Publications

BACKGROUND: Newer glucose-lowering drugs, including sodium glucose co-transporter 2 inhibitors (SGLT2i) and GLP-1 agonists, have a key role in the pharmacologic management of type 2 diabetes. No studies have measured primary nonadherence for these two drug classes, defined as when a medication is prescribed for a patient but ultimately not dispensed to them.

OBJECTIVE: To describe the incidence and predictors of primary nonadherence to SGLT2i (canagliflozin, empagliflozin) or GLP-1 agonists (dulaglutide, liraglutide, semaglutide) using a dataset that links electronic prescribing with health insurance claims.

DESIGN AND PARTICIPANTS: A retrospective cohort design using data of adult patients from a large health system who had at least one prescription order for a SGLT2i or GLP-1 agonist between 2012 and 2019. We used mixed-effects multivariable logistic regression to determine associations between sociodemographic, clinical, and provider variables and primary nonadherence.

MAIN MEASURES: Primary medication nonadherence, defined as no dispensed claim within 30 days of an electronic prescription order for any drug within each medication class.

KEY RESULTS: The cohort included 5146 patients newly prescribed a SGLT2i or GLP-1 agonist. The overall incidence of 30-day primary medication nonadherence was 31.8% (1637/5146). This incidence rate was 29.8% (n = 726) and 33.6% (n = 911) among those initiating a GLP-1 agonist and SGLT2i, respectively. Age ≥ 65 (aOR 1.37 (95% CI 1.09 to 1.72)), Black race vs White (aOR 1.29 (95% CI 1.02 to 1.62)), diabetic nephropathy (aOR 1.31 (95% CI 1.02 to 1.68)), and hyperlipidemia (aOR 1.18 (95% CI 1.01 to 1.39)) were associated with a higher odds of primary nonadherence. Female sex (aOR 0.86 (95% CI 0.75 to 0.99)), peripheral artery disease (aOR 0.73 (95% CI 0.56 to 0.94)), and having the index prescription ordered by an endocrinologist vs a primary care provider (aOR 0.76 (95% CI 0.61 to 0.95)) were associated with lower odds of primary nonadherence.

CONCLUSIONS: One third of patients prescribed SGLT2i or GLP-1 agonists in this sample did not fill their prescription within 30 days. Black race, male sex, older age, having greater baseline comorbidities, and having a primary care provider vs endocrinologist prescribe the index drug were associated with higher odds of primary nonadherence. Interventions targeting medication adherence for these newer drugs must consider primary nonadherence as a barrier to optimal clinical care.

OBJECTIVES: Limited information is available regarding provider- and patient panel-level factors associated with primary care provider (PCP) adoption/prescribing of medication for opioid use disorder (MOUD).

METHODS: We assessed a retrospective cohort from 2015 to 2018 within the Pennsylvania Medicaid Program. Participants included PCPs who were Medicaid providers, with no history of MOUD provision, and who treated ≥10 Medicaid enrollees annually. We assessed initial MOUD adoption, defined as an index buprenorphine/buprenorphine-naloxone or oral/extended release naltrexone fill and sustained prescribing, defined as ≥1 MOUD prescription(s) for 3 consecutive quarters from the PCP. Independent variables included provider- and patient panel-level characteristics.

RESULTS: We identified 113 rural and 782 urban PCPs who engaged in initial adoption and 36 rural and 288 urban PCPs who engaged in sustained prescribing. Rural/urban PCPs who issued increasingly larger numbers of antidepressant and antipsychotic medication prescriptions had greater odds of initial adoption and sustained prescribing (P < 0.05) compared to those that did not prescribe these medications. Further, each additional patient out of 100 with opioid use disorder diagnosed before MOUD adoption increased the adjusted odds for initial adoption 2% to 4% (95% confidence interval [CI] = 1.01-1.08) and sustained prescribing by 4% to 7% (95% CI = 1.01-1.08). New Medicaid providers in rural areas were 2.52 (95% CI = 1.04-6.11) and in urban areas were 2.66 (95% CI = 1.94, 3.64) more likely to engage in initial MOUD adoption compared to established PCPs.

CONCLUSIONS: MOUD prescribing adoption was concentrated among PCPs prescribing mental health medications, caring for those with OUD, and new Medicaid providers. These results should be leveraged to test/implement interventions targeting MOUD adoption among PCPs.

BACKGROUND AND AIMS: The time lag encountered when accessing health-care data is one major barrier to implementing opioid overdose prediction measures in practice. Little is known regarding how one's opioid overdose risk changes over time. We aimed to identify longitudinal patterns of individual predicted overdose risks among Medicaid beneficiaries after initiation of opioid prescriptions.

DESIGN, SETTING AND PARTICIPANTS: A retrospective cohort study in Pennsylvania, USA among Pennsylvania Medicaid beneficiaries aged 18-64 years who initiated opioid prescriptions between July 2017 and September 2018 (318 585 eligible beneficiaries (mean age = 39 ± 12 years, female = 65.7%, White = 62.2% and Black = 24.9%).

MEASUREMENTS: We first applied a previously developed and validated machine-learning algorithm to obtain risk scores for opioid overdose emergency room or hospital visits in 3-month intervals for each beneficiary who initiated opioid therapy, until disenrollment from Medicaid, death or the end of observation (December 2018). We performed group-based trajectory modeling to identify trajectories of these predicted overdose risk scores over time.

FINDINGS: Among eligible beneficiaries, 0.61% had one or more occurrences of opioid overdose in a median follow-up of 15 months. We identified five unique opioid overdose risk trajectories: three trajectories (accounting for 92% of the cohort) had consistent overdose risk over time, including consistent low-risk (63%), consistent medium-risk (25%) and consistent high-risk (4%) groups; another two trajectories (accounting for 8%) had overdose risks that substantially changed over time, including a group that transitioned from high- to medium-risk (3%) and another group that increased from medium- to high-risk over time (5%).

CONCLUSIONS: More than 90% of Medicaid beneficiaries in Pennsylvania USA with one or more opioid prescriptions had consistent, predicted opioid overdose risks over 15 months. Applying opioid prediction algorithms developed from historical data may not be a major barrier to implementation in practice for the large majority of individuals.

STUDY OBJECTIVE: High prescribers of antibiotics and opioids are an important target for stewardship interventions. The goal of this study was to assess the association between high antibiotic and high opioid prescribing by provider type.

DESIGN: A national cross-sectional study.

SETTING: 2015-2017 Department of Veterans Affairs (VA) electronic health record data.

POPULATION: Prescribers were identified as dentists (2017: n = 1346) and medical providers (physicians n = 23,072; advanced practice providers [APP] n = 7705; and other providers [pharmacists/chiropractors] n = 3674) (2017: n = 34,451).

MEASUREMENTS: High prescribing was defined as being in the top 25% of visit-based rates of antibiotic or opioid prescribing (number of prescriptions/number of dental or medical visits). Multivariable random effects logistic regression with clustering by facility was used to assess the adjusted association between high antibiotic and opioid prescribing.

RESULTS: Medical providers prescribed 4,348,670 antibiotic and 10,256,706 opioid prescriptions; dentists prescribed 277,170 antibiotic and 124,103 opioid prescriptions. Among all high prescribers of antibiotics, 40% were also high prescribers of opioids as compared to 18% of those who were not high antibiotic prescribers (p < 0.0001). High prescribing of antibiotics was associated with high prescribing of opioids in medical providers (adjusted odds ratio [aOR] = 2.87, 95% confidence interval [CI] = 2.72-3.04) and dentists (aOR = 8.40, 95% CI 6.00-11.76). Older provider age, specific US geographic regions, and lower VA facility complexity and rurality were also associated with high opioid prescribing by medical providers. In dentists, younger provider age, male gender, specific regions of the United States, and lower number of dentists in a facility were associated with high opioid prescribing. At the facility level, high dental prescribers of antibiotics or opioids were not at the same facilities as high medical prescribers, respectively (p < 0.0001).

CONCLUSIONS: High antibiotic prescribing was associated with high opioid prescribing. Thus, stewardship interventions targeting both medication classes may have higher impact to efficiently reduce prescribing of medications with high public health impact. Provider-targeted interventions are needed to improve antibiotic and opioid prescribing in both dentists and medical providers.

BACKGROUND: The drug supply chain is global and at risk of disruption and subsequent drug shortages, especially during unanticipated events.

OBJECTIVE: Our objective was to determine the impact of coronavirus disease 2019 (COVID-19) on drug purchases overall, by class, and for specific countries.

METHODS: A cross-sectional time series analysis of country-level drug purchase data from August 2014 to August 2020 from IQVIA MIDAS was conducted. Standardized units per 100 population and percentage increase in units purchased were assessed from 68 countries and jurisdictions in March 2020 (when the World Health Organization declared COVID-19 a pandemic). Analyses were compared by United Nations development status and drug class. Autoregressive integrated moving average models tested the significance of changes in purchasing trends.

RESULTS: Before COVID-19, standardized medication units per 100 population ranged from 3990 to 4760 monthly. In March 2020, there was a global 15% increase in units of drugs purchased to 5309.3 units per 100 population compared with the previous year; the increase was greater in developed countries (18.5%; P < 0.001) than in developing countries (12.8%; P < 0.0001). After the increase in March 2020, there was a correction in the global purchase rate decreasing by 4.7% (April to August 2020 rate, 21,334.6/100 population; P < 0.001). Globally, we observed high purchasing rates and large changes for respiratory medicines such as inhalers and systemic adrenergic drugs (March 2020 rate, 892.7/100 population; change from 2019, 28.5%; P < 0.001). Purchases for topical dermatologic products also increased substantially (42.2%), although at lower absolute rates (610.0/100 population in March 2020; P < 0.0001). Interestingly, purchases for systemic anti-infective agents (including antiviral drugs) increased in developing countries (11.3%; P < 0.001), but decreased in developed countries (1.0%; P = 0.06).

CONCLUSION: We observed evidence of global drug stockpiling in the early months of the COVID-19 pandemic, especially among developed countries. Actions toward equitable distribution of medicines through a resilient drug supply chain should be taken to increase global response to future unanticipated events, such as pandemics.

IMPORTANCE: Within the Veterans Health Administration (VA), the use and cost of low-value services delivered by VA facilities or increasingly by VA Community Care (VACC) programs have not been comprehensively quantified.

OBJECTIVE: To quantify veterans' overall use and cost of low-value services, including VA-delivered care and VA-purchased community care.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study assessed a national population of VA-enrolled veterans. Data on enrollment, sociodemographic characteristics, comorbidities, and health care services delivered by VA facilities or paid for by the VA through VACC programs were compiled for fiscal year 2018 from the VA Corporate Data Warehouse. Data analysis was conducted from April 2020 to January 2022.

MAIN OUTCOMES AND MEASURES: VA administrative data were applied using an established low-value service metric to quantify the use of 29 potentially low-value tests and procedures delivered in VA facilities and by VACC programs across 6 domains: cancer screening, diagnostic and preventive testing, preoperative testing, imaging, cardiovascular testing and procedures, and other procedures. Sensitive and specific criteria were used to determine the low-value service counts per 100 veterans overall, by domain, and by individual service; count and percentage of each low-value service delivered by each setting; and estimated cost of each service.

RESULTS: Among 5.2 million enrolled veterans, the mean (SD) age was 62.5 (16.0) years, 91.7% were male, 68.0% were non-Hispanic White, and 32.3% received any service through VACC. By specific criteria, 19.6 low-value services per 100 veterans were delivered in VA facilities or by VACC programs, involving 13.6% of veterans at a total cost of $205.8 million. Overall, the most frequently delivered low-value service was prostate-specific antigen testing for men aged 75 years or older (5.9 per 100 veterans); this was also the service with the greatest proportion delivered by VA facilities (98.9%). The costliest low-value services were spinal injections for low back pain ($43.9 million; 21.4% of low-value care spending) and percutaneous coronary intervention for stable coronary disease ($36.8 million; 17.9% of spending).

CONCLUSIONS AND RELEVANCE: This cross-sectional study found that among veterans enrolled in the VA, more than 1 in 10 have received a low-value service from VA facilities or VACC programs, with approximately $200 million in associated costs. Such information on the use and costs of low-value services are essential to guide the VA's efforts to reduce delivery and spending on such care.

BACKGROUND: Dentists in the United States frequently prescribe opioids for dental-related pain, although evidence shows superior efficacy of nonopioids for pain management. A national sample of US dentists was interviewed to understand the barriers and facilitators to opioid prescribing.

METHODS: Semistructured one-on-one telephone interviews were conducted with dentists sampled from the 6 regions of The National Dental Practice-Based Research Network. Responses were coded into the domains of the Capability, Opportunity and Motivation Model of Behavior. Potential behavior change interventions were identified for targeted themes.

RESULTS: Seventy-three interviews were qualitatively analyzed. Most of those interviewed were general dentists (86.3%) and on average (SD) were in practice for 24.3 (13.0) years. Ten themes were identified within the Capability, Opportunity and Motivation Model of Behavior. Dentists' knowledge of opioid risk, ability to identify substance use disorder behavior, and capability of communicating pain management plans to patients or following clinic policies or state and federal regulations were linked with judicious opioid prescribing. Dentists reported prescribing opioids if they determined clinical necessity or feared negative consequences for refusing to prescribe opioids.

CONCLUSIONS: Dentists' opioid decision making is influenced by a range of real-world practice experiences and patient and clinic factors. Education and training that target dentists' knowledge gaps and changes in dentists' practice environment can encourage effective communication of pain management strategies with patients and prescribing of nonopioids as first-line analgesics while conserving opioid use.

PRACTICAL IMPLICATIONS: Identified knowledge gaps in dentistry can be targets for education, clinical guidelines, and policy interventions to ensure safe and appropriate prescribing of opioids.

BACKGROUND: Benzodiazepines, opioids, proton-pump inhibitors (PPIs), and antibiotics are frequently prescribed inappropriately by primary care physicians (PCPs), without sufficient consideration of alternative options or adverse effects. We hypothesized that distinct groups of PCPs could be identified based on their propensity to prescribe these medications.

OBJECTIVE: To identify PCP groups based on their propensity to prescribe benzodiazepines, opioids, PPIs, and antibiotics, and patient and PCP characteristics associated with identified prescribing patterns.

DESIGN: Retrospective cohort study using VA data and latent class regression analyses to identify prescribing patterns among PCPs and examine the association of patient and PCP characteristics with class membership.

PARTICIPANTS: A total of 2524 full-time PCPs and their patient panels (n = 2,939,636 patients), from January 1, 2017, to December 31, 2018.

MAIN MEASURES: We categorized PCPs based on prescribing volume quartiles for the four drug classes, based on total days' supply dispensed of each medication by the PCP to their patients (expressed as days' supply per 1000 panel patient-days). We used latent class analysis to group PCPs based on prescribing and used multinomial logistic regression to examine patient and PCP characteristics associated with latent class membership.

KEY RESULTS: PCPs were categorized into four groups (latent classes): low intensity (23% of cohort), medium-intensity overall/high-intensity PPI (36%), medium-intensity overall/high-intensity opioid (20%), and high intensity (21%). PCPs in the high-intensity group were predominantly in the highest quartile of prescribers for all four drugs (68% in the highest quartile for benzodiazepine, 86% opioids, 64% PPIs, 62% antibiotics). High-intensity PCPs (vs. low intensity) were substantially less likely to be female (OR: 0.30, 95% CI: 0.21-0.42) or practice in the northeast versus other census regions (OR: 0.10, 95% CI: 0.06-0.17).

CONCLUSIONS: VA PCPs can be classified into four clearly differentiated groups based on their prescribing of benzodiazepines, opioids, PPIs, and antibiotics, suggesting an underlying typology of prescribing. High-intensity PCPs were more likely to be male.

IMPORTANCE: Metrics that detect low-value care in common forms of health care data, such as administrative claims or electronic health records, primarily focus on tests and procedures but not on medications, representing a major gap in the ability to systematically measure low-value prescribing.

OBJECTIVE: To develop a scalable and broadly applicable metric that contains a set of quality indicators (EVOLV-Rx) for use in health care data to detect and reduce low-value prescribing among older adults and that is informed by diverse stakeholders' perspectives.

DESIGN, SETTING, AND PARTICIPANTS: This qualitative study used an online modified-Delphi method to convene an expert panel of 15 physicians and pharmacists. This panel, comprising clinicians, health system leaders, and researchers, was tasked with rating and discussing candidate low-value prescribing practices that were derived from medication safety criteria; peer-reviewed literature; and qualitative studies of patient, caregiver, and physician perspectives. The RAND ExpertLens online platform was used to conduct the activities of the panel. The panelists were engaged for 3 rounds between January 1 and March 31, 2021.

MAIN OUTCOMES AND MEASURES: Panelists used a 9-point Likert scale to rate and then discuss the scientific validity and clinical usefulness of the criteria to detect low-value prescribing practices. Candidate low-value prescribing practices were rated as follows: 1 to 3, indicating low validity or usefulness; 3.5 to 6, uncertain validity or usefulness; and 6.5 to 9, high validity or usefulness. Agreement among panelists and the degree of scientific validity and clinical usefulness were assessed using the RAND/UCLA (University of California, Los Angeles) Appropriateness Method.

RESULTS: Of the 527 low-value prescribing recommendations identified, 27 discrete candidate low-value prescribing practices were considered for inclusion in EVOLV-Rx. After round 1, 18 candidate practices were rated by the panel as having high scientific validity and clinical usefulness (scores of ≥6.5). After round 2 panel deliberations, the criteria to detect 19 candidate practices were revised. After round 3, 18 candidate practices met the inclusion criteria, receiving final median scores of 6.5 or higher for both scientific validity and clinical usefulness. Of those practices that were not included in the final version of EVOLV-Rx, 3 received high scientific validity (scores ≥6.5) but uncertain clinical usefulness (scores <6.5) ratings, whereas 6 received uncertain scientific validity rating (scores <6.5).

CONCLUSIONS AND RELEVANCE: This study culminated in the development of EVOLV-Rx and involved a panel of experts who identified the 18 most salient low-value prescribing practices in the care of older adults. Applying EVOLV-Rx may enhance the detection of low-value prescribing practices, reduce polypharmacy, and enable older adults to receive high-value care across the full spectrum of health services.

OBJECTIVE: Guidelines advocate against tight glycemic control in older nursing home (NH) residents with advanced dementia (AD) or limited life expectancy (LLE). We evaluated the effect of deintensifying diabetes medications with regard to all-cause emergency department (ED) visits, hospitalizations, and death in NH residents with LLE/AD and tight glycemic control.

RESEARCH DESIGN AND METHODS: We conducted a national retrospective cohort study of 2,082 newly admitted nonhospice veteran NH residents with LLE/AD potentially overtreated for diabetes (HbA1c ≤7.5% and one or more diabetes medications) in fiscal years 2009-2015. Diabetes treatment deintensification (dose decrease or discontinuation of a noninsulin agent or stopping insulin sustained ≥7 days) was identified within 30 days after HbA1c measurement. To adjust for confounding, we used entropy weights to balance covariates between NH residents who deintensified versus continued medications. We used the Aalen-Johansen estimator to calculate the 60-day cumulative incidence and risk ratios (RRs) for ED or hospital visits and deaths.

RESULTS: Diabetes medications were deintensified for 27% of residents. In the subsequent 60 days, 28.5% of all residents were transferred to the ED or acute hospital setting for any cause and 3.9% died. After entropy weighting, deintensifying was not associated with 60-day all-cause ED visits or hospitalizations (RR 0.99 [95% CI 0.84, 1.18]) or 60-day mortality (1.52 [0.89, 2.81]).

CONCLUSIONS: Among NH residents with LLE/AD who may be inappropriately overtreated with tight glycemic control, deintensification of diabetes medications was not associated with increased risk of 60-day all-cause ED visits, hospitalization, or death.