Essential team science skills for biostatisticians on collaborative research teams.

Introduction: Despite the critical role that quantitative scientists play in biomedical research, graduate programs in quantitative fields often focus on technical and methodological skills, not on collaborative and leadership skills. In this study, we evaluate the importance of team science skills among collaborative biostatisticians for the purpose of identifying training opportunities to build a skilled workforce of quantitative team scientists. Methods: Our workgroup described 16 essential skills for collaborative biostatisticians. Collaborative biostatisticians were surveyed to assess the relative importance of these skills in their current work. The importance of each skill is summarized overall and compared across career stages, highest degrees earned, and job sectors. Results: Survey respondents were 343 collaborative biostatisticians spanning career stages (early: 24.2%, mid: 33.8%, late: 42.0%) and job sectors (academia: 69.4%, industry: 22.2%, government: 4.4%, self-employed: 4.1%). All 16 skills were rated as at least somewhat important by > 89.0% of respondents. Significant heterogeneity in importance by career stage and by highest degree earned was identified for several skills. Two skills ("regulatory requirements" and "databases, data sources, and data collection tools") were more likely to be rated as absolutely essential by those working in industry (36.5%, 65.8%, respectively) than by those in academia (19.6%, 51.3%, respectively). Three additional skills were identified as important by survey respondents, for a total of 19 collaborative skills. Conclusions: We identified 19 team science skills that are important to the work of collaborative biostatisticians, laying the groundwork for enhancing graduate programs and establishing effective on-the-job training initiatives to meet workforce needs.

Evaluation of antibiotic-resistant bacteria in home kitchens and bathrooms: Is there a link between home disinfectant use and antibiotic resistance?

OBJECTIVE: To determine the relationship between home disinfectant use and the prevalence of antibiotic resistance among environmental isolates of human pathogens. METHODS: Bacteria were cultured from 5 kitchen and 5 bathroom sites using quantitative methods. Antibiotic susceptibility was determined by standard methods. Home disinfectant use was assessed via a questionnaire. RESULTS: The overall total mean log10 counts (total CFU) for the kitchen and bathroom were 4.31 and 4.88, respectively. Gram-positive bacteria were more common in the bathroom (4.05) than in the kitchen (3.60), while Gram-negative bacilli were more common in the kitchen (4.23) than in the bathroom (3.86). The sink and bath drains were the most contaminated sites with 6.16-log10 of total CFU and 6.6-log10 in the kitchen and bathroom, respectively. Households reported cleaning frequency with a variety of commercial products. Most respondents used antibacterial products (eg, soaps, surface disinfectants) in the home. Antibiotic-resistant pathogens were infrequently isolated in the homes evaluated. CONCLUSIONS: Compared to pathogens causing community-acquired clinical infections in the ICARE study, pathogens isolated from households are less likely to demonstrate antibiotic resistance. In addition, no relationship between antibacterial use or frequency of cleaning or disinfection and antibiotic resistance was revealed.

Treatment Effect Estimates From Pilot Trials Are Unreliable.

CONTEXT: The CONSORT guideline defines a pilot trial as a small-scale version of a desired future efficacy trial that is intended to answer the key questions of whether and how a larger study should be done. For example, a pilot trial might evaluate different approaches to data collection or outcome measurement. However, pilot trials are unreliable for assessing treatment efficacy due to the statistical phenomenon called sampling variability. OBJECTIVES: In this tutorial we use computer simulation to demonstrate the influence of sampling variability on efficacy estimates from pilot trials, illustrating why pilot trial designs should not be used to evaluate whether a treatment is promising or not. METHODS: We simulate a 2-arm parallel group trial (N=20 per group) with a survival outcome as an example. Simulations are done under two scenarios: 1) the treatment is efficacious at the level of a hypothetical minimum clinically important difference (hazard ratio [HR] = 0.75); and 2) the treatment is not efficacious (HR=1). RESULTS: As expected, in both simulated scenarios the range of observed results is distributed around the true treatment effect, HR=0.75 or HR=1. Importantly, ∼20% of trials simulated under scenario 1 incorrectly suggest the treatment may be harmful (HR > 1). Under scenario 2, half of the simulated studies incorrectly suggest the treatment is beneficial. CONCLUSION: Treatment effect estimates from pilot trials should not be used to make future development decisions regarding a novel therapy because of the high risk of misleading conclusions.

A Visual Tool to Help Develop a Statistical Analysis Plan for Randomized Trials in Palliative Care.

Collaboration with a statistician about the design of a statistical analysis plan can be enhanced by illustrating how statisticians conceptualize their task. This conceptualization can be represented by a directed acyclic graph (DAG), which illustrates the statistician's approach and also provides an actionable tool to assist in the development of the plan.

Institutional approaches to preventing questionable research practices.

Questionable research practices (QRP) are actions taken by researchers that span a range of concern related to violation of research best practices, and ultimately expose institutions and research participants to risk. Numerous studies have shown that QRP are common. The continued prevalence of QRP indicates that existing approaches for dealing with QRP are falling short. In this editorial we discuss the risks associated with QRP and propose mitigation strategies at the institutional level using a common QRP as an example, questionable treatment of subgroup analyses. We argue that the need for institutional intervention in cases such as this are particularly motivating when both the investigator and the institution have a substantial financial conflict of interest related to intellectual property that requires the investigator's expertise to continue developing. To address this, we propose an expansion of the traditional conflict of interest management process.

Designing Psychosocial Intervention Pilot Studies: A Tutorial for Palliative Care Investigators.

This is a tutorial on designing a persuasive pilot study of a psychosocial intervention (e.g., behavioral symptom management) in the palliative care setting. This tutorial is most relevant for early stages of intervention research that aims to progress toward a randomized controlled trial with a high degree of internal validity. Broadly, a pilot study aims to address multiple elements of feasibility and acceptability so that investigators are well positioned for the next study in their program of research. To assist investigators in writing compelling grant applications we designed this tutorial as an annotated checklist of goals that a pilot study within the palliative care domain should seek to accomplish. These goals include the following: 1) begin with the end in mind, 2) use a formal conceptual model, 3) use measures with strong psychometric properties, 4) justify the timing of study sessions and assessments, 5) test recruitment methods, 6) estimate retention, 7) assess interventionist fidelity, 8) assess acceptability of the intervention, 9) assess feasibility, and 10) identify barriers to the next study. We elaborate on these goals by describing an ongoing pilot study testing the feasibility and acceptability of a psychosocial pain management intervention for patients with advanced cancer. Pilot studies are crucial for building a successful program of research, but they are also limited in terms of their sample size and overall objectives. A persuasive pilot study is one that is limited yet useful rather than limited and trivial.

Aerobic Versus Resistance Training Effects on Ventricular-Arterial Coupling and Vascular Function in the STRRIDE-AT/RT Trial.

Background: The goal was studying the differential effects of aerobic training (AT) vs. resistance training (RT) on cardiac and peripheral arterial capacity on cardiopulmonary (CP) and peripheral vascular (PV) function in sedentary and obese adults. Methods: In a prospective randomized controlled trial, we studied the effects of 6 months of AT vs. RT in 21 subjects. Testing included cardiac and vascular ultrasoundography and serial CP for ventricular-arterial coupling (Ees/Ea), strain-based variables, brachial artery flow-mediated dilation (BAFMD), and peak VO2 (pVO2; mL/kg/min) and peak O2-pulse (O2p; mL/beat). Results: Within the AT group (n = 11), there were significant increases in rVO2 of 4.2 mL/kg/min (SD 0.93) (p = 0.001); O2p of 1.9 mL/beat (SD 1.3) (p = 0.008) and the brachial artery post-hyperemia peak diameter 0.18 mm (SD 0.08) (p = 0.05). Within the RT group (n = 10) there was a significant increase in left ventricular end diastolic volume 7.0 mL (SD 9.8; p = 0.05) and percent flow-mediated dilation (1.8%) (SD 0.47) (p = 0.004). Comparing the AT and RT groups, post exercise, rVO2 2.97, (SD 1.22), (p = 0.03), O2p 0.01 (SD 1.3), (p = 0.01), peak hyperemic blood flow volume (1.77 mL) (SD 140.69) (p = 0.009), were higher in AT, but LVEDP 115 mL (SD 7.0) (p = 0.05) and Ees/Ea 0.68 mmHg/ml (SD 0.60) p = 0.03 were higher in RT. Discussion: The differential effects of AT and RT in this hypothesis generating study have important implications for exercise modality and clinical endpoints.

Behavioral cancer pain intervention using videoconferencing and a mobile application for medically underserved patients: Rationale, design, and methods of a prospective multisite randomized controlled trial.

BACKGROUND: Women with breast cancer in medically underserved areas are particularly vulnerable to persistent pain and disability. Behavioral pain interventions reduce pain and improve outcomes. Cancer patients in medically underserved areas receive limited adjunctive cancer care, as many lack access to pain therapists trained in behavioral interventions, face travel barriers to regional medical centers, and may have low literacy and limited resources. mHealth technologies have the potential to decrease barriers but must be carefully adapted for, and efficacy-tested with, medically underserved patients. We developed an mHealth behavioral pain coping skills training intervention (mPCST-Community). We now utilize a multisite randomized controlled trial to: 1) test the extent mPCST-Community reduces breast cancer patients' pain severity (primary outcome), pain interference, fatigue, physical disability, and psychological distress; 2) examine potential mediators of intervention effects; and 3) evaluate the intervention's cost and cost-effectiveness. METHODS/DESIGN: Breast cancer patients (N = 180) will be randomized to mPCST-Community or an attention control. mPCST-Community's four-session protocol will be delivered via videoconferencing at an underserved community clinic by a remote pain therapist at a major medical center. Videoconference sessions will be supplemented with a mobile application. Participants will complete self-report measures at baseline, post-intervention, and 3- and 6-month follow-ups. CONCLUSIONS: mPCST-Community has the potential to reduce pain and disability, and decrease barriers for cancer patients in medically underserved areas. This is one of the first trials to test an mHealth behavioral cancer pain intervention developed specifically for medically underserved communities. If successful, it could lead to widespread implementation and decreased health disparities.

Two Questions About the Design of Cluster Randomized Trials: A Tutorial.

This is a short tutorial on two key questions that pertain to cluster randomized trials (CRTs): 1) Should I perform a CRT? and 2) If so, how do I derive the sample size? In summary, a CRT is the best option when you "must" (e.g., the intervention can only be administered to a group) or you "should" (e.g., because of issues such as feasibility and contamination). CRTs are less statistically efficient and usually more logistically complex than individually randomized trials, and so reviewing the rationale for their use is critical. The most straightforward approach to the sample size calculation is to first perform the calculation as if the design were randomized at the level of the patient and then to inflate this sample size by multiplying by the "design effect", which quantifies the degree to which responses within a cluster are similar to one another. Although trials with large numbers of small clusters are more statistically efficient than those with a few large clusters, trials with large clusters can be more feasible. Also, if results are to be compared across individual sites, then sufficient sample size will be required to attain adequate precision within each site. Sample size calculations should include sensitivity analyses, as inputs from the literature can lack precision. Collaborating with a statistician is essential. To illustrate these points, we describe an ongoing CRT testing a mobile-based app to systematically engage families of intensive care unit patients and help intensive care unit clinicians deliver needs-targeted palliative care.

A novel decision aid for acute myeloid leukemia: a feasibility and preliminary efficacy trial.

PURPOSE: Acute myeloid leukemia (AML) is a hematologic malignancy characterized by a poor prognosis but also a paradoxical possibility of cure. This renders decision-making complex and imminent. Unfortunately, many patients with AML misestimate their prognosis and treatment risk. While decision aids can improve illness understanding and reduce decisional conflict, there are no validated decision aids for AML. We developed and tested a novel AML decision aid (NCT03442452). METHODS: Patients (n = 20) were recruited at Duke University from May 2018 to February 2019. Participants completed assessments of AML knowledge and decisional conflict, before and after using the electronic decision aid. The primary endpoint was feasibility (endpoint met if > 80% of study participants completed all study components). Secondary analyses of efficacy were conducted using paired t tests for dependent pre-/post-samples. RESULTS: The primary endpoint of feasibility was met (100% of participants completed all study components). Secondary analyses showed improved knowledge and reduced decisional conflict after using the decision aid. Knowledge scores improved from a mean of 11.8 (out of 18) correct items at baseline to 15.1 correct items after using the decision aid (mean difference 3.35; p < 0.0001). Decisional conflict scores reduced significantly from baseline to post-test as well (mean difference - 6.5; p = 0.02). CONCLUSION: These findings suggest that our AML decision aid is a useful tool to improve the patient experience and promote shared decision-making in AML. A randomized efficacy trial is planned.

Some statistical memes which sound correct but aren't quite: Application to the analysis of observational databases used in learning health systems.

We consider four memes, correct within the context of randomized trials but requiring modification for the analysis of the observational databases typically associated with learning health systems: (a) the right answer always requires randomization; (b) a bigger database is always a better database; (c) statistical adjustment always works if based on a large enough database; and (d) always make a formal adjustment when testing multiple hypotheses. The rationale for these memes within the context of randomized trials is discussed, and the memes are restated in a fashion that is consistent with learning health systems.

"No more than one day per week" seems simple enough: So why is conflict of commitment so confusing?

Conflict of interest (COI) and conflict of commitment (COC) are similar in that both concern the "risk that professional judgment or actions regarding a primary interest will be unduly influenced by a secondary interest" (Institute of Medicine). The Institute of Medicine recommends that the circumstances, risks, and evaluative frameworks for COI and COC warrant separate consideration. Here, we discuss some principles (and thus an evaluative framework) underlying COC and hope to provide a clarifying framework which extends across institutions. We propose that: An institution should become concerned about relationships, whether considered individually or as a whole, which because of the amount of the time commitment and/or the amount of the monetary compensation, potentially induces a dual loyalty or otherwise calls into question the ability of the faculty member to discharge their primary responsibility to the institution. The impact of single relationships on single actions constitutes COI; the net impact of multiple relationships and/or the overall ability to fully discharge primary institutional responsibilities constitutes COC.

Vascular Cellular Adhesion Molecule-1 (VCAM-1) and Memory Impairment in African-Americans after Small Vessel-Type Stroke.

BACKGROUND: African-Americans (AA) are 3 times more likely to have small-vessel-type ischemic strokes (SVS) than Whites. Small vessel strokes are associated with cognitive impairment, a relationship incompletely explained by white matter hyperintensity (WMH) burden. We examined whether inflammatory/endothelial dysfunction biomarkers are associated with cognition after SVS in AAs. METHODS: Biomarkers were obtained in 24 subjects (median age 56.5 years, 54% women, median 12 years education). Cognition was assessed more than 6 weeks poststroke using the memory composite score (MCS), which was generated using recall from the Hopkins Verbal Learning Test-II and Brief Visuospatial Memory Test-Revised. A semi-automated, volumetric protocol was used to quantify WMH volume (WMHv) on clinical MRI scans. Potential biomarkers including vascular cell adhesion molecule-1 (VCAM-1), interleukin-1 receptor antagonist, interleukin-6, interleukin-8, interleukin-10, interferon gamma, and thrombin-antithrombin (TAT) were log-transformed and correlated with MCS with adjustment for potential confounders. RESULTS: Among serum biomarkers, only VCAM-1-correlated with poorer memory based on the MCS (r = -.659; P = .0006). VCAM-1 (r = .554; P = .005) and age (r = .479; P = .018) correlated with WMHv; VCAM-1 was independently associated with MCS after adjustment for WMHv, age, and education (P = .023). CONCLUSIONS: The findings of this exploratory analysis suggest that endothelial dysfunction and inflammation as reflected by VCAM-1 levels may play a role in poststroke cognitive impairment. Additional studies are needed to validate this observation and to evaluate this relationship in non-AAs and with other stroke types and compare this finding to cognitive impairment in nonstroke populations.

An analysis of missing items in real-world electronic patient reported outcomes data: implications for clinical care.

PURPOSE: Utilization of electronic patient-reported outcomes (ePROs) in the clinic can improve quality of life and prolong survival in cancer care. However, there remain unanswered questions regarding trends in missing data and the potential effect on real-time patient care. METHODS: This study utilized a prospectively collected dataset of ePROs from oncology clinics that administered the Patient Care Monitor 2.0 (PCM), a validated symptoms survey assessing 78 items for men, and 86 for women. We tabulated the frequency of missing items, by item and domain (emotional, functional and physical symptom-related), and examined these by age, gender, education, race and marital status. RESULTS: Within 20,986 encounters, there were responses to at least 1 PCM item from 6933 unique patients. The highest frequency of missing answers occurred for: "attend a paid job" (10.7%), "reduced sexual enjoyment" (3.8%), and "run" (3.7%). By domain, 12.3% of functional, 8.4% of physical symptom-related, and 1.6% of emotional constructs contained at least one missing item. For functional and physical symptom-related items, missingness was most common in patients >60 years old. CONCLUSION: The frequency of missingness was highest for functional items, like attending a paid job, suggesting that some respondents (e.g., retirees without a paid job) skipped questions that were less applicable to them. More universal issues for cancer patients, such as emotional well-being, had much lower frequencies of missingness. This suggests differential item completion that warrants further study to understand the inherent drivers. Identifying causes of missingness could improve the clinical utility of ePROs and highlight opportunities to personalize care.

Managing the science in the presence of financial conflict of interest.

We consider the question of why (primarily) and how (secondarily) to perform scientific oversight of research performed by investigators with a financial conflict of interest (COI). One way to frame the trade-off of having investigators with financial COI participate in research is through a decision rule: "Our institution is willing for financially conflicted investigators to participate in research around their intellectual property if (a) the science is likely to be significantly better with their participation (or if other significant benefits accrue); and (b) the COI can be adequately managed". A key component of COI management is the demonstration that the underlying science is sufficiently rigorous and transparent, and in turn, a scientific oversight plan is a key element of that demonstration. Scientific oversight plans should be proactively generated, by individuals (i.e., scientific reviewers) who are independent and expert, and they should assess the rigor and transparency of the research, in a fashion which is fair and efficient. Abbreviations: COI: conflict of interest; SOP: scientific oversight plan.

Implementation of a responsible conduct of research education program at Duke University School of Medicine.

Academic medical centers rarely require all of their research faculty and staff to participate in educational programs on the responsible conduct of research (RCR). There is also little published evidence of RCR programs addressing high-profile, internal cases of misconduct as a way of promoting deliberation and learning. In the wake of major research misconduct, Duke University School of Medicine (DUSoM) expanded its RCR education activities to include all DUSoM faculty and staff engaged in research. The program included formal deliberation of the Translational Omics misconduct case, which occurred at Duke. Over 5,000 DUSoM faculty and staff participated in the first phase of this new program, with a 100% completion rate. The article reports on the program's development, challenges and successes, and future directions. This experience at Duke University illustrates that, although challenging and resource intensive, engagement with RCR activities can be integrated into programs for all research faculty and staff. Formal, participatory deliberation of recent cases of internal misconduct can add a novel dimension of reflection and openness to RCR educational activities.

Prognostic significance of primary tumor sidedness in patients undergoing liver resection for metastatic colorectal cancer.

BACKGROUND: Approximately 38% of patients with colorectal cancer will develop isolated liver metastases. Sidedness of colon tumor is identified in non-metastatic and unresected metastatic cancers as predictive of survival, yet its dedicated analysis in resected liver metastases is minimal. Our primary aim was to assess whether left-sided primary tumors improve prognosis in stage IV cancer patients undergoing curative-intent liver metastasectomy; it was hypothesized that it would. METHODS: This is a retrospective, observational cohort study from 1996 to 2016 in a single tertiary-care facility. Survival from diagnosis was calculated via Kaplan-Meier method and compared between the right and left sides via log-rank analysis. RESULTS: Median survival differs significantly between colorectal tumors of the right and left origins after hepatic metastasectomy in 612 patients. In patients with right-sided tumors, median survival from diagnosis was 4.5 years (IQR 4.1-5.3), and 6.3 years (IQR 5.6-6.9) in those with left tumors (HR 1.5, 95% CI 1.38-1.60, p < 0.001). CONCLUSION: As in studies on earlier-stage or unresected metastatic disease, tumor sidedness is an important prognostic factor in patient survival with liver metastasectomy. Clinical risk scores should include side of primary tumor. Further work is needed to determine the molecular basis for this difference.

Pruritus in patients with solid tumors: an overlooked supportive care need.

PURPOSE: Pruritus is a common symptom in cutaneous malignancies, but its impact on patients with solid tumors is unclear. We explored the impact and management of pruritus in patients with solid tumors, using patient-reported outcomes (PRO) data from a real-world registry. METHODS: From 2006 to 2011, patients seen in the Duke Cancer Institute reported their symptoms via the Patient Care Monitor v2.0, a validated PRO tool that includes a 0-10-point question about pruritus severity. From > 25,000 encounters, 203 patients reported severe pruritus (> 6/10) on at least one visit and 506 total visits were abstracted where patients reported either moderate or severe pruritus (> 3/10). From this cohort, we abstracted demographics, diagnosis, stage, cancer therapy, anti-pruritic therapy, and clinicians' responses. RESULTS: Mean age was 59.8 (SD 13.3), 134 (66%) were female, 125 (62%) were Caucasian, and 65 (32%) were African American. Breast cancer was the most common tumor (36.5%), followed by lung cancer (23.2%). Mean pruritus severity score was 6.8 (SD 1.8) for patients on chemotherapy, 6.9 (SD 1.8) for patients on targeted therapy alone or in combination, and 7.1(SD 1.8) for patients off treatment. Overall, 67% of patients reported at least two episodes of moderate-severe pruritus (mean # of visits 4.2 (SD 2.7)). Despite frequent report of severe and persistent pruritus, this was mentioned in just 28% of clinician notes and an intervention was recommended/prescribed in only 7% of visits. CONCLUSIONS: Pruritus is an under-addressed symptom in patients with solid tumors. Additional research is needed to understand the burden of pruritus in affected populations.

Caregiver-guided pain coping skills training for patients with advanced cancer: Background, design, and challenges for the CaringPals study.

BACKGROUND/AIMS: Pain is a major concern of patients with advanced cancer and their caregivers. There is strong evidence that pain coping skills training interventions based on cognitive-behavioral principles can reduce pain severity and pain interference. However, few such interventions have been tested for patients with advanced cancer and their family caregivers. This study aims to test the efficacy of a caregiver-guided pain coping skills training protocol on patient and caregiver outcomes. METHODS: A total of 214 patients age ≥18 with Stage III-Stage IV cancer and moderate to severe pain, along with their family caregivers, are being identified and randomized with a 1:1 allocation to the caregiver-guided pain coping skills training intervention or enhanced treatment-as-usual. Dyads in both conditions receive educational resources on pain management, and the caregiver-guided pain coping skills training intervention includes three weekly 60-min sessions conducted with the patient-caregiver dyad via videoconference. Measures of caregiver outcomes (self-efficacy for helping the patient manage pain, caregiver strain, caregiving satisfaction, psychological distress) and patient outcomes (self-efficacy for pain management, pain intensity and interference, psychological distress) are collected at baseline and post-intervention. Caregiver outcomes are also collected 3 and 6 months following the patient's death. The study is enrolling patients from four tertiary care academic medical centers and one free-standing hospice and palliative care organization. The primary outcome is caregiver self-efficacy for helping the patient manage pain. RESULTS: This article describes challenges in the design and implementation of the CaringPals trial. Key issues for trial design include the identification and recruitment of patients with advanced cancer and pain, and the follow-up and collection of data from caregivers following the patient's death. CONCLUSION: The CaringPals trial addresses a gap in research in pain coping skills training interventions by addressing the unique needs of patients with advanced cancer and their caregivers. Findings from this study may lead to advances in the clinical care of patients with advanced cancer and pain, as well as a better understanding of the effects of training family caregivers to help patients cope with pain.