Understanding the decision to screen for lung cancer or not: A qualitative analysis

Abstract Background Although new screening programmes with low‐dose computed tomography (LDCT) for lung cancer have been implemented throughout the United States, screening uptake remains low and screening‐eligible persons' decisions to screen or not remain poorly understood. Objective To describe how current and former long‐term smokers explain their decisions regarding participation in lung cancer screening. Design Phone interviews using a semi‐structured interview guide were conducted to ask screening‐eligible persons to describe their decisions regarding screening with LDCT. The interviews were transcribed and analysed with conventional content analytic techniques. Setting and participants A subsample of 40 participants (20 who had screened and 20 who had not) were drawn from the sample of a survey study whose participants were recruited by Facebook targeted advertisements. Results The sample was divided into the following five groups based on their decisions regarding lung cancer screening participation: Group 1: no intention to be screened, Group 2: no deliberate consideration but somewhat open to being screened, Group 3: deliberate consideration but no definitive decision to be screened, Group 4: intention to be screened and Group 5: had been screened. Reasons for screening participation decisions are described for each group. Across groups, data revealed that screening‐eligible persons have a number of misconceptions regarding LDCT, including that a scan is needed only if one is symptomatic or has not had a chest x‐ray. A physician recommendation was a key influence on decisions to screen. Discussion and conclusions Education initiatives aimed at providers and long‐term smokers regarding LDCT is needed. Quality patient/provider communication is most likely to improve screening rates.


| INTRODUC TI ON
Lung cancer screening with annual low-dose computed tomography (LDCT) is recommended by the U. S. Preventive Services Task Force (USPSTF) for long-term current and former smokers. 1 Lung cancer is the deadliest form of cancer; nearly 1.8 million people including 154 050 Americans are expected to die from lung cancer in 2019. 2,3 Lung cancer kills more people in the United States than breast, colorectal, pancreatic, and prostate cancers combined. 4 Most die because they are diagnosed at an advanced stage, and, until recently, an effective screening test did not exist.
LDCT of the chest is promising for the early detection of lung cancer. The National Lung Screening Trial (NLST), the largest randomized controlled trial to date on lung cancer screening, found an approximate 20% relative reduction in lung cancer-related mortality in high-risk individuals (age 55-74, current or former smokers who have quit within the past 15 years, and have at least a 30 pack-year history of cigarette smoking) screened with annual LDCT compared to chest x-ray over a 3-year period. 5 More recently, the Dutch-Belgium Lung Cancer Screening [NELSON] Trial also found decreased lung cancer mortality in high-risk individuals screened with LDCT of the chest compared to chest radiography. 6 In 2013, the NLST results led to the Grade B recommendation from the US Preventive Services Task Force of lung cancer screening with LDCT for high-risk individuals (aged 55-80 years with a minimum 30 packyear smoking history who are current or former smokers who quit within the past 15 years). 7 Subsequently, the Centers for Medicare and Medicaid Services issued a coverage determination to cover lung cancer screening among high-risk individuals with an upper age limit of 77. 8 Although new screening programmes with LDCT have been implemented throughout the United States, screening uptake remains low 9 and persons' decisions to screen or not screen for lung cancer remain poorly understood. [10][11][12] The purpose of this study was to describe how current and long-term smokers explain their decisions to participate or not in screening for lung cancer with LCDT.

| PARENT S TUDY: IDENTIF YING FAC TOR S A SSO CIATED WITH LUNG C AN CER SCREENING
Data for this study were collected as part of a larger sequential explanatory mixed methods study designed to identify factors associated with lung cancer screening (referred to as the parent study).
The study consisted of Phase 1, a quantitative phase, and Phase II, a qualitative phase. Both phases of the parent study were approved by the authors' institutional review board, and all research team members had human subjects training. Although Phase II is the focus of this report, Phase 1 is first briefly described as it provided the context of Phase II.

| Phase I: quantitative study
The aim of Phase I was to test an explanatory framework for lung cancer screening participation. As seen in Figure 1 LDCT scan in the past 12 months to screen for lung cancer. They were given a description of LDCT of the chest and provided a photograph of the scanner so as to distinguish the procedure from a chest x-ray.

| Phase II: qualitative study
Phase II was a qualitative study in which telephone interviews were conducted with a sub-set of the participants in Phase I. Participants in Phase II were asked to explain and expand on some of their responses to items in the Phase I survey.

| Sampling and recruitment
Participants who completed the survey were asked if they would be willing to participate in a follow-up phone call and 336 expressed a willingness to do so. Selective sampling was used to recruit a subsample of 40 persons from those who agreed to participate in the telephone interviews. Phase II sampling procedures were aimed at ensuring participants whose screening decisions converged with, or diverged from, the Phase I explanatory framework in a variety of ways were interviewed. Therefore, we selected 20 participants who had screened and 20 who had not. Within each group, we included 10 persons whose scores on the psychological and cognitive/health belief measures were hypothesized to be favourable to screening and 10 persons whose scores on these measures were hypothesized to be unfavourable to screening. The creation of the profiles and the selective sampling of participants is described in more detail in Appendix S1. Those who were selected were contacted by email or phone and scheduled for a phone interview.

| Data collection
Tailored semi-structured interview guides were developed for each participant in Phase II based upon whether they had screened or not combined with their scores on the psychological and/or cognitive and health belief factors subscales in Phase I (see Appendix S2).
The first question on the survey in Phase I was whether participants had a LDCT to screen for lung cancer in the past 12 months. During the interview, the interviewer confirmed their response. If the participant had screened, they were asked how they had decided to be screened. If they had not screened, they were asked how they decided not to screen. The interviewers then referred to the participants' high or low results on the subscales and asked participants to elaborate on their responses. For example, if a participant had scored low on the medical mistrust subscale, they would be asked to describe interactions with providers that lead to mistrust.
The interviews lasted between 15 and 60 minutes. While some participants were reticent and difficult to engage, most were forthcoming with their thoughts and feelings about lung cancer screening. The interviews were digitally recorded and transcribed verbatim by a member of the research team. Each transcript was given an identification number (001-040), and no identifying data were included on the transcripts.

| Data analysis
Data were analysed using conventional content analysis. 13 Conventional content analysis is an inductive process that allows codes and categories to emerge from the data rather than a deductive process confirming or refining a pre-existing theoretical structure. 13  The nursing students and the project manager were each assigned transcripts to code individually. All data relevant to the participants' stated reasons for being screened or not were extracted and divided into text units (eg words, phrases, sentences or short passages capturing a single thought). They labelled each text unit with a short code that captured its essence. A senior researcher verified each of the codes by re-examining the transcripts.
The codes were then entered into data display tables. 14 The tables were constructed by assigning a row to each participant and labelling it with their study ID number. The table had two columnsone displayed each participant's basic demographic information, including their age, gender, ethnicity and smoking status, and one displayed all the codes associated with that participant. The rows were initially divided into two groups according to whether the participant had been screened or not.
The senior researchers met to summarize the information on the data display tables using a process of discussion and consensus.
Because the analysis involved a low level of interpretation, analytic disagreements were few and easily resolved by a re-examination of the data. When examining the interview responses for those who had not been screened, we recognized that they were a heterogeneous group in terms of their screening decisions, which ranged from having made a firm decision not to be screened to having made a firm decision to be screened. We therefore divided the sample into five groups: persons who had decided not to be screened (Group 1), persons who had not deliberately considered being screened but were at least somewhat open to it (Group 2), persons who had deliberately considered being screened but had not made a definitive decision (Group 3), persons who fully intended to be screened but had yet to do so (Group 4) and persons who had been screened at the time of the interview (Group 5).
The codes associated with each group were compared for similarities and differences and grouped into categories. The categories, which represented a variety of reasons that participants in each group decided to be screened or not, were validated with a re-examination of the transcripts. The research team determined that the 40 interviews yielded ample data to provide a robust description of how current and long-term smokers explain their decisions to participate or not in screening for lung cancer with LCDT. A narrative description of each category was prepared and reviewed by the research team.

| Sample
Forty persons participated in the interviews, but data from one person were lost due to recording equipment failure. One participant had indicated on her survey that she had screened but denied this during the interview and was therefore considered a non-screener for this analysis. No participants reported having a lung scan be- Of the final sample of 39, 26 were women and 13 were men.
Participants ranged in age from 55 to 70, with an average age of 62 years. Thirty-five were Caucasian, one was African American, one was American Indian/Alaskan Native, and two were mixed race (ie African American and Native Hawaiian, Caucasian and American Indian). Twelve were divorced, 10 were married, seven were never married, five were living with a partner, and five were widowed. Twenty-eight were not working for pay, six were working full-time, and five were working part-time. Eleven had graduated from college, 21 had completed some college, six had graduated from high school, and one had not graduated from high school.
Eight had an annual income of >$50k, 18 between $25K and $50k, and 13 < $25k. Twenty-one participants were former smokers and 18 currently smoked. Information about earlier screening experiences other than the LDCT or illness history was not collected in the Phase I survey.

| Decisions regarding participating in lung cancer screening
In response to the interviewers' initial questions about decisions regarding lung cancer screening participation, most participants pro-   A few participants in this group stated they had not had the time to get screened, whereas others were deterred by a variety of other factors. The cost of the scan was a concern for some of the participants who said they would have the scan if it was free or covered by insurance. Other participants wanted to be screened but were reluctant to get a scan because they feared the results. A 55-yearold woman who formerly smoked (ID: 014) said, 'I am going to go [to get a lung scan] but right now I haven't gone. Eventually I will go.

| Group 3: deliberate consideration but no definitive decision to be screened
I'm trying to work up the courage'. Other reasons participants put off screening included having other more pressing health issues and having a serious ill family member who required their attention. These participants discussed why they delayed getting a lung scan in light of the fact that they were determined to be screened.

| Group 4: intention to be screened
In some instances, participants were hindered by a physician who was not supportive of the lung scan. A 62-year-old woman who formerly smoked (ID: 017) decided to have a lung scan done 'as a baseline' and went to 'the x-ray place' to have it done. However, she was told to get a 'written order' from her physician for the scan.
The participant indicated that her physician refused to provide the order because he believed the scan was unnecessary as she was asymptomatic. A 63-year-old woman who currently smoked (ID: 021) had decided to have a lung scan, along with a number of other screening procedures, when she received Medicare in a few months following the interview.

| Group 5: had been screened
All persons who had had a lung scan at the time of the interview clearly articulated at least one reason why they chose to be screened and several provided more than one reason. The most common reasons included experiencing pulmonary problems, receiving a recommendation from a physician and valuing early detection.

Experiencing pulmonary problems
Some participants indicated that they had a lung scan because they had a pulmonary illness such as chronic obstructive pulmonary disease (COPD), emphysema, collapsed lung, pulmonary hypertension, or scleroderma or were experiencing 'breathing problems', shortness of breath or a cough. A 64-year-old woman who formerly smoked (ID: 001) said, 'I was screened because I do have Stage 4 COPD, so of course they did test for cancer'. Some of these participants indicated that the scan was a necessity rather than a choice.

Receiving a recommendation from a physician
Some participants indicated that they had been screened on the recommendation of their physician. Some physicians had recommended the scan because the participant was having some pulmo-

Other reasons
A few participants cited other reasons for having a lung scan. Some indicated that they had a lung scan because it was covered by their insurance as a 'preventative tool'. Others indicated they had a lung scan because it was required by Medicaid or was part of a research study. One had a scan because of a family history of lung disease.

| D ISCUSS I ON
Consistent with our prior work, 12 participants' decisions to screen for lung cancer ranged from a firm decision not to screen because they believed it was unnecessary and a waste of time to a decision to screen based on a firm conviction that screening was essential to their health and well-being. Factors that were most likely to promote screening included perceived physician endorsement and experiencing pulmonary symptoms. Factors that impeded decisions to screen included the belief that other tests detect cancer, cost of the scan, fear of the results and other life stressors.
As a result of the findings, the team determine that the groups identified in this analysis aligned with precaution adoption process model (PAPM) 15  found that a provider recommendation is a strong predictor of lung cancer screening participation [16][17][18] and with studies on other types of cancer screening that found that a provider recommendation is the best predictor of breast and colorectal cancer screening. 17,[19][20][21] The results also revealed that several participants held inaccurate beliefs that influenced their decisions about lung cancer screening participation. Our findings support similar research confirming lack of knowledge about the concept of screening. 12,22,23 A common misconception was that those with lung cancer are symptomatic and thus screening was unnecessary in the absence of symptoms. Another misconception was that routine examinations and chest x-rays would reveal lung cancer thereby rendering a LDCT redundant. These two misconceptions discouraged screening and thus could impede detection of lung cancer at early, more treatable stages. Moreover, several participants were convinced that lung cancer screening would be very expensive and not covered by insurance. Although this concern was found in with our prior studies examining screening barriers, 16,17 lung cancer screening is a Grade B recommendation by the USPSTF with zero copayment for Medicare beneficiaries.

| LI M ITATI O N S
The findings should be considered in the context of several study limitations. Determination of screening status was based on selfreport, and a few participants needed to be reminded a lung scan was different than a chest x-ray. However, all participants had been provided a picture of a LDCT scan as well as a description of the scan and its purpose. Another limitation was that prior participation the minority group to be recruited and networking with page administrators who can promote the study, may be needed. 27 We also recognize that there were twice as many women as men in the qualitative sample and this could have had an impact on results. However, our prior work on the role of gender in lung cancer screening behaviour has not revealed any gender-related differences. 28 In addition, in Phase I, we did not collect data on some factors

| CLINIC AL IMPLIC ATIONS
Our study confirms that education regarding lung cancer screening should target both patients and providers and should be widely implemented. Similar to findings by Raz and colleagues, 30 our findings suggest that such efforts should clarify the difference between screening for lung cancer to detect it early in the absence of symptoms and diagnostic testing to identify aetiology when symptoms are present.
Education initiatives should stress that LDCT is the only effective and approved test to screen for lung cancer and dispel the misconception that a chest x-ray is an effective screening test. Moreover, all providers need to be aware of national guidelines and practice recommendations and the evidence that supports screening with LDCT. Targeted provider education should be offered through multiple venues (ie professional organizations, public service announcements, clinical pearls [tips for clinicians to apply in clinical practice], continuing education offerings). Our findings also suggest that high-quality patient-provider communication about lung cancer screening should occur regularly.
Given that our participants' reasons for not screening were varied and in some instances evolving, opportunities to discuss their hesitancy to screen with a provider would likely improve screening rates.

| SUMMARY
During phone interviews, screening-eligible persons were asked to explain their decisions about participation in screening with LDCT. Some revealed they decided to screen because they were firmly convinced of its benefit, some revealed they decided not to screen because they were certain it was unnecessary, and others discussed being open to screening, thinking about screening or intending to screen. Although a variety of factors influenced their decisions, including several misconceptions about LDCT, their health-care providers' role in discussing or endorsing screening was paramount. The findings indicate that education regarding lung cancer screening is needed for screening-eligible persons and their providers and that high-quality patient-provider communication regarding screening is most likely to improve screening rates.

CO N FLI C T O F I NTE R E S T
None.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data consisted of transcripts of semi-structured interviews that might contain information that could identify participants and thus will not be available.