A Randomized Controlled Trial Testing Provision of Fecal and Blood Test Options on Participation for Colorectal Cancer Screening

Early detection is critical for reducing mortality from colorectal cancer and can be achieved through effective screening programs. Many countries have implemented organized, population-based, screening programs with fecal occult blood tests (FOBT), but participation rates can be poor (1–3) and need to be improved for screening to achieve its full potential as a public health strategy.

Previous work that focused on identifying and removing barriers to participation in fecal test–based screening suggests that a blood test could improve screening uptake, particularly for those who dislike fecal sampling (4, 5) or who have a benign, bleeding condition making completion of FOBT inappropriate (6). Research has confirmed that dislike of fecal sampling is a common reason for nonparticipation in screening (7), with one study finding that fecal tests were the least preferred of various testing options including colonoscopy, barium enema, sigmoidoscopy, and virtual colonoscopy (8). Beliefs about colorectal cancer prevention, awareness of the tests, and the benefit of screening (9), as well as test technology variables (9–11) are also likely to be relevant when designing screening programs.

Emerging molecular diagnostic assays offer the potential to detect circulating tumor DNA (ctDNA) in blood by detection of aberrantly methylated genes in colorectal cancer such as SEPT9 (12). We have shown that the genes BCAT1 and IKZF1 are methylated with high frequency in colorectal neoplastic tissues (13, 14) and that methylation in these genes can be detected in the plasma of patients with colorectal cancer (15, 16). The methylated BCAT1/IKZF1 blood test has comparable sensitivity to fecal immunochemical tests for hemoglobin (FIT) for detection of colorectal cancer, and better specificity (17). Use of these blood tests could circumvent fecal aversion or inappropriateness in the case of benign medical conditions causing bleeding, and therefore improve screening program participation.

To date, participation rates with blood or fecal tests have not been compared in a general population already familiar with fecal test–based, centralized screening program. It is important to understand whether a population that has been offered a fecal test would be prepared to screen with a blood test, particularly in those who have previously rejected the fecal test. A blood test could provide the possibility of engaging such nonparticipants. A conservative approach would have a screening blood test deployed within a screening program as a second-line “rescue” strategy to engage those who reject the existing fecal test screening method. Alternatively, it might be offered as an upfront choice. Understanding influences on participation, as well as test mode preferences, would provide useful information on how best to implement a blood test for colorectal cancer screening.

The primary aim of this study was to determine whether provision of a blood screening test, either as an upfront choice, or as a “rescue” offer to fecal test nonparticipants, within a mail-based FIT-based screening program would increase screening uptake compared with a standard screening program that only offers a fecal test. The secondary aim was to understand whether demographics, experience with screening, or differences in anxiety about test sampling influenced preferences and choice.

A random selection of people ages 50–74 years was provided by the Australian Electoral Commission for six electoral regions of South Australia. Each region was matched for gender and 5-year age bands. Study invitees were randomized into three groups (n = 600/group) using a random number generator. The study groups were labeled as control, rescue, and choice.

The study consisted of two consecutive 12-week phases (Fig. 1). In the first phase, all groups were invited to screen and were mailed a free FIT kit, with the choice group also provided with the opportunity of arranging a blood test in place of using the FIT. In the second 12-week phase, the control group received a further reminder to screen with FIT, and the rescue group were similarly reminded but also given information for arranging a blood test instead of the FIT, while the choice group received no reminder.

Prior to the screening invitation, all invitees in each group were mailed an advance notification letter (18) alerting them to the importance of screening for colorectal cancer and advising them that they would shortly be sent an invitation to participate in a screening study, but were not informed of the study design. Individuals were provided with a phone number which they could use to opt-out from the screening offers if they had completed a screening test within the previous 12 months, or had a colonoscopy within the previous 5 years.

Ethical approval for the study was obtained from the Southern Adelaide Clinical Human Research Ethics Committee, with the study conducted in accordance with the Declaration of Helsinki. The trial was prospectively registered with the Australian and New Zealand Clinical Trials Registry (trial number ACTRN 12615000972527). Written informed consent was obtained from all study participants.

Two weeks following the advance notification letter, all study invitees who had not opted-out were mailed a screening pack in a design similar to that used by the Australian National Bowel Cancer Screening Program (NBCSP). The pack included an invitation letter and information sheet describing the study nature, a colorectal cancer screening information booklet, a participant details form, FIT kit, and a survey for completion and return. The FIT kit included two sample collection tubes (OC Sensor, Eiken Chemical Company) and collection sheets, instructions on how to complete the test, and a reply-paid return envelope for the sample tubes and completed forms to be returned by post to the laboratory (Repatriation General Hospital, Daw Park, South Australia, Australia). Invitees were requested to sample from two different bowel motions at home and return the collected samples within 2 weeks. Samples collected incorrectly were not analyzed, and a replacement kit was sent for the opportunity to repeat the test. Analysis of FIT collection tubes for hemoglobin concentration followed the manufacturer's instructions as described previously (19).

In addition, the choice group was provided with written information in the invitation letter with details on how to call the helpline and request a referral for a blood test should they not wish to do the FIT. They were further informed that the blood test accuracy for colorectal cancer detection had a similar performance to the FIT in that it would detect at least 60% of colon and rectum cancers (17). Individuals opting for the blood test were sent an information sheet, a participant details form, a consent form, a list of blood collection center locations, and a referral for the methylated BCAT1/IKZF1 blood test (Colvera, Clinical Genomics Pty Ltd). Phlebotomy was arranged through blood collection centers (Healthscope Ltd), selected within the electoral regions. Blood was collected (18 mL, K3-EDTA) and the extracted plasma was couriered to Clinical Genomics for assay of methylated BCAT1 and IKF1 as described previously (15).

Participation with FIT was recorded as return of the sample collection tubes, even if sampling was performed incorrectly. For the blood test offer, a phone call to request a referral was recorded as “interest in screening by blood test”. Collection of blood was then registered as screening participation.

Within each group, if participation was not recorded within 6 weeks, a reminder letter was sent, consistent with our previous screening studies (11). After 12 weeks, study invitees in the control and rescue groups who had still not participated were eligible for phase II (Fig. 1).

At 12 weeks, the rescue group nonparticipants were mailed a letter inviting them to screen using the blood test. This invitation was identical to that provided to the choice group invitees in phase I. The control group nonparticipants were mailed a further reminder letter encouraging them to complete the previously provided FIT, or request a replacement if needed. Although the existing NBCSP protocol does not include a reoffer of tests after nonparticipation, we offered this to ensure comparability for analyses with the rescue group invitees. All nonparticipants in phase II were sent another reminder letter after further 6 weeks (week 18).

The survey provided at invitation included the following elements:

1. Demographic variables and past experience with screening

   Information on age, gender, marital status, education, employment status, country of birth, cultural identity, health insurance status, and postcode were collected. The postcode enabled calculation of relative socioeconomic disadvantage Socio-Economic Indexes for Areas (SEIFA) information (ref. 20; a higher score indicates a relative lack of disadvantage). Information about previous experience with blood and fecal testing was also collected.

2. Anxiety about samples involved in screening

   Blood-taking anxiety and fear of needles were measured using an adapted version of measures developed by Deacon and Abramowitz (21). A Fecal Aversion scale was adapted for the study from Cole and colleagues (9) consisting of three items for which agreement was scored on a 4 point scale from 1, strongly disagree to 4, strongly agree. An example item is “collecting bowel motions is distasteful”.

3. Awareness of screening test types

   The survey included questions asking respondents whether they had ever heard of using a stool or blood test for screening in line with the first stage in the Precaution Adoption Process Model (22). Allowed responses were “yes,” “no,” or “unsure”.

An improvement in participation rate of 10% for the intervention group was considered a meaningful difference, and it was expected that the control group would have a participation rate of approximately 50% based on our previous experience. A sample size of 600 per group (n = 1,800 in total) provides more than 90% power to detect a difference in participation rates of 10% when assessed with a χ² test of proportions and a two-sided type I error rate of alpha = 0.05.

Screening participation rates at 12 and 24 weeks were compared between groups using the χ² test or Fisher Exact test. Sociodemographic characteristics were compared with one-way ANOVA and χ² tests for continuous and categorical variables, respectively. Logistic regression analysis was performed to determine predictors of screening participation. A P value less than 0.05 (two-tailed) was accepted as statistically significant. The reliability and internal consistency for adapted survey questions were determined with Cronbach's alpha coefficient.

The sociodemographic characteristics of the study population are shown in Table 1. There were no significant differences in any characteristic: age χ²(8) = 13.507, P = 0.096; gender χ²(2) = 2.110, P = 0.348; and socioeconomic disadvantage χ²(2) = 4.176, P = 0.126.

Figure 1 shows the disposition of study invitees in each of the three groups and participation in response to each intervention and reminder letter over the 24 weeks. A small number of invitees in each group (n = 3–15; total n = 30) withdrew after receipt of the advance notification letter; the remainder (n = 1,770) were provided with screening invitation packs and included in subsequent analyses.

Figure 2 shows cumulative participation rates for each group by phase and according to the test used. Participation rates were not different between groups at the end of phase I [χ² (2df) = 1.02; P = 0.601], and were still similar after 24 weeks [χ² (2df) = 2.19; P = 0.334]. The reinvitation strategies applied to the control and rescue groups in weeks 13–24 resulted in increase in participation for the control group by 3.6% (21/588; FIT only) and for the rescue group by 5.3% (31/585; FIT or blood test; P > 0.05). Participation with the blood test when offered as a rescue strategy (23/353, 6.5%) was greater than when provided as an upfront option in the choice group [9/597; 1.5%; χ² (1df) = 17.08; P < 0.001].

The proportion of participants who were female or male did not differ between tests; 52.1% (317/608) versus 47.9% (291/608) for FIT, respectively, [χ² (1df) = 2.22; P = 0.136] and 56.3% (18/32) and 43.8% (14/32) for the blood test [χ² (1df, Yate's corrected) = 0.563; P = 0.451], respectively. The age distribution of participants is shown for each test in Table 2. Participation did not differ by age using FIT [χ² (4df) = 4.10; P = 0.393] nor for the blood test [but numbers were low; χ² (4df) = 0.265; P = 0.992]. Participation with any type of screening test was less in the age group under 65 years compared with those ≥65 years (33.1% vs. 39.3%; χ² = 7.143; P = 0.008).

Total screening uptake was not associated with socioeconomic status score [expressed as a continuous variable, OR, 1.000; 95% confidence interval (CI), 0.998–1.001; P = 0.765]. Similarly, of those who completed FIT (n = 608), socioeconomic status was not a significant predictor (OR, 1.000; 95% CI, 0.999–1.002; P = 0.837), whereas there was a trend for those who completed the blood test (n = 33) to be from the lower socioeconomic status areas. For each unit increase in socioeconomic status score, there was a 0.5% decrease in odds of completing the blood test, but this did not reach statistical significance (P = 0.075).

Table 3 shows the number of invitees in the rescue and choice groups who were given information for arranging a blood test and then attended a collection center. A significantly greater proportion of the rescue group participants requested the blood test compared with the choice group (P < 0.001) but only 23 of 40 (57.5%) of requesters from the rescue group actually attended for the blood test compared with 9 of 10 of the choice group (90%; Fisher test; P = 0.073). Nonetheless, actual participation with the blood test remained higher in the rescue group (Table 3). There were no differences in demographics between those requesting referral for the blood test without participating, and those who attended the collection center (Supplementary Table S1; all P > 0.26).

Surveys were completed by 560 of 1,770 (31.8%) invitees. Irrespective of study group, 83.8% of survey respondents (469/560) subsequently completed a FIT test, 1.1% (6/560) completed a blood test, and 15.2% (85/560) did not complete any test within the 24-week study duration. Survey completion rates were lower in males (40.5% vs. 52.6%; χ² = 6.53; P = 0.011) and in those aged <65 years compared with ≥65 years (27.8% vs. 37.6%; χ² = 18.85; P < 0.001). There was no significant difference between socioeconomic status of survey respondents and nonrespondents (proportions of respondents and nonrespondents in the most disadvantaged socioeconomic status groups were 37.3% and 38.7%, respectively; P = 0.578). There were also no between-group differences in the proportions completing the survey or in the number of survey respondents who did the FIT (Table 4).

The majority of survey respondents were aware of the fecal test for screening (536/557, 96.2%), and had previously completed a FIT (476/538, 88.5%). In contrast, the majority (76.9%) of respondents were unaware of a blood test for screening [426/554; χ²(1df) = 471.9; P < 0.0001], although the majority had blood taken previously by a health professional (502/548 responses, 91.6%).

The relationship between past participation in FIT-based screening and actual participation in this study, by intervention group, is summarized in Table 4. There was no between-group difference in FIT participation between those who had previously participated and those who had not previously completed a FIT. In contrast, experience in having completed any blood test prior to the study was significantly more likely in rescue and choice Groups (P = 0.001).

There was no significant difference in Fecal Aversion (overall range 3–12) between those who returned a FIT within the study duration (n = 433; mean 5.41; SD 1.91) and those who did not [n = 82; mean 5.37; SD 2.06; t (513) = 0.17; P = 0.86]. The Fecal Aversion scale had good internal consistency, with a Cronbach alpha of 0.875 (n = 515 respondents). Analysis of the impact of scores on the measures blood-taking anxiety, fear of needles, and fecal aversion was not possible within the blood test invitees given the very small number who utilized the test.

In fecal-based screening programs, barriers to participation include fecal aversion and unsuitability of the test on medical grounds, such as presence of benign bleeding conditions (4, 6). A blood test for ctDNA might circumvent these two barriers. However, logistics of this approach, specifically attendance at a blood collection center, might create a different barrier to test completion. This barrier is not seen with fecal sampling, which can be undertaken at home. Furthermore, public understanding of, and familiarity with, blood testing for colorectal cancer screening is likely to be poorer than for fecal sampling and impact on potential participation, regardless of whether blood tests are offered as an option or as a rescue strategy. Strategies to implement a blood test, and the resulting participation in a screening context in a country that uses mailed FIT for the centrally organized screening program, have not been previously investigated.

This study assessed two strategies for implementing a blood test within a mailed FIT-based screening program. The blood test used was one that assays for the ctDNA biomarkers, methylated BCAT1 and IKZF1, and which is known to detect approximately two-thirds of patients with colorectal cancer (16, 17). We measured screening participation using a rescue strategy (offer of the blood test to FIT nonparticipants) and a choice strategy (as an upfront alternative option to FIT). We also tested relevant behavioral variables, awareness, and aversion, to see if they related to test usage.

In this screening context, and judging on overall participation rates after 24 weeks, neither the rescue nor choice strategies improved test-independent participation rates. We observed only a small number of individuals completing the blood test. This is in contrast to the Australian research that found that when presented with a hypothetical choice of screening tests, the majority had a strong preference for a blood test compared with a stool test (79% vs. 20%; ref. 4). Likelihood of participating was significantly negatively associated with increased health system interactions, and the blood screening scenarios featured increased interactions (23). Thus, although people may indicate acceptance of blood testing, reported acceptance may not improve participation rates due to perceptions that the approach requires significant time commitment and logistical complexity. Moreover, reported intention to engage in a behavior does not necessarily predict actual behavior. A meta-analysis of 10 meta-analyses of the intention–behavior relationship conducted by Sheeran (24) indicated that intentions only explained 28% of the variance of future behavior. Despite this, we found that among those who requested information about the blood test, the majority attended blood collection. In the study design, we provided multiple blood collection locations within each electorate region to provide convenient access to phlebotomy to reduce this as a barrier for screening.

In those that did complete the blood test, there was a trend toward greater uptake in those from lower socioeconomic status areas. The Australian colorectal cancer screening program, as well as other studies, report that participation with FIT screening is positively correlated with socioeconomic status (25, 26). Our finding may therefore have strategic implications for screening rollout, and targeted rescue strategies could be trialed. It is also likely that preference for FIT and blood test could differ by ethnicity. One study that compared screening test type preferences (colonoscopy vs. blood test) found that there were racial differences (27). We were unable to analyze participation data in relation to ethnicity as this information was unavailable for the entire invited cohort.

Another obvious explanation for limited use of the blood test when provided in this context is the dramatic difference in awareness of the ability to screen with a fecal compared with a blood test (96% vs. 23%). This difference in awareness was likely to affect perceived test efficacy; future research should measure trust in the screening technology and its influence on participation. It is intriguing that participation using the blood test was more likely when it was offered in a rescue strategy than as an upfront option. This emphasizes the importance of context when offering screening tests; the rationale for using the blood test may be clearer to someone failing to engage in the FIT, even when awareness of the fecal test was much higher. Nonetheless, paying particular attention to nonparticipants generates a response in itself and using a blood test to rescue nonparticipants is clearly preferred to FIT.

It is important to determine the best means to implement a new test into an existing screening program. We investigated both sequential offers and choice in this study. In relation to the former, previous studies show that sequential offers of different colorectal cancer screening tests can improve overall participation rates (28, 29). The second strategy raises issues about choice. Australian research shows that offering multiple options creates uncertainty (30), but it may also improve participation by allowing individuals to choose a test which aligns with their preferences (31). Given that most people of screening age are familiar with having blood taken, we believed that this option might reduce barriers associated with having to make a choice because familiarity can overcome perceived barriers (4). In addition, the offer of choice in screening modality has achieved significant support with the American Cancer Society strongly advocating for patients to be given the opportunity to make an “informed” choice of test (32). In this study, the anticipated benefit of offering patients choice in decision-making about screening test (e.g., Indamoni and colleagues; ref. 33), however did not lead to increased participation; no benefit was seen comparing participation at conclusion of phase I between the choice group and either the control or rescue groups. Surprisingly, only 1.5% of those who participated in the choice group opted to receive the blood test. Neither the idea of an option or the nature of the option itself achieved better participation. There have been few previous studies examining the effectiveness of providing a choice in screening tests. None of them has resulted in enhanced participation when the choice is offered at the outset (34, 35). This is referred to as the “choice paradox,” where allowing individuals to choose between screening strategies results in a lower participation rate instead of the expected improvement in uptake (36). Potentially the convenience of the home fecal test compared with the time and expenses associated with the requirement to attend for venesection overrides the other influences. The blood test obviously added complexity to screening logistics because not everyone who requested a blood test referral attended a collection center.

It might also be expected that anxiety about having blood taken or aversion to sampling feces might influence test choice as these have previously been identified as potential barriers, but in this study fecal aversion did not appear to account for completion or nonparticipation with FIT. The number of survey respondents was too small to allow assessment of anxiety about blood collection.

A strength of this study is that it is applicable to population-based organized-screening programs with FIT, a standard practice in many countries (1). However, this effect on participation is not necessarily relevant to personalized screening delivered through direct discussion with a health professional. Furthermore, in such settings, facilities for blood collection are likely to be close at hand and the more complex logistics of blood sampling are negated because a special appointment is not otherwise needed. The blood test is more likely to be taken up at a time when a person attends a health professional and further studies to investigate blood test rescue strategies in this context are warranted.

As the blood test was new, a sizable amount of documentation (nature, value, and risks) was provided when the blood test was offered to achieve adequately informed consent. This was a limitation to the study and may have created information overload and lead to confusion in these groups resulting in blunted participation with the blood test. In addition, the provision of the survey prior to completion of the intervention could have added to this overload and compromised participation; however, such an effect would have applied to all groups. A further limitation was that not all groups within the study received the same number of reminders for screening participation. The Choice group only received a reminder following 6 weeks of nonparticipation, to mimic the usual practice of the Australian National Bowel Cancer Screening Program; while the control and rescue groups received additional contact at the 12 and 18 week timepoints. A further reminder was unlikely to significantly change overall outcomes, but nonetheless it remains a study limitation. In addition, this study was conducted in the general population of Australia, where a sizable proportion of the population is known to be up-to-date with screening (37).

In conclusion, provision of a blood test either as a rescue of FIT nonparticipants or as an upfront choice to FIT, did not increase overall participation in a population familiar with FIT-screening offered in an organized program through a centralized mail-out. This might in part be due to convenience of sampling at home with FIT relative to the inconvenience of needing to attend for a blood test but lack of awareness of the blood test as a screening option may also be a contributing factor. Although the two tests have comparable sensitivity for cancers, the blood test has poor performance characteristics for adenoma detection (17), and therefore FIT should remain the first-line test to be offered in screening programs. However, use of blood tests for screening are an appropriate rescue mechanism for those unwilling or unable to complete FIT for behavioral or medical reasons (6). Furthermore, the context in which the blood test is offered is important as it was taken up more often when offered as a rescue to nonparticipants than as a choice at the initial invitation. It is possible that the blood test might prove more useful when offered in the context of personalized screening where health professionals are directly involved in the invitation and blood collection facilities are close at hand.

S.K. Pedersen has ownership interest (including stock, patents, etc.) in Clinical Genomics. G.P. Young reports receiving a commercial research grant from and is a consultant/advisory board member for Clinical Genomics Technologies. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E.L. Symonds, R. Woodman, G. Chen, J. Ratcliffe, R.J.L. Fraser, C.J. Wilson, G.P. Young

Development of methodology: E.L. Symonds, I. Flight, J. Ratcliffe, R.J.L. Fraser, C.J. Wilson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.L. Symonds, S.K. Pedersen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.L. Symonds, D. Hughes, I. Flight, R. Woodman, G. Chen, J. Ratcliffe, R.J.L. Fraser, C.J. Wilson, G.P. Young

Writing, review, and/or revision of the manuscript: E.L. Symonds, D. Hughes, I. Flight, R. Woodman, G. Chen, J. Ratcliffe, S.K. Pedersen, R.J.L. Fraser, C.J. Wilson, G.P. Young

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.L. Symonds, D. Hughes

Study supervision: E.L. Symonds

Research funding for this study was provided by the Cancer Council SA Beat Cancer Research Project Grant (APP1080541), and the National Health and Medical Research Council Project Grant (APP1101837). Authors G. Chen and I. Flight were supported by a grant funded by the financial support of Cancer Council SA's Beat Cancer Project on behalf of its donors and the State Government of South Australia through the Department of Health together with the support of the Flinders Medical Centre Foundation, its donors and partners.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.