Lead Site

Simulation for Attending Obstetricians to Improve Technical Skills for Managing Postpartum Hemorrhage (R01-HD107451)

Postpartum hemorrhage (PPH) is a rare but potentially catastrophic obstetric emergency, and a leading cause of severe maternal morbidity and preventable mortality in the U.S. Women of racial and ethnic minority groups and living in underserved and remote areas are at greatest risk. Obstetric units must be prepared to manage PPH, and obstetric (OB) attendings must maintain technical skills necessary to control PPH. To address this, simulation is a promising approach. Obstetric simulation has shown effectiveness for team-based training, yet there is compelling evidence for the use of simulation to improve technical skills. Further, prior studies and simulation research from other surgical specialties demonstrate the high likelihood of successfully validating an existing high-fidelity simulator and establishing its effectiveness in training OB attendings in technical skills needed to manage PPH. However, a simulation program directed at OB attendings who work in varied hospital settings across the country is subject to known obstacles for implementing and spreading simulation. We propose a staged, multicenter study to establish simulation, using innovative approaches to mitigate barriers, as an effective means for training OB attendings in three technical skills demonstrated as effective for controlling PPH: Bakri balloon placement, B-Lynch suture placement, and O'Leary suture placement. OB attendings will be assessed at baseline, mentored, and re-assessed. The primary outcomes will be change in assessment scores (measured by validated and widely used OSATS) and change in percentage of OB attendings who achieve competency (score of 60) from baseline to program completion. Our main hypotheses are that the PPH technical skills simulation program will show a significant increase in skill level among OB attendings and that the proposed innovations (affordable, remote, asynchronous) will address known barriers to implementation. We propose the following specific aims: Aim 1: Establish simulation as an effective means for training OB attendings in technical skills for treating PPH by validating a high-fidelity simulator and rigorously evaluating its effectiveness for a PPH technical skills simulation program with an in-person mentoring program. Aim 2: Evaluate three innovations aimed at overcoming common barriers to simulation training for OB attendings: Aim 2A: Validate a novel, affordable medium-fidelity simulator and evaluate its effectiveness in a PPH technical skills program using in-person mentoring; Aim 2B: Evaluate a PPH technical skills simulation program with remote synchronous mentoring; Aim 2C: Evaluate a PPH technical skills simulation program with asynchronous mentoring. Aim 3: Evaluate the effectiveness of a PPH technical skills simulation program in rural and community hospital settings, using the approach that sufficiently addresses the barriers to implementation established in Aim 2 (affordable, remote, asynchronous) and is acceptable to OB attendings. If successful, results will inform best practices in obstetrics, advance integration of simulation in medical education and training standards, and enable dissemination in varied settings nationwide.

Role: Lead Site; Jason Adelman MPI and Dena Goffman MPI
Study Period: 6/1/2022–5/31/2027

Developing and Validating Health IT Safety Measures to Capture Violations of the Five Rights of Medication Safety (AHRQ R01-HS024538)

The 2011 Institute of Medicine report, Health IT and Patient Safety, raised awareness of risks introduced by Health IT systems and called for the development of "new measures for reliably assessing the current state of Health IT safety and monitoring for improvements." Adelman and colleagues developed and validated the first Health IT Safety measure—the Wrong-Patient Retract-and-Reorder (RAR) Measure—endorsed by the National Quality Forum. The Wrong-Patient RAR Measure identifies orders placed for a patient that are retracted within 10 minutes, and then placed by the same provider for a different patient within the next 10 minutes. The measure identified over 5,000 near-miss, wrong-patient electronic orders at one hospital in a single year, which is more than 500 times the average number of errors previously identified by voluntary reporting. In this project, we will use the Retract-and-Reorder automated detection method to develop a set of valid and reliable measures to capture violations of the Five Rights of Medication Safety: right patient, right drug, right dose, right route, and right frequency. In Aim 1, we will develop and validate measures for detecting wrong-drug, wrong-dose, wrong-route, and wrong-frequency electronic orders using a novel methodology that enables us to conduct near real-time interviews with providers after an RAR event occurs. In Aim 2, we will implement the automated measures at a second hospital using a different EHR to evaluate the reliability of the measures. In Aim 3, we will conduct a multi-site observational study describing the overall frequency of these types of errors in different settings and systems. Developing and validating a full set of medication safety measures will provide a greater understanding of the epidemiology of these critical errors, enable regulatory bodies to conduct ongoing surveillance of health system performance, and allow researchers to test interventions aimed at preventing these errors and improving patient safety.

Role: Lead Site; Jason Adelman PI
Study Period: 9/1/16–6/30/22
AHRQ R01-HS024538

Using the Retract-and-Reorder automated detection method, we will develop a set of measures to capture five types of medication safety errors: wrong patient, wrong medication, wrong dose, wrong route, and wrong frequency.

Providing Evidence and Developing a Toolkit to Accelerate the Adoption of Patient Photographs in Electronic Health Records (AHRQ R01-HS024713)

To prevent wrong-patient errors, the Office of the National Coordinator for Health Information Technology (ONC) Patient Identification SAFER Guide recommends displaying patient photographs in EHRs, but the vast majority of healthcare systems have not adopted this safety practice. In a national survey of hospitals, respondents identified lack of evidence that photographs improve safety and workflow challenges as major barriers to adoption. Preliminary data suggest that displaying patient photographs in an EHR significantly decreases the frequency of wrong-patient orders. In this project, we are pursuing the following aims: Aim 1) provide rigorous evidence that patient photographs prevent wrong-patient order errors, using the Wrong-Patient Retract-and-Reorder Measure to identify the outcome; Aim 2) demonstrate generalizability by evaluating the effectiveness of patient photographs across three large health systems using three different EHR systems; and Aim 3) develop a Health IT Toolkit to guide healthcare organizations through the implementation process. Because implementation of patient photographs will likely vary across systems, this project leverages a collaboration among Columbia University, NewYork-Presbyterian (NYP) Hospital, NYP/Brooklyn Methodist Hospital, Johns Hopkins Medicine, and Montefiore Medical Center/Albert Einstein College of Medicine. Based on the functionality of the EHRs, we are conducting a randomized controlled trial in Allscripts and Epic, in which providers are randomized to view EHR screens with versus without patient photos, and we will conduct a pre- versus post-implementation study in Cerner. We are working with the ECRI Institute's Partnership for Health IT Patient Safety to develop the first Health IT Safety Toolkit, Implementing Patient Photographs in EHR Systems, to share lessons learned.

Role: Lead Site; Jason Adelman PI
Study Period: 9/30/2017–7/31/2023
AHRQ R01-HS024713

We are studying whether patient photographs prevent wrong-patient errors in electronic health record systems. Providers are randomly assigned to either see a patient photo or not see patient photos when placing electronic orders.

Effectiveness of Pictographs to Prevent Wrong-Patient Errors in the NICU (NICHD R01-HD094793)

Newborns in the neonatal intensive care unit (NICU) are at high risk for wrong-patient errors. A major contributing factor is the use of temporary, nondistinct first names (e.g., Babyboy/Babygirl) that are assigned to newborns at birth and remain unchanged throughout their hospital stay. Our research found that a distinct newborn naming convention that incorporates the mother’s first name (e.g., Wendysgirl) reduced the risk of wrong-patient orders in the NICU by 36%. However, the distinct naming convention conferred benefit only for singletons—multiples remained at high risk as a result of siblings sharing the same name distinguished by a single character (e.g., 1Wendysgirl, 2Wendysgirl). Displaying patient photographs in EHRs is a promising strategy to improve patient identification, but photographs are unlikely to be effective in the NICU because newborns lack distinguishing physical features. Instead, we are testing Pictographs as “photo equivalents” for newborns in the NICU, displayed at the bedside and in the EHR to serve as a visual cue when providers place orders. Pictographs consist of three elements: an image of an easy-to-remember object; the infant’s given name (when available); and a color-coded border indicating the infant’s sex. Parents select a distinctive Pictograph for their infants for the duration of their hospital stay, with no two infants having the same Pictograph at the same time in the same NICU. We will pursue the following aims: Aim 1) conduct a multi-site, randomized controlled trial to compare the frequency of wrong-patient orders in the NICU between providers assigned to view EHR screens with versus without Pictographs, as identified by the Wrong-Patient Retract-and-Reorder Measure; Aim 2) conduct subgroup analyses of the effectiveness of Pictographs for reducing the frequency of wrong-patient orders among siblings of multiple births; and Aim 3) conduct a qualitative evaluation to examine the perceptions and experience of Pictographs among providers and parents. If proven effective, Pictographs could be a landmark innovation that safeguards newborns in the NICU.

Role: Lead Site; Jason Adelman PI
Study Period: 4/1/2018–3/31/2023
NICHD R01-HD094793

We developed an innovative identifier for newborns called a Pictograph that includes a distinctive object, the baby’s name, and a color-coded border indicating the baby’s sex. Three Pictographs—a Lion, a Spaceship, and a Clover—distinguish triplets whose names in the electronic health record are very similar (1Anitasboy, 2Anitasboy, 3Anitasboy).

We are testing whether Pictographs will prevent wrong-patient errors among newborns, and specifically among multiple birth infants. Providers are randomly assigned to see a Pictograph or not see Pictographs when placing orders in the electronic health record.

Assessing the Risk of Wrong-Patient Errors in an EHR that Allows Multiple Records Open (AHRQ R21-HS023704)

Although patient safety experts recommend limiting providers to one patient record open at a time in EHRs, no evidence supports this recommendation. Therefore, we conducted the first randomized comparative effectiveness trial to assess the risk of wrong-patient orders in a "restricted” configuration that limited providers to open one patient record at a time compared to an "unrestricted” configuration that allowed providers to open up to four records at once. The study included more than 3,000 providers who placed more than 12,000,000 orders, for more than 500,000 patients. Using the Wrong-Patient Retract-and-Reorder Measure, we found that restricting providers to one record open did not decrease the risk of wrong-patient orders, overall and in the emergency department, inpatient, and outpatient settings. In a survey of randomized providers, satisfaction and usability overwhelmingly favored the unrestricted arm. Data collection for this study is complete; however, secondary analysis and manuscript development are ongoing.

Lead Site; Jason Adelman PI
Period: Completed, data analysis ongoing
AHRQ R21-HS023704

Providers were randomly assigned to open 1 patient record (Restricted arm) or up to 4 records (Unrestricted arm) at a time.

Columbia University Patient Safety Research Fellowship (AHRQ T32-HS026121)

The aim of this postdoctoral research fellowship is to support promising clinician-researchers to pursue academic careers and become future leaders in patient safety and health services research. The program is designed to prepare fellows to conduct innovative interdisciplinary research, employ rigorous and varied research methodologies, present and publish study findings, and successfully compete for extramural peer-reviewed research funding. The fellowship offers a unique combination of formal research education, mentored research projects, and exposure to patient safety operations at a large academic medical center. Fellows will have the opportunity to work with a diverse and accomplished group of Columbia faculty with a proven track record of grant funding, interdisciplinary research collaboration, publication, and mentorship. Our Faculty Mentors represent a broad range of clinical and academic disciplines, and have expertise and active research support in focus areas including medical errors, medication safety, healthcare-associated infections, health informatics, quality measurement and outcomes, cost and cost-effectiveness, chronic disease epidemiology, and health disparities. This 2-year training program consists of six core components: 1) Formal Research Education; 2) Mentored Research Projects; 3) Patient Safety Immersion; 4) Bi-Weekly Research Seminars; 5) Grant Proposal Development; and 6) Career Development. Fellows will earn a Master of Science degree in a research-oriented field such as Epidemiology, Clinical Research Methods, or Patient-Oriented Research at Columbia University’s Mailman School of Public Health. Fellows who already have a related Master's degree will have the opportunity to take advanced courses in quantitative and qualitative research methods or other relevant coursework, including informatics. In addition, a distinctive core component of the program is Patient Safety Immersion, designed to provide Fellows with direct experience in patient safety operations at a large academic medical center. Fellows will attend root cause analyses and review adverse event reports at NYP/CUIMC, develop and conduct a fellow-initiated research project based on real-world patient safety hazards, and participate in a patient safety curriculum.

See Patient Safety Research Fellowship for more information.

Role: Multiple PIs/Co-Directors; Jason Adelman, R. Graham Barr
Study Period: 7/1/2018–6/30/2023; Renewal 7/01/2023–6/30/2028
AHRQ T32-HS026121

Pilot Awards

Develop and Validate Novel Automated Measures to Detect Wrong-Imaging Order Errors for Use in Large-Scale Intervention Trials  | 2022 CUIMC Clinical Trialist Early Career Development Scholars Award 

Incorrectly ordered imaging tests are a major cause of missed diagnoses but little is known about why these errors occur. Current methods measuring imaging order errors are limited by reporting bias and the need for patient chart review. This project proposes applying an innovative, systematic approach, the Retract-and-Reorder method, to develop a novel automated and reliable way to identify imaging order errors.

PI: Jerard Z. Kneifati-Hayek, MD, MS
Study Period: 01/2023–12/2024 

A Pragmatic, Multi-Center, Factorial, Randomized Controlled Trial of Electronic Sepsis Alerts for Nurses and Providers | 2022 CUIMC Clinical Trialist Early Career Development Scholars Award 

Sepsis is the most common cause of death of hospitalized patients in the United States. The cornerstone of sepsis therapy is early identification and treatment. International guidelines recommend all hospitals screen patients for sepsis to aid in rapid identification and treatment. Systemic inflammatory response syndrome (SIRS) criteria, or related tools, are recommended by guidelines for screening. However, the optimal method of alerting nurses and providers is unknown. A common method of sepsis screening and alerting involves interruptive “pop-up” alerts in the electronic health record (EHR). Although pre-post studies exist, there has never been a randomized evaluation of a SIRS-based interruptive EHR alert. Furthermore, the optimal alerting strategy is unknown. This proposal is for a pragmatic, multi-hospital, factorial, randomized controlled study examining the effectiveness of SIRS based interruptive EHR alerts. Patients will be randomized to one of four groups: no alert, alert to nurses only, alert to providers only, and alerts to both nurses and providers. Effective alerting has the potential to improve sepsis care and decrease mortality. Ineffective alerting can contribute to alert fatigue, making other alerts in the health system less effective. This study has the potential to improve care for hospitalized patients both locally and around the world.

Granting Institution: Herbert Irving Comprehensive Cancer Center and the Irving Institute for Clinical and Translational Research at Columbia University Irving Medical Center 
Note: The award is supported by NIH/NCI Cancer Center Support Grant P30CA013696 and by the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant Number UL1TR001873

PI: Benjamin L. Ranard, MD, MSHP
Study Period: 01/2023–12/2024 

Development and Implementation of a Novel Machine Learning Algorithm for Early Detection of Sepsis in Hospitalized Patients |  Learning Health System Pilot Award

Granting Institution: Irving Institute for Clinical and Translational Research, ColumbiaDoctors, and the Columbia University Fu Foundation School of Engineering and Applied Science (SEAS) 

PI: Carri W. Chan, PhD, MS
Co-investigator: Benjamin L. Ranard, MD, MSHP
Study Period: 07/2022–7/2024 

Reducing Medication Ordering Errors Through Indications-Based Prescribing | The Doctors Company Foundation

Despite widespread adoption of Computerized Provider Order Entry (CPOE) systems, its impact on reducing medication ordering errors has been limited. The majority of clinical decision support (CDS) tools are designed to identify when an error occurs, and then alert providers, rather than preventing the initial error. A more proactive approach to medication safety would guide users to ordering the appropriate medication and its parameters at the time of order entry. An evaluation of an indication-based ordering prototype demonstrated a significant decrease in ordering errors when compared with standard order entry. Despite this success, indication-based prescribing has not been rigorously evaluated in a real-world setting. The goal of our study is to evaluate a robust indications-based ordering system aimed at decreasing antimicrobial ordering errors in the inpatient setting. The indication-based antimicrobial decision support tool utilizes indication and patient characteristics to guide prescribers to the correct antimicrobial agent, dose, and frequency at time of order entry. To evaluate our intervention, we will randomize providers to standard order entry versus indication-based ordering. If successful, our results can be extrapolated to develop additional indications-based order sets in both the inpatient and  outpatient environments.

PI: Jason Adelman, MD, MS
Co-PI: Anne Grauer, MD, MS
Study Period: Ongoing, 10/2023–10/2025

Collaborating Site

Preventing Wrong-Drug and Wrong-Patient Errors with Indication Alerts in CPOE Systems (AHRQ R01-HS024945)

A core principle of medication safety is making sure the right patient gets the right drug, yet wrong-drug and wrong-patient errors occur at a rate of about 1 per 1000 orders in inpatient and outpatient settings, resulting in millions of errors annually in the US. Indication-based prescribing has the potential to reduce these errors, and improve the completeness of patients’ problem lists. This project develops, validates, and replicates indication alerts in two health systems that utilize two different EHR systems. Indication alerts are triggered when a provider orders a medication indicated for a condition that is not in the patient’s problem list. Medications targeted for alerts are ones that are 1) prone to look-alike and sound-alike errors; 2) commonly prescribed; and 3) have a narrow indication (eg, metformin for diabetes). We will implement a set of 30–50 indication alerts developed by the University of Illinois at Chicago and NewYork-Presbyterian. Using a time series design (pre- versus post-implementation), we will study the effect of indication alerts on the rate of near-miss, wrong-drug and wrong-patient order errors. The study will also assess the impact of indication alerts on the likelihood of adding new diagnoses to the problem list.

Role: Collaborating Site; Bruce Lambert PI; Jason Adelman Site Lead
Study Period: 10/1/2016–9/30/2022
AHRQ R01-HS024945

Generalizability and Spread of an Evidenced-Based Fall Prevention Toolkit: Fall TIPS (AHRQ R18-HS024945)

In the U.S., 1 million hospitalized patients fall annually and approximately 30% of falls result in injury. The Fall TIPS (Tailoring Interventions for Patient Safety) toolkit is an evidence-based fall prevention program integrated into hospital-based EHR systems. Fall TIPS significantly reduced falls in a randomized trial, particularly among older patients who are at greatest risk. In this project, the Fall TIPS program is being implemented in 15 hospitals across three large healthcare systems using different EHR systems in diverse patient populations: Partners, Montefiore, and Columbia/NewYork-Presbyterian. We are using the RE-AIM framework (Reach, Effectiveness, Adoption, Implementation, Maintenance) to conduct an initiation, process, and outcome evaluation. Outcomes from this dissemination study will increase understanding of the overall impact of the Fall TIPS program on providers and hospitalized patients. The study will both advance the development of Fall TIPS as a high-quality, turnkey evidence-based program and obtain data on factors that facilitate the adoption, implementation, and maintenance of hospital-based fall prevention programs. The ultimate objective is to spread best practices related to implementing and sustaining an evidence-based fall prevention program that prevents unintentional injuries and reduces their consequences.

Role: Collaborating Site; Patricia Dykes PI; Jason Adelman Site Lead
Study Period: 4/1/2017–1/31/2020
AHRQ R18-HS024945

Ensuring Safe Performance of Electronic Health Records (AHRQ R01-HS023696)

EHRs are being deployed rapidly across the U.S., but it is uncertain to what extent they will improve safety and quality of patient care. To address this issue, Brigham and Women’s Hospital built the Leapfrog tool, which evaluates the safety performance of EHRs after deployment, particularly focusing on high-impact patient safety and medication safety issues in the inpatient setting. The Leapfrog tool is essentially a "flight simulator" for EHRs with computerized provider order entry (CPOE). Hospitals download simulated patients, attempt to enter simulated predefined groups of orders, and record whether critical decision support appears in response to these scenarios. After taking the test, hospitals get immediate feedback on their performance in high-impact high-prevalence safety areas, including a subset of potentially fatal orders in the test. In this project, the Leapfrog CPOE/EHR will be tested and further refined to cover additional high-impact clinical safety domains based on the experience of four hospitals using four different EHR systems, including Columbia/NewYork-Presbyterian. The updated test will be released for national use through the Leapfrog Group.

Role: Collaborating Site; David Bates PI; Jason Adelman Site Lead
Study Period: 9/1/2018–8/31/2019
AHRQ R01-HS023696

Completed Grants

Making Acute Care More Patient Centered (AHRQ P30-HS023535)

The Patient Safety Learning Laboratory (PSLL) is a collaboration led by David Bates, MD, MSc at the Center for Patient Safety, Research and Practice at Brigham and Women's Hospital and James Benneyan, PhD at the Healthcare Systems Engineering Institute at Northeastern University. The PSLL develops tools to engage patients, family, and professional care team members in reliable identification, assessment, and reduction of patient safety threats in real-time, before they cause harm. The PSLL developed systems approaches to integrating health information technology (IT), stakeholder engagement mechanisms, and process design/engineering methods focused on three core projects: Project 1) Patient-centered Fall Prevention Toolkit; Project 2) Patient Safety Checklist Tool; and Project 3) MySafeCare Patient Safety Reporting System. As a result of increasing implementation and use of health IT and patient/family engagement in their plan of care, the PSLL will provide information, strategies, and tools for utilizing health IT to facilitate patient activation in eliminating harm in hospital settings. At NYP, we implemented the Fall TIPS program in selected pilot units and evaluated the perceptions of patients and staff about the program and the impact of the intervention on reducing falls.

Role: Collaborating Site; David Bates PI
Study Period: 9/30/2014–9/29/2018
AHRQ P30-HS023535

Project RedDE: Reducing Diagnostic Errors in Primary Care Pediatrics (AHRQ R01-HS023608)

Limited research on pediatric diagnostic errors highlights the significance of the problem: 54% of pediatricians report making a diagnostic error at least monthly and 45% report making a diagnostic error that harms patients at least annually. This project was a multisite, prospective, cluster randomized trial testing a quality improvement collaborative intervention within the American Academy of Pediatrics Quality Improvement Innovation Networks (QuIIN). The goal was to reduce the incidence of three pediatric primary care diagnostic errors: missed diagnosis of adolescent depression, missed diagnosis of pediatric elevated blood pressure, and delayed diagnosis of actionable laboratory results. The study tested whether a quality improvement collaborative, consisting of evidence-based best-practice methodologies, mini-root cause analyses, data sharing, and behavior change techniques, reduced diagnostic error rates in a national group of pediatric primary care practices. This project will enhance the understanding of ambulatory pediatric diagnostic errors and serve as a foundation for projects aimed at reductions of pediatric diagnostic errors across settings and diagnoses.

Role: Collaborating Site; Michael Rinke PI
Study Period: 9/30/2014–9/29/2018
AHRQ R01-HS023608