Mapping colorectal cancer screening workflows in primary care

Abstract

Introduction

Colorectal cancer (CRC) is preventable and treatable when detected early, but mortality from the disease remains high due to low uptake of preventive screening. Primary care plays a crucial role in CRC screening, but still often falls short of screening targets. We sought to map the CRC screening workflows of primary care clinics to describe how preventive CRC screening for normal risk patients is conducted, and how these workflows differ across clinic types.

Methods

Eight primary care clinics were recruited with ‘high’ or ‘low’ rates of CRC screening. Clinicians and staff from each clinic were interviewed about their CRC screening workflow. Workflow maps were created based on individual interview responses, then reviewed and revised in collaboration with each clinic. Interview data and resulting workflow maps were evaluated for similarities and differences across clinic characteristics.

Results

CRC screening rates ranged from 54.5% to 90% across the eight clinics. Clinics shared similarities in how they identified patients eligible for CRC screening, how they communicated the need for screening, the types of screening tests offered and how they communicated test results. Clinics with higher rates of CRC screening particularly described more communication of screening need, more emphasis on screening during the clinical conversation and workflow processes to follow up on missing results.

Conclusions

The CRC screening workflows described by clinics shared common pathway stages, but variation in how these stages were executed indicates the benefit of patient decision tools, clinical communication strategies, patient navigation and results tracking to increase CRC screening uptake.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• To date, there have been no studies exploring the workflow/s that primary care clinics use to accomplish this screening. This study describes the CRC screening workflows in eight primary care clinics across the USA.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This information can serve as a baseline for future studies and quality improvement initiatives aimed at increasing CRC screening uptake.

Introduction

Colorectal cancer (CRC) is preventable and treatable with routine screening. Screening tests that are low cost and/or covered by most insurance are effective in detecting early cancer and precancerous polyps. When caught early, the 5-year survival rate for CRC is 91%.¹ While screening has contributed to an overall decline in CRC mortality, it remains underutilised across demographic groups such as race and age. Between 1975 and 2013, increased screening reduced CRC mortality by 50% among whites, but only 26% among blacks.^{2 3} In 2021, only the 65–74-year-old age group achieved the national target screening rate of 80%, while 45–49-year olds and 50–54-year olds only achieved 20% and 50% screening rates, respectively.⁴ CRC was also the second most deadly cancer in the USA in 2022, responsible for over 50,000 deaths. Population-based estimates predict a similar number of CRC deaths again in 2025,⁵ prompting a closer look at the disconnect between effective and available screening versus screening uptake.

Primary care plays a critical role in promoting and providing access to CRC screening. Screening-eligible adults with access to primary care are 3 to 5 times more likely to be up to date on screening than those without,^{6 7} and over 80% of patients who receive a physician recommendation complete CRC screening.^{8 9} Despite this, primary care clinics still struggle to achieve target screening rates for a myriad of reasons, including competing demands within a visit, inadequate tracking systems, lack of knowledge or training in screening modalities and patient reluctance.10,14 The research on barriers to CRC screening in primary care is plentiful, but much of it rests on the self-reported experience of patients or clinicians, leaving gaps in our knowledge of the individual steps taken between the identification of a patient eligible for CRC screening and that patient’s completion of a screening test. Understanding how primary care clinical workflows facilitate each step in this process is the necessary next step to identify where these barriers typically occur and where they can be best addressed. To our knowledge, there are no studies in the current literature that document existing CRC screening workflows in primary care clinics.

Workflow or process mapping is relatively new in healthcare research, with the majority of studies that utilise it being published since 2010.¹⁵ Despite this, it has quickly become an integral component of quality improvement (QI) methodology as a means of setting a baseline understanding for all stakeholders of the existing series of steps to accomplish a particular task.¹⁶ While typically utilised within single clinics or health systems to enact local QI, workflow mapping has also been used across similar sites to characterise common barriers and facilitators to particular workflows related to screening, treatment or referrals and to identify key differences that may impact outcomes.17,20 Used in this way, workflow mapping can facilitate the development of tools or interventions that are generalisable to similar sites.^{21 22} In this study, we used workflow mapping to understand the process of CRC screening across multiple primary care clinics. The purpose of this study was to generate workflow maps for CRC screening activities for normal risk patients in primary care clinics, from the identification of eligible patients to the communication of screening results, to identify common elements for screening completion and strategies to support patient navigation.

Methods

Study design

This was a qualitative study, employing individual and group interviews to understand primary care clinics’ workflows for preventive CRC screening in normal risk patients. Reporting is based on the Consensus Reporting Items for Studies in Primary Care checklist.²³

Participants

Clinics were recruited from the American Academy of Family Physicians (AAFP) National Research Network (NRN) at the DARTNet Institute. Clinics of the NRN were invited via email to indicate their interest in participating in the study by completing a brief survey. The survey collected their self-reported CRC screening rates and clinic characteristics. We sought to enrol a total of eight practices: four with self-reported practice-level CRC screening rates of 70% or greater (‘higher’ screening rate) and four with screening rates below 70% (‘lower’ screening rate) to observe any differences in screening workflows between groups. The cut point of 70% was chosen based on the national US CRC screening rate of 69.9% in 2021,²⁴ prior to the US Preventive Services Task Force 2021 update, which lowered the recommended screening age for normal-risk adults from 50 to 45.²⁵ The two groups are referred to as those with ‘higher’ or ‘lower’ screening rates for ease of discussion. Clinic staff were then purposively sampled to include at least two individuals (in either clinical or non-clinical roles) who were directly involved in some or all aspects of the CRC screening workflow (identifying eligible patients, clinical conversation, test ordering, results follow-up) to participate in study activities.

Data collection

Data were collected in individual interviews, group interviews and surveys. Data were collected in a sequence of study activities conducted with each clinic, represented in figure 1 and detailed below.

Figure 1. Study activities workflow. CRC, colorectal cancer.

Phase 1 interview

Individuals who were identified as having a role in the CRC screening process within the clinic participated in a 1-on-1 interview with a member of the study team. We used a semistructured interview guide (online supplemental Appendix A) to ask individuals about the CRC screening workflow protocol in their clinic as they understood it. Interview questions and probes sought to clarify each step of the workflow, ranging from the identification of eligible patients through to the receipt and communication of test results. Four members of the study team conducted the phase 1 interviews (EAR, CMH, JWL, CBL), and each generated detailed notes of their interview. After the interviewer completed their notes, the audio recording of the call and the interviewer’s notes were peer reviewed by a separate study team member for verification (EAR, CMH, CBL). After their interview, individuals completed a brief survey, which collected their role and years in clinic.

Workflow mapping

Two members of the study team (CMH and JWL) reviewed the notes and transcripts of each phase 1 interview to create a comprehensive visual map representing each clinic’s CRC screening workflow based on review of each individual’s description of the process. Maps were created using Microsoft Visio.²⁶

Phase 2 interview

On completion of clinic-level workflow maps, phase 1 interviewees participated in a phase 2 group interview in which all participating clinic staff members were invited to review the visual workflow map together, provide feedback and suggest adjustments. The interviewers annotated the workflow maps to capture the participants’ feedback. Three members of the study team conducted phase 2 interviews (CMH, JWL, CBL).

Workflow map revision

Following phase 2 interviews, all members of the study team reviewed and discussed the comments and feedback received from clinics during the phase 2 group interview and adjusted the workflow map accordingly. After reviewing all comments from the group interview, JWL produced a final workflow map for each clinic.

Data analysis

We conducted thematic analysis using individual and group discussion data generated during the interviews as well as study team meeting notes to identify patterns and assess how they varied across clinical contexts.²⁷ Deductive coding was guided by the workflow domains outlined in the semistructured interview guides. This combination of process mapping and deductive thematic analysis has been demonstrated in similar studies on healthcare processes^{28 29} and allowed for comparison across clinic characteristics to better evaluate workflow differences that may influence screening outcomes.

Interview recordings, notes and practice workflow maps were stored securely on encrypted servers and accessible only by the study team.

Ethical approval

This study received exempt status (category 2) by the Institutional Review Board at the AAFP due to minimal risks to participation, the absence of collected personal health information and the voluntary nature of participation. Prior to completing study activities, individuals were presented with study information, which included details of the study aims and purpose, the type and duration of study activities, the risks, benefits and voluntary nature of participating, and how their data would be utilised and protected. At the beginning of their first interview, participants were asked to confirm they had received and reviewed this information, had the opportunity to discuss any questions or concerns, and provided verbal consent to proceed with the study activities. Although interviews focused on details of established clinic workflows rather than personal health/healthcare experience, interviews were first conducted individually to mitigate any discomfort related to role hierarchy.

Activities for this study were conducted in 2024.

Patient and public involvement

Patients and the public were not involved in the planning or conducting of the research. At the end of data analysis, each participating clinic received a study report with their final CRC screening workflow map and the deidentified workflow maps of the other clinics.

Reflexivity statement

The research team brings a diverse range of expertise, including practice-based primary care research (all members), over 40 years of clinical primary care experience (JWL), 15 years of research in CRC (CMH), qualitative methodology (EAR) and a decade of research project management (EAR and CBL). We acknowledge that our professional backgrounds and prior experiences may shape how we interpret data and engage with participants. JWL’s long-standing clinical role may influence assumptions about workflow feasibility, while CMH’s focus on CRC research may heighten sensitivity to screening gaps. EAR and CBL’s experience in project management and qualitative methods may guide the structuring and interpretation of workflow narratives. These potential biases were mitigated by engaging all study team members in the interview process and member-checking all interview notes as well as by confirming the accuracy of the workflow maps with the clinic members themselves.

Results

Participants

Eight primary care clinics from across the US participated, with self-reported CRC screening rates ranging from 54.5% to 90% (table 1). Clinics represented various clinic types in rural and urban settings across seven states. The number of participants from each practice ranged from 2 to 5 and included clinical and non-clinical roles such as physicians, clinical staff, front desk attendants and office managers. Clinic and participant characteristics are detailed in table 1.

Table 1. Clinic characteristics.

Clinics				Participants
ID	Clinic type	Location	Reported screening rate	Role	Years in clinic
1	Affiliated with Academic Medical Center, Rural Health Clinic	MO*	54.50%	Medical director/physician	>15
				Charge licensed practical nurse	>15
				Health records technician	1-2
				Certified medical assistant	<1
				Licensed practical nurse	11–15
2	Academic Center-Based Residency Program, Community Health Center	D.C.	60%	Chief resident of QI and research	1–2
				Chief resident of QI and research	3–5
				Physician	6-10
3	Academic Center-Based Residency Program	AZ	62.30%	Physician	11–15
				Physician	6–10
				Population health medical assistant	6–10
				Lead medical assistant	6–10
4	Primary care only	VA	65%	Physician	>15
				Office manager	3–5
				Receptionist	1–2
5	Primary care only	TX	70%	Physician	N/A
				Nurse practitioner	6–10
				Clinical research director	N/A
6	Primary care only	NY	78.30%	Physician	>15
				Office manager	11–15
				Registered nurse	1–2
				Front desk	3–5
7	Primary care only	PA*	85%	Physician	>15
				Office manager	>15
8	Primary care only	PA	90%	Physician	>15
				Office manager	11–15

^*Located in rural area.

QI, quality improvement.

CRC screening workflows

The CRC screening workflows described by clinics shared common pathway stages (figure 2), with some variation in stage execution (see online supplemental Appendix B). These shared high-level stages and variations are described below, while distinct differences between clinics with higher and lower screening rates are detailed in subsequent sections.

Figure 2. Aggregate CRC screening workflows of primary care clinics with intervention recommendations. CRC, colorectal cancer; EHR, electronic health record; FIT, fecal immunochemical (stool) test; GI, gastrointestinal.

Identification of and communication with eligible patients

Patients eligible for CRC screening were often identified by office staff via direct chart review, and some clinics additionally utilised insurance database reports, accountable care organisation (ACO) reports and/or electronic health record (EHR) extensions or applications. Patients with upcoming visits were identified as eligible for screening the day of or up to 2 weeks prior. Clinicians communicated to patients that they were due for screening during visits and explained why screening is important and what types of screening tests are available.

Screening test offered, recommended and ordered

Most clinics offered colonoscopy, the Cologuard test and the fecal immunochemical (stool) test (FIT). They described recommending colonoscopy as the most thorough and accurate screening test, recommending the Cologuard test to patients who were low risk or unlikely to complete a colonoscopy, and utilising the FIT as an inexpensive option for uninsured or screening-resistant patients. Clinicians were typically responsible for ordering a referral for colonoscopy, while clinicians and office staff both could be responsible for ordering a Cologuard test kit or distributing an FIT card.

Communicating results and next steps

Physicians most often communicated positive results to a patient, while clinical or office staff communicated negative results by phone or portal message. Follow-up and future screening recommendations varied depending on results and screening type and were entered into EHR charts on receipt of results. Referral processes for diagnostic colonoscopy varied depending on whether the clinic had access to a gastrointestinal (GI) clinic through their associated health system or an established partner. For clinics with this access, colonoscopy scheduling was typically provided by the clinic to which the patient was referred. If referral was made to a GI clinic outside of the clinic’s health system or established partner, the referring clinic typically took an active role in ensuring that scheduling and follow-up did in fact occur.

While workflows shared similar process stages across all sites, certain key differences were identified in both how these stages were operationalised and in the characteristics of the clinics when comparing data between the groups with ‘high’ (≥70%) vs ‘low’ (<70%) screening rates. Comparison of clinic workflow variation is detailed in online supplemental Appendix B.

Workflow characteristics of clinics with lower screening rates

Of the clinics with lower screening rates, three of these were academic clinics (clinics 1, 2 and 3). To identify eligible patients, clinics in this group described using more staff and tools prior to a scheduled visit. For example, they described using some combination of ACO reports, automated EHR alerts and external applications to identify patients in addition to direct chart review. Some also described designated staff, such as a health records technician or a population health team member, who reviewed patient eligibility and identified patients due for screening in addition to office and clinical staff.

During the communication of screening need, some of these clinics with lower screening rates did not immediately present all available screening options. Clinics 1 and 2 described that they would offer one screening method first and wait for the patient to decline or inquire before offering alternatives. For example, in clinic 1, which served a rural area, gastroenterologists were in short supply with long waitlists, so unless a patient was high-risk for CRC (based on symptoms or family history), only the FIT was offered. In contrast, colonoscopy was offered first in clinic 2, but if a patient seemed resistant or unlikely to complete it, the Cologuard test would be offered next, then the FIT if necessary.

Clinics 1 and 2 did not describe any additional support for patients to complete their test. Clinic 3 reported that they would assist in scheduling a colonoscopy, but only if the GI clinic failed to contact the patient initially. Clinic 4 described that the office manager would walk the patient through the next steps of their Cologuard test during check-out; how it was ordered, what it would look like, when to expect it in the mail and to reach out to the clinic if it did not arrive.

Finally, none of these clinics described workflows for following up on missing screening results, stating that any missing results may be caught or addressed during the next eligibility screening.

Workflow characteristics of clinics with higher screening rates

In contrast, clinics with higher screening rates were all private primary care clinics (clinics 5–8). During the communication of screening need, Clinics 5, 6 and 7 described that clinical staff communicated screening options to eligible patients in addition to the physician, so that patients often had multiple conversations about CRC screening within one visit. Clinical staff also explicitly described emphasising the value of screening and taking time to educate patients about each available test. Physicians from these clinics additionally described their clinical conversations as ‘long’ (clinic 5) or ‘extensive’ (clinic 6). Clinics 5, 7 and 8 also described using real examples in their clinical conversations, such as other patients who benefitted from screening, CRC affecting someone in their own life, or celebrities who have died due to CRC. Physicians in these clinics reported that they would begin the clinical conversation by presenting each screening test option and the pros and cons of each.

Clinics 7 and 8 additionally described supporting patients to complete the Cologuard by walking through the details for completing and returning the test on check-out. Clinic 8 also described instructing patients to reach out for assistance in scheduling with the GI clinic if they were not contacted within a certain time frame.

Finally, clinics 6, 7 and 8 described a specific workflow stage for following up on missing screening results, either through direct outreach when results were not received within a few weeks, or by noting this in the patient’s chart to address at the next visit.

Between the two groups of clinics, the timing of the screening communication and the availability of educational resources were also examined, but no discernible differences were found. Clinics across both groups described communicating screening need at annual physicals only, at any visit or even contacting patients between visits. Similarly, clinics varied in the educational materials they made available to patients, from none at all, to printed pamphlets, to demonstration kits of the tests.

Discussion

Workflow mapping is an effective tool to visualise a standard process, such as preventive screening, and enact QI initiatives specifically tailored to a clinic’s needs and setting.^{15 30} By comparing workflows across multiple sites, differences in workflow characteristics can reveal more generalisable strategies for improvement that can be adopted across diverse settings.^{31 32} However, our results highlight how clinical workflows can appear similar in their pathway stages, yet can vary meaningfully in the way each stage is operationalised across clinics. In this study, we found that primary care clinics of different types and settings describe common pathway stages for preventive CRC screening in normal risk patients (figure 2) and identified key pathway stages where clinics with ‘high’ and ‘low’ screening rates diverged in the ways that they executed these stages. These were Communication of Screening Need, the Clinical Conversation and Follow Up on Missing Results (figure 2). Here, we synthesise these differences and offer recommendations for interventions that may be adapted by primary care clinics and should be further studied for their effect on CRC screening uptake.

Incorporate patient decision-making tools

Clinics with higher screening rates described multiple roles involved in communicating the need for CRC screening to the patient. Clinic staff described presenting information such as the types of screening tests available and what each entails in addition to the clinical conversation that would later be conducted by the physician. This suggests that implementing patient information and decision-aid tools may be particularly effective in the Communication of Screening Need stage by priming the patient for the clinical conversation. Patient-facing shared decision-making applications that aid in choosing a preferred CRC screening test and guide through the logistical steps to complete that test have been shown to nearly double the screening completion rate compared with health education alone.³³ Presenting this type of tool ahead of the clinical conversation may further encourage uptake by prompting more discussion or questions from the patient once they see their physician. Similar examples of ‘communication-priming’ have been shown to be effective for goals-of-care discussions for patients with serious illness, but research on this strategy for preventative services is lacking.^{34 35}

Expand the clinical conversation

Clinics with higher screening rates also described more time and emphasis given to the clinical conversation itself. In addition to describing each screening test option and their pros and cons up front, clinicians also provided anecdotes and examples to illustrate the importance of screening. Studies have found that narrative or storytelling is effective in influencing patients’ decision to complete cancer screening, but these typically focus on patient stories presented outside of the clinical conversation.^{36 37} Similarly, studies on the effect of anecdotal evidence have found that negative anecdotes of medical treatments or procedures are more influential than statistical evidence in patient decision-making,^{38 39} suggesting that there may be a mirrored effect of clinician-provided anecdotes of procedure benefits. Recommendations for effective clinical conversations typically focus on communication strategies such as active listening, shared decision-making and attention to patients’ identity goals, and while these have been found to increase CRC screening uptake, only shared decision-making was described by one physician in this study.^{40 41} Instead, these clinicians described presenting facts thoroughly (test types, pros and cons) as well relatably, highlighting real people and their experiences. Time constraints and competing demands are a documented barrier to CRC screening conversations,^{10 42} but when time allows, the use of narrative strategies may further increase CRC screening uptake.

Include patient navigators and processes for missing results

Clinic workflows diverged most significantly after screening tests were ordered, where clinics with higher rates had processes to follow-up on missing results, and clinics with lower rates did not. Loss to follow-up on test results is a well-documented issue, but studies typically focus on diagnostic testing rather than screening.43,45 However, loss of screening test results poses many of the same threats to health outcomes. Studies have provided recommendations to address the gap between referral and referring clinics, but these often focus on technology and only address a portion of the problem.^{46 47} For CRC screening in particular, the patient is responsible for several steps to follow through with the testing outside of the visit, either by completing a kit at home and mailing or returning it to the clinic or by scheduling with a gastroenterologist and completing a time-consuming preparation process. This presents several points between test ordering and receipt of test results where loss may occur due to the patient, the GI clinic, the mailing system, the lab or the physician’s clinic. These non-clinic opportunities for loss are better addressed by patient navigation interventions, which are effective in increasing CRC screening uptake and completion.48,51 Patient navigation activities, such as assisting with colonoscopy scheduling and follow-up phone calls to ensure Cologuard and FIT kits are received and completed, should be combined with clinical tracking systems to increase screening completion to reduce loss to follow-up.

Finally, it is noteworthy that of all eight clinics, those with the lowest screening rates were academically affiliated clinics (clinics 1–3), and those with the highest screening rates were independent primary care clinics (clinics 4–8). This may be due to the presence of trainees (residents) in the clinic, which has been shown to be a significant factor in loss of test follow-up due to frequent rotations that inhibit familiarity and mastery of a particular clinic’s workflow process.^{43 52 53} Additionally, patients of smaller independent clinics may have a greater rapport with their physician, which is associated with receiving more guideline-consistent care, including completing cancer screening.^{54 55}

Limitations

There are limitations to this study: first, this small pilot sample cannot ensure generalisability to all primary care clinics. However, our sample does include diverse clinic types, settings and locations, and the range of CRC screening rates in these clinics is similar to what has been reported in other studies.^{56 57} Second, CRC screening rates were self-reported by clinics prior to study enrollment and were not verified by the study team. Future studies should include standardised calculation and validation of screening rates for comparison.

Finally, workflow descriptions may not capture the complexities of CRC screening within the overarching, system-level clinic or appointment workflow, including other preventive and diagnostic screening or the variations that may occur over time due to changes in staff roles, the procedures or availability of specific screening tests, insurance coverage, cost, etc. This was mitigated by gathering workflow descriptions from a variety of roles involved in the process and verifying these in the group interview, which has been demonstrated as effective in similar workflow studies^{15 19}; however, roles or system processes important to the workflow may have been missed. For example, for clinics that rely on other departments or EHR automation in certain workflow stages, these subprocesses should be examined to fully map how each domain stage is operationalised. Future studies using a system-based lens a priori could capture these ancillary workflows and provide greater detail to the conceptual domains that were described by clinic teams here.

Conclusion

The results of this study demonstrate the value of primary care in addressing CRC screening and improving CRC outcomes for patients. These findings provide practical pathways and explicit steps in the CRC screening process for researchers and clinical teams alike to investigate and target for further improvements in CRC screening uptake.

Data availability statement

No data are available.