A whole-slide foundation model for digital pathology from real-world data.

Digital pathology poses unique computational challenges, as a standard gigapixel slide may comprise tens of thousands of image tiles1-3. Prior models have often resorted to subsampling a small portion of tiles for each slide, thus missing the important slide-level context4. Here we present Prov-GigaPath, a whole-slide pathology foundation model pretrained on 1.3 billion 256 × 256 pathology image tiles in 171,189 whole slides from Providence, a large US health network comprising 28 cancer centres. The slides originated from more than 30,000 patients covering 31 major tissue types. To pretrain Prov-GigaPath, we propose GigaPath, a novel vision transformer architecture for pretraining gigapixel pathology slides. To scale GigaPath for slide-level learning with tens of thousands of image tiles, GigaPath adapts the newly developed LongNet5 method to digital pathology. To evaluate Prov-GigaPath, we construct a digital pathology benchmark comprising 9 cancer subtyping tasks and 17 pathomics tasks, using both Providence and TCGA data6. With large-scale pretraining and ultra-large-context modelling, Prov-GigaPath attains state-of-the-art performance on 25 out of 26 tasks, with significant improvement over the second-best method on 18 tasks. We further demonstrate the potential of Prov-GigaPath on vision-language pretraining for pathology7,8 by incorporating the pathology reports. In sum, Prov-GigaPath is an open-weight foundation model that achieves state-of-the-art performance on various digital pathology tasks, demonstrating the importance of real-world data and whole-slide modelling.

Evaluation of an Integrated Intervention to Address Clinical Care and Social Needs Among Patients with Type 2 Diabetes.

BACKGROUND: The Providence Diabetes Collective Impact Initiative (DCII) was designed to address the clinical challenges of type 2 diabetes and the social determinants of health (SDoH) challenges that exacerbate disease impact. OBJECTIVE: We assessed the impact of the DCII, a multifaceted intervention approach to diabetes treatment that employed both clinical and SDoH strategies, on access to medical and social services. DESIGN: The evaluation employed a cohort design and used an adjusted difference-in-difference model to compare treatment and control groups. PARTICIPANTS: Our study population consisted of 1220 people (740 treatment, 480 control), aged 18-65 years old with a pre-existing type 2 diabetes diagnosis who visited one of the seven Providence clinics (three treatment and four control) in the tri-county area of Portland, Oregon, between August 2019 and November 2020. INTERVENTIONS: The DCII threaded together clinical approaches such as outreach, standardized protocols, and diabetes self-management education and SDoH strategies including social needs screening, referral to a community resource desk, and social needs support (e.g., transportation) to create a comprehensive, multi-sector intervention. MAIN MEASURES: Outcome measures included SDoH screens, diabetes education participation, HbA1c, blood pressure, and virtual and in-person primary care utilization, as well as inpatient and emergency department hospitalization. KEY RESULTS: Compared to patients at the control clinics, patients at DCII clinics saw an increase in diabetes education (15.5%, p<0.001), were modestly more likely to receive SDoH screening (4.4%, p<0.087), and had an increase in the average number of virtual primary care visits of 0.35 per member, per year (p<0.001). No differences in HbA1c, blood pressure, or hospitalization were observed. CONCLUSIONS: DCII participation was associated with improvements in diabetes education use, SDoH screening, and some measures of care utilization.

Drivers of High-cost Medical Complexity in a Medicaid Population.

BACKGROUND: Efforts to improve outcomes for the 10% of patients using two thirds of health care expenditures increasingly include addressing social determinants. Empiric evidence is needed to identify the highest impact nonmedical drivers of medical complexity and cost. OBJECTIVES: This study examines whether complex, highest cost patients have different patterns of critical life adversity than those with better health and lower utilization. RESEARCH DESIGN: Using a validated algorithm we constructed a complexity/cost risk patient profile. We developed and fielded a life experience survey (Supplemental Digital Content 1, http://links.lww.com/MLR/B920) to a representative sample, then examined how the prevalence of specific adversities varied between complex, high-cost individuals, and others. SUBJECTS: Surveys were sent to 9176 adult Medicaid members in Portland, Oregon. MEASURES: Our primary variable was high medical complexity health cost risk; an alternative specification combined health cost risk and actual utilization/cost. Our survey instrument measured exposure to early and later-life adversities. RESULTS: Compared with healthy individuals in our population, medically complex individuals had significantly higher rates of adversity. The greatest risk of medical complexity and cost was associated with substance use [odds ratio (OR), 4.1], homelessness (OR, 3.0), childhood maltreatment (OR, 2.8), and incarceration (OR 2.4). Those with the highest prior year acute care utilization and cost had the highest rates of these same factors: substance use (62.5%), homelessness (61.7%), childhood maltreatment (55.5%), and incarceration (52.1%). CONCLUSION: Clinical and policy strategies that mitigate high-impact social drivers of poor outcomes are likely critical for improving both health and costs for complex, high-needs patients.

The effect of increased cost-sharing on low-value service use.

We examine the effect of a value-based insurance design (VBID) program implemented at a large public employer in the state of Oregon. The program substantially increased cost-sharing for several healthcare services likely to be of low value for most patients: diagnostic services (e.g., imaging services) and surgeries (e.g., spinal surgeries for pain). Using a difference-in-differences design coupled with granular, administrative health insurance claims data over the period 2008-2012, we estimate the change in low-value service use among beneficiaries before and after program implementation relative to a comparison group not exposed to the VBID. Our findings suggest that the VBID significantly reduced the use of targeted services, with an implied elasticity of demand of -0.22. We find no evidence that the VBID led to substitution to non-targeted services or increased overall healthcare costs. However, we also observe no evidence that the program led to cost-savings.

Patient experiences in behavioral health integrated primary care settings: the role of stigma in shaping patient outcomes over time.

Behavioral health integration (BHI) models seek to improve patient experience and outcomes by bridging physical and behavioral health services. Past BHI research has not focused on stigma in these settings, which has been previously found to impact patient engagement and outcomes. We surveyed patients over a two year period at 12 integrated clinics in Oregon using measures developed by a Patient Advisory Team. Over a quarter of respondents reported stigmatization (26.81%). Compared to non-stigmatized patients, those who reported stigma had five times the odds of reporting unmet health needs (OR=5.14, p<0.0001), three times the odds reporting issues accessing care (OR=2.93, p<0.0001), six times the odds reporting hassle to get care (OR=6.49, p<0.0001), and three times the odds of reporting poor communication between providers (OR=3.45, p<0.0001). After examining the interaction between stigmatization and time, we found that stigmatized patients had lower odds at year two of reporting unmet health needs (OR=0.68, p=0.0034), issues accessing care (OR=0.77, p=0.0400), hassle getting care (OR=0.57, p=0.0001), and poor provider communication (OR=0.77, p=0.0544). We found that stigma remained prevalent for patients seeking care in the integrated clinics studied despite integration. Systems should consider integration efforts and reducing stigmatizing experiences in tandem to truly improve patient outcomes.

The Effect of Medicaid on Management of Depression: Evidence From the Oregon Health Insurance Experiment.

Policy Points: We take advantage of Oregon's Medicaid lottery to gauge the causal effects of Medicaid coverage on mental health care, how effectively it addresses unmet needs, and how those effects differ for those with and without a history of depression. Medicaid coverage reduced the prevalence of undiagnosed depression by almost 50% and untreated depression by more than 60%. It increased use of medications and reduced the share of respondents reporting unmet mental health care needs by almost 40%. There are likely to be substantial mental health consequences of policy decisions about Medicaid coverage for vulnerable populations. CONTEXT: Expanding Medicaid to previously uninsured adults has been shown to increase detection and reduce the prevalence of depression, but the ways that Medicaid affects mental health care, how effectively it addresses unmet needs, and how those effects differ for those with and without a history of depression remain unclear. METHODS: We take advantage of Oregon's Medicaid lottery to gauge the causal effects of Medicaid coverage using a randomized-controlled design, drawing on both primary and administrative data sources. FINDINGS: Medicaid coverage reduced the prevalence of undiagnosed depression by almost 50% and untreated depression by more than 60%. It increased use of medications frequently prescribed to treat depression and related mental health conditions and reduced the share of respondents reporting unmet mental health care needs by almost 40%. The share of respondents screening positive for depression dropped by 9.2 percentage points overall, and by 13.1 for those with preexisting depression diagnoses, with greatest relief in symptoms seen primarily in feeling down or hopeless, feeling tired, and trouble sleeping-consistent with the increase observed not just in medications targeting depression but also in those targeting sleep. CONCLUSIONS: Medicaid coverage had significant effects on the diagnosis, treatment, and outcomes of a population with substantial unmet mental health needs. Coverage increased access to care, reduced the prevalence of untreated and undiagnosed depression, and substantially improved the symptoms of depression. There are likely to be substantial mental health consequences of policy decisions about Medicaid coverage for vulnerable populations.

Odds of Hospitalization for COVID-19 After 3 vs 2 Doses of mRNA COVID-19 Vaccine by Time Since Booster Dose.

This study assesses the association between COVID-19 mRNA booster immunization compared with vaccination with the primary mRNA vaccination series alone and odds of hospitalization for COVID-19.

What does Medicaid expansion mean for cancer screening and prevention? Results from a randomized trial on the impacts of acquiring Medicaid coverage.

BACKGROUND: The Oregon Medicaid lottery provided a unique opportunity to assess the causal impacts of health insurance on cancer screening rates within the framework of a randomized controlled trial. Prior studies regarding the impacts of health insurance have almost always been limited to observational evidence, which cannot be used to make causal inferences. METHODS: The authors prospectively followed a representative panel of 16,204 individuals from the Oregon Medicaid lottery reservation list, collecting data before and after the Medicaid lottery drawings. The study panel was divided into 2 groups: a treatment group of individuals who were selected in the Medicaid lottery (6254 individuals) and a control group who were not (9950 individuals). The authors also created an elevated risk subpanel based on family cancer histories. One year after the lottery drawings, differences in cancer screening rates, preventive behaviors, and health status were compared between the study groups. RESULTS: Medicaid coverage resulted in significantly higher rates of several common cancer screenings, especially among women, as well as better primary care connections and self-reported health outcomes. There was little evidence found that acquiring Medicaid increased the adoption of preventive health behaviors that might reduce cancer risk. CONCLUSIONS: Medicaid coverage did not appear to directly impact lifestyle choices that might reduce cancer risk, but it did provide access to important care and screenings that could help to detect cancers earlier. These findings could have long-term population health implications for states considering or pursuing Medicaid expansion. Cancer 2016;122:791-797. © 2015 American Cancer Society.

The Oregon experiment--effects of Medicaid on clinical outcomes.

BACKGROUND: Despite the imminent expansion of Medicaid coverage for low-income adults, the effects of expanding coverage are unclear. The 2008 Medicaid expansion in Oregon based on lottery drawings from a waiting list provided an opportunity to evaluate these effects. METHODS: Approximately 2 years after the lottery, we obtained data from 6387 adults who were randomly selected to be able to apply for Medicaid coverage and 5842 adults who were not selected. Measures included blood-pressure, cholesterol, and glycated hemoglobin levels; screening for depression; medication inventories; and self-reported diagnoses, health status, health care utilization, and out-of-pocket spending for such services. We used the random assignment in the lottery to calculate the effect of Medicaid coverage. RESULTS: We found no significant effect of Medicaid coverage on the prevalence or diagnosis of hypertension or high cholesterol levels or on the use of medication for these conditions. Medicaid coverage significantly increased the probability of a diagnosis of diabetes and the use of diabetes medication, but we observed no significant effect on average glycated hemoglobin levels or on the percentage of participants with levels of 6.5% or higher. Medicaid coverage decreased the probability of a positive screening for depression (-9.15 percentage points; 95% confidence interval, -16.70 to -1.60; P=0.02), increased the use of many preventive services, and nearly eliminated catastrophic out-of-pocket medical expenditures. CONCLUSIONS: This randomized, controlled study showed that Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years, but it did increase use of health care services, raise rates of diabetes detection and management, lower rates of depression, and reduce financial strain.

Reevaluating Reports of Defensive Medicine.

There is ongoing policy debate about the potential for malpractice liability reform to reduce the use of defensive medicine and slow the growth of health care spending. The effectiveness of such policy levers hinges on the degree to which physicians respond to liability pressures by prescribing medically unnecessary care. Many estimates of this relationship are based on physician reports. We present new survey evidence on physician assessment of their own use of medically unnecessary care in response to medical liability and other pressures, including a randomized evaluation of the sensitivity of those responses to survey framing. We find that while use of such care is potentially quite prevalent, responses vary substantially based on survey framing, with the way the question is phrased driving differences in responses that are often as great as those driven by physician specialty or whether the physician has personally been named in a lawsuit. These results suggest that self-reported use of medically unnecessary care ought to be used with caution in the formulation of malpractice liability system reform.

The role of stigma in access to health care for the poor.

CONTEXT: The Affordable Care Act provides new Medicaid coverage to an estimated 12 million low-income adults. Barriers to access or quality could hamper the program's success. One of these barriers might be the stigma associated with Medicaid or poverty. METHODS: Our mixed-methods study involved 574 low-income adults and included data from an in-person survey and follow-up interviews. Our analysis of the interviews showed that many participants who were on Medicaid or uninsured described a perception or fear of being treated poorly in the health care setting. We defined this experience as stigma and merged our qualitative interviews coded for stigma with our quantitative survey data to see whether stigma was related to other sociodemographic characteristics. We also examined whether stigma was associated with access to care, quality of care, and self-reported health. FINDINGS: We were unable to identify other sociodemographic characteristics associated with stigma in this low-income sample. The qualitative interviews suggested that stigma was most often the result of a provider-patient interaction that felt demeaning, rather than an internalized sense of shame related to receiving public insurance or charity care. An experience of stigma was associated with unmet health needs, poorer perceptions of quality of care, and worse health across several self-reported measures. CONCLUSIONS: Because a stigmatizing experience in the health system might interfere with the delivery of high-quality care to new Medicaid enrollees, further research and policy interventions that target stigma are warranted.

Does health insurance continuity among low-income adults impact their children's insurance coverage?

Parent's insurance coverage is associated with children's insurance status, but little is known about whether a parent's coverage continuity affects a child's coverage. This study assesses the association between an adult's insurance continuity and the coverage status of their children. We used data from a subgroup of participants in the Oregon Health Care Survey, a three-wave, 30-month prospective cohort study (n = 559). We examined the relationship between the length of time an adult had health insurance coverage and whether or not all children in the same household were insured at the end of the study. We used a series of univariate and multivariate logistic regression models to identify significant associations and the rho correlation coefficient to assess collinearity. A dose response relationship was observed between continuity of adult coverage and the odds that all children in the household were insured. Among adults with continuous coverage, 91.4% reported that all children were insured at the end of the study period, compared to 83.7% of adults insured for 19-27 months, 74.3% of adults insured for 10-18 months, and 70.8% of adults insured for fewer than 9 months. This stepwise pattern persisted in logistic regression models: adults with the fewest months of coverage, as compared to those continuously insured, reported the highest odds of having uninsured children (adjusted odds ratio 7.26, 95% confidence interval 2.75, 19.17). Parental health insurance continuity is integral to maintaining children's insurance coverage. Policies to promote continuous coverage for adults will indirectly benefit children.

The Oregon health insurance experiment: when limited policy resources provide research opportunities.

In 2008 Oregon allocated access to its Medicaid expansion program, Oregon Health Plan Standard, by drawing names from a waiting list by lottery. The lottery was chosen by policy makers and stakeholders as the preferred way to allocate limited resources. At the same time, it also gave rise to the Oregon Health Insurance Experiment: an unprecedented opportunity to do a randomized evaluation - the gold standard in medical and scientific research - of the impact of expanding Medicaid. In this article we provide historical context for Oregon's decision to conduct a lottery, discuss the importance of randomized controlled designs for policy evaluation, and describe some of the practical challenges in successfully capitalizing on the research opportunity presented by the Oregon lottery through public-academic partnerships. Since policy makers will always face tough choices about how to distribute scarce resources, we urge thoughtful consideration of the opportunities to incorporate randomization that can substantially improve the evidence available to inform policy decisions without compromising policy goals.

Integrating patient-reported physical, mental, and social impacts to classify long COVID experiences.

Long COVID was originally identified through patient-reported experiences of prolonged symptoms. Many studies have begun to describe long COVID; however, this work typically focuses on medical records, instead of patient experiences, and lacks a comprehensive view of physical, mental, and social impacts. As part of our larger My COVID Diary (MCD) study, we captured patient experiences using a prospective and longitudinal patient-reported outcomes survey (PROMIS-10) and free-text narrative submissions. From this study population, we selected individuals who were still engaged in the MCD study and reporting poor health (PROMIS-10 scores < 3) at 6 months (n = 634). We used their PROMIS-10 and narrative data to describe and classify their long COVID experiences. Using Latent Class Analysis of the PROMIS-10 data, we identified four classifications of long COVID experiences: a few lingering issues (n = 107), significant physical symptoms (n = 113), ongoing mental and cognitive struggles (n = 235), and numerous compounding challenges (n = 179); each classification included a mix of physical, mental, and social health struggles with varying levels of impairment. The classifications were reinforced and further explained by patient narratives. These results provide a new understanding of the varying ways that long COVID presents to help identify and care for patients.

Evaluating childsmile, Scotland's National Oral Health Improvement Programme for children.

In the early 2000s, a Scottish Government Oral Health Action Plan identified the need for a national programme to improve child oral health and reduce inequalities. 'Childsmile' aimed to improve child oral health in Scotland, reduce inequalities in outcomes and access to dental services, and to shift the balance of care from treatment to prevention through targeted and universal components in dental practice, community and educational settings. This paper describes how an embedded, theory-based research and evaluation arm with multi-disciplinary input helps determine priorities and provides important strategic direction. Programme theory is articulated in dedicated, dynamic logic models, and evaluation themes are as follows: population-level data linkage; trials and economic evaluations; investigations drawing from behavioural and implementation science; evidence reviews and updates; and applications of systems science. There is also a growing knowledge sharing network internationally. Collaborative working from all stakeholders is necessary to maintain gains and to address areas that may not be working as well, and never more so with the major disruptions to the programme from the COVID-19 pandemic and response. Conclusions are that evaluation and research are synergistic with a complex, dynamic programme like Childsmile. The evidence obtained allows for appraisal of the relative strengths of component interventions and the reach and impact of Childsmile to feed into national policy.

Decreased Risk of Coronavirus Disease 2019-Related Hospitalization Associated With the Omicron Variant of Severe Acute Respiratory Syndrome Coronavirus 2.

Among 134 223 patients with coronavirus disease 2019 (COVID-19), we assessed how risk of hospitalization changed at different intervals in the pandemic, controlling for prior COVID-19 immunity. In multivariable analysis, outpatients with COVID-19 during the Omicron-predominant time period had significantly lower odds of hospitalization compared to pre-Delta (adjusted odds ratio, 0.26 [95% confidence interval, .22-.32]).

Comparative vaccine effectiveness against severe COVID-19 over time in US hospital administrative data: a case-control study.

BACKGROUND: Research suggests the protection offered by COVID-19 vaccines might wane over time, prompting consideration of booster vaccinations. Data on which vaccines offer the most robust protection over time, and which patients are most vulnerable to attenuating protection, could help inform potential booster programmes. In this study, we used comprehensive hospitalisation data to estimate vaccine effectiveness over time. METHODS: In this case-control study, we used data from a large US health-care system to estimate vaccine effectiveness against severe SARS-CoV-2 infection and examined variation based on time since vaccination, vaccine type, and patients' demographic and clinical characteristics. We compared trends in attenuation of protection across vaccines and used a multivariable model to identify key factors associated with risk for severe breakthrough infection. Patients were considered to have severe COVID-19 if they were admitted to the hospital, had a final coded diagnosis of COVID-19 (according to International Classification of Diseases Tenth Revision code U07.1) or a positive nucleic acid amplification test for symptomatic SARS-CoV-2 during their hospitalisation, and were treated with remdesivir or dexamethasone during hospitalisation. FINDINGS: Between April 1, 2021, and Oct 26, 2021, we observed 9667 admissions for severe COVID-19 (ie, cases). Overall, 1293 (13·4%) of 9667 cases were fully vaccinated at the time of admission, compared with 22 308 (57·7%) of 38 668 controls, who were admitted to hospital for other reasons. The median time between vaccination and hospital admission among cases was 162 days (IQR 118-198). Overall vaccine effectiveness declined mostly over the course of the summer, from 94·5% (95% CI 91·4-96·5) in April, 2021 (pre-delta), to 84·0% (81·6-86·1) by October, 2021. Notably, vaccine effectiveness declined over time, from 94·0% (95% CI 92·8-95·0) at days 50-100 after vaccination to 80·4% (77·8-82·7) by days 200-250 after vaccination. After 250 days, vaccine effectiveness declines were even more notable. Among those who received the BNT162b2 (Pfizer-BioNTech) vaccine, vaccine effectiveness fell from an initial peak of 94·9% (93·2-96·2) to 74·1% (69·6-77·9) by days 200-250 after vaccination. Protection from the mRNA-1273 (Moderna) and Ad26.COV2 (Janssen) vaccines declined less over time, although the latter offered lower overall protection. Holding other factors constant, the risk of severe breakthrough infection was most strongly associated with age older than 80 years (adjusted odds ratio 1·76, 95% CI 1·43-2·15), vaccine type (Pfizer 1·39, 0·98-1·97; Janssen 14·53, 8·43-25·03; both relative to Moderna), time since vaccination (1·05, 1·03-1·07; per week after week 8 when protection peaks, technically), and comorbidities including organ transplantation (3·44, 95% CI 2·12-5·57), cancer (1·93, 1·60-2·33), and immunodeficiency (1·49, 1·13-1·96). INTERPRETATION: Vaccination remains highly effective against hospitalisation, but vaccine effectiveness declined after 200 days, particularly for older patients or those with specific comorbidities. Additional protection (eg, a booster vaccination) might be warranted for everyone, but especially for these populations. In addition to promoting general vaccine uptake, clinicians and policy makers should consider prioritising booster vaccinations in those most at risk of severe COVID-19. FUNDING: None.

The individual and program impacts of eliminating Medicaid dental benefits in the Oregon Health Plan.

OBJECTIVES: We determined how elimination of dental benefits among adult Medicaid beneficiaries in Oregon affected their access to dental care, Medicaid expenditures, and use of medical settings for dental services. METHODS: We used a natural experimental design using Medicaid claims data (n = 22 833) before and after Medicaid dental benefits were eliminated in Oregon in 2003 and survey data for continuously enrolled Oregon Health Plan enrollees (n = 718) covering 3 years after benefit cuts. RESULTS: Claims analysis showed that, compared with enrollees who retained dental benefits, those who lost benefits had large increases in dental-related emergency department use (101.7%; P < .001) and expenditures (98.8%; P < .001) and in all ambulatory medical care use (77.0%; P < .01) and expenditures (114.5%; P < .01). Survey results indicated that enrollees who lost dental benefits had nearly 3 times the odds (odds ratio = 2.863; P = .001) of unmet dental need, and only one third the odds (odds ratio = 0.340; P = .001) of getting annual dental checkups relative to those retaining benefits. CONCLUSIONS: Combined evidence from both analyses suggested that the elimination of dental benefits resulted in significant unmet dental health care needs, which led to increased use of medical settings for dental problems.

Patients With Health-Related Social Needs More Likely to Report Poor Clinic Experiences.

Measuring patients' care experience is necessary to understanding and improving health care quality and is a core component of patient-centered care. In this study, we test whether patient health care experiences differed between patients with and without health-related social needs, above and beyond demographic differences previously studied. This study relies on survey data from 2341 patients who visited 1 of 7 primary care clinics in Portland, Oregon, and surrounding communities during the latter half of 2018. Survey analysis reveal that patients with at least 1 health-related social need had greater odds of reporting staff not always answering questions, not getting all the care they need, not getting the information to manage care, not being treated with respect by their provider, and getting care being a hassle. The findings from this study suggest that patients with health-related social needs are not getting the holistic care they expect in their primary care clinics and find it a hassle to get care regardless of their demographic characteristics and insurance status. This study may help to inform how health care systems and clinics can best serve patients with health-related social needs.