Association of oral disease-modifying agents and their adherence trajectories with annual relapses in multiple sclerosis.

BACKGROUND: Real-world effectiveness can vary across oral disease-modifying agents (DMAs) and their adherence trajectories in patients with multiple sclerosis (MS). However, previous studies have not considered longitudinal adherence patterns while evaluating oral DMAs. OBJECTIVES: This study aimed to evaluate the association of oral DMAs and their adherence trajectories with annualized relapse rate (ARR) in patients with MS. METHODS: This retrospective observational cohort study based on the 2015-2019 MarketScan Commercial Claims and Encounters Database involved continuous enrolled adults (18-64 years) with ≥1 MS diagnosis (ICD-9/10-CM:340/G35) and ≥ 1 oral DMA prescription. Patients were grouped into incident fingolimod (FIN), teriflunomide (TER), and dimethyl fumarate (DMF) users based on the index DMA with a one-year washout period. Annual DMA adherence trajectories based on the monthly Proportion of Days Covered (PDC) one year after treatment initiation were identified using Group-Based Trajectory Modeling (GBTM). The validated claims-based ARR was evaluated during the one-year follow-up period using generalized boosted model-based inverse probability treatment weights with negative binomial regression model. RESULTS: The study cohort consisted of 994 MS patients who initiated with FIN (23.0%), TER (22.3%), and DMF (54.7%) during the study period. GBTM grouped eligible patients into three adherence trajectories: complete adherers (59.2%), slow decliners (23.8%), and rapid decliners (17.0%). The proportion of complete adherers varied across the oral DMAs (FIN: 67.1%, TER: 55.4%, and DMF: 57.4%). The negative binomial regression modeling revealed that, while there was no difference in ARR across the three DMAs, rapid decliners (adjusted incidence rate ratio[aIRR]: 1.6, 95% CI: 1.1-2.4) had a higher rate of relapses compared to completely adherent patients. The type of oral DMAs did not moderate the relationship between ARR and the adherence trajectory groups. CONCLUSIONS: Adherence trajectories classified as rapid decliners were associated with a higher ARR than complete adherers after adjusting for their type of oral DMAs. Longitudinal medication adherence patterns are critical in reducing relapse rates in MS.

Cumulative Anticholinergic Burden and its Predictors among Older Adults with Alzheimer's Disease Initiating Cholinesterase Inhibitors.

BACKGROUND: Cumulative anticholinergic burden refers to the cumulative effect of multiple medications with anticholinergic properties. However, concomitant use of cholinesterase inhibitors (ChEIs) and anticholinergic burden can nullify the benefit of the treatment and worsen Alzheimer's disease (AD). A literature gap exists regarding the extent of the cumulative anticholinergic burden and associated risk factors in AD. Therefore, this study evaluated the prevalence and predictors of cumulative anticholinergic burden among patients with AD initiating ChEIs. METHODS: A retrospective longitudinal cohort study was conducted using the Medicare claims data involving parts A, B, and D from 2013 to 2017. The study sample included older adults (65 years and older) diagnosed with AD and initiating ChEIs (donepezil, rivastigmine, or galantamine). The cumulative anticholinergic burden was calculated based on the Anticholinergic Cognitive Burden scale and patient-specific dosing using the defined daily dose over the 1 year follow-up period after ChEI initiation. Incremental anticholinergic burden levels were dichotomized into moderate-high (sum of standardized daily anticholinergic exposure over a year (TSDD) score ≥ 90) versus low-no (score 0-89). The Andersen Behavioral Model was used as the conceptual framework for selecting the predictors under the predisposing, enabling, and need categories. A multivariable logistic regression model was used to evaluate the predictors of high-moderate versus low-no cumulative anticholinergic burden. A multinomial logistic regression model was also used to determine the factors associated with patients having moderate and high burdens compared to low/no burdens. RESULTS: The study included 222,064 older adults with AD with incident ChEI use (mean age 82.24 ± 7.29, 68.9% females, 83.6% White). Overall, 80.48% had some anticholinergic burden during the follow-up, with 36.26% patients with moderate (TSDD scores 90-499), followed by 24.76% high (TSDD score > 500), and 19.46% with low (TSDD score 1-89) burden categories. Predisposing factors such as age; African American, Asian, or Hispanic race; and need factors included comorbidities such as dyslipidemia, syncope, delirium, fracture, pneumonia, epilepsy, and claims-based frailty index were less likely to be associated with the moderate-high anticholinergic burden. The factors that increased the odds of moderate-high burden were predisposing factors such as female sex; enabling factors such as dual eligibility and diagnosis year; and need factors such as baseline burden, behavioral and psychological symptoms of dementia, depression, insomnia, urinary incontinence, irritable bowel syndrome, anxiety, muscle spasm, gastroesophageal reflux disease, heart failure, and dysrhythmia. Most of these findings remained consistent with multinomial logistic regression.  CONCLUSION: Four out of five older adults with AD had some level of anticholinergic burden, with over 60% having moderate-high anticholinergic burden. Several predisposing, enabling, and need factors were associated with the cumulative anticholinergic burden. The study findings suggest a critical need to minimize the cumulative anticholinergic burden to improve AD care.

Patients' willingness to pay for naloxone: A national cross-sectional survey of prescription opioid users with chronic pain in the United States.

BACKGROUND: Millions of U.S. people have been heavily affected by opioids. In March 2023, the Food and Drug Administration approved naloxone as an over-the-counter medication. This has allowed more access to patients at high risk of opioid overdose. However, the patient's willingness to pay for naloxone at the pharmacy counter has not been assessed. OBJECTIVES: This study aimed to characterize factors associated with the willingness to pay for naloxone among the patient group. METHODS: A cross-sectional Qualtrics online panel survey instrument was developed. This survey was distributed to patients in the United States, aged ≥ 18 years, with any chronic pain and taking opioids. The survey included demographics, and clinical characteristics (pain assessment, opioid use, and knowledge of naloxone). In addition, willingness to pay was assessed using a 7-point Likert scale ranging from strongly disagree to strongly agree. An ordinal logistic regression model was used to examine demographic and clinical characteristics. RESULTS: A total of 549 subjects completed the survey (women [53.01%], white or Caucasian (83.61%), age mean [SD] 44 [13]). Women were associated with less willingness to pay (adjusted odds ratio [aOR] 0.685 [95% CI 0.478-0.983], P = 0.0403). Compared with the high household income group (≥ $150,000), low household income ≤ $25,000 (aOR 0.326 [95% CI 0.160-0.662], P = 0.0020) or income between $25,000 and 74,999 (aOR 0.369 [95% CI 0.207-0.657], P = 0.0007) was associated with less likelihood of willing to pay. Patients with a previous diagnosis of obstructive sleep apnea were associated with a higher likelihood of willingness to pay (aOR 1.685 [95% CI 1.138-2.496], P = 0.0092). Each unit increase in pain was also associated with a higher likelihood of willingness to pay (aOR 1.247 [95% CI 1.139-1.365], P < 0.0001). CONCLUSIONS: Demographics and clinical factors were associated with willingness to pay for naloxone. This study's findings are useful in the development of interventions to address pharmacy-based naloxone distribution programs.

Pain-related hospitalization and emergency room visit following initial analgesic prescription after outpatient surgery.

PURPOSE: Our study examined the association between outpatient postsurgical analgesic prescription and risk of insufficiently managed pain characterized by pain-associated hospital admission and emergency room (ER) visit. METHODS: Eligible individuals were children 1-17 years of age who filled an incident analgesic following an outpatient surgery during 2013-2018. Pain-associated hospital admission or ER visit were measured within 30 days following the outpatient surgical procedure. A hierarchical multivariable logistic regression model with patients nested under prescribers was fitted to test the association between incident analgesic prescription and risk of having pain-associated hospital admission or ER visit. RESULTS: Of 14 277 children meeting the inclusion criteria, 6224 (43.6%) received an incident opioid and 8053 (56.4%) received an incident non-opioid analgesic prescription respectively. There were a total of 523 (3.7%) children undergoing surgical procedures that had pain-related hospital admissions or ER visits with 5.1% initiated on non-opioid analgesics and 1.8% on opioid analgesics. The multilevel model indicated that initial opioid analgesic recipients were 32% less likely of having a pain-associated hospital admission or ER visit [aOR: 0.68 (95% CI: 0.3-0.8)]. CONCLUSION: Majority of postsurgical patients do not require additional pain management strategies. In the 3.7% of patients requiring additional pain management strategies, those initiated on non-opioid analgesics are more likely to have a pain-associated hospital admission or ER visit compared with their opioid recipient counterparts.

Risk of incident cardiovascular events with disease-modifying anti-rheumatic drugs among adults with rheumatoid arthritis: a nested case-control study.

OBJECTIVE: This study examined the risk of cardiovascular disease (CVD) associated with the disease-modifying anti-rheumatic drugs (DMARDs) in rheumatoid arthritis (RA). METHOD: This nested case-control study used the MarketScan database (2012-2014), involving adult RA patients (aged ≥18 years) initiating either a conventional synthetic (cs) DMARD, biologic DMARD, or targeted synthetic (ts) DMARD between January 1, 2013 and December 31, 2014 (cohort entry) and had no CVD history. Cases were individuals with incident CVD identified using diagnosis codes or procedure codes from medical claims. For each case, 10 age- and sex-matched controls were selected using the incident density sampling with replacement. Prescriptions of DMARDs were measured 90 days before the event date. Conditional logistic regression examined the association of risk of CVD with DMARDs in combination treatment or individual use, with reference to methotrexate (MTX) monotherapy, adjusting for baseline confounders. Subgroup analyses were performed separately in DMARD combination therapy users or individual DMARD users, respectively. RESULTS: In total, 270 cases of incident CVD and 2700 controls were included (mean [standard deviation (SD)] age: 54 [1]; 75.6% women). The commonly prescribed DMARD therapies were csDMARD monotherapy (n = 795, 27.04%), followed by  tumor necrosis factor inhibitors (TNFi) monotherapy (n = 367, 12.48%), and TNFi in combination with MTX (n = 314, 10.68%). Compared with MTX monotherapy, overall use of DMARD agents was not associated with the differential risk of CVD, including various types of DMARD combination regimens. The findings were similar across subgroup analyses. CONCLUSIONS: The study found no differential risk of CVD with DMARDs in combination therapy or monotherapy compared to MTX monotherapy in patients with RA. Key Points • This study evaluated the risk of cardiovascular disease (CVD) associated with the disease-modifying anti-rheumatic drugs (DMARDs) in rheumatoid arthritis (RA). • Findings suggest no differential CVD risk with DMARDs in combination with MTX or used individually compared with MTX monotherapy in patients with early RA. • Further efforts should focus on a better understanding of the mechanism of DMARD combination treatments with MTX in modifying CV risk.

Adherence to Recommended Metabolic Monitoring of Children and Adolescents Taking Second-Generation Antipsychotics.

OBJECTIVE: Clinical guidelines recommend periodic monitoring for adverse metabolic effects associated with second-generation antipsychotic medications. The authors sought to evaluate adherence to the guideline-recommended metabolic monitoring schedule for children and adolescents prescribed second-generation antipsychotics. METHODS: The authors used a national electronic medical records database for a retrospective study of children and adolescents ages 1-17 years (N=9,620) who were prescribed second-generation antipsychotics in January 2010-December 2018. Adherence to guideline-recommended monitoring of body mass index (BMI), blood glucose, and cholesterol was categorized as full, partial, and no monitoring. Full monitoring of patients was defined as strict metabolic monitoring, following the guideline-recommended schedule. Patients who received any monitoring, but not meeting the full monitoring criteria, were considered partially monitored. Three multinomial logistic regression models were fitted for each metabolic parameter to identify predictors associated with monitoring status. RESULTS: BMI was the metabolic parameter with the highest adherence to guideline-recommended monitoring (full monitoring, 4.7% of patients; partial monitoring, 44.8%), followed by blood glucose (full monitoring, 6.5%; partial monitoring, 29.4%) and cholesterol (full monitoring, 0.8%; partial monitoring, 22.4%). Being Black (vs. non-Black), having a comorbid mood disorder (vs. none), receiving olanzapine as the index second-generation antipsychotic (vs. aripiprazole), and receiving an antidepressant as a concurrent medication (vs. none) were associated with a higher likelihood of receiving both full and partial monitoring of all three metabolic parameters. CONCLUSIONS: Both full and partial adherence to guideline-recommended monitoring of children and adolescents prescribed second-generation antipsychotics were poor. However, children and adolescents at increased metabolic risk tended to be more closely monitored.

Application of machine learning in predicting survival outcomes involving real-world data: a scoping review.

BACKGROUND: Despite the interest in machine learning (ML) algorithms for analyzing real-world data (RWD) in healthcare, the use of ML in predicting time-to-event data, a common scenario in clinical practice, is less explored. ML models are capable of algorithmically learning from large, complex datasets and can offer advantages in predicting time-to-event data. We reviewed the recent applications of ML for survival analysis using RWD in healthcare. METHODS: PUBMED and EMBASE were searched from database inception through March 2023 to identify peer-reviewed English-language studies of ML models for predicting time-to-event outcomes using the RWD. Two reviewers extracted information on the data source, patient population, survival outcome, ML algorithms, and the Area Under the Curve (AUC). RESULTS: Of 257 citations, 28 publications were included. Random survival forests (N = 16, 57%) and neural networks (N = 11, 39%) were the most popular ML algorithms. There was variability across AUC for these ML models (median 0.789, range 0.6-0.950). ML algorithms were predominately considered for predicting overall survival in oncology (N = 12, 43%). ML survival models were often used to predict disease prognosis or clinical events (N = 27, 96%) in the oncology, while less were used for treatment outcomes (N = 1, 4%). CONCLUSIONS: The ML algorithms, random survival forests and neural networks, are mainly used for RWD to predict survival outcomes such as disease prognosis or clinical events in the oncology. This review shows that more opportunities remain to apply these ML algorithms to inform treatment decision-making in clinical practice. More methodological work is also needed to ensure the utility and applicability of ML models in survival outcomes.

Trends in Repeat Opioid Analgesic Prescription Utilization for Acute Pain in Children: 2013-2018.

OBJECTIVE: Our study examined the change in repeat opioid analgesic prescription trends in children and adolescents experiencing acute pain between 2013 and 2018. METHODS: Eligible individuals were children and adolescents between 1 and 17 years of age enrolled in a Medicaid Managed Care plan and filled an incident opioid analgesic prescription from 2013 to 2018. A repeat opioid prescription was defined as receiving a subsequent opioid prescription within 30 days from the end of the incident opioid prescription. A generalized linear regression analysis was conducted to examine changes in repeat opioid analgesic dispensing over time at quarterly intervals from January 1, 2013, to December 31, 2018. RESULTS: The cohort comprised 17,086 children and adolescents receiving an incident opioid analgesic. Of these, 1780 (10.4%) filled a repeat opioid analgesic prescription. There was a significant decline in the repeat opioid analgesic trend from 11.5% in Q1 2013 to 9.6% in Q4 2018. Stratified analyses by age, sex, and race and ethnicity in a sub-cohort of patients undergoing surgical procedures showed that a significant decline in repeat opioid utilization over time has been observed in all racial/ethnic groups stratified by age and sex, with the most significant decline found in non-Hispanic White children and Hispanic adolescents. At the end of the 6-year follow-up, the racial and ethnic variations in repeat opioid utilization associated with surgical procedures had significantly reduced in children yet persisted among adolescents. CONCLUSIONS: Approximately 10% of incident pediatric opioid analgesic recipients received a repeat opioid prescription. There has been a moderate but steady decline (∼7% per quarter) in repeat opioid analgesic utilization between 2013 and 2018.

Association Between Abnormal Metabolic Parameters and Receiving Subsequent Interventions in Children and Adolescents Initiating Second-Generation Antipsychotics.

Objectives: This study aimed to examine the association between abnormal readings of metabolic parameters detected during second-generation antipsychotic (SGA) treatment and the likelihood of receiving subsequent adverse drug event interventions. Methods: This was a nested case-control study conducted on patients 1-17 years of age with at least two prescriptions of SGAs between January 2010 and January 2019 using TriNetX EMR data. Following an incident density sampling procedure, patients who received the SGA metabolic adverse event intervention (mAEI) (case) were matched with three nonrecipients (controls). The abnormal readings of metabolic parameters within 30 days before the initiation of mAIEs were further identified. These metabolic parameters include body mass index (BMI) and laboratory parameters such as cholesterol, high-density lipoprotein, low-density lipoprotein, triglycerides, blood glucose, HbA1c, thyroid hormones, liver enzymes, and prolactin. The association of abnormal metabolic parameters with subsequent mAEIs was assessed using a conditional logistic regression model, after adjusting for demographic and other clinical risk factors. Results: One thousand eight hundred eighty-four children and adolescents met the inclusion criteria and were prescribed SGA mAEIs. The most common types of mAEIs prescribed were weight management pharmacotherapy (40.6%), switching from a high or medium metabolic risk profile SGA to a low-risk one (30.9%), nonpharmacological treatment (25.4%), and switching from SGA polytherapy to monotherapy (11.7%). The conditional logistic regression analysis on matched mAEI recipients and nonrecipients showed that patients with an abnormal BMI had 43% higher odds of receiving mAEI (odds ratio [95% confidence interval]: 1.43 [1.13-1.79]). However, the presence of an abnormal laboratory reading was not associated with the initiation of mAEIs. Conclusions: The prescribing of mAEIs were associated with the presence of obesity, but not with abnormal readings of other metabolic parameters, suggesting that additional data are needed to clarify the long-term implication of SGA metabolic adverse events other than weight gain and to inform the appropriate timing for interventions.

Performance of advanced machine learning algorithms overlogistic regression in predicting hospital readmissions: A meta-analysis.

Objectives: Machine learning algorithms are being increasingly used for predicting hospital readmissions. This meta-analysis evaluated the performance of logistic regression (LR) and machine learning (ML) models for the prediction of 30-day hospital readmission among patients in the US. Methods: Electronic databases (i.e., Medline, PubMed, and Embase) were searched from January 2015 to December 2019. Only studies in the English language were included. Two reviewers performed studies screening, quality appraisal, and data collection. The quality of the studies was assessed using the Quality in Prognosis Studies (QUIPS) tool. Model performance was evaluated using the Area Under the Curve (AUC). A random-effects meta-analysis was performed using STATA 16. Results: Nine studies were included based on the selection criteria. The most common ML techniques were tree-based methods such as boosting and random forest. Most of the studies had a low risk of bias (8/9). The AUC was greater with ML to predict 30-day all-cause hospital readmission compared with LR [Mean Difference (MD): 0.03; 95% Confidence Interval (CI) 0.01-0.05]. Subgroup analyses found that deep-learning methods had a better performance compared with LR (MD 0.06; 95% CI, 0.04-0.09), followed by neural networks (MD: 0.03; 95% CI, 0.03-0.03), while the AUCs of the tree-based (MD: 0.02; 95% CI -0.00-0.04) and kernel-based (MD: 0.02; 95% CI 0.02 (-0.13-0.16) methods were no different compared to LR. More than half of the studies evaluated heart failure-related rehospitalization (N = 5). For the readmission prediction among heart failure patients, ML performed better compared with LR, with a mean difference in AUC of 0.04 (95% CI, 0.01-0.07). The leave-one-out sensitivity analysis confirmed the robustness of the findings. Conclusion: Multiple ML methods were used to predict 30-day all-cause hospital readmission. Performance varied across the ML methods, with deep-learning methods showing the best performance over the LR.

Outpatient opioid prescribing by Alzheimer's diagnosis among older adults with pain in United States.

OBJECTIVE: To examine opioid prescribing practices for pain in older adults with and without Alzheimer's Disease and Related Dementias (ADRD). METHODS: This cross-sectional study used National Ambulatory Medical Care Survey data (2014-2016, and 2018). Adults aged ≥ 50 years with pain were analyzed. Prescribing of opioid and concomitant sedative prescriptions (including benzodiazepines, Z-drugs, and barbiturates) were identified by the Multum lexicon code. Multivariable logistic regression evaluated the risk of opioid prescribing or co-prescribing of opioid and sedative associated with ADRD in older adults with pain. RESULTS: There were 13,299 office visits in older adults with pain, representing 451.75 million visits. Opioid prescribing occurred in 27.19%; 30% involved co-prescribing of opioids and sedatives. ADRD was not associated with opioid prescribing or co-prescribing of opioid and sedative therapy. CONCLUSIONS: Opioid and sedatives are commonly prescribed in older adults with pain. Longitudinal studies need to understand the etiology and chronicity of opioid use in older patients, specifically with ADRD.

Real-world Comparative Effectiveness of Methotrexate-based Combinations for Rheumatoid Arthritis: A Retrospective Cohort Study.

PURPOSE: Guidelines recommend using disease-modifying antirheumatic drugs (DMARDs) in combination with methotrexate (MTX) for patients with rheumatoid arthritis (RA) after monotherapy. Little is known about the real-world comparative effectiveness of these MTX-DMARD combinations. This study compared the effectiveness of various MTX-based DMARD combinations for patients with RA initiating MTX-DMARD combination therapy using administrative claims database. METHODS: This retrospective cohort study included adults (aged ≥18 years) with RA who initiated MTX combination treatment with conventional synthetic DMARDs (csDMARDs), tumor necrosis factor inhibitor (TNFi) biologic DMARDs (bDMARDs), non-TNFi bDMARDs, or targeted synthetic DMARDs (tsDMARDs) between July 1, 2012, and December 31, 2013 (index date), from the MarketScan Commercial Claims Data. Patients had continuous enrollment from the 6 months of preindex period until the 12 months of postindex period. The MTX-based DMARD combination therapy cohort was defined as ≥1 MTX prescription in the first 30 days from the index date and ≥14 days overlapping use of the prescription fills of the MTX and the index DMARD. Effectiveness was measured by using the claims algorithm (dosing, switching, addition, oral glucocorticoid use, or multiple glucocorticoid injection). Propensity score analysis with the inverse probability of treatment weighting (PS-IPTW), estimated by using the generalized boosted machine learning method, was used to balance the distribution of baseline variables between the combination groups. Multivariable logistic regression using PS-IPTW was conducted to compare the effectiveness of the combination groups. Sensitivity analysis evaluated the modified effectiveness algorithms or the time to the first treatment failure. FINDINGS: A total of 3174 adult patients with RA starting an MTX-DMARD combination therapy were identified (mean [SD] age, 50 [9] years), including 1568 (49%) initiating a csDMARD + MTX, 1343 (42%) initiating TNFi + MTX, and 240 (8%) initiating non-TNFi bDMARD + MTX, and 23 (1%) initiating tsDMARD + MTX. Owing to the small sample, the tsDMARD combination group was not included in the comparative analysis. Algorithm-based therapy effectiveness was found in 9.95% of the csDMARD + MTX, 20.48% of the TNFi + MTX, and 20.83% of the non-TNFi + MTX groups. PS-IPTW showed that the csDMARD combination is less effective (adjusted odds ratio, 0.422; 95% CI, 0.341-0.524) than the TNFi combination; however, the non-TNFi biologic combination had similar effectiveness (aOR, 1.063; 95% CI, 0.680-1.662) compared to the TNFi combination. Sensitivity analyses confirmed the main results. IMPLICATIONS: Among RA patients initiating MTX-DMARD combinations, both non-TNFi biologics and TNFi-based combinations with MTX were equally effective, but csDMARD + MTX was less effective than the TNFi plus MTX.

Assessing treatment switch among patients with multiple sclerosis: A machine learning approach.

Background: Patients with multiple sclerosis (MS) frequently switch their Disease-Modifying Agents (DMA) for effectiveness and safety concerns. This study aimed to develop and compare the random forest (RF) machine learning (ML) model with the logistic regression (LR) model for predicting DMA switching among MS patients. Methods: This retrospective longitudinal study used the TriNetX data from a federated electronic medical records (EMR) network. Between September 2010 and May 2017, adults (aged ≥18) MS patients with ≥1 DMA prescription were identified, and the earliest DMA date was assigned as the index date. Patients prescribed any DMAs different from their index DMAs were considered as treatment switch. . The RF and LR models were built with 72 baseline characteristics and trained with 70% of the randomly split data after up-sampling. Area Under the Curves (AUC), accuracy, recall, G-measure, and F-1 score were used to evaluate the model performance. Results: In this study, 7258 MS patients with ≥1 DMA were identified. Within two years, 16% of MS patients switched to a different DMA. The RF model obtained significantly better discrimination than the LR model (AUC = 0.65 vs. 0.63, p < 0.0001); however, the RF model had a similar predictive performance to the LR model with respect to F- and G-measures (RF: 72% and 73% vs. LR: 72% and 73%, respectively). The most influential features identified from the RF model were age, type of index medication, and year of index. Conclusions: Compared to the LR model, RF performed better in predicting DMA switch in MS patients based on AUC measures; however, judged by F- and G-measures, the RF model performed similarly to LR. Further research is needed to understand the role of ML techniques in predicting treatment outcomes for the decision-making process to achieve optimal treatment goals.

Impact of the AACAP practice parameters on the metabolic adverse event monitoring for second-generation antipsychotics (SGAs) in children and adolescents.

INTRODUCTION: The objective of our study was to evaluate the impact of the publication of the American Academy of Child and Adolescent Psychiatry (AACAP) practice parameters for SGA metabolic monitoring in 2011 on SGA metabolic monitoring uptake among pediatric SGA recipients. METHODS: This was a retrospective study of children 1-17 years of age who initiated SGA treatment from Jan 2010 to December 2018 using a national Electronic Medical Records database. A segmented regression of interrupted time-series (ITS) analysis was conducted to analyze the change of metabolic monitoring rates for Body Mass Index (BMI), Blood Glucose (BG), and Total Cholesterol (CHL) 9 quarters pre- and 26 quarters post-the publication of the AACAP practice parameters. RESULTS: The analytical cohort included 9620 children and adolescents who initiated SGA treatment during the study period. The ITS results showed that the publication of the AACAP practice parameter for SGA metabolic monitoring was associated with a 12.61 percentage points (p < 0.0002) immediate increase in BMI monitoring rate, (increased from 29.10% in Q4 2011 to 40.10% in Q3 2012). There was a positive trend of BMI monitoring rate prior to the publication of AACAP practice parameters, which continued during the post-publication period. Neither immediate nor sustained changes in the association of monitoring rates for BG and CHL were observed after the issuance of the guidelines. CONCLUSION: The publication of AACAP practice parameters for SGA monitoring was associated with a significant improvement in the monitoring for BMI, but not for BG and CHL in children and adolescents.

Factors influencing prescribing the first add-on disease-modifying antirheumatic drugs in patients initiating methotrexate for rheumatoid arthritis.

Background: Advances in Disease-Modifying Antirheumatic Drugs (DMARDs) have expanded the treatment landscape for Rheumatoid Arthritis (RA). Guidelines recommend adding either conventional synthetic (cs), biologic (b), or targeted synthetic (ts) DMARDs to methotrexate (MTX) for managing RA. Limited evidence exists regarding the factors that contribute to adding a DMARD agent to the MTX regimen. This study examined the factors associated with adding the first DMARD in RA patients initiating MTX. Methods: This retrospective cohort study utilized the MarketScan data (2012-2014) involving adults (aged ≥18) with RA initiating an MTX (index date) between Jul 1, 2012 and Dec 30, 2013, and with continuous enrollment for the 6-month pre-index period. The combination therapy users received the first treatment addition of DMARD starting from day 30 after the index MTX over one year period. The study focused on the addition of csDMARDs, Tumor Necrosis Factor Inhibitors (TNFi) bDMARDs, non-TNFi bDMARDs, or tsDMARDs. Baseline covariates were measured in the 6-month pre-index and grouped into predisposing, enabling, and need factors, as per the Andersen Behavior Model. Multivariable logistic regression examined the factors associated with the addition of TNFi compared to adding a csDMARD. An additional regression model evaluated the factors associated with adding any biologic (combining TNFi and non-TNFi biologics). Results: Among 8350 RA patients starting MTX, 31.92% (n = 2665) initiated any DMARD within the 1-year post-index period. Among RA patients initiating a DMARD prescription after starting MTX, 945 (11.32%) received combination therapy with treatment addition of a DMARD to MTX regimen; majority added TNFi (550, 58%), followed by csDMARD (352, 37%); non-TNF biologic (40, 4%), or tsDMARD (3, 0.3%). The tsDMARD group was limited and was not included for further analysis. The multivariable model found Preferred Provider Organization insurance coverage (odds ratio [OR], 1.43; 95% confidence interval (CI), 1.06-1.93), chronic pulmonary disease (OR, 1.98; 95% CI, 1.14-3.44), liver disease (OR, 5.24; 95% CI, 1.77-15.49), and Elixhauser score (OR, 0.91; 95% CI, 0.86-0.97) were significantly associated with the addition of TNF-α inhibitors. The separate multivariable model additionally found that patients from metropolitan areas (OR, 1.50; 95% CI, 1.04-2.16) were positively associated with adding any biological agent. Conclusions: TNFi are often added to MTX for managing RA. Enabling and need factors contribute to the prescribing of a TNFi add-on therapy in RA. Future research should examine the impact of these combination therapies on RA management.

Comparative adherence trajectories of oral disease-modifying agents in multiple sclerosis.

STUDY OBJECTIVE: This study compared the adherence trajectories of fingolimod (FIN), teriflunomide (TER), and dimethyl fumarate (DMF) users with multiple sclerosis (MS) as there is limited evidence regarding the comparative adherence patterns of different oral disease-modifying agents (DMAs). DESIGN: A retrospective cohort study DATA SOURCE: 2015-2019 IBM MarketScan Commercial Claims Database. PATIENTS: Adults (≥18 years) with MS (International Classification of Diseases [ICD]-9/10-Clinical Modification [CM]:340/G35) diagnosis and ≥1 DMA prescription. INTERVENTION: Incident FIN-, TER-, or DMF use based on the index DMA with 1 year of washout period. MEASUREMENTS: The DMA adherence trajectories based on the proportion of days covered (PDC) were examined using the Group-Based Trajectory Modeling (GBTM) one year after the treatment initiation. Generalized boosting models (GBM)-based inverse probability treatment weights (IPTW) were incorporated in multinomial logistic regression to assess the comparative adherence trajectories across oral DMAs with FIN group as a reference category. MEASUREMENTS AND MAIN RESULTS: The study cohort consisted of 1913 patients with MS who were initiated with FIN (24.2%, n = 462), TER (24.0%, n = 458), and DMF (51.9%, n = 993) during 2016-2018. The adherence rate (PDC ≥ 0.8) among FIN, TER, and DMF users was found to be 70.8% (n = 327), 59.6% (n = 273), and 61.0% (n = 606), respectively. The GBTM grouped patients into three adherence trajectories: Complete Adherers-59.1%, Slow Decliners-22.6%, and Rapid Discontinuers-18.3%. The multinomial logistic regression model involving GBM-based IPTW revealed that DMF (adjusted odds ratio [aOR]: 2.32, 95% confidence interval [CI]:1.57-3.42) and TER (aOR: 2.50, 95% CI: 1.62-3.88) users had higher odds to be rapid discontinuers relative to FIN users. In addition, TER users were more likely (aOR: 1.50, 95% CI: 1.06-2.13) to be slow decliners compared with FIN users. CONCLUSION: Teriflunomide and DMF were associated with poorer adherence trajectories than FIN. More research is needed to evaluate the clinical implications of these adherence trajectories of oral DMAs to optimize the management of MS.

Incremental health care expenditures for non-Hodgkin lymphoma in comparison with other cancers: Analysis of national survey data.

BACKGROUND: Non-Hodgkin lymphoma (NHL) is among the most common cancers in the United States, with an estimated annual incidence of more than 80,000 and a high survival rate. However, limited national data exist regarding the health care burden of NHL. OBJECTIVE: To evaluate the incremental health care expenditures among patients with NHL using the Medical Expenditure Panel Survey (MEPS) data compared with patients with other cancers. METHODS: This observational cross-sectional study included all patients with NHL (≥ 18 years) and all individuals diagnosed with other cancers from the MEPS 2014-2019. The components of health care expenditures included hospital inpatient care, office-based visits, outpatient care, emergency department, prescription medications, dental, home health, and other expenditures. Patients with NHL and those diagnosed with other cancers were identified from the full-year consolidated MEPS Household Component 2014-2019. Descriptive weighted analysis was used to compare the health care expenditure components between individuals with NHL and all other cancers. A 2-part model using probit and generalized linear models with a log link function was used to estimate the incremental increase in total health care expenditures for NHL compared with all other cancers. RESULTS: According to the MEPS, there were 0.74 million patients with NHL (95% CI = 0.62-0.86) and 27.91 million patients with other cancers (95% CI = 26.69-29.13) annually. Most of the patients with NHL were White (78.36%), male (60.67%), and older than 65 years (45.8%). The unadjusted analysis indicated a total annual expenditure of $21,698 (95% CI = $16,752-$26,645) for NHL, which was significantly higher than the annual expenditure for patients with other cancers ($15,029 [95% CI = $14,476-$15,582]). Most of the total health expenditure of both the NHL group and the other cancers group was distributed in 3 categories of hospital inpatient care (29.15% vs 26.29%), office-based visits (28.10% vs 25.08%), and prescription medications (19.03% vs 22.57%). Based on the 2-part model adjusted for all covariates, the annual health care expenditure for NHL was $7,284 (95% CI = $1,432-$13,135), higher than the expenditure of patients diagnosed with all other cancers. Among the health care expenditure components, the office-based visits were $2,641 higher for patients with NHL compared with the other cancers group (95% CI = $1,129-$4,153). CONCLUSIONS: The economic burden of NHL is higher compared with other cancers. Most of the NHL expenditures were attributable to hospital inpatient services and office-based visits. The study findings can inform value-based care considerations because of a better understanding of utilization and care patterns for NHL. DISCLOSURES: Dr Aparasu has received research funding from Astellas Inc., Incyte Corp., Gilead, and Novartis Inc. for projects unrelated to the current work. The other authors declare no conflicts of interest for this article. We confirm that this work is original and has not been published elsewhere, nor is it currently under consideration for publication elsewhere.

Association Between Initial Opioid Prescription Duration and 30-Day Risk of Receiving Repeat Opioid Among Children.

OBJECTIVE: Our study evaluated the association between initial opioid prescription duration and receipt of a repeat opioid prescription in children. METHODS: Eligible individuals were children between 1 and 17 years of age who enrolled in a Medicaid Managed Care plan and filled an incident opioid prescription during 2013 to 2018. An incident prescription was defined as receipt of an opioid analgesic without a prior use for 12 months. A repeat opioid prescription was defined as receipt of a subsequent opioid prescription within 30 days since the end of incident opioid prescription. A hierarchical multivariable logistic regression model was fitted to test the association between incident opioid prescription duration and the likelihood of receiving a repeat prescription. RESULTS: The cohort consisted of 17,086 children receiving an incident opioid prescription in which 6272 (36.7%) received 1 to 3 days' supply, 8442 (49.4%) received 4 to 7 days' supply, 1434 (8.4%) received 8 to 10 days' supply, and 938 (5.5%) received >10 days' supply. Of these incident opioid recipients, 1780 (10.4%) filled a repeat opioid prescription. The multilevel model results indicated that, children receiving 4 to 7 days' supply (adjusted odds ratio [aOR]: 0.98 {0.9-1.1}), 8 to 10 days' supply (aOR: 1.03 [0.8-1.3]), and >10 days' supply (aOR: 0.85 [0.7-1.1]) had comparable likelihoods of receiving a repeat prescription as those receiving 1 to 3 days' supply. DISCUSSION: Nearly 10% of children who filled an opioid prescription for acute pain received a repeat prescription. Initial prescription duration was not associated with the risk of receiving a repeat prescription.

Machine learning methods to predict 30-day hospital readmission outcome among US adults with pneumonia: analysis of the national readmission database.

BACKGROUND: Hospital readmissions for pneumonia are a growing concern in the US, with significant consequences for costs and quality of care. This study developed the rule-based model and other machine learning (ML) models to predict 30-day readmission risk in patients with pneumonia and compared model performance. METHODS: This population-based study involved patients aged ≥ 18 years hospitalized with pneumonia from January 1, 2016, through November 30, 2016, using the Healthcare Cost and Utilization Project-National Readmission Database (HCUP-NRD). Rule-based algorithms and other ML algorithms, specifically decision trees, random forest, extreme gradient descent boosting (XGBoost), and Least Absolute Shrinkage and Selection Operator (LASSO), were used to model all-cause readmissions 30 days post-discharge from index pneumonia hospitalization. A total of 61 clinically relevant variables were included for ML model development. Models were trained on randomly partitioned 50% of the data and evaluated using the remaining dataset. Model hyperparameters were tuned using the ten-fold cross-validation on the resampled training dataset. The area under the receiver operating curves (AUROC) and area under precision-recall curves (AUPRC) were calculated for the testing set to evaluate the model performance. RESULTS: Of the 372,293 patients with an index hospital hospitalization for pneumonia, 48,280 (12.97%) were readmitted within 30 days. Judged by AUROC in the testing data, rule-based model (0.6591) significantly outperformed decision tree (0.5783, p value < 0.001), random forest (0.6509, p value < 0.01) and LASSO (0.6087, p value < 0.001), but was less superior than XGBoost (0.6606, p value = 0.015). The AUPRC of the rule-based model in the testing data (0.2146) was higher than the decision tree (0.1560), random forest (0.2052), and LASSO (0.2042), but was similar to XGBoost (0.2147). The top risk-predictive rules captured by the rule-based algorithm were comorbidities, illness severity, disposition locations, payer type, age, and length of stay. These predictive risk factors were also identified by other ML models with high variable importance. CONCLUSION: The performance of machine learning models for predicting readmission in pneumonia patients varied. The XGboost was better than the rule-based model based on the AUROC. However, important risk factors for predicting readmission remained consistent across ML models.

Dose-dependent relation between metformin and the risk of hormone receptor-positive, her2-negative breast cancer among postmenopausal women with type-2 diabetes.

PURPOSE: Metformin has demonstrated a chemoprotective effect in breast cancer but there is limited evidence on the effect of cumulative exposure to metformin and the risk of hormone receptor-positive and human epidermal growth factor receptor 2-negative (HR + /HER2-) breast cancer. This study assessed this risk with dose and intensity of metformin in postmenopausal women with type-2 diabetes mellitus (T2DM). METHODS: This nested case-control study used the Surveillance, Epidemiology, and End Results-Medicare data (2008-2015). Cohort entry was the date of incident T2DM diagnosis. Cases were those diagnosed with HR + /HER2- breast cancer (event date) as their first/only cancer. Non-cancer T2DM controls were matched using variable-ratio-matching. Cumulative dose and average intensity of metformin were measured during the 1-year lookback period. Dose(mg) was categorized as: (1)0, (2)0-30,000, (3)30,001-136,000, (4)136,001-293,000, and (5) > 293,000, and intensity(mg/day) as: 0, 1-500, and > 500. Covariates were conceptualized using the Andersen Behavioral Model. Conditional logistic regression was used to assess the risk of HR + /HER2- breast cancer with metformin-use. RESULTS: There were 690 cases and 2747 controls. The median duration of T2DM was 1178 days in controls and 1180 days in cases. Higher cumulative dose categories: 4 (adjusted odds ratio(aOR) = 0.72, 95% CI 0.55-0.95,p = 0.02), and 5 (OR = 0.60, 95% CI 0.42-0.85,p < 0.01) had significantly lower odds of HR + /HER2- breast cancer compared to category 0. The highest intensity category of metformin had 39% lower odds of HR + /HER2- breast cancer (OR = 0.61, 95% CI 0.46-0.82,p < 0.01) compared to the 0 mg/day group. CONCLUSIONS: Higher metformin exposure was associated with reduced risk of HR + /HER2- breast cancer, adding to the evidence supporting metformin's chemoprotective effect.