 37], and social determinants of health challenges [38, 39] since both rural patients and people who smoke tend to have more social needs challenges than other groups [40, 41].

As with other research [39, 42] there was also confusion related to insurance coverage and costs associated with LCS. These perceptions run contrary to established policies and regulations dictating coverage for Medicare beneficiaries, Affordable Care Act policies given the USPSTF B grade for service recommendation, and coverage of LCS for Medicaid beneficiaries in Colorado [43, 44]. This barrier was identified following the initial launch of LCS while payers adapted their coverage and data management systems, but issues have substantially declined with more years of experience. It is important to explore this barrier more to determine if insurance issues are linked with inappropriate coverage refusals by payers or if initial coverage challenges contributed to sustained misperceptions regarding LCS policy.

This paper further contributes to the existing literature because of its focus on implementation within the practice. Of note, we found smoking identification, counseling, and referral to resources and assistance were systematic processes involving staff with follow-up support by the clinician with patient decision making; whereas, these processes for LCS were largely absent. Although the shared decision making may still rest with the clinician, other parts of the process – such as calculating pack years – could be added to staff responsibilities and be systematically collected. As found by Slatore, et al. [45] in their survey of practitioners in Oregon, the process was essentially left up to the clinician. They also found that registry and EMR systems lacked support for this effort [45]. This was especially true for rural settings in their study. This finding aligns with research showing that a centralized LCS program may be better suited to manage annual and follow-up screening [46], while maintaining communication with referring primary care clinicians regarding patient management. Other research describes how to set up a team-based approach for other cancer screenings and recommends that the same be done for LCS [47].

In examining perspectives of different participants in this process, we found similar barriers across groups regarding the practical difficulties of completing LCS and the variable response among patients to shared decision making regarding LCS. Patients were able to explain in detail their reasoning for accepting or declining LCS. Some wanted and took the opportunity to get LCS because it might help catch cancer early and prevent death, and others had belief systems that were inconsistent with screening such as it being not possible to change (“it’s my time when it’s my time to go”) or not necessary (“deal with it when something comes up”). Shared mental models about what something is, how it works, and why it is important are an important implementation concern [48]. One potential option is to systematically elicit patient perceptions as part of the shared decision making process, such as through a pre-visit questionnaire or having the patient watch a pre-visit video [49, 50]. Knowing their perceptions would illuminate to what extent patient values and perspectives are influencing the process of getting LCS versus logistical or financial concerns and help the clinician focus their consultation time on correcting misperceptions or problem-solving barriers to screening.

We found the RE-AIM model was helpful to conceptualize and categorize factors related to LCS [51, 52]. Using RE-AIM made it evident that some dimensions were discussed less often than others; for example, practice staff paid less attention to the effectiveness of LCS and the maintenance of continuing LCS. This has significant implications for implementation and points to areas for intervention in terms of clinical consultations regarding the process and value of LCS. In particular, the cascade effect of the interaction among several sequential factors in the RE-AIM model (e.g., if a patient was identified; if so, then approached for discussion, etc.) was particularly salient.

Our findings have several implications for implementation of LCS initiatives directed at rural primary care practices and their patients. First, context is a critically important factor, which we define broadly: in addition to physical settings and available resources and workflows, there are also the more subjective contextual issues such as history and patient and clinician values. Patient perspectives and preferences may be different and require different strategies. For example, in our study, some patients voiced concerns with getting health care interventions, consistent with a minimizer perspective [53], which may be more common in rural areas [26]. Second, processes for implementing LCS in rural settings need to be pragmatic, not overly time consuming, fit into existing workflows, and tap available resources in rural primary care as opposed to those applicable in large integrated care settings where much of the research on LCS has been conducted. This may be particularly challenging in rural settings that have less access to centralized LCS programs that take a more active role in managing the full LCS process and helping patients navigate the complexities. One important workflow option could involve clinical staff assessing smoking status and offering smoking cessation, including calculating eligiblility for LCS, while clinicians maintain responsibility for the LCS discussion with the patient. Third, outcomes could be enhanced by educating staff regarding 1) current eligibility/and reimbursement details; 2) differences between screening for high-risk patients versus diagnostic follow-up of symptomatic patients; and 3) importance of implementing of high quality shared decision making and smoking cessation counseling and not just “checking the box” [54, 55]. Finally, many rural settings do not have state of the art EHR systems or other technologies; enhancing automated identification and prompting systems for identification and follow up with LCS eligible patients, and conducting ongoing audit and feedback would likely enhance success.

Limitations of this study include the relatively small sample size in Colorado, although we did find thematic saturation and feel the sample was sufficient for this qualitative exploration. Our findings need replication – especially as they were obtained during COVID-19 and at a time when policy and reimbursement issues around LCS remain poorly understood. This as well as the relatively new and somewhat changing and uncertain specifics around requirements for reimbursement might well produce findings that could vary over time. Our study also has strengths and unique contributions, especially the use of an implementation science approach to LCS with use of RE-AIM and multiple perspectives for the analysis [56].

Conclusion

In conclusion, there are multiple contextual factors that affect implementation of LCS and performance of shared decision making in rural settings. Several of the perceived challenges were shared across different types of participants. Future research should attempt to replicate and expand our findings in different settings and evaluate interventions based on proposed recommendations regarding more robust assessment of smoking by staff and integrating additional tools to educate candidates for LCS.