A pancreatic cancer risk prediction model (Prism) developed and validated on large-scale US clinical data.

BACKGROUND: Pancreatic Duct Adenocarcinoma (PDAC) screening can enable early-stage disease detection and long-term survival. Current guidelines use inherited predisposition, with about 10% of PDAC cases eligible for screening. Using Electronic Health Record (EHR) data from a multi-institutional federated network, we developed and validated a PDAC RISk Model (Prism) for the general US population to extend early PDAC detection. METHODS: Neural Network (PrismNN) and Logistic Regression (PrismLR) were developed using EHR data from 55 US Health Care Organisations (HCOs) to predict PDAC risk 6-18 months before diagnosis for patients 40 years or older. Model performance was assessed using Area Under the Curve (AUC) and calibration plots. Models were internal-externally validated by geographic location, race, and time. Simulated model deployment evaluated Standardised Incidence Ratio (SIR) and other metrics. FINDINGS: With 35,387 PDAC cases, 1,500,081 controls, and 87 features per patient, PrismNN obtained a test AUC of 0.826 (95% CI: 0.824-0.828) (PrismLR: 0.800 (95% CI: 0.798-0.802)). PrismNN's average internal-external validation AUCs were 0.740 for locations, 0.828 for races, and 0.789 (95% CI: 0.762-0.816) for time. At SIR = 5.10 (exceeding the current screening inclusion threshold) in simulated model deployment, PrismNN sensitivity was 35.9% (specificity 95.3%). INTERPRETATION: Prism models demonstrated good accuracy and generalizability across diverse populations. PrismNN could find 3.5 times more cases at comparable risk than current screening guidelines. The small number of features provided a basis for model interpretation. Integration with the federated network provided data from a large, heterogeneous patient population and a pathway to future clinical deployment. FUNDING: Prevent Cancer Foundation, TriNetX, Boeing, DARPA, NSF, and Aarno Labs.

Exploring Breast Cancer Systemic Drug Therapy Patterns in Real-World Data.

PURPOSE: To explore medications and their administration patterns in real-world patients with breast cancer. METHODS: A retrospective study was performed using TriNetX, a federated network of deidentified, Health Insurance Portability and Accountability Act-compliant data from 21 health care organizations across North America. Patients diagnosed with breast cancer between January 1, 2013, and May 31, 2022, were included. We investigated a rule-based and unsupervised learning algorithm to extract medications and their administration patterns. To group similar administration patterns, we used three features in k-means clustering: total number of administrations, median number of days between administrations, and standard deviation of the days between administrations. We explored the first three lines of therapy for patients classified into six groups on the basis of their stage at diagnosis (early as stages I-III v late as stage IV) and the sensitivity of the tumor's receptors to targeted therapies: hormone receptor-positive/human epidermal growth factor 2-negative (HR+/ERBB2-), ERBB2-positive (ERBB2+/HR±), or triple-negative (TN; HR-/ERBB2-). To add credence to the derived regimens, we compared them to the National Comprehensive Cancer Network (NCCN): Breast Cancer (version 2.2023) recommendations. RESULTS: In early-stage HR+/ERBB2- and TN groups, the most common regimens were (1) cyclophosphamide and docetaxel, administered once every 3 weeks for three to six cycles and (2) cyclophosphamide and doxorubicin, administered once every 2 weeks for four cycles, followed by paclitaxel administered once every week for 12 cycles. In the early-stage ERBB2+/HR± group, most patients were administered carboplatin and docetaxel with or without pertuzumab and with trastuzumab (for six or more cycles). Medications most commonly administered in our data set (7,798 patients) agreed with recommendations from the NCCN in terms of medications (regimens), number of administrations (cycles), and days between administrations (cycle length). CONCLUSION: Although there is a general agreement with the NCCN Guidelines, real-world medication data exhibit variability in the medications and their administration patterns.

Evolving Effect of the COVID-19 Pandemic on Cancer-Related Encounters.

PURPOSE: This is an update to a previously published report characterizing the impact that efforts to control the COVID-19 pandemic have had on the normal course of cancer-related encounters. METHODS: Data were analyzed from 22 US health care organizations (members of the TriNetX global network) having relevant, up-to-date encounter data. Although the original study compared encounter data pre-COVID-19 (January-April 2019) with the corresponding months in 2020, this update considers data through April 2021. As before, cohorts were generated for all neoplasm patients (malignant, benign, in situ, and of unspecified behavior), all new incidence neoplasm patients, exclusively malignant neoplasm patients, and new incidence malignant neoplasm patients. Data on the initial cancer stage were available for calendar year 2020 from about one third of the study's organizations. RESULTS: Although COVID-19 cases fluctuated through 2021, newly diagnosed cancers closely paralleled the prepandemic base year 2019. Similarly, screening for breast, colorectal, and cervical cancers quickly recovered beginning in May 2020 to prepandemic numbers. Preliminary data for the initial cancer stage showed no significant difference (P > .10) in distribution for breast or colon cancers between 2019 and 2020. CONCLUSION: Although the number of COVID-19 cases fluctuated, the steep declines observed during March and April 2020 in screening for breast and colon cancer and patients with newly diagnosed cancer did not continue through the rest of 2020 and into April 2021. Screening and new incidence cancer numbers quickly rose compared with prepandemic levels. The concern that more patients with advanced-stage cancer would be seen in the months following the drastic dips of March-April 2020 was not realized as the major disruption to normal cancer care was limited to these 2 months.

Recommendations for patient similarity classes: results of the AMIA 2019 workshop on defining patient similarity.

Defining patient-to-patient similarity is essential for the development of precision medicine in clinical care and research. Conceptually, the identification of similar patient cohorts appears straightforward; however, universally accepted definitions remain elusive. Simultaneously, an explosion of vendors and published algorithms have emerged and all provide varied levels of functionality in identifying patient similarity categories. To provide clarity and a common framework for patient similarity, a workshop at the American Medical Informatics Association 2019 Annual Meeting was convened. This workshop included invited discussants from academics, the biotechnology industry, the FDA, and private practice oncology groups. Drawing from a broad range of backgrounds, workshop participants were able to coalesce around 4 major patient similarity classes: (1) feature, (2) outcome, (3) exposure, and (4) mixed-class. This perspective expands into these 4 subtypes more critically and offers the medical informatics community a means of communicating their work on this important topic.

Effects of the COVID-19 Pandemic on Cancer-Related Patient Encounters.

PURPOSE: While there are studies under way to characterize the direct effects of the COVID-19 pandemic on the care of patients with cancer, there have been few quantitative reports of the impact that efforts to control the pandemic have had on the normal course of cancer diagnosis and treatment encounters. METHODS: We used the TriNetX platform to analyze 20 health care institutions that have relevant, up-to-date encounter data. Using this COVID and Cancer Research Network (CCRN), we compared cancer cohorts identified by querying encounter data pre-COVID (January 2019-April 2019) and current (January 2020-April 2020). Cohorts were generated for all patients with neoplasms (malignant, benign, in situ, and of unspecified behavior), with new incidence neoplasms (first encounter), with exclusively malignant neoplasms, and with new incidence malignant neoplasms. Data from a UK institution were similarly analyzed. Additional analyses were performed on patients with selected cancers, as well as on those having had cancer screening. RESULTS: Clear trends were identified that suggest a significant decline in all current cohorts explored, with April 2020 displaying the largest decrease in the number of patients with cancer having encounters. Of the cancer types analyzed, lung, colorectal, and hematologic cancer cohorts exhibited smaller decreases in size in April 2020 versus 2019 (-39.1%, -39.9%, -39.1%, respectively) compared with cohort size decreases for breast cancer, prostate cancer, and melanoma (-47.7%, -49.1%, -51.8%, respectively). In addition, cancer screenings declined drastically, with breast cancer screenings dropping by -89.2% and colorectal cancer screenings by -84.5%. CONCLUSION: Trends seen in the CCRN clearly suggest a significant decrease in all cancer-related patient encounters as a result of the pandemic. The steep decreases in cancer screening and patients with a new incidence of cancer suggest the possibility of a future increase in patients with later-stage cancer being seen initially as well as an increased demand for cancer screening procedures as delayed tests are rescheduled.

Using a Federated Network of Real-World Data to Optimize Clinical Trials Operations.

Clinical trials, whether industry, cooperative group sponsored, or investigator initiated, have an unacceptable rate of failure as a result of the inability to recruit sufficient numbers of patients. Even those trials that are completed often require time-consuming protocol amendments to achieve accrual goals. These inefficiencies in clinical trial research result in increasing costs and prolong the time needed to bring improved treatments to cancer clinical practice. TriNetX has developed a clinical research collaboration platform-deployed by a federated network of health care organizations (HCOs), pharmaceutical firms (Pharma), and contract research organizations (CROs)-to enable data-driven clinical research study design to reduce accrual failure and protocol amendment. Currently, the network extends to 55 HCOs and covers 84 million patients, mostly within the United States, but with a growing international presence. (Many of the HCOs in United States are Clinical and Translational Science Awardees and/or National Cancer Institute-designated cancer centers.) The TriNetX business model includes Pharma and the CROs as sponsors whose subscriptions financially support the network, including the software and hardware costs of the HCOs. Furthermore, as each HCO network member has their data harmonized with the TriNetX model upon joining, data sharing among them does not require any technical processes to establish connectivity. To date, on the basis of the data on the network, HCOs have been presented approximately 757 studies by Pharma and CROs, and four data-sharing subnetworks have been formed among member HCOs.

Effects of documentation-based decision support on chronic disease management.

OBJECTIVE: To evaluate whether a new documentation-based clinical decision support system (CDSS) is effective in addressing deficiencies in the care of patients with coronary artery disease (CAD) and diabetes mellitus (DM). STUDY DESIGN: Controlled trial randomized by physician. METHODS: We assigned primary care physicians (PCPs) in 10 ambulatory practices to usual care or the CAD/DM Smart Form for 9 months. The primary outcome was the proportion of deficiencies in care that were addressed within 30 days after a patient visit. RESULTS: The Smart Form was used for 5.6% of eligible patients. In the intention-to-treat analysis, patients of intervention PCPs had a greater proportion of deficiencies addressed within 30 days of a visit compared with controls (11.4% vs 10.1%, adjusted and clustered odds ratio =1.14; 95% confidence interval, 1.02-1.28; P = .02). Differences were more pronounced in the "on-treatment" analysis: 17.0% of deficiencies were addressed after visits in which the Smart Form was used compared with 10.6% of deficiencies after visits in which it was not used (P <.001). Measures that improved included documentation of smoking status and prescription of antiplatelet agents when appropriate. CONCLUSIONS: Overall use of the CAD/DM Smart Form was low, and improvements in management were modest. When used, documentation-based decision support shows promise, and future studies should focus on refining such tools, integrating them into current electronic health record platforms, and promoting their use, perhaps through organizational changes to primary care practices.