Implementation and Evaluation of a Novel Colorectal Cancer Decision Aid Using a Centralized Delivery Strategy

Abstract

Background

Colorectal cancer is the second leading cause of cancer-related deaths in the US; however, colorectal cancer screening reduces both incidence and mortality rates. Patient decision aids are an evidence-based strategy to support patients making health related decisions. Colorectal cancer screening decision aids can be unsuccessful due to provider preferences for colonoscopy and lack of effective decision aid implementation strategies within clinical settings.

Methods

We conducted a hybrid implementation-effectiveness study testing the feasibility of using an existing centralized preventive health screening outreach infrastructure to implement a novel CRC decision aid across a health care system. Participants included primary care patients at one of three study clinics. Implementation was assessed by determining whether patients remembered receiving the DA and were aware of CRC screening options. Effectiveness was measured by comparing overall screening rates between the control and intervention groups.

Results

Using a centralized delivery system was a feasible and efficient method for implementing decision aids to a large academic health system. More than 90% of the intervention group remembered receiving the decision aid and 80% found it helpful in their decision making process. The decision aid was successful in improving colorectal cancer screening knowledge, however, overall CRC rates significantly decreased between the control and intervention periods (51% vs. 39% respectively, p = 0.03).

Conclusions

Centralized delivery is a feasible method for DA implementation. Although DAs increase knowledge, the true effectiveness of CRC DAs in clinical settings is unknown as result of the number in screening tests, diversity in DA format, and the variability in dissemination and implementation practices.

Introduction

Colorectal cancer is the second leading cause of cancer-related deaths in the US. However, many deaths are preventable with routine colorectal cancer screening.¹ The United States Preventive Services Task Force (USPSTF) recommends colorectal (CRC) cancer screening for asymptomatic adults from ages 50-75, yet more than a third of Americans are overdue for screening.^2-4 There are several types of screening tests for CRC, including invasive imaging tests (e.g. colonoscopy) and less invasive stool blood testing.⁵ Both colonoscopy and stool blood testing reduce the incidence and mortality rates of colorectal cancer equally; however, there are no clinical trials comparing their effectiveness side-by-side. Most physicians generally recommend colonoscopy,^6,7 which does not always align with patient preference. Offering patients a choice in screening options has been shown to increase overall screening rates.⁸

Patient decisions aids (DAs) are an evidence-based strategy to support patients making a variety of medical decisions including CRC screening decisions. DAs support patients in the process of decision making by providing information on diagnosis and/or treatment options and helping patients clarify their values and treatment goals.^9-11 DAs assume various forms including: paper (e.g. leaflets, pamphlets, and booklets), electronic media (e.g. audio recordings, and videos) and web-based interactive portals.¹²

There are several colorectal cancer DAs are currently available.¹³ Numerous randomized control trials have shown CRC DAs to be efficacious in increasing CRC screening rates.^8,10,11 However, implementing CRC DAs in routine clinical care comes with many challenges. First, there is a lack of DA delivery structures that facilitates implementation. Second, there is a strong physician bias for colonoscopy because the 10 year follow-up period is easier to manage versus more frequent follow-up for FIT and because physicians generally believe colonoscopy to be a more robust test.^7,12-14

Herein, we describe a pilot study designed to overcome and acknowledge these implementation challenges. First, at the clinic level, we utilized a novel centralized delivery mechanism.¹⁵ The metro region of the University of Colorado healthcare system (UCHealth Metro) uses a centralized approach for managing preventive health screenings throughout its primary care clinics. This model preserves limited clinical resources by outsourcing preventive health patient contact to non-medical staff. We collaborated with the staff of this existing infrastructure to test the system-wide implementation and adoption of a CRC screening DA. Second, to overcome clinician barriers, we developed an innovative CRC DA that acknowledges physician preference for colonoscopy as the best screening test but also recognizes the value of offering patients a choice. Our DA endeavored to facilitate screening completion while preserving patient choice by reinforcing the importance of CRC screening regardless of test chosen.^16-18 We incorporated design elements in the DA to empower patients and activate them to complete screening.

Within the UCHealth Metro system, CRC screening-eligible patients are only offered colonoscopy. The current default of colonoscopy only in our system presented us with a unique opportunity to investigate a naturalistic intervention of implementing a choice-based CRC screening approach. Therefore, the purpose of this pilot study was twofold. First, we aimed to determine if the established centralized outreach infrastructure is an appropriate mechanism to implement DAs across a large academic health care system. Second, we evaluated the acceptability and preliminary effectiveness of the DA among patients eligible for CRC screening.

Methods

Overview

We performed a pilot hybrid implementation and effectiveness study of a DA for CRC screening using a centralized delivery system. There were two study periods – a control and an intervention. The enrollment period for the control period was from May 2015 thru November 2015, and the intervention period was from February 2016 thru July 2016. The study protocol was approved by the Colorado Multiple Institutional Review Board (COMIRB).

Decision Aid Development

We used a stepwise process of DA development which conforms to the International Patient Decision Aid Standards (IPDAS).^19,20 We included only colonoscopy and stool blood testing in the DA as these are the most common tests and because research suggests too many options can create confusion.⁸ Moreover, colonoscopy is most physicians’ preferred screening methodology, and most physicians do not consider “no screening” to be an acceptable screening option.^6,7 We created a paper and video DA incorporating feedback from patients and developed several iterations of the DAs, revising each version based on feedback received. We followed the recommendations for clear communication, which ensured both the paper and video versions were written at a 5^th grade level, using the Flesch-Kincaid scoring available in Microsoft word.^21-23

Fuzzy trace theory (FTT) guided the development of the DA. FTT suggests that patients make decisions based on the bottom-line meaning they derive from information.²⁴ DAs using a FTT framework focus on presenting information using gisted, or summarized, statements that convey the main points. FTT de-emphasizes the importance of quantitatively oriented DAs by arguing that many patients lack the numeracy skills to accurately interpret numerical risk and even those who do possess higher numeracy tend to make decisions based on their gist interpretations.²⁴

Because we used FTT as the guiding theoretical foundation of the DAs, it was important to validate the gist by confirming patients understood the bottom-line message intended by the DAs. We achieved this goal by showing the final version of the DAs to eight patient volunteers and asking them to summarize the main idea of the DAs and answer questions regarding the information in the DAs. We also asked ten experts, including clinicians and shared decision-making researchers, to review the DAs and provide feedback. We found that patients understood the gist and experts felt the gist of the DAs accurately represented the information they convey to patients. Additionally, we incorporated other elements as recommended by the IPDAS criteria (Please see our, website www.patientdecision.org, to view paper and video decision aids).^19,20

Study setting

The centralized patient outreach program for preventive health screenings at UCHealth Metro is called Ambulatory Health Promotion (AHP). AHP employs outreach specialists responsible for tracking and scheduling preventive health screening tests as well as chronic disease management tests and follow-up appointments. AHP utilizes electronic medical record software which generates a registry of patients due for health screenings across the UCHealth Metro network. Once identified, patients due for screening are contacted by mail and telephone to discuss and schedule screenings. Utilization of non-medical outreach specialists has been shown to improve screening rates for preventive healthcare,²⁵ decrease time and financial burdens at the clinic level^26,27 and improve patient satisfaction.^26,28

We recruited three of seven primary care sites (two general internal medicine and one family medicine practice) that are part of UCHealth Metro region. These practices include a total of 27 academic physicians and/or full time clinical equivalent. The three primary care clinics selected for the study represent the largest patient populations in UCHealth Metro central primary care.

Participants

Eligible participants met the USPSTF Grade A recommendation for CRC screening, which includes asymptomatic adults age 50-75.²⁹ Ineligibility criteria included conditions which increased risk of CRC, such as a family history of CRC, any type of ostomy pouch, and patients in CRC surveillance. We additionally excluded patients with end-stage renal disease and patients using anticoagulants because these higher risk conditions require providers to initiate screening within the UCHealth Metro system. Therefore, to use AHP for our implementation strategy it was necessary to exclude these patients from the study. Additionally, eligible participants had a primary care visit at one of the 3 study clinics within the last 18 months.

Study Design and Measures

The study design was a pilot hybrid implementation and effectiveness trial which simultaneously tested an implementation strategy and collected data on a clinical intervention.³⁰

We evaluated our implementation strategy by first observing a control period where standard preventive screening outreach practices were used and then an intervention period where we employed our implementation strategy. In the control period, AHP mailed eligible patients AHP’s usual CRC screening outreach materials. The mailing consists of a letter reminding patients they are due for CRC screening and was signed by the patient’s primary care provider. The mailing also included the Centers for Disease Control (CDC) pamphlet, “Colorectal Cancer Screening Saves Lives” (CDC Publication #99-6948). The CDC brochure provides basic information about colorectal cancer and the different screening tests available but does not offer guidance in choosing a test. During the intervention period, patients received the same letter but did not receive the CDC pamphlet. Instead, they received a paper version of the study DA and a DVD containing the video version of the DA. For both the control and intervention periods, after the initial mailings were sent, AHP personnel made a maximum of four attempts to reach each patient by phone and offered to schedule them for screening. The control patients were presented with colonoscopy as the only screening option, while the intervention patients were asked about the material they received and offered a choice between FIT and colonoscopy. We also asked the AHP team to document reasons why patients declined testing in both the control and intervention. We conducted chart reviews 6 months after enrollment to determine if patients completed CRC screening.

Patient Survey

After AHP outreach, we mailed all patients a printed questionnaire consisting of CRC screening knowledge questions.³¹ The intervention group also received the Acceptability of Patient Decision Aids Scale.³² We mailed the questionnaire to patients up to three times. If no response was received by mail, patients were called up three times in an effort the complete the survey by phone.

Implementation Evaluation

In the intervention period, the AHP staff evaluated the implementation of the DA during the outreach call by determining whether patients remembered receiving the DA and asking recipients if they read the DA. Additionally, we observed the implementation efforts by AHP to determine the feasibility of using a centralized delivery process for DA.

Screening and Survey Evaluation

We compared screening rates between the control and intervention groups using chi-square tests for homogeneity, given the data arise from two different populations, and Fisher’s exact test given small cell sizes. Additionally, we compared the difference in mean knowledge scores using a two sample t-test.

Results

Sample Description

The pilot study consisted of 585 patients recruited from the three study clinics (Figure 1). Three hundred twenty-five patients were initially identified for the control and 270 for the intervention. After accounting for exclusions and patients who could not be contacted (see Figure 1), the control group consisted of 178 patients. Two-thirds of the control patients were female and the average age was approximately 58. The intervention group consisted of 148 patients. Fifty-one percent of the patients were female and the average age was 58 (Table 1).

Figure 1.

Study recruitment diagram. The study design was an implementation hybrid pre/post design. Recruitment for the control was from May 2015-November 2015 and the intervention from February 2016-July 2016.

Table 1.

Demographics

		Control	Intervention
Sex		n (%)	(%)
	Male	66 (37)	72 (49)
	Female	111 (63)	76 (51)
	Total	177	148
Age		Mean (Range)	Mean (Range)
	Male	57.48 (50-75)	58.88 (50-74)
	Female	57.93 (50-74)	56.95 (50-73)
	Total	57.76 (50-75)	57. 89 (50-74)

^*p=0.56 comparing mean age between groups

Implementation Outcomes

We evaluated the implementation process through feedback from AHP regarding time spent on distributing the DA to patients. AHP easily integrated distribution of the DA into their regular workflow for CRC screening by exchanging the CDC pamphlet for the DA in the mailings. The study project manager remained in constant communication with AHP throughout the duration of the study. The only implementation issues that arose were questions about patient eligibility, which were easily resolved based on the USPTF guidelines. The AHP team reported that the implementation evaluation questions asked to the intervention group did not negatively influence the time spent conducting or documenting outreach calls.

AHP mailed 260 DAs to patients in the intervention group. It was determined 14 these patients were ineligible for the study and AHP was unable to contact 98 patients (Figure 1). Of the remaining 148 patients, 81 scheduled CRC screening and 67 declined screening. Reasons for refusal included: not being ready for the test, not wanting screening, transportation barriers, not having time for screening, and insurance or financial barriers. We collected implementation data from a total of 89 patients. Among those who scheduled a screening test, 53% (n=47) were unaware of CRC screening other than colonoscopy prior to learning about screening options from the DA. After learning about FIT as a screening option, 48% (n=43) of patients stated a preference for colonoscopy while 39% (n=35) stated a preference for FIT. Twelve percent of patients expressed no preference (n=3).

Screening Outcomes

The overall screening rate for the pilot at 6-month follow-up was 51% (90/176) for the control period and 39% (58/148) for the intervention period (p =0.03). A total of 108 patients in the control period scheduled colonoscopies at the time of AHP outreach. About 75% of these patients completed colonoscopy screening within 6 months of AHP outreach. Sixty-seven patients in the control initially declined colonoscopy screening, but 9 of these people completed screening within 6 months of AHP outreach. Of the 148 patients in the intervention group, who were given a choice between colonoscopy and FIT, 43 scheduled colonoscopy, 38 scheduled FIT and 67 declined to schedule screening. Almost 72% of patients requesting colonoscopy completed screening, 58% of patients requesting FIT completed screening, and 6% of the patients who initially declined screening ultimately completed CRC screening within 6 months of AHP outreach. The difference in completion rate between colonoscopy (72%) and FIT (58%) was statistically significant (p<0.001). Women in the control group were 77% less likely to complete colonoscopy screening than men (OR 0.23, p<0.001). There was no significant difference in screening rates between men and women observed in the intervention group. Age did not influence screening behaviors in either group.

CRC Screening Knowledge Survey Outcomes

The response rates for the mailed survey were 37% and 15%, respectively, for the control and intervention periods. We found a difference in CRC screening knowledge between groups. The intervention group performed better on all questions except one (Table 2). The average knowledge score in the control group was 73% (SD=16.6) versus 87% (SD=15.8) in the intervention (p<0.001). More than 90% of respondents in the intervention group reported that they remembered receiving the DA, and three-quarters of them reported reading some or all of the DA.

Table 2.

Colon Cancer Screening Knowledge Survey Questions

	Control Percent Correct	Interventi on Percent Correct	P-value
Colon cancer deaths decrease with colonoscopy and stool blood testing (True)	79%	85%	0.809
Colon cancer tests are performed before symptoms occur (True)	90%	100%	0.253
Screening tests for colon cancer include colonoscopy and stool blood testing (True)	71%	92%	0.178
Colonoscopies are preformed yearly (False)	92%	96%	0.498
People who have colonoscopies can have complications from the test (True)	74%	73%	0.461
People who have a positive stool blood test need a colonoscopy (True)	85%	92%	0.147
Stool tests are done every 5 years (False)	60%	80%	0.001
Overall Knowledge Score	73%	87%	0.001

We also surveyed patients on the acceptability and usefulness of the DA. Over 80% of respondents found the DA extremely helpful in helping them decide about CRC screening and more than 70% reported they would definitely recommend the DA to someone else making a decision about CRC screening (Figure 2).

Figure 2.

Patient Decision Aid Scale. A total of 26 respondents returned surveys but not answered all questions.

Discussion

In this pilot implementation and effectiveness study, we found using a centralized system to be an effective method for delivering DAs to patients. The integration of DA delivery required little effort beyond the standard work of the AHP team. This finding supports the acceptability of centralized delivery of PtDAs within a large academic healthcare system. The evaluation of the implementation found that 90% of patients remembered receiving the DA and more than 80% found it helpful in supporting their decisions about CRC screening. Surprisingly, we observed a decrease in overall CRC screening rates between the control and intervention groups which contradicts previous studies that found increased screening rates by offering choice in screening methods.⁸

Although we found centralized delivery to be an easy and effective method of DA delivery, mailing DAs to patients is a relatively passive approach to shared decision making. A recent clinical trial found a substantial decrease in adherence to CRC screening when clinic staff was not available to assist patients through the decision making process.³³ One reason for our observed decrease in screening rates may have been a result of the absence of trained staff members to review the DA with patients and facilitate decision making. Further, current systems barriers mean that patients who request FIT screening do not receive follow-up or reminder calls.

In our pilot study, patients who selected colonoscopy received much more contact than patients who select FIT testing. The pharmacy and the clinic staff contacted colonoscopy patients to remind them of their appointment multiple times. No such mechanism exists for FIT testing. Patients are mailed the test and no further contact or reminders are offered. This lack of active follow-up may have influenced the lower overall screening rates in the intervention group. Further, physicians prefer and recommend colonoscopy, which might have influenced patients’ decisions to choose and complete colonoscopy over FIT screening.

Additionally, the naturalistic pre/post study design allowed for secular trends in staffing, clinic resources, and real-world challenges associated with daily clinic processes that we were unable to control for in a pre/post design. We feel the naturalistic approach is an important aspect of the study design as help to highlight the many challenges in implementation. Further it strengthens the argument that a centralized delivery and implementation strategy may overcome implementation barriers at the clinic level. In the future, we would recommend that AHP staff actively follow-up with patients requesting FIT in an effort to activate patients to complete screening.

In addition to finding the DA helpful in CRC screening decision making, we also demonstrated the effectiveness of the DA at improving knowledge through the mailed survey. The intervention group scored better on most knowledge questions than the control group. However, there was a lower rate of screening completion in the intervention group. This finding is consistent with other studies measuring the effect of CRC DAs on screening rates. Both Trevena³⁴ and Dolan¹⁰ found patients who independently reviewed a CRC DA performed better on knowledge and decisional conflict scales but were significantly less likely to complete screening. These findings add credence to the notion that implementation of DAs may require active facilitation to be maximally effective.

Limitations

This study has limitations. First, this study was conducted at a single site, which limits its generalizability. However, the novel implementation approach with centralized delivery is important in the implementation evaluation. Second, due to constraints and barriers posed by the electronic medical record, we were unable to collect covariates such as race, income, and insurance status, which may have provided the opportunity for a more detailed analysis of screening behavior and barriers. Third, we were unable to contact 35% of eligible patients. One reasons for this is a high proportion of this is due to incorrect contact information in the medical record. Lastly, the response rates on the mailed survey were low. This might be due to fatigue from the number of contacts patients had with both AHP outreach and the study team via phone and mail. Fourth, the pre/post study design is a weak design at risk for secular trends. However, for implementation of a new standard, designs such as this are necessary. Despite these limitations, the pilot did yield results that improve our understanding of the implementation of DAs and shared decision making techniques in a real-world implementation trial of a non-selected population.

Conclusions

To our knowledge, this is the first colorectal cancer screening DA that highlights that patients have choices in screening options while specifically emphasizing that physicians generally recommend and prefer colonoscopy. More than half of the participants in the intervention period were unaware of the different CRC screening options prior to reviewing the DA. Even though the DA offered a choice, most respondents stated a preference for colonoscopy. Acceptability data from the pilot shows that patients found the DA helpful in facilitating a choice between screening options and would recommend the DA to peers.

We found using a centralized delivery system to be an effective and efficient method of DA implementation in a large academic health system. Although we failed to increase CRC screening rates, we were able to draw some conclusions about the implementation of DAs for CRC screening. Namely, we demonstrated that it is feasible to implement DAs using a centralized delivery method. DAs are efficacious at increasing knowledge but evidence is mixed on the true effectiveness of CRC DAs in the real world.^{8,10,11,33,35} This is a result of the heterogeneity in screening options, diversity in DA format, and the variability in dissemination and implementation practices. Further research is needed that accounts for the variations in screening options and DA format.

Table 3.

Test Scheduled vs. Test Completed by Study Group

		Test Completed
	Scheduled (n)	Colonoscopy (n)	FIT (n)	Neither (n)
Control
Colonoscopy	108	81	0	27
Declined	67	9	0	58
Intervention
Colonoscopy	43	31	1	11
FIT	38	0	22	16
Declined	67	2	2	63