Publications

AIMS: We evaluated the relationship between the timing of insulin initiation and cardiovascular diseases (CVD) risk in Pennsylvania Medicaid enrollees with type 2 diabetes (T2D).

METHODS: We included 17,873 enrollees (age 47.4 ± 10.3 years; range 18-64 years) initially treated with non-insulin glucose-lowering agents (GLAs) in 2008-2016. Based on clinical guidelines, we identified early (N = 1,158; 6%; insulin initiation ≤ 6 months after first-line GLAs), in-time (N = 569; 3%; 6-12 months), delayed (N = 2,761; 15%; >12 months), and non-insulin users (N = 13,385; 75%). The Prentice-Williams-Peterson (PWP) models with inverse probability weighting estimated CVD risk across the four groups and the change in risk after insulin initiation.

RESULTS: Regardless of time to insulin initiation, insulin users had higher CVD risks after first-line GLAs than non-insulin users (aHR: early: 2.0 [1.5-2.5], in-time: 1.8 [1.2-2.6], delayed: 1.9 [1.6-2.3]). However, we found only a borderline increase in CVD risk after insulin initiation vs. before in early (aHR: 1.4 [1.1-1.8]) and delayed users (aHR: 1.3 [1.0-1.7]), and no increase in in-time users (aHR: 1.3 [0.9-2.0]).

CONCLUSIONS: We observed no gains in CVD benefits from insulin initiation in the early stages of pharmacotherapy possibly because CVD developed before insulin initiation. Additional management of hypertension and dyslipidemia may be important to reduce CVD risk in this young and middle-aged T2D cohort.

IMPORTANCE: State decisions not to expand Medicaid under the Patient Protection and Affordable Care Act could reduce emergency access to acute care hospitals.

OBJECTIVE: To determine the relationship between state Medicaid expansion and emergency access to acute care hospitals in the United States.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study linked hospital-level data from the Centers for Medicare & Medicaid Services from 2007 to 2017 to US Census data for all 50 US states and the District of Columbia. Geospatial analyses and difference-in-differences regression models were used to compare temporal changes in the size of the population without 30-minute access to acute care hospitals between 32 states that expanded Medicaid with the population without access in 19 that did not, before and after expansion. Analyses focused on the total population and those with low incomes; secondary analyses examined emergency access to safety-net hospitals.

EXPOSURES: State-level Medicaid expansion.

MAIN OUTCOMES AND MEASURES: Population without emergency access to an acute care hospital, defined as living outside a 30-minute drive of any hospital.

RESULTS: States that did not expand Medicaid experienced an increase in the population without access to hospitals overall (without expansion: 6.76% to 6.79% [0.03%]; vs with expansion: 5.65% to 5.35% [-0.30%]; difference-in-differences, 0.33%; 95% CI, 0.33%-0.34%; P < .001) and for low-income persons (without expansion: 7.43% to 7.39% [-0.04%]; vs with expansion: 6.25% to 6.15% [-0.10%]; difference-in-differences, 0.06%; 95% CI, 0.05%-0.07%; P < .001). If access changes in nonexpansion states were the same as expansion states, an estimated 421 000 more persons overall and 48 000 more low-income persons would have retained access. States that did not expand Medicaid experienced an increase in the population without access to safety-net hospitals overall (46.91% to 47.70% [0.79%] vs 33.94% to 33.07% [-0.87%]; difference-in-differences, 1.66%; 95% CI, 1.64%-1.66%; P < .001) and for low-income persons (45.28% to 46.14% [0.86%] vs 33.00% to 32.23% [-0.77%]; difference-in-differences, 1.63%; 95% CI, 1.63%-1.67%; P < .001). If access changes in nonexpansion states were the same as expansion states, an estimated 2 242 000 more persons overall and 364 000 more low-income persons would have retained access.

CONCLUSIONS AND RELEVANCE: States that did not expand Medicaid under the Patient Protection and Affordable Care Act were associated with worse emergency access to acute care hospitals compared with states that expanded Medicaid.

OBJECTIVE: This study sought to characterize recent trends in mental health visits of adult outpatients to primary care physicians (PCPs), specialty mental health providers (SMHPs), and other providers (non-primary care physicians, specialists other than SMHPs, nurse practitioners, and physician assistants). Trends determined by degree of patients' psychological distress and in the types of treatments received within different settings were also examined.

METHODS: Data were from the household component of the nationally representative Medical Expenditure Panel Survey for the 2008-2011 and 2012-2015 periods for adults ages ≥18 years (N=13,111) who had a mental health outpatient visit. Bivariate logistic regression was used to compare means between the two periods.

RESULTS: The percentage of adults having mental health outpatient visits increased between the two periods, largely driven by an increase in visits with providers other than SMHPs and PCPs, which rose from 11.9% (N=667) to 15.5% (N=1,048). Outpatient mental health visits with PCPs decreased from 29.0% (N=1,802) to 26.8% (N=1,945). The proportion of respondents with mental health outpatient visits increased both among those with high psychological distress and among those with low or no psychological distress (from 30.7% [N=1,332] to 36.2% [N=1,491] and from 6.0% [N=4,516] to 6.9% [N=5,772], respectively). The percentage of respondents receiving only psychotropic medication decreased over the two periods.

CONCLUSIONS: Mental health outpatient visits for adults increased between 2008 and 2015, and visits with SMHPs remained relatively stable during that time. A greater understanding of recent trends in types of outpatient mental health services may help identify targets for future mental health workforce studies.

Objectives: Using data from the Healthcare Cost and Utilization Project (HCUP), we estimated prevalence of individual substance use disorders (SUDs) and psychiatric comorbidities among pregnant women with opioid use disorder (OUD) in the New York State from 2009 to 2014.Methods: In this cross-sectional study, pregnancy outcome and gestational age at delivery were estimated, and OUD diagnosis during pregnancy or at delivery discharge was identified. Prevalence of SUDs and psychiatric comorbidities were then calculated.Results: Among 1,463,302 pregnant women, 8324 (0.57%) were diagnosed with OUD during pregnancy or at delivery. The most frequent SUDs or psychiatric comorbidities among pregnant women with OUD were non-opioid SUD (78.2%), followed by tobacco use disorder (74.9%), generalized anxiety disorder (38.0%), major depressive disorder (36.9%), cannabis use disorder (28.3%) and cocaine use disorder (27.4%).Conclusions: Most pregnant women with OUD were diagnosed with at least one non-opioid SUD and tobacco use disorder. Generalized anxiety disorder and major depressive disorder were also common, suggesting that mental health screenings should be prioritized for pregnant women with OUD.

PURPOSE: Compared with conventional therapy (enoxaparin followed by warfarin), the direct-acting oral anticoagulant apixaban is thought to offer similar protection against recurrent venous thromboembolism (VTE) with lower bleeding risk. However, evidence regarding the heterogeneity of treatment effect from real-world data is lacking. The study described here aimed to compare the effectiveness and safety of use of apixaban versus warfarin in patients with VTE.

METHODS: We conducted a retrospective cohort analysis of commercial and Medicare supplemental databases (data coverage period, 2014-2017) among patients with a diagnosis of VTE who were new users of apixaban or warfarin. We controlled for confounding using propensity score [PS] 1:4 matching. Cox proportional hazard models were used to obtain hazard ratios (HRs) and 95% confidence intervals (CIs). Heterogeneity of treatment effect was assessed among patients with provoked VTE versus unprovoked VTE.

RESULTS: After PS matching, a total of 36,907 patients were included in the cohort (n = 8,094 apixaban users and n = 28,813 warfarin users). In Cox regression models, the use of apixaban versus warfarin was associated with lower risks of recurrent VTE (HR, 0.54; 95% CI, 0.45-0.65) and major bleeding events (HR, 0.67; 95% CI, 0.54-0.84); these results remained consistent in patients with provoked VTE and those with unprovoked VTE.

CONCLUSION: This population-based analysis of patients with VTE extends results of randomized clinical trials indicating lower risks of development of recurrent VTE and major bleeding events with use of apixaban versus warfarin in real-world settings. The observed benefits of apixaban extended to selected subgroups of the VTE population, including patients with provoked VTE.

OBJECTIVE: The potential for beta blocker use to reduce joint pain and analgesic use in osteoarthritis (OA) patients has not been well established. The objective of this study was to estimate the association between beta blocker use and knee pain, areas of joint pain, and analgesic use among participants with symptomatic knee OA.

DESIGN: We selected participants with symptomatic knee OA from the Osteoarthritis Initiative. Outcome measures included knee pain (e.g., WOMAC pain subscale), areas of joint pain (e.g., widespread joint pain), and analgesic use (e.g., use of strong pain prescriptions including opioids). We decomposed time-varying beta blocker use into within-person and between-person variation, and included these components in linear mixed effects models for repeated outcome measures of knee pain, joint pain, and analgesic use over 8 years.

RESULTS: Among 1,168 participants, 15% reported beta blocker use at baseline. Beta blocker users (5.2, 95% CI [4.7, 5.8]) had similar estimated mean WOMAC pain scores as other anti-hypertensive users (4.9, 95% CI [4.6, 5.2]), with an estimated within-person difference of 0.1 (95% CI [-0.3, 0.4]). Proportion of participants reporting widespread joint pain was similar between beta blocker users and other anti-hypertensive users (40.1% vs 40.3%; within-person effect, odds ratio [OR] = 0.87, 95% CI [0.63, 1.22]). Reported use of strong prescription pain medication was also similar between beta blocker users and other anti-hypertensive users (7.7% vs 8.2%; within-person effect, OR = 1.39, 95% CI [0.75, 2.55]).

CONCLUSIONS: We found no evidence that beta blockers confer a clinically meaningful reduction in knee pain severity, areas of joint pain, or analgesic use among participants with symptomatic knee OA.

PURPOSE: Little is known about adherence to antidepressant treatment during acute and continuation phase of depression among older adults with dementia and newly diagnosed major depressive disorders (MDD). This study estimated the extent of and factors associated with adherence to acute and continuation phase antidepressant treatment among older adults with dementia and newly diagnosed MDD.

METHODS: We conducted a retrospective cohort study using the Medicare 5% sample claims data (2012-2013) among older adults (age≥65 years) with dementia who were newly diagnosed with MDD. Intake period of our study was from 01-May-2012 through 30-April-2013. The dependent variables of this study were acute and continuation phase depression treatment adherence. Factors associated with acute and continuation phase antidepressant treatment adherence were identified using multiple logistic regression analyses.

RESULTS: The final study sample consisted of 6239 [adherent: N=4644 (74.44%)] and 5617 [adherent: N=3584 (63.81%)] older adults with dementia and MDD during the acute and continuation phase treatment, respectively. During the acute phase, only race/ethnicity was significantly associated with adherence to depression treatment, whereas race/ethnicity and baseline antipsychotic use were significantly associated with adherence to depression treatment during the continuation phase.

CONCLUSION: Approximately, 74% and 64% older adults with dementia and MDD were adherent to acute and continuation phase antidepressant treatment in this nationally representative sample of Medicare beneficiaries, and we identified several modifiable and non-modifiable factors associated with adherence.

IMPORTANCE: Since 2011, immune checkpoint inhibitors (ICIs) have been effective treatment options for advanced melanoma. Little is known about how risks of immune-related adverse events (irAEs) vary by ICIs.

OBJECTIVE: To compare the risk of irAEs across different treatment regimens for advanced melanoma using network meta-analysis.

DATA SOURCES: PubMed/MEDLINE, Embase, Web of Science, and Scopus were searched for all randomized clinical trial (RCT) articles published from January 1, 2010, through June 30, 2019.

STUDY SELECTION: Studies included phases 2 and 3 RCTs in the treatment of advanced melanoma that compared ICIs (ipilimumab, nivolumab, and pembrolizumab) with chemotherapy drugs (eg, dacarbazine, carboplatin, and paclitaxel) or different ICI regimens.

DATA EXTRACTION AND SYNTHESIS: Different treatment regimens were compared using bayesian network meta-analysis with Markov chain Monte Carlo simulation with noninformative prior distribution and random-effects generalized linear models.

MAIN OUTCOMES AND MEASURES: Primary outcomes were the cumulative incidence of any irAEs (regardless of severity) and severe irAEs (grades 3-5). Based on the pooled odds ratios (ORs) and 95% credible intervals (95% CrI), the probability of being associated with the lowest irAE risks was estimated for each treatment regimen.

RESULTS: Nine RCTs with 8 different treatment regimens for advanced melanoma and involving a total of 5051 patients were included. Overall, the 3 ICI treatment regimens associated with the lowest risk of any or severe irAEs were pembrolizumab, 2 mg/kg, every 3 weeks; nivolumab, 3 mg/kg, every 2 weeks; and pembrolizumab, 10 mg/kg, every 3 weeks. Compared with ipilimumab, 10 mg/kg, every 3 weeks, only nivolumab, 3 mg/kg, every 2 weeks, was associated with a decreased risk for any irAEs (OR, 0.34; 95% CrI, 0.13-0.94). A decreased risk for severe irAEs was observed for ipilimumab, 3 mg/kg, every 3 weeks (OR, 0.35; 95% CrI, 0.14-0.74); pembrolizumab, 10 mg/kg, every 2 weeks (OR, 0.22; 95% CrI, 0.05-0.95) and 10 mg/kg every 3 weeks (OR, 0.20; 95% CrI, 0.06-0.68); and nivolumab, 3 mg/kg, every 2 weeks (OR, 0.20; 95% CrI, 0.07-0.48) compared with ipilimumab, 10 mg/kg, every 3 weeks. An increased risk for severe irAEs was associated with nivolumab, 1 mg/kg, every 3 weeks combined with ipilimumab, 3 mg/kg, every 3 weeks compared with other ICI regimens (ORs ranging from 4.09 [95% CrI, 1.73-10.99] to 7.40 [95% CrI, 1.12-49.29]) except ipilimumab, 10 mg/kg, every 3 weeks.

CONCLUSIONS AND RELEVANCE: These findings suggest that for patients with advanced melanoma at high risk of irAEs, pembrolizumab, 2 mg/kg, every 3 weeks, nivolumab, 3 mg/kg, every 2 weeks, and pembrolizumab, 10 mg/kg, every 3 weeks may be the preferred treatment regimens (with respect to irAE risks) among the ICI regimens reported, whereas ipilimumab, 10 mg/kg, every 3 weeks alone and nivolumab, 1 mg/kg, every 3 weeks combined with ipilimumab, 3 mg/kg, every 3 weeks should be used with caution. A network analysis may be valuable for clinical decision-making when evidence from head-to-head comparisons is lacking.