Abstract

Introduction: Recent research has suggested that lung cancer screening might reduce lung cancer mortality and that significant numbers of participants quit smoking after screening. Concerns have been raised about the validity of self-reported smoking measures among cancer screening participants who may be motivated to underreport smoking. However, only one lung cancer screening study has biochemically verified self-reported smoking behavior.

Methods: This study validated self-reported smoking prevalence against urinary cotinine levels measured with the NicAlert® strip (Nymox, Maywood, NJ). The sample consisted of 55 consecutive participants enrolled in the Jewish Hospital Lung Cancer Screening & Early Detection Study (JHLCSEDS), a randomized trial comparing spiral CT and chest X-ray for lung cancer screening. Study participants were a mean of 59 years of age, and were predominantly Caucasian (96%) and male (55%).

Results: Using urinary cotinine levels for determination of nicotine exposure (cut-off=100 ng/ml), sensitivity and specificity of self-reported smoking status were 91% and 95%, respectively. Total misclassification rate was 7% (4 participants). However, three misclassified participants reported current use of nicotine replacement. Eliminating these cases from the analysis revealed sensitivity of 100% and specificity of 95% with only one misclassified case (a false positive).

Summary: In conclusion, results suggest that participants in lung cancer screening trials provide valid reports of smoking status.

Introduction

• Early detection offers the best opportunity for lung cancer survival among high-risk individuals who smoke.
  
• Studies suggest that participation in lung cancer screening is associated with increased rates of smoking cessation after screening.
  
• However, previous studies of smoking behavior among participants in lung cancer screening have relied solely on self-reports of smoking behavior.
  
• Although studies of adult smokers have generally demonstrated high concordance between self-reported smoking and biochemical verification, self-reported smoking behavior could be less accurate among high-risk participants in lung cancer screening, who may be motivated to under-report smoking.

Purpose and Hypotheses

• The aim of this research study is to examine concordance between self-reported smoking behavior and a biochemical “gold standard” measure of smoking behavior in a sub-sample of study participants in a randomized trial of lung cancer screening.
  
• Hypothesis I: Self-reported smoking behavior will be concordant with the biochemical index of smoking.
  
• Hypothesis II: Responses to two self-reported smoking behavior questionnaires will be concordant.
  

Procedure

Participants were enrolled in the JHLCSEDS parent study. Only those due for annual screenings in 9/04 were eligible for this sub-study. Parent study eligibility criteria are summarized below:

Inclusion Criteria
Current or former smokers
Between 40 - 70 years old
≥ 40 pack-year cumulative smoking history
Significant baseline airflow obstruction (FEV1/FVC ratio of < 70% or FEV1 of < 80% predicted normal values)
Baseline abnormal or stable chest X-ray

Exclusion Criteria
Existing disease at time of JHLCSEDS enrollment limiting expected life span to < 5 years
History of cancer (except non-melanotic skin cancer or cervical carcinoma in situ)
Pregnancy
Presence of symptoms of lung cancer at time of enrollment (e.g., significant weight loss or unexplained hemoptysis)

• Sub-study information mailed to all 104 JHLCSEDS participants due for 9/04 screening, followed by phone contacts to determine interest in participation
  
• Urine samples immediately analyzed using NicAlert® strip (Nymox, Maywood, NJ) for urinary cotinine levels
  
• Self-reported smoking status abstracted from parent study annual questionnaire, previously completed by each participant
  
• 55 consecutive interested participants met with research assistant immediately before annual screening to provide informed consent, complete sub-study questionnaire, and provide urine sample ($20 gift certificates issued): Total N = 55
  

Measures

• Prior to sub-study participation, each participant had already completed an Annual Survey of Smoking Behavior (ASSB) as part of the JHLCSEDS parent study. A single item – self-report of smoking status – was used to compare self-reported smoking status before and after being informed of the purpose of the sub-study.

• For the sub-study, each participant completed the Current Sub-study Smoking Behavior questionnaire (CSSB) prior to biological sample collection, including items measuring:

    (1) readiness to quit
    (2) smoking point prevalence (24 hr, 7 day, 30 day)
    (3) number of cigarettes smoked per day
    (4) use of other forms of tobacco
    (5) environmental smoke exposure
    (6) nicotine replacement usage
    (7) employment-related tobacco exposure

• Finally, after completing the CSSB, each participant provided a urine sample for cotinine analysis using NicAlert® strips.
  

Sample Description (N = 55)
Sex
• 54% Male
• 46% Female

Screening Group
• 67% Spiral CT
• 33% Chest X-ray

Race
• 95% Caucasian
• 2% African American
• 2% Hispanic
• 2% Other

Mean Age (SD) & Range
• 59 (8.4) / 43 – 73

Smoking Status on ASSB
• 58% current smokers
• 6% quit < 6 months ago
• 36% quit > 6 months ago

Current Smokers’ Plans to Quit Smoking
• N = 32
• 28% Plan to quit within 30 days
• 47% Plan to quit within 6 months
• 25% Not considering quitting

Current Smokers’ Daily Cigarettes Smoked
• N = 32
• Median = 16
• Range = 1 - 50

Reported Use of Other Tobacco
• 5% of smokers (3 report cigars, 1 “other”)
• 0% of ex-smokers

Exposure to 2nd Hand Smoke in Past 24 Hours
• 66% of smokers
• 30% of ex-smokers

Current Use of Nicotine Replacement Therapy
• 5% of smokers (1 gum, 1 patch, 1 “other”)
• 13% of ex-smokers (1 gum, 2 lozenges)

Results

CSSB Compared to ASSB:
100% Agreement in Sub-Study vs. Annual Self-Reports (kappa = 1.00, p < .001)

Cotinine & Self-Reported Smoking Status

• Sensitivity of Self-Report: 86%
  
• Specificity of Self-Report: 95%
  
• Positive Predictive Value of Self-Report: 97%

• Negative Predictive Value of Self-Report: 87%

(assuming smoking prevalence in the lung cancer screening population is similar to the prevalence in this study)

The 3 cases identified as positive by the cotinine test but negative by self-report were found to be non-smokers using nicotine replacement therapy, and thus were erroneously treated as smokers above. Excluding these 3 cases from the analysis resulted in the following:

• Sensitivity of Self-Report: 100%
• Specificity of Self-Report: 95%
  
• Positive Predictive Value of Self-Report: 97%
  
• Negative Predictive Value of Self-Report: 100%
  

(assuming smoking prevalence in the lung cancer screening population is similar to the prevalence in this study)

Conclusions

• Self-reported smoking behavior among participants in a randomized trial of lung cancer screening exhibited a high level of agreement with smoking status ascertained by urinary cotinine levels.
  
• Sensitivity of self-reported smoking status measured against biochemical verification ranged from 91% to 100%. Specificity ranged from 95% to 100%. These results were consistent with previous research (e.g., Jarvis et al., 1987).
  
• The two separate self-reports of smoking behavior (ASSB and CSSB, given before and during sub-study participation, respectively) were perfectly consistent.
  
• Extensive use of cotinine measurements for biochemical verification of smoking status may not be justified, as the validity of self-reported smoking status is consistently high in population-based studies (Rebagliato, 2002). Validity of self-reported smoking status was supported in this study with a high-risk sample who may have been motivated to under-report smoking.

Limitations

• The sample for this study was relatively small, consisting of only 55 of the 1000+ participants from the parent study of lung cancer screening.
  
• A convenience sample of 55 consecutive participants was utilized, rather than a random sample of participants from the parent study.
  
• It is possible that participation in the JHLCSEDS parent study may have increased the likelihood of accurate self-report among study participants.