Studying prescription drug use and outcomes with medicaid claims data: strengths, limitations, and strategies

Management and outcomes of heart failure hospitalization among older adults in the United States and Japan

AIMS

Despite advances in therapies, the disease burden of heart failure (HF) has been rising globally. International comparisons of HF management and outcomes may reveal care patterns that improve outcomes. Accordingly, we examined clinical management and patient outcomes in older adults hospitalized for acute HF in the United States (US) and Japan.

METHODS

We identified patients aged >65 who were hospitalized for HF in 2013 using US Medicare data and the Japanese Registry of Acute Decompensated Heart Failure (JROADHF). We described patient characteristics, management, and healthcare utilization and compared outcomes using multivariable Cox regression during and after HF hospitalization.

RESULTS

Among 11 193 Japanese and 120 289 US patients, age and sex distributions were similar, but US patients had higher comorbidity rates. The length of stay was longer in Japan (median 18 vs. 5 days). While Medicare patients had higher use of implantable cardioverter defibrillator or cardiac resynchronization therapy during hospitalization (1.32% vs. 0.6%), Japanese patients were more likely to receive cardiovascular medications at discharge and to undergo cardiac rehabilitation within 3 months of HF admission (31% vs. 1.6%). Physician follow-up within 30 days was higher in Japan (77% vs. 57%). Cardiovascular readmission, cardiovascular mortality and all-cause mortality were 2.1-3.7 times higher in the US patients. The per-day cost of hospitalization was lower in Japan ($516 vs. $1323).

CONCLUSIONS

We observed notable differences in the management, outcomes and costs of HF hospitalization between the US and Japan. Large differences in length of hospitalization, cardiac rehabilitation rate and outcomes warrant further research to determine the optimal length of stay and assess the benefits of inpatient cardiac rehabilitation to reduce rehospitalization and mortality.

Adherence and persistence to tafamidis treatment among Medicare beneficiaries in the presence of a patient assistance program

Tafamidis is the only disease-modifying therapy approved to treat patients in the United States with transthyretin amyloid cardiomyopathy (ATTR-CM), which most commonly affects patients aged ≥ 65 years. The manufacturer operates a patient assistance program (PAP) to support access to tafamidis. This study conducted Privacy Preserving Record Linking (PPRL) using Datavant tokens to match patients across Medicare prescription drug plan (PDP) and PAP databases to evaluate the impact of PAPs on treatment exposure classification, adherence, and persistence determined using Medicare PDP data alone. We found 35% of Medicare PDP patients received tafamidis through the PAP only; 14% through both Medicare PDP and the PAP, and 51% through Medicare PDP only. Adherence and persistence were comparable between these cohorts but underestimated among patients who received ≥ 2 prescriptions through Medicare PDP and ≥ 1 through the PAP when solely using Medicare data versus pooled Medicare and PAP data (modified Medication Possession Ratio: 84% [69% ≥ 80% adherent] vs. 96% [93%]; Proportion of Days Covered: 77% [66% ≥ 80% adherent] vs. 88% [88%]; mean days to discontinuation: 186 vs. 252; total discontinuation: 13% vs. 11%). Cross-database PPRL is a valuable method to build more complete treatment journeys and reduce the risk of exposure misclassification in real-world analyses.

Multilevel Interventions and Dental Attendance in Pediatric Primary Care: A Cluster Randomized Clinical Trial

Importance

Untreated tooth decay is disproportionately present among low-income young children. While American Academy of Pediatrics (AAP) guidelines require pediatric clinicians to implement oral health care, the effectiveness of these oral health interventions has been inconclusive.

Objective

To test the effectiveness of multilevel interventions in increasing dental attendance and reducing untreated decay among young children attending well-child visits (WCVs).

Design, Setting, and Participants

The Pediatric Providers Against Cavities in Children's Teeth study is a cluster randomized clinical trial that was conducted at 18 pediatric primary care practices in northeast Ohio. The trial data were collected between November 2017 and July 2022, with data analyses conducted from August 2022 to March 2023. Eligible participants included Medicaid-enrolled preschoolers aged 3 to 6 years attending WCVs at participating practices who were enrolled at baseline (WCV 1) and followed-up for 2 consecutive examinations (WCV 2 and WCV 3).

Interventions

Clinicians in the intervention group received both the practice-level (electronic medical record changes to document oral health) and clinician-level (common-sense model of self-regulation theory-based oral health education and skills training) interventions. Control group clinicians received AAP-based standard oral health education alone.

Main Outcomes and Measures

Dental attendance was determined through clinical dental examinations conducted by hygienists utilizing International Caries Detection and Assessment System criteria and also from Medicaid claims data. Untreated decay was determined through clinical examinations. A generalized estimating equations (GEE) approach was used for both clinical examinations and Medicaid claims data.

Results

Eighteen practices were randomized to either intervention or control. Participants included 63 clinicians (mean [SD] age, 47.0 [11.3] years; 48 female [76.2%] and 15 male [23.8%]; 28 in the intervention group [44.4%]; 35 in the control group [55.6%]) and 1023 parent-child dyads (mean [SD] child age, 56.1 [14.0] months; 555 male children [54.4%] and 466 female children [45.6%]; 517 in the intervention group [50.5%]; 506 in the control group [49.5%]). Dental attendance from clinical examinations was significantly higher in the intervention group (170 children [52.0%]) vs control group (150 children [43.1%]) with a difference of 8.9% (95% CI, 1.4% to 16.4%; P = .02). The GEE model using clinical examinations showed a significant increase in dental attendance in the intervention group vs control group (adjusted odds ratio, 1.34; 95% CI, 1.07 to 1.69). From Medicaid claims, the control group had significantly higher dental attendance than the intervention group at 2 years (332 children [79.6%] vs 330 children [73.7%]; P = .04) but not at 3 years. A clinically but not statistically significant reduction in mean number of untreated decay was found in the intervention group compared with controls (B = -0.27; 95% CI, -0.56 to 0.02).

Conclusions and Relevance

In this cluster randomized clinical trial, children in the intervention group had better dental outcomes as was evidenced by increased dental attendance and lower untreated decay. These findings suggest that intervention group clinicians comprehensively integrated oral health services into WCVs.

Trial Registration

ClinicalTrials.gov Identifier: NCT03385629.

Literature Review of Studies Using the National Database of the Health Insurance Claims of Japan (NDB): Limitations and Strategies in Using the NDB for Research

The use of the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB) for research has increased over time. Researchers need to understand the characteristics of the data to generate quality-assured evidence from the NDB. In this review, we mapped and characterized the limitations and related strategies using the NDB for research based on the descriptions of published NDB studies. To find studies that used Japanese healthcare claims data, we searched MEDLINE, EMBASE, and Ichushi-Web up to June 2023. Additionally, we hand-searched the NDB data publication list from the Ministry of Health, Labour and Welfare (2017-2023). We abstracted data based on the NDB data type, research themes, age of the study sample or population, targeted disease, and the limitations and strategies in the NDB studies. Ultimately, 267 studies were included. Overall, the most common research theme was describing and estimating the prescriptions and treatment patterns (125 studies, 46.8%). There was a variation in the frequency of themes according to the type of NDB data. We identified the following categories of limitations: (1) lack of information on confounders/covariates, outcomes, and other clinical content, (2) limitations regarding patients not included in the NDB, (3) misclassification of data, (4) lack of unique identifiers and register of beneficiaries, and (5) others. Although the included studies noted several limitations of using the NDB for research, they also provided some strategies to address them. Organizing the limitations of NDB in research and the related strategies across research fields can help support high-quality NDB studies.

Prevalence and Risk of Behavioral Symptoms among Patients with Insomnia and Alzheimer's Disease: A Retrospective Database Analysis

OBJECTIVES

This study evaluated the prevalence and patterns of behavioral symptoms, including agitation/aggression (AA), psychotic symptoms (PS), anxiety/mood disorders (MD), and delirium among patients with Alzheimer's disease (AD) and their association with diagnosed insomnia.

DESIGN

A retrospective cohort analysis was conducted using the MarketScan Multi-State Medicaid Database 2016-2020.

SETTING AND PARTICIPANTS

Patients aged ≥50 with newly diagnosed AD (N = 56,904) were identified during 2017-2019 and categorized into insomnia and non-insomnia groups based on billing codes recorded in medical and pharmacy claims.

METHODS

The index date was defined as the earliest date of diagnosis/medication of insomnia. The new diagnosis of AD had to be established within 12 months before (baseline) or 3 months after the index date. Point prevalence of behavioral symptoms was estimated during baseline and the 12-month follow-up period. Propensity score matching was performed to match patients with and without insomnia. Multivariable conditional logistic regression was used to assess the risk of diagnosis of behavioral symptoms among insomnia and non-insomnia groups.

RESULTS

The study cohort included 7808 patients with newly diagnosed AD (mean age = 79.4, SD = 9.6 years). The point prevalence of behavioral symptoms was as follows: among those with insomnia (n = 3904), in the baseline, AA = 9.0%, PS = 12.5%, and MD = 57.8%, and during the follow-up, AA = 13.9%, PS = 16.3%, and MD = 72.1%; among those without insomnia (n = 3904), in the baseline, AA = 6.2%, PS = 9.2%, and MD = 41.4%; and during the follow-up, AA = 7.4%, PS = 10.4%, and MD = 49.2%. The likelihood of being diagnosed with any behavioral symptoms in the follow-up period was significantly higher among patients with insomnia than those without [adjusted odds ratio (OR), 2.7; 95% confidence interval (CI), 2.4-3.1].

CONCLUSIONS AND IMPLICATIONS

In patients with AD, prevalence of behavioral symptoms and likelihood of being diagnosed with behavioral symptoms were significantly higher among patients with diagnosed insomnia. Further investigation is needed to understand the relationship between insomnia and behavioral symptoms in patients with AD.

Accuracy of administrative claims prescription fill data to estimate glucocorticoid use and dose in patients with rheumatoid arthritis

PURPOSE

To assess accuracy of administrative claims prescription fill-based estimates of glucocorticoid use and dose, and approximate bias from glucocorticoid exposure misclassification.

METHODS

We identified adults with rheumatoid arthritis with linked Medicare and CorEvitas registry data. An algorithm identifying glucocorticoid use and average dose over 90 days from Medicare prescription fills was compared to physician-reported measures from a CorEvitas visit during the same period, using weighted kappa to compare doses (none, ≤5 mg, 5-10 mg, >10 mg/day). A deterministic sensitivity analysis examined the effect of exposure misclassification on estimated glucocorticoid-associated infection risk from a prior study.

RESULTS

We identified 621 observations among 494 patients. Prescription fills identified glucocorticoid use in 41.9% of observations versus 31.1% identified by CorEvitas physician-report. For glucocorticoid use (yes/no), prescription fills had sensitivity 88.1% (95% CI 82.7-92.3), specificity 79.0% (74.8-82.7), PPV 65.4% (59.3-71.2), NPV 93.6% (90.6-95.9), and 81.8% agreement with CorEvitas, with kappa 0.61 (moderate to substantial agreement). There was 89.5% agreement between prescription fills and physician-reported doses, with weighted kappa 0.56 (moderate agreement). Applying these results to a prior Medicare study evaluating glucocorticoid-associated infection risk [risk ratio 1.44 (95% CI 1.41-1.48)] led to an externally adjusted risk ratio of 1.74 when accounting for exposure misclassification, representing -17% bias in infection risk estimate.

CONCLUSIONS

This study supports the use of claims data to estimate glucocorticoid use and dose, but investigators should account for exposure misclassification, which may lead to underestimates of glucocorticoid risks. Our results could be applied to adjust risk estimates in other studies that use prescription fills to estimate glucocorticoid use.

Age, period and cohort-related trends in prescription opioid use in the USA, 1999-2018

BACKGROUND

To provide insight into the longitudinal dynamics of opioid use throughout the overdose crisis, this study estimated the separate influences of age, period and cohort on prescription opioid use during 1999-2018 in the USA.

METHODS

Data from 10 cycles of the cross-sectional National Health and Nutrition Examination Survey were used to conduct an age-period-cohort analysis of the prevalence of prescription opioid use (n=63 500 across 1999-2018). Temporal trends were graphically visualised. The median polish approach was used to estimate age, period and cohort-related effects on prescription opioid use.

RESULTS

Prescription opioid use broadly increased across the lifespan, with steeper prevalence increases observed from young adulthood to mid-adulthood. Period-related variation was consistent with recognised nationwide declines in opioid prescribing. While there was no evidence of systematic cohort effects, compared with individuals born in 1951-1954, those born during 1963-1966 had greater prescription opioid use (prevalence ratio (PR)=1.23, 95% CI: 1.05 to 1.43), whereas the 1991-1994 and 1999-2002 cohorts had lower prescription opioid use (PR91-94=0.70, 95% CI: 0.50 to 0.98; PR99-02=0.72, 95% CI: 0.63 to 0.81).

CONCLUSION

In the USA, longitudinal trends in prescription opioid use during 1999-2018 were predominantly driven by age and period influences. The cohort of youngest baby boomers experienced greater prescription opioid use, whereas recent-born cohorts have had lower use. As the overdose crisis continues evolving, such population-level characterisations of age, period and cohort dynamics are instrumental in understanding opioid use and can inform prevention and intervention approaches by identifying population groups more likely to use opioids who, thus, may also experience related outcomes.

A cohort of patients in New York State with an alcohol use disorder and subsequent treatment information - A merging of two administrative data sources

OBJECTIVE

Despite the high prevalence of alcohol use disorder (AUD) in the United States, limited research is focused on the associations among AUD, pain, and opioids/benzodiazepine use. In addition, little is known regarding individuals with a history of AUD and their potential risk for pain diagnoses, pain prescriptions, and subsequent misuse. Moreover, the potential risk of pain diagnoses, prescriptions, and subsequent misuse among individuals with a history of AUD is not well known. The objective was to develop a tailored dataset by linking data from 2 New York State (NYS) administrative databases to investigate a series of hypotheses related to AUD and painful medical disorders.

METHODS

Data from the NYS Office of Addiction Services and Supports (OASAS) Client Data System (CDS) and Medicaid claims data from the NYS Department of Health Medicaid Data Warehouse (MDW) were merged using a stepwise deterministic method. Multiple patient-level identifier combinations were applied to create linkage rules. We included patients aged 18 and older from the OASAS CDS who initially entered treatment with a primary substance use of alcohol and no use of opioids between January 1, 2003, and September 23, 2019. This cohort was then linked to corresponding Medicaid claims.

RESULTS

A total of 177,685 individuals with a primary AUD problem and no opioid use history were included in the dataset. Of these, 37,346 (21.0%) patients had an OUD diagnosis, and 3,365 (1.9%) patients experienced an opioid overdose. There were 121,865 (68.6%) patients found to have a pain condition.

CONCLUSION

The integrated database allows researchers to examine the associations among AUD, pain, and opioids/benzodiazepine use, and propose hypotheses to improve outcomes for at-risk patients. The findings of this study can contribute to the development of a prognostic prediction model and the analysis of longitudinal outcomes to improve the care of patients with AUD.

Characteristics and Trends in HIV Testing Among Medicaid Enrollees Diagnosed as Having Schizophrenia

OBJECTIVE

People with schizophrenia have more HIV risk factors and higher rates of HIV infection than the general U.S. population. The authors aimed to examine HIV testing patterns in this population nationally and by demographic characteristics and presence of high-risk comorbid conditions.

METHODS

This retrospective longitudinal study compared HIV testing between Medicaid-only enrollees with schizophrenia and without schizophrenia during 2002-2012 (N=6,849,351). Interrupted time series were used to analyze the impacts of the 2006 federal policy change recommending expanded HIV testing. Among enrollees with schizophrenia, multivariable logistic regression was used to estimate associations between testing and both demographic characteristics and comorbid conditions. Sensitivity analyses were also conducted.

RESULTS

Enrollees diagnosed as having schizophrenia had consistently higher HIV testing rates than those without schizophrenia. When those with comorbid substance use disorders or sexually transmitted infections were excluded, testing was higher for individuals without schizophrenia (p<0.001). The federal policy change likely increased testing for both groups (p<0.001), but the net change was greater for those without schizophrenia (3.1 vs. 2.2 percentage points). Among enrollees with schizophrenia, testing rates doubled during 2002-2012 (3.9% to 7.2%), varied across states (range 17 percentage points), and tripled for those with at least one annual nonpsychiatric medical visit (vs. no visit; adjusted OR=3.10, 95% CI=2.99-3.22).

CONCLUSIONS

Nationally, <10% of enrollees with schizophrenia had annual HIV testing. Increases appear to be driven by high-risk comorbid conditions and nonpsychiatric encounters, rather than by efforts to target people with schizophrenia. Psychiatric guidelines for schizophrenia care should consider HIV testing alongside annual metabolic screening.

Improving Access to High-Value, High-Cost Medicines: The Use of Subscription Models to Treat Hepatitis C Using Direct-Acting Antivirals in the United States

State payers may face financial incentives to restrict use of high-cost medications. Yet, restrictions on access to high-value medications may have deleterious effects on population health. Direct-acting antivirals (DAAs), available since 2013, can cure chronic infection with hepatitis C virus (HCV). With prices upward of $90,000 for a treatment course, states have struggled to ensure access to DAAs for Medicaid beneficiaries and the incarcerated, populations with a disproportionate share of HCV. Advance purchase commitments (APCs), wherein a payer commits to purchase a certain quantity of medications at lower prices, offer payers incentives to increase access to high-value medications while also offering companies guaranteed revenue. This article discusses the use of subscription models, a type of APC, to support increased access to high-value DAAs for treating HCV. First, the authors provide background information about HCV, its treatment, and state financing of prescription medications. They then review the implementation of HCV subscription models in two states, Louisiana and Washington, and the early evidence of their impact. The article discusses challenges to evaluating state-sponsored subscription models, and it concludes by discussing implications of subscription models that target DAAs and other high-value, high-cost medicines.

Adherence to cardiovascular disease risk factor medications among patients with cancer: a systematic review

PURPOSE

The most common cause of mortality for many cancer survivors is cardiovascular disease (CVD). This requires a shift in thinking where control of CVD risk factor-related comorbidity is paramount. Our objective was to provide an understanding of adherence to medications for the management of CVD risk factor-related comorbidities among cancer survivors.

METHODS

We systematically searched for articles indexed in MEDLINE (via PubMed), Embase, Cochrane (Wiley), PsycINFO, and Scopus (via Elsevier) for articles published from inception to October 31, 2019, and updated the search on June 7, 2021. English language, original research that assessed medication adherence to common CVD risk factor-related comorbidities among cancer survivors was included. We assessed risk of bias using the Mixed Methods Appraisal Tool.

RESULTS

Of the 21 studies included, 57% focused on multiple cancer types. Seventy-one percent used pharmacy-based adherence measures. Two were prospective. Adherence was variable across cancer types and CVD risk factor-related comorbidities. Among the studies that examined changes in comorbid medication adherence, most noted a decline in adherence following cancer diagnosis and throughout cancer treatment. There was a focus on breast cancer populations.

CONCLUSIONS

CVD risk factor-related medication adherence is low among cancer survivors and declines over time. Given the risk for CVD-mortality among cancer survivors, testing of interventions aimed at improving adherence to non-cancer medications is critically needed.

IMPLICATIONS FOR CANCER SURVIVORS

For many cancer survivors, regularly taking medications to manage CVD risk is important for longevity. Engaging with primary care throughout the cancer care trajectory may be important to support cardiovascular health.

Utilization of chronic lung disease treatment before the respiratory syncytial virus season as palivizumab prophylaxis qualifier in the American Academy of Pediatrics Guidelines

Guidelines from the American Academy of Pediatrics recommend palivizumab immunoprophylaxis for children with CLD in their second year of life if they continue to need treatment within 6 months before the RSV season. The utilization patterns of treatment (chronic corticosteroid therapy, diuretic therapy, or supplemental oxygen) are not well understood. We examined variations in CLD treatment for ten consecutive 20-day segments preceding RSV season onset. Among infants and children with CLD (n = 19,026), 35.2% received one or more medical treatments for CLD any time within 200 days before entering the second RSV season: 8.6%, 3.2%, and 29.7% received supplemental oxygen, diuretics, and corticosteroids, respectively. Utilization decreased as infants' age increased with corticosteroids surpassing oxygen and diuretics. To avoid the capture of intermittent use of corticosteroids for acute infections, we found requiring a minimum of 45 days cumulative exposure was reasonable to determine chronic use. What is Known: • Guidelines from the American Academy of Pediatrics recommend palivizumab immunoprophylaxis for children with CLD in their second year of life if they continue to need treatment within 6 months before the RSV season. • The utilization patterns of treatment (chronic corticosteroid therapy, diuretic therapy, or supplemental oxygen) are not well understood. A definition of chronic corticosteroid therapy in this setting is not available. What is New: • Among infants and children with CLD of prematurity, 35.2% received one or more medical treatments for CLD any time within 200 days before entering the second RSV season: 8.6%, 3.2%, and 29.7% received oxygen, diuretics, and corticosteroids, respectively. Utilization decreased as infants' age increased with corticosteroids surpassing oxygen and diuretics. • A minimum of 45 days cumulative corticosteroid use within the past 90 days would accurately capture chronic use to fulfill criteria for immunoprophylaxis while limiting the inclusion of intermittent use of corticosteroids for acute infections.

Calibration of Chronic Lung Disease Severity as a Risk Factor for Respiratory Syncytial Virus Hospitalization

BACKGROUND

Guidelines assume children with chronic lung disease (CLD) who require medical support within 6 months before the second respiratory syncytial virus (RSV) season remains at high risk of severe RSV disease. We determined the number of days since the last treatment (DSL) when the risk of RSV hospitalization among children with CLD becomes equivalent to the risk for those not qualified for immunoprophylaxis.

METHODS

The study cohort was assembled using Medicaid billing records from 1999 to 2010 linked to Florida and Texas birth certificate records. We developed DSL-trend discrete time logistic regression models within a survival analysis framework, adjusting for use of immunoprophylaxis, to compare the hospitalization risk of CLD infants at 4 age points to that of term infants at 1 month of age with siblings.

RESULTS

The study cohort included 858 830 healthy term and 5562 preterm infants with CLD. Among 1-month-old term infants, the RSV hospitalization risk averaged across all covariate strata was 14.8 (95% confidence interval [CI], 13.5-16.1) per 1000 patient season-months. Risk for preterm CLD children reached the threshold derived from term infants when DSL was 76 (95% CI, 22-198.5), 52 (95% CI, 6.5-123), 35 (95% CI, 0-93.5), and 12 (95% CI, 0-61.5) at the respective ages of 12, 15, 17.2, and 21 months.

CONCLUSIONS

The 180-day threshold used to define CLD severity at season start can be shortened to 120 days, 90 days, and 60 days for children with CLD at age 15, 17.2, and 21 months, respectively.

A best-worst scaling experiment to identify patient-centered claims-based outcomes for evaluation of pediatric antipsychotic monitoring programs

Objective

This article employs a best‐worst scaling (BWS) experiment to identify the claims‐based outcomes that matter most to patients and other relevant parties when evaluating pediatric antipsychotic monitoring programs in the United States.

Data Sources

Patients and relevant parties, with pediatric antipsychotic oversight and treatment experience, completed a BWS experiment, including policymakers (n = 31), foster care alumni (n = 28), caseworkers (n = 23), prescribing clinicians (n = 32), and caregivers (n = 18).

Study Design

Respondents received surveys with a scenario on antipsychotic monitoring programs and ranked 11 candidate claims‐based outcomes as most and least important for program evaluation.

Data Analysis

Stratified by respondent group, best‐worst scores were calculated to identify the relative importance of the claims‐based outcomes. A conditional logit examined whether candidate outcomes for safety, quality, and unintended consequences were preferred over reduction in antipsychotic treatment, the outcome used most often to evaluate antipsychotic monitoring programs.

Principal Findings

Safety indicators (eg, antipsychotic co‐pharmacy, cross‐class polypharmacy, higher than recommended doses) ranked among the top three candidate outcomes across respondent groups and were an important complement to antipsychotic treatment reduction. Foster care alumni prioritized “antipsychotic treatment reduction” and “increased psychosocial treatment.” Caseworkers, prescribers, and caregivers prioritized “increased follow‐up after treatment initiation.” Potential unintended consequences of an antipsychotic monitoring program ranked lowest, including increased use of other psychotropic medication classes (as a substitute), increased psychiatric hospital stays, and increased emergency room utilization. Results of the conditional logit model found only caregivers significantly preferred other indicators over antipsychotic treatment reduction, preferring improvements in follow‐up care (5.78) and psychosocial treatment (4.53) and reduction in prescriptions of higher than recommended doses (3.64).

Conclusions

The BWS experiment supported rank ordering of candidate claims‐based outcomes demonstrating the opportunity for future studies to align outcomes used in antipsychotic monitoring program evaluations with community preferences, specifically by diversifying metrics to include safety and quality indicators.

Structural and Social Determinants of Health Factors Associated with County-Level Variation in Non-Adherence to Antihypertensive Medication Treatment

Background: Non-adherence to antihypertensive medication treatment (AHM) is a complex health behavior with determinants that extend beyond the individual patient. The structural and social determinants of health (SDH) that predispose populations to ill health and unhealthy behaviors could be potential barriers to long-term adherence to AHM. However, the role of SDH in AHM non-adherence has been understudied. Therefore, we aimed to define and identify the SDH factors associated with non-adherence to AHM and to quantify the variation in county-level non-adherence to AHM explained by these factors. Methods: Two cross-sectional datasets, the Centers for Disease Control and Prevention (CDC) Atlas of Heart Disease and Stroke (2014–2016 cycle) and the 2016 County Health Rankings (CHR), were linked to create an analytic dataset. Contextual SDH variables were extracted from the CDC-CHR linked dataset. County-level prevalence of AHM non-adherence, based on Medicare fee-for-service beneficiaries’ claims data, was extracted from the CDC Atlas dataset. The CDC measured AHM non-adherence as the proportion of days covered (PDC) with AHM during a 365 day period for Medicare Part D beneficiaries and aggregated these measures at the county level. We applied confirmatory factor analysis (CFA) to identify the constructs of social determinants of AHM non-adherence. AHM non-adherence variation and its social determinants were measured with structural equation models. Results: Among 3000 counties in the U.S., the weighted mean prevalence of AHM non-adherence (PDC < 80%) in 2015 was 25.0%, with a standard deviation (SD) of 18.8%. AHM non-adherence was directly associated with poverty/food insecurity (β = 0.31, P-value < 0.001) and weak social supports (β = 0.27, P-value < 0.001), but inversely with healthy built environment (β = −0.10, P-value = 0.02). These three constructs explained one-third (R² = 30.0%) of the variation in county-level AHM non-adherence. Conclusion: AHM non-adherence varies by geographical location, one-third of which is explained by contextual SDH factors including poverty/food insecurity, weak social supports and healthy built environments.

The impact of cognitive impairment on survival and medication adherence among older women with breast cancer

INTRODUCTION

The purpose of this study was to examine the impact of preexisting cognitive impairments on survival and medication adherence, and whether chronic medication adherence mediates or moderates the association between cognitive impairments and mortality in patients with breast cancer.

METHODS

This retrospective cohort study of older female patients diagnosed with breast cancer was conducted using the Surveillance, Epidemiology, and End Results Medicare Linked Database. We examined the risk of mortality from cancer and non-cancer causes in patients with and without a history of cognitive impairment. In addition, we examined if chronic medication adherence rates differ between these groups of patients and if medication adherence mediates or moderates the association between cognitive impairments and non-cancer mortality.

RESULTS

Mortality from cancer-specific (HR 1.13, 95% CI 1.04-1.23) and non-cancer causes (HR 1.16, 95% CI 1.11-1.21) as well as all-cause mortality (HR 1.30, 95% CI 1.23-1.38) was significantly higher in patients with cognitive impairments compared to those without cognitive impairment. Both groups showed low adherence levels to chronic medication before and after the breast cancer diagnosis. Further analysis did not show that medication adherence mediates or moderates the relationship between cognitive impairment and non-cancer mortality (p value > 0.05).

CONCLUSION

The results of this study indicate that older female patients with cognitive impairments and a breast cancer diagnosis have a heightened risk of cancer-specific and non-cancer mortality. Our findings do not indicate that chronic medication adherence plays a role in the association between a history of cognitive impairment and mortality, it is still necessary to further investigate this issue.

Who are we missing? Underrepresentation of data sources used for pharmacoepidemiology research in the United States

PURPOSE

Research using healthcare databases often includes patients frequently excluded from clinical trials; yet it is not known whether commonly used data represents the overall population or specific sub-populations of interest. We aimed to examine population representativeness from data sources in recent research studies in the United States (US).

METHODS

We identified data sources from abstracts accepted to the 34th International Conference on Pharmacoepidemiology & Therapeutic Risk Management. The final sample included research studies using ≥1 data source from the US. We classified data sources broadly as claims, linkage, electronic health records (EHR), survey, distributed data network, and other. Studies using claims and EHRs were further classified into more specific categories, including special populations of interest (eg, children).

RESULTS

We identified 356 abstracts. The majority used claims data (n = 201, 56.5%), followed by data linkages (n = 46, 12.9%), and EHR data (n = 39, 11.0%). Among EHR studies, most (n = 16, 41.0%) came from network data sources (eg, Kaiser Permanente). Almost half (49.4%) of claims-based studies used commercial claims data sources, followed by Medicare (22.1%), Medicaid (11.3%), and Medicare Supplemental (6.1%). Only 15% of studies included children in the study population (n = 53), with 8% focused on a pediatric topic (n = 27).

CONCLUSIONS

We find that certain populations in the US are under-represented in pharmacoepidemiology, particularly Medicaid enrollees and children. Researchers should strive to utilize data sources that may be more representative of the US population, particularly vulnerable populations.

Validation of an Optimized Algorithm for Identifying Persons Living With Diagnosed HIV From New York State Medicaid Data, 2006-2014

Algorithms are regularly used to identify persons living with diagnosed human immunodeficiency virus (HIV) (PLWDH) in Medicaid data. To our knowledge, there are no published reports of an HIV algorithm from Medicaid claims codes that have been compared with an HIV surveillance system to assess its sensitivity, specificity, positive predictive value, and negative predictive value in identifying PLWDH. Therefore, our aims in this study were to 1) develop an algorithm that could identify PLWDH in New York State Medicaid data from 2006-2014 and 2) validate this algorithm using the New York State HIV surveillance system. Classification and regression tree analysis identified 16 nodes that we combined to create a case-finding algorithm with 5 criteria. This algorithm identified 86,930 presumed PLWDH, 88.0% of which were verified by matching to the surveillance system. The algorithm yielded a sensitivity of 94.5%, a specificity of 94.4%, a positive predictive value of 88.0%, and a negative predictive value of 97.6%. This validated algorithm has the potential to improve the utility of Medicaid data for assessing health outcomes and programmatic interventions.

The present and future scope of real-world evidence research in diabetes: What questions can and cannot be answered and what might be possible in the future?

The last decade has witnessed an exponential growth in the opportunities to collect and link health-related data from multiple resources, including primary care, administrative, and device data. The availability of these "real-world," "big data" has fuelled also an intense methodological research into methods to handle them and extract actionable information. In medicine, the evidence generated from "real-world data" (RWD), which are not purposely collected to answer biomedical questions, is commonly termed "real-world evidence" (RWE). In this review, we focus on RWD and RWE in the area of diabetes research, highlighting their contributions in the last decade; and give some suggestions for future RWE diabetes research, by applying well-established and less-known tools to direct RWE diabetes research towards better personalized approaches to diabetes care. We underline the essential aspects to consider when using RWD and the key features limiting the translational potential of RWD in generating high-quality and applicable RWE. Only if viewed in the context of other study designs and statistical methods, with its pros and cons carefully considered, RWE will exploit its full potential as a complementary or even, in some cases, substitutive source of evidence compared to the expensive evidence obtained from randomized controlled trials.

Economic burden associated with inadequate antidepressant medication management among patients with depression and known cardiovascular diseases: insights from a United States-based retrospective claims database analysis

Aims: The current study examined the association between insufficient major depressive disorder (MDD) care and healthcare resource use (HCRU) and costs among patients with prior myocardial infarction (MI) or stroke.Methods: This was a retrospective study conducted using the MarketScan Claims Database (2010-2015). The date of the first MI/stroke diagnosis was defined as the cardiovascular disease (CVD) index date and the first date of a subsequent MDD diagnosis was the index MDD date. Adequacy of MDD care was assessed during the 90 days following the index MDD date (profiling period) using 2 measures: dosage adequacy (average fluoxetine equivalent dose of ≥20 mg/day for nonelderly and ≥10 mg/day for elderly patients) and duration adequacy (measured as the proportion of days covered of 80% or higher for all MDD drugs). Study outcomes included all-cause and CVD-related HCRU and costs which were determined from the end of the profiling period until the end of study follow-up. Propensity-score adjusted generalized linear models (GLMs) were used to compare patients receiving adequate versus inadequate MDD care in terms of study outcomes.Results: Of 1,568 CVD patients who were treated for MDD, 937 (59.8%) were categorized as receiving inadequate MDD care. Results from the GLMs suggested that patients receiving inadequate MDD care had 14% more all-cause hospitalizations, 4% more all-cause outpatient visits, 17% more CVD-related outpatient visits, 13% more CVD-related emergency room (ER) visits, higher per patient per year CVD-related hospitalization costs ($21,485 vs. $17,756), higher all-cause outpatient costs ($2,820 vs. $2,055), and higher CVD-related outpatient costs ($520 vs. $434) compared to patients receiving adequate MDD care.Limitations: Clinical information such as depression severity and frailty, which are potential predictors of adverse CVD outcomes, could not be ascertained using administrative claims data.Conclusions: Among post-MI and post-stroke patients, inadequate MDD care was associated with a significantly higher economic burden.

Defining a recovery-oriented cascade of care for opioid use disorder: A community-driven, statewide cross-sectional assessment

BACKGROUND

In light of the accelerating and rapidly evolving overdose crisis in the United States (US), new strategies are needed to address the epidemic and to efficiently engage and retain individuals in care for opioid use disorder (OUD). Moreover, there is an increasing need for novel approaches to using health data to identify gaps in the cascade of care for persons with OUD.

METHODS AND FINDINGS

Between June 2018 and May 2019, we engaged a diverse stakeholder group (including directors of statewide health and social service agencies) to develop a statewide, patient-centered cascade of care for OUD for Rhode Island, a small state in New England, a region highly impacted by the opioid crisis. Through an iterative process, we modified the cascade of care defined by Williams et al. for use in Rhode Island using key national survey data and statewide health claims datasets to create a cross-sectional summary of 5 stages in the cascade. Approximately 47,000 Rhode Islanders (5.2%) were estimated to be at risk for OUD (stage 0) in 2016. At the same time, 26,000 Rhode Islanders had a medical claim related to an OUD diagnosis, accounting for 55% of the population at risk (stage 1); 27% of the stage 0 population, 12,700 people, showed evidence of initiation of medication for OUD (MOUD, stage 2), and 18%, or 8,300 people, had evidence of retention on MOUD (stage 3). Imputation from a national survey estimated that 4,200 Rhode Islanders were in recovery from OUD as of 2016, representing 9% of the total population at risk. Limitations included use of self-report data to arrive at estimates of the number of individuals at risk for OUD and using a national estimate to identify the number of individuals in recovery due to a lack of available state data sources.

CONCLUSIONS

Our findings indicate that cross-sectional summaries of the cascade of care for OUD can be used as a health policy tool to identify gaps in care, inform data-driven policy decisions, set benchmarks for quality, and improve health outcomes for persons with OUD. There exists a significant opportunity to increase engagement prior to the initiation of OUD treatment (i.e., identification of OUD symptoms via routine screening or acute presentation) and improve retention and remission from OUD symptoms through improved community-supported processes of recovery. To do this more precisely, states should work to systematically collect data to populate their own cascade of care as a health policy tool to enhance system-level interventions and maximize engagement in care.

Smoking and Smoking Cessation Among Persons with Tobacco- and Non-tobacco-Associated Cancers

PURPOSE

To examine smoking and use of smoking cessation aids among tobacco-associated cancer (TAC) or non-tobacco-associated cancer (nTAC) survivors. Understanding when and if specific types of cessation resources are used can help with planning interventions to more effectively decrease smoking among all cancer survivors, but there is a lack of research on smoking cessation modalities used among cancer survivors.

METHODS

Kentucky Cancer Registry data on incident lung, colorectal, pancreatic, breast, ovarian, and prostate cancer cases diagnosed 2007-2011, were linked with health administrative claims data (Medicaid, Medicare, private insurers) to examine the prevalence of smoking and use of smoking cessation aids 1 year prior and 1 year following the cancer diagnosis. TACs included colorectal, pancreatic, and lung cancers; nTAC included breast, ovarian, and prostate cancers.

RESULTS

There were 10,033 TAC and 13,670 nTAC survivors. Smoking before diagnosis was significantly higher among TAC survivors (p < 0.0001). Among TAC survivors, smoking before diagnosis was significantly higher among persons who: were males (83%), aged 45-64 (83%), of unknown marital status (84%), had very low education (78%), had public insurance (89%), Medicaid (85%) or were uninsured (84%). Smoking cessation counseling and pharmacotherapy were more common among TAC than nTAC survivors (p < 0.01 and p = 0.05, respectively).

DISCUSSION

While smoking cessation counseling and pharmacotherapy were higher among TAC survivors, reducing smoking among all cancer survivors remains a priority, given cancer survivors are at increased risk for subsequent chronic diseases, including cancer. Tobacco cessation among all cancer survivors (not just those with TAC) can help improve prognosis, quality of life and reduce the risk of further disease. Health care providers can recommend for individual, group and telephone counseling and/or pharmacotherapy recommendations. These could also be included in survivorship care plans.

Identifying Smoking Status and Smoking Cessation Using a Data Linkage Between the Kentucky Cancer Registry and Health Claims Data

PURPOSE

Linkage of cancer registry data with complementary data sources can be an informative way to expand what is known about patients and their treatment and improve delivery of care. The purpose of this study was to explore whether patient smoking status and smoking-cessation modalities data in the Kentucky Cancer Registry (KCR) could be augmented by linkage with health claims data.

METHODS

The KCR conducted a data linkage with health claims data from Medicare, Medicaid, state employee insurance, Humana, and Anthem. Smoking status was defined as documentation of personal history of tobacco use (International Classification of Diseases, Ninth Revision [ICD-9] code V15.82) or tobacco use disorder (ICD-9 305.1) before and after a cancer diagnosis. Use of smoking-cessation treatments before and after the cancer diagnosis was defined as documentation of smoking-cessation counseling (Healthcare Common Procedure Coding System codes 99406, 99407, G0375, and G0376) or pharmacotherapy (eg, nicotine replacement therapy, bupropion, varenicline).

RESULTS

From 2007 to 2011, among 23,703 patients in the KCR, we discerned a valid prediagnosis smoking status for 78%. KCR data only (72%), claims data only (6%), and a combination of both data sources (22%) were used to determine valid smoking status. Approximately 4% of patients with cancer identified as smokers (n = 11,968) and were provided smoking-cessation counseling, and 3% were prescribed pharmacotherapy for smoking cessation.

CONCLUSION

Augmenting KCR data with medical claims data increased capture of smoking status and use of smoking-cessation modalities. Cancer registries interested in exploring smoking status to influence treatment and research activities could consider a similar approach, particularly if their registry does not capture smoking status for a majority of patients.

Identifying incident cancer cases in dispensing claims: A validation study using Australia's Repatriation Pharmaceutical Benefits Scheme (PBS) data

Introduction

Dispensing claims are used commonly as proxy measures in pharmacoepidemiological studies; however, their validity is often untested.

Objectives

To assess the performance of a proxy for identifying cancer cases based on the dispensing of anticancer medicines and estimate the misclassification of cancer status and potential for bias researchers may encounter when using this proxy.

Methods

We conducted our validation study using Department of Veterans’ Affairs (DVA) client data linked with the New South Wales (NSW) Cancer Registry and Repatriation Pharmaceutical Benefits Scheme data. We included DVA clients aged ≥65 years residing in NSW between July 2004 and December 2012. We matched clients with a cancer diagnosis to clients without a diagnosis based on demographic characteristics and available observation time. We used dispensing claims for anticancer medicines dispensed between July 2004 and December 2013 as a proxy to identify clients with cancer and calculated sensitivity, specificity, positive predictive values and negative predictive values compared with cancer registrations (gold standard), overall and by cancer site. We illustrated misclassification by the proxy in a cohort of people initiating opioid therapy. Using the proxy, we excluded people with cancer from the cohort, in an attempt to delineate people potentially using opioids for cancer rather than chronic non-cancer pain.

Results

We identified 15,679 new cancer diagnoses in 14,112 DVA clients from the cancer registry and 62,663 clients without a diagnosis. Sensitivity of the proxy based on dispensing claims was 30% for all cancers and around 20% for specific cancers (range: 10-67%). Specificity was above 90% for all cancers. The dispensing proxy correctly identified 26% of people with a cancer diagnosis who initiated opioid therapy and failed to identify 74% those with a cancer diagnosis; the proxy was most robust for clients with breast cancer where 61% were correctly identified by proxy.

Conclusions

Using dispensing of anticancer medicines to identify people with a cancer diagnosis performed poorly. Excluding patients with evidence of anticancer medicine use from cohort studies may result removal of a disproportionate number of women with breast cancer. Researchers excluding or otherwise using anticancer medicine dispensing to identify people with cancer in pharmacoepidemiological studies should acknowledge the potential biases introduced to their findings.

Keywords

cancer, diagnosis, proxy, dispensing records, validation study

Use of electronic health record data from diverse primary care practices to identify and characterize patients' prescribed common medications

We use prescription of statin medications and prescription of warfarin to explore the capacity of electronic health record data to (1) describe cohorts of patients prescribed these medications and (2) identify cohorts of patients with evidence of adverse events related to prescription of these medications. This study was conducted in the WWAMI region Practice and Research Network (WPRN)., a network of primary care practices across Washington, Wyoming, Alaska, Montana and Idaho DataQUEST, an electronic data-sharing infrastructure. We used electronic health record data to describe cohorts of patients prescribed statin or warfarin medications and reported the proportions of patients with adverse events. Among the 35,445 active patients, 1745 received at least one statin prescription and 301 received at least one warfarin prescription. Only 3 percent of statin patients had evidence of myopathy; 51 patients (17% of those prescribed warfarin) had a bleeding complication. Primary-care electronic health record data can effectively be used to identify patients prescribed specific medications and patients potentially experiencing medication adverse events.

Characteristics of Medicaid Recipients Receiving Persistent Antipsychotic Polypharmacy

Antipsychotic polypharmacy (APP) is a common strategy despite guidelines advising against this practice. This article seeks to quantify the prevalence and correlates of APP using Medicaid Analytic eXtract files from 2003 to 2004. Nineteen percent of Medicaid recipients who received an antipsychotic were treated with APP. Individuals who received APP were more likely to be white, male, disabled, between the ages of 18-29, diagnosed with a psychotic disorder, and diagnosed with a higher number of psychiatric conditions. Geographic variation in APP rates was also observed. Quality improvement initiatives may help reduce APP for medically vulnerable patients.

Safety of ibuprofen in infants younger than six months: A retrospective cohort study

Objective

We hypothesized (1) that gastrointestinal (GI) and renal adverse events (AE) would occur more often in infants first prescribed ibuprofen before rather than after six months of age and (2) that ibuprofen would be associated with more adverse effects than acetaminophen in infants younger than six months.

Methods

We created two partly overlapping retrospective cohorts of infants aged less than six months when California Medicaid first paid for ibuprofen or acetaminophen between 2004 and 2010. In the first cohort we compared the incidence rate ratio (RR) of GI and renal AE between those infants first prescribed ibuprofen before six months of age with those first prescribed ibuprofen after six months of age. In the second cohort we compared the RR of GI and renal AE between infants younger than six months prescribed ibuprofen (+/-acetaminophen) with those prescribed only acetaminophen.

Results

We identified 41,669 prescriptions for ibuprofen and 176,991 prescriptions for acetaminophen in 180,333 eligible infants (median age 2.1 months). We did not observe higher RR of any AE in infants first prescribed ibuprofen before rather than after six months of age. Most infants prescribed ibuprofen were also prescribed acetaminophen. Any GI (adjusted (a)RR 1.25, 95% CI 1.13–1.38) and moderate or severe GI AE (aRR 1.24, 95% CI 1.09–1.40) were more common in infants younger than six months who were prescribed ibuprofen versus acetaminophen alone. Severe GI (aRR 0.63, 95% CI 0.27–1.45) and renal AE (aRR 1.84 95% CI 0.66–5.19) were not different between the ibuprofen (+/-acetaminophen) and acetaminophen-only groups.

Conclusions

GI and renal AEs were not higher in infants younger than six months who were prescribed ibuprofen compared with those aged six to 12 months. AEs were increased in infants younger than six months who were prescribed ibuprofen compared with infants who were prescribed acetaminophen alone.

State-level estimation of diabetes and prediabetes prevalence: Combining national and local survey data and clinical data

Many statisticians and policy researchers are interested in using data generated through the normal delivery of health care services, rather than carefully designed and implemented population-representative surveys, to estimate disease prevalence. These larger databases allow for the estimation of smaller geographies, for example, states, at potentially lower expense. However, these health care records frequently do not cover all of the population of interest and may not collect some covariates that are important for accurate estimation. In a recent paper, the authors have described how to adjust for the incomplete coverage of administrative claims data and electronic health records at the state or local level. This article illustrates how to adjust and combine multiple data sets, namely, national surveys, state-level surveys, claims data, and electronic health record data, to improve estimates of diabetes and prediabetes prevalence, along with the estimates of the method's accuracy. We demonstrate and validate the method using data from three jurisdictions (Alabama, California, and New York City). This method can be applied more generally to other areas and other data sources.

Long-term etanercept retention patterns and factors associated with treatment discontinuation: a retrospective cohort study using Canadian claims-level data

To examine 12-month retention rates over 6 years of etanercept patients in Canada, and to identify factors associated with treatment discontinuation. A retrospective cohort study was conducted using longitudinal prescription drug claims data from IQVIA Private Drug Plan database (PDP), Ontario Public Drug Plan database (OPDP), and Régie de l'assurance maladie du Québec database (RAMQ). Between 07/2008 and 06/2010, bio-naïve patients who initiated etanercept were identified and followed for 72 months. Twelve-month retention rates were estimated in one-year increments and factors associated with time to discontinuation over the 72-month period were identified using a Cox proportional hazards regression model. The study identified 4528 etanercept patients (61% female, 85% rheumatic diseases, and 15% psoriasis). Twelve-month etanercept retention rates increased significantly for patients following their first year on therapy (p < 0.0001), with 66% of patients retained at year 1 vs. 79, 82, 84, 83, and 79% at years 2, 3, 4, 5, and 6, respectively. 17.1% (n = 771) of patients were retained for the entire 72-month study. Patients with psoriasis were at increased risk (HR 1.199; p < 0.0001); while public drug coverage plan patients (OPDP HR 0.721; p < 0.0001 and RAMQ HR 0.537; p < 0.0001) were at decreased risk of treatment discontinuation. Twelve-month etanercept retention rates increased significantly for patients following their first year on therapy. Indication and drug coverage plan were associated with patients' time to etanercept discontinuation. With a better understanding of factors associated with retention, programs can be designed to address the specific needs of at-risk groups while supporting patients stable on therapy.

Updating the Evidence of the Interaction Between Clopidogrel and CYP2C19-Inhibiting Selective Serotonin Reuptake Inhibitors: A Cohort Study and Meta-Analysis

INTRODUCTION

We previously found that patients who initiate clopidogrel while treated with a cytochrome P450 (CYP) 2C19-inhibiting selective serotonin reuptake inhibitor (SSRI) have a higher risk of subsequent ischemic events than patients treated with other SSRIs. It is not known whether initiating an inhibiting SSRI while treated with clopidogrel will also increase risk of ischemic events.

OBJECTIVE

The aim of this study was to assess clinical outcomes following initiation of a CYP2C19-inhibiting SSRI versus initiation of other SSRIs among patients treated with clopidogrel and to update existing evidence on the clinical impact of clopidogrel-SSRI interaction.

METHODS

Using five US databases (1998-2013), we conducted a cohort study of clopidogrel initiators who encountered treatment with SSRI during their clopidogrel therapy. Patients were matched by propensity score (PS) and followed for as long as they were exposed to both clopidogrel and index SSRI group. Outcomes were a composite ischemic event (myocardial infarction, ischemic stroke, or a revascularization procedure, whichever came first) and a composite major bleeding event (gastrointestinal bleed or hemorrhagic stroke, whichever came first). Results were combined via random-effects meta-analysis with previous evidence from subjects initiating clopidogrel while on SSRI therapy.

RESULTS

The PS-matched cohort comprised 2346 clopidogrel users starting CYP2C19-inhibiting SSRI therapy and 16,115 starting other SSRIs (mean age 61 years; 59% female). Compared with those treated with a non-inhibiting SSRI, the hazard ratio (HR) for patients treated with a CYP2C19-inhibiting SSRI was 1.07 (95% confidence interval [CI] 0.82-1.40) for the ischemic outcome and 1.00 (95% CI 0.42-2.36) for bleeding. The pooled estimates were 1.11 (95% CI 1.01-1.22) for ischemic events and 0.80 (95% CI 0.55-1.18) for bleeding.

CONCLUSIONS

We observed similar estimates of association between the two studies. The updated evidence still indicates a small decrease in clopidogrel effectiveness associated with concomitant exposure to clopidogrel and CYP2C19-inhibiting SSRIs.

A decline in depression treatment following FDA antidepressant warnings largely explains racial/ethnic disparities in prescription fills

BACKGROUND

The Food and Drug Administration's 2004 antidepressant warning was followed by decreases in antidepressant prescribing for youth. This was due to declines in all types of depression treatment, not just the intended changes in antidepressant prescribing patterns. Little is known about how these patterns varied by race/ethnicity.

METHOD

Data are Medicaid claims from four U.S. states (2002-2009) for youth ages 5-17. Interrupted time series analyses measured changes due to the warning in levels and trends, by race/ethnicity, of three outcomes: antidepressant prescription fills, depression treatment visits, and incident fluoxetine prescription fills.

RESULTS

Prewarning, antidepressant fills were increasing across all racial/ethnic groups, fastest for White youth. Postwarning, there was an immediate drop and continued decline in the rate of fills among White youth, more than double the decline in the rate among Black and Latino youth. Prewarning, depression treatment visits were increasing for White and Latino youth. Postwarning, depression treatment stabilized among Latinos, but declined among White youth. Prewarning, incident fluoxetine fills were increasing for all groups. Postwarning, immediate increases and increasing trends of fluoxetine fills were identified for all groups.

CONCLUSIONS

Antidepressant prescription fills declined most postwarning for White youth, suggesting that risk information may have diffused less rapidly to prescribers or caregivers of minorities. Decreases in depression treatment visits help to explain the declines in antidepressant prescribing and were largest for White youth. An increase in incident fluoxetine fills, the only medication indicated for pediatric depression at the time, suggests that the warning may have shifted prescribing practices.

Oral anticoagulation use by patients with atrial fibrillation in Germany

Atrial fibrillation (AF) is the most common significant cardiac rhythm disorder. Oral anticoagulation (OAC) is recommended by guidelines in the presence of a moderate to high risk of stroke. Based on an analysis of claims-based data, the aim of this contribution is to quantify the stroke-risk dependent OAC utilisation profile of German AF patients as well as the possible causes and the associated clinical outcomes of OAC under-use. Our data set was derived from two large mandatory German medical insurance funds. Risk stratification of patients was based on the CHADS2-score and the CHA2DS2-VASc-score. Two different scenarios were constructed to deal with factors potentially disfavouring OAC use. Causes of OAC under-use and its clinical consequences were analysed using multivariate analysis. Observation year was 2008. A total of 183,448 AF patients met the inclusion criteria. This represents an AF prevalence of 2.21%. The average CHADS2-score was 2.8 (CHA2DS2-VASc-score: 4.3). On between 40.5 and 48.7% of the observed patient-days, there was no antithrombotic protection by OAC, other anticoagulants or aspirin. Older female patients with a high number of comorbidities had a higher risk of OAC under-use. Patients who had already experienced a thromboembolic event had a lower risk of OAC under-use. In the observation year, 3,367 patients experienced a stroke (incidence rate 1.8%). In our multi-level Poisson random effects estimate, OAC use decreases the stroke rate by almost 80% (IRR 0.236). In conclusion, OAC under-use is widespread in the German market. It is associated with severe clinical consequences.

Novel Methods and Data Sources for Surveillance of State-Level Diabetes and Prediabetes Prevalence

States bear substantial responsibility for addressing the rising rates of diabetes and prediabetes in the United States. However, accurate state-level estimates of diabetes and prediabetes prevalence that include undiagnosed cases have been impossible to produce with traditional sources of state-level data. Various new and nontraditional sources for estimating state-level prevalence are now available. These include surveys with expanded samples that can support state-level estimation in some states and administrative and clinical data from insurance claims and electronic health records. These sources pose methodologic challenges because they typically cover partial, sometimes nonrandom subpopulations; they do not always use the same measurements for all individuals; and they use different and limited sets of variables for case finding and adjustment. We present an approach for adjusting new and nontraditional data sources for diabetes surveillance that addresses these limitations, and we present the results of our proposed approach for 2 states (Alabama and California) as a proof of concept. The method reweights surveys and other data sources with population undercoverage to make them more representative of state populations, and it adjusts for nonrandom use of laboratory testing in clinically generated data sets. These enhanced diabetes and prediabetes prevalence estimates can be used to better understand the total burden of diabetes and prediabetes at the state level and to guide policies and programs designed to prevent and control these chronic diseases.

Drug survival of second biological DMARD therapy in patients with rheumatoid arthritis: a retrospective non-interventional cohort analysis

Background

Since persistence to first biological disease modifying anti-rheumatic drugs (bDMARDs) is far from ideal in rheumatoid arthritis (RA) patients, many do receive a second and/or third bDMARD treatment. However, little is known about treatment persistence of the second-line bDMARD and it is specifically unknown whether the mode of action of such a treatment is associated with different persistence rates. We aimed to assess discontinuation-, re-initiation- or continuation-rates of a 2nd bDMARD therapy as well as switching-rates to a third biological DMARD (3rd bDMARD) therapy in RA patients.

Method

Analysis was based on German claims data (2010–2013). Patients were included if they had received at least one prescription for an anti-TNF and at least one follow-up prescription of a 2nd bDMARD different from the first anti-TNF. Patient follow-up started on the date of the first prescription for the 2nd bDMARD and lasted for 12 months or until a patient’s death.

Results

2667 RA patients received at least one anti-TNF prescription. Of these, 451 patients received a second bDMARD (340 anti-TNF, mean age 52.6 years; 111 non-anti-TNF, mean age 55.9 years).

During the follow-up, 28.8% vs. 11.7% of the 2nd anti-TNF vs. non-anti-TNF patients (p < 0.001) switched to a 3rd bDMARD; 14.1% vs. 19.8% (p = 0.179) discontinued without re-start; 3.8% vs.1.8% (p = 0.387) re-started and 53.5 vs. 66.7% (p < 0.050) continued therapy. Patients in the non-anti-TNF group demonstrated longer drug survival (295 days) than patients in the anti-TNF group (264 days; p = 0.016).

Independent variables associated with earlier discontinuation (including re-start) or switch were prescription of an anti-TNF as 2nd bDMARD (HR = 1.512) and a higher comorbidity level (CCI, HR = 1.112), whereas previous painkiller medication (HR = 0.629) was associated with later discontinuation or switch.

Conclusions

Only 56.8% of RA patients continued 2nd bDMARD treatment after 12 months; 60% if re-start was included. Non-anti-TNF patients had a higher probability of continuing 2nd bDMARD therapy.

Electronic supplementary material

The online version of this article (doi:10.1186/s12891-017-1684-0) contains supplementary material, which is available to authorized users.

Use of prasugrel vs clopidogrel and outcomes in patients with acute coronary syndrome undergoing percutaneous coronary intervention in contemporary clinical practice: Results from the PROMETHEUS study

BACKGROUND AND OBJECTIVES

We sought to determine the frequency of use and association between prasugrel and outcomes in acute coronary syndrome patients undergoing percutaneous coronary intervention (PCI) in clinical practice.

METHODS

PROMETHEUS was a multicenter observational registry of acute coronary syndrome patients undergoing PCI from 8 centers in the United States that maintained a prospective PCI registry for patient outcomes. The primary end points were major adverse cardiovascular events at 90days, a composite of all-cause death, nonfatal myocardial infarction, stroke, or unplanned revascularization. Major bleeding was defined as any bleeding requiring hospitalization or blood transfusion. Hazard ratios (HRs) were generated using multivariable Cox regression and stratified by the propensity to treat with prasugrel.

RESULTS

Of 19,914 patients (mean age 64.4years, 32% female), 4,058 received prasugrel (20%) and 15,856 received clopidogrel (80%). Prasugrel-treated patients were younger with fewer comorbid risk factors compared with their counterparts receiving clopidogrel. At 90days, there was a significant association between prasugrel use and lower major adverse cardiovascular event (5.7% vs 9.6%, HR 0.58, 95% CI 0.50-0.67, P<.0001) and bleeding (1.9% vs 2.9%, HR 0.65, 95% CI 0.51-0.83, P<.001). After propensity stratification, associations were attenuated and no longer significant for either outcome. Results remained consistent using different approaches to adjusting for potential confounders.

CONCLUSIONS

In contemporary clinical practice, patients receiving prasugrel tend to have a lower-risk profile compared with those receiving clopidogrel. The lower ischemic and bleeding events associated with prasugrel use were no longer evident after accounting for these baseline differences.

The quality of Medicaid and Medicare data obtained from CMS and its contractors: implications for pharmacoepidemiology

Background

Administrative claims of United States Centers for Medicare and Medicaid Services (CMS) beneficiaries have long been used in non-experimental research. While CMS performs in-house checks of these claims, little is known of their quality for conducting pharmacoepidemiologic research. We performed exploratory analyses of the quality of Medicaid and Medicare data obtained from CMS and its contractors.

Methods

Our study population consisted of Medicaid beneficiaries (with and without dual coverage by Medicare) from California, Florida, New York, Ohio, and Pennsylvania. We obtained and compiled 1999–2011 data from these state Medicaid programs (constituting about 38% of nationwide Medicaid enrollment), together with corresponding national Medicare data for dually-enrolled beneficiaries. This descriptive study examined longitudinal patterns in: dispensed prescriptions by state, by quarter; and inpatient hospitalizations by federal benefit, state, and age group. We further examined discrepancies between demographic characteristics and disease states, in particular frequencies of pregnancy complications among men and women beyond childbearing age, and prostate cancers among women.

Results

Dispensed prescriptions generally increased steadily and consistently over time, suggesting that these claims may be complete. A commercially-available National Drug Code lookup database was able to identify the dispensed drug for 95.2–99.4% of these claims. Because of co-coverage by Medicare, Medicaid data appeared to miss a substantial number of hospitalizations among beneficiaries ≥ 45 years of age. Pregnancy complication diagnoses were rare in males and in females ≥ 60 years of age, and prostate cancer diagnoses were rare in females.

Conclusions

CMS claims from five large states obtained directly from CMS and its contractors appeared to be of high quality. Researchers using Medicaid data to study hospital outcomes should obtain supplemental Medicare data on dual enrollees, even for non-elders.

Trial Registration

Not applicable.

Electronic supplementary material

The online version of this article (doi:10.1186/s12913-017-2247-7) contains supplementary material, which is available to authorized users.

Index to Predict In-hospital Mortality in Older Adults after Non-traumatic Emergency Department Intubations

INTRODUCTION

Our goal was to develop and validate an index to predict in-hospital mortality in older adults after non-traumatic emergency department (ED) intubations.

METHODS

We used Vizient administrative data from hospitalizations of 22,374 adults ≥75 years who underwent non-traumatic ED intubation from 2008-2015 at nearly 300 U.S. hospitals to develop and validate an index to predict in-hospital mortality. We randomly selected one half of participants for the development cohort and one half for the validation cohort. Considering 25 potential predictors, we developed a multivariable logistic regression model using least absolute shrinkage and selection operator method to determine factors associated with in-hospital mortality. We calculated risk scores using points derived from the final model's beta coefficients. To evaluate calibration and discrimination of the final model, we used Hosmer-Lemeshow chi-square test and receiver-operating characteristic analysis and compared mortality by risk groups in the development and validation cohorts.

RESULTS

Death during the index hospitalization occurred in 40% of cases. The final model included six variables: history of myocardial infarction, history of cerebrovascular disease, history of metastatic cancer, age, admission diagnosis of sepsis, and admission diagnosis of stroke/ intracranial hemorrhage. Those with low-risk scores (<6) had 31% risk of in-hospital mortality while those with high-risk scores (>10) had 58% risk of in-hospital mortality. The Hosmer-Lemeshow chi-square of the model was 6.47 (p=0.09), and the c-statistic was 0.62 in the validation cohort.

CONCLUSION

The model may be useful in identifying older adults at high risk of death after ED intubation.

Impact of an Interaction Between Clopidogrel and Selective Serotonin Reuptake Inhibitors

Clopidogrel is a pro-drug that requires activation by the cytochrome P450 (CYP) enzyme system. Patients receiving clopidogrel are often treated with selective serotonin reuptake inhibitors (SSRIs) for co-existing depression. SSRIs that inhibit the CYP2C19 enzyme have the potential to reduce the effectiveness of clopidogrel. Using 5 US databases (1998 to 2013), we conducted a cohort study of adults who initiated clopidogrel while being treated with either an SSRI that inhibits CYP2C19 (fluoxetine and fluvoxamine) or a noninhibiting SSRI. Patients were matched by propensity score and followed for as long as they were exposed to both clopidogrel and the index SSRI group (primary analysis) or for 180 days after clopidogrel initiation (sensitivity analysis). Outcomes included a composite ischemic event (myocardial infarction, ischemic stroke, or a revascularization procedure) and a composite major bleeding event (gastrointestinal bleed or hemorrhagic stroke). The final propensity score-matched cohort comprised 9,281 clopidogrel initiators on CYP2C19-inhibiting SSRIs and 44,278 clopidogrel initiators on a noninhibiting SSRIs. Compared with those treated with a noninhibiting SSRI, patients on a CYP2C19-inhibiting SSRI had an increased risk of ischemic events (hazard ratio [HR] 1.12; 95% confidence interval [CI] 1.01 to 1.24), which was more pronounced in patients ≥65 years (HR 1.22; 95% CI 1.00 to 1.48). The HR for major bleeding was 0.76 (95% CI 0.50 to 1.17). In conclusion, the findings from this large, population-based study suggest that being treated with a CYP2C19-inhibiting SSRI when initiating clopidogrel may be associated with slight decrease in effectiveness of clopidogrel.

Adherence to antiepileptic drugs among diverse older Americans on Part D Medicare

INTRODUCTION

Older minority groups are more likely to have poor AED adherence. We describe adherence to antiepileptic drugs (AEDs) among older Americans with epilepsy.

METHODS

In retrospective analyses of 2008-2010 Medicare claims for a 5% random sample of beneficiaries augmented by minority representation, epilepsy cases in 2009 were those with ≥1 claim with ICD-9345.x or ≥2 with 780.3x, and ≥1 AED. New-onset cases had no such claims or AEDs in the year before the 2009 index event. We calculated the Proportion of Days Covered (PDC) (days with ≥1 AED over total follow-up days) and used logistic regression to estimate associations of non-adherence (PDC <0.8) with minority group adjusting for covariates.

RESULTS

Of 36,912 epilepsy cases (19.2% White, 62.5% African American (AA), 11.3% Hispanic, 5.0% Asian and 2% American Indian/Alaskan Native), 31.8% were non-adherent (range: 24.1% Whites to 34.3% AAs). Of 3706 new-onset cases, 37% were non-adherent (range: 28.7% Whites to 40.5% AAs). In adjusted analyses, associations with minority group were significant among prevalent cases, and for AA and Asians vs. Whites among new cases. Among other findings, beneficiaries from high-poverty ZIP codes were more likely to be non-adherent than their counterparts, and those in cost-sharing drug benefit phases were less likely to be non-adherent than those in deductible phases.

CONCLUSION

About a third of older adults with epilepsy have poor AED adherence; minorities are more likely than Whites. Investigations of reasons for non-adherence, and interventions to promote adherence, are needed with particular attention to the effect of cost-sharing and poverty.

Validating a proxy for disease progression in metastatic cancer patients using prescribing and dispensing data

BACKGROUND

Routine data collections are used increasingly to examine outcomes of real-world cancer drug use. These datasets lack clinical details about important endpoints such as disease progression.

AIM

To validate a proxy for disease progression in metastatic cancer patients using prescribing and dispensing claims.

METHODS

We used data from a cohort study of patients undergoing chemotherapy who provided informed consent to the collection of cancer-treatment data from medical records and linkage to pharmaceutical claims. We derived proxy decision rules based on changes to drug treatment in prescription histories (n = 36 patients) and validated the proxy in prescribing data (n = 62 patients). We adapted the decision rules and validated the proxy in dispensing data (n = 109). Our gold standard was disease progression ascertained in patient medical records. Individual progression episodes were the unit of analysis for sensitivity and Positive Predictive Value (PPV) calculations and specificity and Negative Predictive Value (NPV) were calculated at the patient level.

RESULTS

The sensitivity of our proxy in prescribing data was 74.3% (95% Confidence Interval (CI), 55.6-86.6%) and PPV 61.2% (95% CI, 45.0-75.3%); specificity and NPV were 87.8% (95% CI, 73.8-95.9%) and 100% (95% CI, 90.3-100%), respectively. In dispensing data, the sensitivity of our proxy was 64% (95% CI, 55.0-77.0%) and PPV 56.0% (95% CI, 43.0-69.0%); specificity and NPV were 81% (95% CI, 70.05-89.0%) and 91.0% (95% CI, 82.0-97.0%), respectively.

CONCLUSION

Our proxy overestimated episodes of disease progression. The proxy's performance is likely to improve if the date of prescribing is used instead of date of dispensing in claims data and by incorporating medical service claims (such as imaging prior to drug changes) in the algorithm. Our proxy is not sufficiently robust for use in real world comparative effectiveness research for cancer medicines.

Validity of a stroke severity index for administrative claims data research: a retrospective cohort study

Background

Ascertaining stroke severity in claims data-based studies is difficult because clinical information is unavailable. We assessed the predictive validity of a claims-based stroke severity index (SSI) and determined whether it improves case-mix adjustment.

Methods

We analyzed patients with acute ischemic stroke (AIS) from hospital-based stroke registries linked with a nationwide claims database. We estimated the SSI according to patient claims data. Actual stroke severity measured with the National Institutes of Health Stroke Scale (NIHSS) and functional outcomes measured with the modified Rankin Scale (mRS) were retrieved from stroke registries. Predictive validity was tested by correlating SSI with mRS. Logistic regression models were used to predict mortality.

Results

The SSI correlated with mRS at 3 months (Spearman rho = 0.578; 95 % confidence interval [CI], 0.556–0.600), 6 months (rho = 0.551; 95 % CI, 0.528–0.574), and 1 year (rho = 0.532; 95 % CI 0.504–0.560). Mortality models with the SSI demonstrated superior discrimination to those without. The AUCs of models including the SSI and models with the NIHSS did not differ significantly.

Conclusions

The SSI correlated with functional outcomes after AIS and improved the case-mix adjustment of mortality models. It can act as a valid proxy for stroke severity in claims data-based studies.

Electronic supplementary material

The online version of this article (doi:10.1186/s12913-016-1769-8) contains supplementary material, which is available to authorized users.

Medication Adherence Patterns after Hospitalization for Coronary Heart Disease. A Population-Based Study Using Electronic Records and Group-Based Trajectory Models

Objective

To identify adherence patterns over time and their predictors for evidence-based medications used after hospitalization for coronary heart disease (CHD).

Patients and Methods

We built a population-based retrospective cohort of all patients discharged after hospitalization for CHD from public hospitals in the Valencia region (Spain) during 2008 (n = 7462). From this initial cohort, we created 4 subcohorts with at least one prescription (filled or not) from each therapeutic group (antiplatelet, beta-blockers, ACEI/ARB, statins) within the first 3 months after discharge. Monthly adherence was defined as having ≥24 days covered out of 30, leading to a repeated binary outcome measure. We assessed the membership to trajectory groups of adherence using group-based trajectory models. We also analyzed predictors of the different adherence patterns using multinomial logistic regression.

Results

We identified a maximum of 5 different adherence patterns: 1) Nearly-always adherent patients; 2) An early gap in adherence with a later recovery; 3) Brief gaps in medication use or occasional users; 4) A slow decline in adherence; and 5) A fast decline. These patterns represented variable proportions of patients, the descending trajectories being more frequent for the beta-blocker and ACEI/ARB cohorts (16% and 17%, respectively) than the antiplatelet and statin cohorts (10% and 8%, respectively). Predictors of poor or intermediate adherence patterns were having a main diagnosis of unstable angina or other forms of CHD vs. AMI in the index hospitalization, being born outside Spain, requiring copayment or being older.

Conclusion

Distinct adherence patterns over time and their predictors were identified. This may be a useful approach for targeting improvement interventions in patients with poor adherence patterns.

Expenditures for Persons Living With HIV Enrolled in Medicaid, 2006-2010

BACKGROUND

Costs of care for persons living with HIV have been high historically. Cost estimates based on data from 1 health care site may underestimate total expenditures; using insurance claims avoids this limitation. We used Medicaid claims data to comprehensively assess payments for care for persons living with HIV between 2006 and 2010.

METHODS

Five sites from the HIV Research Network (HIVRN) provided information on patients with Medicaid coverage. Medicaid data were obtained from the sites' states (MD, NY, and MA) and 3 surrounding states and matched to HIVRN medical record-based data. Individuals less than 18, those with Medicare, and those in Medicaid managed care plans were excluded. Medicaid and HIVRN data were compared to ascertain concordance in capturing any inpatient event and any antiretroviral (ART) medication use.

RESULTS

Of 6892 unique HIVRN identifiers, 6196 (90%) were linked to Medicaid data. The analytic sample included 11,341 person-years of Medicaid claims data from 3695 individuals in fee-for-service (FFS) programs. The mean annual FFS payment for all services was $47,434; mean annual FFS payment for only medical services was $38,311. Concordance between Medicaid and HIVRN data was excellent for ART use, but HIVRN data did not record a substantial proportion of years in which Medicaid recorded inpatient use.

CONCLUSIONS

Estimated Medicaid payment amounts in this study are higher than some previous estimates. More complete capture of expensive inpatient hospitalizations in Medicaid data may partially explain this finding. Although inpatient care and ART medications contribute the most, expenditures for nonmedical services are substantial.

Factors Associated With Musculoskeletal Injuries in Children and Adolescents With Attention-Deficit/Hyperactivity Disorder

BACKGROUND

Musculoskeletal injuries may be associated with attention-deficit/hyperactivity disorder (ADHD) symptom severity, comorbid psychiatric or medical conditions, and the prescribed psychostimulant.

METHODS

A population-based, retrospective cohort design was employed using South Carolina's Medicaid claims data set covering outpatient and inpatient medical services and medication prescriptions over an 11-year period (January 1, 1996, through December 31, 2006) for patients ≤ 17 years of age with ≥ 2 visits for ICD-9-CM diagnostic codes for ADHD. A cohort of 7,725 cases was identified and analyzed using logistic regression to compare risk factors for those who sustained focal musculoskeletal injuries and those who did not.

RESULTS

The risk of sustaining sprains, arthropathy and connective tissue disorders, or muscle and joint disorders was significantly related to being diagnosed with comorbid hypertension (adjusted odds ratios [aORs] = 1.60, 2.09, and 1.46, respectively) and a substance use disorder (aORs = 1.58, 1.38, and 1.28). Having a substance use disorder was also related to incident fractures and dorso/spinal injuries (aORs = 1.42 and 1.21). Diagnosed hypertension was related to incident concussions (aOR = 2.00), a diagnosed thyroid disorder was related to an increased risk of sprain and concussion (aORs = 1.44 and 2.05), a diagnosed anxiety disorder was related to an increased risk of dorso/spinal disorders (aOR = 1.71), and diagnosed diabetes was related to incident bone and cartilage disorders (aOR = 1.61).

CONCLUSIONS

Comorbid hypertension, substance use disorders, and thyroid disorders deserve increased clinical surveillance in children and adolescents with ADHD because they may be associated with an increased risk of more than one musculoskeletal injury.