Shared Decision-Making for Lung Cancer Screening

Abstract

Background

Lung cancer screening (LCS) reduces lung cancer mortality, but it also carries a range of risks. Shared decision-making (SDM) is a process of engaging patients in their health care decisions and is a vital component of LCS.

Research Question

What is the quality of SDM among patients recently assessed for LCS?

Study Design and Methods

Cross-sectional study of screened patients recruited from two academic tertiary care centers in the South Central Region of the United States. Self-reported surveys assessed patient demographics, values related to outcomes of LCS, knowledge, SDM components including receipt of educational materials, and decisional conflict.

Results

Recently screened patients (n = 266) possessed varied LCS knowledge, answering an average of 41.4% of questions correctly. Patients valued finding cancer early over concerns about harms. Patients indicated that LCS benefits were presented to them by a health care provider far more often than harms (68.3% vs 20.8%, respectively), and 30.7% reported they received educational materials about LCS during the screening process. One-third of patients had some decisional conflict (33.6%) related to their screening decisions, whereas most patients (86.6%) noted that they were involved in the screening decision as much as they wanted. In multivariate models, non-White race and having less education were related to lower knowledge scores. Non-White patients and former smokers were more likely to be conflicted about the screening decision. Most patients (n = 227 [85.3%]) indicated that a health care provider had discussed smoking cessation or abstinence with them.

Interpretation

Among recently screened patients, the quality of decision-making about LCS is highly variable. The low use of educational materials including decision aids and imbalance of conveying benefit vs risk information to patients is concerning. A structured approach using decision aids may assist with providing a balanced presentation of information and may improve the quality of SDM.

Take-home Points.

Study Question: What is the quality of SDM among patients recently assessed for LCS?

Results: After SDM encounters for LCS, most participants displayed poor knowledge, did not receive a decision aid, and remained with decisional conflict.

Interpretation: Among recently screened patients, the quality of decision-making about LCS is highly variable but low, and use of decision aids is low.

Two large randomized controlled trials, the National Lung Screening Trial¹ and the Dutch Belgian Randomized Lung Cancer Screening (NELSON [Nederlands Leuvens longkanker Screenings Onderzoek])² trial, showed a reduction in mortality resulting from lung cancer using low-dose CT (LDCT) screening in appropriate high-risk individuals. The Centers for Medicare and Medicaid Services initiated reimbursement requirements for lung cancer screening (LCS) that include a patient counseling and shared decision-making (SDM) visit before the patient undergoes LDCT for LCS,³ which most organizations incorporate into their LCS process.^{4, 5, 6} In the case of LCS in which a small benefit, an absolute risk reduction of 0.3%, may be attenuated by potential harms, the use of SDM is especially important.⁷ SDM incorporates three essential elements between a patient and health care provider: recognizing and acknowledging that a decision is required, knowing and understanding the best available evidence, and incorporating the patient’s values and preferences into the decision.⁸

Specific components of SDM are outlined by the Centers for Medicare and Medicaid Services, including harms and benefits of LCS, education on the importance of adherence to annual screening, use of a decision aid, and counseling for smoking cessation or continued abstinence.³ The components of SDM often are difficult to implement in routine clinical practice, and existing studies predominantly focus on review of content, but do not directly address the quality of the process.^{9, 10, 11} An understanding of current SDM implementation for LCS may clarify areas where patients can be meaningfully involved in decision-making. In this study, we described the quality of SDM among patients recently screened for lung cancer. Based on limited previous research, we hypothesized that the quality of the LCS decision-making process would be poor. We further explored how patient factors are related to the quality of LCS decisions.

Methods

This was a cross-sectional study. We used the Strengthening the Reporting of Observational Studies in Epidemiology Statement checklist to guide our reporting of the results.¹²

Study Settings

Patients were recruited from two academic tertiary care centers in the south central region of the United States. One center is an academic medical center encompassing four hospitals, more than 70 clinic locations, and 900 faculty serving south and southeast Texas regions. The second center is located in a large metropolitan area, specializes in cancer care, and is affiliated with a health system including five hospitals and more than 40 clinics. Participants were provided a consent statement before completing the surveys. The project was approved by the institutional review board at both centers.

Consecutive patients were invited via postal mail, email, or both to participate in the study from May 2018 through July 2018 after undergoing LDCT scanning for LCS at two centers. Screening had to be completed within 12 months before the survey. Participants from the cancer center came from two main groups, (1) a tobacco treatment program located at the cancer center and (2) the center’s LCS program, and were identified through the screening database. These participants were referred for LCS by their primary care physician, cancer prevention center, pulmonary medicine, thoracic surgery, and thoracic oncology. Patients in the cancer center received a written order for LCS. Patients from the second center, an academic medical center, were referred for LCS during primary care encounters. Providers from this group have an institutional-approved lung screening pamphlet to assist with SDM. Additionally, when placing the order for LCS examination, providers attest that SDM occurred, and links to the Agency for Healthcare Research and Quality (AHRQ) (https://effectivehealthcare.ahrq.gov/decision-aids/lung-cancer-screening/patient.html) and shouldiscreen.com aids are available for use during the visit. Providers also have a video that covers the elements of SDM visits to use as discussion points.

Study Measures

Demographics

Patient age, race, education, smoking history, family history of lung cancer, LCS history, comorbidity, eligibility criteria, and risk factors for lung cancer were collected.

Patients’ Values Related to Outcomes of LCS

Patients ranked the importance they associated with five outcomes of LCS: finding cancer early when it is potentially curable, avoiding false alarms, avoiding additional harmful testing, avoiding overdiagnosis and treatment, and avoiding radiation exposure resulting from LCS.⁹ Scales ranged from 0 (“not at all important”) to 10 (“extremely important”).

Knowledge of Lung Cancer Screening

Patients answered a 16-item measure of LCS knowledge adapted from previous studies¹³ addressing knowledge of risk factors for lung cancer (questions 1-3), eligibility and frequency of screening (questions 4-8), benefits of LCS (questions 9-11), and potential harms of LCS (questions 12-16).

SDM Process, Involvement in Decision-Making, Receipt of Patient Educational Materials, and Discussion of Risk and Smoking Cessation and Abstinence

Patients completed an adapted version of the Shared Decision Making Process Survey, a four-item patient self-reported measure of the degree to which a health care provider discussed the benefits and harms of LCS and explained the options (to be screened or not screened) and of the desire to be screened.¹⁴ Patients were asked about their perceived involvement in the decision to be screened, if they received any materials describing the benefits and harms of LCS before being screened, and if their personal chances of lung cancer developing were discussed. We included a question about the importance the patient placed on having a conversation with a health care provider about the benefits and harms of LCS. Finally, patients were asked about smoking cessation or the importance of continued tobacco abstinence.

Patient Decisional Conflict

Decisional conflict is an individual’s perception of uncertainty about the course of action to take when the choices involve risk, loss, regret, or a challenge to personal life values. Often, it indicates an individual’s level of comfort with a decision. The SURE (Sure of myself; Understand information; Risk-benefit ratio; Encouragement) scale, a four-item short version of the Decisional Conflict Scale,¹⁵ also was completed by patients¹⁶ that addressed knowing the benefits and risks of screening, being clear about which benefits and risks matter most, having enough support to make a decision, and feeling sure about the choice. A score of 4 indicates no decisional conflict.

Statistical Analysis

We performed descriptive statistics to report baseline characteristics of the participants. Continuous variables were summarized as means with SDs and categorical variables were reported as frequencies and percentages. The χ ² test and Fisher exact test were used to compare demographic characteristics. For the patients’ values related to outcomes of LCS, the Wilcoxon signed-rank test was used to compare the difference between scores because the data did not have a normal distribution. Each value question was compared with the item about the importance of finding lung cancer early. We also provided percentages for scores of 9 or 10 vs ≤ 8 (reported in Table 6). Three multivariate regression models were used to evaluate the factors associated with different outcomes: ordinary least squares regression was used for the outcome of screening knowledge (percentage of correct answers) and logistic regressions were used for the outcomes of SDM process (scores of 3 and 4 vs 0-2) and decisional conflict (1 = some decisional conflict; 0 = no decisional conflict). All descriptive analyses were conducted in STATA version 15 software (StataCorp LP), and multivariate regression was performed in SAS version 9.4 software (SAS Institute, Inc.). The significance level was set at P ≤ .05.

Table 6.

Distribution of Patient Characteristics for Knowledge of Lung Cancer Screening, SDM Process, and Decisional Conflict for Current and Former Smokers Screened for Lung Cancer (n = 266)

Variable	Knowledge of Lung Cancer and Screeninga		SDM Processb		Decisional Conflict About Screening Decisionc
	Percentage	95% CI	OR	95% CI	OR	95% CI
Screening settings (academic medical center is reference group)
Cancer tobacco treatment program (n = 137 patients)	9.05	2.55-15.54	0.54	0.23-1.27	1.59	0.65-3.88
Cancer center lung cancer screening program (need label; n = 90 patients)	7.69	0.92-14.46	0.82	0.33-2.02	1.06	0.41-2.72
Time since last screening (< 3 mo is reference group), mo
3-6	–2.77	–8.89 to 3.34	0.67	0.30-1.52	0.68	0.27-1.73
6-12	–5.42	–11.11 to 0.27	0.59	0.27-1.28	0.48	0.20-1.15
Age (<65 is reference group)
65+	3.04	–1.14 to 7.23	1.02	0.59-1.78	1.78	0.97-3.25
Sex (1 = male and 0 = female; male as reference)
Female	1.8	–2.36 to 5.97	1.54	0.88-2.68	1.18	0.65-2.15
Race or ethnicity (White not Hispanic is reference group)
Other than White or Hispanic	–7.93	–14.48 to –1.37	2.51	1.03-6.10	4.31	1.36-13.7
Education (college graduate is reference group)
High school graduate or less	–11.42	–17.13 to –5.71	0.68	0.32-1.44	1.05	0.47-2.33
Some college or trade school	–10.11	–14.64 to -5.58	0.77	0.42-1.41	1.38	0.71-2.67
No. of times screened for lung cancer (1 is reference group)
2	–3.75	–9.12 to 1.63	0.83	0.42-1.65	1.68	0.79-3.56
3	2.46	–2.6 to 7.52	1.26	0.64-2.45	1.85	0.92-3.73
Current smoker (1 = yes and 0 = no; yes as reference)
No	1.41	–2.91 to 5.72	0.97	0.55-1.71	1.93	1.04-3.55
Pack-year smoking history	–0.01	–0.11 to 0.08	1.02	1.00-1.03	1.01	0.99-1.02

SDM = shared decision-making. Boldface values indicate characteristic is significantly related to the outcome variable at P ≤ .05.

^aPercentage correct responses using ordinary least square regression.

^b1 = scores of 3 and 4 and 0 = scores from 0 to 2 using logistic regression.

^c1 = some decisional conflict and 0 = no decisional conflict using logistic regression.

Results

Participants

A flow diagram of participant inclusion in the study is shown in Figure 1. A total of 676 patients were identified by the research team for the study. Of these patients, seven were found to be ineligible because of not meeting eligibility criteria, and 43 were not invited because we had reached enrollment targets. A total of 626 patients were invited to participate in the study. Two hundred sixty-six (42.5%) consented to participate and completed the survey.

Figure 1.

Flow diagram showing participant inclusion in the study. LCS = lung cancer screening; LDCT = low-dose CT.

Characteristics of Screened Patients

Table 1 lists participants’ characteristics at baseline. The mean ± SD age of patients was 64.8 ± 5.6 years, 52% of participants were men, and 87% of participants were non-Hispanic White people. Compared with county demographic data for 55- to 77-year-olds (demographics.texas.gov), the participants were more likely to be non-Hispanic White people. A family history of lung cancer was reported by 38% of patients, 39% were current smokers, and the mean pack-year history was 40.4 years. Overall, 82 patients (32%) reported having been screened once, and about half (n = 125 [48%]) were screened within the previous 6 months.

Table 1.

Baseline Characteristics of Screened Patients (n = 266)

Characteristic	Data
Sex
Male	137 (52)
Female	128 (48)
Race/Ethnicity
Asian/Pacific Islander	5 (2)
Black	14 (5)
Hispanic	8 (3)
White not Hispanic	231 (87)
Other	7 (3)
Education
Less than high school	6 (2)
Graduated high school/GED	46 (17)
Some college/trade school	96 (36)
Graduated college or more	117 (44)
Family history of lung cancer	101 (38)
Smoked ≥ 100 cigarettes in lifetime	264 (99)
Current smoker	103 (39)
No. of times screened
1	82 (32)
2	76 (29)
> 2	101 (39)
Time since last screening, mo
< 3	47 (18)
3-6	78 (30)
> 6-≤12	137 (52)
Diagnosis of COPD or different cancer	94 (35)
Age, y	64.8 ± 5.6
Years smoked cigarettes	39.5 ± 10.4
Average no. of cigarettes smoked per day	20.8 ± 10.2
Pack-year history	40.4 ± 22.0
Years since quit smoking	7.3 ± 6.8
No. of times screened for lung cancer	2.5 ± 1.9

Data are presented as No. (%) or mean ± SD. Some frequencies do not sum to 266 because of missing data. GED = general educational development.

Patients’ Values Related to Outcomes of Screening

Scores from the values questions evaluating screening harms were significantly different from the single question evaluating screening benefit (Fig 2). For instance, most indicated finding lung cancer early as “extremely important.” Patients’ values related to avoiding any of the harms of LCS were less important than finding cancer early (P < .001 for all comparisons). Up to one-third of patients rated four specific questions focused on the potential harms of LCS (false-positive results, harms resulting from further testing, radiation exposure, and overdiagnosis) as “not important” or “neutral” in their decision for LCS (range, 21.2%-38.8%).

Figure 2.

Graph showing patients’ values related to outcomes of lung cancer screening.

Knowledge of Lung Cancer Screening

Responses to the knowledge survey are given in Table 2. Overall, patients answered an average of 41.4% of the questions correctly. Across patients, the range of correct responses was 0% to 93.8%. Among specific knowledge questions, patients overwhelmingly understood that smoking was the leading cause or modifiable risk factor for lung cancer in the United States (n = 223 [84.2%]), but a minority (n = 79 [29.8%]) recognized lung cancer as the leading cause of cancer deaths. Most patients (n = 173 [66.3%]) correctly identified smoking abstinence as the best way to lower lung cancer risk. Of concern, 81 patients (31.0%) indicated that screening rather than smoking abstinence was the best way to lower the chances of lung cancer developing and dying of lung cancer.

Table 2.

Patients’ Knowledge of Lung Cancer Screening (n = 266)

Question	Correct Responses
Where does lung cancer rank as a cause of cancer death in the United States? (No. 1 cause of cancer deaths)	79 (29.8)
What is the best way to lower the chance of lung cancer developing or dying of lung cancer? (Don’t smoke)	173 (66.3)
What is the leading cause of lung cancer in the United States? (Smoking)	223 (84.2)
Do health professional groups recommend all current and former smokers be screened for lung cancer? (No)	52 (19.6)
Is screening recommended for someone who quit smoking more than 15 y ago? (No)	23 (8.7)
How often do health professional groups recommend people be screened for lung cancer? (Every year)	131 (49.4)
Is screening recommended for someone who has other health problems that may shorten their life? (No)	27 (10.2)
Is screening recommended for someone who is not able or willing to be treated for lung cancer? (No)	78 (29.3)
Without screening, is lung cancer often found at a later stage when cure is less likely? (Yes)	229 (86.1)
How much does screening for lung cancer lower the chances of dying of lung cancer? (About 20%)	7 (2.6)
Can lung cancer screening find other health conditions? (Yes)	212 (79.7)
How much radiation is a person exposed to through lung cancer screening? (More than a chest radiograph)	25 (9.4)
How many people with abnormal lung cancer screening results will have lung cancer? (Most will not have lung cancer)	33 (12.5)
Can lung cancer screening suggest that you have lung cancer when you do not? (Yes)	118 (44.4)
Can lung cancer screening miss a possible lung cancer? (Yes)	156 (58.9)
Will all tumors found by lung cancer screening grow to be life threatening? (No)	171 (64.3)

Data are presented as No. (%). Frequencies do not sum to total per group because of missing data. Correct response noted in parentheses.

Questions about the benefits of LCS were answered correctly more often or benefits were overestimated by patients, whereas knowledge of potential harms was variable. Few patients (n = 23 [8.7%]) knew that LCS was not recommended for someone who quit smoking more than 15 years ago. Many patients understood that screening can result in false-positive results (n = 118 [44.4%]) and false-negative results (n = 156 [58.9%]), although others were “not sure” about screening performance. The correct interval for screening was identified by 131 patients (49.4%). Finally, 33 patients (12.5%) understood than most people with abnormal scan results will be found not to have lung cancer, an indicator of understanding of the predictive value of abnormal results.

Shared Decision-Making Process About Screening

Participants were asked to recall specific aspects of the SDM visit. Only 45 patients (17.1%) scored the maximum on the Shared Decision Making Process Survey (Table 3). Most patients (n = 163 [61.5%]) indicated that a health care provider asked if they wanted to be screened. Most patients (n = 181 [68.3%]) indicated that a health care provider talked with them “some” or “a lot” about reasons to be screened, similar to the number of patients (n = 163 [61.7%]) who indicated that a health care provider did not discuss reasons why the patient may not want to be screened, whereas 46 patients (17.4%) indicated that harms were discussed “a little.” Most patients (n = 185 [69.6%]) indicated that a health care provider explained they could choose not to be screened, whereas 20.3% did not receive an explanation.

Table 3.

Patients’ Responses to Shared Decision Making Process Survey (n = 266)

Question	Response
How much did a health care provider talk with you about the reasons you might want to be screened for lung cancer?
A lot	72 (27.2)
Some	109 (41.1)
A little	36 (13.6)
Not at all	48 (18.1)
How much did a health care provider talk with you about reasons you may not want to be screened for lung cancer?
A lot	10 (3.8)
Some	45 (17.1)
A little	46 (17.4)
Not at all	163 (61.7)
Did a health care provider explain to you that you could choose not to be screened for lung cancer?
Yes	185 (69.6)
No	54 (20.3)
I’m not sure	27 (10.2)
Did a health care provider ask you if you wanted to be screened for lung cancer or not?
Yes	163 (61.5)
No	80 (30.2)
I’m not sure	22 (8.3)
Total scoresa
0	43 (16.4)
1	36 (13.7)
2	52 (19.8)
3	87 (33.1)
4	45 (17.1)

Data are presented as No. (%). Numbers may not sum to total due to missing data.

^aResponses of “a lot,” “some,” and “yes” were scored as 1, and other responses were scored as 0. The rescored items then were summed.

Overall, 81 patients (30.7%) received some form of educational materials about LCS, whereas the remainder did not or were not sure. When considering their degree of involvement in making the screening decision with their health care provider, 226 patients (86.6%) indicated that they were involved as much as they wanted, whereas 23 patients (8.8%) were less involved and 12 patients (4.6%) were more involved. Finally, 83 patients (31.3%) indicated that their health care provider discussed the patient’s chances of lung cancer developing. When asked about the importance of having a conversation with a health care provider about the benefits and harms of LCS, 156 patients (59.1%) indicated that it was “very important,” 72 patients (27.3%) indicated that it was “somewhat important,” and 36 patients (13.6%) responded that it was “not overly important.” Discussions of smoking cessation or abstinence were noted by 227 patients (85.3%).

Decisional Conflict About the Screening Decision

Using the SURE scale, 174 patients (66.4%) reported low decisional conflict related to their choice to be screened (total scores in Table 4). Scores for the remaining 88 patients (33.6%) indicated that they were conflicted about their choice.

Table 4.

SURE: Patients’ Decisional Conflict Scores Related to Screening Decision (n = 266)

Item	Response
Did you feel SURE about the best choice for you?
Yes	223 (84.5)
No	12 (4.6)
Not sure	29 (11.0)
Did you feel that you knew the benefits and risks of each option?
Yes	194 (73.5)
No	31 (11.7)
Not sure	39 (14.8)
Were you clear about which benefits and risks matter most to you?
Yes	189 (72.1)
No	36 (13.7)
Not sure	37 (14.1)
Did you have enough support and advice to make a choice?
Yes	210 (79.6)
No	22 (8.3)
Not sure	32 (12.1)
Total scorea
0	35 (13.4)
1	15 (5.7)
2	14 (5.3)
3	24 (9.2)
4	174 (66.4)

Data are presented as No. (%). Frequencies do not sum to total because of missing data. SURE = Sure of myself; Understand information; Risk-benefit ratio; Encouragement scale.

^aTotal scores are sum of items where yes = 1 and no or not sure = 0. A score of 4 indicates no decisional conflict.

Patient Factors Related to Knowledge, Shared Decision-Making, and Decisional Conflict

Table 5 shows the distribution of patient characteristics for knowledge of LCS, adequacy of the SDM process, and decisional conflict of patients. In multivariate models adjusting for screening site and length of time since screening was completed, several patient factors were related to patients’ knowledge of LCS (Table 6). Patients who self-identified as non-White had lower percent of correct responses (7.9 percentage points) regarding the knowledge measured compared with White patients. Lesser education also was related to lower knowledge scores. Patient characteristics were not related to the Shared Decision Making Process measure when patients scoring the maximum (a score of 4) were compared with other patients. When a lower threshold was used to determine a high-quality shared decision-making process (scores of 3 or 4, compared with lower scores), two patient factors were significant. Non-White patients were more likely to report a higher-quality shared decision-making process than White patients (OR, 2.51; 95% CI, 1.03-6.10). Similarly, as the patient’s pack-year smoking history increased, the odds of reporting a higher-quality shared decision-making process also increased (OR, 1.02; 95% CI, 1.00-1.03). Finally, non-White patients compared with White patients (OR, 4.31; 95% CI, 1.36-13.70) and former smokers compared with current smokers (OR, 1.93; 95% CI, 1.04-3.55) were more likely to report being conflicted about their screening decision.

Table 5.

Patient Characteristics Related to Knowledge of Lung Cancer Screening, SDM Process, and Decisional Conflict for Current and Former Smokers Screened for Lung Cancer (n = 266)

	Knowledge of Lung Cancer and Screening		SDM Process		Decisional Conflict About Screening Decision
	No.	Percent Correct Responsesa	No.	Scores of 3 and 4b	No.	Scores of 1c
Screening settings
Academic medical center	34	34.2 ± 18.7	34	61.8	34	58.8
Cancer tobacco treatment program	127	43.3 ± 15.9	131	44.3	132	71.2
Cancer center lung cancer screening program	80	42.8 ± 17.3	82	56.1	82	67.1
Time since last screening, mo
<3	42	44.5 ± 19.4	42	64.3	43	76.7
3-6	72	42.3 ± 16.9	75	52.0	75	69.3
6-12	127	40.7 ± 16.2	130	45.4	130	64.6
Age, y
<65	123	40.1 ± 18.1	126	50.0	127	63.0
65+	118	43.7 ± 15.7	121	51.2	121	73.6
Sex
Male	126	42.1 ± 17.8	128	46.1	128	67.2
Female	115	41.6 ± 16.2	119	55.5	120	69.2
Race or ethnicity
White	214	43.0 ± 17.2	219	48.9	219	65.8
Other than White or Hispanic	27	32.9 ± 12.5	28	64.3	29	86.2
Education
High school graduate or less	44	34.7 ± 12.2	47	51.1	47	63.8
Some college or trade school	89	37.5 ± 16.5	88	47.7	90	72.2
College	108	48.4 ± 16.9	112	52.7	111	66.7
No. of times screened for lung cancer
1	73	38.9 ± 18.1	78	52.6	78	59.0
2	72	38.9 ± 16.6	73	45.2	72	72.2
3	96	46.4 ± 15.5	96	53.1	98	72.5
Current smoker
Yes	90	39.2 ± 16.0	93	48.4	94	59.6
No	151	43.5 ± 17.5	154	52.0	154	73.4
Pack-year smoking historyd
<40	113	41.4 ± 17.7	117	42.7	118	64.4
≥40	128	42.2 ± 16.4	130	57.7	130	71.5

Data are presented as mean ± SD or percentage, unless otherwise indicated. Sample size does not sum to 266 for some patient characteristics because of missing data. SDM = shared decision-making.

^aPercent correct of 16 total items.

^bScores of 3 or 4 indicate a good SDM process about lung cancer screening.

^cScores of 1 indicate some decisional conflict experienced by patient related to the screening decision.

^dTreated as continuous variable in regression models, but reported here using a cutpoints in this table to aid in interpretation.

Discussion

In this study of patients having completed LCS, we found that patients (1) valued finding cancer early over concerns about harms, (2) were varied in their knowledge of LCS, (3) indicated that the benefits of screening were addressed by their health care providers far more than harms, (4) generally reported not receiving or remembering receiving educational materials about LCS at the time they were screened, and (5) felt they were involved in decision-making as much as they wanted. Additionally, about one-third of patients reported some decisional conflict related to their screening decisions. Most patients (more than 85%) reported that a health care provider discussed smoking cessation or abstinence as part of their LCS experience.

Patients’ poor knowledge of LCS is concerning. Many incorrectly endorsed LCS, rather than smoking abstinence, as the most effective approach to lower lung cancer risk. Few recognized that most abnormal LDCT results do not result in a patient being diagnosed with lung cancer, suggesting an ongoing need to help patients understand the implications of screening results. Most patients incorrectly answered the eligibility criteria for LCS, which may not have been emphasized or discussed because of a variety of factors, including bias of the screening facility to offer screening; provider determined eligibility, provider skipped this discussion because time constraints, or both; or provider bias regarding which patients want to, should, or are able to engage in SDM.^8,17 Important patient subgroup differences also were noted, specifically, that non-White and less educated patients scored lower on LCS knowledge. Prior data suggest that those with the lowest level of formal education have the poorest understanding of LCS.¹⁸ It is the responsibility of providers to ensure that patients receive high-quality SDM and use tools at appropriate literacy levels. Strategies to enhance knowledge, and especially for those with the lowest education levels, should be explored further.

The quality of SDM conversations in this study was inconsistent, and few patients reported a discussion of the harms of screening with a health care provider. In prior screening studies, > 90% of providers discussed the benefits of screening, whereas < 30% discussed the potential harms of screening.^{18, 19, 20} In our study, some suggestion was found that non-White patients experienced a better SDM process, but these findings were not consistent across different thresholds for determining a high-quality process. It is concerning that some patients in this study were conflicted about their screening decision based on responses to the SURE questionnaire. Overall, SDM interventions have been shown to improve knowledge and reduce decisional conflict.²¹ Our study did not assess interventions before and after SDM to determine this, but overall results of knowledge and decisional conflict were poor. Future studies should consider the impact of decisional conflict on subsequent screening and follow-up of abnormal findings, in particular among non-White people and former smokers, because it is reasonable to expect that some patients will experience regret over their screening choice.

Interestingly, one-third of the patients did not indicate that the SDM process was important to them, although most answered that knowing the benefits of screening was important. Studies show that patients and providers overestimate benefit and underestimate harm of screening, which is consistent with our findings.^22,23 It is unclear whether the importance for knowing and higher knowledge regarding benefits by respondents in our study represent a selection bias of respondents, a reflection of provider bias conveyed during the SDM visit, a true recall bias phenomenon of screening surveys, or a combination thereof. Also, all patients in this study participated in SDM whereas in a study among Medicare beneficiaries, only 9.0% who underwent LCS had a prior visit with a SDM claim.²⁴

In our study, about a third of participants were told their risk of lung cancer developing. Providers are challenged with choosing among several risk assessment calculators with variable risk thresholds for screening. The United States Preventive Services Task Force is considering recommending use of three externally validated models: ≥ 1.3%-NCCN (National Comprehensive Cancer Network), ≥ 1.51%-PLCOM₂₀₁₂ (Prostate, Lung, Colorectal, and Ovarian modified model), and ≥ 1.2%-LCDRAT (Lung Cancer Death Risk Assessment Tool).²⁵ Additionally, competing provider and patient priorities, difficulty accessing decision aids, limited patient comprehension, and anticipated patient emotions all have been cited as barriers to realizing this component of SDM.^{26, 27, 28, 29} In this study, the reported low use of any educational materials including decision aids was not anticipated, given that this is a component of the SDM visit listed by the Centers for Medicare and Medicaid Services and the participating screening programs had decision aids available. Other studies have shown similarly poor use of decision aids despite wide availability.^5,19 Improving the use of decision aids that address individual risk assessments may balance patients’ knowledge of risks and benefits, may provide more accurate expectations of the LCS process, and may help patients to reach choices that are more consistent with their values.³⁰

Our study has some limitations. Patients were recruited from academic-affiliated practices and were located in only one geographic area. Compared with participants in the National Lung Screening Trial,^1,31 patients in this study were slightly older, included a larger percentage of women, were less likely to be current smokers, and had a much lower pack-year smoking history (56 pack-years vs. 40.4 pack-years, respectively), and more had a college education (44% vs 31%, respectively). Recall bias also is a concern, although we adjusted for timing of screening in the analyses. We do not have information about who conducted the SDM conversation and the settings for these conversations or the types of smoking cessation interventions the patients received. We did not collect data on screening results, and the impact of this information on the study outcomes is not known. The data reflect only patients who completed a screening examination. Data on patients who may have had an SDM discussion with a provider and opted not to be screened may differ from those who were screened.

Interpretation

In this study of patients recently screened for lung cancer, we found the quality of SDM to be highly variable. Overall, respondents were satisfied with their SDM experience, yet many were conflicted about the decision to be screened. An imbalance was found between provision of benefit and risk information and low use of educational materials including decision aids. A structured approach using decision aids may assist with providing a balanced, consistent presentation of information and could improve knowledge. The importance of emphasizing smoking cessation and abstinence with patients must remain a priority for clinicians.