A prospective, population-based intervention study design was chosen to evaluate the effects of FARKOR. All participants will give written informed consent prior to study entry. Recruitment starts in October 2018 and runs until the end of March 2020. First results are expected to be available in October 2020.
Participants
The recruitment of participants is multimodal. Participating statutory health insurance (SHI) companiesFootnote 1 will inform all patients who have been diagnosed with CRC in the last 18 months that their relatives may have a higher risk of developing CRC compared to the general population. The recently diagnosed CRC-patient can inform his/her relatives about the participation in FARKOR. Moreover, physicians will be able to certify for the FARKOR program by undergoing a specific internet based training program on the identification of individuals with a family history of CRC. As FARKOR physician, they recruit subjects between 25 and 50 years of age to participate in the screening program. Finally, there will be a public campaign for the program as well as a website (https://www.darmkrebs-in-der-familie.de/), which allows interested individuals to choose a close-by FARKOR physician.
Inclusion criteria are: Age between 25 and 50 years, membership of a participating SHI, residence in Bavaria, and written informed consent. Individuals for whom adequate CRC related screening measures already exist are excluded from the study. Exclusion criteria are: Previous diagnosis of CRC, familial adenomatous polyposis or chronic inflammatory bowel diseases (ulcerative colitis, Crohn’s disease) or a known family history of hereditary non-polyposis colorectal carcinoma (HNPCC).
The FARKOR screening process
Participants will be invited by participating physicians to participate in a short standardized interview regarding their family history, which is based on a simplified version of the Amsterdam and Bethesda criteria [20], and to sign a written informed consent form. Based on this short interview consisting of five simple questions concerning the diagnosis of CRC in first- and second-degree relatives, participants are either classified as potential risk carrier if they indicate having at least one first- or second-degree relative with CRC or as risk free. In a shared decision-making, potential risk carriers and their physicians discuss the risks and benefits of possible CRC screening options, which consist of an iFOBT, a screening colonoscopy, or deferring further CRC screening measures to a later point in life. Patients with a positive iFOBT result will undergo diagnostic workup by colonoscopy and both diagnostic and screening colonoscopies are accompanied by a comprehensive documentation of both macroscopic and histologic findings and of potential complications. In case of pathologic findings, patients will receive treatment according to the established guidelines [21]. Additionally, potential risk carriers will be encouraged to undergo an in-depth interview regarding their family history of CRC in which detailed information concerning age, incidence and mortality of CRC in all first- and second-degree relatives is ascertained.
The full FARKOR process is shown in Fig. 1 and consists of the following steps: (1) informed consent, (2) short questionnaire on family history of CRC, (3) shared decision making, (4) offer to complete an in-depth questionnaire on family history of CRC, and (5) further screening measures (iFOBT, colonoscopy, deferral and timing of future screening procedures). All steps in this screening process are accompanied by an internet-based documentation which will be completed by the physicians participating in the program.
Additionally to this FARKOR screening process, all participants (with and without family history of CRC) will be invited to complete an online lifestyle questionnaire based on an instrument previously developed for the RAPS study (Risiko-adaptierte Präventions-Strategien für Darmkrebs) [16]. It consists of questions on pre-existing conditions, participation in prevention programs, lifestyle habits (e.g. smoking, alcohol consumption, nutrition, exercise, drug use…) as well as anthropometric and sociodemographic information (including age, sex, height, weight, and educational level).
Data collection and data protection
The data collection consists of (1) an internet-based documentation of interviews, results of the shared decision-making, and findings of the proposed screening measures, (2) an internet-based lifestyle questionnaire, which is self-administered by the participants, and (3) health insurance data on the incidence of colorectal adenoma and CRC in Bavaria between the years 2015 and 2020. A central trust center will be performing data linkage and pseudonymization. It will also coordinate all requests from participants regarding transparent information, communication and modalities for the exercise of the rights of the data subject according to the EU General Data Protection Regulation (GDPR) including the right to erasure. The analysis will use anonymized data. The analysis dataset will be stored for 10 years following the guideline of good epidemiological practice (GEP) [22].
Statistical analysis
Primary endpoint
The prevalence of CRC (excluding carcinoma in situ) and of precancerous lesions will be evaluated in participants who underwent a screening colonoscopy. Based on health insurance records, it will also be possible to estimate the incidence of CRC and of precancerous lesions (1) in FARKOR participants with familial CRC risk who decide to defer CRC screening measures to a later point in life (2) in FARKOR participants who are classified as risk free, and (3) in individuals in the target population who do not participate in the FARKOR study.
Secondary endpoint
The number of complications arising through screening colonoscopies performed in FARKOR will be considered as secondary endpoint.
Comparing the prevalence of CRC and of precancerous lesions in study participants to a general screening population
In order to answer the primary research question, the prevalence of CRC and of precancerous lesions in study participants who undergo colonoscopy will be compared to the findings in a screening population of individuals for whom CRC screening is already implemented in Germany. In particular, the findings concerning the prevalence of CRC and of clinically relevant precancerous lesions (advanced adenoma, defined as adenoma ≥ 10 mm, containing high-grade dysplasia, or villous histology) for colonoscopies performed in FARKOR will be compared to the prevalence observed in screening colonoscopies which are proposed to individuals who are aged 55 years and older in Germany.
Modeling long-term outcomes and estimating long-term effects
Due to its short duration of only 18 months (October 2019 until March 2020), the study cannot provide direct evidence concerning the effect of a risk-adapted screening program on CRC mortality in the target population of individuals between 25 and 50 years of age. Therefore, CRC mortality will be estimated indirectly based on information on the stages of detected CRC cases and on the characteristics of precancerous lesions in both FARKOR participants and in individuals in the target population who do not participate in the screening program. A baseline estimate of CRC mortality in the target population for a time period in which no risk-adapted screening was proposed will be estimated based on health insurance data collected in the three years before the start of the FARKOR study.
A hierarchical model describing the long-term outcomes of the FARKOR program is presented in Fig. 2. The unknown parameters in this hierarchical model can be estimated through Bayesian inference. In order to provide more precise estimates and to reduce some of the uncertainties, it is important to consider not only the direct sources of information (the study database and the health insurance data) but also data from indirect sources of information to make use of all relevant information. Thus, data from the RaPS study [16], the Munich Cancer Registry, the Bavarian Cancer Registry, and the scientific literature will be considered as indirect sources of information.
A detailed description of all statistical methods is provided in a statistical analysis plan, which will be published before the end of data collection.
Finally, additional parameters will be evaluated, including the number of complications due to performed colonoscopies, participation rates as a function of region, patient characteristics and the medical specialties of the recruiting physicians, and the positive predictive value of the iFOBT in individuals between 25 and 50 years of age with a family history of CRC.
Economic evaluation
The systematic evaluation of long-term health-economic consequences of CRC screening in the population of interest uses a decision-analytic state-transition model [23, 24]. The model reflects the German health care context. It simulates the natural history of colorectal adenoma and cancer development in the target population and assesses the impact of different CRC screening strategies including iFOBT and colonoscopy with diagnostic follow-up and treatment.
The decision model will be applied and calibrated to the German epidemiological setting using the data of the Bavarian Cancer Registry and other German cancer registry data sources. Direct medical costs for screening procedures, diagnostic tests, treatment and follow-up as well as resource utilizations are documented within FARKOR. Outpatient costs will be based on the ’Standardized Rating Scale of the Federal Association of Physicians under the Statutory Health Insurance’ and the ’Statutes of Medical Fees’. Inpatient costs will be derived based on the German Diagnosis Related Groups System.
The economic analyses use a lifelong time horizon. The perspective of the German statutory health insurance (SHI) will be adopted and a 3% annual discount rate will be applied to costs and health outcomes. Outcomes for benefits and harms will include life-years gained (LYG), CRC cases and CRC-related deaths averted, additional complications due to colonoscopy (physical harm) and positive test results (psychological harm). Economic outcomes will include lifetime costs and discounted incremental cost-effectiveness ratios (ICER). Comprehensive sensitivity analyses assess the uncertainty of the results.
The modeling process follows the guidelines of health technology assessment (HTA) and comparative effectiveness research. The analysis will comply with international guidelines and recommendations for decision-analytic modeling, such as the Joint Task Force guidelines of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) and Society for Medical Decision Making (SMDM) [24, 25], international key principles for HTA, reporting guidelines (CHEERS) and the EUnetHTA guideline for health economic evaluations [26–29].
The decision-analytic state-transition cohort model will be programmed and validated using the decision-analytic software package TreeAge Pro 2017 (TreeAge Software Inc., Williamstown, MA, USA).
Sample size considerations
The statistical analysis will use a population-based data set. In the sample size consideration, it is only possible to use educated guesses for specific parameters (see Fig. 3). The intervention is implemented in a population of about 3.5 million people (Bavarian population between 25 and 50 years of age) of which about 90% (3.15 million people) are SHI members, who represent the target population. Health care statistics show that about 1.5 million individuals in the target population get in contact with a practitioner or a CRC screening related physician every year. We expect that 90,000 (6% of the 1.5 million) will accept to participate in the program and that 6% of these 90,000 will report a positive family history of CRC. This leads to 5,400 potential risk carriers recruited through contact with physicians. Based on the number of CRC cases in Bavaria, it can be estimated that a further 3,200 potential risk carriers will be recruited through their first-degree relatives who are contacted by their SHI. This results in an expected total of 8,600 participants with a family history of CRC. About 40% will decide for a colonoscopy (3,440). The remaining 60% of these subjects will decide for an iFOBT (5,160) of whom 30% will be positive an undergo colonoscopy (1,548). It is expected that 33% of the colonoscopies will produce findings that result in polypectomy and histology (1,150). Given these numbers, we expect to be able to calculate confidence intervals of fair precision for prevalence of CRC or advanced adenoma (±0.01).