A streamlined approach to classifying and tailoring implementation strategies: recommendations to speed the translation of research to practice.

BACKGROUND: Implementation science emerged from the recognized need to speed the translation of effective interventions into practice. In the US, the science has evolved to place an ever-increasing focus on implementation strategies. The long list of implementation strategies, terminology used to name strategies, and time required to tailor strategies all may contribute to delays in translating evidence-based interventions (EBIs) into practice. To speed EBI translation, we propose a streamlined approach to classifying and tailoring implementation strategies. MAIN TEXT: A multidisciplinary team of eight scholars conducted an exercise to sort the Expert Recommendations for Implementing Change (ERIC) strategies into three classes: implementation processes (n = 25), capacity-building strategies (n = 20), and integration strategies (n = 28). Implementation processes comprise best practices that apply across EBIs and throughout the phases of implementation from exploration through sustainment (e.g., conduct local needs assessment). Capacity-building strategies target either general or EBI-specific knowledge and skills (e.g., conduct educational meetings). Integration strategies include "methods and techniques" that target barriers or facilitators to implementation of a specific EBI beyond those targeted by capacity building. Building on these three classes, the team collaboratively developed recommendations for a pragmatic, five-step approach that begins with the implementation processes and capacity-building strategies practice-settings are already using prior to tailoring integration strategies. A case study is provided to illustrate use of the five-step approach to tailor the strategies needed to implement a transitional care intervention in skilled nursing facilities. CONCLUSIONS: Our proposed approach streamlines the formative work required prior to implementing an EBI by building on practice partner preferences, expertise, and infrastructure while also making the most of prior research findings.

Understanding mechanisms of racial disparities in breast cancer: an assessment of screening and regular care in the Carolina Breast Cancer Study.

PURPOSE: Screening history influences stage at detection, but regular preventive care may also influence breast tumor diagnostic characteristics. Few studies have evaluated healthcare utilization (both screening and primary care) in racially diverse screening-eligible populations. METHODS: This analysis included 2,058 women age 45-74 (49% Black) from the Carolina Breast Cancer Study, a population-based cohort of women diagnosed with invasive breast cancer between 2008 and 2013. Screening history (threshold 0.5 mammograms per year) and pre-diagnostic healthcare utilization (i.e. regular care, based on responses to "During the past ten years, who did you usually see when you were sick or needed advice about your health?") were assessed as binary exposures. The relationship between healthcare utilization and tumor characteristics were evaluated overall and race-stratified. RESULTS: Among those lacking screening, Black participants had larger tumors (5 + cm) (frequency 19.6% vs 11.5%, relative frequency difference (RFD) = 8.1%, 95% CI 2.8-13.5), but race differences were attenuated among screening-adherent participants (10.2% vs 7.0%, RFD = 3.2%, 0.2-6.2). Similar trends were observed for tumor stage and mode of detection (mammogram vs lump). Among all participants, those lacking both screening and regular care had larger tumors (21% vs 8%, RR = 2.51, 1.76-3.56) and advanced (3B +) stage (19% vs 6%, RR = 3.15, 2.15-4.63) compared to the referent category (screening-adherent and regular care). Under-use of regular care and screening was more prevalent in socioeconomically disadvantaged areas of North Carolina. CONCLUSIONS: Access to regular care is an important safeguard for earlier detection. Our data suggest that health equity interventions should prioritize both primary care and screening.

Unpacking the relationship between shared decision-making and decisional quality, decision to screen, and screening completion in lung cancer screening.

OBJECTIVES: Lung cancer screening is a complex and individualized decision. To understand how best to support patients in this decision, we must understand how shared decision-making is associated with both decisional and behavioral outcomes. METHODS: Observational cohort study combining patient survey data with electronic health record data of lung screening-eligible patients who recently engaged in a shared decision-making discussion about screening with a primary care clinician. RESULTS: Using multivariable analysis (n = 529), factors associated with higher lung cancer screening decisional quality include higher knowledge (OR = 1.33, p < .0001), lower perceived benefits (OR = 0.90, p = .0004), higher perceived barriers (OR = 1.07, p < .0001), higher self-efficacy (OR = 1.13, p < .0001), and higher levels of perceiving the discussion was shared (OR = 1.04, p < .0001). Factors associated with the patient's decision to screen include older age (OR = 1.12, p = .0050) and higher self-efficacy (OR = 1.11, p = .0407). Factors associated with screening completion included older age (OR = 1.05, p = .0050), higher knowledge (OR = 1.24, p = .0045), and higher self-efficacy (OR = 1.12, p = .0003). CONCLUSIONS: Shared decision-making in lung cancer screening is a dyadic process between patient and clinician. As we continue to strive for high-quality patient-centered care, patient decision quality may be enhanced by targeting key factors such as high-quality knowledge, self-efficacy, and fostering a shared discussion to support patient engagement in lung cancer screening decisions.

Uptake of Lung Cancer Screening CT After a Provider Order for Screening in the PROSPR-Lung Consortium.

BACKGROUND: Uptake of lung cancer screening (LCS) has been slow with less than 20% of eligible people who currently or formerly smoked reported to have undergone a screening CT. OBJECTIVE: To determine individual-, health system-, and neighborhood-level factors associated with LCS uptake after a provider order for screening. DESIGN AND SUBJECTS: We conducted an observational cohort study of screening-eligible patients within the Population-based Research to Optimize the Screening Process (PROSPR)-Lung Consortium who received a radiology referral/order for a baseline low-dose screening CT (LDCT) from a healthcare provider between January 1, 2015, and June 30, 2019. MAIN MEASURES: The primary outcome is screening uptake, defined as LCS-LDCT completion within 90 days of the screening order date. KEY RESULTS: During the study period, 18,294 patients received their first order for LCS-LDCT. Orders more than doubled from the beginning to the end of the study period. Overall, 60% of patients completed screening after receiving their first LCS-LDCT order. Across health systems, uptake varied from 41 to 87%. In both univariate and multivariable analyses, older age, male sex, former smoking status, COPD, and receiving care in a centralized LCS program were positively associated with completing LCS-LDCT. Unknown insurance status, other or unknown race, and lower neighborhood socioeconomic status, as measured by the Yost Index, were negatively associated with screening uptake. CONCLUSIONS: Overall, 40% of patients referred for LCS did not complete a LDCT within 90 days, highlighting a substantial gap in the lung screening care pathway, particularly in decentralized screening programs.

Self-Collection for Primary HPV Testing: Perspectives on Implementation From Federally Qualified Health Centers.

INTRODUCTION: Primary testing for high-risk human papillomavirus (HPV) by self-collection could result in higher rates of cervical cancer screening. Federally qualified health centers (FQHCs) in the US serve a large proportion of women who have low income and no health insurance and are medically underserved - risk factors for being insufficiently screened for cervical cancer. Although the implementation of self-collection for HPV testing is not yet widespread, health care entities need to prepare for its eventual approval by the US Food and Drug Administration. We conducted focus groups and interviews among clinical and administrative staff and leadership to gather data on key logistical concerns that must be addressed before implementing self-collection for HPV testing in FQHCs. METHODS: We identified focus group and interview participants from 6 FQHCs in North Carolina. We conducted focus groups with clinical and administrative staff (N = 45) and semistructured interviews with chief executive officers, senior-level administrators, chief medical officers, and clinical data managers (N = 24). Transcripts were coded by using codebooks derived from research questions and notes taken during data collection. Themes emerged on implementation of self-collection for HPV testing. We applied the constructs from the Consolidated Framework for Implementation Research (CFIR) to themes to identify domains of potential barriers and facilitators to implementation. RESULTS: Clinical personnel reported that offering self-collection for HPV testing is acceptable and feasible and can increase cervical cancer screening rates. Uncertainties emerged about accuracy of results, workflow disruptions, financial implications, and effects on clinic quality measures. CONCLUSION: Implementing self-collection for HPV testing was considered feasible and acceptable by participants. However, important health service delivery considerations, including financial implications, must be addressed before integrating self-collection for HPV testing into the standard of care.

Associations With Virtual Visit Use Among Patients Receiving Radiation Therapy.

Purpose: The objective of this study was to test for patient characteristics associated with virtual versus office visits among radiation oncology patients. Methods and Materials: Using the electronic health record, we extracted encounter data and corresponding patient information for the 6 months before and 6 months of COVID-19-enabled virtual visits (October 1, 2019, to March 22, 2020 vs March 23, 2020, to September 1, 2020) at a National Cancer Institute-Designated Cancer Center. Encounters during COVID-19 were categorized as in-person or virtual visits. We compared patient demographic variables including race, age, sex, marital status, preferred language, insurance status, and tumor type during the pre-COVID-19 period as a baseline versus during the COVID-19 period. Multivariable analyses examined associations between these variables and virtual visit use. Results: We analyzed 4974 total encounters (2287 before COVID-19 and 2687 during COVID-19) for 3960 unique patients. All (100%) pre-COVID-19 encounters were in-person. During COVID-19, 21% of encounters were via virtual visits. There were no differences identified in pre- versus during-COVID-19 patient characteristics. However, we found significant differences in patient characteristics for in-person versus virtual encounters during COVID-19. On multivariable analysis, virtual visit use was less common among patients who were Black versus White (odds ratio [OR], 0.75; 95% CI, 0.57-0.99; P = .044) and not married versus married (OR, 0.76; 95% CI, 0.59-0.98; P = .037). Patients with head and neck (OR, 0.63; 95% CI, 0.41-0.97; P = .034), breast (OR, 0.36; 95% CI, 0.21-0.62; P ≤ .001), gastrointestinal/abdominal (OR, 0.31; 95% CI, 0.15-0.63; P = .001), or hematologic malignancy (OR, 0.20; 95% CI, 0.04-0.95; P = .043) diagnoses were less likely to be scheduled for virtual visits relative to patients with genitourinary malignancy. No Spanish-speaking patients engaged in a virtual visit. We did not identify differences in the insurance status or sex of patients scheduled for virtual visits. Conclusions: We found significant differences in virtual visit use by patient sociodemographic and clinical characteristics. Further investigation into implications of differential virtual visit use including social and structural determinants and subsequent clinical outcomes is indicated.

Test performance metrics for breast, cervical, colon, and lung cancer screening: a systematic review.

BACKGROUND: Multiple quality metrics have been recommended to ensure consistent, high-quality execution of screening tests for breast, cervical, colorectal, and lung cancers. However, minimal data exist evaluating the evidence base supporting these recommendations and the consistency of definitions and concepts included within and between cancer types. METHODS: We performed a systematic review for each cancer type using MEDLINE, Embase, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) from 2010 to April 2020 to identify guidelines from screening programs or professional organizations containing quality metrics for tests used in breast, cervical, colorectal, and lung cancer screening. We abstracted metrics' definitions, target performance levels, and related supporting evidence for test completeness, adequacy (sufficient visualization or collection), accuracy, and safety. RESULTS: We identified 11 relevant guidelines with 20 suggested quality metrics for breast cancer, 5 guidelines with 9 metrics for cervical cancer, 13 guidelines with 18 metrics for colorectal cancer (CRC), and 3 guidelines with 7 metrics for lung cancer. These included 54 metrics related to adequacy (n = 6), test completeness (n = 3), accuracy (n = 33), and safety (n = 12). Target performance levels were defined for 30 metrics (56%). Ten (19%) were supported by evidence, all from breast and CRC, with no evidence cited to support metrics from cervical and lung cancer screening. CONCLUSIONS: Considerably more guideline-recommended test performance metrics exist for breast and CRC screening than cervical or lung cancer. The domains covered are inconsistent among cancers, and few targets are supported by evidence. Clearer evidence-based domains and targets are needed for test performance metrics. REGISTRATION: PROSPERO 2020 CRD42020179139.

Predictive Modeling for Adverse Events and Risk Stratification Programs for People Receiving Cancer Treatment.

Emergency department visits and hospitalizations are common among people receiving cancer treatment, accounting for a large proportion of spending in oncology care and negatively affecting quality of life. As oncology care shifts toward value- and quality-based payment models, there is a need to develop interventions that can prevent these costly and low-value events among people receiving cancer treatment. Risk stratification programs have the potential to address this need and optimally would consist of three components: (1) a risk stratification algorithm that accurately identifies patients with modifiable risk(s), (2) intervention(s) that successfully reduce this risk, and (3) the ability to implement the risk algorithm and intervention(s) in an adaptable and sustainable way. Predictive modeling is a common method of risk stratification, and although a number of predictive models have been developed for use in oncology care, they have rarely been tested alongside corresponding interventions or developed with implementation in clinical practice as an explicit consideration. In this article, we review the available published predictive models for treatment-related toxicity or acute care events among people receiving cancer treatment and highlight challenges faced when attempting to use these models in practice. To move the field of risk-stratified oncology care forward, we argue that it is critical to evaluate predictive models alongside targeted interventions that address modifiable risks and to demonstrate that these two key components can be implemented within clinical practice to avoid unplanned acute care events among people receiving cancer treatment.

Automatic speech recognition performance for digital scribes: a performance comparison between general-purpose and specialized models tuned for patient-clinician conversations.

One promising solution to address physician data entry needs is through the development of so-called "digital scribes," or tools which aim to automate clinical documentation via automatic speech recognition (ASR) of patient-clinician conversations. Evaluation of specialized ASR models in this domain, useful for understanding feasibility and development opportunities, has been difficult because most models have been under development. Following the commercial release of such models, we report an independent evaluation of four models, two general-purpose, and two for medical conversation with a corpus of 36 primary care conversations. We identify word error rates (WER) of 8.8%-10.5% and word-level diarization error rates (WDER) ranging from 1.8%-13.9%, which are generally lower than previous reports. The findings indicate that, while there is room for improvement, the performance of these specialized models, at least under ideal recording conditions, may be amenable to the development of downstream applications which rely on ASR of patient-clinician conversations.

Provider response and follow-up to parental declination of HPV vaccination.

OBJECTIVE: Parents often decline HPV vaccination, but little is known about how healthcare providers should promote vaccination at a later visit for secondary acceptance. We examined the associations of two factors, providers' response to declination during the visit and follow-up after the visit, with secondary acceptance. METHODS: We conducted a cross-sectional survey of US parents whose 9- to 17-year-old child had not yet completed the HPV vaccination series. Parents who declined HPV vaccination during an initial discussion with a provider (n = 447) reported whether their provider engaged in any active response during the visit (e.g., giving information, trying to change their mind) or any follow-up after the visit (e.g., scheduling another visit). We conducted multivariable logistic regression to determine whether an active response or follow-up was associated with secondary acceptance of HPV vaccination. RESULTS: Only about one-third of parents reported an active response during the visit (35%) or follow-up after the visit (39%) following HPV vaccination declination. Parents had higher odds of secondary acceptance of HPV vaccine if they received any provider follow-up after the visit (43% vs. 20%, aOR:3.19; 95% CI:2.00:5.07). Receipt of an active provider response was not associated with secondary acceptance. More parents thought a provider should actively respond and follow-up (61% and 68% respectively), compared with those who received such a response (both p < .01). CONCLUSIONS: Providers' follow-up after the visit may be important for promoting secondary acceptance of HPV vaccination. Parents who decline HPV vaccination often prefer to receive an active response or follow-up from a provider.

Cancer screening in the U.S. through the COVID-19 pandemic, recovery, and beyond.

COVID-19 has proved enormously disruptive to the provision of cancer screening, which does not just represent an initial test but an entire process, including risk detection, diagnostic follow-up, and treatment. Successful delivery of services at all points in the process has been negatively affected by the pandemic. There is a void in empirical high-quality evidence to support a specific strategy for administering cancer screening during a pandemic and its resolution phase, but several pragmatic considerations can help guide prioritization efforts. Targeting guideline-eligible people who have never been screened, or those who are significantly out of date with screening, has the potential to maximize benefits now and into the future. Disruptions to care due to the pandemic could represent an unparalleled opportunity to reassess early detection programs towards an explicit, thoughtful, and just prioritization of populations historically experiencing cancer disparities. By focusing screening services on populations that have the most to gain, and by careful and deliberate planning for the period following the pandemic, we can positively affect cancer outcomes for all.

Primary Care Physician Perspectives on Recommending E-cigarettes to Smokers: a Best-Worst Discrete Choice Experiment.

BACKGROUND: Recent clinical trials suggest that e-cigarettes may be more effective for smoking cessation than traditional cessation aids, yet primary care physician (PCP) practices regarding e-cigarette recommendations for smokers have not been studied in-depth. OBJECTIVE: To identify factors influencing PCP recommendation of e-cigarettes for smoking cessation. DESIGN: Discrete choice experiment and survey. PARTICIPANTS: Florida PCPs. MEASURES: The survey included a discrete choice experiment in which PCPs indicated whether they would recommend e-cigarettes for each of 8 hypothetical patient profiles with the following contrasting characteristics: e-cigarette use, interest in approved cessation methods, smoking intensity, prior experience with approved cessation medications, quit intention, age, and comorbidity. Responses were summarized using descriptive statistics and standardized scores (SS). KEY RESULTS: The sample (n = 216) was predominately male (76%), white (66%), and non-Hispanic (78%), and most respondents had held their medical degree for 20+ years (77%). The response rate was 28.7%. Most PCPs thought e-cigarettes were at least somewhat effective for smoking cessation (66%) and lowering disease risk (65%); 31% perceived e-cigarettes to be equally/more effective than traditional cessation aids. PCPs were split regarding whether e-cigarettes were less (50%) or equally harmful (38%) as cigarettes. Yet, few were very confident in their ability to counsel patients on e-cigarettes risks (27%) or benefits (15%). PCPs recommended e-cigarettes in 27% of patient profiles they evaluated. The most important factors influencing the decision to recommend or not recommend e-cigarette were patients' prior use of nicotine replacement therapy with (SS = 0.22, 95% CI = 0.17-0.27) and without use of other medications for cessation (SS = 0.18, 95% CI = 0.13-0.23), and being middle age (50 years old) with chronic obstructive pulmonary disease (SS = 0.16, 95% CI = 0.10-0.23). CONCLUSIONS: Considering the increased patient use of e-cigarettes and increasing use for cessation, this study highlights the need for guidelines and education to aid PCPs' counseling of patients about e-cigarette use.

Automated rating of patient and physician emotion in primary care visits.

OBJECTIVE: Train machine learning models that automatically predict emotional valence of patient and physician in primary care visits. METHODS: Using transcripts from 353 primary care office visits with 350 patients and 84 physicians (Cook, 2002 [1], Tai-Seale et al., 2015 [2]), we developed two machine learning models (a recurrent neural network with a hierarchical structure and a logistic regression classifier) to recognize the emotional valence (positive, negative, neutral) (Posner et al., 2005 [3]) of each utterance. We examined the agreement of human-generated ratings of emotional valence with machine learning model ratings of emotion. RESULTS: The agreement of emotion ratings from the recurrent neural network model with human ratings was comparable to that of human-human inter-rater agreement. The weighted-average of the correlation coefficients for the recurrent neural network model with human raters was 0.60, and the human rater agreement was also 0.60. CONCLUSIONS: The recurrent neural network model predicted the emotional valence of patients and physicians in primary care visits with similar reliability as human raters. PRACTICE IMPLICATIONS: As the first machine learning-based evaluation of emotion recognition in primary care visit conversations, our work provides valuable baselines for future applications that might help monitor patient emotional signals, supporting physicians in empathic communication, or examining the role of emotion in patient-centered care.

The Effects of Personalizing Colorectal Cancer Risk Communication on Risk Perceptions and Health Behavior Intentions: a Randomized Trial of Average-Risk Adults.

Risk assessment tools may help individuals gauge cancer risk and motivate lifestyle and screening behavior changes. Despite the evermore common availability of such tools, little is known about their potential utility in average-risk population approaches to cancer prevention. We evaluated the effects of providing personalized (vs. generic) information concerning colorectal cancer (CRC) risk factors on average-risk individuals' risk perceptions and intentions to engage in three risk-reducing behaviors: CRC screening, diet, and physical activity. Further, we explored whether the receipt of CRC-specific risk assessment feedback influenced individuals' breast cancer risk perceptions and mammography intentions. Using an online survey, N = 419 survey respondents aged 50-75 with no personal or family history of CRC were randomized to receive an average estimate of CRC lifetime risk and risk factor information that was either personalized (treatment) or invariant/non-personalized (control). Respondent risk perceptions and behavioral intentions were ascertained before and after risk assessment administration. No differences were observed in risk perceptions or behavioral intentions by study arm. However, regardless of study arm, CRC screening intentions significantly increased after risk assessment feedback was provided. This occurred despite a significant reduction in risk perceptions. Results support the role simple cancer risk assessment information could play in promoting screening behaviors while improving the accuracy of cancer risk perceptions. Providing cancer risk assessment information may decrease individuals' perceptions of cancer risk to more realistic levels while simultaneously facilitating screening intentions among an average-risk population, regardless of whether provided risk information is personalized.

Improving secure messaging: A framework for support, partnership & information-giving communicating electronically (SPICE).

BACKGROUND: Patient-centered communication benefits patients and is widely endorsed. However, it is primarily associated with face-to-face contexts, although patients are increasingly using electronic platforms, such as secure messaging in patient portals, to communicate with providers. PURPOSE: Given the popularity of secure messaging and its ability to impact the patient-provider relationship, this study aimed to determine which attributes of patient-centered communication are most desired by cancer patients using secure messaging. METHODS: A 26 balanced incomplete block design discrete choice experiment was conducted using the best-worst scaling technique. Respondents were asked to select their most and least preferred attributes of two simulated patient-provider exchanges within each of eight choice sets. RESULTS: 210 respondents indicated that either level of partnership (high and low) and either level of information-giving (high and low) were most preferred, while response times greater than 24 hours and low levels of support were least favored. CONCLUSIONS: Similar to face-to-face communication, patients value aspects of patient-centered communication in the secure messaging setting and desire them to be included in provider replies. PRACTICE IMPLICATIONS: Patient-centered communication is important to patients using secure messaging. Providers should incorporate SPICE (Support, Partnership, and Information-giving while Communicating Electronically).

Patient-Oncologist Communication Regarding Oral Chemotherapy During Routine Office Visits.

PURPOSE: Although studies in other clinical areas have shown that patient-clinician communication can positively influence adherence to medications, little is known about how oncologists address medication counseling during routine office visits. We describe patient-oncologist office-based discussions of oral chemotherapy treatment. METHODS: Transcripts of 24 patient-oncologist office visits were obtained from a national database. Patients were aged ≥ 19 years and prescribed capecitabine for colorectal cancer. We developed a structured coding worksheet using medication-counseling concepts previously identified as important to medication adherence and a grounded approach. Two coders reviewed transcripts for oncologists' provision of medication information, assessment of patients' adherence to medication, and the provision of self-management support for management of adverse effects. We assessed interrater reliability with Cohen κ statistics. We describe the counseling concepts present within patient-oncologist conversations and present illustrative quotes to describe how they were discussed. RESULTS: Oncologists generally provided patients who had yet to initiate therapy comprehensive medication information; those in the midst of treatment received less information. Oncologists discussed patients' continued use of the medication (or discontinuation) among all patients who had initiated therapy (N = 18). How the patient was taking the medication (ie, therapy implementation) was less commonly discussed. Medication adverse effects were also discussed in all encounters. Self-management strategies were commonly provided, albeit mostly in response to a presenting symptom and not preemptively. Patients' use of concurrent medications, financial access to therapy, and assessments of logistical arrangements were discussed more sporadically. CONCLUSION: Using audio recordings from a national sample of patient-oncologist office visits, we identified several potentially important opportunities to enhance medication counseling among patients prescribed capecitabine for the treatment of colorectal cancer.

The Association Between Medicare Low-Income Subsidy and Anticancer Treatment Uptake in Advanced Lung Cancer.

BACKGROUND: High out-of-pocket costs may impact anticancer treatment uptake. The Low-Income Subsidy (LIS) program can reduce patient out-of-pocket cost for Medicare Part D-covered treatments. We examined whether the LIS increased uptake and reduced time to initiate orally administered anticancer drugs in patients with advanced non-small cell lung cancer (NSCLC). METHODS: Using Surveillance, Epidemiology and End Results (SEER)-Medicare data, we identified older adults (aged 65 years and older) diagnosed with advanced NSCLC from 2007 through 2013 and categorized them as full LIS, partial LIS, or non-LIS. We used propensity-score weighted (IPTW) Cox proportional hazards regression to assess the likelihood of and time to initiate Part D treatments. Part B medication uptake was our negative control because supplemental insurance reduces out-of-pocket costs for those drugs. All statistical tests were two-sided. RESULTS: Among 19 746 advanced NSCLC patients, approximately 10% initiated Part D treatments. Patients with partial or no LIS were less likely to initiate Part D treatments than were those with full subsidies (partial LIS vs full LIS HRIPTW = 0.77, 95% confidence interval = 0.62 to 0.97; non-LIS vs full LIS HRIPTW = 0.87, 95% confidence interval = 0.79 to 0.95). Time to initiate Part D treatments was also slightly shorter among full-LIS patients (full LIS mean [SD] = 10.8 [0.04] months; partial LIS mean [SD] = 11.3 [0.08] months; and non-LIS mean [SD] = 11.1 [0.03] months, P < .001). Conversely, patients with partial or no LIS had shorter time to initiation of Part B drugs. CONCLUSIONS: Patients receiving the full LIS had higher orally administered anticancer treatment uptake than patients without LIS. Notably, patients with partial LIS had the lowest treatment uptake, likely because of their low incomes combined with high expected out-of-pocket spending. High out-of-pocket costs for Part D medications may be a barrier to treatment use for patients without full LIS.

Challenges and opportunities using online portals to recruit diverse patients to behavioral trials.

We describe the use of an online patient portal to recruit and enroll primary care patients in a randomized trial testing the effectiveness of a colorectal cancer (CRC) screening decision support program. We use multiple logistic regression to identify patient characteristics associated with trial recruitment, enrollment, and engagement. We found that compared to Whites, Blacks had lower odds of viewing the portal message (OR = 0.46, 95% CI = 0.37-0.57), opening the attached link containing the study material (OR = 0.75, 95% CI = 0.62-0.92), and consenting to participate in the trial (OR = 0.85, 95% CI = 0.67-0.93). We also found that compared to Whites, Asians had lower odds of viewing the portal message (OR = 0.53, 95% CI = 0.33-0.64), opening the attached link containing the study material (OR = 0.76, 95% CI = 0.54-0.97), consenting to participate in the trial (OR = 0.68, 95% CI = 0.53-0.95), and completing the trial's baseline questionnaire (OR = 0.59, 95% CI = 0.36-0.90). While portals offer an opportunity to mitigate human bias in trial invitations, because of racial disparities-not only in who has a portal account, but in how they interact with trial recruitment and enrollment material within the portal-using portals alone for trial recruitment may generate study samples that are not racially diverse.

Detecting conversation topics in primary care office visits from transcripts of patient-provider interactions.

OBJECTIVE: Amid electronic health records, laboratory tests, and other technology, office-based patient and provider communication is still the heart of primary medical care. Patients typically present multiple complaints, requiring physicians to decide how to balance competing demands. How this time is allocated has implications for patient satisfaction, payments, and quality of care. We investigate the effectiveness of machine learning methods for automated annotation of medical topics in patient-provider dialog transcripts. MATERIALS AND METHODS: We used dialog transcripts from 279 primary care visits to predict talk-turn topic labels. Different machine learning models were trained to operate on single or multiple local talk-turns (logistic classifiers, support vector machines, gated recurrent units) as well as sequential models that integrate information across talk-turn sequences (conditional random fields, hidden Markov models, and hierarchical gated recurrent units). RESULTS: Evaluation was performed using cross-validation to measure 1) classification accuracy for talk-turns and 2) precision, recall, and F1 scores at the visit level. Experimental results showed that sequential models had higher classification accuracy at the talk-turn level and higher precision at the visit level. Independent models had higher recall scores at the visit level compared with sequential models. CONCLUSIONS: Incorporating sequential information across talk-turns improves the accuracy of topic prediction in patient-provider dialog by smoothing out noisy information from talk-turns. Although the results are promising, more advanced prediction techniques and larger labeled datasets will likely be required to achieve prediction performance appropriate for real-world clinical applications.

Randomised trial to evaluate the effectiveness and impact of offering postvisit decision support and assistance in obtaining physician-recommended colorectal cancer screening: the e-assist: Colon Health study-a protocol study.

INTRODUCTION: How to provide practice-integrated decision support to patients remains a challenge. We are testing the effectiveness of a practice-integrated programme targeting patients with a physician recommendation for colorectal cancer (CRC) screening. METHODS AND ANALYSIS: In partnership with healthcare teams, we developed 'e-assist: Colon Health', a patient-targeted, postvisit CRC screening decision support programme. The programme is housed within an electronic health record (EHR)-embedded patient portal. It leverages a physician screening recommendation as the cue to action and uses the portal to enrol and intervene with patients. Programme content complements patient-physician discussions by encouraging screening, addressing common questions and assisting with barrier removal. For evaluation, we are using a randomised trial in which patients are randomised to receive e-assist: Colon Health or one of two controls (usual care plus or usual care). Trial participants are average-risk, aged 50-75 years, due for CRC screening and received a physician order for stool testing or colonoscopy. Effectiveness will be evaluated by comparing screening use, as documented in the EHR, between trial enrollees in the e-assist: Colon Health and usual care plus (CRC screening information receipt) groups. Secondary outcomes include patient-perceived benefits of, barriers to and support for CRC screening and patient-reported CRC screening intent. The usual care group will be used to estimate screening use without intervention and programme impact at the population level. Differences in outcomes by study arm will be estimated with hierarchical logit models where patients are nested within physicians. ETHICS AND DISSEMINATION: All trial aspects have been approved by the Institutional Review Board of the health system in which the trial is being conducted. We will disseminate findings in diverse scientific venues and will target clinical and quality improvement audiences via other venues. The intervention could serve as a model for filling the gap between physician recommendations and patient action. TRIAL REGISTRATION NUMBER: NCT02798224; Pre-results.