Implementation of a ColoRectal Cancer Screening Tool in US Primary Care Practices - Usual Quality Improvement (10 Clinics) vs Normalization Process Theory-Participatory Learning in Action (10 Clinics)

Overview

Although implementation intentions (I2)-based tools enhance colorectal cancer (CRC) screening uptake, prior studies have not tested their implementation into routine primary care delivery. In this study, investigators will conduct a cluster-randomized trial in 20 US primary care clinics. Specific aims for the project will be: 1) to test whether a Normalization Process Theory-informed Participatory Learning in Action (NPT-PLA intervention) implementation of a proven implementation Intentions-based colorectal cancer screening tool ("I2") improves screening uptake (i.e. screening order and completion) within 6 months of patient enrollment versus usual quality improvement (control) implementation; and 2) to evaluate the facilitators and barriers of each implementation arm using the 2022 expanded Normalization Process Theory (NPT) framework. Multi-disciplinary clinic 'implementation teams' that include clinic staff and patients whose preferred language is Spanish will meet monthly during the first 6 months of clinic participation and aim to integrate into routine primary care the "I2" CRC screening tool, using the NPT-PLA intervention or control approach. The I2 tool addresses the "when," "where" and "how" details of stool sample or colonoscopy screening. The I2 tool will be delivered via an on-line survey or (if patients prefer) by paper form customized for use in English or Spanish. At least 100 patients in each clinic will be enrolled in the first 6 months of clinic participation (2000 in total). All patients eligible for CRC screening will be offered the I2 tool. Their choices will be communicated automatically to clinics for order entry. Primary (Aim 1) outcomes will be CRC screening orders placed (by clinic staff); completion of the I2 tool and CRC screening completion (by patients) over 6 months of patient follow-up. For Aim 2, surveys based on the NPT domains (the "NOMAD") will be used to assess staff comprehension of their role in implementing the I2-based CRC screening tool, its salience, their buy-in, feasibility of altering workflows, and the potential impact of using the tool in their setting. Investigators will conduct summative qualitative focus group discussions in all participating clinics after 6 months of clinic participation. The study will provide important information on barriers and facilitators of embedding NPT-PLA interventions in "real-world" primary care clinical settings.

The study is a Type 3 Hybrid cluster-randomized trial to be conducted in 20 practices that are members of the DARTNet Institute/American Academy of Family Physicians National Research Network. Investigators will use this approach because the implementation Intervention they are testing is delivered at the clinic level; and because they have already tested the I2- tool that they are implementing and found it is effective. Clinics will be randomly assigned to 1 of the 2 study arms (NPT-PLA or usual QI 'control'), so that neither clinic staff nor DARTNet/AAFP NRNstudy staff know in advance the arm to which a clinic will be assigned. Clinics will each participate for one year. Clinics will enter the study over the first 3 years of the study in a rolling fashion, starting in the 4th quarter of year 1 (2 clinics randomized 1:1 every 3 months thereafter until all have been enrolled), with a maximum of 8 clinics simultaneously participating at any one time. Individual patients needing CRC screening will be enrolled in the first 6 months ("active intervention" period) of a clinic's participation and followed for 6 months after they are provided access to the I2 screening tool (between inception and 6 months during the period of clinic participation), to monitor if they used the tool, and if CRC screening was ordered and completed. During the active intervention period (first 6 months of participation), all clinics in both arms will form clinic-based 'implementation teams' that will meet monthly to plan how they will implement the I2 tool. On entry to the study, each clinic will work with DARTNet staff in both arms of the study to produce a workflow map that describes how they currently conduct CRC screening. Control clinics will be provided a prespecified workflow (the "pilot protocol") that follows the approach that the investigators developed in our 2019-22 pilot study; and asked to integrate the I2 tool into their existing clinic workflow. NPT-PLA clinics will also be given the pilot protocol, but they will also use a survey-based assessments of their clinic's capacity to implement the I2 tool using the NPT based NOMAD tool. Control clinics will also complete this, but will not use the results to guide 'usual QI' processes in their clinic. Thereafter, intervention teams will meet monthly to adapt the pilot protocol using NPT-PLA methods and use their adapted approach to implement the I2 tool in their clinical context. Finally, the investigators will use NPT (2022) informed qualitative interviews at the end of the active intervention in all participating clinics to conduct a summative evaluation of the implementation process. Study arms and interventions are further described in that section of this Record Twenty clinics will be randomly assigned 1:1 to treatment group (NPT-PLA -10 clinics) or 'usual QI' (10 clinics) at the time of the clinic's study entry (which will take place during the first 4 years of the study), based on random numbers generated by study biostatisticians and delivered to DARNet/AAFP NRN study staff responsible for recruitment and enrollment. We will use stratified randomization,with 3 strata: a) federally qualified health clinics or other free clinics, b) academic medical centers or other large health systems and c) other physician owned practices. Study biostatisticians who analyze the study data will not be told which clinics are in which treatment group. Each clinic will participate for a total of 12 months. Clinics will be recruited via email and a one-page description of the project. Clinic 'implementation team' members Clinic staff and patients in both treatment groups will participate in clinic-based "implementation teams" ("iteams"). Investigators will recruit clinic staff members (front desk staff, nursing assistants, LPNs, RNs, physicians and representative patients) to participate in each clinic's 'implementation team'. Patients will be identified by clinic managers, and invited via email/patient portal to participate in the clinic's implementation team. NOTE: since patients who participate in implementation teams will differ from other patients of the clinic in terms of their awareness of and motivation to complete I2, investigators will not include their CRC screening related outcomes in our analytic data set; however, they will be demographically similar to patients of the clinic who are eligible for CRC screening and will have access to the I2 tool for their own use. . All staff and patients participating in these implementation teams will provide written informed consent. Patients whose CRC outcomes are monitored (within assigned clinic) To address Aim 1 outcomes, CRC screening outcomes will be monitored for clinic patients eligible for CRC screening. Please see the Eligibility section for inclusion/exclusion criteria for these patients. Each clinic will create a list of eligible patients that is created from that clinic's EMR. This list will indicate which patients of the clinic are eligible for screening and who will age into or out of the eligible age range during the first 6 months of clinic participation, so that those who have not completed I2 or have ordered/completed CRC screens can be presented with the I2 tool, appropriately. Those patients not excluded will be considered "enrolled". Each clinic will enroll at least 100 patients; however, investigators will aim to enroll all who are eligible to be screened. Each patient will be followed for 6 months after their initial invitation to complete I2. Delivery of Intervention Please see "Arms and Interventions" Section for a description of the study arms and associated intervention. After assignment to 1 of the 2 study arms and collection of baseline survey data, clinics in each arm will start to meet to discuss how to implement the I2 tool. Prior to the initial meeting, Team Leads at each site in both arms (the Practice Manager, Head Nurse or Lead Physician) will complete a checklist to identify: how CRC screening eligible patients are identified and when; how and when screening options are presented and decided upon; if and how patient navigation occurs after screening recommendations are made for each type of screening test; how barriers and/or facilitators at different stages in clinical workflows alter clinician orders and/or patient completion of CRC screening; and what QI methods the clinic uses to monitor and optimize CRC screening, including for patients whose preferred language is Spanish. DARTNet Study staff will use this information to create a 'workflow map' describing how the clinic carries out CRC screening for it's patients. Each team will start their discussions based on a prescribed protocol to implement I2 in their clinic workflow. NPT-PLA (active intervention) clinics Team Leads will be trained to facilitate PLA meetings, and will adapt their implementation of the I2 tool following those methods. Usual QI (control) clinic Team Leads will use their usual 'quality improvement' methods to follow the 'pilot protocol' as closely as possible. . Teams in both arms of the trial will meet monthly for 6 months, following their respective approaches (NPT-PLA or usual QI) to increase I2 tool use by patients and CRC screening overall. In the 6th month, teams in both arms will take the NOMAD survey, which is a validated survey including 23 Likert-style questions (range 0-10 indicating agreement to the statement) that correspond to NPT core constructs. In the 7th month, all clinics will review results of the TIDieR survey concerning intervention fidelity (this will be completed by one of the Team Leads in advance of the session) and participate in a focus group discussion, with questions guided by the 2022 NPT framework to discuss the clinic's implementation of I2 using their assigned approach. Analysis Plan CRC screening outcomes will be monitored for all enrolled patients, and will be extracted from data routinely collected in each participating clinic's electronic medical record, addended by I2 tool outcome data if that form has also been completed. Cleaning and management routines (e.g., logical checks for continuous variables, compliance with skip patterns, missing data codes) will be conducted using SAS v9.4.132, paying attention to the range and distribution of the data. Investigators will examine that distribution for all baseline quantitative measures, comparing the clinics assigned to NPT-PLA vs 'usual QI' clinics. The unit of analysis will be individual patients within clinics. Investigators will exclude screening outcomes for patients who participate in the 'implementation teams', since participation on the i-team may bias the patient towards completion of I2 and CRC screening, differently than other clinic patients. Investigator will calculate unadjusted risk ratios and use Chi-squared test (likelihood ratio method) to compare the main outcomes for patients in each treatment arm. They will use means or median tests of equality to assess inequality in age, gender, employment, insurance coverage between NPT-PLA vs usual QI clinics, to determine the effectiveness of randomization procedures to balance these patient-level factors between treatment arms. Demographic data e.g. age and sex will also be included in data extractions from participating clinic EMRs to include in adjusted analytic models of the main outcome responses if, at baseline, there are any differences between treatment group demographics. A priori, investigators will include patient's preferred language. Other variables will be selected for inclusion in the models (if they vary between groups at baseline) by backward elimination procedure using the Bayesian information criterion (BIC). Potential multicollinearity will be examined by variance inflation factor (VIF). Investigators will use linear mixed models using a logistic model (primary screening outcome completed vs not), including a random effects term to account for patients' clinic of origin. Another sensitivity analysis with interactions between included variables and treatment group (NPT-PLA vs Usual QI control) will be examined and used to stratify results if significant. Investigators will report the results of this test for preferred language. They will repeat this analysis for the secondary aim outcome (completion of the I2 tool), as described above; however, for this outcome they will use an ordinal logistic model. For aim 2 outcomes, investigators will use both quantitative and qualitative data. They will follow a similar approach to that described for Aim 1 quantitative analyses. They will analyze pre-post changes in NOMAD construct scores, and compare 6-month NOMAD scores between treatment groups. They will use linear mixed models, analogously to the approach used for Aim 1 outcomes, but (because NOMAD construct scores are quantitative) they will use a Gaussian model. If these scores are not normally or homogenously distributed between groups, they will use appropriate data transformations so as to enable use of use parametric modeling. DARTNet staff responsible for qualitative coding will review monthly summaries of clinic "checkin calls" (for which investigators will collect field notes) and graphical material from NPT-PLA clinic PLA sessions. Research staff (PI LeMaster and DARTNet staff) are experienced in qualitative analysis, and will examine the qualitative data generated during the study (focus group transcripts, meeting and field notes), as well as PLA and other clinic team meeting notes to identify emerging qualitative themes and how they vary across clinical contexts. Investigators will conduct a deductive framework analysis, structured according to the NPT (2022) construct and using a project specific set of validated coding rules. They will use the newly available empirical coding manual for NPT that includes 12 primary and 16 secondary constructs which are organized into a "Context-Mechanism-Outcome (CMO) structure. These identify, characterize, and explain the course of implementation processes through which strategic intentions are translated into behaviors, and how enactment of those behaviors lead to different outcomes and maintenance of CRC screening practices. This analysis will provide insight into differences in implementation pathways experienced in each treatment group (NPT-PLA vs control). Thus, investigators are using an extension of the original NPT model to evaluate the implementation process in each clinic. The CMO constructs and their cross-walk to NPT constructs are as follows: * Contexts are events in systems unfolding over time within and between settings in which implementation work is done (primary NPT constructs: strategic intentions, adaptive execution, negotiating capability, reframing organizational logic). * Mechanisms motivate and shape the work that people do when they participate in implementation processes (primary NPT constructs: coherence-building, cognitive participation, collective action, reflexive monitoring). * Outcomes are the effects of implementation work in context-that make visible hoq things change as implementation processes proceed (primary NPT constructs: intervention performance, normative restructuring, relational restructuring, sustainment). Investigators will identify which of these constructs was dominant across clinics in each arm and during clinic participation; how these were expressed in each and across clinic contexts; and (prior to report completion/results dissemination) discuss these themes with participating implementation team members, to check that their perspectives are taken into account in the analysis. During the analysis of all qualitative data, investigators will pay active attention to any data that fall outside the NPT constructs to appraise its relevance for understanding the implementation processes. DARTNet staff not involved in the design of the study will undertake the initial coding and refer unclear coding issues to Dr. LeMaster (who is highly expert in qualitative analysis). After initial coding is completed, these staff will meet with Dr. LeMaster and the study AnalysisTeam monthly during year 5 of the study to discuss qualitative themes emerging from these data. Sample size: Investigators base their primary power analysis on the Aim 1 outcome "CRC screening completed", since this is the final endpoint of the implementation of the I2 tool; and "CRC orders completed" must precede "CRC screening completed". Investigators are primarily interested in increases in screening uptake in each study arm. Assuming that the effect size of the intervention is at least that of our 2021-22 pilot study comparing NPT-PLA to usual QI in 4 clinics (2 participated in each arm) in which the proportion overall who completed CRC screening in the NPT-PLA arm=50%, and 37% in usual QI control clinics, Mantel-Haenzsel odds ratio adjusting for language = 1.72 (95% C.I. 1.08 - 2.75), two-sided p=0.02), if investigators recruit 10 NPT-PLA and 10 usual QI control clinics in a cluster randomized trial, and if clinics will each enroll at least 100 patients within 6 months of starting participation, with intra-class correlation coefficient=0.02, the study will have at least 82% power, when using a two-sided test at the 0.05 significance level, to detect a significant difference between arms in the proportion of patients for whom CRC screening is completed .