About

Nikhil K. Mull, MD, is a Professor of Clinical Medicine in the Department of Medicine at the Perelman School of Medicine at the University of Pennsylvania. He serves as the Medical Director of the Penn Medicine Center for Evidence-based Practice (CEP). His clinical expertise focuses on the utilization, education, and dissemination of evidence-based clinical medicine. Dr. Mull's work involves improving healthcare practices through systematic reviews and research on healthcare algorithms, racial and ethnic disparities in health and healthcare, and patient safety interventions. His contributions include leading research on machine learning prognostic models, healthcare worker implicit bias training, and automated interventions to reduce inappropriate inpatient ordering, among others.

Research topics

• Physics
• Mechanical engineering
• Engineering
• Oceanography
• Geology
• Environmental science

Selected publications

• An Emergency Department Nudge-Based Strategy to Screen and Treat Patients With Alcohol Misuse

  Annals of Emergency Medicine · 2026-05-01

  articleOpen access

  STUDY OBJECTIVE: To examine the effectiveness and implementation of a multicomponent strategy to increase emergency department (ED) initiation of evidence-based treatment for patients with alcohol misuse. METHODS: Setting was an academic health system with 6 hospitals. Interventions occurred in the ED at 4 of the hospitals, with the other 2 serving as controls. We launched interventions in 2 phases: (1) ED discharge order set with clinical decision support (May 20, 2024); (2) screening for patient concerns about alcohol use and facilitating treatment conversations (August 21, 2024). Multivariate logistic regression assessed changes relative to baseline in the primary outcome, proportion of ED patients with an alcohol-related discharge diagnosis who were discharged with a naltrexone prescription. A difference-in-difference analysis compared intervention hospitals to controls. RESULTS: Across the 43-month study period, there were 8,909 (2.0%) ED patients discharged with an alcohol-related diagnosis code. At intervention hospitals, there were 13 (0.2%) discharged with a naltrexone prescription at baseline, 18 (2.7%) during phase 1, and 81 (3.2%) during phase 2. At control hospitals, the rate of naltrexone prescribing was flat across these periods (0.0%, 0.0%, and 0.3%, respectively). In the multivariate model, patients with alcohol-related diagnoses at intervention hospitals were more likely to be discharged with naltrexone in phase 1 (odds ratio [confidence interval] = 12.3 [6.0 to 25.7]) and phase 2 (14.6 [8.4 to 27.4]) compared to baseline. The difference-in-difference analysis showed a 2.9% [2.4% to 3.5%] greater absolute increase in naltrexone prescribing among intervention hospitals. CONCLUSION: A triage-based ED protocol that integrated universal screening, electronic health record banners, and clinical decision support increased initiation of naltrexone to treat alcohol misuse and alcohol use disorder.

  PublisherDOI

• 649 DEVELOPMENT AND VALIDATION OF A PROGNOSTIC MODEL TO PREDICT LENGTH OF STAY IN ACUTE PANCREATITIS UTILIZING TWO LARGE PROSPECTIVE COHORTS

  Gastroenterology · 2026-05-01

  article
  PublisherDOI

• P-1123. Optimizing use of Cefazolin for Surgical Prophylaxis in Patients with Penicillin Allergy Labels

  Open Forum Infectious Diseases · 2026-01-01

  articleOpen access

  Abstract Background Cefazolin (CFZ), a common antibiotic used in surgical prophylaxis, is not always utilized when appropriate in patients with a penicillin (PCN) allergy due to concerns regarding potential cross-reactivity. Use of alternative antibiotics has been associated with increased risk of surgical site infections and delays in start of surgery due to longer infusions.Figure 1:WorkflowFigure 2:Antimicrobial prophylaxis recommendations Methods Our team completed a multi-step evaluation including contextual inquiry to identify potential opportunities for optimization within the current workflow and collaborated with key stakeholders to identify areas for intervention.Table 1:Cefazolin prescribing rates in individuals with documented penicillin allergies Results Contextual inquiry demonstrated several key barriers including various workflows for ordering of perioperative antibiotics, identification of a CPOE alert regarding potential of cross-allergenicity, and lack of clarity for who is ultimately responsible for decision as to which antibiotic to use. As an initial intervention, the Antimicrobial Stewardship team collaborated with orthopedics and allergy clinicians to provide guidance on risk of cross-reactivity with CFZ/common PCN-based antibiotics and the importance of updating the medical record with accurate allergy information. This educational initiative demonstrated increased CFZ prescribing which was sustained for the following two quarters, however, still left room for optimization. Several strategies were developed and proposed to key stakeholders including integrated pathways detailing workflow and antibiotic recommendations (figures 1 and 2). After discussion, the decision was made to continue with utilization of the current order set with optimized verbiage to nudge providers to use CFZ in individuals with documented PCN allergies. In addition, individual provider ordering summaries were distributed to the individual orthopedic and anesthesiology providers detailing the rate of CFZ prescribing in these individuals as compared to their top quartile peers. Prescribing rates of CFZ increased from a baseline of 63% to 75% after optimization of the order sets and 80% post distribution of the provider summaries (table 1). Conclusion This collaboration demonstrated the importance of engaging end users in strategy meetings as well optimizing the verbiage in CPOE systems to support recommendations. Disclosures Amanda Binkley, PharmD, BCIDP, AAHIVP, Shionogi: Advisor/Consultant

  PublisherOA PDFDOI

• 649 DEVELOPMENT AND VALIDATION OF A PROGNOSTIC MODEL TO PREDICT LENGTH OF STAY IN ACUTE PANCREATITIS UTILIZING TWO LARGE PROSPECTIVE COHORTS

  Gastrointestinal Endoscopy · 2026-05-01

  article
  PublisherDOI

• Su1536: DISEASE-RELATED PARAMETERS DO NOT EXPLAIN HALF THE STORY A PROGNOSTIC MODEL TO PREDICT LENGTH OF STAY IN ACUTE PANCREATITIS PERFORMS POORLY

  Gastroenterology · 2025-05-01

  article
  PublisherDOI

• Response to: Disagreement with Categorization of Clinical Trials included in Treadwell et al. Systematic Review and Meta‐Analysis

  Headache The Journal of Head and Face Pain · 2025-07-25

  letter
  PublisherDOI

• Patient Hand Hygiene Before Meals

  Journal of Nursing Care Quality · 2025-04-01 · 2 citations

  reviewOpen access

  BACKGROUND: Hand hygiene is recognized as an effective way to prevent health care-associated infections (HAIs). However, there is limited attention to patient hand hygiene (PHH). PURPOSE: The purpose of this systematic review was to summarize evidence, interventions, and outcomes of PHH before meals. METHODS: Literature was searched from 1999 to 2024 in 4 databases. The Grading of Recommendations, Assessment, Development, and Evaluation was used to appraise the strength of evidence. RESULTS: Ten reports were included in the review. Five categories of PHH interventions were identified: direct observation, reminders, education, policy change, and bundles of more than one intervention. There is insufficient evidence to establish a direct causal link between PHH before meals and a reduction in HAIs. CONCLUSIONS: The limited and moderate level of evidence highlights a significant gap in understanding PHH. Hand hygiene is a fundamental infection prevention strategy that warrants additional research in hospitalized patient populations to determine the clinical efficacy and causal effects on HAIs.

  PublisherOA PDFDOI

• Examining the Impact of Implementing High-Reliability Organization Principles on Patient Safety Outcomes

  2025-05-07 · 1 citations

  reportSenior author

  Objectives. High-reliability organizations (HROs) operate in complex, high-hazard domains for extended periods without serious accidents or catastrophic failures. Interventions are designed to change thinking about patient safety and system performance through distinct HRO principles. The purpose of this review was to determine the effectiveness of implementing HRO principles on patient safety outcomes. Methods. We followed rapid review processes of the Agency for Healthcare Research and Quality Evidence-based Practice Center Program. We searched PubMed and the Cochrane Library from January 2019 to May 2024, supplemented by a narrowly focused search for unpublished reports. We included any comparative study that evaluated HRO implementation in healthcare organizations using a multicomponent framework. Findings. One rapid evidence review and one pre-post primary study were included. The 2022 review summarized and updated the results of the Department of Veterans Affairs (VA) 2019 evidence brief on implementation of HRO principles (N=23 studies). The authors found that multicomponent HRO interventions delivered for at least two years were associated with improved patient safety outcomes, but the overall strength of evidence was low. Although the updated search identified two additional effectiveness studies, the authors did not report any new insights or conclusions. The new pre-post study identified in our review examined a single VA medical center (Truman). At Truman, the National Center for Patient Safety implemented a comprehensive high-reliability hospital (HRH) model based on HRO principles during a 3-year period. HRH implementation positively impacted patient safety culture, patient safety event reporting (particularly low-harm events), 30-day standardized mortality rate (SMR), and complication rate (CR). However, six months post-intervention, both the SMR and CR were observed to increase. The study was judged to be at serious risk of bias. No strength of evidence assessment was conducted owing to the limited studies included. Conclusions. While some data suggests that sustained implementation of HRO principles in organizations may improve patient safety outcomes, the published evidence is limited, making it challenging to draw conclusions about the effectiveness of specific interventions.

  PublisherDOI

• A systematic review of machine learning-based prognostic models for acute pancreatitis: Towards improving methods and reporting quality

  PLoS Medicine · 2025-02-24 · 9 citations

  reviewOpen accessCorresponding

  BACKGROUND: An accurate prognostic tool is essential to aid clinical decision-making (e.g., patient triage) and to advance personalized medicine. However, such a prognostic tool is lacking for acute pancreatitis (AP). Increasingly machine learning (ML) techniques are being used to develop high-performing prognostic models in AP. However, methodologic and reporting quality has received little attention. High-quality reporting and study methodology are critical for model validity, reproducibility, and clinical implementation. In collaboration with content experts in ML methodology, we performed a systematic review critically appraising the quality of methodology and reporting of recently published ML AP prognostic models. METHODS/FINDINGS: Using a validated search strategy, we identified ML AP studies from the databases MEDLINE and EMBASE published between January 2021 and December 2023. We also searched pre-print servers medRxiv, bioRxiv, and arXiv for pre-prints registered between January 2021 and December 2023. Eligibility criteria included all retrospective or prospective studies that developed or validated new or existing ML models in patients with AP that predicted an outcome following an episode of AP. Meta-analysis was considered if there was homogeneity in the study design and in the type of outcome predicted. For risk of bias (ROB) assessment, we used the Prediction Model Risk of Bias Assessment Tool. Quality of reporting was assessed using the Transparent Reporting of a Multivariable Prediction Model of Individual Prognosis or Diagnosis-Artificial Intelligence (TRIPOD+AI) statement that defines standards for 27 items that should be reported in publications using ML prognostic models. The search strategy identified 6,480 publications of which 30 met the eligibility criteria. Studies originated from China (22), the United States (4), and other (4). All 30 studies developed a new ML model and none sought to validate an existing ML model, producing a total of 39 new ML models. AP severity (23/39) or mortality (6/39) were the most common outcomes predicted. The mean area under the curve for all models and endpoints was 0.91 (SD 0.08). The ROB was high for at least one domain in all 39 models, particularly for the analysis domain (37/39 models). Steps were not taken to minimize over-optimistic model performance in 27/39 models. Due to heterogeneity in the study design and in how the outcomes were defined and determined, meta-analysis was not performed. Studies reported on only 15/27 items from TRIPOD+AI standards, with only 7/30 justifying sample size and 13/30 assessing data quality. Other reporting deficiencies included omissions regarding human-AI interaction (28/30), handling low-quality or incomplete data in practice (27/30), sharing analytical codes (25/30), study protocols (25/30), and reporting source data (19/30). CONCLUSIONS: There are significant deficiencies in the methodology and reporting of recently published ML based prognostic models in AP patients. These undermine the validity, reproducibility, and implementation of these prognostic models despite their promise of superior predictive accuracy. REGISTRATION: Research Registry (reviewregistry1727).

  PublisherOA PDFDOI

• Behavioral interventions for migraine prevention: A systematic review and meta‐analysis

  Headache The Journal of Head and Face Pain · 2025-02-19 · 21 citations

  reviewOpen accessSenior author

  OBJECTIVES/BACKGROUND: This study was undertaken to synthesize evidence on the benefits and harms of behavioral interventions for migraine prevention in children and adults. The efficacy and safety of behavioral interventions for migraine prevention have not been tested in recent systematic reviews. METHODS: An expert panel including clinical psychologists, neurologists, primary care physicians, researchers, funders, individuals with migraine, and their caregivers informed the scope and methods. We searched MEDLINE, Embase, PsycINFO, PubMed, the Cochrane Database of Systematic Reviews, clinicaltrials.gov, and gray literature for English-language randomized trials (January 1, 1975 to August 24, 2023) of behavioral interventions for preventing migraine attacks. Primary outcomes were migraine/headache frequency, migraine disability, and migraine-related quality of life. One reviewer extracted data and rated the risk of bias, and a second verified data for completeness and accuracy. Data were synthesized with meta-analysis when deemed appropriate, and we rated the strength of evidence (SOE) using established methods. RESULTS: For adults, we included 50 trials (77 publications, N = 6024 adults). Most interventions were multicomponent (e.g., cognitive behavioral therapy [CBT], biofeedback, relaxation training, mindfulness-based therapies, and/or education). Most trials were at high risk of bias, primarily due to possible measurement bias and incomplete data. For adults, we found that any of three components (CBT, relaxation training, mindfulness-based therapies) may reduce migraine/headache attack frequency (SOE: low). Education alone that targets behavior may improve migraine-related disability (SOE: low). For three other interventions (biofeedback, acceptance and commitment therapy, and hypnotherapy), evidence was insufficient to permit conclusions. We also found that mindfulness-based therapies may reduce migraine disability more than education, and relaxation + education may improve migraine-related quality of life more than propranolol (SOE: low). For children/adolescents, we included 13 trials (16 publications, N = 1444 children), but the evidence was only sufficient to conclude that CBT + biofeedback + relaxation training may reduce migraine attack frequency and disability more than education alone (SOE: low). CONCLUSION: Results suggest that for adults, CBT, relaxation training, and mindfulness-based therapies may each reduce the frequency of migraine/headache attacks, and education alone may reduce disability. For children/adolescents, CBT + biofeedback + relaxation training may reduce migraine attack frequency and disability more than education alone. Evidence consisted primarily of underpowered trials of multicomponent interventions compared with various types of control groups. Limitations include semantic inconsistencies in the literature since 1975, differential usage of treatment components, expectation effects for subjectively reported outcomes, incomplete data, and unclear dosing effects. Future research should enroll children and adolescents, standardize intervention components when possible to improve reproducibility, consider smart study designs and personalized therapies based on individual characteristics, use comparison groups that control for expectation, which is a known challenge in behavioral trials, enroll and retain larger samples, study emerging digital and telehealth modes of care delivery, improve the completeness of data collection, and establish or update clinical trial conduct and reporting guidelines that are appropriate for the conduct of studies of behavioral therapies.

  PublisherOA PDFDOI

Frequent coauthors

• Emilia Flores

  University of Pennsylvania Health System

  40 shared

• Laurel Glaser

  27 shared

• Mika Epps

  University of Pennsylvania Health System

  27 shared

• Kathleen Hopkins

  27 shared

• David A. Pegues

  University of Pennsylvania

  27 shared

• Matthew J. Ziegler

  27 shared

• Bikas Nag

  Desun Hospital & Heart Institute

  25 shared

• Christina Bennett

  Oregon Medical Research Center

  25 shared

Labs

• Nikhil K. Mull LabPI