2.1 Overview
The INSPIRE-Lung Study is designed as a randomized, controlled, community-based trial with two parallel groups and a primary endpoint of lung screening uptake by 6 months post intervention. Randomization will be performed as block randomization with a 1:1 allocation. See Fig. 1 for Participant Timeline. This study has two components: (1) to assess the ability of FBTA to reach high-risk individuals eligible for lung screening; and (2) to examine the comparative-effectiveness of LungTalk and a non-tailored lung screening information video in a national sample of screening-eligible, community-based individuals using an RCT design. Reach is defined as the absolute number, proportion, and representativeness of participating individuals assessed for lung screening knowledge, awareness and uptake, and reasons why or why not. Effectiveness is defined as increased knowledge, decreased perceived barriers to lung screening, occurrence of a patient-clinician discussion about the option to screen, and screening uptake, if the decision is to screen. Potential moderators of effectiveness (i.e., smoking status, gender, age, family history of lung cancer, provider recommendation, stigma, mistrust, fatalism, fear, worry, lung screening health beliefs) will also be assessed. This study was approved by the Institutional Review Board of Hackensack Meridian Health (IRB Protocol #: Pro-2022-0860). In addition, all methods will be performed in accordance with the guidelines and ethical principles that are fundamental to human subject protection and electronic written informed consent will be obtained from all study participants.
2.2 Study Setting
We will leverage FBTA to recruit community-based lung screening-eligible individuals. Using the Centers for Disease Control and Prevention Smoking and Tobacco Use statistics,(25) we chose states with a relatively high (e.g., 15.9% or greater) adult smoking rate representing all U.S. census regions. We will employ FBTA in five states including Indiana, Kentucky, Pennsylvania, Oklahoma, and Oregon. We chose these five states because they are geographically diverse across the U.S., their populations are racially and ethnically diverse, and they have moderate-to-high adult smoking rates increasing the likelihood of reaching screening-eligible individuals. We are partnering with the GO2 for Lung Cancer (GO2) to identify Centers of Excellence in Lung Cancer Screening (as designated by GO2) (26) to connect individuals who seek a screening referral request but do not have a primary care clinician. There are more than 800 Centers of Excellence in Lung Cancer Screening nationwide, and there are currently 126 in the five states in which we will conduct the study. These centers are well-established, well-connected, dedicated to high quality screening and care, and have strong collaborators with their local primary care networks, making them the ideal national partner to connect participants without a primary care clinician to one for facilitating the patient-clinician discussion about screening. They can also link individuals to primary care clinicians regardless of insurance status through federally qualified health centers and community health centers fostering access to high quality screening for low income and other vulnerable subpopulations.
2.3 Sample Eligibility Criteria
Eligibility criteria mirror the current USPSTF lung screening guidelines: 1) aged 50 to 80 years; 2) ≥ 20-pack-year smoking history; 3) individuals who currently smoke or quit smoking within the past 15 years.(6) Participants will be excluded if they are non-English speaking, have previously undergone LDCT for early detection of lung cancer, have a lung nodule or nodules that are currently being followed, or if they have been diagnosed with lung cancer.
2.4 Recruitment Rationale and Procedures
We will use a highly successful recruitment strategy via FBTA (27) to recruit screening-eligible individuals from Indiana, Kentucky, Pennsylvania, Oklahoma, and Oregon. The Facebook user’s interest list includes a wide range of details a user can select when setting up and/or maintaining their profile that they have an interest in such as groups, hobbies, lifestyle choices, behaviors, points of view, specific organizations and more. This allows us to purposively sample people who are age 50 years and older, indicate smoking or smoking cessation as an interest and reside in a particular state, city, or zip code. Using this approach, as we have in prior studies, (8, 10, 13, 28) we will target our advertisement on Facebook using the following keywords: cigarette, tobacco, nicotine replacement therapy, nicotine gum, electronic cigarette, smoking, vaping. Guided by the safety and monitoring guidelines for researchers using social media,(29–30) our approach includes design and close monitoring of the FBTA to ensure all methodologic and ethical standards are upheld. Currently, we have a potential reach of 550,000 potentially screening-eligible individuals in the 5 states above. As an example, with a $5,000 recruitment budget and an ad campaign that runs for 14 days, Facebook analytics estimate that 28,000 to 82,000 people per day will see the ad in their daily news feed in the five states and 262 to 758 unique Facebook users will click on the embedded eligibility survey link within the advertisement on a daily basis.
2.5 Description of the Intervention
Individuals who enroll into the study will be randomized to one of two arms: LungTalk or the non-tailored lung screening educational video. LungTalk is a computer-tailored health communication and decision-making tool that is theoretically grounded in the Conceptual Model on Lung Cancer Screening Participation. (24) See Fig. 2. This model links the Health Belief Model to the Precaution Adoption Process Model and includes key psychological variables (e.g., stigma, mistrust, fatalism, fear and worry) as factors that may influence an individuals’ decision to screen, or not, for lung cancer. (24) The tool as a whole serves as a cue to action for a screening-eligible individual to engage in a discussion with their clinician about the option to screen, or not, for lung cancer.
LungTalk is an interactive computer program that takes approximately 8 to 12 minutes to complete depending upon specific tailoring variables selected by the user and includes embedded audio, video and animation segments with tailoring algorithms for scripts presented from a master content library. (8, 13)LungTalk tailors initially on smoking status. Early in the program, the user is asked to indicate if they currently smoke cigarettes or if they quit smoking and subsequent content is aligned with an individual’s current smoking status. In addition, LungTalk tailors on the top three barriers to lung screening identified in our prior research testing the Conceptual Model for Lung Cancer Screening Participation (R15 CA208543).(24) Those barriers are: (1) cost/insurance issues; (2) worry; and (3) not receiving a recommendation from a healthcare clinician. After viewing an embedded video of how a lung scan is performed, the user will be asked if they perceive cost/insurance, worry, or lack of receiving a clinician recommendation as a barrier to getting screened. At this point, the user is able to choose none, one, two, or all three barriers. Based upon the user’s responses, LungTalk will play a brief video that includes messaging to address the perceived barrier. LungTalk concludes by offering the option of saving or printing a tailored summary at the end for individuals to use as a discussion prompt with their clinician.(8, 13) This print-out highlights key points related to lung health and screening, tailored by smoking status, offers question prompts to initiate a discussion with their clinician, and tailors the question prompts based upon questions that remain important to the user that they wish to discuss further with their clinician. (8, 13) Messages in LungTalk are presented at an 8th grade reading level, and in consideration of different ways people like to learn, the content is narrated as well as presented as key text on the screen.
The comparator/attention control condition is a non-tailored 5-minute video from the GO2 for Lung Cancer about lung screening designed for the lay individual. (31) This video was chosen as the comparator to serve as an attention control condition for the nonspecific effects of the intervention by balancing attention, treatment contact, and delivery channel so that a precise test of the hypothesized active component (the tailoring in LungTalk) of the intervention comparator can be made. Attention control conditions require two components – clinical attention and a therapeutic benefit; choosing the GO2 for Lung Cancer’s lung screening video meets both requirements.
2.6 Intervention Delivery
Intervention delivery will occur online. Eligible participants will be randomly assigned to either the intervention (LungTalk) or attention control (non-tailored educational video) after completion of the baseline survey with a 1:1 allocation as per a computer-generated randomization schedule stratified by smoking status (current or former). After recruitment, participants will be redirected to the REDCap platform to complete the informed consent and baseline survey. Participants then will be randomized and directed to their assigned intervention condition (i.e., LungTalk or attention control).
2.7 Data Collection
This study is focused on both the implementation of a social media-based communication platform to increase awareness about lung screening as well as the effectiveness of a tailored health communication and decision support tool (LungTalk). For the implementation of a social media-based communication platform, the primary outcome is reach. For the effectiveness of LungTalk, the primary outcome is screening uptake.
Assessment of Reach. Inherent in FBTA are a number of standard analytics (32) that will facilitate our ability to assess the number, proportion, and representativeness of individuals who are exposed to both health communications about lung screening. Using the FB analytics component (32) of our FBTA, our assessment plan will measure the total reach of the FBTA to increase awareness of the option to screen for lung cancer among screening-eligible individuals. Quantitative data collected from FB analytics during the recruitment campaign will assess reach by detailing: 1) reach; 2) link clicks; and 3) impressions. Reach in FB analytics is defined as the number of people who saw the FBTA at least once. (32) This can be further analyzed by hour, day, specific number of days, week, and campaign length. In addition, reach can be further stratified by specific location (i.e., state, city, town, county, zip code). (32)
Assessment of Effectiveness. Our assessment plan will compare the effectiveness of a tailored (LungTalk) health communication and decision support tool versus non-tailored health communication tool delivered online to improve: 1) total knowledge about lung screening; 2) lung cancer screening health beliefs; 3) occurrence of a patient-clinician discussion about lung screening; and 4) screening uptake. After enrollment, we will conduct a baseline survey using REDCap Survey with validated measures (35) used in our prior work assessing knowledge, lung cancer screening health beliefs, occurrence of a patient-clinician discussion about lung screening and stage of adoption for lung screening among 500 screening-eligible individuals in 5 states representing socioeconomically, ethnically, and geographically diverse locations. Grounded in the Precaution Adoption Process Model, stage of adoption is defined as seven stages an individual may be classified when presented with a health decision: (1) unaware, (2) aware but unengaged, (3) undecided, (4) decided not to act, (5) decided to act, (6) action, and (7) maintenance. Stratified by smoking status, participants will then be randomized to LungTalk or attention control. One week after delivery of the intervention, participants will complete an online follow-up survey to assess changes in knowledge, lung cancer screening health beliefs, occurrence of a patient-clinician discussion about lung screening, and screening uptake. At six months, participants will complete another online survey to assess occurrence of a patient-clinician discussion about lung screening and screening uptake. See Table 1 for Measures of Assessment.
Table 1
To assess the ability of FBTA to reach high-risk individuals eligible for lung screening |
Constructs | Assessment or Measure |
Reach | Total # of people who saw the FBTA at least once. |
Link Clicks | Total # of clicks on the link within the FBTA that led to the REDCap survey platform of the study. |
Impressions | Total # of times the FBTA was on screen (may include multiple views of the ad by the same person/people). |
To examine the comparative-effectiveness of LungTalk and a non-tailored lung screening information video in a national sample of screening-eligible, community-based individuals using an RCT design |
| Timeline of Assessment |
Constructs | Assessment or Measure | # of Items | Baseline | 1 wk | 6 mo |
Knowledge | Knowledge: Lung Cancer Screening | 9 | X | X | |
Perceived Risk | Perceived Risk of Lung Cancer Scale11 | 3 | X | X | |
Perceived Benefits | Perceived Benefits of Lung Cancer Screening Scale11 | 6 | X | X | |
Perceived Barriers | Perceived Barriers to Lung Cancer Screening Scale11 | 17 | X | X | |
Self-Efficacy | Self-Efficacy for Lung Cancer Screening Scale11 | 9 | X | X | |
Occurrence of Patient-Clinician Discussion | Self-report of Occurrence of a Patient-Clinician Discussion about Lung Cancer Screening | 1 | X | X | X |
Screening Uptake | Self-report via the stages of adoption algorithm for screening with verification process | 1 | X | X | X |
In order to mimic real-world implementation of a social media campaign to increase screening uptake, participant incentives will not be offered for the baseline survey. However, after enrollment in the study and viewing the intervention to which the participant has been randomized, participants will learn that they will receive a monetary gift card upon completion of the follow-up surveys at two time points following intervention ($50 after 1-week survey post-intervention; $25 after 6-month survey post-intervention).
Retention. Several techniques to increase engagement in the study over the 6-month follow-up period will be employed such as: 1) communicating clearly the requirements of the study during the recruitment phase; 2) obtaining alternative contact information such as phone numbers (i.e., home, work, cell) and participant email address for follow-up; and 3) sending out an electronic newsletter to report the progress of the study.
2.8 Data Analysis and Interpretation
Analysis of Reach. Using data from the Facebook analytics collected during the targeted advertisement period, we will analyze reach in the following ways: (1) number and percentage of individuals age 50 and older who currently or formerly smoke in the population in which the advertisement is marketed; (2) percentage of eligible participants who agree to participate in the study; (3) compare differences between those participating and those not participating on smoking status (i.e., current vs. former), age, gender, geography and other key variables collected on the screening survey; (4) record reasons that participants refuse to participate in the study; (5) estimate attrition at 1 week and 6-month follow-up time periods; and (6) compare differences between those completing and those not completing the study on sociodemographic and health status variables, geography, baseline scores on knowledge, lung cancer screening health beliefs, and stage of adoption for lung cancer screening. Facebook generates analytics related to the advertisement and includes descriptive statistics such as proportions and means to assess reach, link clicks, and impressions, as described above, for the FBTA. (32)
Analysis of Effectiveness. Our analyses were defined a priori to address the study aims. We will use descriptive statistics such as means, standard deviations, and frequency distributions/distributional assumptions to examine data quality, identify patterns of missing and out-of-range values, and evaluate the assumptions of statistical tests. Specifically, we will examine all aspects of data quality to ensure statistical integrity and accuracy including: 1) data skewness, kurtosis, and parametric assumptions; 2) intention-to-treat (ITT) principles; 3) missing data considerations; and 4) control of overall alpha to avoid inflated experiment-wise Type-I error due to multiple statistical tests. Remediation of normal distribution assumption violations will be accomplished using methods such as data transformations (e.g., log or square root for positively skewed variables), Box-Cox family transformations, or kernel estimation techniques to determine the best-fitting parametric density, (36) or other methods as appropriate. Assessment of internal consistency reliability of all scales will be carried out using the Cronbach’s alpha coefficient. We will apply the intention-to-treat (ITT) principle in handling missing data on screening uptake. A study participant will be coded as ‘no screening uptake’ unless otherwise verified by our participating sites. We will examine violations of the missing-at-random assumption. If missing is not completely at random, then covariates associated with missingness will be incorporated into data analysis to minimized. A related approach is the use of mixed-effects models, which is capable of handling binary as well as continuous outcomes,(37) to use all available behavioral outcomes data since HLM does not carry out list-wise deletion by default, thus the statistical power loss due to missing data may be minimal. Additionally, missing outcome data (assuming up to 20% of the respondents will be unreachable at our three-month follow-up) as well as missing assessments may be amenable to imputation by several techniques that can handle both continuous and categorical missing data.(38–40) We may also use the Pattern-Mixture Model to examine whether or not missed follow-up assessments are associated with baseline characteristics with safeguards to minimize model overfit. Finally, to control for potentially inflated Type-I error rate due to multiple comparisons, analyses will incorporate a multiple comparisons method, such as a False Discovery Rate-controlling procedure, which is more powerful than simple Bonferroni corrections.(40)
Total knowledge scale scores, Lung Cancer Screening Health Beliefs (total scale scores for perceived risk, perceived benefits, perceived barriers, self-efficacy), and stage of adoption are continuous variables. Screening referral request and screening uptake are dichotomous variables. Prior to group comparisons, measures will be described by timepoint, both overall and by group. Within-group changes will be assessed using the standardized response mean (SRM) effect sizes (mean change divided by SD of change). The primary outcome is lung screening uptake but forward movement in stage of adoption (i.e., the change score) will also be evaluated as a secondary analysis. For between-group comparisons, continuous outcome variables will be compared using two-sided independent-sample t-tests (accompanied by the standardized mean difference effect size, i.e., difference between group means divided by baseline SD) and dichotomous outcome variables will be compared using the Chi-square test, or two-sided Fisher’s exact test, if 20% or more cells have expected counts less than 5 (accompanied by the odds ratio effect size). In the case of differential attrition, as noted above, regression models will also be used to assess outcomes by adjusting for baseline covariates that differ significantly between participants who do and do not complete follow-up assessments. Moderators of intervention effectiveness will be assessed by regressing outcome variables (e.g., knowledge score) on randomization arm, the potential moderator, and an interaction term, where a significant interaction effect is indicative of moderation. Significant moderator effects will further be explored and described with stratified analyses.
Sample Size Justification and Power Analysis. With a sample size of n = 250 per intervention group, and assuming an ITT analysis (with a default of no screening uptake unless otherwise verified), we will be able to detect a difference between the LungTalk and the attention control intervention condition groups with an 81% statistical power if the difference in lung screening uptake is 31% in the LungTalk group compared to 10% in the attention control intervention condition group, in a test of independent proportions and a two-sided type-I error rate of 5%. This 31% versus 10% difference is based on preliminary data testing LungTalk in a sample of community-based screening-eligible individuals in Indiana in 2018 using the same ITT procedure. With a sample size of n = 250 per group and up to 20% missing assessment data (n = 200 available for analysis) on total knowledge scale and total perceived risk scale scores, we will have an 80% statistical power if the difference is d = 0.28 (in standardized effect size units, or Cohen d), in an independent-sample t-test with a two-sided type-I error rate of 5%. A 0.25 effect is considered a ‘small’ effect size in psychology-based research, thus a conservative estimate of the statistical evidence that can be supported in our study design.
We expect to start enrollment in June 2023 and conclude the study in the Summer of 2027. Data analysis will be completed by December 2027.