INTRODUCTION

The United States Preventive Services Task Force (USPSTF) guidelines recommend lung cancer screening (LCS), with an annual low dose–computed tomography scan of the chest, for individuals who meet age and smoking history criteria.1 Despite receiving a Grade B from the USPSTF, this recommendation has not been well adopted, with only 5.8% of eligible patients undergoing LCS annually.2 This is in contrast to screening for breast cancer, which similarly has a Grade B designation by the USPSTF, but has an uptake rate of 76% among eligible individuals.3 Medicare policy since 2015 has stipulated that LCS must be offered in the context of “shared decision making, in an effort to promote informed, value-concordant decisions.”4 A similar Medicare policy does not exist for breast, colorectal, or prostate cancer screening, and its impact on LCS adherence rates remains unknown.

Shared decision making is an approach whereby clinicians and patients make decisions together; physicians discuss available screening options, review the likely benefits and harms of each, and elicit their patients’ goals and values to arrive at a decision that is congruent with patient preferences.5,6 Shared decision-making conversations can be supported by the use of decision aids, which help mediate clinician-patient discussions, provide educational information, consider potential benefits and harms, and assist patients in arriving at decisions that reflect their goals of care and personal values.6

Many individuals who are eligible for LCS have comorbid conditions (eg, cardiovascular or cerebrovascular disease, chronic obstructive pulmonary disease) because of the shared risk factor with lung cancer from tobacco exposure.7-11 These comorbid conditions may alter the risk-benefit ratio from LCS by influencing the risk of complications from lung cancer diagnostic and curative procedures, modifying eligibility for lung cancer treatment, and negatively influencing quality of life and life expectancy.10-12 The influence of comorbidities among individuals who are eligible for LCS is acknowledged in many guidelines; the USPSTF recommends that annual screening continue until age 80 years or until patients “develop a health problem that substantially limits life expectancy or the ability or willingness to have curative lung surgery.”1 The American College of Chest Physicians more directly advises that LCS not be offered to individuals with “comorbidities that substantially limit their life expectancy and adversely influence their ability to tolerate the evaluation of screen detected findings.”12 Given the medically complex history of individuals with comorbidities, LCS shared decision making requires a clinician to consider the nuances of each condition while keeping in mind a holistic view of the patient.

The extent to which comorbidities influence the considerations of primary care physicians (PCPs) when engaging in shared decision making with patients otherwise eligible for LCS remains unknown. The goal of this qualitative study therefore was to learn more about how PCPs reason through the LCS decision-making process with complex patients having comorbidities.

METHODSParticipant Selection

We conducted a qualitative study between October 2020 and February 2021 with PCPs to assess their views on the potential impact of patient comorbidities on LCS shared decision-making conversations. We identified attending PCPs from the faculty roster of 4 internal medicine practices affiliated with the Mount Sinai Health System in New York City. The PCPs recruited and interviewed were all participants in a larger study described in detail elsewhere.13 The study was approved by the institutional review board at the Icahn School of Medicine at Mount Sinai (protocol 19-02795), and all participants gave informed consent.

Setting and Data Collection

A study research coordinator enrolled and obtained the consent of participants. After giving consent, the PCPs participated in a 45-minute recorded Zoom (Zoom Video Communications Inc) video conference interview with a PCP and clinical researcher (M.S.K.) and a health/social psychologist and behavioral scientist (M.A.D.), both of whom were experienced in qualitative data assessment.

Several domains of the theoretical domains framework (TDF)14 were assessed during the interviews. The TDF is validated for use in health care settings as part of implementation research to identify barriers and facilitators when introducing new or changing existing clinical practices. By integrating 128 constructs across 14 domains, the TDF synthesizes information from 33 behavioral change theories to create a comprehensive lens to view the processes involved in and the factors that influence behavior. We used the TDF domains to create a semistructured interview guide to focus discussions with the PCPs. Question prompts examined PCPs’ understanding of how comorbidities influence LCS and explored the presence and extent of shared decision-making discussions in the context of comorbidities. The study interview guide and a complete list of question prompts are included in the Supplemental Appendix. The interviewers used a semistructured interview method, and all audio recordings were transcribed professionally, deidentified, and audited for accuracy.

Analysis

Three investigators (J.S., M.S.K., M.A.D.) all independently conducted a thematic analysis following the approach of Braun and Clarke.15 The Dedoose web-based application (Socio-Cultural Research Consultants, LLC) was used to organize and analyze the research data.16 After coding the first 3 transcripts, the investigators discussed, refined, and revised the identified themes by consensus, and applied the codebook to all subsequent transcripts. Every 5 transcripts were reviewed by the coders to ensure appropriate intercoder reliability. Throughout this process, the completeness of the codebook was verified. No new themes emerged after 15 interviews, indicating thematic saturation had been achieved. At the end of the study, a fourth study team member (O.M.) reviewed the themes relevant to considerations of patient comorbidity to confirm that each was unique and internally consistent.17

RESULTS

Our sample consisted of 15 PCPs from 4 academic-affiliated primary care practices. Their characteristics are shown in Table 1. On average, they had been practicing for 15 years and had ordered 70 low-dose computed tomography scans.

Table 1.

Physician Characteristics (N = 15)

Analyses of the interviews to assess impact of patient comorbidities on LCS shared decision-making practices revealed 3 main themes: (1) to discuss or not to discuss this screening, (2) shared decision making is not a simple conversation, and (3) ultimately, the decision is up to the patient. We discuss these themes and subthemes, and provide illustrative quotes below.

Theme 1. To Discuss or Not to Discuss

Theme 1 addressed PCPs’ use of subjective clinical judgment, or the gestalt of the patient’s clinical case, to decide whether to initiate discussions of LCS with a patient with comorbidities. PCPs reported doing a mental cost-benefit analysis to decide whether to raise the topic in this context. They described self-created estimates assessing the patient’s health, life expectancy, and quality of life. In addition, the PCPs considered psychosocial factors—whether the patient had the support system, mobility, and cognitive capacity to be able to adhere to follow-up diagnostic and treatment protocols if anything were to be found on LCS.

The PCPs quickly synthesized their perceptions of all of these factors to decide whether a discussion of LCS was likely to be productive and beneficial. It should be noted that, at times, the clinical judgment heuristics overrode USPSTF guidelines,1 so that some patients who met criteria for LCS were not informed that it was an option. This internal heuristic theme is exemplified by a comment from one PCP: “That patient with 20 medications and 10 different specialists, and all these comorbidities, chances are life expectancy [isn’t] that high, so maybe the utility of lung cancer screening is pretty low and less than the other things they have going on. So, I use my judgment. I don’t have a specific [criterion] that I say, ‘Oh, if they have this, this, and this, I won’t ask them.’ I just kind of use judgment.”

Theme 1 had 4 subthemes, each an essential variable in the PCPs’ heuristic assessment. The subthemes were (1a) How sick is my patient?, (1b) How long does my patient have to live?, (1c) How likely is my patient to follow up?, and (1d) How does patient quality of life factor into this?

Subtheme 1a. How Sick Is My Patient?

The PCPs described assessing the patient’s current health status to judge whether LCS should be raised (Table 2). Overall, they stated that they would likely not raise the topic with patients having high disease burden. PCPs believed that such patients did not need more on their plate, and decided that other issues should and would take priority in their appointment with the patient. Some of the vignettes the PCPs described show extreme, and often rare, levels of comorbidity. The ways in which they thought through their decisions, however, are translatable to patients with varying levels of comorbidity. For example, one PCP discussed a patient with stage 2 cancer and stated they would not recommend that the patient undergo LCS. Although cancers of this stage are generally treatable, this scenario suggests that the PCPs ascribe to a “one diagnosis at a time” approach to patient care and management. When they believed that their patient needed to focus on another condition, that view often precluded broaching LCS decision-making conversations with the patient.

Table 2.

Illustrative Quotes of Subtheme 1a, How Sick Is My Patient?

Subtheme 1b. How Long Does My Patient Have to Live?

The PCPs reported using the patient’s clinical history to predict how long a patient had to live (Table 3). Without consulting validated life expectancy prognosticators, PCPs relied on heuristics to assess patient life expectancy and risk.18 In doing so, they used subjective metrics for patient history to calculate life expectancy. After determining life expectancy subjectively, PCPs described placing subjective cutoff points, ranging between 1 and 10 years, at which the patient’s life could be prolonged with LCS. These estimates were then used to determine whether screening was worth discussing with the patient.

Table 3.

Illustrative Quotes of Subtheme 1b, How Long Does My Patient Have to Live?

Subtheme 1c. How Likely Is My Patient to Follow Up?

The PCPs also considered the patient’s likelihood of being able to follow through on treatment should they receive a lung cancer diagnosis (Table 4). The psychosocial factors considered included the patient’s support system, transportation barriers, the patient’s past adherence, and how often the patient expressed reluctance to attend medical appointments. PCPs inferred from current and past trends in behavior whether the patient would adhere to treatment follow-up. They did not consider whether a cancer diagnosis would alter that behavior.

Table 4.

Illustrative Quotes of Subtheme 1c, How Likely Is My Patient to Follow Up?

Subtheme 1d. How Does Patient Quality of Life Factor In to This?

Questions concerning the patient’s current quality of life came up (Table 5). The PCPs used quality of life as a key component in determining whether LCS would be the best option for their patient, considering the patient’s engagement in and enjoyment of activities at present and the potential of LCS outcomes to change that for better or worse.

Table 5.

Illustrative Quotes of Subtheme 1d, How Does Quality of Life Factor in to This?

Theme 2. Shared Decision Making Is Not a Simple Discussion

Once PCPs decided that LCS was worth raising with a patient, they turned their focus to how the conversation would take place (Table 6). Some PCPs were able to have open and objective discussions of the pros and cons, whereas others struggled with how to balance honesty with putting a thumb on the scales.

Table 6.

Illustrative Quotes of Theme 2, Shared Decision Making Is Not a Simple Discussion

Theme 3. Ultimately, the Decision Is Up to the Patient

Most PCPs believed that ultimately, their role was more to advise patients on LCS than to make the final decision on whether to proceed (Table 7). They believed that at the end of the process, the patient was the one who should make the final decision even if the clinician personally disagreed with that decision. The PCPs stated that the patients were the ones who had to live with the decision, and as such, it was theirs to make.

Table 7.

Illustrative Quotes of Theme 3, Ultimately, the Decision Is Up to the Patient

DISCUSSION

Comorbidities may influence the risk-benefit ratio of LCS by affecting workup complications, quality of life, life expectancy, and patient eligibility for lung cancer treatment, all of which should be discussed during shared decision making. In our qualitative study of PCPs from 4 academic-affiliated primary care practices at a large urban health care center, we observed LCS shared decision-making conversations differed substantially for patients with complex comorbidities. For patients with complex comorbidities, the differences in these discussions were not limited to the content of the conversations; importantly, PCPs were found to take additional steps before initiating LCS discussions that ultimately reduced the frequency of shared decision-making conversations.

Many PCPs reported using clinical judgment when deciding whether to discuss LCS rather than strictly adhering to the guidelines to engage all eligible patients in shared decision-making conversations. Clinical judgment was influenced by patient characteristics (eg, comorbidities); life expectancy; opinions about the patient’s personality, including their expectations of and attitude toward medical care; and the presence of a support system. Patients who were perceived as likely to adhere to recommendations and as having a high quality of life were more likely to be advised to complete LCS as opposed to those who had previously reported frustration or dissatisfaction with their state of health and well-being. Although PCPs believed their judgments helped determine which patients were good candidates for LCS, they were, in effect, determining which patients were eligible for shared decision-making conversations and as such, which patients deserved a say in the decisions concerning their care.

A limitation of our study is that we included only internists, and findings may not be generalizable to the practices of physicians of other specialties, including pulmonologists. In addition, all of the PCPs were recruited from practices affiliated with the same urban health system, further limiting generalizability of findings; however, our sample of PCPs was diverse, with varying years of clinical practice, and the use of one medical center can also be viewed as a strength because it eliminates any potential for differences in the LCS referral processes (eg, centralized vs decentralized) affecting their practices of shared decision making. Although some did reference patient anxiety as a potential factor influencing the shared decision-making process, PCPs were not directly asked and did not specifically address the LCS decision-making process in the context of patients having psychiatric comorbidities. Given that national reports show smoking rates of people with psychiatric conditions are double those without,19 it is interesting that when asked to describe previous experiences of shared decision making with complex patients, PCPs did not include any patients with noteworthy psychiatric conditions. Future research should investigate the impact of mental health comorbidities on gestalt clinical judgments made by PCPs when determining patient eligibility for LCS.

Other studies investigating LCS shared decision making have found low PCP engagement with patients in these conversations. Previous studies have attributed the paucity of these conversations to factors such as time allowed per visit,20,21 individual PCP biases,22,23 and inadequate physician knowledge of the shared decision-making process,24,25 all of which identify the need for communication skills training.24,26-29 Our study presents a possible unseen factor precluding shared decision-making conversations: the clinical judgments made a priori by the PCP. Our findings indicate PCPs make clinical judgments using the content typical of shared decision-making conversations ahead of time and without the patient to decide whether to engage the patient in a discussion about LCS. They reported making these clinical judgments most commonly in cases where patients had complex comorbidities.

The reliance on subjective clinical judgment, or the gestalt of the presenting patient, is not surprising given that there is little evidence-based guidance on the benefits of LCS in interpreting an individuals’ life expectancy and the salience and severity of comorbidity. There is also little guidance on how these assessments should influence clinical decision making on LCS referral, as it is challenging to combine in a single conversation the need for screening, the patient’s life expectancy, and the potential complications of comorbidities. Studies to determine when outcomes of LCS are no longer favorable have yielded conflicting results. In one study of individuals from the Prostate, Lung, Colorectal, and Ovarian Screening Trial, individuals with a history of coronary artery disease had an increased likelihood of high-risk complications after a positive LCS result, and individuals with chronic obstructive pulmonary disease were at high risk for respiratory complications after LCS.30 In a study of 1,741 veterans who underwent invasive evaluation after a positive LCS result, comorbidities did not increase the likelihood of high- or intermediate-risk complications.31 The conflicting data on LCS in patients with comorbidities could explain in part PCPs’ conversation hesitancy and perhaps temper their confidence in providing clinical guidance and recommendations on this screening to their complex patients.

Lung cancer screening is the only type of cancer screening where eligibility involves having a behavioral risk factor (smoking) that also increases the patient’s likelihood of multiple comorbidities. The increased risk from a smoking history, and its resulting comorbidities, likely contributed to the Medicare decision to have a policy of shared decision making that takes into account individual patient history.4 Although a request for shared decision making may seem straightforward, the conversation can take many forms, and guidelines about how to shape the conversation are vague. Not only are there multiple ways to communicate options to patients, there exist a variety of validated option-presenting methods and models, including the Enhanced Autonomy Model and the Independent Choice Model.32

The confusion about implementing the shared-decision policy on LCS4 puts clinicians in a problematic situation for which the only solution—eliminating the policy—is denied by the circumstances inherent to the situation. The policy was added to address the increased prevalence of complex patients and patients with comorbidities in the screening population, but it creates confusion about patient eligibility and distracts from the importance of LCS. It also possibly contributes to PCP ambivalence in recommending this screening to patients with complex comorbidities. The ambivalence of PCPs to discuss LCS with these patients could perhaps be seen in their responses to those with a strong opinion on this screening. Regardless of whether the patient was for or against undergoing LCS, many PCPs deferred to the patient and let them decide. Such findings stand starkly in contrast to current updates on preventive medicine models emphasizing quaternary prevention, in which PCPs are advised to be discerning with screening so that patients are not overscreened and overtreated.33 PCPs deferring to their patients to make a decision, especially in light of the heightened focus on clinician judiciousness, may reflect the lack of confidence the physicians feel in making LCS decisions. It could be that PCPs deferred to patients with comorbidities to make their own choice because the PCPs themselves were unsure of whether LCS would be appropriate for the patient.

Future research that includes real-world data from a generalizable population, as well as modeling studies, will determine the specific impact of comorbidities on complications from the workup of positive LCS tests. In addition, research should aim to understand PCP selection bias when engaging patients in shared decision-making conversations and the extent to which PCPs consider patient goals when deciding whether to discuss screening. Doing so is particularly important to create decision support strategies in cases of older adults with multiple comorbidities, and efforts should be made to evaluate the outcomes of actionable frameworks, such as ones discussed by Boyd et al.34 Initiatives to create tools that standardize shared decision-making conversations will likely improve the frequency with which truly patient-centered decisions are to be reached.20-22,24,28,29 These tools have been shown to not only increase the rate of shared decision conversations, but also to improve patient outcomes of care as streamlined and centralized processes have been found to improve LCS adherence and follow-up.35 The inclusion of LCS as a Healthcare Effectiveness Data and Information Set (HEDIS) measure in 2024 will likely spur the development of centralized interventions to promote screening.36

There should also be increased research into outcomes from involving members of the care team, other than PCPs, to contribute to shared decision-making conversations. The outcome of a Medicare policy decision in 2022 that expands the type of health care professional able to engage patients in shared decision making beyond physicians and nurse practitioners remains unknown.4 At present, it may be that other members of the health care team do not have sufficient time or training to conduct the nuanced information sharing, collaboration, and supported deliberation required of shared decision making.37 With the proper education and skills training, however, there is opportunity to effectively create a new role that acts in tandem with the rest of the care team and is focused specifically on LCS decision-support counseling. As lung cancer treatment options evolve and better-tolerated therapies are developed, patients will have an increase in their choices that will need to be reflected in the LCS shared decision-making conversations. With expanding treatment options, clinicians will need to pursue additional education, and patient-clinician shared decision making will likely be not just one discussion, but rather an ongoing conversation with continual patient education sessions, as decisions may change with new treatment options. Using a single member of the care team to provide LCS decision support will relieve other members of the growing responsibilities involved in patient education and may be an effective time management strategy as LCS conversations become more involved. With supplemental staff facilitating LCS shared decision-making discussions, there will be increased opportunities to improve patient access to support, including the potential to expand LCS education beyond the clinical setting and into the community. Population health tools and community-based education programs could be developed with health professionals trained in LCS education and decision support to lead group information sessions, after which there could be an opportunity for 1-on-1 shared decision making for those eligible and interested.38

CONCLUSIONS

Our study is the first to characterize how PCPs consider chronic diseases and fitness for LCS in the context of comorbidities. Our findings indicate that PCPs make clinical judgments to assess whether the patient is a good candidate for LCS before approaching the patient, rather than determining whether the patient is well suited for this screening during a shared decision-making discussion. Once the conversation begins, PCPs do try to present unbiased and honest information. When faced with a patient with a strong opinion on LCS, however, regardless of whether the patient was for or against, the PCP deferred to the patient’s wishes.

Our findings show PCPs taking additional steps before discussing LCS with patients with complex comorbidities, indicating hesitancy in recommending this screening for this group. PCPs deferring to patient opinions rather than providing education and counseling support also indicates reluctance to impose their own opinion on LCS for patients with comorbidities. These behaviors are likely accentuated by conflicting data on LCS outcomes in patients with comorbidities, and by the dearth of evidence-based directions for guiding physicians’ approach to the screening in this population. Our findings support the call for continued research to determine the impact of comorbidities on LCS risks and benefits as well as its clinical application. PCPs need more evidence-based information on LCS in cases of complex comorbidities to be able to effectively conduct shared decision making with this population. As much as these steps are necessary to reduce PCP hesitancy to discuss LCS with patients, it will also be key to conduct research, establish protocols, and develop tools to reduce selection bias for patient-clinician shared decision-making conversations. Ambivalence of PCPs in referring patients for LCS likely reflects their uncertainty about its benefit and confusion from complicated guideline recommendations. Efforts should be made to streamline protocols and make LCS guidelines more clear for patients and clinicians, as the additional policy designed to emphasize caring for a high-risk population may indeed be undercutting the very people for whom it was designed to protect.