Examining Whether Lung Screening Changes Risk Perceptions: NLST Participants at 1-year Follow-up

Abstract

Purpose

The National Lung Screening Trial (NLST) Research Team reported reduced lung cancer mortality among current and former smokers with a minimum 30-pack/year history, screened with spiral CT scans compared with chest x-ray. The objectives of this study are to examine, at one-year follow-up: 1) risk perceptions of lung cancer and smoking related diseases and behavior change determinants, 2) whether changes in risk perceptions differ by baseline screening result; and 3) whether changes in risk perceptions affect smoking behavior.

Methods

A 25-item risk perceptions sub-study questionnaire was administered to a subset of participants at 8 American College of Radiology Imaging Network (ACRIN)/NLST sites, prior to initial and one-year follow-up screens. Items assessed risk perceptions of lung cancer and smoking related diseases, cognitive and emotional determinants of behavior change, and knowledge of smoking risks.

Results

Among 430 NLST participants (M age=61.0, 55.6% male, 91.9% white), half were current smokers at baseline. Overall, risk perceptions, and associated cognitive and emotional determinants of behavior change, did not change significantly from prescreen trial enrollment to 1 year follow-up and did not significantly differ by screening test result. Changes in risk perceptions were not associated with smoking status changes (9.7% quitting, 6.6% relapse) at one-year follow-up.

Conclusions

Lung screening did not change participants’ risk perceptions for lung cancer or smoking related disease. A negative screening test, the most common result of screening, did not appear to decrease risk perceptions nor provide false reassurance to smokers.

INTRODUCTION

Each year in the U.S. over 220,000 individuals are diagnosed with lung cancer, and 87% of lung cancer deaths are attributable to smoking.¹ Smokers are over ten times more likely to develop lung cancer compared to nonsmokers.^{2, 3} The prognosis for patients with lung cancer is poor, largely because only 15% of lung cancer patients are diagnosed at an early stage of disease.¹ The National Lung Screening Trial (NLST) compared lung cancer mortality rates among high risk current and former smokers with a minimum 30-pack/year smoking history and found that low dose computed tomography (CT) screening reduced lung cancer mortality by 20% relative to screening with chest x-ray.⁴ The National Comprehensive Cancer Network (NCCN) recently released guidelines recommending annual low-dose helical CT screening for high-risk individuals.⁵

There is significant apprehension about widespread use of lung CT. Bach and colleagues, in a recent systematic review, concluded that CT screening significantly reduced lung cancer mortality but that uncertainty exists surrounding its potential harms.⁶ The cost-effectiveness of CT screening is estimated to be $126,000/QALY for current smokers and $169,000/QALY for former smokers,⁷ which is largely the result of the substantial number of false positives resulting in downstream costs and risks associated with additional work-ups. In the NLST, over three rounds of screening, 39.1% of the low dose helical CT screens and 16.0% of the chest X-ray screens received at least one positive result; over 90% of these were deemed false positives.⁴ Some have argued that funds for lung cancer prevention would be better spent on proven primary prevention efforts rather than screening.^8–10

The NCCN’s revised lung screening guidelines explicitly include smoking cessation counseling,⁵ which could add a meaningful risk reduction component. Even at the time of a lung cancer diagnosis, quitting smoking can improve treatment efficacy.^11–13 Undergoing screening might enhance abstinence motivation and quitting among current smokers and prevent relapse among former smokers, especially if the screening result is not normal. Alternatively, a negative CT screening test might falsely reassure smokers and make them less motivated to quit smoking.¹⁴ The effect of undergoing CT screening, on smoking behaviors, remains inconclusive; some studies demonstrate higher than expected quit rates and others show little or no change in smoking.^{14– 16}

Thus, lung screening, and its result, may influence smokers’ intentions and motivation to quit,^14,17–18 and risk perception changes may mediate cessation motivation and behavior change.¹⁹ Park and colleagues demonstrated that, prior to their initial screen, NLST participants’ risk perceptions were associated with their behavioral intentions. Specifically, among current smokers, risk perceptions correspondingly increased with intentions to quit smoking; among former smokers, risk perceptions were lower among those who were committed to remain abstinent. Individuals’ perceptions of their risk of lung cancer and other smoking related diseases (SRDs) may be affected by screening or its results; recent studies suggest that lung screen results change short-term lung cancer risk perceptions by increasing risk perceptions for individuals who have non-negative tests.^{20, 21}

The sustained effect of lung screening on the effect of risk perceptions, and in turn, smoking behavior changes, is unclear. Thus, our aim was to assess the effects of lung screening and test result, on risk perceptions that might underlie smoking behavior changes one-year following an initial lung screen. We hypothesized that the process of undergoing screening would increase risk and thus promote abstinence, particularly among participants receiving a positive baseline test result. A theoretically-based risk perception questionnaire²² was administered to a subset of NLST participants prior to their initial and 1 year screens, providing an opportunity to assess changes in risk perception and smoking behavior and the impact of baseline screen results on these constructs. This study contributes to past research on risk perceptions in the context of lung screening in that: 1) our findings are based on a comprehensive risk measure that examines both lung and SRD risk perceptions; 2) we include an examination of cognitive and emotional behavior change constructs; and 3) we present longitudinal data on risk perception changes and smoking outcomes.

METHODS

Study Design

This longitudinal study examined changes in lung cancer and SRD risk perceptions among NLST participants, and the effects of change in risk perceptions and other theoretically-based behavior change determinants on smoking behaviors, from baseline to one-year follow-up. The NLST is a collaboration between ACRIN and the NCI Lung Screening Study.⁴

Cohort and data collection

At the time of recruitment into the trial, participants were 55–74 years of age, current or former (quit within 15 years) smokers with a history of 30-pack years minimum, and had no history of lung cancer. 23 ACRIN study sites participated in the NLST/ACRIN arm of NLST; 8 sites participated in the sub-study. From 12/03–3/04, participants at the 8 sites were offered the opportunity to complete the risk perception sub-study questionnaire at the time of baseline and 1 year follow-up. At baseline, 630 participants completed the survey; of these, 430 (68%) also completed the survey prior to their 1 year follow-up screen. Nearly 90% of participants who completed the ACRIN/NLST follow-up forms also completed the risk perception questionnaire. Baseline smoking status and recruitment site significantly differed for noncompleters vs. completers; significantly more non-completers were current smokers.

Measures

Baseline sociodemographic, medical and smoking history variables

The trial baseline questionnaire²³ included measures of sociodemographics, medical history (family history of lung cancer, personal history of cancer or SRD), smoking history (smoking status (“Do you smoke now?”), number of years smoked, number of years quit, average number of cigarettes smoked per day, and nicotine dependence during “the time period when you smoked the most” (FTND²⁴; score range = 0–10; low to high dependence)). Baseline screen results were categorized as 1) negative, defined as having no major abnormalities, 2) significant abnormalities not suspicious for lung cancer, or 3) positive, defined as showing a nodule or other abnormality suspicious for lung cancer.

Follow-up Smoking Outcome

One year follow-up questionnaires²³ collected information on smoking status (“Do you now smoke cigarettes (one or more cigarettes per week)?”). Self-reported smoking status is a reliable means of assessing smoking status among participants undergoing lung screening²⁵

Risk perception sub-study questionnaire

Details on the questionnaire were published previously.²² In brief, using behavioral change constructs from the Health Belief Model,²⁶ the Precaution Adoption Process Model²⁷, the Self Regulation Model,²⁸ and the work of Kreuter & Strecher²⁹, we developed a questionnaire.²²

Smoking Risk Perceptions Scale (SRPS)

Ten items assessed individual’s personal (individual risk), and comparative risk (risk related to others) for lung cancer and other SRDs. Given that lung screening may detect lung cancer and/or a SRD, risk perceptions for lung cancer and SRDs were analyzed separately, by two 5-item scales. To assess lung cancer and SRD risk, two questions assessed the likelihood (‘very unlikely’ to ‘very likely’) and danger (‘strongly disagree’ to ‘strongly agree’) of developing lung cancer and SRD, and three questions, using the referent groups average person, others of the same age and sex, and other former/current smokers, assessed comparative risk (lung risk 5-item scale alpha= .87; 70% variance; SRD risk 5-item alpha= .86; 66% variance).

Cognitive and emotional determinants of smoking cessation

Self-efficacy was measured by one question about confidence to quit smoking/remain quit (5-point Likert scale, ‘not at all’ to ‘extremely’). Perceived benefits of quitting smoking/staying quit were measured by 3 questions about the benefits of quitting in terms of decreasing risk for lung cancer, other SRDs, and increasing life expectancy (4-point Likert scale, ‘not at all’ to ‘very much’); these items were combined to create a composite score (range = 3–12; alpha=.89). Perceived severity was assessed by 4 questions about the health consequences and severity of lung cancer and other SRDs (5-point Likert scale, ‘not at all’ to ‘extremely’); these items were combined to create a composite score (range 4–16; alpha=.85). Perceived benefits of lung screening were assessed by two questions about the curability and benefits of lung cancer detected by screening (4-point Likert scale ‘very few’ to ‘most,’ range 1–4). Worry about lung cancer and other SRDs was assessed by 4 questions about intensity (4-point Likert scale, ‘not at all’ to ‘extremely’) and frequency of worry (4-point Likert scale, ‘not at all’ to ‘all of the time’); items were combined to create a composite score (range = 4–16; alpha=.89).

Knowledge

Knowledge of smoking risks was assessed by 2 questions asking participants to quantify a smoker’s risk of developing lung cancer (“A smoker who smokes one pack of cigarettes a day is at how many more times risk of developing lung cancer, compared to a non-smoker?” and “Among 100 smokers, how many will get lung cancer because they smoke?”).

Analyses

All statistical analyses were conducted using SAS/STAT version 9.2. Complete case analyses were run. To ensure that the substudy participants were balanced across randomization arms, we compared baseline characteristics by study arm. Baseline findings demonstrated that former and current smokers significantly differed on sociodemographic factors, smoking characteristics, and risk perceptions,²² so all analyses were conducted separately by smoking status. Thus, study arm and baseline smoking status group differences were assessed using Chi Square test/Fisher’s Exact tests for categorical variables, and two sample t-tests for continuous variables separately for former and current smokers. Mean differences and 95% confidence intervals were calculated to compare baseline to 1 year changes in risk perceptions, cognitive and emotional behavioral change determinant constructs, and knowledge. Separately for former and current smokers, one-way ANOVAs were conducted to assess the changes in lung cancer and SRD risk perceptions across the categories of baseline screen results. Logistic regression models, separate for current and former smokers, examined the effect of baseline to 1 year follow-up changes in risk perceptions, for lung cancer and SRDs, on 1 year smoking status; unadjusted and adjusted analyses (adjusting for study arm, baseline test result, gender, and race) were conducted.

RESULTS

Participant characteristics

Approximately half of participants were current smokers (Table 1). Baseline participant characteristics and risk perception outcomes did not differ significantly by study arm. Current and former smokers significantly differed on many sociodemographic and smoking characteristics. Former smokers had been quit of average of nearly 7 years; however, former smokers reported a higher average number of cigarettes per day (29.77) compared to current smokers (26.27). Baseline screen results were negative for 82.49% of current smokers and 87.32% of former smokers; negative for lung cancer, but positive for other lung abnormalities in 7.83% of current smokers and 3.29% of former smokers; and positive for suspected lung cancer in 9.68% of current smokers and 9.39% of former smokers. At 1 year follow-up, no participants had a confirmed lung cancer diagnosis.

Table 1.

Baseline characteristics: sociodemographic, smoking and medical

	Current Smoker (N=217)	Former Smoker (N=213)	P Value
Sociodemographic characteristics Age-Mean (SD)	60.30 (4.40)	61.80 (5.24)	0.001
Gender-N (%)      Male	106 (48.85)	133 (62.44)	0.01
Female	111 (51.15)	80 (37.56)
Education-N (%)      < High School	15 (6.91)	12 (5.63)	0.25
High school graduate/GED	75 (34.56)	54 (25.35)
Post high school training, some college	70 (32.26)	79 (37.09)
> College	50 (23.04)	61 (28.64)
Other/Unknown	7 (3.23)	7 (3.29)
Household Income-N (%)      < $15,000	30 (13.82)	13 (6.10)	<0.01
$15,000–34,999	58 (26.73)	40 (18.78)
35,000–64,999	55 (25.35)	71 (33.33)
$65,000 and above	40 (18.43)	30 (14.08)
Unknown	34 (15.67)	59 (27.70)
Marital Status-N (%)      Married/Living as Married	122 (56.22)	154 (72.30)	<0.01
Widowed/Divorced/Separated	85 (39.17)	50 (23.47)
Never Married	10 (4.61)	8 (3.76)
Unknown	0 (0.00)	1 (0.47)
Race-N (%)      White	192 (88.48)	203 (95.31)	0.02
Black or African-American	23 (10.60)	8 (3.76)
Other	2 (0.92)	2 (0.94)
Smoking characteristics Pack years-Mean (SD)	55.12 (20.71)	54.93 (23.73)	0.93
# Years Smoking-Mean (SD)	42.15 (6.07)	37.50 (7.08)	<0.01
# Years Quit-Mean (SD)	N/A	6.88 (4.85)
Average # of Cigarettes per Day-Mean (SD)	26.27 (9.21)	29.77 (12.24)	<0.01
24-hour quit attempts in the past year-N (%)      None	108 (49.77)	N/A
1 or more	88 (40.55)	N/A
Unknown or N/A	21 (9.68)	N/A
FTND Score*-Mean(SD)	5.41 (2.23)	5.20 (2.61)	0.38
Medical characteristics Had family history of lung cancer-N (%)	50 (23.04)	39 (18.31)	0.23
Had personal history of cancer-N (%)	22 (10.14)	23 (10.80)	0.82
Had personal history of SRD-N (%)	57 (26.27)	44 (20.66)	0.17
Baseline screening result-N (%)      Negative	179 (82.49)	186 (87.32)	0.12
Significant for non-lung cancer abnormalities	17 (7.83)	7 (3.29)
Positive	21 (9.68)	20 (9.39)

Note:

^*Fagerstrom Test of Nicotine Dependence. Nicotine dependence during “the time period when you smoked the most” (FTND²⁴; score range = 0–10; low to high dependence

Changes in risk perception and other determinants of smoking cessation

Risk perception of lung cancer and SRDs are listed in Table 2. Lung cancer risk perceptions did not change significantly from baseline to 1 year follow-up (current smoker mean = 17.91 vs. 18.09, former smoker means= 16.28 at baseline and follow-up). Similarly, SRD risk perceptions did not change over time (means = 18.52 vs. 18.76 for current smokers, 16.76 vs. 16.86 for former smokers). In terms of determinants of smoking behavior change, cognitive and emotional constructs (perceived confidence, perceived disease severity, perceived benefits of quitting and lung screening, and worry), and knowledge of smoking risks did not change. The only factor that changed significantly was an elevated sense of the perceived benefits of quitting smoking among former smokers (Mean change= 0.43; CI = 0.20, 0.67).

Table 2.

Changes in risk perception constructs, from baseline to 1 year follow-up

	Current Smoker at Baseline						Former Smoker at Baseline
	N*	Baseline, mean (SD)	12 months, mean (SD)	Mean change	95% CI	N*	Baseline, mean (SD)	1 year, mean (SD)	Mean change	95% CI
Risk Perceptions
Lung cancer risk (5–25)	215	17.91 (3.93)	18.09 (3.85)	0.18 (0.21)	(−0.23,0.59)	212	16.28 (3.80)	16.28 (3.96)	0.00 (0.22)	(−0.43, 0.43)
SRD risk (5–25)	215	18.52 (4.06)	18.76 (4.05)	0.25 (0.25)	(−0.25, 0.74)	212	16.76 (4.00)	16.86 (4.29)	0.09 (0.26)	(−0.43, 0.62)
Cognitive and emotional constructs
Perceived benefits of quitting smoking (3–12)	207	9.11 (2.10)	9.00 (2.12)	−0.09 (0.15)	(−0.39, 0.21)	205	9.75 (1.99)	10.23 (1.82)	0.43 (0.12)	(0.20, 0.67)
Confidence to quit smoking/ remain quit (1–5)	203	2.71 (1.12)	2.84 (1.21)	0.13 (0.08)	(−0.04, 0.29)	192	4.66 (0.74)	4.56 (0.85)	−0.09 (0.06)	(−0.22, 0.03)
Perceived severity of lung cancer and SRDs (4–20)	193	18.24 (2.12)	18.26 (1.98)	−0.01 (0.13)	(−0.26, 0.24)	196	18.55 (1.70)	18.28 (2.11)	−0.22 (0.13)	(−0.47, 0.03)
Perceived benefits of lung screening (1–5)	207	4.83 (1.44)	4.88 (1.45)	0.05 (0.09)	(−0.14, 0.23)	207	4.94 (1.41)	4.91 (1.36)	−0.02 (0.12)	(−0.25, 0.20)
Worry about lung cancer and SRDs (4–16)	192	10.09 (2.84)	10.23 (2.81)	0.14 (0.16)	(−0.18, 0.46)	200	9.04 (2.72)	9.13 (2.52)	0.10 (0.16)	(−0.21, 0.40)
Knowledge of smoking risks
Percent of smokers who will get lung cancer (0–100%)	184	37.23 (23.39)	36.58 (21.97)	−0.65 (1.32)	(−3.25, 1.95)	194	40.82 (22.54)	42.16 (22.27)	1.34 (1.67)	(−1.95, 4.63)
One pack/day smokers risk of developing lung cancer	207	3.04 (1.08)	3.06 (1.11)	0.02 (0.07)	(−0.11, 0.16)	204	3.17 (0.99)	3.26 (0.96)	0.10 (0.08)	(−0.05, 0.25)

^*Participants with a missing baseline and/or 1 year construct are not included in the table for that construct.

Changes in risk perception by lung screen results

Mean changes in risk perception, from baseline to 1 year follow-up, across baseline smoking status and baseline screening result, are shown in Table 3. The change in current and former smokers’ risk perceptions of lung cancer did not differ significantly by baseline screening test results (Mean change =0.17 and −0.08 (negative), 0.71 and 0.86 (significant for non-lung abnormalities) −0.14 and 0.40 (positive). Similarly, the change in current and former smokers’ risk perceptions of SRDs also did not differ significantly by baseline screening test result (Mean change =0.25 and −0.01(negative), 0.41 and 2.00 (significant for non-lung abnormalities) and 0.10 and 0.40 (positive).

Table 3.

Changes in risk perceptions, from baseline to 1 year follow-up, by baseline screen results

Baseline Screening Result for Current Smokers
	Negative N=177		Significant for non-lung cancer abnormalities N=17		Positive N=21
	Mean change	95% CI	Mean change	95% CI	Mean change	95% CI	P-value
Changes in Lung Cancer Risk	0.17	(−0.29, 0.63)	0.71	(−0.83, 2.25)	−0.14	(−1.45, 1.16)	0.69
Changes in Smoking-Related Disease Risk	0.25	(−0.30, 0.80)	0.41	(−1.10, 1.92)	0.10	(−1.74, 1.93)	0.97
Baseline Screening Result for Former Smokers
	Negative N=185		Significant for non-lung cancer abnormalities N=7		Positive N=20
	Mean change	95% CI	Mean change	95% CI	Mean change	95% CI	P-value
Changes in Lung Cancer Risk	−0.08	(−0.53, 0.38)	0.86	(−1.73, 3.44)	0.40	(−1.14, 1.94)	0.62
Changes in Smoking-Related Disease Risk	−0.01	(−0.56, 0.54)	2.00	(−2.47, 6.47)	0.40	(−1.61, 2.41)	0.37

Changes in risk perception as predictors of smoking status at 1 year follow-up

Changes in smoking status were observed from baseline to 1 year follow-up among participants with a known smoking status at 1 year follow-up; 9.7% of baseline current smokers reported quitting smoking, and 6.6% of baseline former smokers reported smoking resumption (data not shown). Table 4 shows that among current smokers at baseline, smoking status at 1 year was not significantly associated with changes in risk perceptions of lung cancer (OR = 1.09, CI=.93,1.26) or SRD (OR=1.05, 95% CI=0.93, 1.19) Among former smokers at baseline, smoking status at 1 year was not significantly associated with changes in risk perceptions of lung cancer (OR=1.07, CI=0.90, 1.28) or SRD (OR=0.98, CI=0.85, 1.13).

Table 4.

Changes in risk perception as predictors of smoking at 1 year follow-up

MODELING CONTINUED SMOKING	Odds Ratio	95% Confidence Interval
Current smokers at baseline	(n=215)
Changes in lung cancer risk perceptions	1.09	(0.93, 1.26)
Changes in SRD risk perceptions	1.05	(0.93, 1.19)
MODELING RELAPSE	Odds Ratio	95% Confidence Interval
Former smokers at baseline	(n=212)
Change in lung cancer risk perceptions	1.07	(0.90, 1.28)
Changes in SRD risk perceptions	0.98	(0.85, 1.13)

Note: Analyses were run for unadjusted and adjusted (study arm, gender, and race) models; findings in models did not differ. Unadjusted models are presented.

DISCUSSION

In this longitudinal observational study of a subset of NLST participants followed for 1 year after their initial lung screen, we examined the impact of lung screening on perceived lung cancer risks and SRD risks and how changes in risk perceptions affected smoking behavior. In a subset of NLST participants, we conclude that undergoing screening per se did not have a significant impact on risk perceptions or sustained smoking behavior change.

Overall, risk perceptions for lung cancer and SRDs did not change at 1 year following lung screening. Our findings were similar to work conducted by Vierikko³⁰ and Sinicrope²⁰ which determined that there were no significant changes in perceived lung cancer risk by 6–12 month following lung cancer screening test. Sinicrope also reported that perceived lung cancer risk rose one month following the screening. It is therefore possible that there might have been an alteration in risk perception immediately following screening, however, if this is so, our results indicate that this elevation is not sustained. Indeed, essentially no theoretically-based determinants of smoking behavior change or knowledge were significantly altered, with the exception of an elevated sense of the perceived benefits of quitting smoking among former smokers.

Twelve month follow-up self-reported quit rates (9.7% for current smokers at baseline) were somewhat higher than general population annual quit rates;³¹however, this is to be expected considering that participants who enroll in screening trials are highly motivated to quit smoking and maintain abstinence.^32,33 Quit rates, of current smokers at baseline and relapse rates of former smokers at baseline, were relatively similar to rates reported in recent screening studies.^{17, 32, 34}

Importantly, a negative initial screen did not appear to decrease current or former smokers’ risk perceptions, indicating that a negative test does not provide false reassurance. This is very important considering that, in real-world clinic settings; most lung screen results will be negative. These findings are concordant with recent research examining the effect of repeated negative scans on smoking behaviors. Townsend and colleagues reported that over a 3-year period repeated positive scans were associated with increased rates of smoking cessation.¹⁴ Similarly, Anderson and colleagues reported that over a 6-year period consistently negative scans were not associated with a lower likelihood of prolonged abstinence among smokers or a higher likelihood of relapse to smoking among former smokers.¹⁶

Some study limitations must be noted. First, the generalizability of these results is limited by a racially homogenous group of individuals who elected to participate in a screening trial at the 8 sites participating in this sub-study. Individuals who pursue screening based on a physician recommendation are likely to differ; indeed, individuals who have participated in lung screening studies have reported higher levels of risk perceptions^33,35 and quitting motivation.³² Second, we were not able to collect data on risk perceptions immediately following lung screening to determine if there were short-term transient changes in these outcomes. Regardless, our results indicate that, had there been any proximal changes, they were not sustained. Lastly, we were likely underpowered to be able to detect significant differences in risk perceptions by baseline screen test result.

In summary, it appears that undergoing lung cancer screening does not by itself modify risk or other underlying determinants of smoking behavior change. At the initial screening, risk perceptions were associated with smoking intentions,²² but by 1 year follow-up risk perceptions were not predictive of smoking behavior change. We cannot rely on the screening itself to alter risk, but rather need to intervene to leverage high risk individuals’ intentions to quit smoking. Given the high cost of screening, which increases if cessation rates are unaffected by screening, it is critical that tobacco treatment is integrated into the screening process.