Publications by Year: 2026

OBJECTIVES: To examine the association between receipt of psychotherapy and behavior reduction among long-stay nursing home residents with Alzheimer's disease and related dementias (ADRD).

DESIGN: Retrospective cohort study using Minimum Data Set assessments linked to Medicare: Master Beneficiary Summary File, Part B Carrier, and Part D Event file (2017-2018).

SETTING AND PARTICIPANTS: All US Medicare- or Medicaid-certified nursing homes. Traditional Medicare beneficiaries enrolled in Medicare Parts B and D, aged ≥65 with ADRD and any problematic behaviors (physical, verbal, or other), stratified to residents with both ADRD and co-occurring psychiatric disorders, and ADRD only.

METHODS: The unit of analysis was 2 consecutive quarters with indication of a problematic behavior in the first. Outcomes were reductions in any, physical, verbal, or other behaviors. The treatment was receipt of psychotherapy in both quarters vs neither. Covariates included predisposing, enabling, and need factors. Propensity score weighting balanced resident characteristics, and generalized estimating equation logistic models were applied.

RESULTS: The cohort included 175,165 resident-quarter observations from 99,584 unique long-stay residents with ADRD and problematic behaviors. Psychotherapy was received in 9% of observations, most often in short (73%), quarterly (59%), or monthly (35%) sessions delivered by psychologists (56%) or social workers (34%). After weighting, psychotherapy was associated with greater likelihood of behavior reduction (ADRD + psychiatric disorders: 17%, P < .01; ADRD only: 33%, P < .05). In ADRD-only residents, psychotherapy was associated with 36% reduction in physical behaviors and 28% in verbal behaviors. In those with ADRD + psychiatric disorders, reductions were seen in physical (31%) and other behaviors (18%).

CONCLUSIONS AND IMPLICATIONS: Fewer than 10% of nursing home residents with ADRD and problematic behaviors received psychotherapy, yet its receipt was significantly associated with behavior reduction. Findings underscore the need to expand access to psychotherapy in nursing homes and to further investigate the psychotherapy characteristics that contribute to effectiveness.

The lytic-latent balance is a major viral persistence strategy and obstacle to curing viral diseases, yet its mechanisms remain poorly understood. Following lytic infection in non-neuronal cells, herpes simplex virus (HSV) establishes latency specifically in neurons and is reactivated by stresses. Here we identify forkhead box (FOX) transcription factors that can strongly activate or repress replication of HSV-1 and other alphaherpesviruses, and show that neurons express activating Fox (e.g., Foxf) genes poorly but repressive Fox (Foxk) genes abundantly while non-neuronal or stressed neuronal cells exhibit higher expression of activating Fox genes. Remarkably, knockdown of Foxk1 or overexpression of activating Fox genes induces reactivation from latency in male mouse neuronal culture and in vivo. Of note, FOX proteins bind the viral genome globally and nonsequence-specifically and interact with epigenetic cofactors for gene regulation. FOXF1 interacts with CBP and P300 to acetylate and open viral chromatin. FOXK1 works with SIN3A, a cofactor of histone deacetylation, and MAX to suppress HSV-1 and antagonize activating FOX proteins. Therefore, the viral lytic-latent balance is controlled by the relative abundance of counteracting host transcription factors that recruit different epigenetic regulators to the viral genome.

The molecular pathways linking genetic variants to Parkinson's disease (PD) onset and progression remain incompletely defined; however, risk alleles in multiple genes, including GBA1, strongly implicate lipid metabolism. To systematically identify causal biomarker signatures, we analyzed comprehensive metabolome profiles from blood plasma in 149 PD patients and 150 controls, along with complementary genetic, RNA-sequencing, and metabolic data from other available clinical and pathologic cohorts. Using colocalization and summary-data-based Mendelian randomization, we tested whether expression and metabolic quantitative trait loci mediate the association between implicated genetic variants and PD risk. We further integrated differential metabolomics and proteomics from blood and brain to reveal pertinent mechanisms. We show that common PD risk variants at the serine palmitoyltransferase small subunit B (SPTSSB) locus, a key regulator of de novo sphingolipid biosynthesis, are associated with increased SPTSSB brain expression and elevated plasma ceramides. Additional analyses strongly support our hypothesis that a common SPTSSB causal variant is responsible for PD risk as well as the expression and metabolic quantitative trait loci. Multiple sphingolipids and fatty acid derivatives were perturbed in PD, and we identified both unique and shared features with the Alzheimer's disease metabolome. A PD acylcarnitine signature was further replicated in human postmortem brain tissue, when comparing those with or without preclinical Lewy body pathology. Integrated analysis of complementary brain proteomic profiles revealed dysregulation of mitochondrial processes dependent on acylcarnitines, including fatty acid beta-oxidation, the tricarboxylic acid cycle, and oxidative phosphorylation. Our results identify promising biomarkers and reveal a causal chain linking genetic variation to altered gene/protein expression, lipid dysmetabolism, and the manifestation of PD.

Cardiovascular adverse effects of drugs have significant practical implications for patient management. While cardiovascular adverse effects have commonly been associated with oncologic therapeutics, a growing body of evidence suggests that non-oncologic medications can also be associated with significant cardiovascular harm. These adverse effects range from arrhythmias, conduction abnormalities, QT prolongation, heart failure, myocardial infarction, or structural cardiomyopathy. Non-oncologic drugs that have been implicated include antibiotics (e.g. macrolides, fluoroquinolones), antidiabetics (e.g. thiazolidinediones), non-steroidal anti-inflammatory drugs, drugs for gastrointestinal and urological conditions, and most importantly, cardiovascular drugs. In this narrative review, we focus on the most common non-oncologic drugs that cause cardiovascular adverse effects, their proposed underlying mechanisms with particular emphasis on their clinical manifestations and clinical implications for everyday cardiovascular practice.

IMPORTANCE: Early and accurate identification of clinical warning signs in young infants may help avert sepsis morbidity and mortality in resource-limited settings.

OBJECTIVE: To systematically review evidence on the association and accuracy of clinical signs to diagnose sepsis or predict mortality in young infants aged 0 to 59 days to inform management in settings with limited laboratory diagnostics.

DATA SOURCES: MEDLINE, Embase, CINAHL, Global Index Medicus, and Cochrane CENTRAL Register were searched from inception through May 2023, with updated searches on September 5, 2024. An umbrella search of systematic reviews was conducted in January 2024.

STUDY SELECTION: Included studies reported data on 24 infant clinical signs informed by current World Health Organization (WHO) Integrated Management of Childhood Illness (IMCI) and hospital-based algorithms for the care of sick young infants reporting odds ratios (OR), risk ratios, or sensitivity and specificity.

DATA EXTRACTION AND SYNTHESIS: Data were extracted independently by 2 reviewers. Quality assessment used the Newcastle-Ottawa, Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2), and Quality Assessment of Prognostic Accuracy Studies (QUAPAS) scales. OR data were pooled using random-effects models. Data analysis was performed from July to September 2025.

MAIN OUTCOMES AND MEASURES: OR of all-cause mortality, culture-confirmed sepsis, or clinical sepsis (with access to laboratory investigations).

RESULTS: Of 7641 studies, 52 studies with 140 885 participants were included. A total of 16 clinical signs were significantly associated with mortality, 11 with culture-confirmed sepsis, and 13 with clinical sepsis. For mortality, the 5 strongest associations were weak, abnormal, or absent cry (OR, 20.48; 95% CI, 6.59-63.67); not able to feed at all (OR, 18.32; 95% CI, 6.00-55.97); not feeding well (OR, 13.39; 95% CI, 6.97-25.72); drowsiness or unconsciousness (OR, 12.46; 95% CI, 6.06-25.62); and prolonged capillary refill (OR, 12.06; 95% CI, 2.77-52.53). The top 5 signs associated with culture-confirmed sepsis were not feeding well (OR, 4.52; 95% CI, 1.10-18.59); prolonged capillary refill (OR, 3.59; 95% CI, 2.05-6.28); lethargy (OR, 3.44; 95% CI, 1.89-6.26); drowsiness or unconsciousness (OR, 3.07; 95% CI, 2.01-4.68); and feeding intolerance (OR, 2.95; 95% CI, 1.67-5.21).

CONCLUSIONS AND RELEVANCE: All current WHO IMCI clinical signs were significantly associated with mortality or culture-confirmed sepsis. Several signs not in IMCI were identified that may improve identification of life-threatening illness in young infants in resource-limited settings where clinical sign algorithms are the primary diagnostic tool.

One of the key pillars of Ending the HIV Epidemic is ensuring adherence to oral HIV pre-exposure prophylaxis (PrEP). Men who have sex with men (MSM) who also have substance use disorders experience multiple challenges to maintaining PrEP adherence. We developed a digital pill system (DPS) linked to a personalized adherence intervention, PrEPSteps, to address barriers to PrEP adherence, and tested the feasibility and acceptability of this system, as well as its potential for an effect on PrEP adherence. We enrolled MSM with moderate to severe substance use disorder who were on oral PrEP in a two-arm pilot randomized controlled trial. Both arms received the DPS co-encapsulated with oral PrEP. Participants in the intervention arm also received "PrEPSteps" - a personalized cognitive-behavioral adherence intervention. Primary outcomes were feasibility and acceptability of DPS + PrEPSteps. To explore potential intervention effects, adherence changes from baseline to 3-month follow-up were compared across arms. At 6-month follow-up, adherence was assessed via self-report. Thirty-six participants were enrolled, 32, completed the run-in period, 28 were randomized, and 27 completed the 3-month intervention period. Of those, 26 completed six-month follow-up. Operation of the DPS and PrEPSteps was feasible, with consistent data recording throughout the 3-month intervention period. Qualitative interviews in the intervention arm at 3 months demonstrated PrEPSteps was acceptable. Intervention arm participants had 14% higher PrEP adherence (b = 13.67, 95%CI [.77-26.57], p = .039) at 3 month follow up. This effect persisted at six months, suggesting that PrEPSteps has the potential to improve PrEP adherence and help individuals sustain adherence benefits over time.Trial registration: www.ClinicalTrials.gov identifier: NCT03512418.

INTRODUCTION: Patients admitted for hip fracture surgery may receive care from a team led either by the operating surgeon or a hospitalist.

OBJECTIVE: To describe the prevalence of the hospitalist care model for hip fracture admissions and its association with patient outcomes.

DESIGN: We conducted a retrospective cohort study of patients admitted in 2018-2019 for fracture of the head and neck of the femur (ICD10 S72.0x-2x). We compared outcomes at hospitals with low versus high use of the hospitalist care model. Our exposure was the hospital-level adoption of hospitalist care, categorized into quartiles.

PARTICIPANTS: Fee-for-service Medicare patients 66 years old or greater.

MAIN MEASURES: Length of stay (LOS), professional services (Part B) inpatient spending, specialty consultation, discharge to home, all-cause 7- and 30-day readmissions, and 30-day mortality.

KEY RESULTS: A total of 294,150 patients with hip fracture were admitted to 2466 hospitals. Patients cared for in low-use (Q1) versus high-use (Q4) hospitals did not differ meaningfully in demographic characteristics or comorbidities. Hospitals ranged in use of the hospitalist care model from 12% in low-use (Q1) hospitals to 81% in high-use (Q4) hospitals. Low-use hospitals had significantly higher inpatient consult use (unadjusted: Q1 vs Q4, 1.06 vs 0.63 consults, p < 0.0001; adjusted: -0.36, p < 0.001) and length of stay (unadjusted: Q1 vs Q4, 6.04 vs 5.94 days, p < 0.0001; adjusted: -0.09 days, p < 0.05), but no significant difference in adjusted analyses for spending, likelihood of discharge home, 7- and 30-day readmission, or 30-day mortality.

CONCLUSIONS: Hospitalist care for older adults admitted for hip fracture surgery is both common and associated with slightly shorter length of stay.

We describe the development and delivery of a seminar to train emerging geriatric leaders from diverse disciplines in quality improvement science utilizing the Institute for Healthcare Improvement Model for Improvement. We implemented a six-session synchronous virtual quality improvement seminar employing a "flipped classroom" approach consistent with experiential learning, involving pre-work, in-session application through exercises and work-time, and homework, over three cohorts of learners (2022-2024). Content was adapted through formative evaluation in year one, with additional refinements in years two and three. Of those who completed a summative evaluation survey in year three (N = 23, 46% response), 60.8% pursued a quality improvement project described as completed (21.7%) or in progress (39.1%); 21.7% had presentations or publications about their projects. Most participants rated seminar topics and tools as somewhat or very useful (on a 1-3 scale), particularly the aims statement (M = 2.95, SD = 0.22), and project planning tools (M = 2.81, SD = 0.40). Participants agreed or strongly agreed they were using seminar information now (82.6%) and/or in the future (91.3%), and that their project improved patient care (60.8%). A quality improvement seminar for interprofessional emerging geriatric leaders was well received and may contribute to dissemination of scholarly products and improvements in patient care.

Hepatitis C virus (HCV) infection is associated with unfavorable multidrug- and rifampicin-resistant (MDR/RR) tuberculosis (TB) outcomes. We examined whether this association would decrease in settings where no participants were lost-to-follow-up or where all adhered to regimens comprised of priority TB drugs. We analyzed data from 1530 participants with HCV testing in the endTB observational cohort (NCT03259269). We estimated the relative risk of death, treatment failure, and loss-to-follow-up comparing participants with and without HCV, using inverse probability weighting to adjust for confounding. We then estimated relative risks of HCV on death and failure in weighted pseudopopulations representing hypothetical interventions eliminating loss-to-follow-up and ensuring adherence to strong MDR/RR-TB regimens. The unadjusted risk difference comparing participants with and without HCV was 14.1% (95% confidence interval [CI] 8.0%, 20.1%), decreasing to 11.0% (95%CI, 3.0%, 19.1%) after weighting. In pseudopopulations without loss-to-follow-up or with adequate adherence to strong regimens, the risk differences were 7.7% (95% CI, 0.8%, 16.2%) and 7.0% (95% CI, -1.6%, 17.3%), respectively. Adjustment for baseline confounders attenuated the association between HCV and unfavorable outcomes, suggesting these factors partly explain the disparity. Further attenuation after eliminating loss-to-follow-up suggests that improving treatment retention in MDR/RR-TB care may reduce outcome disparities among patients with HCV.