Our Publications

This retrospective cohort study used electronic health records (EHRs) and administrative data from the VA Corporate Data Warehouse, which contains observations from more than 5 million veterans per study year. The data comprised 27,787,152 observations among 7,215,711 unique veterans receiving VA primary care from 2016 to 2021. Primary outcomes were change in model performance; and percent of veterans with CAN ≥90th percentile with a palliative care visit. This study found that CAN algorithm performance declined from 2016 to 2021 due to shifts in outcome prevalence and distributions of health care utilization and demographic covariates. Close surveillance of clinical risk algorithms and quality metrics derived from algorithm-generated risk scores could mitigate suboptimal resource allocation or decision-making.

This study used 2022 Medicare fee-for-service claims and other sources of clinician and geographic data, and described care patterns for ambulatory services among dual vs non-duals. The authors found that duals received fewer ambulatory, in particular subspecialty, services than non-duals, while duals received slightly more services through primary care physicians and advanced practice professionals. Among duals, most services were provided in office rather than federally qualified health centers or rural health clinics by a disproportionally small number of geographically dispersed practices and clinicians. High-dual practices and clinicians differed from non-high dual counterparts with respect to multiple characteristics. These differences highlight the opportunity to improve access and outcomes among duals by targeting policy and practice changes toward certain groups and clinicians.

In this cohort study of more than 7 million hospitalized Medicare beneficiaries in 2022, Z codes were used more often for less clinically complex patients and patients with higher levels of historical health care utilization. These patterns lead to a spurious negative association between documented social risk and predicted mortality. Existing documentation patterns for social risk are susceptible to measurement bias through multiple mechanisms and have important implications for risk adjustment, value-based payment, and health care resource allocation.

In this multipronged, behaviorally informed intervention for clinicians and practices, the authors found increases in referrals to high-value settings for radiology procedures, but not for four other service types. Physician- and practice-focused interventions may be an effective and scalable avenue for generating site-of-service savings for certain service types, with moderate changes in referral behavior translating into large cost savings.

In this cohort study of 14,280 patients the authors investigated whether early participation in the first Medicare bundled payment episode for outpatient surgery was associated with reduced spending and improved quality. The authors found that for patients who received outpatient spine surgery, hospital participation in bundled payments was associated with reduced spending and improved quality for outpatient spine surgery, although longer-term evaluation of this and other outpatient conditions is needed.

This retrospective cohort study used a hospital year-level dataset on short-term acute care US hospitals from 2014 to 2022. Hospital-level and area-level measures were used to define Safety Net Hospitals. Hospital characteristics under each definition, overlap across definitions, and stability of SNH samples produced by each definition from were described. Different Safety Net Hospital definitions produced different samples, and candidate measures had variable overlap and stability over time. These findings highlight the trade-offs when considering different options to define Safety Net Hospitals.

This paper by The Parity Center Collaborators, Dr. Sri Lekha Tummalapalli and Dr. Yuvaram Reddy, investigates the effects of stratified benchmarks

In this Viewpoint article, Amol and coauthors discuss the implications of the US historically low investment in Primary care, including that US adults are less likely to have established relationships with primary care clinicians than adults in peer nations.

In this paper, the authors critically examine seven prominent and controversial health care algorithms and show that popular approaches that aim to improve the fairness of health care algorithms can in fact worsen outcomes for individuals across all racial and ethnic groups. The authors offer an alternative algorithm design that mitigates these harms.

The purpose of this trial is to examine the effectiveness of a multi-component nudge intervention to increase breast cancer screening among eligible primary care patients. Outcomes will be evaluated using generalized estimating equations (GEE) to address clinic-level clustering. GEE models will also be fit to evaluate the impact of the intensification nudge on mammogram completion. If successful, this trial provides a blueprint for leveraging scalable multi-component nudge interventions for clinicians and patients to improve cancer screening rates.