Building the algorithm

An extensive literature review of the available smoking prevention and cessation resources was conducted to gather the necessary information to develop the proposed web-based algorithm. These resources included: the Transtheoretical Model (Stages of Change) [27], the US Department of Health Clinical Guidelines for Treating Tobacco Use and Dependence [18], the 5 A’s model [28], published smoking cessation paper-based algorithms used in clinical settings [15], as well as other published scientific literature [14, 16, 17]. Tobacco specialists with clinical and research expertise also contributed to the design of the proposed algorithm. Brief communicatory statements were inserted into the algorithm to help the CCPs initiate and sustain an engaging provider-patient conversation when providing smoking prevention and cessation services to cancer patients.

The algorithm was designed with the use of a simple top-down flowchart-like tree structure with multiple possible branches built into the Research Electronic Data Capture (REDCap) platform where the output (personalized smoking prevention and cessation counseling approach and corresponding pharmacotherapy) is based on a set of binary decisions (Fig. 1). Through the utilization of the powerful research electronic data capture of REDCap and its comprehensive data validation features, we were able to ensure the accuracy and integrity of the collected data, enhancing the reliability of the web-based algorithms built upon it. This robust validation process helped reduce errors and biases, ultimately leading to a more precise, robust, and trustworthy clinical decision-making process for smoking prevention and cessation.

Fig. 1

Some selected screenshots from the web-based algorithm

The web-based algorithm was initially developed in English and subsequently adapted culturally and linguistically to Spanish. We used neutral Spanish to minimize regional variations and dialectical differences among CCPs in Colombia and Peru. The ultimate goal of this adaptation was to foster engagement and empowerment among CPPs making informed choices aligned with their cultural values. The algorithm was pre-tested among 10 CCPs (5 in each country). Feedbacks from these ten individuals were key for refining the design of the web-based algorithm before testing its usability and acceptability in a larger study sample.

Participants, setting, and procedure

Twenty-nine CCPs from the National Cancer Institute in Colombia (INC-Colombia) and 24 from the National Institute of Neoplastic Diseases in Peru (INEN) were invited to pilot-test the web-based algorithm three months after completing a smoking prevention and cessation training program that covered the epidemiology of tobacco use in LATAM, cancer and other diseases associated with cigarette smoking, behavioral-based smoking prevention and cessation interventions, and cessation medications and therapies. The sample size for this pilot study was dependent on the sixty CCPs (30 at INC-Colombia and 30 at INEN) who were originally invited to participate in the STOP training program for smoking prevention and cessation. This convenience sample was considered appropriate by the leadership at INC-Colombia and INEN to avoid any disruption in their daily workflow and clinical activities.

The inclusion criteria for this study were (1) being appointed at one of the above centers; (2) ≥ 18 years old; (3) had direct contact with cancer patients; and (4) proficiency/access to internet with a computer, tablet, or smartphone.

Before implementing the study, an introductory 1-hour virtual meeting was conducted to educate CCPs about the functionality and use of the web-based algorithm. Afterward, participants completed an online pre-test data collection form. The link and quick response (QR) code for accessing the algorithm were distributed via email. Additionally, printed cards with the QR code were provided to CCPs to facilitate their access to the algorithm.

All study participants were asked to use the web-based algorithm with at least ten patients every month for a total period of 3 months, after which the study participants completed an online post-test form. The pre-test and post-test were identical, collecting information about demographics (e.g., age, sex assigned at birth, level of education, profession, and time devoted to patient care), as well as CCPs’ self-efficacy, and practice for smoking prevention and cessation:

Self-efficacy domain: The questions (SI appendix, Table S1) included in this domain were adapted from the International Association for the Study of Lung Cancer (IASLC) survey [29] and the National Health Interview Survey (NHIS) [30]. This domain had seven questions and assessed the CCPs’ self-efficacy to provide smoking prevention and cessation to their cancer patients.

Practice domain: The practice domain was also adapted from the IASLC and NHIS surveys and had six questions [29, 30]. This domain assessed the smoking prevention and cessation practices of the CCPs (SI appendix, Table S2).

Usability: We used the System Usability Scale (SUS) to measure CCPs’ usability perception of the web-based algorithm [31]. The SUS is a widely used metric comprising 10 standard questions on a 5-point Likert scale from Strongly Agree to Strongly Disagree (Table 1). The web-based tool or system with a benchmark cut-off score equal to or greater than 68 is perceived to have a high usability.

Table 1 The system usability scale

After completion of the post-test at three months, CCPs were invited to participate in a focus group to share their perceptions about the usability and acceptability of the web-based algorithm. A total of 4 focus groups were conducted (2 in Colombia and 2 in Peru) at the INC-Colombia and INEN facilities by a trained moderator who asked open-ended questions about CCPs’ views of the algorithm and requested feedback on specific design elements of the web-based tool. Peruvian data were collected in April 2019, while Colombian data was collected in September 2019. Each focus group lasted less than 60 min. Participants were informed of the study procedures, risks, and benefits and provided written consent to participate in the focus group discussions and for the audio recording. All focus groups were digitally recorded, transcribed verbatim, de-identified, translated to English, and saved as Microsoft Word files.

Analysis planDemographic information

Descriptive statistics were used to summarize the CCPs’ demographic characteristics. Bivariate analyses between the Colombian and Peruvian CCPs’ demographic characteristics were computed using the Chi-squared test (or Fisher’s exact test, when appropriate) for categorical variables and a two-sample t-test for the continuous variable.

Quantitative data

Self-efficacy: Self-efficacy questions also had responses on a 5-point Likert scale and a score range of 7–35. Change in the CCP’s smoking prevention and cessation efforts self-efficacy was evaluated using the Wilcoxon signed Ranked test. The effect size of the change in self-efficacy was computed as the absolute value of the difference between the post-test and pre-test self-efficacy scores divided by the mean of the standard deviations for each.

Practice: The practice domain had questions with responses on a 5-point Likert scale and a possible score range of 6–30. We used the Wilcoxon signed Rank test to measure a change in the CCP’s smoking prevention and cessation practice before and after using the web-based algorithm. The effect size of the change in practice was calculated as the absolute value of the difference between the post-test and pre-test practice scores divided by the mean of the standard deviations for each.

Usability: Each CCP’s SUS score was calculated and then converted to percentile ranks. The SUS scores of the CCPs were combined, and the mean and standard deviation were calculated for each and both countries.

The significance threshold was set at 0.05. All quantitative analyses were conducted using Stata v. 17.0/SE (StataCorp, College Station, Texas).

Qualitative data

Two coders performed a round of coding on the four transcripts. Initially, each coder reviewed and coded the transcripts individually. Through this, a preliminary list of codes was established. Finally, both coders collaborated on subsequent pass coding processes where all transcripts were reviewed thoroughly, and a list of codes was refined based on the inter-coder agreement. Both coders established agreement through subjective discussion throughout the final coding process. Instances of disagreement were resolved through coder discussion, a revisiting of the transcripts and notes, and recoding to encompass both coders’ perspectives if appropriate. A thematic analysis was performed manually to identify the major themes that reflected the acceptability and usability (i.e., reported merits, flaws, and suggested modifications) of the web-based algorithm tested by the CCPs.

Ethical statement

The study was approved by the INC in Colombia (Cod INC-STOP Protocol/EX- 00812), the University of Texas MD Anderson Cancer Center Houston (Protocol No. PA17-0878), and the University of Texas Health Science Center Houston (Protocol No. HSC-SPH-20-1339). Protocol approval was not required from the INEN in Peru as the STOP program was an educational effort. Signed informed consent was obtained from all participating CCPs.