Introduction
Fifteen to 20% of colorectal cancer (CRC) cases are related to familial or hereditary factors [
1‐
3]. These are mostly familial CRC (FCC), a family-history based diagnosis without a known genetic cause, and to a smaller extent the autosomal dominantly inherited Lynch syndrome, comprising 2 to 4% of CRC cases [
1,
4].
Identification of these hereditary and familial factors in CRC patients provides an opportunity to engage in effective screening and surveillance protocols for those patients as well as for their relatives at risk. A detailed assessment of the family history for CRC can help to identify these patients. To judge whether they should be referred for genetic counseling, persons are generally tested against referral criteria, such as the Amsterdam II and the Revised Bethesda criteria [
5‐
7]. However, physicians and nurses might have limited knowledge about these referral criteria and the family history might be incompletely explored [
8‐
15]. Consequently, in daily practice only 10–30% of patients at risk are referred for genetic counseling and for screening and surveillance recommendations [
3,
8,
12‐
17].
To identify persons at an increased risk of FCC or hereditary CRC, we previously developed and validated an online patient-administered questionnaire that provides a full and detailed family history [
18]. This questionnaire was incorporated in a family history tool that creates an automated referral recommendation for genetic counseling, in case of suspected Lynch syndrome and for surveillance colonoscopies, in case of FCC [
5].
To evaluate its effectiveness, we designed a multicenter trial with a stepped-wedge design, in outpatient clinics for patients with CRC. Each hospital continued standard procedures for identifying patients at risk (control strategy) and then switched to offering the family history tool to included patients (intervention strategy). We anticipated that the implementation of this tool would increase the proportion of newly diagnosed patients with CRC receiving a CRC screening or surveillance recommendation for hereditary CRC or FCC for themselves and/or their relatives, provided by genetic counseling.
Discussion
As FCC and hereditary CRC syndromes are not always recognized in CRC patients, many persons at risk do not receive appropriate CRC screening and surveillance recommendations. We developed and validated a family history tool that could assist in recognizing CRC patients and their relatives with FCC or a hereditary CRC syndrome. In this multicenter trial with a stepped-wedge design we observed that implementation of this tool in multidisciplinary outpatient clinics for patients with CRC did not increase the proportion of CRC patients with hereditary CRC or FCC receiving surveillance recommendations for themselves or their relatives, compared to standard practice.
What are the potential reasons for the intervention strategy not to be effective in detecting those patients compared to the control strategy? To clarify this, we should first try to explain why the proportion of patients who underwent genetic counseling was similar in both strategies (14%). First, an unexpectedly large proportion of patients received surveillance recommendations in the control strategy (12%, whereas we expected 4%), leaving limited room for improvement. One factor could clarify why the control strategy performed much better than what was previously reported in literature [
3,
8,
12‐
17]. We could have increased awareness by offering a list with the referral criteria to all hospitals at the start of the control strategy.
Secondly, only a disappointing proportion of 58% of eligible patients were offered the family history tool in the intervention strategy. The reasons were not explored, but this could be due to the fact that nurses found the tool time-consuming, study enrolment was not a priority in the first work-up of a patient with CRC, nurses did not always counsel the patient (i.e. after an endoscopically resected T1 carcinoma) and thus not always invited them for participation and selection bias might have occurred (i.e. inviting mainly those with a striking family history of CRC who would be easily recognized without the tool). Moreover, a proportion of patients did not wish to use the tool after it was offered.
A third explanation for the comparable genetic counseling rates may be the false reassurance, resulting from introduction of the family history tool. Physicians may have thought that the tool, that was offered by nurses, would identify everyone with a referral indication. Yet there may also be patients in a ‘grey referral area’, who would usually be referred despite not strictly fulfilling referral criteria, such as a patient with multiple relatives with CRC above the age of 70. Such patients are commonly diagnosed as having a familial risk of CRC. However, in several hospitals nurses were not used to referring patients, as the physician usually did this, and might not have recognized these ‘grey’ referral indications that were not detected by the tool. This is confirmed by the finding that there was a relatively small proportion of patients diagnosed with FCC in the intervention strategy, although this could also have another reason. A year after study commencement (when all hospitals had started the intervention strategy), a new nationwide guideline on the detection of hereditary CRC was introduced, recommending immunohistochemical analysis of mismatch-repair proteins on all CRC cases under the age of 70 years [
5]. This could have changed results in the intervention strategy, as patients with a suspect family history but without loss of mismatch-repair proteins might not have been referred, despite still fulfilling criteria for FCC.
Despite the genetic counseling rates being similar, the proportion of referred patients who received screening and surveillance recommendations was (non-significantly) smaller in the intervention strategy. We might clarify this. The family history tool mostly focusses on identifying patients with Lynch syndrome, by taking into account all Lynch syndrome-associated tumors. If genetic testing does not detect Lynch syndrome, especially in patients who fulfill referral criteria based on familial occurrence of extra-colonic Lynch associated tumors, often no surveillance recommendations will be given. This is most likely the reason why in 16/67 (23.9%) referred patients in the intervention strategy a syndrome was excluded, compared to only 6/41 (14.6%) patients in the control strategy. Combining that with the above-mentioned finding that the intervention strategy identified fewer patients with FCC, fewer patients received a surveillance recommendation in the intervention strategy.
Several limitations should be acknowledged. First of all, our aim was to identify more patients and relatives with hereditary CRC syndromes and FCC, although our tool only identified patients with Lynch syndrome and FCC, not those with polyposis syndromes. However, as FCC and Lynch syndrome contribute to the majority of all CRC syndromes and are easily missed due to the lack of specific endoscopic features, we decided that identification strategies should mainly focus on these two groups. For that purpose the tool is accurate, as was previously reported [
18]. Another factor that might influence results is that patients could have returned to the referring hospital that was not in our region, resulting in unknown data of genetic counseling. By crosschecking referrals with all genetic centers in the surrounding of the participating hospitals, we believe that having missed referral data within the participating region is unlikely. Besides, this most likely then would have happened equally in both strategies.
Several family history tools that include a referral decision aid have been developed before, but the tool we developed is the first to verify both FCC and all Lynch syndrome associated tumors [
22‐
26]. Only three previous studies have evaluated the effectiveness of a CRC family history tool and compared it to a control setting [
22,
25,
26]. An unexpected effect, similar to ours, was seen in a recent Dutch clustered randomized controlled trial in CRC patients [
26]. The authors investigated the efficacy of a comparable tool, complemented by a website with information on familial risk of CRC, risk calculators for patients, and a genetic counseling decision support intervention for high-risk patients. This intervention did not improve referral for cancer prevention measures, compared to a control setting. The authors suggested that giving a recommendation for genetic testing through an online application instead of through a physician or nurse could decrease adherence to the recommendation. However, in our study health care providers advised patients on genetic referral, after being informed by the tool. This neither resulted in an effective outcome. In the American Family Healthware Impact Trial, Rubinstein et al. studied the effect of an online tool that stratifies risks for six common diseases, including CRC [
25]. In their randomized controlled trial in healthy persons in a primary care setting, the intervention did not result in an increased rate for CRC screening compared to a control setting. Screening rates were already high at baseline, as in our study, resulting in a low power to detect an intervention effect. The third study, the primary care based cluster randomized GRAIDS trial, showed that a family history tool resulted in more patients referred for genetic counseling compared to the current best practice [
22]. However, this study involved very small numbers. It thus seems that all those tools, despite a successful validation, do not always work when implemented in real-life health care situations. Reasons for limited success rates should be further evaluated.
Based on our explanations, we can provide suggestions for adjustments in the implementation of the family history tool to improve efficacy. One could consider assessing the proportion of detected hereditary CRC and FCC cases before implementing this tool. If this proportion is less than the generally expected 15–20% of CRC patients with FCC or hereditary CRC, it could be worthwhile to use the tool. To reduce the time and workload needed for implementation, implementation within a pre-existing electronic patient file is desired. To enhance continuity, one or two trained persons can be assigned to apply the tool and decide on genetic referrals, including the above-mentioned ‘grey’ referrals [
27,
28]. As we felt that patients as well as health care providers did not prioritize genetic referrals in the early stages of the CRC work-up, these trained persons could implement the tool after surgery instead of at the time of diagnosis and the first visit to the clinic. These adjustments should first be evaluated in comparative effectiveness studies. Moreover, one could consider evaluating its efficacy in other healthcare situations, such as in CRC screening programs or in primary care. In these situations the tool might be more effective, as these persons have not yet been seen by a CRC specialist, resulting in a higher likelihood of detecting unrecognized familial and hereditary factors [
29].
We conclude that the family history tool in this multicenter prospective comparative cohort study does not necessarily increase the number of CRC patients and their relatives diagnosed with FCC or hereditary CRC syndromes, and therefore will not automatically increase the number of persons who receive screening or surveillance recommendations to prevent new CRC cases. Other interventions should be considered to facilitate the identification and enrollment of these patients in surveillance programs.