Telehealth Access and Substitution in the VHA.

BACKGROUND: In response to COVID-19, the Veterans Health Administration (VHA) expanded telehealth availability, allowing veterans to receive care at home. We explore the extent of substitution of telehealth for in-person care among medical centers (facilities), providers, and patients. We explore the extent to which patient preferences drive telehealth utilization, and compare access to care (as measured by waiting times) for telehealth and in-person visits. METHODS: We use VHA electronic health records to identify scheduled outpatient mental health (MH) appointments from January 2019 through February 2023 focusing on care delivered by social workers, psychologists, and psychiatrists. For each quarter, we compute the proportion of completed appointments that were delivered via phone or video by each facility, provider, and patient and show the changes in these proportions before, during, and after the onset of COVID-19. To explore patient preferences, we match providers of patients with high rates of telehealth utilization and examine the extent to which those providers deliver in-person care. To examine access to care, we compute waiting times for in-person, video, and phone new patient appointments. We investigate differences between urban and rural patients, and patients of different ages. KEY RESULTS: Telehealth for MH grew dramatically in the VHA after the onset of COVID-19. While some facilities provided more telehealth than others, all facilities (as of early 2023) provided some telehealth MH services. Approximately 86% of individual providers provided telehealth, with 27% scheduling MH appointments almost exclusively as telehealth appointments and 59% providing a mix. Patients exhibited more polarization, with 36% scheduling only in-person visits for almost all their MH visits and 56% of them scheduling exclusively telehealth, and only 8% of them utilizing a mix of modalities. Of those who exclusively received telehealth care, a majority of them utilized video (80%) over phone (20%). Take-up of MH among younger patients was higher relative to older patients. Urban patients used telehealth more than rural patients. Patient preferences rather than provider preferences drove utilization of patients who almost exclusively utilized telehealth. Between April 2021 and February 2023, the average difference in waiting time for in-person and video appointments was less than 1 day, with comparable appointment volumes, suggesting that the supply of and demand for in-person and video were not different enough to merit waiting longer. Telehealth was chosen over in-person more among urban and younger patients, as older and rural patients exhibited higher willingness to wait for in-person over video appointments. By contrast, appointment volumes and waiting times for phone appointments were lower across all groups, suggesting that phone may not be as substitutable for in-person visits in MH. CONCLUSIONS: We find that the VHA has made telehealth widely available, providing access to many veterans. While telehealth utilization has increased, face-to-face care persists for MH services, suggesting that one modality may not serve all purposes and preferences for care. Patient preferences drive the modality decision among those who exclusively use MH care via telehealth. For those who persist in mostly utilizing in-person care, there may be various factors influencing those preferences such as issues with limited internet connectivity, language barriers, and digital literacy, especially for older and rural patients who utilize in-person care more than those who are younger and more urban. Further investigation is required to investigate the optimal mix of modalities which may allow for potential increases in patient satisfaction, quality of care, and clinic efficiency.

Clinical effectiveness and cost-impact after 2 years of a ketogenic diet and virtual coaching intervention for patients with diabetes.

AIM: We previously evaluated the impacts at 5 months of a digitally delivered coaching intervention in which participants are instructed to adhere to a very low carbohydrate, ketogenic diet. With extended follow-up (24 months), we assessed the longer-term effects of this intervention on changes in clinical outcomes, health care utilization and costs associated with outpatient, inpatient and emergency department use in the Veterans Health Administration. MATERIALS AND METHODS: We employed a difference-in-differences model with a waiting list control group to estimate the 24-month change in glycated haemoglobin, body mass index, blood pressure, prescription medication use, health care utilization rates and associated costs. The analysis included 550 people with type 2 diabetes who were overweight or obese and enrolled in the Veterans Health Administration for health care. Data were obtained from electronic health records from 2018 to 2021. RESULTS: The virtual coaching and ketogenic diet intervention was associated with significant reductions in body mass index [-1.56 (SE 0.390)] and total monthly diabetes medication usage [-0.35 (SE 0.054)]. No statistically significant differences in glycated haemoglobin, blood pressure, outpatient visits, inpatient visits, or emergency department visits were observed. The intervention was associated with reductions in per-patient, per-month outpatient spending [-USD286.80 (SE 97.175)] and prescription drug costs (-USD105.40 (SE 30.332)]. CONCLUSIONS: A virtual coaching intervention with a ketogenic diet component offered modest effects on clinical and cost parameters in people with type 2 diabetes and with obesity or overweight. Health care systems should develop methods to assess participant progress and engagement over time if they adopt such interventions, to ensure continued patient engagement and goal achievement.

Vector-based kernel weighting: A simple estimator for improving precision and bias of average treatment effects in multiple treatment settings.

Treatment effect estimation must account for observed confounding, in which factors affect treatment assignment and outcomes simultaneously. Ignoring observed confounding risks concluding that a helpful treatment is not beneficial or that a treatment is safe when actually harmful. Propensity score matching or weighting adjusts for observed confounding, but the best way to use propensity scores for multiple treatments is unknown. It is unclear when choice of a different weighting or matching strategy leads to divergent inferences. We used Monte Carlo simulations (1000 replications) to examine sensitivity of multivalued treatment inferences to propensity score weighting or matching strategies. We consider five variants of propensity score adjustment: inverse probability of treatment weights, generalized propensity score matching, kernel weights (KW), vector matching, and a new hybrid that is easily implemented-vector-based kernel weighting (VBKW). VBKW matches observations with similar propensity score vectors, assigning greater KW to observations with similar probabilities within a given bandwidth. We varied degree of propensity score model misspecification, sample size, treatment effect heterogeneity, initial covariate imbalance, and sample distribution across treatment groups. We evaluated sensitivity of results to propensity score estimation technique (multinomial logit or multinomial probit). Across simulations, VBKW performed equally or better than the other methods in terms of bias, efficiency, and covariate balance measured via prognostic scores. Our simulations suggest that VBKW is amenable to full automation and is less sensitive to PS model misspecification than other methods used to account for observed confounding in multivalued treatment analyses.