About

Douglas Barthold is a health economist and Associate Professor at the University of Washington, specializing in the intersection of health policies, healthcare utilization, and health outcomes, with particular focus on aging, dementia, and prescription drug policy. He is faculty at the Comparative Health Outcomes, Policy, and Economics (CHOICE) Institute, serves as Co-Director of the Program in Health Economics and Outcomes Research Methodology (PHEnOM), and is a Board Member of the Washington State Prescription Drug Affordability Board. His research explores how health insurance design and prescription drug costs impact management of chronic diseases, including mid-life risk factors for dementia. Barthold holds a Ph.D. in Economics from McGill University and previously conducted postdoctoral research at the Schaeffer Center for Health Policy and Economics at the University of Southern California. His expertise includes chronic disease management, health policy, health economics, and health services research, with a focus on aging, Alzheimer’s disease, and dementia prevention and management. He teaches courses in health economics and has contributed to understanding disparities in pharmaceutical therapy utilization and effectiveness, medication adherence, and the impact of drug regulation and policy on health outcomes.

Research topics

• Medicine
• Internal medicine
• Gerontology
• Psychology
• Endocrinology

Selected publications

• Angiotensin II–Stimulating Antihypertensive Medications and Dementia-Related Neuropathology

  JAMA Network Open · 2026-02-11 · 2 citations

  articleOpen access

  Importance: Antihypertensive medications that stimulate angiotensin II type 2 or 4 receptors (angiotensin II-stimulating medications) may be associated with lower risk of dementia. Objective: To examine associations between cumulative exposure to angiotensin II-stimulating vs angiotensin II-inhibiting antihypertensive medications and neuropathology, accounting for blood pressure. Design, Setting, and Participants: This community-based autopsy cohort study from the Adult Changes in Thought cohort was conducted at Kaiser Permanente Washington between February 24, 1994, and November 25, 2022, among 756 participants who had blood pressure measurements and at least 1 person-year (PY) of angiotensin II-stimulating or -inhibiting antihypertensive medication exposure prior to death. Statistical analysis was performed between September 2024 and August 2025. Exposure: Angiotensin II-stimulating antihypertensive medications (angiotensin II receptor blockers, dihydropyridine calcium channel blockers, thiazides) and angiotensin II-inhibiting antihypertensive medications (angiotensin-converting enzyme inhibitors, β-blockers, nondihydropyridine calcium channel blockers) were ascertained from paper-based medical records (before 1977) and electronic prescription fill data (after 1977). The primary exposure was cumulative angiotensin II PYs, and the secondary exposure was long-term use (≥15 years). Main Outcomes and Measures: Neuropathology outcomes were classified as Alzheimer disease related, vascular brain injury, or other. Exploratory outcomes included quantitative measures of Aβ42 and phosphorylated tau. Data were analyzed using multivariable modified Poisson, proportional odds, and linear regression models and accounted for potential selection bias. Results: The sample included 756 participants (mean [SD] age at death, 89.2 [6.4] years; 440 women [58.2%]; mean [SD] follow-up, 22.2 [13.5] years). Compared with exposure to 5 additional PYs of angiotensin II-inhibiting antihypertensive medications, exposure to 5 additional PYs of angiotensin II-stimulating antihypertensive medications was associated with a 6% lower risk for arteriolosclerosis (relative risk [RR], 0.94; 95% CI, 0.89-0.99), with long-term use associated with a 24% lower risk (RR, 0.76; 95% CI, 0.63-0.91). For exploratory outcomes, PYs of angiotensin II-stimulating antihypertensive medications were associated with less quantitative phosphorylated tau burden in several brain regions (temporal lobe [adjusted ratio of geometric means, 0.79; 95% CI, 0.62-1.00], hippocampus [adjusted ratio of geometric means, 0.83; 95% CI, 0.71-0.97], cornu ammonis subfield 1 [adjusted ratio of geometric means, 0.86; 95% CI, 0.74-0.99], and transentorhinal cortex [adjusted ratio of geometric means, 0.83; 95% CI, 0.70-0.98]) but not with Aβ42 quantitative measures. Conclusions and Relevance: In this community-based autopsy cohort study, angiotensin II-stimulating antihypertensive medications were associated with lower risk of neuropathological burden, supporting findings from epidemiologic dementia studies. Additional mechanistic research examining the effects of individual antihypertensive classes on Alzheimer disease-related biomarkers is warranted.

  PublisherOA PDFDOI

• Patient preferences for long-acting antiretroviral treatment among people with HIV in Kenya: a discrete choice experiment

  BMC Infectious Diseases · 2026-02-11

  articleOpen access

  BACKGROUND: For people with HIV (PWH), long-acting antiretroviral therapies (LA-ART) are promising treatment alternatives to daily oral regimens, with potential to improve adherence and achieve viral suppression. Understanding patient preferences is crucial for successful and efficient implementation and scaling of LA-ART in resource-limited settings such as Kenya. We conducted a discrete choice experiment (DCE) to elicit preferences for LA-ART attributes among PWH in Kenya. METHODS: We recruited 700 PWH taking daily oral ART from Kenyatta National Hospital and two Sex Workers Outreach Program clinics in Nairobi. In 17 choice scenarios, participants chose between their current daily oral regimen and two hypothetical LA-ART alternatives defined by seven attributes: delivery mode (long-acting oral, subcutaneous or intramuscular injection, implant), administration location (clinic, chemist, home), frequency (weekly, every 1, 2, 3, 6, or 12 months), delivery-site pain (none, mild, moderate), pre-treatment viral suppression (required, not required), pre-treatment negative reaction testing (required, not required), and late-dose leeway (short, long). We used conditional logistic regressions with interactions between mode and pain to determine the relative importance in participants' choices across attributes. RESULTS: Participants had a median age of 36 years (interquartile range [IQR]: 28-46); 64% were female, 83% were virally suppressed, 51% were from key populations (e.g., sex workers or men who have sex with men), and median time on ART was 9 years (IQR: 5-15). Participants generally preferred the hypothetical LA-ART options over their current daily oral ART, and the interaction of delivery mode and pain was the most important attribute combination. Oral LA-ART was the most preferred mode; 1-year implants with mild pain was the next preferred option. Participants favored administration at clinics to chemists or home and preferred less frequent dosing. CONCLUSIONS: LA-ART would be highly acceptable in Kenya, with oral LA-ART and administration at clinics as the preferred formulation and location. Our findings provide valuable evidence to guide the development of novel LA-ART products. Future research should evaluate preference heterogeneity and investigate ways to effectively scale LA-ART in Kenya and similar settings, while taking into account patient preferences. CLINICAL TRIAL NUMBER: Not applicable.

  PublisherDOI

• Challenges of the Inflation Reduction Act for long-term care pharmacy: Examining impact and policy solutions

  Journal of Managed Care & Specialty Pharmacy · 2026-02-10

  articleOpen accessSenior author

  The Inflation Reduction Act of 2022 contains several health care provisions aimed at reducing costs for Medicare beneficiaries. Although these goals are critical, the legislation may have unintentional adverse impacts on long-term care (LTC) pharmacies. This article examines how specific provisions of The Inflation Reduction Act of 2022 may destabilize the LTC pharmacy sector, creating barriers to access for LTC residents.We analyze the financial implications of the Medicare Drug Price Negotiation Program, illustrating how the shift to maximum fair price reimbursement combined with inadequate dispensing fees could result in an estimated revenue decline of more than 85% for drugs selected within the program. We also outline how new payment flows for the maximum fair price involving the Medicare Transaction Facilitator will likely extend reimbursement timelines, creating liquidity challenges for pharmacies operating on thin margins.Beyond finances, we discuss operational conflicts with the Medicare Prescription Payment Plan. Specifically, we argue that alerting requirements are incompatible with LTC workflows and pose unnecessary administrative burden on pharmacies. To address these vulnerabilities, we propose near-term legislative and regulatory remedies, including appropriating funds to ensure timely manufacturing payments and codifying the standard default refund amount to protect acquisition discounts. Finally, we recommend structural reforms that shift reimbursement away from volume-based models to value-based care. By aligning payment with patient outcomes, policymakers can ensure the financial sustainability of LTC pharmacies and the continued delivery of highly effective pharmaceutical care to the nation's aging population.

  PublisherOA PDFDOI

• Patient preferences for long-acting HIV treatment: a preference heterogeneity assessment

  BMC Infectious Diseases · 2025-02-19 · 7 citations

  articleOpen access

  BACKGROUND: Long-acting antiretroviral therapy (LA-ART) is an emerging alternative to daily oral ART pills that may improve HIV treatment adherence. We studied preference heterogeneity for LA-ART among people with HIV (PWH) in western Washington State and Atlanta, Georgia to determine how preference heterogeneity was related to individual characteristics. METHODS: We recruited 699 PWH to complete a survey including 17 choice-tasks, each of which included two hypothetical LA-ART alternatives and current daily oral therapy. Each hypothetical alternative was defined by mode (long-acting [LA] oral pills, subcutaneous injections, intramuscular injections, and implants), frequency, treatment location (home, clinic, or pharmacy), injection pain, pre-treatment time undetectable, pre-treatment reaction testing, and late-dose leeway. We fitted a latent class model to the data and investigated associations between class membership and individual characteristics. RESULTS: Our sample had three classes which were defined by their treatment preferences. Two classes preferred LA-ART over current treatment: the LA-Implant class (29%) and the LA-Oral-or-Injection class (35%). In contrast, the Daily-or-LA-Oral class (36%) preferred current treatment or LA oral pills taken at home. Compared to the third class, participants from the other two were younger, more educated, less adherent to current ART, and less averse to injections. Further, LA-Implant participants were less likely to be virally suppressed and had easier clinic access. LA-Oral-or-Injection participants had a higher prevalence of psychotic disorders. CONCLUSION: These results provide a deeper understanding of the preference landscape for LA-ART and can aid in the development of interventions better aligned with individual preferences.

  PublisherOA PDFDOI

• EE301 Cost-effectiveness Analysis of Zavegepant in Acute Migraine Treatment

  Value in Health · 2025-07-01 · 2 citations

  articleSenior author
  PublisherDOI

• Adaptation and pilot testing of a discrete choice experiment and survey on patient preferences for long-acting antiretroviral therapies for HIV in Kenya

  AIDS Care · 2025-11-22 · 2 citations

  articleOpen access

  Long-acting antiretroviral therapy (LA-ART) could increase viral suppression among people living with HIV (PLWH) by promoting ART uptake and adherence. We adapted and pilot-tested a discrete-choice experiment (DCE) and an associated survey that were previously implemented in the United States, based on key informant interviews with Kenyan HIV experts and research team discussion about changes needed for the Kenyan context. We conducted five waves of pilot-testing, each with 9-11 PLWH receiving care at the Kenyatta National Hospital Comprehensive Care Clinic, iteratively updating the DCE (which elicited participant preferences) and the survey (which collected data on participant characteristics) as required. We enrolled 50 participants (median age 36, 56% male) between May and November 2022. Overall understanding of the DCE instructions and choice tasks was good, with the majority needing no assistance and most others needing minimal assistance. Most edits made during pilot testing were to questions in the survey rather than the DCE text or design. Preliminary results assessed using conditional logistic regression were similar to results of the US DCE: Across all participants, LA-ART alternatives were preferred over current daily oral ART, with the strongest preference for a long-acting oral pill. Full implementation with 700 participants is underway.

  PublisherDOI

• PCR101 Association of Chronic Conditions and Medication Burden with Patient Preferences for Long-Acting Antiretroviral Therapy in HIV Treatment

  Value in Health · 2025-07-01

  article
  PublisherDOI

• Utilization of Low- and High-Value Health Care by Individuals With and Without Cognitive Impairment

  The American Journal of Managed Care · 2024-07-01 · 2 citations

  articleOpen access1st authorCorresponding

  OBJECTIVES: Cognitive impairment and dementia have rising prevalence and impact the health care utilization and lives of older adults. Receipt of low-value (LV) care and underutilization of high-value (HV) care by individuals with these cognitive disorders may have negative consequences for patient health, health system efficiency, and societal welfare. Evidence on health care value among cognitively impaired individuals is limited; we thus ascertained receipt of LV and HV health care in older adults with normal cognition, cognitive impairment without dementia (CIND), and dementia. STUDY DESIGN: Retrospective cohort study of Health and Retirement Study data linked to Medicare claims (1996-2018). METHODS: We examined the association between cognitive decline and the receipt of 5 LV and 7 HV services vs individuals with no change in cognition. RESULTS: Receipt of LV care ranged from 4% to 13% regardless of cognitive status. Cognitive decline (from unimpaired to either CIND or dementia) was associated with decreased probability of receipt of 1 LV service (colorectal cancer screening at 85 years and older [5-percentage-point reduction; P = .047]) and 3 HV services (glucose-lowering drugs [7-percentage-point reduction; P = .029], statins [32-percentage-point reduction; P = .045], and antiresorptive therapy [61-percentage-point reduction; P = .019]). CONCLUSIONS: LV service receipt is wasteful and may be harmful, but it was not consistently associated with cognitive status. Lack of HV care for those with cognitive impairment could be a missed opportunity to improve well-being or reduce preventable adverse events. Our results suggest opportunities for improving the quality of care received by all older adults, including those with cognitive impairment.

  PublisherOA PDFDOI

• Caregiving for dementia: trends pre-post onset and predictive factors of family caregiving (2002–2018)

  Health Affairs Scholar · 2024-02-16 · 4 citations

  articleOpen access

  Persons living with Alzheimer's and other related forms of dementia rely heavily on care from family and friends for assistance with daily activities ("family care"), but little is known about care transitions over time. We analyzed data from the Health and Retirement Study to describe caregiving patterns, from 2 years before dementia onset and up to 6 years after. Using sociodemographic data from the interview prior to dementia onset, we determined if there are significant factors that predict receipt of family care at dementia onset. We found that one-third (33%) of people living with dementia were receiving help with daily activities 2 years prior to their first positive dementia screen and this increased to 60% during the first positive screen. Nearly all of those receiving assistance received family care. We found multiple significant predictors of receiving family care at onset, including race, education, access to private health insurance, number of activities of daily living that were difficult, number of chronic conditions, and already receiving help. This demonstrates potential gaps in dementia care, and which subpopulations may benefit most from targeted interventions for household members who do not have adequate caregiving resources or programs that provide additional formal care.

  PublisherOA PDFDOI

• Medicare expenditures among spouses of persons with dementia

  Journal of the American Geriatrics Society · 2024-07-08 · 1 citations

  articleOpen access

  BACKGROUND: Spouses of persons living with dementia face intense strains on their well-being compared with similarly aged adults and spouses of partners with no dementia. This strain can impact spouses' health and healthcare needs, and therefore affect their healthcare utilization and expenditures. METHODS: Using data from the Health and Retirement Study linked with Medicare claims, we matched dyads of spouses and their partners with dementia (SPWD) to a comparison group of similar spouses and their partners with no dementia (SPWND). We then examined Medicare expenditures for spouses in the 5 years following their partner's dementia onset month using a two-part regression model. RESULTS: SPWD cumulative total Medicare expenditures were, on average, $60,043 in the 5 years post dementia onset, compared to $56,068 for SPWND. This difference ($3974, 95% CI = [-$3,199; $11,477]) was not significant. However, there were significant differences in the 5th year's total expenditures (+$2,748 [$321; $5,447]), driven by inpatient expenditures ($1,562 [$22; $3,277]). CONCLUSIONS: Despite the differences in partner's dementia status, we found no significant difference in the 5-year cumulative Medicare expenditures between SPWD and SPWND. Compared to previous studies, we likely captured an earlier stage of dementia more consistently for a broader population which may be less straining on spouses. Further research should examine patterns of expenditures in later years and around critical timepoints in caregiving, such as partner transitions to formal long-term care settings and death, to better understand healthcare expenditures for spouses of persons living with dementia.

  PublisherOA PDFDOI

Frequent coauthors

• Zachary A. Marcum

  University of Washington

  63 shared

• Shelly L. Gray

  Seattle University

  53 shared

• Paul K. Crane

  University of Washington

  41 shared

• Eric B. Larson

  Human Longevity (United States)

  38 shared

• Willem A. van Gool

  University of Amsterdam

  36 shared

• Edo Richard

  Amsterdam University Medical Centers

  36 shared

• Eric P. Moll van Charante

  36 shared

• Jan Willem van Dalen

  Radboud University Medical Center

  36 shared

Education

• Ph.D., Economics

  McGill University