Subjects and setting
Potential participants were identified from a prospective database of ≥5 year lymphoma survivors (ADAPT) held at The Christie NHS Foundation Trust. HL survivors were eligible for the study if they were currently aged 18-80 and had been treated with chemotherapy regimens that contained alkylating agent/s associated with excess risk of SMN, and/or radiotherapy where the breast/s, lung/s or bowel were included in the radiation field. We excluded those with a diagnosis of dementia, learning difficulties or a SMN for which they were undergoing palliative treatment. A postal questionnaire and participant information sheet was sent to 281 eligible individuals followed by reminder letters to those who had not returned the questionnaire within three weeks. Return of the questionnaire was taken as consent to participate.
Study questionnaire development
The study questionnaire was designed for use in this study, but used or adapted a number of published scales, as described below. The questionnaire can be found in Supplementary data.
Willingness to undergo lung cancer screening
Participants were asked to rate the strength of their willingness to participate in a future lung cancer screening programme with the question ‘If you were invited to go for a lung cancer screening test, would you go?’ The response options were ‘yes definitely’, ‘yes probably, ‘probably not’ and ‘definitely not’.
Lung cancer screening related health beliefs
Lung cancer screening related health beliefs were measured using the Lung Cancer Screening Health Belief Scales (LCSHBS), developed to measure health beliefs impacting lung cancer screening uptake using the HBM framework and psychometrically tested in ever smokers.  The LCSHBS comprise of four scales measuring perceived risk of developing lung cancer and perceived benefits, perceived barriers and self-efficacy (an individuals’ belief in their capacity to execute a behaviour) for undergoing lung cancer screening. Although the Extended Health Belief Model includes separate constructs for perceived risk and perceived severity, the LCSHBS do not include a perceived severity scale because cancer is always perceived to be severe. To adapt the LCSHBS for this study population, items relating to cost, lack of a regular healthcare provider and booking a scan appointment were removed and never smokers were instructed not to complete the items in the perceived barriers scale which relate to a personal history of smoking. Prior to completing the scales, participants were provided with a short statement describing a lung cancer screening test.
Items in the perceived risk, perceived benefits and perceived barriers scales were scored using 5-point Likert scales indicating agreement (strongly agree/agree/neither agree nor disagree/disagree/strongly disagree). The self-efficacy scale had a 4-point Likert scale indicating level of confidence (very confident/somewhat confident/slightly confident/ not at all confident). The following are examples of items included in the scales: ‘It is likely that I will get lung cancer in the next five years’ (perceived risk scale); ‘Having a lung scan would lower my chances of dying from lung cancer’, ‘Having a lung scan would help me plan for the future’ (perceived benefits scale); ‘I might put off a lung scan because no one in my family had lung cancer’, ‘I might put off having a lung scan because I think I am too old to benefit from screening for lung cancer’ (perceived barriers scale; ‘How confident are you that you could find transportation to get to and from the clinic/hospital to have a lung scan?’, ‘How confident are you that you could get a lung scan even if you were anxious about the results?’ (self-efficacy scale).
Cronbach’s alpha was used to estimate internal consistency for each of the LCSHBS subscales and was found to be .90 for the 3-item perceived risk scale, .84 for 6–item perceived benefits scale, .89 for the 7-item self-efficacy scale, .94 for the 15-item perceived barrier scale for ever smokers and .91 for the 12-item perceived barrier scale for never smokers.
Participants’ age, gender and full postcode were extracted from electronic medical records. The questionnaire included questions about ethnicity, current employment status and level of education.
Other psychosocial and health related factors
Cancer worry was measured using an item adapted from the Cancer Worry Chart  which is considered to measure cancer worry severity: ‘In the last 4 weeks, how often were you bothered by thoughts or worry about your chances of getting cancer again in the future?’ (response options not at all / slightly / moderately / quite a bit / extremely). Dispositional optimism was measured using the Revised Life Orientation Test (LOT-R),  in which a higher score represents a higher level of dispositional optimism. Self-rated health was measured with a single item taken from the SF-12 Health Survey.  Optimistic bias was measured using an existing question relating to developing melanoma  adapted for this study: Compared to the average person of your age and sex, how likely is it in your opinion that you will develop [lung] cancer? (response options: much less likely / a bit less likely / about the same / a bit more likely / much more likely / I don’t know). We developed items to measure presence of a close family history of lung cancer (in parents or siblings), prior uptake of breast or bowel cancer screening and prior knowledge of lung cancer as a late effect of HL treatment. To investigate 6-year lung cancer risk values in our participants, demographic and lung cancer risk factor data were entered into the PLCOall2014 (Prostate, Lung, Colorectal, Ovarian) lung cancer risk calculator. The PLCOall2014 calculator is analogous to the PLCOm2012 calculator designed for ever smokers which is currently used to determine eligibility to undergo lung cancer screening in the UK, however PLCOall2014 also calculates 6-year lung cancer risk in never smokers. When it was used to calculate 6-year lung cancer risk in 65,711 never smokers in the PLCO cohort, the maximum risk observed was 1.47% falling below the ≥1.51% risk threshold for screening. 
Descriptive statistics were used to analyse the demographic and clinical characteristics of participants at risk of lung cancer, their knowledge of lung cancer risk, cancer screening behaviours and future lung cancer screening willingness and responses to the LCSHBS.
The demographic characteristics of participants versus non-participants were compared using Chi-squared test for gender and Mann-Whitney U-test for age and Index of Multiple Deprivation (IMD) decile. To identify the psycho-social factors associated with lung cancer screening hesitancy - defined as those responding ‘yes probably’ or ‘probably not’ to the lung cancer screening willingness question - a binary logistic regression analysis was performed. The dependent variable was screening willingness and participants for whom complete data was available for the independent variables were included in the analysis. Independent variables included socio-demographics, psychological variables (cancer worry and LOT-R scale score) and LCSHBS scores. Independent variables were entered into the multivariable logistic regression model regardless of whether they were associated with screening hesitancy on univariate analysis.
For the logistic regression, LCSHBS scoring for perceived risk, perceived benefits and perceived self-efficacy was reversed so that higher scores represented lower risk perception, lower perceived benefits and lower-self-efficacy. Scores for perceived barriers were retained so that higher scores represented higher perceived barriers. This change was made because we hypothesised that higher perceived barriers would increase screening hesitancy, whilst higher perceived risk, benefits and efficacy scores would reduce hesitancy. IMD decile was categorised as low (deciles 1-5) or high (deciles 6-10). A P value < .05 (two-tailed) was considered statistically significant for all analyses. Statistical analyses were performed using SPSS 23.0 (IBM, Chicago, IL)