About

Justin E. Bekelman, M.D., is the Marietta and Howard Stoeckel Professor of Radiation Oncology, Medicine, and Medical Ethics and Health Policy at the Perelman School of Medicine at the University of Pennsylvania. He is a member of the Abramson Cancer Center and a Senior Fellow at the Leonard Davis Institute for Health Economics. Dr. Bekelman is the founding Director of the Penn Center for Cancer Care Innovation at the Abramson Cancer Center. His research focuses on improving the length and quality of life for patients with cancer by studying healthcare delivery, payment reform, and cancer care transformation. His work integrates methods from innovation, epidemiology, clinical trials, health economics, and public policy, and has been published in high-impact scientific journals and featured internationally. Dr. Bekelman has received funding from prominent organizations such as the National Cancer Institute, the American Cancer Society, and the Patient-Centered Outcomes Research Institute, and has served as an advisor to the Centers for Medicare and Medicaid Services on cancer care payment reform and innovation.

Research topics

• Medicine
• Internal medicine
• Family medicine
• Emergency medicine
• Gynecology
• Oncology
• Nursing
• Intensive care medicine

Selected publications

• Results of integrating an automated electronic social determinants of health screening survey in a gynecologic oncology practice: a pilot study

  International Journal of Gynecological Cancer · 2026-02-25

  article
  PublisherDOI

• Defining the role of pragmatic clinical trials in cancer clinical research: outcomes of a collaborative workshop hosted by the European Organisation for Research and Treatment of Cancer

  The Lancet Oncology · 2025-05-01 · 6 citations

  reviewOpen access
  PublisherOA PDFDOI

• Gynecologic oncology referral rates of adnexal masses suspicious for ovarian cancer in an academic health system: A cohort study.

  Journal of Clinical Oncology · 2025-05-28

  article

  5544 Background: Ovarian-Adnexal Reporting Data System (O-RADS) is an international lexicon and risk stratification tool. O-RADS 4 or 5 lesions are complex adnexal masses that a 10-90% risk of malignancy, and national guidelines recommend gynecologic oncology referral. Our objective was to examine patient, clinician, and imaging factors associated with referral to gynecologic oncology for complex adnexal masses. Methods: This retrospective cohort study was exempt from IRB review. We identified all patients with O-RADS 4 or 5 lesions on ultrasound (US) or MRI from July 1, 2020 to December 31, 2023. Our primary outcome was referral to gynecologic oncology. We gathered patient demographic data and ordering clinician characteristics from electronic health records. We performed descriptive statistics and multivariate logistic regression of patient demographics and ordering clinician characteristics associated with gynecologic oncology referral. Results: Our cohort included 373 patients with O-RADS 4 or 5 lesions and no prior gynecologic oncology care. The referral rate to gynecologic oncology was 68%, and referral within 30 days of abnormal imaging was 43%. Time from abnormal imaging to referral ranged from 0 to 407 days (mean 15.3, median 4 days). In multivariate analyses, the likelihood of referral to gynecologic oncology was higher among patients with repeat abnormal imaging compared to those with single instance of abnormal imaging (aOR 20.61, 95%CI 2.63-161.6), O-RADS 5 lesions compared to O-RADS 4 lesions (aOR 9.15, 95%CI 3.47-24.85) and detection on MRI compared to US (aOR 7.79, 95%CI 1.57-38.65). The likelihood of referral to gynecologic oncology was lower among non-white patients (aOR 0.24, 95%CI 0.08-0.76). There were no differences by Hispanic ethnicity, rurality, insurance, or language. Referral was higher among patients whose imaging was ordered by an internal medicine clinician (aOR 3.89, 95%CI 1.48-10.20) compared to ob/gyn. Conclusions: One-third of patients with complex adnexal masses were not referred to gynecologic oncology. Disparities in referral to gynecologic oncology for complex adnexal masses rates based on patient race and ordering clinician specialty highlight the need for system-based approaches including clinician education or automated referrals. Gynecologic oncology referral after O-RADS 4/5. Multivariate OR (95%CI) Postmenopausal (≥55 years) 1.89 (0.95-3.74) Race - White Reference - Black 0.57 (0.27-1.21) - Asian 1.03 (0.30-3.58) - Some other race 0.24 (0.08-0.76) Ordering specialty - Obstetrics/Gynecology Reference - Emergency Medicine 0.93 (0.33-2.62) - Internal Medicine 3.89 (1.48-10.20) - Family Medicine 1.62 (0.66-3.98) - Other specialty 0.87 (0.27-2.76) Has PCP 1.66 (0.81-3.42) O-RADS - 4 Reference - 5 9.15 (3.27-24.85) Imaging - MRI 7.79 (1.57-38.65) - US Reference Repeat abnormal imaging 20.61 (2.63-161.79)

  PublisherDOI

• Algorithm-Based Palliative Care in Patients With Cancer

  JAMA Network Open · 2025-02-21 · 11 citations

  articleOpen access

  Importance: Among patients with advanced solid malignant tumors, early specialty palliative care (PC) is guideline recommended, but strategies to increase PC access and effectiveness in community oncology are lacking. Objective: To test whether algorithm-based defaults with opting out and accountable justification embedded in the electronic health record (EHR) increase completed PC visits. Design, Setting, and Participants: This 2-arm cluster randomized clinical trial was conducted from November 1, 2022, to December 31, 2023. Eligible patients from 15 urban or rural clinics within a large community oncology network in Tennessee had advanced lung or noncolorectal gastrointestinal cancer and were identified by an automated EHR algorithm adapted from national guidelines. Data were analyzed between November 1, 2023, and March 4, 2024. Intervention: At sites randomized to control, clinicians received weekly reports detailing PC referral rates compared with peer clinicians (peer comparison) and referred patients to PC at their discretion. At sites randomized to intervention, clinicians also received default PC orders using the EHR. Clinicians who opted out of PC consultation were asked to provide justification (accountable justification). If clinicians did not opt out, a study coordinator contacted patients to introduce and schedule PC visits using a standardized, predefined script. Main Outcomes and Measures: The primary outcome was a completed PC consultation within 12 weeks of enrollment. Exploratory outcomes included quality of life, feeling heard and understood, and intensive end-of-life care. Outcomes were analyzed using clustered generalized linear and logistic regression models. Results: The trial enrolled 562 patients (mean [SD] age, 68.5 [10.1] years; 288 male [51.2%]), of whom 433 (77.0%) had lung cancer. There were 130 of 296 patients (43.9%) randomized to the intervention group and 22 of 266 (8.3%) randomized to the control group who completed PC visits (adjusted odds ratio, 8.9 [95% CI, 5.5-14.6]; P < .001). Among 179 patients who died at the 24-week follow-up, 6 of 92 (6.5%) in the intervention group compared with 14 of 87 (16.1%) in the control group received systemic therapy within 14 days of death (adjusted odds ratio, 0.3 [95% CI, 0.1-0.7]; P = .05). There were no differences in quality of life, feeling heard and understood, or late hospice referral. Conclusions and Relevance: In this randomized clinical trial of algorithm-based EHR defaults, the intervention increased PC consultations and decreased end-of-life systemic therapy. The intervention provides a scalable implementation strategy to increase specialty PC referrals in the community oncology setting. Trial Registration: ClinicalTrials.gov Identifier: NCT05590962.

  PublisherOA PDFDOI

• The Impact of Machine Learning Mortality Risk Prediction on Clinician Prognostic Accuracy and Decision Support: A Randomized Vignette Study

  Medical Decision Making · 2025-07-04

  articleOpen access

  Background Machine learning (ML) algorithms may improve the prognosis for serious illnesses such as cancer, identifying patients who may benefit from earlier palliative care (PC) or advance care planning (ACP). We evaluated the impact of various presentation strategies of a hypothetical ML algorithm on clinician prognostic accuracy and decision making. Methods This was a randomized clinical vignette survey study among medical oncologists who treat metastatic non-small-cell lung cancer (mNSCLC). Between March and June 2023, clinicians were shown 3 vignettes of patients presenting with mNSCLC. The vignettes varied by prognostic risk, as defined from the Lung Cancer Prognostic Index (LCPI). Clinicians estimated life expectancy in months and made recommendations about PC and ACP. Clinicians were then shown the same vignette with a hypothetical survival estimate from a black-box ML algorithm; clinicians were randomized to receive the ML prediction using absolute and/or reference-dependent prognostic estimates. The primary outcome was prognostic accuracy relative to the LCPI. Results Among 51 clinicians with complete responses, the median years in practice was 7 (interquartile range 3.5–19), 14 (27.5%) were female, 23 (45.1%) practiced in a community oncology setting, and baseline accuracy was 54.9% (95% confidence interval [CI] 47.0–62.8) across all vignettes. ML presentation improved accuracy (mean change relative to baseline 20.9%, 95% CI 13.9–27.9, P &lt; 0.001). ML outputs using an absolute presentation strategy alone (mean change 27.4%, 95% 16.8–38.1, P &lt; 0.001) or with reference dependence (mean change 33.4%, 95% 23.9–42.8, P &lt; 0.001) improved accuracy, but reference dependence alone did not (mean change 2.0% [95% CI −11.1 to 15.0], P = 0.77). ML presentation did not change the rates of recommending ACP nor PC referral (mean change 1.3% and 0.7%, respectively). Limitations The singular use case of prognosis in mNSCLC, low initial response rate. Conclusions ML-based assessments may improve prognostic accuracy but not result in changed decision making. Implications ML prognostic algorithms prioritizing explainability and absolute prognoses may have greater impact on clinician decision making. Trial Registration: CT.gov: NCT06463977 Highlights While machine learning (ML) algorithms may accurately predict mortality, the impact of prognostic ML on clinicians’ prognostic accuracy and decision making and optimal presentation strategies for ML outputs are unclear. In this multicenter randomized survey study among vignettes of patients with advanced cancer, prognostic accuracy improved by 20.9% when clinicians reviewed vignettes with a hypothetical ML mortality risk prediction, with absolute risk presentation strategies resulting in greater accuracy gains than reference-dependent presentations alone. However, ML presentation did not change the rates of recommending advance care planning or palliative care referral (1.3% and 0.7%, respectively). ML-based prognostic assessments without explanations improve prognostic accuracy but do not change decisions around palliative care referral or advance care planning.

  PublisherDOI

• Defining the role of pragmatic clinical trials in cancer clinical research: outcomes of a collaborative workshop hosted by the European Organisation for Research and Treatment of Cancer

  Universität Zürich, ZORA · 2025-05-01

  articleOpen access
  PublisherDOI

• Phase III Randomized Trial of Proton vs. Photon Therapy for Patients with Non-metastatic Breast Cancer Receiving Comprehensive Nodal Radiation: A Radiotherapy Comparative Effectiveness (Radcomp) Consortium Trial: Health-Related Quality of Life Outcomes

  International Journal of Radiation Oncology*Biology*Physics · 2025-10-14 · 3 citations

  articleSenior author
  PublisherDOI

• Rideshare services for transportation assistance in gynecologic oncology: a quality improvement study

  BMC Health Services Research · 2025-02-14 · 1 citations

  articleOpen access

  OBJECTIVE: Transportation barriers can lead to delays in care and suboptimal treatment. Our objective was to assess the utilization of a novel transportation pilot intervention in gynecologic oncology. METHODS: Since May 2022, we have provided donor-funded transportation to patients receiving gynecologic cancer treatment at 5 University of Pennsylvania practices. Patients are screened for transportation barriers at first visit and re-screened during care. Patients who screen positive are referred to the intervention, a HIPAA-compliant ride-sharing service. There are no income or insurance restrictions; distance was limited to 25 miles. We report descriptive statistics on ride completion, distance traveled, and cost. RESULTS: In the 15-month pilot, 133 of 4,376 patients (3%) screened positive, and 48 (1%) patients received rides. Of 85 patients who screened positive, but did not receive ride-sharing, 43 (51%) had transportation assistance through their insurance, 12 (14%) lived more than 25 miles away, and the remainder (30; 35%) identified alternative transportation. Patients who received transportation assistance were more likely to be older, self-identify as a race other than white, have Medicare or Medicaid insurance, and have a higher ECOG score than the overall patient population. Eight patients received a single ride, and the rest (n = 40) received multiple rides (range 2-30) for total of 417 rides. The mean time via ride-sharing was 19.5 min shorter than public transportation, and mean cost of a one-way trip was $25.75 (range $13.83-129.91). CONCLUSION: A rideshare service served a socially-vulnerable population and reduced commute times to oncology visits, which may contribute to more equitable access to cancer care. Further research on clinical outcomes is needed to understand the impact of transportation assistance on equitable cancer care delivery.

  PublisherOA PDFDOI

• Using a Simplified Survey to Assess Lung Cancer Screening Eligibility: The Effects of Messaging and Incentives on Survey Response

  American Journal of Respiratory and Critical Care Medicine · 2025-05-01

  article

  Abstract Rationale: Uptake of annual lung cancer screening (LCS) remains suboptimal, due in part to challenges with identifying adults who meet eligibility criteria for lifetime smoking intensity (i.e., 20 pack-years or greater). In our prior work, we found brief yes/no questions can accurately estimate LCS eligibility (Rendle et al. JAMA Netw Open. 2023), but uncertainty remains on how best to ensure equitable and widespread response to these questions. To help fill this gap, we conducted a randomized trial (NCT06133816) to test the effects of different types of messaging with or without incentives on survey completion. Methods: Using electronic health record data from a large urban healthcare system, we identified primary care patients who met age eligibility for LCS and had not been screened or diagnosed with lung cancer. We randomly assigned eligible patients to one of 12 different study arms and invited them to complete the brief survey via text message. All participants randomly received 1 of 3 introductory messages and 1 of 2 tobacco use normalizing messages, and half were offered a lottery-based incentive for survey completion (2x3x2 factorial design). We modeled main effects (incentive vs no incentive) using independent proportions z-test and marginal response probability across arms using random effects. We report results overall and by race and sex to assess equity of effectiveness. Results: Of the 13,245 patients contacted between July-November 2023, 46% (6,125) confirmed identity and were randomized. Of those randomized, 74% (4,541/6,125) completed the survey. We observed a significantly higher response rate in Black and Asian adults than White adults (77.3% vs 79.4% vs 73.9%; p&amp;lt;0.01) and in males than females (75.8% vs 72.5%; p&amp;lt;0.01). Contrary to hypothesized effects, we observed a higher response rate in participants who were not offered an incentive in comparison to those who were (51.4% vs 48.6%; p&amp;lt;0.001). In stratified analysis, this association remained significant in males, females, and Non-Black adults, but not Black adults. When comparing survey completion across the 12 arms, we observed marginal differences (ranging from 66-79%), but nothing reaching statistical significance (Figure 1). Conclusions: Our study found that using text messaging to assess LCS eligibility resulted in equitable and high responses rates, supporting this approach as an effective, low-touch option for overcoming barriers to assessing LCS eligibility in diverse communities. Future research will assess the accuracy of using this brief survey to estimate LCS eligibility and lung cancer detection in this large sample of patients.

  PublisherDOI

• Sequential EHR interventions to increase genetic testing for breast and ovarian cancer predisposition across diverse patient populations in gynecology practices at Penn Medicine.

  Journal of Clinical Oncology · 2025-05-28

  article

  10589 Background: Genetic testing (GT) identifies individuals who may benefit from increased surveillance and risk reduction strategies. GT is under-utilized, especially in those without a personal history of cancer and in minority populations. Methods: We identified a patient cohort meeting NCCN criteria for genetic testing utilizing electronic health record (EHR) phenotyping in 2 diverse gynecological practices. NCCN criteria included individuals with a personal history of ovarian cancer, early-onset (&lt;50) breast cancer diagnosed before 2021, or a family history (FH) of ovarian cancer or male breast cancer. Participants with prior genetics visits were excluded. Patient nudges and provider messaging strategies were introduced to boost genetic counseling consultation. Nudges included patient portal messaging (PP) followed by texts using the Way To Health platform (WH) in those that did not respond to PP. For non-responders to patient directed nudges, genetic counseling consult orders were placed using Epic’s Pend &amp; Send tool and sent to their gynecologist. Endpoints included the open rate for PP, response rate for the WH text, and the number of genetic counseling appointments completed. Differences between the clinics were calculated by Chi Square. Results: Of 1055 patients identified and who received a PP message regarding genetic counseling, 81% had a FH of ovarian cancer. Characteristics of the patient populations differed across clinics: Clinic D (n=505): 71.3% Black, 18% White and 67% &lt; 45 years. Clinic R (n=550):10% Black, 83.5% White and 63% &gt; 60 years.79% opened PP and 22.1% replied to PP, more in Clinic R (26.7% vs 17.0%, p&lt;0.001). Patient engagement by PP or WH was 59.8% (631/1055), more in Clinic R (67.1% vs 51.9%, p&lt;0.001). Of those that connected by patient nudges, 62.8% (396/631) declined additional follow-up (more in Clinic R, 41.5% then Clinic D, 33.3%, p=0.014), either due to incorrect family history in EHR, prior genetic testing, or, in the majority of cases, because they were not interested (296/631) (46.9%). Provider nudges added little to patient nudges with regard to GT uptake. 25% (266/1055) scheduled and 14.9% (157/1055) of the cohort completed GT appointments with no difference between the two (Clinic D 13.9% vs Clinic R 15.8%, p=NS). Conclusions: Patient directed nudges led to engagement of nearly 60% of patients in two diverse gynecology practices. 25% of individuals scheduled and 14.9% completed appointments, with continued follow-up. Although engagement in PP and WH differed between the two clinics, the number of visits did not. An EHR-based approach to identifying patients and encouraging genetic testing is a relatively low effort, scalable strategy to increase reach and encourage engagement in genetic counseling. However, a majority of patients either did not respond or did not wish to be tested. Clinical trial information: NCT05721326 .

  PublisherDOI

Recent grants

• Admin_Core

  NIH · $9.0M · 2022–2027

• NIH Grant K07CA163616

  NIH · $875k · 2018

• NIH Grant F32CA123964

  NIH · $61k · 2007

Frequent coauthors

• Ravi B. Parikh

  Penn Center for AIDS Research

  113 shared

• Lawrence N. Shulman

  eHealth Initiative

  104 shared

• Peter Gabriel

  96 shared

• Jason A. Efstathiou

  Harvard University

  80 shared

• Katharine A. Rendle

  Abramson Cancer Center

  74 shared

• Anthony L. Zietman

  Mass General Brigham

  73 shared

• Neha Vapiwala

  University of Pennsylvania

  64 shared

• John P. Christodouleas

  University of Pennsylvania

  61 shared

Awards & honors

• Elected member of the American Society for Clinical Investig…
• Member of the National Cancer Policy Forum of the National A…