Jeremy D. Goldhaber-Fiebert’s research while affiliated with Stanford Medicine and other places

Background: The introduction of novel treatments and treatment approaches for breast cancer has reduced breast cancer mortality and unnecessary toxicity. As care improvements develop, it is crucial to understand whom they reach. We used insurance claims data to analyze the dissemination of four major trends in breast cancer treatment into the population. Methods: We used SEER-Medicare data to study four changes in the treatment of local and regional breast cancer in patients diagnosed between 2010 and 2018: (1) adjuvant paclitaxel-trastuzumab (APT) for patients with HER2+ local disease, (2) pertuzumab for patients with HER2+ regional disease, (3) increasing use of neoadjuvant chemotherapy for patients with triple-negative or HER2+ regional disease, and (4) decreasing use of adjuvant chemotherapy for patients with hormone receptor-positive, HER2-negative regional disease. Patients who received no systemic therapy were excluded. We used multilevel logistic regression to identify demographic (age, comorbidity, race/ethnicity, rurality, median income of residence census tract [low, middle, high]), tumor (hormone receptor status), and care (NCI vs other academic vs community hospital, breast cancer specialization of treating oncologist) factors associated with receipt of updated care. Breast cancer specialization of a physician was defined by terciles as the proportion of breast cancer patients among their patients in SEER-Medicare with colon, lung, or breast cancer. 95% confidence intervals are shown. Results: We identified 2,385 patients with localized HER2+ breast cancer; 1,937 patients with regional HER2+ breast cancer; 3,742 patients with regional triple-negative or HER2+ breast cancer; and 14,611 patients with regional hormone receptor-positive, HER2-negative breast cancer. In these respective populations, from 2010 to 2018, use of the APT regimen increased from 31% (24-38%) to 73% (67-78%), pertuzumab from 0% to 70% (65-75%), neoadjuvant chemotherapy from 24% (19-28%) to 61% (57-66%), and adjuvant chemotherapy decreased from 37% (34-39%) to 26% (24-28%). In multivariate analyses, higher median income of residence census tract of the patient and greater specialization level of treating oncologist were significantly associated with receipt of updated care for all four trends in breast cancer treatment. Additionally, greater cancer specialization of hospital type was associated with receipt of updated care for the APT regimen, pertuzumab, and neoadjuvant chemotherapy. We develop a model that, after adjusting for patient and clinical characteristics, estimates the probability of receiving updated care, according to tumor subtype and stage, for a SEER-Medicare patient from a low-income area treated by an oncologist with low breast cancer specialization vs one from a high-income area treated by an oncologist with high specialization: for APT, 40% (37-42%) vs 76% (75-77%), for pertuzumab 28% (25-32%) vs 63% (60-66%), for neoadjuvant chemotherapy 24% (23-26%) vs 56% (54-57%), and for omission of adjuvant chemotherapy 62% (60-63%) vs 72% (71-73%). In this model, specialization alone substantially mitigates the disparity: for example, a patient from a low-income area treated by an oncologist with high specialization has an estimated 63% (59-67%) probability of receiving appropriate APT. Conclusions: This study reveals disparities in the dissemination of major advances in breast cancer treatment across the SEER-Medicare population from 2010-2018. Patients from lower-income areas, cared for at community hospitals, and treated by oncologists with less breast cancer focus were less likely to receive updated care. These results provide important context for targeting efforts to ensure all eligible patients benefit from the latest advancements in breast cancer care. Citation Format: Jennifer Caswell-Jin, Marissa Reitsma, Hao Tang, James Dickerson, Shannon Phillips, Allison W. Kurian, Becky Staiger, Jeremy D. Goldhaber-Fiebert. Disparities in the dissemination of new breast cancer treatments in the United States [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P4-04-01.

Introduction Economic evaluations identify the best course of action by a decision maker with respect to the level of health within the overall population. Traditionally, they identify 1 optimal treatment choice. In many jurisdictions, multiple technologies can be covered for the same heterogeneous patient population, which limits the applicability of this framework for directly determining whether a new technology should be covered. This article explores the impact of different decision frameworks within this context. Methods Three alternate decision frameworks were considered: the traditional normative framework in which only the optimal technology will be covered (normative); a commonly adopted framework in which the new technology is recommended for reimbursement only if it is optimal, with coverage of other technologies remaining as before (current); and a framework that assesses specifically whether coverage of the new technology is optimal, incorporating previous reimbursement decisions and the market share of current technologies (positivist). The implications of the frameworks were assessed using a simulated probabilistic Markov model for a chronic progressive condition. Results Results illustrate how the different frameworks can lead to different reimbursement recommendations. This in turn produces differences in population health effects and the resultant price reductions required for covering the new technology. Conclusion By covering only the optimal treatment option, decision makers can maximize the level of health across a population. If decision makers are unwilling to defund technologies, however, the second best option of adopting the positivist framework has the greatest relevance with respect to deciding whether a new technology should be covered.

Background Of the hundreds of millions of COVID-19 cases globally, most have been non-fatal, though “Long COVID” after acute infection has been documented in many. While studies report post-COVID increases chronic disease incidence including diabetes mellitus (DM), they frequently underrepresent racial/ethnic minorities and lack controls for potential confounds (e.g., increased DM testing after COVID-19). Methods We conducted a case-control cohort time-to-event study of 29,470 individuals incarcerated in 31 California state prisons. The main outcome was incident diagnosed DM among individuals incarcerated continuously since January 1, 2019 with no DM diagnosis prior to March 1, 2020 (beginning of the unexposed period of observation). The main exposure was a positive COVID-19 test, with the exposure period beginning 31 days afterwards (post-acute period). Covariates included age, gender, race/ethnicity, BMI, and blood glucose at the start of the pandemic, frequency of healthcare contacts prior to the pandemic, and COVID-19 testing frequency prior to testing positive. We excluded individuals who lacked BMI or blood glucose measurements prior to or during the pandemic or were never tested for COVID-19 along with those who had been prescribed blood glucose-altering medications or had a diagnosed condition that could alter blood glucose. We estimated multivariate Cox proportional hazard models: 1) exposure variable and all covariates; 2) adding interactions between the exposure and each covariate. We assessed whether confounding due to changes in DM testing post-COVID could explain our results. Results COVID-19 infection significantly increased the rate of incident DM (main effects model HRR: 1.17 [95%CI: 1.03-1.34]; no significant interaction effects were observed). If all individuals in our study had had a COVID-19 infection, the 2-year cumulative risk of DM would have been 3.2% [2.5%-3.9%] compared to 2.7% [2.1%-3.4%] if none had been infected. While our findings were consistent to different definitions of the post-acute COVID period, confounding due to changes in DM testing post-COVID may imply that the effect is halved (HRR: 1.08-1.10). Conclusion COVID-19 may increase the risk of incident DM long after acute infection, warranting additional provider awareness and clinical consideration.

Background: Of the hundreds of millions of COVID-19 cases globally, most have been non-fatal, though "Long COVID" after acute infection has been documented in many. While many studies have reported post-COVID increases chronic disease incidence including hypertension (HTN), racial/ethnic minority study participants are underrepresented and studies often lacks controls for potential confounds. Methods: We conducted a case-control cohort time-to-event study of 39,746 individuals incarcerated in 31 California state prisons. The main outcome was incident diagnosed HTN among individuals incarcerated continuously since January 1, 2019 with no HTN diagnosis prior to March 1, 2020 (beginning of the unexposed period of observation). The main exposure was a positive COVID-19 test, with the exposure period beginning 31 days afterwards (post-acute period). Covariates included age, gender, race/ethnicity, BMI, and blood pressure at the pandemic's start, frequency of healthcare contacts prior to the pandemic, and COVID-19 testing frequency prior to testing positive. We excluded individuals who lacked BMI or blood pressure measurements prior to or during the pandemic or were never tested for COVID-19 along with those who had been prescribed blood pressure-altering medications. We estimated multivariate Cox proportional hazard models: 1) exposure variable and covariates; 2) adding interactions between the exposure and each covariate. We assessed whether confounding due to changes in HTN testing post-COVID could explain our results. Results: COVID-19 infection significantly increased the rate of incident HTN (main effects model HRR: 1.44 [95%CI: 1.32-1.57]; including interactions HRR: 2.05 [1.50-2.79]). If all individuals in our study had a COVID-19 infection, the 2-year cumulative risk of hypertension was 7.1% [5.7%-8.5%] versus 5.0% [4.2%-5.9%] if none had been infected. The largest absolute effects of COVID-19 on HTN incidence were in those with higher BMIs, higher pre-pandemic blood pressure levels and older ages. Our findings remained consistent with different definitions of the post-acute period and to confounding due to changes in HTN testing post-COVID. Conclusion: COVID-19 increases the risk of incident hypertension long after acute infection. As so many people have had COVID-19 and elevated blood pressure, additional provider awareness and clinical consideration are warranted.

Importance: Many acceptable quality deceased donor kidneys go unused every year. Older transplant candidates are more vulnerable to rapid health decline. Objective: To assess the cost-effectiveness of increasing the rate of kidney transplantation in older patients with end-stage kidney disease by using acceptable quality deceased donor kidneys. Design: This cost-effectiveness analysis utilizes a microsimulation model of the kidney transplantation process for older adult candidates over a lifetime horizon. Setting: Health state transition probabilities are derived from Scientific Registry of Transplant Recipient data. Costs and quality-of-life weights are derived from published literature and United States Renal Data System annual reports, all of which are varied in a probabilistic sensitivity analysis. Participants: A synthetic population of deceased donor kidney transplant candidates 65 or older. Interventions: Increasing the rate of transplantation in 5% increments higher than the status quo rate from 5% to 25% using acceptable quality deceased donor kidneys. Main Outcomes and Measures: The primary outcomes are the number of key waitlist and post-transplant outcomes, costs, quality-adjusted life-years (QALYs), incremental cost-effectiveness ratios (ICERs). Results: We estimated there would be 141 fewer waitlist deaths per 10,000 candidates if the rate of deceased donor transplantation were increased by 25% from the status quo rate. Increasing the rate of deceased donor transplantation by 25% costs $8,100 per QALY gained or is cost-saving from the healthcare sector and modified healthcare sector perspectives, respectively. From the healthcare sector perspective, a 25$40,000 per QALY gained. Conclusions and Relevance: Increasing the rate of kidney transplantation in older adults, even using acceptable quality deceased donor kidneys, would be cost-effective or cost-saving.

Background The United States aims to reduce the incidence of human immunodeficiency virus (HIV) and hepatitis C virus (HCV) infections by 90% by 2030. Objective To identify strategies for achieving incidence reduction goals by scaling interventions that address the syndemic of substance use disorder, HIV infection, and HCV infection among people who inject drugs (PWID). Design Stochastic agent-based multiplex network model. Setting Urban areas in the United States. Participants People who inject drugs. Interventions Scenarios scaled interventions from current baseline to moderate and high values. Prevention and cessation interventions were increased 15 (moderate) and 30 (high) percentage points. Test and treat interventions increased the percentage of PWID with current HCV infection achieving sustained virologic response per year from 3% to 16% (moderate) and 28% (high) and the percentage of PWID with HIV that were virally suppressed from 44% to 58% (moderate) and 71% (high). Measurements HIV and HCV infection incidence among PWID over ten years, and quality-adjusted life-years, discounted at 3% annually, over 80 years. Results High coverage across all three intervention strategies resulted in an 86% (95% Uncertainty Interval (UI): 72-96%) decrease in new HIV infections, a 90% (95% UI: 87-94%) decrease in new HCV infections, and an increase of 1.8 (95% UI: 1.6-2.0) discounted quality-adjusted life-years among PWID. Moderate coverage across all three strategies yielded 62% (95% UI: 39-81%) and 68% (95% UI: 61-74%) decreases in new HIV and HCV infections among PWID, respectively. Increasing cessation of injection consistently produced the largest gains in quality-adjusted life-years. Limitations Model did not examine specific interventions or economic costs. Parameters were not representative of all urban areas or all PWID. Conclusion Increases in survival and health-related quality of life for PWID can be achieved by scaling syndemic-focused intervention strategies. Primary Funding Source This project was funded by the Centers for Disease Control and Prevention, National Center for HIV, Viral Hepatitis, STD, and TB Prevention Epidemiologic and Economic Modeling Agreement (NEEMA; award #NU38PS004651) and the National Institute on Drug Abuse.

PURPOSE Creating value-aligned treatment pathways in breast cancer requires understanding the cost and cost-effectiveness of new therapies. To address uncertainty in the optimal treatment sequence, we developed a decision model to assess the cost-effectiveness of various treatment sequences for patients with human epidermal growth factor receptor 2 (HER2)–low metastatic breast cancer who are eligible for trastuzumab deruxtecan (T-DXd) and sacituzumab govitecan (SG) under current US Food and Drug Administration labeling. METHODS We derived disease progression and therapy data from the Destiny-Breast04 trial and sourced cost and quality-of-life data from the published literature. Our simulation modeled 57-year-old women with HER2-low, endocrine refractory, and triple-negative metastatic breast cancer eligible for third-line treatment. We evaluated four sequences: chemotherapy (chemo) → chemo, T-DXd → chemo, chemo → T-DXd, and T-DXd → SG. Outcomes included quality-adjusted life years (QALYs), total lifetime costs (2020 US dollars [USD], 3% annual discount), and incremental cost-effectiveness ratios. Sequences that cost <$150,000 USD to gain an additional QALY were considered cost effective. RESULTS Chemo → chemo has the lowest cost at$176,000 (USD) per patient and yields 0.82 QALYs. T-DXd → chemo costs $282,000 (USD) and yields 1.08 QALYs, with an incremental cost-effectiveness ratio of$408,000 (USD) per QALY gained. T-DXd → SG costs $304,000 (USD) and yields 1.09 QALYs, with an incremental cost-effectiveness ratio of$2,200,000 (USD) per QALY gained. Drug cost drives the cost differences between each strategy. For T-DXd → chemo to be cost effective at the $150,000 (USD) per QALY threshold, we estimate that a 41% price reduction for T-DXd is needed. CONCLUSION At its current price, T-DXd is not cost effective for HER2-low metastatic breast cancer. Price reductions can make this drug cost effective. Optimal value-based sequencing in this patient population uses a single antibody-drug conjugate rather than back-to-back conjugates.

Discrete Event Simulation (DES) is a flexible and computationally efficient approach for modeling diverse processes; however, DES remains underutilized in healthcare and medical decision-making due to a lack of reliable and reproducible implementations. We developed an open-source DES framework in R to simulate individual-level state-transition models (iSTMs) in continuous time accounting for treatment effects, time dependence on state residence, and age-dependent mortality. Our DES implementation employs a modular and easily adaptable structure, with each module corresponding to a unique transition between health states. To simulate the evolution of the process (i.e., individual state transitions), we adapted the next-reaction algorithm from the stochastic chemical reactions literature. Simulation-time dependence (age-dependent mortality) and state residence time dependence (transition from Sick to Sicker) are seamlessly incorporated into the DES framework via validated non-parametric and parametric sampling routines (e.g., inversion method) of event times. Treatment effects are integrated as scaling factors of the hazard functions (proportional hazards). We illustrate the framework's benefits by implementing the Sick-Sicker Model and conduct a cost-effectiveness analysis and probabilistic sensitivity analysis. We also obtain epidemiological outcomes of interest from the DES output, such as disease prevalence, survival probabilities, and distributions of state-specific dwell times. Our DES framework offers a reliable and accessible alternative that enables the simulation of more realistic dynamics of state-transition processes at potentially lower implementation and computational costs than discrete time iSTMs.

Simulation models inform health policy decisions by integrating data from multiple sources and fore-casting outcomes when there is a lack of comprehensive evidence from empirical studies. Such models have long supported health policy for cancer, the first or second leading cause of death in over 100 countries. Discrete-event simulation (DES) and Bayesian calibration have gained traction in the field of Decision Science because they enable efficient and flexible modeling of complex health conditions and produce estimates of model parameters that reflect real-world disease epidemiology and data uncertainty given model constraints. This uncertainty is then propagated to model-generated outputs, enabling decision makers to determine the optimal strategy to recommend, assess confidence in the recommendation, and estimate the value of collecting additional information. However, there is limited end-to-end guidance on structuring a DES model for cancer progression, estimating its parameters using Bayesian calibration, and applying the calibration outputs to policy evaluation and other downstream tasks. To fill this gap, we introduce the DES Model for C ancer I nterventions and P opulation H ealth in R (DESCIPHR), an open-source framework and codebase integrating a flexible DES model for the natural history of cancer, Bayesian calibration for parameter estimation, and screening strategy evaluation. We also introduce an automated method to generate data-informed parameter prior distributions and enhance the accuracy and flexibility of a neural network emulator-based Bayesian calibration algorithm. We anticipate that the adaptable DESCIPHR modeling template will facilitate the construction of future decision models evaluating the risks and benefits of health interventions. Key points for decision makers For simulation models to be useful for decision-making, they should accurately reproduce real-world outcomes and their uncertainty. The DESCIPHR framework and code repository address a gap in open-source resources to fit an individual-level model for cancer progression to real-world data and forecast the impact of cancer screening interventions while accounting for data uncertainty. The codebase is designed to be highly adaptable for researchers who wish to apply DESCIPHR for economic evaluation or for studying methodological questions.